A Role for International Vertical Joint

Ventures Within An Automobile

Industry Paradigm

A case study of MG Rover in comparison with

the global automobile industry

Cardiff
UNIVERSITY

PRI FYSGOL

CaeRDY[§)

A thesis submitted for the Degree of Doctor of Philosophy

Cardiff University

by

Michael S Wynn-Williams

Cardiff Business School

Cardiff University

October 2007
 UMI Number: U584197

All rights reserved

INFORMATION TO ALL USERS

The quality of this reproduction is dependent upon the quality of the copy submitted.

In the unlikely event that the author did not send a complete manuscript and there are
missing pages, these will be noted. Also, if material had to be removed, a note will
indicate the deletion.

Disscrrlation Publishing

UMI U584197
Published by ProQuest LLC 2013. Copyright in the Dissertation held by the Author.
Microform Edition © ProQuest LLC. All rights reserved.
This work is protected against unauthorized copying under
Title 17, United States Code.

ProQuest LLC
789 East Eisenhower Parkway
P.O. Box 1346
Ann Arbor, Ml 48106-1346
 Abstract

This research puts forward a conceptual framework for automobile companies by

constructing an automobile industry paradigm. The precepts of the paradigm are based
upon investigations of transaction costs and economies of scale. This results in a full-
function model of vertical integration for an automobile firm operating with prescribed
levels of production output. The research then uses historical case studies and empirical
data to look at the extent to which automobile companies are competitive within the
paradigm, particular reference being made to the main case study company, MG Rover.
Contextual empirical data was gathered using semi structured elite interview techniques
conducted with senior executives throughout the automobile industry in the UK and Japan.
Through survivor analysis and investigation of leading plant sizes the research found that
although there is a minimum efficient scale (MES) for a company, this being the
prototypical automobile firm with optimal output MESp, there are additional advantages to
be gained from output expansion. Conversely, firms that are wholly uncompetitive can
replicate the prototypical model by exploiting divisibilities due to differences between
human and capital assets as well as international factor endowment differentiation. The
organisational structure for achieving this is the international vertical joint venture (IVJV).
There is evidence to suggest that MG Rover was moving towards this type of structure but
was precluded from doing so due to its financial collapse in April 2005. Since then, the
purchasers of MG Rover’s assets, SAIC and NAC of China, have separately continued with
variations of the IVJV structure in order to approximate to the requirements of the
automobile industry paradigm.
DECLARATION

This work has not previously been accepted in substance for any degree and is not
being concurrently submitted in candidature for any degree.

Signed..- .(candidate)

Date.... 1.991

STATEMENT 1

This thesis is the result of my own investigations, except where otherwise stated.
Other sources are acknowledged by footnotes giving explicit references. A
bibliogr^hv is appended.

SignedffT. (candidate)

STATEMENT 2

I hereby give consent for my thesis, if accepted, to be available for photocopying and for
inter-library loan, and for the title and summary to be made available to outside
organisations.

Signed.. (candidate)

Ac?
 Acknowledgements

This is my third thesis for this PhD. My first thesis was as magnificent and tragic as the
Titanic, the second as industrious as a cross-Channel ferry. For this thesis, the last of the
line, I hope I have imbued it with the strength of an ice-breaker and the grace of a
schooner: it certainly feels as heavy as a battleship.

I was fortunate to gain funding from the EPSRC, which included several trips to Japan, and
I hope the organisation finds value in their investment. I was later sponsored by Trend
Tracker Ltd, with thanks to Chris Oakham. In conducting the research I would like to thank
everyone who allowed me to interrogate them. All interviewees gave their time freely and I
have nothing but fond memories of the fieldwork. At Cardiff Business School, Dr Paul
Nieuwenhuis assisted with the automotive related material and can be said to be the Father
of the Budd Paradigm upon which this research is based. Professor Garel Rhys was
inspirational and had a far greater impact on the research than the brevity of his advice
might suggest. My examiners were Professor Max Munday and Dr David Bailey, who were
generous in their mercy but firm in their expectations. Toiling away in the background
were Elsie Phillips and Laine Clayton, whose quiet efficiency and ready sympathy was not
unnoticed. Most of all, I am grateful to Professor Trevor Boyns who supervised beyond-
the-call-of-duty and without whose tireless goading this thesis would never have attained
the standard that it did.

Throughout this six-year sentence of penal servitude in the cause of automotive research I
have been sustained by a host of people, many of whom will not be aware of the support
they have given me. It was Bethan Roberts who first suggested this course in my life,
perhaps in a moment of idle jest, and Dr Chris Nightingale, who put my application into a
credible form. In the second year of the research, when my morale hit rock bottom, the
arrival of Jyoti Butel (cheers mate!) and Stefan L^ng raised my spirits high enough to give
me the energy to persevere. My parents were unstinting in their support, as were my
godfather and his wife, Derek and Sheila Watson. Somewhere in the background there was
always Roger Waters, though my admiration for him is probably the safer for having never
met him. There are also those who have fallen on the way, such as Rhossili, Micra and
Fiesta; they were little troopers to the end. Ultimately, though, there is one who has stuck
by me over the past decade with a project of her own: me. Well, Kerry Ann Kludsikofsky,
you have passed.
 Contents
Section Title Page

1: Introduction 1

1.1 Growth of a global industry 1

1.2 Defining the automobile industry paradigm 5
2: Methodology 14

2.1 The role of research questions 15

2.1.1 Scope of investigative methods 15

2.1.2 Qualitative and quantitative research designs 20
2.2 Case study design 22

2.2.1 Critical perspectives 22

2.2.2 Defining the case study as a research strategy 23
2.2.3 Kuhn and paradigm revolutions 25
2.2.4 Case Studies as paradigm and methodology 27
2.2.5 Types of case study 28
2.2.6 Stake and collective case studies 29
2.2.7 Yin and the matrix 29
2.3 Data collection 32

2.3.1 Elite interviews 33

2.3.2 Linking of interviews - by content 37
2.3.3 Linking of interviews - by respondent 39
2.3.4 Data recording and coding 42
2.4 Validity, reliability and generalisability 44

2.4.1 Reliability 44
2.4.2 Internal validity 45
2.4.3 External validity and generalisability 46
2.5 Ethical concerns 48

2.6 Foundation for research design 51

3: Research Design 53

3.1 Qualitative case study 53

3.2 Selection of the case study company 55

3.3 Company access 58
3.3.1 Case study access - MG Rover 58
3.3.2 Contextualindustry access 59
3.3.3 Access hold-ups 61

3.4 Other documented information 62

3.5 Triangulation 63

3.5.1 Literary and empirical source triangulation 65

3.5.2 Interview triangulation 66
3.5.3 Interview frequency 69

3.6 Ethics 71
3.7 Initial phase 73

3.7.1 Morgan Motor Company - an introduction to fieldwork 73

3.7.2 Trade union perspective 74
3.7.3 International motor shows 75

3.8 Main interviews 77

3.8.1 MG Rover 77
3.8.2 Contextual industry 78
3.9 Recording and coding 80

3.9.1 Recording 80
3.9.2 Coding 81
3.10 Conclusion 83

4: Considerations of Scale 85
4.1 Economies of size 86

4.1.1 Economies of scale 87

4.1.2 Economies of scope 90
4.2 Production costs 93

4.2.1 Short-run costs 94

4.2.2 Long-run costs 96
4.2.3 Alternative long-run average cost curves 100
4.2.4 Integrating differentproduction processes 101
4.2.5 Cost curves for themulti-plant firm 103
4.2.6 Survivor analysis 106
4.3 Economies of scale in the automobile industry 111
4.3.1 Survivor analysis in the automobile industry 113
4.3.2 Body construction 123
4.3.2.1 The Budd Paradigm 124
4.3.2.2 Economies of scale in body pressings 126
4.3.3 Powertrain 129
4.3.4 Final assembly 132
4.3.4.1 Final assembly economies of scale 133
4.3.5 Research and development: 136
economies of scale and scope 139
4.3.5.1 Vehicle research and development
4.4 Combining the LAC curves 145

4.5 Conclusion 152

5: Structuring the Automobile Industry Paradigm 153

Part I: Company Structure 154

5.1 Vertical integration and the full-function company
155
5.1.1 Component supply industry and functional relationships 157
5.1.2 Transactions as mechanical friction 158
5.1.3 Contracts and the market 161
5.1.4 The imperfect contract 163
5.1.5 Specificity 165
5.1.6 Site asset specificity 166
5.1.7 Physical asset specificity 168
5.1.8 Human asset specificity 169
5.1.9 Unitary and multidivisional governance structures 172
Part II: Constructing the Automobile Industry
185
Paradigm
5.2 Constructing the automobile industry paradigm
186
5.2.1 Paradigmatic view of the automobile industry 189
5.2.2 Vertical integration of the automobile paradigm 192
5.2.3 Defining the functions 194
5.2.4 Integration with powertrain 197
5.2.5 Integrating body production - 198
the case of GM and Fisher Body
5.2.6 Integration with design 204
5.3 Resultant size and shape
208
5.4 Empirical evidence of vertical integration 211
5.4.1 Vertical integration - contextual sources 212
5.4.2 Vertical integration - MG Rover 220
5.5 Conclusion 224

6: Approximating to the AutomobileIndustry 227

Paradigm
6.1 Introduction 228

6.1.1 Sub-scale production 230

6.2 Internal approximations to the automobile paradigm 233
6.2.1 Extending use of the production system 233
6.2.2 Economies of scope and scale in flexible production 234
6.3 External approximations to the automobile paradigm 238

6.3.1 Hybrid organisational structure 240

6.3.2 Co-operative organisation structure 241
6.3.3 Human assets in link alliances 246
6.4 Opportunities raised by globalisation 250

6.4.1 Globalisation of trade 252

6.4.2 Critics of globalisation 254
6.4.3 Foreign direct investment 257
6.4.4 Dynamic theories of EDI 260
6.4.5 The eclectic paradigm 265
6.4.6 Strategic EDI 269
6.4.7 Ford: a failure of global structure 273
6.5 Reassessing the full-function structure 277

6.5.1 Vertical joint venture at GM 278

6.5.2 Human assets and the full-function model 281
6.5.3 International vertical joint venture: Rootes Group 282
6.6 Conclusion 286

7: A Sustainable Model for MG Rover 288

Part I: The Contextual Automobile Industry 291

7.1 Internal approximations to the automobile industry 291

paradigm
7.2 Structure and opportunities of globalisation - 294

Institutions
7.3 Globalisation and the contextual automobile industry 297
7.4 Collaboration 302
7.5 Conclusion - adaptation to the automobile industry 305

paradigm by contextual companies

Part II: MG Rover 307

7.6 The rise and fall of British Leyland/Rover Group 307

7.6.1 The consolidation of MG and Rover 307

7.6.2 Internal rescue strategies - 1975 to 1978 309
7.6.3 External rescue strategy - link and scale alliances 312
7.6.4 BAe purchases Rover Group 321
7.6.5 External assistance for internal improvement - 324
BMW
7.7 MG Rover - autonomy and sustainability 328

7.7.1 MG Rover internal approximations - 329

literary sources
7.7.2 MG Rover internal approximations - 331
contemporary views
7.7.3 MG Rover and globalisation - empirical data 332
7.7.4 MG Rover globalisation and industry - empirical data 332
7.7.5 Collaboration - MG Rover literary sources 336
7.7.6 Collaboration - MG Rover empirical data 343
7.8 MG Rover assets - reconstructing the 350

international vertical joint venture

7.9 Conclusion 356

8: Conclusions 359
8.1 Current state of the industry 360

8.2 Paradigm revolution for the automobile industry 363

8.3 Implications for national policy 366

8.4 Applicability to other industries 367

8.5 Suggestions for subsequent research 370

8.5 Closing remarks 371

References 372
Appendix 1 396
Appendix 2 399

Appendix 3 404

Appendix 4 405

Appendix 5 406

Appendix 6 407

Appendix 7 409

Appendix 8 410

Appendix 9 413

Appendix 10 415

Appendix 11 417

Appendix 12 418

Appendix 13 419

Appendix 14 420
 List of Figures
Figure Title Page

1.1 Fundamental arguments to the automobile industry 6

paradigm
2.1 Basic types of case study 31

3.1 Frequency of interviews 69

4.1 Fixed cost, variable cost and total cost in the short-run 94

4.2 Plant cost in the short-run 95

4.3 Alternative plant sizes for long-run planning 97
4.4 Marginal cost in the long and short-run 99
4.5 Alternative LAC curves 101
4.6 Matching machine outputs 102
4.7 LAC for two plants 105
4.8 Total global production, 1975-2005 115
4.9 Global production share of seven output ranges 116
4.10 Global production share of five output ranges 118
4.11 Production share for the industry leaders, 1975-2005 119
4.12 Global production output for top 5 manufacturers, 120
1975-2005
4.13 Theoretical economies of scale for panel stamping 126
4.14 Economies of scale for BIW production 129
4.15 Economies of scale for powertrain production 132
4.16 Economies of scale for final assembly 135
4.17 LAC curve for R&D, various industries 138
4.18 Economies of scale for R&D 144
4.19 LAC curves for a prototypical integrated car firm 146
4.20 Automobile company exploiting acceptable 147
economies of scale
4.21 SATC curve for a prototypical integrated automobile firm 148
4.22 SATC and LAC curves for automobile industry firms 151
5.1 Typical full-function producer 156

5.2 U-Form functional governance structure 173

5.3 M-Form multidivisional govemance structure 176
5.4 Hybrid U-Form and M-Form govemance stmctures 182
5.5 Typical full-function automobile producer 193
5.6 Full-function model quantified 208
5.7 Automobile industry paradigm 210
6.1 Pre-global trade 253

6.2 Direction of trade 1997 - inter-triad and intra-triad 255

trade movement proportions
6.3 Vertical joint venture between A O Smith and GM 280
6.4 Vertical joint venture between Rootes Group and 284
Iran Khodro: Iranian standpoint, 1993
7.1 Honda and BL joint venture 1981 - Triumph Acclaim 316
7.2 Honda and BL link and scale alliance 1987 - Project XX 318
7.3 Proposed IVJV between MG Rover and SAIC 342
7.4 IVJV style relationship between Ricardo 2010 R&D 354
and SAIC
8.1 Automobile production output and forecast 2000-2020 363

8.2 Proposed IVJV between Harland and Wolff and a 369

shipbuilding company
 List of Tables
Table Title Page

1.1 Automobile producing countries, 2006 3

2.1 Applications of research strategies 16

3.1 Range of potential alternative cases to MG Rover 56

3.2 Contextual interview categories 62

4.1 Four major automobile functions 112

4.2 Development of MES in the automobile industry 113

4.3 Top thirteen automobile producing nations 114
4.4 Top twenty manufacturers, 2005 122
4.5 Typical Budd plant investment 125
4.6 Japanese powertrain capacity in the US 2004 131
4.7 Japanese assembly operations in the US 2004 133
4.8 Global R&D Expenditure 2005/6 140
4.9 R&D expenditure per model 142
5.1 Honda - areas of control and derived benefits 212

5.2 Honda - functional integration and its impact 215

5.3 Automobile companies and dependency/autonomy 217
5.4 Divisional control of operations and strategy 219
5.5 MG Rover - theoretical and practical considerations 221
of autonomy
6.1 Plant site selection criteria 268

7.1 Institutions of globalisation - effectiveness 295

7.2 Institutions of globalisation - effect on Honda 296

7.3 Global automobile industry and levels of international 298
integration
7.4 Domestic and global structures of Honda 299
7.5 Degrees of product adaptation to overseas markets 300
7.6 Honda - Impact of joint ventures 304
7.7 Honda and BL comparison 314
7.8 Honda-Rover alliance - 317
BL/Rover share of vehicle programmes
7.9 Rover plant utilisation - 1987 322
7.10 Rover Group’s share of UK market 327
7.11 Rover Group financial losses 327
7.12 Longbridge production output 1999-2004 330
7.13 Elements of global industry - strengths and weaknesses 333
7.14 MG Rover-market characteristics and strategic responses 334
7.15 Differentiation of global resources 335
7.16 MG Rover R&D expenditure, 2003-5 340
7.17 Global R&D expenditure 2003/4 341
7.18 MG Rover R&D productivity 341
7.19 MG Rover - search for a partner 344
7.20 MG Rover - Impact of joint ventures 347
 1: Introduction

The global automotive industry is a valid area of research since it unquestionably continues
to have a substantial influence on the world’s economy. Despite this, its future trajectory is
not well understood. This researcher has always been an enthusiastic spectator of the
industry and much of this research therefore springs out of a personal quest to understand
the most fundamental forces that structure the industry. It was a particular advantage that
the researcher already had a substantial, if amateur, knowledge of the industry and a close
interest in its fortunes. The project was begun as an attempt to demonstrate that diversity
and independence had a place in the globalisation of the industry. In the initial phase of the
research the focus was directed towards the survival of the UK’s one remaining and
independent mass market car manufacturer, MG Rover. The hope was to reveal some
strategy by which the company could sustain itself in defiance of prevailing opinion. As
both the research and the company’s fortunes developed the focus was widened to include
any automobile firm that was considered to be of a type that would be vulnerable to the
vicissitudes of the industry.

1.1 Growth of a global industry

The world’s automotive industry has long played an important economic role. It is now one
of the largest industries in the world but even in 1946 its size prompted Peter Drucker to
declare:

“.. .the automobile industry stands for industry all over the globe. It is to the twentieth
century what the Lancashire cotton mills were to the early nineteenth century: the industry
of industries” (Drucker, 1972, p. 176).

Over thirty years later, in 1981, the managing director of Volvo, Pehr G Gyllenhammar,
stated:

“I personally believe that the automobile industry marks the limit of the sustainability of
industrial society. If you, as a country or a nation, state as a fact that you are not
competitive within this industry, then you have also abdicated from industrial society”
(Malmberg, 1991, p.212).

Data from the International Organisation of Motor Vehicles Manufacturers (QICA) shows
that the impact of the industry has not softened since then. In 2005, total vehicle
production reached 66m units (46m of which were cars), an increase of 3.1% over 2004
when employment had reached 8.4m in vehicle production and the related component
industries. OICA estimates that as much as five times as many more people are employed
indirectly, resulting in total employment of around 50m. This makes a substantial
contribution to the global economies, equivalent to turnover of €1.9 trillion, making
possible investments in research and development (R&D) and production of around €85bn.
Governments too, benefit, the 26 OICA member countries collecting €430bn in revenues.

The greater part of the automotive industry is made up of automobile production,

amounting to around 50m units in 2006 (ACEA, 2007). Production output is not
distributed evenly around the world. Western Europe taking the larger share at 33% of the
total. Within the European Union (EU) there has been some shift in production towards the
newest member countries so that Romania, Slovenia and the Czech Republic have gained
production while only Germany of the established European countries has also continued
to grow (OICA, 2006a). In sum, total production for the European region has remained
fairly steady.

Elsewhere in the world shifts in general automotive production have been detected. In
2006 automotive vehicle production in North America fell by 2.7% while Afiica grew by
16.3%, although nearly half of these vehicles were trucks. African output was dominated
by South Africa which manufactured nearly 340,000 automobiles, out of a total of 458,000
for the continent as a whole. The Asia-Pacific region showed growth of 9.3% in total
automotive output, China being conspicuous with general automotive industry growth of
26.3% and automobile output of 4.3m units. Despite that rate of increase Japan remains the
biggest producer in the region with 9.8m automobiles produced, a rise of 8.2% on the year
before (JAMA, 2007). Table 1.1 below shows the countries producing over a million units
a year.
 Table 1.1 Automobile producing countries, 2006

Country Automobile Production

Output (millions)
Japan 9.8
Germany 5.4
United States 4.4
China 4.3
South Korea 3.5
France 2.7
Brazil 2.1
Spain 2.1
India 1.4
UK 1.4
Canada 1.4
Mexico 1.3
Russia 1.2

Source: Automotive News, 2007a

Table 1.1 shows that although emerging economies such as China occupy strong positions
in the global automobile industry, other countries that have long historical links to the
industry continue to be strong players. It is necessary, therefore, to distinguish between the
general automotive industry and the specific characteristics of the automobile industry.
Historically, all vehicles could be made using similar methods, an unstressed body being
attached to a load bearing platform, or chassis. Automobile manufacturing took on a
distinct form of production technology when the all-steel load bearing unitary body was
developed in the early part of the 20^^ Century. This manufacturing process is highly
restrictive, both in the form of the final product and in the method of production, so it is
best suited to automobiles which have a well-defined purpose i.e. the comfortable and
affordable transport of small groups of passengers at their own convenience. Nieuwenhuis
and Wells (1997; 2007) have termed the body manufacturing process the Budd Paradigm
after the originator and it is the economies of scale here that set those economies of scale
for the automobile industry as a whole.
These large economies of scale create high entry barriers in the industry which result in the
incumbents retaining the advantage and high exit barriers which keep the incumbents in
place (Altshuler et a l, 1986). For example, as the Chinese automotive industry has begun
to rise it has required substantial assistance from the govemment in order to elevate
indigenous manufacturers to a level where they could compete against the established,
global competition. The govemment has done this by obliging foreign manufacturers to
enter into joint ventures (JVs) with local firms as a mechanism for transferring technical
capability and equalising output between local and foreign firms. Chief amongst these JVs
is that between VW and SAIC, which sold 694,406 units in 2006 (Automotive News,
2007b). However, domestic Chinese manufacturers that are not in JVs with overseas
manufacturers still had difficulty entering the local automotive industry and accounted for
just 27% of the Chinese market in 2006, this being spread over 20 different manufacturers
(Business Week, 2007). Even Chery, the largest independent Chinese manufacturer with
7.2% of the market, has joint ventures with Fiat and Chrysler along with substantial
engineering consultancy input from Lotus of the UK and AVL of Austria (Chery, 2007;
Automotive News, 2007b)

There is no sign that entry barriers to the industry are coming down. New technological
challenges for the industry, particularly with regard to safety and exhaust emissions, have
meant that product development costs have risen. Bailey (2007) gives a range between
£400m and £lbn for bringing a complete new model programme to the market:

“As a result, large scale production over different brands using a platform-sharing

approach is vital to generate the cash for future model development.” (Bailey, 2007, p.139)

The established manufacturers therefore enjoy advantages in size that enable them to
spread the costs of product development over a greater number of units in addition to
accessing economies of scale in production. This has led to a trend towards higher outputs
and consolidation as major manufacturers merge to form larger groups. Maxton and
Wormald (1985) predicted that only five global automakers would eventually survive, two
in the US and three in Japan. Ford is a case in point, acquiring
Jaguar (1989), Aston Martin (1994), Volvo (1999) and Land Rover (2000) to form the
Premier Automotive Group (PAG) division. There have been similar mergers throughout
the industry, such as Daimler and Chrysler (1998), Renault and Nissan (1999) and BMW
and Rover Group (1994).

However, this appears to have been the peak of the merger activity and since 2000, of
those listed, only the Renault-Nissan relationship has remained intact.
PricewaterhouseCoopers (2005) found that by 2004 the value of mergers between
automotive manufacturers had fallen to $2.3bn, which was 9% by value of the total merger
activity in the global automotive industry and down from 29% in 2002. A later annual
survey by KPMG (2007) found that while larger firms would continue to hold the
advantage consolidation would no longer be achieved through full bilateral mergers but by
strategic alliances. This suggested that firms would be able to access economies of scale
but retain their own organisational structures. In 2005 Automotive News (2005J) counted
21 independent automobile manufacturers that were also operating in partnerships and in
2006 the publication listed 42 automobile manufacturers with output ranging from 14,000
a year (Hindustan Motors) to 9m (Toyota) (Automotive News, 2007). This is not to negate
the pressure for companies to consolidate into groups but it does suggest that the theories
concerning consolidation need further articulation. The Economist (2005) particularly
noted that Toyota had come to prominence achieving economies of scale while largely
retaining the same integrated organisational structure.

1.2 Defining the automobile industry paradigm

Received wisdom has contended that due to economies of scale in the industry there is an
irresistible trend towards consolidation such that smaller manufacturers would either have
to join together or else leave the industry. Yet this convergence on a few dominant
manufacturing groups has not been found to be inevitable and there have been instances in
other industries where firms have exploited economies of scale through a network of
alliances that maintained their independence, Scania offering one such instance in the
commercial vehicle industry. Furthermore, some recently formed large groups have
subsequently demerged, DaimlerChrysler being notable here. If it could have been shown
that MG Rover had the potential to be an exception
to the behemoths of industry then it would have given hope for diversity and
independence, not only in the automobile industry but also in other industries. Although
ultimately this hope was proven to be forlorn in the case of MG Rover as an independent
automobile manufacturer, the research does find that a novel form of international alliance
could, perhaps, have sustained MG Rover as a developer of new models on behalf of a
dedicated production partner that was situated in a distinct region of the world.

This thesis sets out to clarify the structure of the industry by demonstrating that mass
market automobile production is technically distinct from other manufacturing processes
due to the unique characteristics of the production technology. The thesis describes how
the consequent economies of scale, which imply matching scale in related production
processes, are well documented in the literature and thus indicate the trend for
convergence. However, the existence of economies of scale does not in itself compel an
organisational structure, this thesis considering scale and corporate structure as
conceptually separate. Convergence on a single form of organisational structure is not
explicitly articulated in the existing literature and this thesis analyses the make-or-buy
decision that motivates vertical integration in terms of transaction cost analysis, as
espoused by O E Williamson.

The purpose of this thesis is to then propose a model for an automobile firm in terms of its
output and its organisational structure. Figure 1.1 shows how the thesis will take these two
separate and distinct theoretical approaches to demonstrate how the two types of firm, the
prototypical and the industry leader, operate within the paradigm for the automobile
industry.
 Figure 1.1 Fundamental arguments to the automobile paradigm

Production technology Transaction costs

Economies of Vertical
Scale Integration

Prototypical
Firms

Industry
Leading
Firms

From this study it will be possible to derive a systematic understanding of the shared
theoretical orientation of automobile companies before they are adapted to fit their own
specific contexts. The thesis will suggest a generic conceptual framework for automobile
companies by constructing an automobile industry paradigm. The paradigm includes the
main economic factors influencing the industry and so provides the defining features of an
automobile company. The paradigm therefore describes an automobile company that
comprises all the requisite functions and is exploiting the available economies of scale in
each of them. The research will then look at how automobile companies compare with the
theoretical precepts of the automobile industry paradigm and the extent to which they meet
the parameters of the paradigm. Of particular interest will be the options for companies
that are wholly uncompetitive. The research will consider how they might attempt to
approximate to the paradigm using alternative organisational structures. The thesis thereby
encompasses three research issues:
 1. optimal automobile company size with reference to economies of scale.
2. Optimal automobile company organisational structure in terms of vertical
integration.
3. The degree to which uncompetitive firms can approximate to the optimal size and
structure.

Since these issues determine the research methodology, the methodological options will be
discussed in Chapter 2 and the resultant research design in Chapter 3, in advance of the
theories concerning firm size and structure. The research is focussed on the concept of the
company so Chapter 2 shows how the case study approach can be most appropriately
applied. The case study approach is discussed as a research strategy involving a palate of
different research tools and designs. The nature of the data required is crucial in indicating
the appropriate gathering techniques and although much of the theory concerning the
automobile industry paradigm was derived from literary sources of data, primary data was
also gathered from industry sources in order to provide a context for the theory.

In the case of this research programme, interview based fieldwork is put forward as the
appropriate research tool and, since the research questions are concemed with overall
industry strategies, the interviews have been conducted with those who have most
influence or contact with those strategies. As these interviewees can be considered the elite
in their fields the term ‘elite interviews’ is used. Furthermore, in order to allow these elites
to expound their knowledge the semi-structured interview is considered the most
appropriate technique to use. In order to develop the emerging information the interviews
are linked, each one informing the next in a systematic manner reminiscent of the
‘laddering’ technique used in marketing research. Concerns over validity, generalisability
and reliability, are also discussed in this chapter.

The specifics of the research design used in the research programme are then described in
Chapter 3, restating the research questions and the reasons for the selection of MG Rover
as the main case study company. As the research commenced, MG Rover offered more
than accessibility; it was also an apposite opportunity to assess the research questions and
the emerging theory of the automobile paradigm. In order to put the company into its
industrial context the same interview protocols were
extended into studies of a wide variety of automobile companies operating within the
definition of the same paradigm. ,

The dependence on elite interviews meant that company access had to be carefully
managed, not only in the main case study but also in the contextual industry and related
organisations. The use of multiple sources of data from the case study and contextual
companies countered the effect of bias, triangulation being the term for validation of data
by corroboration and cross-referencing. The interviews took place throughout the research
programme, being continuously transcribed, coded and analysed. The paradigm was
formed through an iterative process of data analysis and theoretical discussion. Thus, the
early interviews aided in the initial formation of the paradigm while later interviews
refined it further and provided a contextual perspective to the theoretical concepts. Since
the data emanating from these interviews is related to the theories that comprised the
automobile paradigm theory the data is used to contextualise the theories in the specific
chapters where they are presented.

Having established the nature of the relevant data and how it would be used, the next two
chapters build up the conceptual framework. Chapter 4 looks at the industry from the
perspective of size and thus the presence of economies of scale. For the modem mass
market automobile industry the defining characteristic is the steel load-bearing body, for
which the production technology is the body-in-white (BIW) process. The other processes
inherent to automobile production are research and development (R&D), powertrain and
final assembly. The defining role played by the BIW process has been termed the Budd
Paradigm (Nieuwenhuis and Wells, 1997; 2007).

This thesis uses Stigler’s survivor analysis to comprehend the prevailing economies of
scale in the industry. The research categorised total global output into output ranges in
order to reveal any change in the share of the total for each output range. The results did
not conclusively define the economies of scale in the industry but they did indicate two
broad trends. Firstly, that the output range above 2.5m units was continuing to expand in it
share, and secondly, that although the 0.5m to Im range was shrinking there were
manufacturers within it that had survived for many years and were even growing. It
seemed that this apparent polarisation in the industry was
due to the continuing expansion in global output which meant that the industry had
yet to stabilise. ^

The research then estimates the minimum efficient scale (MES) for an automobile firm as
a whole from the minimum efficient plant size (MEPS) for each process. The data was
gathered from archive and company sources for those firms that the survivor analysis had
identified as showing longevity. For added contextualisation the data gathered by this
research was then compared with the data found in the literature and company sources.
Taking into account the trend for the larger companies to be expanding market share, the
results indicated the presence of two types of company. The first is a company with output
of around 600,000 units a year that exploits, as far as is realistic, the available economies
of scale. This might be termed the prototypical firm since it forms the basic unit of the
industry. The problem for such a company is that fluctuations in output can lead to
damaging variations in costs, so the second type of firm is one that has taken a strategic
course for expansion and become an industry leading firm. Although larger firms can
suffer diseconomies of scale, partly due to rising R&D costs, their higher output and wider
model ranges are found to diversify risk and reduce the effects of fluctuations in output
volume.

However, the MEPS found for each process does not ex ante necessitate their vertical
integration within a single company’s boundary, notwithstanding the conclusions
concerning MES. On the basis of scale alone there is no reason why the other processes,
being technically unrelated to BIW, should not be part of the external supply industry and
so it is not possible at this stage to define which elements should be included in the
vertically integrated automobile industry paradigm. For this reason it is necessary to
evaluate the extent of the vertical integration of a paradigmatic automobile company using
a different theoretical approach.

Where the preceding chapter examined the scale of the different operations, the principles
of vertical integration that might be common throughout the automobile industry dictate
which elements, or processes, should be included in the proposed automobile industry
paradigm. The organisation of the automobile company is discussed in Chapter 5 which
investigates the structuring forces within industry. It follows O E Williamson’s use of
transaction cost analysis (TCA) to examine the

10
dichotomous issue of whether a company should buy a product on the market or make
the product itself in an internal company function. ^

Internalisation of functions results in vertical integration and this thesis proposes a model
of comprehensive integration, terming this structure the full-function model in order to
avoid references to full vertical integration which has no predefined limits. The cost
advantage of internalisation is that it reduces the economic friction of market-based
relationships between the functions, which is experienced in the form of opportunistic
behaviour by participants as they exploit asset specificities. This enables the opportunists
to extract ex post advantages unforeseen in the original contract and Klein (2000) thereby
demonstrates that opportunism is primarily due to the human actors, though they may use
the physical assets as the mechanism by which to extract the benefits. This has
implications for governance structures, such as Chandler’s concept of the multi-divisional
M-Form of corporation, which are here viewed as coping mechanisms for managing the
integrated corporate structure.

The considerations of economies of scale from the preceding chapter are then applied to
the full-function model of the automobile industry to produce a size and structure for an
automobile company. The discipline of the resulting format is such that this thesis
considers it to represent a paradigm in automobile manufacturing, which in this thesis is
called the automobile industry paradigm. Using the MBPS and MBS data a model for a
prototypical firm is constructed which then forms the basic unit for the industry. Since
firms expand by the addition of complete plants the firms can be considered to grow by
units of MBPS and, in combination, MBS. However, for firms that have not reached the
prototypical state, the units of MBPS may not exist for all process and the firms will have
to find approximations of the size and structure advantages enjoyed by prototypical firms
within the paradigm. Chapter 6 therefore discusses the range of options for approximating
to the automobile paradigm. These include exploiting economies of scale through
extension of the product cycle and more intensive use of the production facilities.
Alternatively, from a structural point of view, an uncompetitive company could only
achieve lower structural costs if it is able to find a novel form of organisation that can
replicate the cost advantages of vertical integration. This means forming a relationship
with an external company, but not through the mechanism of the market, which would
carry with it the economic

11
frictions that the full-function structure of the paradigm is designed to avoid. The research
investigates ways in which the full function organisation can be divisible between partner
firms whilst retaining the advantages of vertical integration.

From Dussauge et al. (2004) two forms of alliance are found to be particularly promising,
the scale alliance that shares functions in order to exploit economies of scale, and the link
alliance that permits mutual learning by the two parties. Scale alliances are found to be
long lasting due to the persistent requirement for scale while link alliances have been
found to last only until the need for knowledge is satisfied or one partner asymmetrically
gains an ex post advantage over the other. However, this research uses TCA to suggest that
link alliances are more enduring if the partners have ownership of complete and
complementary functions. Any ex post learning benefits would then accrue solely to that
partner and so eliminate the risk of opportunism. The structure by which this can be
achieved is the vertical joint venture (VJV), each partner being assigned to discrete
functions which make up the full function structure.

Chapter 6 also looks at international opportunities to reveal any divisibility of the full
function structure. This focuses on the foreign direct investment (FDI) literature, starting
with the work of Hymer and elaborated on by Kindleberger. However, as Schott (2003)
argued with regard to international factor endowment differentiation, international
opportunities are not fixed. For a more dynamic aspect, Vernon’s product cycle and stage
theories of progressive internationalisation are introduced. Taking the different theories to
be mutually exclusive Dunning draws their aspects together within the wide ranging
eclectic theory, which covers international opportunities in ownership, location and
internalisation (OLI).

Strategic theories of FDI, as put forward principally by Cowling and Sugden, suggest that
firms may engage in FDI in order to enhance their competitive position relative to rival
firms. This research found that firms will expand output capacity in order to diversify risk,
much of this expansion taking place overseas. However, FDI is only incidental to this
strategy since the same result could be achieved within the home market if it could support
the output volume. Although strategic theories of FDI are informative, this research is
interested primarily in the underlying economic factors

12
that impinge on the industry and as such it is these that remain the focus of the study.
Instead, factor endowment differentiation can be applied to the VJV structure such that the
respective partners take responsibility for those functions in which they are already
endowed with the relevant factor advantage. The discussion culminates in the proposal for
structural and global division of the full-function model in the form of an international
vertical joint venture (IVJV).

The empirical support for the IVJV is then presented in Chapter 7 of the thesis. The
chapter is divided into two parts. The first deals with the contextual companies, all of them
firms competitive within the paradigm or else part of consolidated groups that are
competitive. The second part of the chapter then charts the decline of MG Rover’s
predecessors, BL and Rover Group, to provide an historical perspective on MG Rover’s
condition during the period 2000-2005. The reasons for the ultimate demise of MG Rover
in 2005 will be discussed along with the partial failure of the company to enact an IVJV.
As an adjunct to the collapse of MG Rover, Chapter 7 also discusses the manner in which
NAC and SAIC, two Chinese automotive manufacturers, purchased the respective physical
and human assets of MG Rover yet both pursued variations of the IVJV structure, thus
replicating the advantages of the automobile industry paradigm. More recently the two
companies have been exploring merger options, a possibility that is discussed in terms of
its impact on the emerging rVJV structure.

Chapter 8 does not set out to revisit the arguments and conclusions given in each of the
preceding chapters but to demonstrate how the research findings change the established
understanding of the industry. The current state of the industry is discussed, including the
opportunities that exist for companies within the paradigm as it stands. This has
implications for companies of all sizes, but most particularly small firms and the national
policies that are intended to encourage their sustainability. The paradigmatic view of
industry can also be used to gain a similar understanding of other industries. The
discussion is then widened to evaluate how this paradigm might change, or a new one
develop, if oil based fuel were to be phased out. Although this represents a future for the
industry that cannot be known at this stage, it is hoped that through this research a good
understanding of the present paradigm, and the opportunities it offers, will provide an
informative context for the next paradigm.

13
 2: Methodology

The research programme is shaped in accordance with two factors: the methodology to
prescribe the overall structure, and the research design to describe the construction of the
research in question. The first deals with the theories pertaining to rational investigation
and draws on authoritative sources, the second applies them in the practical setting of the
specified research. To clarify this distinction, methodology and research design have here
been separated into two chapters. The purpose of this, the methodology chapter, is to
establish the scientific credentials of the case study research strategy and so validate its use
in the research.

This chapter will show that the research questions are the ultimate guide to the selection of
the case study method and the relevant literature. The research questions of this thesis are
concerned with the basic structuring of the automobile industry and so determine the
direction of the research programme. For this reason the research questions open this
chapter to show how they guide the structure of the research with reference to its
fundamental methodology. From methodology the research design is described before
continuing with chapters that take up the purpose of the research questions by reviewing
the literature and bringing in empirical data.

The methodology will delineate the options that are available before the final design is
discussed in the following chapter. Having determined the structure of the research from
the research questions the subsequent chapters will discuss the literature relevant to the
subject of the study, i.e. the structuring of the global automobile industry. The
methodology and design chapters also determine the nature of the empirical data gathered
during the fieldwork phase and the literature in each of the subsequent chapters will be
contextualised by this primary data. This again emphasises the sequence of chapters in this
thesis from methodology and design to the theory of the automobile industry paradigm.

The implementation of the case study approach is complex and requires careful evaluation
before using it as a solid foundation for the research. The case study methodology is
deceptively simple and as a research strategy it can include a range of

14
research techniques, some derived from other disciplines. Having reiterated the research
questions, the case study approach must itself be distinguished from others that might be
available. It then remains to discuss the range of research techniques that can be used
within its framework. Chapter 3 will show how the final research design was selected.

2.1 The role of research questions

The research questions for this thesis were set out in the preceding, introduction chapter.
The initial questions seek to determine the limiting forces of the automobile industry in
terms of size and organisational structure, while subsequent questions seek to examine
whether there might be exceptions. In essence they are a call for clarity in developing a
view of the industry and a point of commencement for the literature search. The
subsequent literature search is therefore concerned with defining the structuring effects of
economies of scale and scope alongside transaction costs, then finding how these
conditions may be satisfied through a novel firm structure. Since the research questions
determine the kind of information that is required it is necessary to first define how it
should be obtained. The research questions therefore guide the methodology, which then
informs the literature search and fieldwork.

Stake (2000) acknowledges the importance of research questions but terms them as issues,
the select questions that plan and structure the study. With the case study approach taking
a holistic view of both case and context it is the research questions that define the elements
of interest. This suggests that the case study approach is not a methodology in itself, for it
is the research questions, or issues, that guide us in that direction. Indeed, there appears to
be agreement that this is so, with some authors portraying the case study instead as a
research strategy (Kitay and Callus, 1998; Yin, 1994).

2.1.1 Scope of investigative methods

There are, of course, a variety of investigative methods available but each has its own
particular application. Yin (1994) puts forward four research strategies: experimental,
survey, archival, historical and case study (see Table 2.1). The customary view has

15
been that each has its own application, whether exploratory, descriptive, explanatory or
causal. Instead, Yin (1994) believes that each strategy can be used in any of the
applications, but emphasises that it is the form of the research questions that determines
the choice of application, categorising them into “who”, “what”, “where”, “how” and
“why”. The author also introduces two more elements for consideration: control over the
subject of the study and time relevance as deciding factors in selecting the research
strategy. Control over the subject concerns the ability the researcher has to manipulate the
conditions of the study, while time relevance concerns the degree to which the focus is on
contemporary events. Table 2.1 shows how five separate research strategies, one of them
the case study, may satisfy these conditions in different ways.

Table 2.1 Applications of research strategies

Research Form of Control over Contemporary

Strategy research subject events

question
Experiment How, why Yes Yes
Survey Who, what, No Yes
where, how
many, how
much
Archival Who, what, No Yes/no
analysis where, how
many, how
much
History How, why No No
Case Study How, why No Yes

Source: Yin, 1994

The experimental research strategy puts the emphasis on control and it attempts to expose
underlying mechanisms. For Yin (1994) this is about answering “how” and “why”
questions by controlling the variables. Plutchik (1983, p. 16) states that the purpose of the
approach is:

“ .. .to increase our understanding of and our ability to control and predict events”.

16
In Yin’s definition the experimental research strategy is concerned with a high level of
control over current variables and Plutchik provides four reasons why control is
advantageous: to determine relations between the variables, to extend the conditions of the
study, to improve reliability and to test a theory. The experimental strategy is therefore an
approach that takes the variables out of their context and places them in an artificial
situation. The experimental researcher can then isolate the independent variable and
manipulate it so that its influence on the dependent variable is as conclusive as possible. It
is necessary to ensure that the variables are valid representations of the wider population
that the experiment is designed to study, so sampling bias is a particular concern. If the
experiment is valid then it is then possible to generalise the findings to the original
context.

The survey strategy also involves investigating samples of a defined population but
studies them in context rather than an isolated situation where the variables may be
controlled. It is concerned with “making statements” about a particular field of study
(Fife-Schaw, 1995, p. 100). Surveys are intended to measure a target phenomenon that
exists within that population or is experienced by it. For example, a survey might measure
business confidence amongst managers of the current economic climate and by using
statistical techniques a consensus can be arrived at. The strategy is highly structured and
standardised so is not conducive to recording singular judgements or assigning weight to
those judgements. In measuring business confidence it would be difficult for a survey to
account for the status or experience of the respondent, neither could it pursue novel
responses into greater depth. Surveys are a reactive approach; there is no attempt to
actively manipulate variables although they do benefit from the rigour of their robust
design.

According to Yin (1994) archival strategies pose similar questions to survey techniques
and are interested in quantifying the subject of the analysis. Here, though, the
investigation is conducted through the analysis of records and so uses data that has already
been recorded. When the data has been retrieved it can be manipulated using the same
analytic methods as if it had been gathered specifically for the research. It is appropriate
when the research is seeking to describe the prevalence of a phenomenon that occurred in
a widespread form. Yin (1994) suggests that research into epidemics is well suited to
archive research. The approach can be of benefit when investigating

17
organisations where access to individuals may not be consistent or may present short term
access problems (Bryman, 1989). In such situations it can be useful in providing
background material or else data that can be more readily found in archive form and
therefore relieves the researcher of the need to gain access to the subject of the research.

However, the researcher has little control over what the archive can yield and since any
information is filtered through these records it is not possible to use the standardised
techniques or sampling seen in survey techniques, although the archive material may
include survey data. Archive research is also dependent upon the availability and quality
of the material. However, when it is possible to access data of the same quality as new
primary data then it is an opportunity to reinterpret in the light of new information or
theories. For example, it might lend itself to research in politics where sufficient access
might be difficult to achieve but large amounts of data are within the public domain.

Historical research has a similar research design in that it draws on previously collected
data but it takes a more holistic, time bound approach. As Chase (1995, p.315) states, the
context is of primary importance:

“From the perspective of the researcher these are ‘natural’ conditions, that is, they have
not been determined by the researcher.”

Although surveys are also conducted in context their rigid structure is used to regulate the
degree to which the context interacts. Alternatively, archival research is only coincidental
with the time period, there should not be a time relation between the data and its context.
Historical analysis, however, is open to the interaction between subject and a time related
context. For example. Lord and Hohenfeld (1979) looked at how baseball player
performance was affected by job security anxiety during a period in the mid-1970s when
the negotiation procedures in the game were being changed. The research used archival
research in the form of game data but tied this to the context of the period. It was therefore
possible to test out the predictive powers of equity theory, i.e. whether the players would
perform less well because they felt less valued. This is different to archival research which
obtains the data in order to bring

18
depth to the sample and is not expressly concerned with taking an overview of the subject
within its context. However, like archive research, it is dependent on finding data and
historical analysis that is consistent with the design and purpose of the research
programme so it is at risk of any bias that exists in the original data.

Like the history strategy, the case study approach investigates the subject of the research
within its context. Indeed, the distinctive feature of the case study research strategy is that
in studying the subject in its context it is not only distinguishable from the other strategies
but also in direct opposition to the experiment research strategy. In another publication,
Yin summarised the distinction as follows:

“Compared to other methods, the strength of the case study method is its ability to
examine, in-depth, a ‘case’ within its ‘real-life’ context.” (Yin, 2006, p.111.)

Case studies are marked out by the focus on contemporary context and also the lack of
control over events. Historical analysis, though useful when defining contextual issues,
does not focus on the contemporary because it is interested in a complete time period, one
where the predictive potential of the data can be fulfilled. The experimental strategy is
restricted by the need to take control over events and it attempts to isolate the research
from the context, while the survey approach regulates the context by using a standardised
and targeted design. When the subject of the study exists as an integral part of its context it
is vital to master the nature of the context so that its interrelationship with the subject can
be understood. To not do this would be to allow the research to become the victim of
capricious circumstance.

The emphasis here is on having control over the research design through a deep
understanding of the nature of the context/subject relationship but not active control of
how this relationship can be manipulated and tested. The purpose, then, is to construct a
research design that can elucidate a bounded system, or in Stakes words (1998, p.88):

“ ...the researcher temporarily subordinates other curiosities so that the case may reveal its
story”.

19
Stake is instructing that it is not acceptable to simply investigate without purpose; certain
views of the data must be suppressed so that the relevant ones might emerge. No
prescriptive method is available but this is because the appropriate method is determined
by the nature of the case and, as Yin (1994) would add, the guidance of the research
questions. Denscombe (1998) contributes to this argument by stating that a case study is
appropriate when the researcher has little control over events so it is not possible to
regulate, a priori, the methods suitable for a specified case study. Equally, a case study
research strategy is not necessarily qualitative or quantitative, it depends on the research
questions and the units of analysis as will be discussed in the following sub-section.

2.1.2 Qualitative and quantitative research designs

The complexity that the case study approach preserves and investigates tends to lend itself
to the qualitative research design. Creswell (2003) describes qualitative research as being
emergent, flexible, interpretive, broad ranging and conducted in its “natural setting” or
context. Yin (1994), however, points out that these qualities can be applied to any of the
five research strategies (see Table 2.1). It is quite feasible for psychological experiments
to collect qualitative data on human perceptions, or surveys to collect quantitative data. It
is therefore the nature of the data requirement that determines whether the data should be
quantitative or qualitative, and a case study may even mix the two:

“ ...case studies can be based on any mix of quantitative and qualitative evidence.” (Yin,
1994, p. 14)

Lee (1999, p.38) particularly looked at the application of qualitative research to

organisations and found it to be useful for describing, interpreting and explaining:

“ ...qualitative research may be the best choice when the identification of new theoretical
propositions of managerial actions is deemed necessary.”

In contrast, quantitative data is more useful for generalising or calibrating when the data is
measurable on a scale. This is not always possible with economic concepts

20
which may use numbers to describe particular phenomena or characteristics, but the exact
relationship of the measurements may not be exactly known. For example, Manager A
may state that production is efficient at output IX while Manager B at another firm may
give output 2X; although 2X is double that of output IX, the efficiency difference may not
be precisely quantifiable on the same scale.

Other economic concepts such as risk and opportunism will ultimately affect the financial
performance of a company, but not in a way that is quantifiable in precise terms. Miles and
Huberman (1994) take Lee’s interpretive line and argue that qualitative analysis can infer
causality. Miles (1979) describes “goal-system state analysis” where the organisation’s
goals are defined quantitatively but the analysis on how they might be achieved is
qualitative. Since the connection between the goals and the means by which they are
achieved is complex it can uncover a rich seam of conceptual insights. Qualitative data in
this context:

“ ...lend themselves to the production of serendipitous findings and the adumbration of

unforeseen theoretical leaps.” (Miles, 1979, p.590)

This might be made explicit by management in an organisational setting when indicating

how different concepts fit together, a process which gains in power if the management is
particularly highly placed or influential. When context is important the research is aided
by this interpretation, either by the researcher or the respondent, because quantitative
analysis that attempted to take into account all economic factors would be impossibly
complex. There may still be data that is quantitative in nature, but it is in the treatment of
this data that the research is qualitative. The method by which qualitative and quantitative
approaches are differentiated for this research programme will be discussed in the next
chapter while the theoretical aspects of the case study approach will be further examined
in this chapter in the section below.

21
2.2 Case study design

The case study has often been viewed as a soft option, a useful pedagogical tool but
lacking the rigour necessary for research. It is often used to elucidate particular
arguments, or in journalistic lines of enquiry to fill columns in national newspapers. De
Vaus (2001, p.219) is quite blunt in summarising the historical status of the case study
design approach:
“ .. .for many years the case study has been the ugly duckling of research design”.

A well-chosen case study can be an excellent rhetorical device for expounding a particular
view or opinion. Its use in such a situation invites the application of a model answer to
match the construction of the scenario, a symmetry that is unlikely to be available in
research. Michael Edwardes, on completing his tour of duty as head of the British car
manufacturer BL, recorded his experiences in the book “Back From The Brink” in 1983.
However, this was in no way an academic assessment of the period, being more
autobiographical than objective. The author supplied the scenario and gave his view of it,
predictably without contradiction. This is not to say that the author’s recollections cannot
serve as a valuable resource, but that the views expressed require validation.

2.2.1 Critical perspectives

The populist use of the case study as a narrative gives rise to ex post theories, descriptions
of the processes but no opportunity to uncover the mechanisms of cause and effect.
Academic research needs to dismantle, in some way, the mechanisms that work the case
under investigation and put forward theories that increase the body of knowledge. The
popular criticism of case studies is that they do not do this. While they are fine as a
pedagogical tool, or for generating hypotheses for future study, they are seen as lacking the
rigour that cements the foundation of ‘hard facts’. Consequently, generalisations are often
consigned to the level of speculation by the author. For example, when Edwardes made his
initial report to the National Enterprise Board in 1977, he listed over thirteen factors that
he felt were contributing to British Leyland’s lack of competitivity (Edwardes, 1983).

22
However, it would be premature to dismiss the case study approach based on its popular
usage alone. When dealing with social phenomena, and this includes the commercial
world, it can be difficult, if not impossible, to separate the subject of the study from its
context. The case study permits, even compels, research in situ and theoreticians have
attempted to bring scientific rigour to it. Hakim (1987) describes five kinds of case study:
individual, community, social group, organisation and events. Each type recognises the
value in studjdng certain kinds of material within its context, a context with which it is
inextricably linked. However, this systematic identification of case study types suffers
from using ex post categorisations of studies completed, rather than providing a
prescriptive framework for use as a methodology.

Yet case study research does have academic standing and Hamel (1991) and Tellis (1997)
remind us of the first use of case studies by the Chicago School for researching the city in
the early years of the twentieth century. Indeed, de Vaus (2001) points out that case studies
have been fundamental to advances made in the social sciences while conceding that there
has been a lack of systematic evaluation of the precise role of the approach. The
subsequent decline in its use has been attributed to the rise in status of quantitative
methods, but concerns about their applications in social research brought about a renewal
of interest in case studies (Tellis 1997). It therefore remains an active research approach.

2.2.2 Defining the case study as a research strategy

When the subject of research is being studied in its context then the usual scientific
controls cannot be applied, yet the inherent complexity is the very substance of the
research. Kitay and Callus point out (1998, p. 104):

“The great strength of the case study design is that... it allows the researcher to place the
information in a wider context.”

Denscombe (1998, p.39) also states that the chief advantage of the case study is this
chance to:

“ ...deal with the subtleties and intricacies of complex social situations.”

23
Not only the context but also the internal elements of the subject matter, as noted by
Rossman and Rallis (1998, p. 175):

“Case studies are uniquely intended to capture the complexity of a particular event,
program, individual or place.”

With the case, the research programme seeks to understand that unit as a whole, informed
by the context within which it is found. The case might be an organisation, an individual,
a person, an event or a decision making process. In medical research the case almost
seems to choose itself: it has to be studied in order that a cure might be found, but the
research could involve various techniques from bacteriology to homeopathy. This is not to
suggest that any subject of study can be termed a case or else there would be an ad hoc
selection of the approach for research. The case has to be specific enough to describe a
system, or as Stake (1998) defines it, the case must be a bounded system. Denscombe
(2003, p.38) agrees and describes a case as having “a fairly self-contained entity” with
“fairly distinct boundaries”, thus implying that the specifying of the boundary is unlikely
to be exact. The case has an internal consistency and set of interrelated factors that
distinguish it from its context, yet it must interrelate with its context in a distinguishable
manner. If the case could be removed from its context then experimental methods could
be applied, but instead the context is a necessary part of the story.

The defining of the bounded system throws the ontological considerations into sharp
relief. Whereas in the natural sciences the ontology is implied when the premises are
specified, for the case study the scope of the research has to be confirmed. The impact of a
case study goes beyond its immediate context, for even the context has a wider
environment, but just as a natural science experiment isolates the independent and
dependent variables from the infinity of existence so the scope of the case study must be
reined in and only the salient factors considered. Kitay and Callus (1998) tell us that this
can be achieved by specifying the units of analysis, even if the nature of the particular
project means that they might only be defined in the course of the study. De Vaus (2001,
p.220) puts greater emphasis on the units of analysis and defines the case as:

24
 . .the unit of analysis about which we collect information”.

This conclusion seems to deny a distinct methodology when one considers that the wide
number of techniques allowed under the auspices of case study research, from qualitative
to quantitative, implies that the case study approach is little more than an umbrella term of
little inherent scientific credibility. In other words, it is as if the scientific researcher has
failed to isolate the object of the study and so has left it in its context as a “case” but
intends to employ scientific methodology henceforth. This is unfair since the case in its
context is precisely the object of study and as such guides the research strategy towards
certain techniques and precludes others. This falls within the remit of Kuhn’s polemic on
the paradigmatic structure of science and this will now be explored.

2.2.3 Kuhn and paradigm revolutions

Paradigms are frequently cited as providing a structuring framework for scientific

research. Kuhn (1970) is the main proponent of the paradigmatic view of scientific
thought. He emphasises the revolutionary nature of scientific progress which is rooted in
the sociological and methodological structure of scientific communities. Kuhn sees
“normal science” as being one where research is based on previous scientific
achievements that have come to be accepted by a scientific community as the foundation
for future practice. Kuhn then relates this to his own concept of paradigms where scientific
practice, comprising law, theory application and instrumentation, form the basis for
cohesive traditions of scientific research. For Kuhn, science itself is defined by paradigms
and within a paradigm normal science can be practised (Chalmers, 1983). Chalmers
clarifies this notion of normal science as:

“ ...a puzzle solving activity governed by the rules of a paradigm.” (Chalmers, 1983,
p.92).

According to Kuhn, normal, or paradigm based, research is said to possess three foci:
definition of a revealing class of facts, a comparison to the paradigm’s predictions and
continuing articulation of the paradigm theory. Kuhn refers to a mature scientific

25
community, a community made up of factions loyal to rival paradigms. In comparing
these paradigms the scientific historian can discover elements that:

“ ...the member of that community may have abstracted from their more global paradigms
and deployed as rules in their research” (Kuhn, 1970 p.43).

This dilutes the original concept of a paradigm as being a self-contained and self defining
scientific tradition, suggesting that it is more akin to a node of activity or a recognised
reference point for the wider physical science community. Indeed, Kuhn prefers to use the
term “disciplinary matrix”, the members of which agree upon formal expressions of
symbolic generalisations such as basic formulas, a foundation of basic conceptual models
and commonly held values. This allows the use of shared principles to be applied in
different analytical settings and paradigms.

Kuhn states that the period of normal science within the paradigm is an opportunity to
pursue detailed research without having to waste resources defending the fundamentals
that underpin the paradigm, these being already generally accepted. This would lead to a
paradigm taking up permanent residency within its field of research were it not for the
periodic revolutions that enable human understanding to lurch forward. It is the well-
defined, self-referential nature of paradigms that means a competing paradigm will be
exclusive and the transition to it must be revolutionary. Kuhn sees a dominating paradigm
providing a clear set of laws governing the activity of scientists, until they lose faith with
the laws and the paradigm in a revolutionary change. As Williamson (1998, p.25) puts it:

“ .. .it takes a theory to beat a theory.”

Lakatos points out that this means that revolution is extra-scientific:

“Scientific change is a kind of religious change.” (Lakatos, 1970, p.93.)

The message Kuhn has for practitioners within a paradigm is to engage in detailed
research but be aware that a developing mismatch between the paradigm and nature will
necessarily entail the creation of an entire new paradigm. Thus, paradigmatic

26
science can be said to be an organiser of theory, data collection and testing, permitting
progressively focused research without the need to reiterate accepted fundamental
concepts. This is a view that constrains future research by binding it to established
customs. The next sub-section of this thesis will explore the definition of paradigmatic
science in relation to case study research.

2.2.4 Case Studies as paradigm and methodology

Kuhn’s theory of the paradigmatic nature of science was discussed as a device for
describing the conceptual frameworks of scientific study. The same reasoning can be
applied in order to bring a coherent structure to the case study as a methodology. It is
apparent that there is a scientific community and literature growing around Robert Stake
and Robert Yin, it being seemingly impossible to refer to the case study literature without
quoting extensively from the two authors.

The case study is distinguished by the crucial role played by contextual factors, and
indeed this puts it into conflict with natural science research strategies. A well-
documented feature of paradigms is that they are mutually exclusive and therefore the
change from one to the other is revolutionary. Although the exact character of the case
study research strategy is not strictly defined, the continuing debates converging on a
precise definition would be the “mopping up” described by Kuhn (1970) and a feature of
normal science. There is also general agreement by adherents of the approach that the case
study permits a variety of methods to be conducted in its name:

“ .. .in practice they often require researchers to make use of more than one, and often
many, different research methods.” (Kitay and Callus 1998).

Indeed, the very complexity of the case study suggests that more than one method is de
rigeur, that a pallet of methods is not an option but a requirement. This would mean that
the case study methodology employs particular research techniques at a sub-category
level. It would therefore be quite acceptable for, say, survey techniques to be used as a
contributory part of the case study method if it was part of an organised portfolio of valid
research techniques. Referring to Chalmers (1983) and his “puzzle solving”, if a
methodology is the conceptual structure that permits the activity

27
then the case study achieves that through its ability to solve puzzles. Since puzzles come
in many forms the efforts to find solutions will not be uniform. For this reason, case
studies will not all be designed according to a single format, instead offering different
types of case study for the puzzle solver.

2.2.5 Types of case study

Stake (1998) describes three types of case study: intrinsic, instrumental and collective. An
intrinsic treats the subject of the study as a unique case and the purpose of the research is
to gain an insight into a singular phenomenon. Denscombe (1998) points out that a
strength of the case study is the ability to concentrate the limited resources of research in
one small area and we can see that this is particularly true of the intrinsic study. There are
still possibilities for generalising the intrinsic case. As a case study a small company has
intrinsic interest, but because it exists within the same commercial context as other firms
in the industry there is an opportunity to gain perspective on the industry at large.

This leads into the notion of the instrumental case and how a case can be used to gain
understanding of a promulgated theory. The case is chosen as being exemplary of a theory,
but difficulties over the unique characteristics of the case will persist. The nature of the
context each company finds itself in will limit the ability to hold it up as being typical of a
theory, the purity of the exemplar being contaminated by the relationship with its context.

Although intrinsic and instrumental cases are single case studies where resources can be
concentrated efficiently, at the same time the wider perspective is lost. De Vaus (2001)
offers the concept of the critical case which, like Stake’s instrumental case, is an apposite
test of a clearly defined theory. Again, though, the case cannot be separated from its
context so it is not possible to use it to falsify a theory, only to provide a supporting (or
contrary) argument. Furthermore, it might be impossible to distinguish the uniqueness of a
case unless deliberate reference is made to other cases within the same context. This
suggests a grouping together of cases for investigation, as described by Stake’s collective
case.

28
2.2.6 Stake and collective case studies

Stake (1998) puts forward the notion of collective studies, essentially multiple
instrumental cases. By studying the theory across cases it is hoped that the idiosyncrasies
of each firm can be controlled for and general principles derived. The research should be
wary of alighting on contrasting cases since that could simply give rise to multiple
intrinsic cases, the uniqueness of one shedding no light on the other. A collective case
study would therefore need to cover more than two cases and they should be conceptually
related to one another.

Care should also be taken when basing the research on just two cases for design reasons.
In selecting cases for collective study there can be problems of sampling bias since the
most representative cases may not be the most accessible. If the researcher is forced to
accept those cases that do afford access then a certain amount of sampling bias might have
to be confronted. The problem is also relevant within the case study where the ease of
access to particular departments and personnel may differ, or the role of the gatekeeper
may have a strong influence on the data that can be selected. The danger that the
representative might have the power to channel information is only mitigated by the risks
being apparent to the researcher, who then puts in place devices to counter the effect. For
example, the researcher could endeavour to build an independent network of contacts.

Robert Yin acknowledges that the complexity of the case study approach is not simply
between the case and its context, but also between the cases in a collective study. Rather
than referring to multiple cases as collective he prefers the term matrix to represent the
interrelation of the cases.

2.2.7 Yin and the matrix

Yin (1994) describes greater complexity in how a case study research strategy is put
together and applies a matrix structure to the categorisation of case studies. There are
single case and multiple case designs, and they may each include holistic or embedded
units of analysis. A unit of analysis is the object of the study and includes events,
individuals, occupations and so on. The single case is similar to the

29
instrumental and intrinsic cases above, but Yin divides it into the three sub-sections of
critical, extreme/unique or revelatory cases.

The extreme/unique case study is analogous to the intrinsic case and, as before, it is
challenging to apply this in research of commercial organisations that are inextricably
linked to their industrial contexts. When the opportunity arises for exploratory research
that has not been previously possible, then the single case study can be termed revelatory.
However, all single case designs suffer from their uniqueness and subsequent lack of
generalisability. The multiple case design addresses this point by following replication
logic so that the cases reveal either similar results for mutual support, or contrasting
results for predicted reasons. This should not be confused with sampling logic: the cases
are not samples selected from a wider population. A useful comparison here can be made
with the method of dialectics. Benson (1983, p.333) supports the idea of extending
dialectics beyond Marxist applications:

“Dialectical analysis must be concerned with the emergence within a social formation of
new, incompatible components.”

While a comparison of different companies does not approach the scale of argument
implied by the social tensions that drive the Marxist dialectical view it is, nonetheless,
instructive. If the data from the cases is set against each other in a dialectical style then the
data from the approaches can be used to reach an acceptable resolution.

Looking inside the case studies it can be found that, for either single or multiple case
studies, the choice of unit of analysis is a problem that characterises the research strategy.
These units of analysis might be holistic or embedded (Yin, 1994). The holistic design can
be used when the case and the unit of analysis occupy the same area of research, the
holistic case therefore comprising the bounded system. Yin (2006) suggests that one
example of an holistic case in education would be researching the implementation of a
school-wide programme. Embedded case studies use multiple units of analysis to research
the case, Yin returning to his example of a school wide programme but this time the data
would be collected from a number of different classes within the school, the classes
comprising the embedded research

30
design. Neither holistic nor embedded designs are exclusive to single or multiple-case
designs, Figure 2.1 illustrating the resultant matrix of four types of case study.

Figure 2.1 Basic types of case study

Single-case Multiple-case

Holistic Type 1 Type 3

Embedded Type 2 Type 4

Source: Yin, 1994

The intrinsic, instrumental and collective approaches should be seen as having a dynamic
relationship with each other. An intrinsic case study can give rise to theories that can be
tested in instrumental or collective studies. De Vaus (2001) sees this as vital in generating
causal explanations. De Vaus also provides another perspective on Yin’s (1989)
suggestions, distinguishing between cases as single entities and cases that contain multiple
aspects, the previously discussed holistic and embedded designs. A company could be
treated as a single entity at the holistic level, looking at the actions of the company as a
whole piece. This would include such concerns as level of sales, productivity, financial
reports and so on. The embedded design takes the components of the case as the units of
analysis. This could involve investigating how the different divisions of the company
work together to formulate a company policy. This approach would need to recognise that
a synergy exists between the divisions and so all the divisions require consideration.

31
2.3 Data collection

In order to maintain the relationship between the case and its context it is necessary to call
on non-invasive forms of data gathering. Case studies are distinguished by gathering data
from contemporary events using observation and interviews (Yin, 1994). The interview
situation can, to some extent, be manipulated by the researcher but not to the same degree
as is possible with an experimental research strategy. The control element of case study
research concerns the choice of questions by the researcher, but to manipulate the
interview would be to inhibit the discovery of new data. For example, Eleanor Miller
(1986) discovered in her study of female street hustlers that while the context remained
constant each respondent had their own story to tell. Submission of a questionnaire would
have handed control to the researcher but at the same time created a rigid process too
inflexible to allow personal expression. The respondents can be seen as experts in their
field and it would be presumptuous for the researcher to narrow the scope of their answers
before they had even begun. With research on companies, where data is gathered from
managers experienced in the industry, the respondents are undeniably the experts in their
areas and need to be allowed freedom of expression. The case study approach allows this
when used in conjunction with semi-structured elite interviews.

Using semi-structured elite interviews the researcher can exercise greater control over the
progression of one interview to the next. The methodology therefore sought out a
systematic approach. The linking of interview content can follow the logic of the
interview questions, pursuing lines of inquiry to a point of redundancy where no new
significant information is discovered. In an attempt to apply a conceptual framework the
‘laddering’ approach from marketing is investigated. In its original format it attempts to
help an interviewee to articulate an answer by laying down a trail of questions. In this
chapter, the thesis explores the approach as a way to establish a linking of questions across
interviews in order to bring about a continuous line of inquiry across multiple
interviewees. This is analogous to the original marketing purpose, though not a direct
application of the method.

Imposing a system on the order of the interviews is often impeded by access problems.
‘Snowballing’ is discussed here as one method for maintaining a flow of

32
interview contacts since it taps into a network of contacts by asking an interviewee to
make an introductions. It is well suited to qualitative research where statistical sampling is
not relevant (Bryman, 2001) but its primary utility is in aiding the research to identify a
series of contacts that were not previously known. This is not necessarily applicable to
organisations which are led by identifiable members of the management team and it can
often be a simple matter of discovering who they are from company literature or the mass
media. This is particularly the case when the interviewees being sought are the most
influential within the management teams. The following sub-sections will look at the
status of interviewees, how the content of interviews can be developed across different
settings by an adaptation of laddering and the manner in which interviewees can be
identified.

2.3.1 Elite interviews

When a research programme seeks to uncover the foundations of the units of analysis it
necessarily focuses on the key decision makers. As defined by Marshall and Rossman
(1999, p. 108) the interview embodies the fundamental aspects of qualitative research:

“The participant’s perspective on the phenomenon of interest should unfold as the

participant views it, not as the researcher views it.”

In the matter of industry, the interview process lends itself to gathering a wide range of
qualitative views from a variety of sources. The purpose here is to reveal the sweep of
factors that impinge on a company operating in the contextual industry, not the minutiae
of operational specifics. As such, the interest is in the key decision makers and their view
of the issues facing them, not the practitioners of established processes. This requires a
method that allows the participants to express themselves, as opposed to responding to
specific questions, and this is best achieved in the interview format.

By focusing on key decision makers it is possible to gain an insight into a wider

perspective, one that deals with the contextual considerations of the company and the
industry. This yields data at a level appropriate to the issues being investigated. As Fitz
and Halpin (1995) point out, a lot of research has been conducted to give a

33
‘voice’ to the disadvantaged, while the powerful in society have the personal resources to
deflect individual attention. Those with power can be called the elite and elite interviews
are intended to give them their voice. Fitz and Halpin (1995, p.67) conducted interviews
focussing on education policy and there were four main objectives:

1. Interview policy makers to obtain additional insight into published policy

documents.
2. Clarify existing published accounts and open a new narrative.
3. Identify networks of individuals.
4. Clarify the ‘assumptive world’ of policy makers.

The first two items are concemed with obtaining a different perspective in the interests of
validation, while the third reveals the decision making framework. The fourth item can
reveal hitherto unknown data because it allows the interviewees to state their priorities
and the underlying issues. It should be noted that the interviewee’s status is not of interest
as such, it is the content of the interview that is being sought. Dexter (1970) prefers the
term “non-standardised” to elite to differentiate it from orthodox interview situations in
the following ways:

1. The interviewee defines the situation in question.

2. The interviewee structures their own account.
3. The interviewee decides what items have priority and relevance.

In a standardised interview the researcher has prepared the problem and is seeking
answers to it, so the unique or illuminating view of the expert can produce anomalies that
are difficult to deal with. Stanley Payne, for example, cautions that the presence of a
specialist can skew the results in the direction of their view, either by direct influence on
their fellow respondents or by statistical bias (Payne, 1951). Conversely, with a non-
standardised interview the interviewee plays the role of teacher and it is their insight, no
matter how unique, that is the point of interest.

Understanding the use of the term “non-standardised” also highlights the distinction that
should be drawn between different types of elite: political and business sources.

34
Moyser and Wagstaffe (1987) devote the majority of the introduction to their book
discussing the normative implications of researching elites in society since political
leaders derive their mandate from the general population. Political elites represent the
interests of those they lead while business elites represent the interests of the company.
This is particularly pertinent to the issues of industry where the interests of production
workers may be entirely at odds with concepts of firm structure. For such a study, the
executive management, those with policy making powers or having intimate knowledge of
it, are taken to be a self-contained elite with a formally defined scope of power. It is also
assumed that the management adhere to a unified vision of the company and so can
represent it, though not all to the same levels of authority.

Dexter contrasts the elite interviewee with the key informant whose status as a source of
data is different. An informant is one who acts more as a proxy for the researcher, having
access to information that can then be passed back as data and playing an important role
as an addition to other sources. They can be taken to be individuals who hold information
but lack the power to act on it, in other words, they do not have strategic control over the
industry or companies but do have intimate knowledge of the issues. Such informants can
include analysts, government agents, industry representatives and public relations
personnel. The purpose of including such informants is to construct a contextual picture
and triangulate the elite interviews by cross-referencing.

The elite interviewee needs to be given space within the interview in which to express
these insights. When looking at how a company might need to shift its orientation to
confront issues, the researcher should avoid putting a preconceived structure on the
interview with planned questions. An elite interview is therefore able to capture
anomalies, indeed this is often the raison d ’etre of the approach.

A problem with the elite interview comes from allowing the interviewee too much control
over the interview process. Marshall and Rossman (1999) go so far as to warn that the
interview may even be manipulated to the degree that it is turned back on the interviewer.
Elite interviewees are often articulate individuals who welcome the chance for a verbal
duel; Fitz and Halpin were particularly intimidated by such professional communicators
during their research in education:

35
“What researchers confront therefore, is not simply the power of office, but power

exercised in and through language and communication.” (Fitz and Halpin, 1995, p.68.)

A recommended way to counter this is for the interviewer to be capable of shrewd

questioning based on a deep knowledge of the subject. This can be achieved by dividing
the data collection into two phases. The first involves basing the research questions on a
thorough understanding of the issues published in the literature and devising proposed
interview protocols which are then conducted on informants of peripheral interest. This
could include elite-style interviews but with interviewees outside the companies under
study. This prepares the ground for the second phase, the elite interviews themselves, by
which time the interviewer has the skills and knowledge to tackle an elite interview at the
centre of the case study.

The interview itself can be controlled by use of two techniques. The first concerns the
degree of structure to the interview. There are two extremes to this, either structured or
unstructured. Moyer and Wagstaffe (1987, p. 18) make the claim that elite interviewing
does not prescribe one technique but:

“ .. .a whole family comprising varied alternatives.”

This may be reasonable when discussing social science research which can take many
forms, but in the case of company policy the range of research tools is significantly
narrowed. This is because the interest is in the elite group as conveyors of information
they have in their possession, not as a social phenomenon.

By implication, the elite interview obviates the use of a structured approach because the
interviewee is the expert and to restrict them to answering prepared questions would
negate the benefits of interviewing them in the first place. At the other extreme, the
unstructured interview presents precisely the control problems that are pathological to the
elite approach. The compromise is to use a semi-structured approach comprising open
ended questions that facilitate and guide the interview process. The essence of this
approach is to be flexible in the use of the questions, the respondent being free to answer
to the depth they feel is suitable, avoiding questions

36
to which they are averse and even changing the order in which they are answered. Fitz
and Halpin (1995) found that a printed sheet of questions could act as a prompt sheet
during the course of the interview, and could also be sent to the interviewee in advance to
allow them to visualise the overall shape of the interview, allowing them to prepare and
prioritise their answers. The results of one interview could then inform the formulation of
the next interview protocol with the topics appropriately amended.

2.3.2 Linking of interviews - by content

The process of one interview influencing the protocols of the next is one that can and
should be formalised. However, as Dexter (1970) notes, unlike standardised interviews it
is not possible to impose a mechanical method to structuring the schedule of elite
interviews. This is particularly so when there is an exploratory element to the research and
the data from one interview can be used to inform the next. The danger is that without a
conceptual framework to this accumulation of data the path will have no coherent
direction, each step being merely dictated by the one immediately preceding. At the same
time, a prescribed approach would be too rigid for what is, in part, a self-generating
research strategy. Since it is the case study that guides the search for data it will
necessarily be that each research programme must adapt an existing theory to one suited
to linking the content of the interviews as they develop in that specific case. One such
candidate is the laddering technique, a system previously used for linking the data
generated within a single interview rather than between successive interviews.

Reynolds and Gutman (1988) are major proponents of the laddering technique, a valuable
tool in marketing research concerning personal values at the micro or psychological level.
It is derived from the Means-End Chain model of consumer behaviour which assumes that
choices are guided by values in the form of desired end states and that the diversity of
products available for a choice decision is reduced by grouping them into sets (Gutman,
1982).

To make this fit a research interview concerning corporate strategies it is necessary to

delete the plethora of products and consumer values, substituting firm strategies and
management priorities. This stretches the Means-End Model beyond its design

37
parameters because its purpose is to uncover a covert and personal psychological process,
whereas corporate strategic decisions are based on overt information searches. Of course,
views are still being sought, but laddering was devised to help bring the respondent to an
answer they would normally find difficult in articulating. An elite interview has the
opposite problem in needing a method for managing the expertly articulated flow of
responses.

Laddering is designed to provide a hierarchical value map, a type of cognitive map, by

exposing the underlying personal motivations. Corporate strategy is based on objective
information and it is the linkages that comprise a decision map that is of interest. Strictly,
laddering is designed to elicit an answer from the respondent, while in elite interviews the
interviewee is able to provide and structure the answer by their own volition. Yet
laddering is still an attractive technique for managing an elite interview because it
provides a template for progressively probing questions for a valuable result.

The derivation of the method is useful because laddering gives a systematic approach to
building up the interview and providing the linkages that reveal the processes
underpinning the corporate decisions. The concept can be extended further by using one
interview to not only inform but also to build towards the next, in this sense laddering
across interviews. In this way, any initial interviews are more than pilot studies for
researcher familiarisation, they are an integral part of the entire research programme and
are the first rungs on the laddering process. This enables discovery and exploration of
themes across a variety of interviews to aid triangulation and the generalising of
conclusions.

Reynolds and Gutman (1988) note two problems with the laddering approach, this being
where the issues are too sensitive for the respondent to deal with or where the respondent
does not have an answer to give. Sensitive issues in this research programme pertain to
commercial confidentialities and so are more of an access problem rather than an obstacle
to the course of the interview itself. Neither should there be difficulty in delivering an
answer, the elite interviewee being articulate in either giving the answer or explaining
why it is not in their remit to be able to give one. Despite this, there are limits to the
loquacity of even the most articulate

38
respondent and so some of the various counter-methods suggested by the authors can still
be employed to good effect. Six of these are listed, given here with their applications to
elite interviews:

1. Evoking the situational context - suggesting an issue for discussion.

2. Postulating the absence - used to ‘unblock’ an answer, gaining perspective
on an issue by suggesting the converse of it.
3. Negative laddering - reversal of orthodox laddering, e.g. “Why wouldn’t
you want to be independent?”
4. Age regression - an historical view, gaining perspective on current issues
in comparison with the past.
5. Third person probe - comparison with the activities of other companies.

6. Redirecting - communication check by repeating responses for

confirmation.

Again, since this research programme proposes using an adapted form of laddering, the
interview techniques are being put to a different purpose. In consumer research they are
needed for eliciting responses but here they are useful in gaining novel perspectives. It is
these new perspectives that can then be carried across to the next view, creating links as
part of a cohesive research design (see Chapter 3).

2.3.3 Linking of interviews - by respondent

At the planning stage of the research there are two questions concerning the collection of
data: who to interview and what should be the sequence of the interviews. In practice,
difficulties of access will work against such a neat plan and this creates two further
problems: how to identify possible subjects and how to gain access.

‘Snowballing’ is one procedure for identifying potential subjects. According to Faugier

and Sargeant (1997) the generic term is link-tracing but the commonly used technique is
known as snowball sampling. This is where one subject proposes the next in order for the
research to gain access to subjects that would not, ordinarily, be available. The authors
acknowledge that this introduces sampling bias due to the

39
deliberate selection of subjects but it also comes with an in-built validity check as the
subjects can reinforce each other. Indeed, Etter and Pemeger (2000) compared
snowballing with random sampling for research on smoking and found little evidence for
bias in snowballing.

In any event, snowballing is not a necessary feature of research based on elite interviews.
Elites are by their nature highly visible, often quoted in the media and their influence
thereby easily inferred. Galaskiewicz (1987) found that a magazine and newspaper search
was an effective way of becoming sensitised to the target elite. This provides the coherent
path to the interview schedule, one that would, if anything, be thrown off course if
snowballing were the primary linking mechanism.

Identification is not, though, a guarantee of accessibility. Brannen (1987) notes that

gaining access is a political process and so took the top-down approach. This meant
gaining access to the chairman first and then using that success as leverage to
progressively access those further down the hierarchy. This is a high risk approach which
can founder on an initial failure, a lower risk approach being to search for any initial
access and then follow that line of enquiry up as far as the data trail will go. Once
exhausted, another line of enquiry can be commenced, though a top-down approach can
be used concurrently as a supporting method.

Thomas (1995, p.4) found that:

“ ...business elites are quite good at insulating themselves from unwanted disturbance”.

The author goes on to warn that the first line of defence for an executive is comprised of
the company gatekeepers, posted to monitor the activities of outsiders. These are often
official spokespersons or representatives. Gatekeepers can be helpful in passing the
researcher to the most appropriate subjects, particularly when their own status is seen to
be enhanced by this, but they can also jealously guard their contacts if they think the
research may be a threat to their own position. When this happens they can present a
serious obstacle that may have to be circumvented. Dexter (1970) was able on one
occasion to find a higher authority to represent his interests to the gatekeeper

40
and so reverse the gatekeeper’s earlier obstinacy. An alternative is to simply contact the
target subject directly. Thomas (1995) does not support this and recommends finding an
intermediary with personal contacts in the company, or making direct contact with
sympathetic non-elite staff at the company who can then represent the researcher’s
request to the target elite. Whatever the approach being made, without a gatekeeper the
researcher is left with little option but that of simple ‘cold calling’.

When contact is granted with the target subject, preparation is essential. Thomas
(1995) recommends that the initial contact should be personalised in three ways:

1. Promote the subject’s view as being of intrinsic value, avoiding being

passed on to another.
2. Express the value of the subject’s view as a source of knowledge.
3. Define the nature of the problem in specific terms.

Dexter (1970) cautions against giving too much detail in the initial contact, since that
encourages the target subject to pass the enquiry on to a specialist, and emphasises that
the request should appear relevant to the subject’s area of expertise. This also entails
having some understanding of the subject’s work practice and how an interview can be
conveniently accommodated by them. Thomas (1995) adds that this is not simply good
manners as elite subjects tend to put a high value on their time.

Once the first interview has been successfully completed a recommendation can be
sought to the next subject, on opportunity to make use of snowball sampling although
here it is a way of gaining access to a target subject already identified. Galaskiwicz (1987)
formalises this into the reputational method where interviewees were asked to rate the
reputations or influence of other interview subjects, though this function is partially
satisfied in the corporate structure by the hierarchical status of the interviewee. The
building of a contact network also imposes a structure on the data gathering because the
easiest contacts will come first, progressively building towards the most influential of the
elite in the higher echelons of the hierarchy. Retired executives can provide particularly
fertile ground for this since they offer greater accessibility and they may recommend
contacts still employed by the company.

41
There are problems in relying on sympathetic subjects representing the researcher’s
interests. Much of the research that uses elite interviews has been in the field of political
science, and researchers wam against the pollution of one source by another. Dexter
(1970) gives the example of an unpopular union leader, many times passed over for
promotion, who “threatened” to adopt the author as a protege. This would have had a
disastrous effect on the gaining of future contacts and could have been as deleterious as an
obstructive gatekeeper. To a lesser extent, the interviewee may have a tangential role to
the main thrust of the company’s policies or that they may “talk up” their contributions.
Such digressions can be countered by firstly relying on recommendations and secondly by
cross-referencing interviews for triangulation. This allows the researcher to increase the
scope of interviews conducted and include those outside the subject companies without
compromising validity.

2.3.4 Data recording and coding

The major challenge of any interviewing technique is extracting the data in a form that is
ready for analysis. In order to link the analysis to the original data an appropriate
recording method should be used, one that captures the richness of the information
gathered without compromising the interview itself.

Simple note taking is presumed to be a non-invasive form of data recording, it being

possible to take notes during a break in the conversation, or even during it if the
interviewer has the necessary skills. This method is also particularly suited to interviews
that focus on the content, where it is information that is being gathered rather than the
process of the interview being studied. Dexter (1970), from experience, considered it
rarely worth using a tape recorder and considered them invasive, but relented in view of
the need to capture the interactions between interlocutors. The author’s point about the
high costs of transcribing the interviews remains apposite, but it should be noted that the
latest technology is more discrete and therefore less invasive.

Corbetta (2003, p.280) takes a more supportive view, perhaps reflecting the technological
improvement. The author warns against a written summary during the course of the
interview:

42
 . .as the result would be incomplete, dull or even incomprehensible”.

A written summary does not preserve the original content of the interview, and the author
even exhorts researchers to transcribe idiosyncrasies of speech, such as dialect or errors.
This is appropriate to social research questions where the form of the interview is of
interest, but for a content driven interview the advantage of a full transcription is that the
data can be interrogated for additional data. It has been shown how one interview can
inform the next, but with a complete transcription it is then possible to return to previous
interviews to uncover material that may not have seemed important at the time. Contrary
to this, a written summary only records what was judged important at that time and
unwittingly discards potentially valuable material.

This does not contradict the view of Dexter (1970) that transcriptions are costly in time,
putting forward a figure of nine hours transcription to one hour of tape, but if it is the
researcher who transcribes then it is more than a mechanical process of converting the
verbal to the written. Indeed, transcription presents another opportunity to re-engage with
the interviewee, albeit passively. This does not so much unveil subtleties in the form of
the replies, as might be useful in social research, but content information can be recovered
that might not have been pursued at the time of the interview. A written summary would
miss the same information but recovery would depend on a spontaneous recollection by
the researcher after an extended period. This hardly denotes reliable recording of data.

An additional time cost comes with the need to code the data for analysis, yet this too is a
chance to revisit the data. It also means that the judgment on what data should be omitted
can be left until late in the fieldwork phase, or all the data can be coded at the same time
to enhance the uniformity of the coding process. Having already completed several
iterations of the data, courtesy of the transcription process, codes can be used to reflect the
themes that have emerged throughout the entire data set.

43
2.4 Validity, reliability and generalisability

However impeccably a research strategy may have been devised, ultimately it must first
show validity by measuring what it set out to, then do it reliably to show the data was not
arrived at by chance, and finally prove its utility by having the potential to be generalised
beyond the subject of the case. It is by confronting the demands of validity, reliability and
generalisability that a research programme shows its value. The three factors are as
relevant to qualitative as much as to quantitative research, but the manner in which they
are dealt with is quite distinct. While quantitative data can be set against formal, statistical
standards, the qualitative study cannot always rely on external corroboration. Dey (1993,
p.221) even says:

“...the qualitative analyst may have to play all the roles in the analytic drama.”

Thus the onus is often on the researcher to organise the data so that it is mutually
evaluating. The key characteristic of a qualitative case study is the revealing of the
complexity of the case. In recognising this, triangulation can be used to regulate the data
by introducing an element of cross-referencing. The very process of exploring and
revealing the richness of the data serves to reinforce its value. Triangulation is not a
formal process, despite the term it does not prescribe necessarily the use of three sources
for every datum, but it does involve cross-checking of the data.

2.4.1 Reliability

Corbetta (2003) states that reliability is concerned with the stability of the results over
time, and yet any study involving human interaction will show the same variability as the
participants. Quite simply, it is not possible to retest people and get the same result. This
is particularly exacerbated in a qualitative study where the very process of articulating
information in one interview may lead the subject to revise their position in the next
meeting. Corbetta refers instead to equivalence, a correlation between results. This can be
effectively addressed by confining the interviews to the same topics, even if the specific
questions are an integral part of the each interview. In developing the coverage of the
topic, as would happen in an exploratory phase, a good

44
correlation would be expected between successive interviews, though not ultimately
between the first and last. This means that correlation of data is conducted tactically rather
than strategically. Naturally, this correlation could not be expressed quantitatively so it is
the purpose of the data analysis to pick it out and demonstrate it by juxtaposition of
sources.

Dey (1993) feels that without the possibility that the research will be replicated then
presentation of corroborating data is not credible support for reliability. Since very few
are replicated, it being impractical for a different research programme to put the same
interview questions to the same case study participants, support for reliability must be
found elsewhere. He illustrates this with the analogy of a watch that the wearer does not
want to subject to testing, so assurances of reliability are given by explaining the
mechanism. Similarly, by meticulous description of methodology, Dey claims that the
trust of the scientific community might be engaged. If this suggests that faith is needed for
the research design to be accepted then Miles and Huberman (1994, p.278) suggest that
the reader should actively question the design as it is revealed.

2.4.2 Internal validity

Internal validity is of particular importance to the case study design because as a bounded
system it must primarily be true to itself. For Miles and Huberman (1994) internal validity
is simply a question of truth value, whether the findings are faithfully linked to the data.
At the most basic level it might be asked whether the case selected actually is
representative of what the researcher claimed to be investigating. Denscombe (1998)
further cautions that because the researcher is in close proximity with the case this can
lead to “observer effect” where those being observed act differently to how they would
normally. These mutually exclusive views, a negligent selection of the case or a
misrepresentation of the data by the participants, both delink the empirical data from the
findings. It is quite possible that interviewees at the case studies might unwittingly
misrepresent their companies, by understating the nature of the challenges or overstating
the achievements. The multiple case study design is intended to neutralise this by
comparing the data from various companies, and

45
triangulation enhanced by including interviews taken in the contextual industry and
institutions.

Corbetta (2003) gives a more specific warning concerning content validity, that of
whether the full range of meanings for a concept is fully defined. Even when the general
research questions have been stated there may be disagreement over the meaning of the
terms involved. In a semi-structured interview this problem can be turned round by
presenting it to the respondents for discussion. In this format the interviewee has some
leeway in applying the emphasis that they perceive as suitable, so in combination with
other respondents a more comprehensive picture of the concept might be constructed.

2.4.3 External validity and generalisability

External validity is perceived as a critical weakness of the case study research strategy as
it relates to the generalisability of the study. Indeed, it is in this area that it has to justify its
wider utility. It is a peculiar problem of case studies because the context of the case is the
external, so it is important that the methodology takes account of the validity of the link.
De Vaus (2001) notes that collection and analysis of data must be guided by theory on how
the case fits with its context. The theory then defines the external validity of the case
study. The author presents two kinds of case study, explanatory and descriptive. The role
of theory in a descriptive case is of lesser importance, while explanatory case studies
present opportunities for theory testing or theory building.

Yin (1994, p.30) also puts deductive theory testing at the centre of case studies and calls
this analytic generalisation, in contrast to statistical generalisation. Statistical methods
allow a Level One Inference by appeal to formulaic confidence levels, which is not open
to case studies because they are not sampling units. Instead, the external validity of a case
study is analogous to Level Two Inference. This encourages the researcher to seek the
representative case that will illuminate the understanding of the theory. The theory that
guides the researcher has arisen from the generalisation of a previous study, but if the
theory is confirmed by the data it cannot strictly speaking be said to be proven, only
supported. In a research programme, the theory might arise

46
from an inductive review of the literature so that it can then form the basis of the
fieldwork.

Stake (1995) claims that generalisations are possible even from single cases, though they
are not as robust as those from multiple cases. In a later publication (1998), the author
warns that a case study is but a small step towards a grand generalisation and that the
researcher should not be tempted away from the essential features of the case to create
clear but unsupported theories. Whether supported or not, refining the theory with case
studies is an open ended affair and additional cases may be amassed and the theory
refined. This leads to theory building where subsequent cases can be chosen for their
capability to further illuminate the original propositions by recognising the differences
and commonalities of the various cases. If multiple cases are researched in parallel then a
degree of replication might be achieved, with even greater robustness if they support the
new theory but not a rival one (Yin 1994).

Corbetta (2003) looks at the taxonomy of external validity. The author states that
criterion-related validity links the indicator to the concept, for example the interview data
to other observations. In practice, as Corbetta states, since this validity is not measurable
it is the correlation between two indicators, not strictly that between the indicator and its
concept. Corbetta eases this task by sub-dividing criterion-related validity into predictive
and concurrent validity. Predictive validity is the most impressive test of validity, the
potential for the case study findings to illuminate the future, but commercial activities
rarely progress with identifiable steps. At worst, the exogenous shocks may force a new
strategy on the case study company, or it might even cease trading altogether.

Concurrent validity releases the research from being held hostage to the future by taking
comparable data simultaneously. This supports the research strategy of using multiple
case studies because the same topics are being addressed across the case studies, albeit
from different perspectives. By taking comparable data at the same time the state of one
case might suggest a future state of the other. This, then, imports an element of predictive
validity. In addition to this, the approach includes the notion of replication described by
Yin (1994).

47
In recognising that external validity has a particular role to play in the richness and
complexity of case study research, the part of the researcher should also be
acknowledged. The case can only be known through the researcher and this causes
problems for validity. Construct validity, therefore, addresses the accusation of personal
subjectivity to which case study research is vulnerable. The researcher needs to show that
the data reported is salient and not minor material that happened to appeal to the
researcher personally. This recalls a defining feature of case studies, the richness of the
data, which complicates the research process and yet also provides mutually supporting
data. It is by thoroughly investigating and presenting this that the researcher is able to deal
with the various problems of internal and external validity.

2.5 Ethical concerns

Authors such as Marshall and Rossman (1999) discuss the importance of ethical
considerations in social research when it impinges on sensitive or illegal activities. They
say the researcher faces a complex dilemma:

“Is the researcher’s primary responsibility to the research task, to those being observed, to
those assisting with the observation, or to society as a whole?” (p.93.)

Although this is rather a broad perspective, in company research there is a need to both
respect commercial sensitivities and maintain the confidence of sources in order to
facilitate the progress of the fieldwork. As case study research looks at the subject of the
research in situ it places the researcher in an unusual position:

“Qualitative researchers are guests in the private spaces of the world. Their manners
should be good and their code of ethics strict.” (Stake, 2003, p. 154.)

Punch (2005) categorises the literature on ethics into two types: there are the prescriptive
guidelines put out by institutional bodies, and then there are the historical accounts of
issues arising during research from which lessons may be leamt. There is no reason why a
prescriptive guide should not apply to a case study strategy, but it is dependent upon the
design of that study. If the methodology were using psychological

48
testing then it would be appropriate to refer to guidelines put out by an organisation such
as the British Psychological Society (BPS). The BPS publishes general guidelines for
research on humans, which set out broad areas of concern (British Psychological Society,
1992):

1. Deception - the distinction between withholding information as part of the

study and deliberately falsifying the purpose.
2. Debriefing - releasing the participant from mental anxiety.
3. Risk - protecting the participant from unreasonable hazard.

Laudable though it is to propose guidelines, even these generic principles require adapting
to research in the commercial world where the informant’s personality is not the focus.
Here the ethical standards do not have the same moral foundation, they pertain more to
the smooth running of the research. In exploratory fieldwork there is unlikely to be
deception on the part of the researcher since it is the participant who has the knowledge,
and any subsequent debriefing would be to allay the researcher’s confusion rather than the
mental anguish of the participant. Risk is an issue, but it is more open ended: the risk that
a confidence might be broken by the researcher, putting the participant’s career in danger.

Punch (2005) instead allows the researcher to appeal to the experience of previous studies
for support if formal guidelines are unsuitable. As Merriam (1988) points out, because
case studies are emergent designs it is difficult to anticipate where the research will lead
and thus what ethical dilemmas might arise. Problems might continue to the publication
stage where it is necessary to identify the case because it is the case’s peculiarities that are
of interest, yet at the same time acknowledge that confidentialities might have to be
observed or identities veiled. Ryen (2004), therefore, discards the prescriptive approach
and looks at the three main ethical issues that face a research programme. The first
concerns codes and consent, specifically informed consent, so that the participants agree
to the research design as it relates to them. We can suggest that this could be fulfilled by
initial contact with the participant containing information on the style of the interview, the
topics that will be raised and how the data will be used. Post-interview this could be
followed-up with a courtesy letter of thanks and a review of how the data has advanced
the research. Ryen’s

49
second ethical issue, confidentiality, can be confronted at the same time by offering to
agree to formal confidentiality before the interview and stating bow the respondent is to
be identified in the research report. If these two issues are satisfactory to the participant
then the third ethical issue, trust, can also be resolved. It is behoven upon the researcher
to leave a good impression on the participants so as to facilitate the course of subsequent
research.

It seems, then, that although a set of prescriptive ethical rules can be applied to a case
study research programme there are nevertheless issues, such as Ryen suggests, that
should be accounted for. Just as the units of analysis lead the researcher to the appropriate
research design, so they can also direct the research to the appropriate ethical measures. If
a company is the unit of analysis, then the phases of the research where ethical concerns
might arise can be categorised as:

1. Access to the company.

2. Data collection process - interviews.
3. Publication of the research.

In the first phase, the research is at the mercy of the goodwill of the company so the
ethical issues are concerned with the research programme and its approach to the
company. Reference could be made to the BPS advice on deception when setting out how
the fieldwork proposal is formulated, including the structure of the data gathering and its
aims. It is also necessary to observe rules of etiquette which, though not strictly within
the field of ethics, colour the impression the research makes on the organisation.

In the second phase the ethical concerns are with both the researcher and the respondent,
although the respondent might not specify their nature. For example the respondent might
be concerned about professional reputation or the confidentiality of the information and
this would inhibit the transfer of data. At the same time, in a multiple case study design,
the researcher must be cautious about quoting data from one company to another, though
there is nothing unethical about this in principle.

50
For the final phase, the ethical responsibilities pertaining to publication are entirely with
the researcher. If confidences are broken at this stage it is unlikely to have a deleterious
effect on the research programme as it stands, but may adversely affect subsequent
studies. The professional reputation of academic research should be maintained in order
that the future of research in the field might be secure.

The overriding approach to dealing with the ethical issues arising from the three phases is
to neutralise them before they appear. When it is acknowledged that case study research
is, in the words of Merriam (1988), an emergent design, then it should also be
acknowledged that ethical problems will arise during the course of the research connected
with the situations to which they are attached. Just as the research design creates a
framework for the data to become known, so careful planning can reduce the potential for
ethical difficulties. This should be done by involving the participants in the research
design so that they are aware of how the study is expected to progress, the instruments of
this being clear statements of intent and offers to discuss concerns that the participants
might have. In this way the overt and emergent ethical issues that are unique to the case
study can be brought into relief and dealt with to mutual satisfaction.

2.6 Foundation for research design

The purpose of the foregoing discussion on methodology has been to present and discuss
the basis of the approaches taken to this research programme. Although a study of this
kind on company structure seems to lend itself to the case study approach this does not
absolve it from rigorous justification. The research questions play a role in this by
indicating the appropriate units of analysis and the methods by which they may be
evaluated. The research questions also indicate whether the research design should be
orientated towards quantitative or qualitative approaches. These can both be contained
within the case study strategy and this chapter has shown how the case study approach
provides a framework within which a coherent investigation may be conducted. In
essence, the research questions posed in this thesis form the foundation of the programme
and as such precede the research design, the following theoretical concepts and the
empirical data.

51
This chapter has not taken into account the unique aspects of this research programme, so
it has not evaluated the aspects specific to the research design that was put in place. The
next chapter, Chapter 3, will demonstrate how this final research design was
implemented. It is from the base established in this chapter that the details of the research
design may be defined, and these will be presented in the next chapter of the thesis.

52
 3: Research Design

While the previous chapter established the methodological foundation to the research
strategy, this chapter is concerned with the practicalities of the data collection. The case
study design will be described along with the selection of candidate companies. Although
ultimately the case study focuses on a single company, the industrial context is made up
of many other companies operating under the same conditions. The context can be further
enhanced by appeal to other sources connected to the industry but not directly involved in
the companies comprising it.

This chapter will also lay out the progress of the research programme in order to show the
full structure of the data collection. The data itself will be utilised and analysed in the
subsequent chapters to which it is related, while some of the research material will be
made available in the appendices to the thesis.

3.1 Qualitative case study

The purpose of this research programme has been to investigate the scale and integration
characteristics of an automobile company. Although the interview protocols were
concerned with detailed questions, the overall research questions focused on three broad
areas:

1. Automobile company size according to technical efficiencies.

2. Automobile company organisational structure.
3. The degree of latitude in achieving the theoretically determined size and structure

(i.e. the potential for approximating to the most competitive size and structure).

As discussed previously in sub-section 2.1.2, a study is qualitative or quantitative

depending on the treatment of the data rather than the nature of the data itself. In this
thesis the data sourced from the literature are used in an emergent fashion (Creswell,

53
2003) to define the limits of company size and structure, with the empirical evidence
bringing wider contextual data. The interviews, in accordance with Lee (1999), permit
those who are influential in the industry to bring their own interpretation to the problems
posed by the research questions.

Although technical efficiency of production is itself a quantitative problem, its treatment

in this research programme is essentially qualitative since the precise differences between
different production sites cannot be accurately quantified but they can be explained and
comprehended. Relative technical efficiencies therefore tend to be based on expert
judgement rather than precise measured quantities. Company structure considerations
would also be difficult to quantify given the inclusion of economic concepts of
opportunism and global factor endowment differentiation. This does not mean that the
quantities do not exist but that it would be difficult to define a priori what dimensions
they should take. For example, if quantitative analysis found that a small company (eg.
Porsche) was making high financial returns on low production output then the study
would conclude that mass market manufacturers should follow the same pattern.
Qualitative analysis based on the judgement of knowledgeable interviewees would,
however, find that such a company was dependent upon a successful cost recovery pricing
as part of its prestige image and thus not an option available to mass market
manufacturers. The qualitative approach is therefore more suited to the building of a
multi-dimensional theoretical model because it can use input from experts to interpret the
numerous perspectives.

Miles (1979) embraces this complexity when putting forward the “goal-system state
analysis” where the means for achieving a target, such as company size and structure, can
be multifaceted. This lends itself to the qualitative approach where the research can infer
causality based on the data collected in interviews. This was often expressed in the
interviews with “What if...” style questions to elicit responses that would illustrate the
range of options available within the industry. Naturally, only semi structured interviews
could include such flexibility within a systematic study. Indeed, the case study company
presented a series of possible scenarios which were then discussed during interviews at
MG Rover. The next section of the thesis will look at the selection of MG Rover as the
case study that is apposite to the research questions.

54
3.2 Selection of the case study company

The subject of this case study, MG Rover, was chosen for its ability to exemplify the
issues facing automobile manufacturers within the industry. The choice among UK
companies was severely limited by the fact that, of the few independent manufacturers
left, at the time this thesis was commenced in 2001 only MG Rover operated as a volume
producer. Moreover, it was only by pure serendipity that it was independent at all, having
being sold by the previous owners, German car producer BMW, to Phoenix Venture
Holdings in 2000.

Other potential case study subjects might have become available by looking outside the
UK. The last full year of MG Rover production was 2004 when around 108,000 vehicles
were produced (Automotive News, 2006b). At this level of output, most other
manufacturers were either protected (eg. by government decree) or were engaged in
licensed production of models developed by other companies. The next largest company
in terms of annual output that was both integrated and fully exposed to global competition
was Porsche but it is not a mass market manufacturer, which means that it is not fully
committed to the mass production technology, and in any case its second largest selling
model was a joint venture with VW for an SUV.

MG Rover production reached a peak of 147,000 units in 2002 (Automotive News,

2005f), at a time when this research programme was firmly established. Then, the next
highest integrated manufacturer was SsangYong Motor with a little over 160,000 units
produced. That company, though, specialised in SUVs, which are not customarily made
according to the mass market passenger car production technology. Table 3.1 shows a
selection of the companies that most closely resembled MG Rover in 2002, demonstrating
scale in terms of output and degree of vertical integration.

55
 Table 3.1 Range of potential alternative cases to MG Rover

Company Output Vertical Integration

Lancia, Italy 105,297 Fiat Group dependent
Shared platforms
237,024 Shared engines
Proton, Malaysia Government dependent
Saab, Sweden 124,892 GM dependent
GM platforms
GM engines

Sources: Automotive News (2004f)

OICA (2006)

From Table 3.1 it is clear that there are few automobile companies that are sustainable
near the level of output MG Rover temporarily achieved in 2002, or even the 200,000
annual output the company targeted. Those manufacturers that are using the same mass
market automobile production technology with these quantities of output are doing so as
part of larger groups. Lancia of Italy is dependent upon Fiat for the basic vehicle
architecture and the powertrains. Proton strongly resembles MG Rover in terms of output
and product characteristics, yet even though it is independent of other manufacturers it
benefits from material government support. Saab had a similar level of output to MG
Rover but its two models are solely in the executive segment and, although it had a
measure of autonomy within General Motors (GM), this element of independence has
now been lost. Other vehicle manufacturers that appear to have comparable production,
such as AutoGaz of Russia with 2002 production of 198,135 (Automotive News, 2004f),
have not been included since they are not using the same production technology. Indeed,
AutoGaz specialise in commercial vehicles, as do most of the Chinese producers at a
similar output level.

Mazda was an attractive proposition as a prospective case study company since it had a
similar range of products to MG Rover; its MX5 sports car is often cited as the spiritual
successor to MG’s MGB. The problem in using Mazda as an alternative case was whether
it could be considered independent or not. Ford bought a 24.5 per cent share in the
company in 1979, representing a strategic investment rather than an attempt at control.
The two companies have cooperated on various occasions, such as assembly of the Ford
Probe sports car and the establishment of the Korean company

56
Kia Motors in 1985, but Mazda retained an autonomous product plan and innovative
engine design (e.g. rotary and Miller Cycle engines). In 1996 Ford increased its
ownership to 33.4 per cent (Mazda, 2004) and now counts Mazda as one of its brands
(Ford, 2004). Furthermore, production by Mazda was far in excess of anything achieved
by MG Rover, amounting to 773,798 in 2002 (Automotive News, 2004f). Mazda was
thereby rejected as a valid alternative to MG Rover on the grounds of its lack of
independence and being far larger.

In 2002, all the mass market automobile manufacturers that could claim to be
independent and vertically integrated enjoyed annual production above 1 million units a
year. This ranged from GM with 8,277,000 units, though many of them would have been
sports utility vehicles (SUVs) based on commercial vehicle platforms, down to BMW
with 1,090,258 units, a minority proportion of which were also SUVs (Automotive News,
20041). Companies of this size also tend to be globally dispersed. This meant that they
would not have contrasted sufficiently from the proposed theorectical model to
unequivocally falsify it. MG Rover was not only operating far below the scale accepted as
the industry standard, but its main corporate functions were also tightly integrated on one
site. The fact that, as a UK based company, it offered easier access was merely a bonus.

The resultant research owes its design to Yin’s multiple case embedded case study
quadrant of the 2x2 matrix. The embedded units of analysis are represented by
investigations into scale and integration of different company functions. The different
companies in the research are each separate cases but with the main emphasis on MG
Rover as the one most likely to test the limits of the proposed theoretical model. The
remaining companies therefore provided the contextual background, revealing the
prevailing norms in the industry and providing support for the model. Research resources
were particularly focussed on Honda as a contextual case since the company had, at one
time, been closely allied to British Leyland and the Rover Group, the predecessors of MG
Rover. This provided an opportunity to make comparisons between two cases that had
shared a strategy from which they had since diverged. This suggested that one, or even
both, of the companies had also diverged from the industry norms of scale and
integration. The remaining contextual data was gathered

57
from other automobile manufacturers as well as external sources closely involved in the
industry.

3.3 Company access

Over the two year period that the fieldwork encompassed, access problems fluctuated.
Starting with no established contacts with any of the target subjects, a basic procedure
was used to gain access to potential informants. This was conducted in three successive
stages:

• Identification of potential subjects - from media interviews, citations

• Correspondence by letter
• Arrangement of interview

3.3.1 Case study access - MG Rover

This three-phase approach evolved after experiencing a failure to maintain access with
MG Rover once initial contact had been made with the company’s education co ordinator
who headed the Education Partnership, a division of the company that interacted with
educational institutions. This was useful in gaining access to a tour of the factory by
joining a school group and it was thought that this would be a precursor to deeper contact
throughout the company. Although the risk of failure was heightened by the dependency
on this gatekeeper, her role as a specialist in liaising with educational institutions was
expected to mitigate it. This proved incorrect as the gatekeeper felt uncomfortable making
introductions at higher management levels and further contact through this avenue was
precluded.

Fortunately, at the time the research was being conducted, MG Rover was receiving
extensive exposure in the media and certain personnel could be easily identified from
interviews. The first of these was the PR representative, Mr Stewart McKee, Head of
International Media Relations Sales and Marketing, and although he could also have
become a restrictive gatekeeper he was found to be more sympathetic to independent

58
researcher contact with management. Contact was then made with Dr Chris Millard, head
of strategy, and later Mr Rob Oldaker, director of product development. All three were
identified first from the media and then contacted by letter (see Appendix 3 for McKee
introductory letter). Having established a good relationship with the company, other
interviewees were identified from the company’s media website once the required
password entry requirements had been granted. As before, each approach was followed
by a letter of introduction.

This approach, however, was not suitable for interviewees who had retired from the
company. There were three of these in the study: Mr John Bacchus, ex-BL director, was
discovered after a speculative letter to British Motor Industry Heritage Trust at Gaydon
where he happened to work part-time, and from that interview an introduction was made
to Mr Peter Woods and Mr Malcolm Harbour, also ex-BL directors. Mr Bacchus and Mr
Woods had also been intimately involved in the relationship between BL and Honda. In
each case, as before, an introductory letter was used seeking an interview.

3.3.2 Contextual industry access

In order to form an understanding of MG Rover’s industrial environment the research

included contextual companies and institutions. To make this understanding as
comprehensive as possible, this included a wide range of automobile firms in the UK and
Japan. The research in Japan permitted a broad view of the global industry while
exploiting this researcher’s capabilities with the Japanese language and culture.

Due to Honda’s close involvement with MG Rover’s predecessors, BL and Rover Group,
the Japanese company was targeted as a key provider of contextual information. Honda
has no education specialist and the company does not entertain the same intensity of
public exposure. A speculative letter was written to Honda Motor Europe in the UK
which was passed on to Mr Chris Rogers, Head of Corporate Affairs, PR Division.
Although an interview with him was not immediately secured, regular correspondence
was maintained until he made an introduction to the director of PR for Honda Motor
Europe, Mr Takeshi Sumita. First contact with Honda Japan was made by accessing the
media website, sending emails to a selection of PR

59
representatives and receiving a reply from Mr Tatsuya lida. Since this introduction was
expedited more quickly than that with Honda Europe, this interview took place before
that with Mr Sumita, though the contact with Honda Europe had been made earlier.
Subsequent interviewees were identified from the media website and approached in the
established manner.

Other contextual interviews were arranged in a slightly more ad hoc approach.

Institutions such as the Japan Automobile Manufacturers Association (JAMA), the
Japanese Ministry of Economy, Trade and Industry (METI), Mitsubishi Research Institute
(MRI) and the UK’s Society of Motor Manufacturers and Traders (SMMT) were
contacted with speculative letters and a correspondence established with the
representative to which the request was then referred. This was because specific contact
details had not been initially available. Car company personnel were identified from the
media as before, though Mr David Bloom of Jaguar, Japan, was introduced by Mr Wood
of Rover Group, and from Mr Bloom an introduction was made to Mr Anthony
Millington of ACEA in the style of snowball sampling. Although contact details for Mr
Millington were not specifically available, the use by his organisation of standardised
email address formats meant that his details could be discovered by simple trial and error
with email addresses. There was one introduction made by an academic at Cardiff
Business School, this was to Professor Shiromi Shioji of Kyoto University, but this then
led to introductions with Professors Yasuo Sugiyama and Tsutomu Demizu.

The letters requesting interviews were all of the same basic type (see Appendix 3), unless
including additional gratitude for the role of an intermediary. The letter summarised the
purpose of the research programme and how the interviewee would be able to contribute
to it. Confidentiality was assured along with a statement that the purpose of the project
was intended to benefit the business community in general (see Appendix 11, letter to
Chris Millard). Letters to Japanese contacts further emphasised the mutual promotion of
international understanding, it being known from personal experience that these are
concerns to which the Japanese tend to be sensitive. The correspondence continued after
the interviews with letters of thanks.

60
3.3.3 Access hold-ups

There were two major problems in gaining access, both concerning hold-ups created by
the gatekeepers. The first, with MG Rover’s educational representative, has already been
described and this was circumvented by repeated independent efforts at new contacts.
The second hold-up was very similar, caused by Mr lida of Honda, who gave one very
hospitable interview but when the second field trip to Japan had been arranged he
announced that no more interviews at Honda would be granted. This was a serious
setback since he was in a much more commanding position than MG Rover’s education
representative. While Mr lida might not have been responsible for granting permission
for interviews he could have made further contact with personnel very difficult.

The tactic to solve this problem was to identify the CEO of Honda from media sources
and appeal to him directly. This appeal was made in a diplomatic form, containing
effusive praise of Mr lida and stressing the role of the research in helping Honda promote
international understanding. The expectation was that the current CEO would not have
time to be interviewed in person but that he would encourage Mr lida to continue aiding
the research, simultaneously putting pressure on Mr lida to cooperate whilst also
enhancing his reputation with his superiors. It was thought that discretion was the most
effective way of solving a difficult situation without putting Mr lida in an invidious
position. It transpired, however, that the person identified as the CEO of Honda, Mr
Hiroyuki Yoshino, had retired six months previously, but this had the advantage that he
was available for interview after all. It was likely that the diplomatic move was entirely
successful since Mr lida gave another lengthy interview and was present during the
interview with Mr Yoshino.

There were few such critical problems with the other contextual interviews since they
were not, on an individual basis, as central to the research as the company interviews. For
example, though the interview at Morgan was useful it could have been replaced with one
at, say, TVR, a similar scale of operation. Moreover, a refusal that precluded further
interviews at such an organisation was never a fatal setback because depth of this kind
was not being sought. In any event, refusals were rare, the only notable one being Sir
Michael Edwardes, the retired president of British Leyland who had

61
presided over the early years of the joint venture with Honda. This was not considered an
important omission to the data, other sources being quite capable of covering the same
historical perspective.

3.4 Other documented information

In the course of collecting interview data other documented material, internet based
information and letters became available and these have been cited as applicable in the
same manner as other literary sources. Such information was not collected systematically
but was requested at all interviews, being obtained in an ad hoc manner.

MG Rover was not a fertile source of additional data; as a recently independent private
company it appeared to be fairly conservative in publishing its own information. Access
was granted to the media website and it was a useful source of contact details, though the
company appeared to be using this mostly as a marketing tool. Some historical documents
concerned with BL and Rover Group are in the public domain, such as government
reports, and even if their relevance to MG Rover was tangential they have been quoted
where appropriate. After the data gathering phase of the research, letters of thanks elicited
two responses of material interest from Mr Oldaker, a letter and an email.

Honda publish literature such as an annual FactBook comprising a precis of output data
and historical background. It also allowed access to the company’s media website which,
during the early part of the research programme, was found to be a valuable source of
diverse material. The company rationalised the website sometime in 2004 whereupon it
became little more useful in content than the MG Rover site. Fortunately, the company
also publishes books celebrating its successes and although a certain amount of company
bias appears in the form of quasi-slogans and even hubris (e.g. “we call it the art of living,
the Honda way” Honda, 1999, p.296) the publications were a useful source of
information on the company philosophy, historical background, anecdotes and numerical
data.

62
Documented information from the other contextual fieldwork was of considerable utility.
Aston Martin provided figures that showed the rise in output under the patronage of Ford
and compared this with Fiat’s acquisition of Ferrari (see Appendix 6). Other companies
were keen to promote their environmental concerns, the visit to Mazda yielded no
interviews but plenty of information on this subject along with financial reports.
Environmental sensibilities were also advertised in this way by Toyota and Honda,
although the interviews did not focus on this as it was not within the remit of the study.
Institutions such as the SMMT and the European Automobile Manufacturers Association
(ACEA) in Japan were a useful source of data on industry trends, independent of the
manufacturers. This was further enhanced by data from the trade unions Amicus and the
TGWU, although these were not of the highest quality and even contained a large number
of typographical errors on occasion. The trade unions in Japan were not contacted, in part
due to their reputation of being more enterprise based and therefore biased towards the
company, but also because the British unions were unable to make an introduction.
Instead, METI provided publications on the Japanese automotive industry and its
expansion overseas which the UK’s Department of Trade and Industry (DTI) did not
make available, though the SMMT partially filled this role. There was little in the way of
useful correspondence from the contextual sources, although Mr Mark Aston of Morgan
sent a brief letter of some interest.

3.5 Triangulation

Triangulation is a metaphor referring to the use in navigation of three back bearings from
known points in the distance. The three lines drawn back from the known points cross
each other at a point that then denotes the position of the navigator. In theory the
intersection of two lines only are necessary, but in practice the presence of errors means
that the third line acts as an additional check on the position. This is an apt description of
the way that different data collection methods are used in qualitative research to
illuminate the same concept, the contrasting lines of inquiry converging on the same
conclusion. For example:

63
“...organizational researchers can improve the accuracy of their judgements by collecting

different kinds of data bearing on the same phenomenon.” (Jick, 1979, p.602)

The navigation metaphor should not be taken too literally, however, since the number of
enquiries and the manner in which they are conducted is not pre-determined. Indeed, for
any kind of triangulation there may only be two independent approaches being made.
Denzin (1978) gives four types of triangulation:

• Triangulation of data - various data drawn from different sources or times.

• Investigator triangulation - input by various researchers.
• Triangulation of theories - use of multiple hypotheses and perspectives.
• Methodological triangulation - use of various methods.

Triangulation of data can include visual data being triangulated with verbal data (Flick,
2004). In a study such as this research programme, which makes use of qualitative elite
interviews in order to comprehend the economic cost base of a typical automobile
company it is clear that triangulation of data predominates. It was the converse for
investigator triangulation, there being but one researcher for the programme and so
investigator triangulation was not an option. Concerning triangulation of theories, it is an
approach where:

“...the results of an investigation employing a method associated with a research strategy

are cross-checked against the results of using a method associated with the other research
strategy.” (Bryman, 2001, p.447)

Although this research programme took two theoretical approaches, one for the size of
the company and the other for its structure, they were discrete from one another and so
cross-checking between them was not possible. There was some methodological
triangulation using literary sources for building the conceptual framework and sourcing
archival data, with the data from the fieldwork then being used to provide an empirical
context.

64
Triangulation is intended to bring mutual validation, but Flick (2004) warns that there is a
specificity to the different methods and sources which must be recognised. Just as
theories on company size and company structure can both describe the same organisation,
they do so from entirely separate theoretical perspectives. Denzin has subsequently taken
the view that the typology of triangulation may not so much a tool for validation but more
a strategy for gaining a more sophisticated understanding of the subject. Janesick (2000)
evolves Denzin’s view from the two-dimensional triangulation analogy to one of
crystallisation, one that describes having multiple views onto the subject like the facets of
a crystal. It is not, therefore, about validation but the gaining of additional knowledge
(Flick, 2004). Nevertheless, triangulation is the generally accepted term for describing a
research design bolstered by additional sources of information.

The data gathering of this fieldwork was designed to create a network of references that
would illuminate the different aspects of organisational size and structure. This involves
not only searches of the available literature but also empirical data from the research
interviews. The purpose of the interviews was for them to both triangulate with the
archive data from the literature and to triangulate the case study company, MG Rover,
with the fieldwork conducted in the contextual automobile industry. The interview
protocols achieved this level of triangulation by seeking insights into certain fundamental
questions across all the interviews. As a group, all the interviews were triangulated with
the literature since the questions were derived from it. Triangulation amongst the
interviews was then effected on two levels: between the literary and empirical sources,
and between the individual intemviews.

3.5.1 Literary and empirical source triangulation

The research questions concerning company size, structure and the degree to which these
excluded approximations led into a review of the literature to discover the prevailing
theories. Since it is assumed that the underlying technology in the mass market
automobile industry has been constant for over half a century, notwithstanding process
improvements, the data found in the literature is considered to be of archival, rather than
historic, relevance (see 2.1.1). The archive data can then be used to inform the conceptual
framework concerning company size and structure.

65
This has been particularly relevant in the matter of survivor analysis which assesses the
development in company size over a period of time. Such a longitudinal approach would
not have been feasible for a research programme of this kind due to time and resource
constraints. Since a problem with archive data is that it is filtered through the original
research first, triangulation can be employed by using different sources of archive data
and then bringing in empirical data gathered during the fieldwork phase of this research
programme.

3.5.2 Interview triangulation

The interviews were all conducted with those who were influential within the industry.
They were also selected to represent a wide range of perspectives with the intention that
they would triangulate with each other. Within the MG Rover case study interviews, the
interviewees, including ex-employees, represented a range of competencies. For the
company these were as follows:

• Senior Management • Human resources

• Public relations • International sales
• Engineering • Independent analyst
• Strategy

Some of the interviewees were able to speak on behalf of more than one area of
knowledge, for example senior managers Mr Bob Beddow and Mr Rob Oldaker were
also qualified to talk about HR and engineering respectively. Mr John Bacchus had been a
senior manager with experience of strategic planning and a strong background in
accountancy yet, as a retiree of the firm now working in academia, he also had a level of
independence, often expressed as cynicism. Most of all, top management have the ability
to convey an overview of the company, so the degree to which interviewees were able to
digress from the remits of their formal job titles is not of particular relevance; the
importance lies in the range of experience they could bring to the research. This created
the conditions for a degree of consensus which might not have been possible if
interviewees had stayed within their technical competencies, for example if Mr Beddow
had restricted himself to HR concerns. The result would have

66
been a plethora of unrelated views that resisted a unifying view of the company’s
situation.

The interviews that took place outside of the case study company were to provide
contextual data and bring another level of triangulation. The interview protocols were
designed to be comparable and again the interviews were grouped into similar categories:

• Senior Management • Independent analysis

• Public relations • Political Institutions
Engineering • Industry Institutions

In these interviews, “strategy” was not found as part of a formal job title, although the
subject was discussed on many occasions. Instead, there is the addition of institutional,
political and research job categories. This disparate array of data would support a
particularly robust framework should a unified argument concerning the automobile
industry emerge. For these interviews, the subject matter was more general in theme,
making reference to the challenges of the industry. Indeed, a point of interest is that such
was the dominance of British Leyland and Rover Group in the UK in the past that many
of the contextual interviewees had experience with the firm and so were able to make
informed comparisons.

As previously noted, a great deal of the contextual research was conducted at Honda. Due
to the limitations of access at Honda it was not possible to match the competencies of the
informants with those at MG Rover. This is countered by the wider overview that the
interviewees had, most notably Mr Yoshino. The functionary competencies covered were
as follows:

• Top management Strategy

• Public relations Independent analysis
• Engineering

67
The range is further narrowed by the engineering bias at Honda, Mr Yoshino being no
different to previous CEOs in rising to the top through Honda Research and
Development, while the strategy team were also from an engineering background. The
independent analyst was an academic, Tsutomu Demizu, who had a very close
knowledge of the company and its development.

Although widening the scope of interviews within companies can unveil a unified view
of the company, there is still a danger of group bias. This might even be formally
expressed by the public relations personnel who are the appointed spokespersons for the
company and the message it wishes to convey. Their contribution is, by corporate
intention, triangulated with the other informants within the company but it does not
tackle the group bias, indeed it may reinforce it.

The prominent role played by Honda as a contextual company, and one that had enjoyed
close links with the predecessors of MG Rover, meant that rich data could be gathered for
comparison. MG Rover and Honda were both independent companies but were
diametrically opposed in terms of scale and global spread. Many of the challenges facing
the British company were the same ones that Honda had faced in the past. Taking the
notion of independence as an example, it could be a manifestation of management hubris
at MG Rover, or a positive way of expressing a failure to find a collaborative partner, but
Honda’s success shows independence has some merit. Thus claims that independence can
bring a competitive advantage are validated by Honda even if they had appeared suspect
at MG Rover.

68
3.5.3 Interview frequency

The case study research strategy related MG Rover to the different contextual
companies at both data gathering and analysis stages. The purpose of the laddering
interview method was more than to aid data gathering within the companies, it also
informed the interviews at the other case studies. Since the initial interviews were
conducted with MG Rover this meant that the topics were more refined by the time of
the fieldwork in Japan over a year later. Figure 3.1 shows the relative timing of the case
study interviews along with the contextual interviews, indicating how the contextual
interviews preceded those at MG Rover although with the bulk of the fieldwork was
contemporaneous. Contact was maintained with MG Rover personnel even after the
financial collapse of the company and Figure 3.1 records interviews with those who
went on to work for the Chinese buyers of the assets. The illustration gives an
impression of the approximate frequency of the interviews in total, drawn as a line
graph to illustrate the manner in which interviews linked to each other.

Figure 3.1 Frequency of interviews

No of interviews

10
Contextual

5—

MG Rover

2002 2003 2004 2005 2006 2007

Years

In all, the fieldwork took in fifty-four elite interviewees lasting one and half hours on
average. The first interview took place in July, 2002 and the last in February, 2007. This
comprised eighteen interviews at MG Rover and related organisations, with

69
another thirty-six in the contextual industry. Details of all interviewees are listed in
Appendix 1, and where more than one interview took place the appropriate date will be
indicated when it is referred to in the text.

A particular problem with this style of qualitative data gathering was in deciding how
many interviews would be necessary to draw valid conclusions. The statistical measures
of confidence that quantitative studies can call on were naturally not available. The
number of interviews is also an ill-defined requirement in the qualitative methodological
literature, and the advice of one academic that ‘you know when you know’ was
considered especially unhelpful. This research therefore devised two requirements: first
that the research questions should be answered and, second, that the body of knowledge
should be advanced.

For the first requirement the research questions were formulated from the existing
literature, with specific questions arising during the laddering build-up of successive
interviews. In order that the questions could be sufficiently answered, the interview data
needs to be mutually supporting through triangulation. To the degree that this is
successful the existing literature is validated, but for the research to advance into fresh
territory the second requirement should be fulfilled in order to expand the body of
knowledge. Elite interviews are fertile ground for making advancements because the
respondents are themselves engaged in the future of the business. This should not,
however, include undisciplined speculation and triangulation helps to control against this.

Concerning specific organisations, MG Rover was the subject of the greatest number of
elite interviews, namely fourteen. This comprised eight executives being questioned on
different occasions, three more who had retired from the firm and another three working
for the successors to the assets of MG Rover. Amongst the contextual cases Honda
predominated with ten executives participating. The rest were distributed across various
sources and provided additional perspective. As the interviewees were aware of the
purpose of the research they often made comparisons on their own initiative. The location
of the interviews is not of material consequence either; some of the interviews that
informed on the issues facing the British industry taking place in Japan.

70
A further requirement for interview numbers is suggested by the laddering metbodology.
As successive interviews are linked to one another the process can be considered to have
reached its conclusion when no new information is found. In reality this is unlikely, since
there is always the possibility of finding one more interviewee who can add knowledge,
even if their relevance to the chain of interviews is tenuous. However, when the laddering
progresses up the management hierarchy then the limit is reached at the highest effective
interview. The data gathered at such an interview should still be in accordance with the
two basic requirements to answer the research questions and expand the body of
knowledge, yet at the same time this most elite of interviews should yield data of the
highest quality. If the two requirements are satisfied by this interview then that ladder can
be considered to have reached its limit. The quality of the data gathered by this research
programme was particularly enhanced by the participation of highly placed executives,
such as departmental heads of contextual institutions, such as the ex-CEO of Honda and
the heads of strategy and engineering at MG Rover.

3.6 Ethics

The major ethical problem with research on corporate organisations concems the
treatment of commercial confidentiality. The automobile business is highly competitive
and the financial position can be volatile. One hardly needs to review the history of MG
Rover to understand that this point is particularly pertinent. This research was not
concerned with the details of specific company projects and so commercial sensitivities
were unlikely to be transgressed, yet reassurances were given in order to induce a trusting
atmosphere to the sessions and encourage communication.

In the course of the introductory correspondence an offer was made to sign any formal
agreements to confidentiality. The ultimate purpose of the research was also declared.
This simple act appeared to be adequate since no formal, or even informal, agreement
was ever demanded. First interview contact was often with the PR representatives but this
did not appear to be a deliberate policy by the companies, it was simply that they were
more easily identified and available for interview. Those representatives were

71
present at later interviews in a supporting role; MG Rover’s Dr Chris Millard, for
example, requested the presence of the PR representative at his first interview (17
September 2004) but this did not impede the flow of information. Indeed, the
representative was present only for the first quarter of the interview. Honda’s PR
representative was present at the interview with Mr Yoshino but in the role of usher rather
than guardian.

An interesting point about the observance of secrecy was that companies claimed to have
no secrets to hide, at least with regard to the issues under investigation. The motivation
tended to be to explain policies and the thinking behind the stories that existed in the
public domain. The only subject that appeared to be precluded was by Honda concerning
the earlier joint venture with Rover Group, and although no questions were scheduled the
subject was frequently broached on the volition of the Honda interviewees. The manner
in which this purportedly sensitive subject was openly discussed is an indicator of the
generally communicative attitude the interviewees took in relation to all issues. On only
two occasions did problems occur, both with PR representatives: firstly with MG Rover’s
Mr Greg Allport, and then with Honda’s Mr Tatsuya lida, who temporarily refused further
contact (see Section 3.3.3 above for a full account). Neither of these were injurious to the
research.

As a precaution, the contextual interviews provided the additional information that might
make up for any sensitive information the interviewees would wish to avoid. Without the
constraints of commercial sensitivities many of the informants, particularly the
independent analysts, would be expected to be at greater liberty to speak. That little
contradictory data was gathered from such sources is testament to the comprehensiveness
of the company interviews.

72
3.7 Initial phase

Although there was no single pilot study, the nature of ladder interviewing meant that the
early data collection formed the foundation for the subsequent fieldwork. The first of
these was with Mr Mark Aston, Assistant Managing Director of The Morgan Motor
Company. Despite the obvious fact that this company operates at a scale unrelated to
mass market car production, and was thereby outside the parameters of the mass
production technology, the company was chosen for the start of the research because it
was hoped that it would provide a valuable introduction to the conduct of fieldwork.
Subsequent interviews gathered an overview of the industry from the trade union
perspective and ad hoc interviews at motor shows.

3.7.1 Morgan Motor Company - an introduction to fieldwork

Morgan sports cars are described as “traditional”, and here that is taken to describe a
sports car of 1930s vintage constructed in the artisan craft production style. Yearly
production is measured in hundreds, approximately the same amount a mass production
facility produces in a single day. The company is also majority owned by the Morgan
family, family-owned stakes in volume manufacturers being rare and none in an absolute
majority (see Appendix 2). This suggests a sentimental attachment between the family
and the firm that may offset the economic demands of automobile production, Harvey-
Jones (1990) being particularly fhistrated at the conservative character of the management
of the company when it came to modem innovations.

While the small size of the company precludes the research from accepting Morgan as
representative of the mass production industry, this does not prevent the case from being
instmctive. Importantly, for an ab initio researcher, it is an assembly line in miniature, the
entire process being clearly visible from a good vantage point on the factory driveway.
Equally, one well-placed interviewee, in this case the assistant managing director Mr
Aston, can reveal a managerial vantage point that provides a perspective of the industry
as a whole. The interview was semi-stmctured, as all subsequent ones were, with
questions derived from a thorough review of the available literature. The questions were
not grouped together into subject areas as were later

73
interview protocols but Mr Aston was provided with an advance copy in order for him to
prepare for the interview. The utility of this tactic led to it being adopted as policy
throughout the remainder of the research programme even if the data obtained from
Morgan was not useful in formulating a model for the mass market industry.

3.7.2 Trade union perspective

A danger of interviewing inside the industry is the possibility of reiterating orthodox

views held by the industry. Even aiming for triangulation by selecting participants widely
within the companies would not be sufficient in eliminating the most fundamental
company bias. Union leaders were seen as a useful source of data because they conduct
their own research focusing on the long term viability of the industry with regards to
employment rather than company-centric measures of financial success or personal
advancement.

Amicus-AEEU was chosen on the basis that it is the UK’s largest manufacturing trade
union, representing 35,000 members in the motor vehicle sector, the other main union
being the Transport and General Workers Union (TGWU) (Amicus, 2004). The National
Officer for the Motor Vehicle Industry, Duncan Simpson, was identified from the union
website and contacted by letter seeking an interview. At Mr Simpson’s behest, this was
arranged at a time and location convenient to him, which happened to be during a union
trip to Swansea. The interview took place at his hotel in the evening. The details are
relevant because the informality of the event made it possible for Mr Simpson to
introduce his colleague, Mr Dave Osborne, National Secretary Vehicle Building and
Automotive for the TGWU. This was fortuitous since Mr Osborne was the more active
participant in the interview, expressing his views candidly and even aggressively. This
was no impediment to the gathering of data, resulting in a reappraisal of the research
programme and a later interview with Mr Osborne and his research assistant at the head
office in London.

The interview protocol for the first interview was in the same format as the previous one
for Morgan, but the content dealt with the British motor industry in general and MG
Rover in particular. As before, the protocol was submitted in advance and on this
occasion this had an added advantage in that an assistant to Mr Simpson was able to

74
provide him with written answers to some of the questions in the form of a briefing. Mr
Simpson contributed to the interview but in general it was dominated by Mr Osborne. A
record was made in the form of interview notes, but it was found that this inhibited the
flow of the conversation and did not produce a comprehensive account. The later
interview with Mr Osborne corrected for this by use of a tape recorder.

3.7.3 International motor shows

As an opportunity to create a network of potential contacts international motor shows

were chosen as an effective forum for making introductions and possibly arranging
interviews. This was because motor shows set aside preview days, before opening to the
public, when industry executives make themselves available en masse for meetings to
publicise their work. This would be an excellent opportunity to introduce the research to
as many contacts as possible in the shortest time and so ease the process of requesting a
formal interview later. Three motor shows were attended: Geneva (March 2003), Tokyo
(November 2003), and Birmingham NEC (June 2004).

The Geneva show was particularly attractive because it is, by reputation, more informal
than the other international motor shows. The public relations (PR) departments of two
companies, MG Rover and Honda, were contacted in the preceding weeks in order to
gain a feel for how meetings at the show could be arranged and who would be available.
It was soon found that meetings were permissible on an ad hoc basis as personnel
became available, though as a precaution meetings were pre-arranged with PR
representatives of the two companies. These were Mr Stewart McKee of MG Rover, and
Mr Chris Rogers of Honda. It was, perhaps, a mark of MG Rover’s status at the motor
show that Mr McKee was able to be generous with his time in granting an interview. The
interview protocol was a generic one to be used at all meetings at the show, based on the
first union interview with Mr Osborne but with specific company references emphasised.
The interviews were recorded in a written format. It might also be an indicator of
Honda’s status that no interviews or introductions could be arranged during the course of
the show.

Two ad hoc interviews were conducted, one with an industry analyst, Mr Pat Devereux,
and the second with Mr Greg Allport, the MG Rover General Manager PR

75
and Events. The interview with Mr Devereux was informal and interesting but did not
jdeld much useful data; the interview with Mr Allport also yielded very little data due to
his excessive observance of commercial secrecy, even concerning information available
in the public domain.

Although data was gathered at the Geneva motor show and introductions made,
considering the time and distance involved it was probably not cost effective. It was
hoped that the lessons learnt from this would improve the efficacy of attending the two
subsequent shows. The Tokyo show was taken as a convenient starting point for
fieldwork that had been previously arranged. It was not, therefore, a crucial feature of the
trip but it was hoped that some valuable introductions could be made. This did not
transpire, all industry executives being unavailable for anything but media interviews.
Fortunately, interviews unrelated to the show yielded valuable data so the disappointment
of the show had no material effect on the data gathered during that visit.

At the British Motor Show in Birmingham it is worth noting that MG Rover policy at the
show was to make company personnel available on the stands instead of hiring external
publicists. It was therefore possible to hold follow-up interviews with Mr McKee and Mr
Rob Oldaker, Director of Product Development, who had been previously interviewed at
length ( 6 January 2004). Other industry personnel were again unavailable.

Across the three shows valuable contacts were made, but these could probably have been
made without attending the shows themselves. From the experience it is not advisable to
rely on ad hoc meetings unless attendance at the show is convenient for other purposes.
Attendance at such events is, perhaps, more suited to the maintaining contacts through
informal meetings.

76
3.8 Main interviews

Since each interview evolved from the previous one and informed the following one it is
not possible to identify a formal pilot study. Even so, the early phase was constructive in
laying the parameters for the main thrust of the research and helped to codify the
interview protocols. In the early phase, the interview protocols came to be substantially
changed for the main company and contextual industry research. The material covered in
the interviews was consistent, although with a semi-structured format it was usually
tailored to reflect the focus of the particular interview.

The full list of interviews is shown in Appendix 1. They have been placed in two groups,
one being for MG Rover and the other for the contextual industry. The interviews are in
chronological order within the groups but it should be noted that they were not conducted
in a planned order but as interviewees made themselves available. This means that the
interviews do not follow on from each other within the groups but across them, so a
Kyoto University interview could follow and be informed by an interview with an MG
Rover executive, for example. Another point is that the groupings are by interview
subject rather than current employment, so some of the respondents are ex-employees of
the companies who could provide historical depth. This is appropriate for an industry that
takes many years to develop.

3.8.1 MG Rover

The interview protocols were mostly sub-divided thematically: chronologically in the

case of John Bacchus (5 August 2003) to reflect the historical antecedents to MG Rover,
and thence categorically for all subsequent interviews in order to focus on emerging
topics. For the interview with Mr Alistair Morton the chronological format was found to
be unsuitable since the interview concerned contemporary concems rather than historical
events. The first categorical interview was with John Bacchus (10 June 2003) which
introduced the following headings under which specific questions were grouped:

77
• Globalisation
• UK Industry
• Japanese Industry
• Foreign Direct Investment (FDI)
• Joint Ventures

The purpose of this was to reflect Mr Bacchus’ experience in formulating the joint
venture between Rover and Honda by focussing on the relative qualities of the national
industries and contrasting joint ventures with FDI. The subsequent interview was with an
MEP, Malcolm Harbour, so the category of Government was added to investigate the
political angle. At the same time, the comparison between UK and Japanese industries
was dropped as this was not central to this interview.

The later MG Rover interviews introduced the concept of culture as a defining

commercial feature by asking about brand values. There was also an increased focus on
joint ventures, with allusions being made to a developing relationship with Tata of India.
Little was made of the specifics of any particular deal, partly for reasons of commercial
sensitivity, but more importantly because this research is not concemed with the minutiae
of extant relationships but how the executives of the company would like a deal to be
structured. Tata is a good example of this: the relationship arose after the collapse of a
much vaunted deal with China Brilliance (now known in the West as Brilliance China)
and before news of a possible joint venture with Shanghai Automotive Industry
Corporation (SAIC). To be concemed with contractual details would be to take an
historical stance, one that would quickly go out of date, without giving due prominence to
the principles at stake.

3.8.2 Contextual industry

Interviews took place with industry representatives in order to provide context and a
wider view of the issues. The thirty-six interviews can be divided into four sub-groups
(see Table 3.2). The data obtained covered a wide spectmm, from long-term political
concems to company details.

78
 Table 3.2 Contextual interview categories

Type of Organisation Organisation

Automobile companies Aston Martin
Bentley
Honda
Jaguar Japan
Morgan
Rolls-Royce
Toyota
Independent Researchers Automotive Engineering International
Kyoto University
Lean Enterprise Research Centre
Mitsubishi Research Institute
Osaka University
Redwood
Industry Institutions ACEA
JAMA
SMMT
Political Bodies METI
TGWU
Malcolm Harbour MEP

The automobile companies, apart from Toyota, were all in the specialist sector and one,
Morgan, has been previously cited as a craft producer. That said, the other companies use
current production technology, in Bentley’s case a production system based on Toyota’s
TPS, and Jaguar has production volumes from three factories broadly similar to MG
Rover’s capacity at one. Amongst them only Morgan and Toyota were independent, but
the other four had highly individual brand images which suggests that a measure of
autonomy within a larger group was an issue with which they were dealing. This might
then further help to define the limits of independence and suggest alternative courses of
action for the case study company.

The interviews with Honda came after the first phase of interviews and over half of the
MG Rover interviews had taken place. The structure therefore benefited from the
experience and they followed the later categorical format. As all but one of the interviews
took place in Japan it was necessary to present the protocols to the interviewees months
in advance, which meant that there was a common format for each of the three fieldwork
tours in Japan, resulting in three formats. Naturally, being

79
semi-structured, content was allowed to vary between interviews to allow one to inform
the next.

Although there were fewer meetings with Honda the earlier experience was put to good
effect in yielding focussed data from the start. This was further enhanced by the
hierarchical position of the respondents, Mr Hiroyuki Yoshino having been President and
Chief Executive Officer from 1998 to 2003, and Mr Saburo Kobayashi being the then
current deputy director of corporate planning. The interviews tended to be organised in a
formal way with requirements for advance notice of questions, though once underway the
interviews progressed smoothly and informally, even with Mr Yoshino. The problem with
the organised approach was that interviews could only take place during the brief visits to
Japan, seven weeks in total, and executives were not always available at those times. This
was most noticeable during the Tokyo Motor Show, despite advice from the company
that this would be a convenient time to arrange meetings.

3.9 Recording and coding

The purpose of an interview record is to provide a faithful account of the interview which
is then amenable to subsequent inspection. It should not intrude into the interview but the
resultant record should be complete enough to serve the purposes of the research. Coding
is used to identify the salient points and facilitate analysis. The analysis is not, though, a
separate activity since it includes the recording and coding processes, continuing for
repeated interactions with the data.

3.9.1 Recording

The first interviews were recorded by taking notes during the course of the interview and
expanding them immediately after while the material remained fresh. There were two
problems with this approach. The first was that notes were only taken when it was
opportune, yet the most inconvenient time was while the most important points were
being made. This created a bias in the data away from the most crucial towards the

80
less important. The second problem was that material could only be revisited through the
filter of the original criteria, so it was not possible to glean additional material from that
which had been noted at the time. From then on it formed part of the modus operandii of
the data gathering to record the interviews on audio tape so that, at the very least,
repeated listening could be conducted.

The recording of data went further than this as all the interviews that were recorded, bar
one, were transcribed. The only one that was not transcribed was with Professor Yasuo
Sugiyama, who insisted in speaking in Japanese, despite having both published numerous
papers and taking the interview questions in English during what was a rather fractious
session. Fortunately, this researcher has a reasonable knowledge of the Japanese language
and a basic understanding of his answers did emerge, though they were not of a relevance
that warranted the commitment of translating and transcribing in full.

3.9.2 Coding

The list of codes was built up during the coding process in order to accommodate the
novel orientations of each interview (see Appendix 12). There were two categories of
coding, topical codes and modifier codes. The topical codes followed the topics that
structured the protocols for the interviews. Although the earlier interviews were
structured chronologically the data was coded according to the topical categories of the
later fieldwork. Thus, there were seven original topics and corresponding codes:

Brand - B
Economies of scale - ES
Globalisation - GL
Government - GV
Independence - 1
Joint ventures - JV
Market - MK
National industry - NX

81
Since the interviewees were free to introduce topics they felt were of importance, four
more topical codes were introduced:

• Company Structure - horizontal or vertical integration (HI or VI),

independent (I), multinational (MNE)
• Culture - C
• Environment - E
• Local - LC

The application of these topical codes was intentionally wide ranging. The code ES,
representing economies of scale, for example, was applied to any comparison of scale to
output in whatever function, it was not necessarily tied to quantifiable production output.
The ES code could therefore be attached to engineering output, design staffing levels and
sustainable levels of sales.

In order to refine the coding further, modifying codes were used. A typical example
would be a country code, so when UK is attached to NI, the code for issues relating to
the national industry of the UK is arrived at. Similarly with the government topic, GV,
which could be modified by the code for representation (RP) and thereby show how an
institution might represent the interests of the industry. If greater precision was required
then the country modifier could be attached also, resulting in, for example, GVRPJ,
which could signify the Japanese government representation through the offices of
METI.

The distinction between topical and modifier codes was not inflexible, it was quite
possible for a topic to become a modifier. This tended not to happen with the main topics,
such as GL, NI and MK but ES often crossed over. From time to time the coding GLES
appeared, signifying the coupling of globalisation and economies of scale. An example of
this would be Toyota’s search for production capacity outside Japan.

For further emphasis, on occasion intensifier codes were used. This was represented by a
plus or minus sign (+/-), and the query sign (?). The purpose of these was to

82
underline the particularly positive or negative mood of the interviewer. The plus/minus
signs were most commonly used in attachment to the independence code (I), denoting
perhaps a fierce defence of a company’s independence or a belief that it was not
sustainable. The query symbol was much rarer, it being used to denote a comment by the
interviewee that was of equivocal validity. For example, Malcolm Harbour asserted that
Jaguar had been courageous in selecting aluminium for the body of the new large car and
so exemplified a degree of independence in body design: considering the level of
investment involved it was equally likely that it was a decision made by parent company
Ford to experiment with the novel material.

The coding of the data is a link between collection and analysis since it comprises the
second iteration of the data after the initial gathering stage. Data analysis takes this
further using the codes to organise the material into a coherent argument synthesised
from the various interviews. As the following chapters of this thesis have been arranged
around specific topics, the relevant data has been allocated to those chapters to which it
applies. Thus, the theoretical discussions in the next chapter, concerning economies of
scale, are followed by data gathered from the literature and the empirical research.

3.10 Conclusion

This chapter has taken the theoretical considerations on methodology presented in

Chapter 2 and applied them to the practical design of this research programme. It detailed
the study of automobile companies and industry institutions as the units of analysis,
resulting in a multiple case embedded design with the main case study company, MG
Rover, being compared and contrasted with the companies and institutions that make up
the contextual environment. It detailed the reasons for selecting the case study company,
MG Rover, as well as the role of the contextual companies and institutions in
triangulating the data. Semi-structured elite interviews were the main tool of empirical
research in order to provide a contemporary context for the archive data found in the
literature. The chapter charted the progress of the research, outlining the problems and
the strategies for overcoming them. The

83
recording and coding of the data described in the chapter will be put to use in
subsequent chapters according to subject matter.

84
 4: Considerations of Scale

In the course of defining the generic structure of a mass market automobile

manufacturer it is necessary to establish the forces that form the company. These give
the firm the structure and size of its operations. The organisational structure of the firm
is defined by the degree of vertical integration, involving the decision whether to source
production extemally or produce intemally, known as the make-or-buy decision. The
size of operations is defined by the available economies of scale which encourage the
firm to adjust output to a point where costs per unit are minimised.

Different industries will have widely variable integration and scale considerations.
There can also be variety within industries as different firms utilise resources in their
own unique ways. However, if an industry has a defining technology then all the firms
operating that system will be drawn into the same set of economic conditions. The firms
will then converge on a generic firm structure.

This chapter of the thesis will illustrate the underlying economic considerations of scale
and then demonstrate how they impinge on the automobile industry. As a multiproduct
production process, including major components such as engines, transmissions and
steel load-bearing bodies, automobile manufacturing comprises a complex array of
industrial activities. Each activity has its own optimum level and these need to be
matched in order that the total production process should exploit the available
economies of scale. Beyond the production processes there are also economies of scope
which bring additional advantages to a firm when adding new products to its range
thereby utilising existing company facilities more efficiently. Although economies of
scope are not derived from the production technology, and therefore do not determine
the economies of scale in the industry or the ideal size of firms, by raising overall
efficiency they can engender greater sustainability.

85
4.1 Economies of size

Large firms enjoy a number of benefits over smaller rivals, Pratten (1971) pointing to
the power that comes with market dominance. A firm might find it has some monopoly
power in setting market prices or extracting lower prices from its suppliers. A larger firm
might have political power, able to influence legislation and tax revenue policies in its
favour. Intemally, larger firms also benefit from team production where members of the
team can specialise in tasks (Parkin et al., 1997).

Two further advantages accrue to a large company in connection with its production
capability: economies of scale and economies of scope. Economies of scale are related to
technical efficiencies and industries that experience its effects find that unit costs of
production decrease as output is increased. It is important for planning purposes to note,
however, that beyond a certain point increased size of plant may be less efficient; they
are then said to be experiencing diseconomies of scale. This consideration of optimal
size is a core theme of this thesis since economies of scale indicate the ideal plant size
for a firm, and thereby the ideal size for the firm as a whole.

Economies of scope exist when a large firm gains efficiency by more intensive use of its
facilities. Although economies of scope are not prescriptive in structuring the firm, and
therefore not core to this thesis, they do indicate where a firm might improve its
sustainability. For example, the cost of selling an additional product through a shared
distribution system might be less than for a rival selling the same product through a
dedicated distribution system. The product itself can also benefit by sharing components
or even the specialist knowledge in research and development (R&D).

86
4.1.1 Economies of scale

Caimcross (1966, p. 106) describes two different areas where scale may be found:
external economies and intemal economies. Extemal economies occur when firms have
grown large enough that the industry can support its own infrastructure made up of, for
example, specialist support and trained labour (Sloman, 2001, p.95). Caimcross (1966,
p. 107) also includes economies of concentration, when firms congregate in a single
location, a tendency recently promoted by the growth of just-in-time (JIT) methods
(Womack et al., 1990). Economies of concentration, then, inhibit geographic spread and
promote vertical integration by introducing cost advantages when companies are in close
proximity. The decision of a firm to vertically integrate or not will be examined in the
next chapter of this thesis. The converse is economies of disintegration, the breaking
down of the production process into specialist functions that are better served by
separate firms or even industries. This does not necessarily promote global spread but it
does enable it and suggests opportunities for divisibility within the production
transformation process as a whole. Divisibility of production will be covered in Chapter
7 of this thesis with specific reference to vertical joint ventures.

Intemal economies of scale, on the other hand, are specific to the firm and Caimcross
(1966, p. 108) lists five of them:

• Financial economies - the ability to raise capital more cheaply, with increasing
retums to scale so that as the firm expands these advantages show a
proportionally larger increase.
• Marketing economies - increasing retums to scale through a mix of growing
monopoly powers and increasing use of sales capacity.
• Risk bearing economies - reducing exposure to risk by expanding with
increasing retums to scale, though a firm can stmcture itself to similar effect by
diversifying output, markets and supply.
• Managerial economies - manifested through delegation of detail or functional
specialisation. Delegation only comes with size, management having its own

87
 indivisibility, yet beyond a certain size there may result an unwieldy bureaucracy
and diseconomies of scale.
• Technical economies of scale - based on production technology.

Technical economies of scale particularly relate to plant size, though the manner in
which the technology is used can still distinguish one firm from another. Caimcross
(1966, p. 109) puts forward three ways in which a firm may utilise the technology to its
own advantage, resulting in economies of superior technique, increased dimensions and
linked processes. Dimensional advantages are clear in, say, a blast fiimace where a
doubling of its dimensions will increase its potential intemal capacity eight times. In
contrast, there are instances when there are clear limits to dimensional advantages if they
are held back by the nature of the product, such as fumiture manufacturing, where there
are both constraints on the natural size of wood supplied and the humanly suitable
dimensions of the final product.

There is also a lower limit to what might be considered to be a production system.

Caimcross (1966, p. 106) states that there is a qualitative difference between the lone
worker and the mechanised production system. To work on a variety of tasks the lone
worker must be a generalist, able to vary the intensity and allocation of work but is not
able to delegate functions. At larger volumes a production system must be put in place
by an organisation, with specialisation in tasks by workers and machines. This is
particularly characterised by indivisibilities, the inability to break a task down into
smaller units. A specifically sized plant comprising such a system can produce for a
range of outputs but average unit production costs will be at a minimum for only
specified levels of output. These levels may represent a comparatively narrow band of
output compared to the full range that the system can physically operate at, or the
minimum may even be denoted by a single level of output. It is at this point that the
specific plant is being operated at the efficient level and this denotes the optimum output
level for the plant, i.e. the level of output that generates the minimum short run average
cost. Either side of this point average costs are higher, either because the plant is
underemployed or because it is being overworked and diminishing retums to the

88
various factors set in. The firm may accept this if it achieves the planned output, but it
should be aware that average costs are not at their minimum.

This is not the case in the long-term. Here, the company can plan for a plant that
provides the level of output considered appropriate for the demand it believes exists for
the product. It is assumed that the larger plant will offer lower average costs otherwise
the smaller plants could simply be multiplied. It is also assumed that as the scale of the
plant increases, drawing in greater factor resources, the relative factor prices will not be
affected (Pratten, 1971). Each plant has its own short-run average cost curve but for
planning purposes, when all possible plant cost curves are plotted together, then the
long-run average cost curve traces a line around all of them, often being referred to as
the envelope curve. Indeed, terming it the long-run average cost curve is misleading
since it is not intended to show how costs change over a period of time and Pratten
(1971) prefers to label it the scale curve. Nevertheless, long-run average cost (LAC)
curve is the generally accepted terminology. The optimum plant size is the one whose
minimum short-run average cost point coincides with the minimum for the long-run
average cost curve and is thereby able to exploit all the available economies of scale.
This is known as the minimum efficient plant size, or MBPS.

Although economies of scale are available for almost any process, the search for the
technical optimum, the output level either side of which would incur cost disadvantages,
has received most attention because it renders itself amenable to quantifiable analysis. If
the technical economies of scale are very large they become a dominant feature of the
firm and give the impression that this is the source of the firm’s competitive advantage.
However, this may mask a lack of economies of scale elsewhere, even diseconomies,
particularly in a multi-plant firm.

89
4.1.2 Economies of scope

Economies of scope are available when the production system is established and a more
intensive approach is made to its usage. To achieve economies of scope the minimised
costs of production for two goods would have to be less when the system is shared than
if the goods were produced separately (Panzar and Willig, 1981). Typical examples
include the production of mutton and wool, or a meat packing firm using its refrigerators
to also store ice cream. However, for the increase in scope to bring economies there
must be spare capacity available or, as Clarke (1987, p. 113) describes it, the inputs
should be used without “complete congestion”. Such inputs are quasi public in that they
have characteristics in common, as opposed to private inputs which are mutually
exclusive. If the production of one good squeezes out the production of another to the
detriment of the cost structure then economies of scope are not being exploited. Equally,
shared production does not mean that economies of scope exist since it might be a case
of filling otherwise redundant production capacity. This may even lead to diseconomies
of scope but this is acceptable as long as the access to economies of scale compensates
for it. Altematively, exploiting economies of scope may not simply utilise existing scale
but lead to its increase, as when an R&D function is expanded to allow the engineering
teams to take on a wider range of projects. This might result in diseconomies of scale in
R&D which are compensated for by economic advantages elsewhere, such as economies
of scale in production or economies of scope in the product range.

Where economies of scope exist they can have integration implications for the firm
(Panzar and Willig, 1981). Multi-product firms that are supplying goods which have
quasi-public inputs have a motivation to integrate in order to access the economies of
scope. Credit unions, for example, enjoy economies of scope in supplying a range of
financial products and so tend to integrate them (Youn Kim, 1986). However, there is a
degree of commonality between financial products that may not be so apparent in the
manufacturing industry. The computer industry can be categorised into software and
hardware production yet the integration of these functions into unified firms is rare.

90
Apple Inc. is one of the few examples of integration but this is less to do with
economies of scope and more to do with the unique features of its Mac OS X operating
system. As Clarke (1987) points out, specialised processes are integrated to reduce
transaction costs rather than to exploit any apparent economies of scope.

Hagel and Singer (1999) looked at how a firm might be divisible according to the
different operations. The unrelated operations have differing economic considerations:
for example, customer management is concemed with scope whereas production is
concemed with scale. This can create a conflict between the operations which is
resolved through compromise:

“To protect its manufacturing scale, for example, a company may prohibit its
salespeople from selling another company's products, thus limiting their ability to
achieve economies of scope.” (Hagel and Singer, 1999, p. 136)

Applied to a multi-product firm with separate functions the emphasis on either scope or
scale may be delineated. To take, as an example, fumiture production: the two
manufacturing processes for the wooden frame and the soft coverings are mutually
exclusive but the designs are sourced from a common R&D function. The design
function has the capability to work on many different styles of fumiture, from sofas to
sideboards, the knowledge in one being relevant to another and thus exploiting
economies of scope. For the two manufacturing functions there is little in the way of
quasi-public inputs except, perhaps, in being conducted within a single building, and so
the main consideration is with scale. The firm’s priority, then, is to synchronise the
output of the two manufacturing functions in order to achieve optimal output for both.

Scale does not, by itself, compel intemal integration of the two fumiture manufacturing
functions since scale is derived from the production technologies and these are
technically unrelated in this example. There may be economies of scope in management
which could have some influence over intemalising different functions. However
management is not specific to production systems so intemalising functions for this

91
reason would result in different fumiture firms intemalising different functions
according to their particular circumstances. As a firm expands its scope it can then suffer
control losses and confound strategic decision making which might be addressed using a
multidivisional M-form of organisational stmcture (Armour and Teece, 1978).

Clarke (1987) cautions against viewing a firm as if it were one stmctured by economies
of scope for two reasons. First is that shareable inputs will compel vertical coordination
without the necessity for full vertical integration. Second is that diversifying in order to
fill existing capacity comes within the definition of economies of scale. Although Clarke
concedes that the concept of economies of scope can include all resource-utilisation
arguments this can obscure the difference in perspective of the respective scale and
stmcture approaches. To retain a clear distinction between the two approaches this thesis
will discuss them as separate entities.

Retuming to the example of the fumiture industry, at the core of the firm are the
manufacturing functions and integration will occur if it is found to be more cost
effective to integrate the functions rather than make market transactions regulated by
contract. The same principle applies to the intemalisation of the R&D function and so all
fumiture firms, given the same economic environment, will tend towards the same
fundamental organisational stmcture of functions. Economies of scope in management
can therefore be considered constant, notwithstanding variation in their actual capability.
For the stmcture of a multiproduct firm, with technically unrelated functions, integration
is concemed with transaction costs and this will be discussed further in Chapter 5.
Chapter 4 will continue by looking at the influence of economies of scale on plant and
overall firm size.

92
4.2 Production costs

The economic fundamentals upon which all companies are founded are based on the
role of production costs. At the simplest, the cost of production is a multivariable
function taking in all the sources of production costs. Koutsoyiannis (1979) expresses
this as follows:

C = fn(X,T,Pf) where C = total cost

X = output
T = technology
Pf = prices of factors of production

This can be further simplified if it is accepted that costs are only a function of output as
long as other cost factors are constant. If this were represented graphically, then changes
in costs due to output would be found by reading along the cost curve, whereas changes
in the hitherto constant factors would be shown by a complete shift of the cost curve.
Technology, for example, includes notions of entrepreneurial and technical innovation
resulting in a change in the type or method of production, therefore resulting in a shift in
the cost curve.

In the long-run all costs may be varied and so they inform the management as to how to
plan for the future. The costs are calculated ex ante and are used to consider the optimal
production system for an intended level of output. Long-run cost curves include the
firm’s intemal economies of scale but extemal economies of scale are beyond its control
and so changes there, inducing changes in factor prices, will bring about shifts in the
long-mn cost curve. In the short-mn, however, the production system has been selected.
The management is now faced with the fixed costs of the production system in place,
comprising the capital equipment which cannot be changed in the short-run, and the
variable costs over which management does have operational control.

93
4.2.1 Short-run costs

Total costs (TC) of production in the short-run are assumed to be either fixed costs (FC)
or variable costs (VC) (Sloman, 2001, p.8 6 ). VC are costs that vary with output volume
in the short-run, taking in supply materials, direct labour and fixed capital running costs.
FC are taken to be those costs that are not amenable to being varied in the short-run,
such as depreciation, fixed maintenance schedules and administrative salaries, even if
they are variable in the long-term. TC is therefore described by the formula:

FC+VC=TC

Initially, variable factors of production exhibit an increase in productivity as more are

employed, as a result of increases in marginal product, resulting from increasing retums
to the variable factor. Beyond a certain level of employment, however, inefficiencies
develop and diminishing retums to the variable factor set in. VC can therefore be
represented by an S-shaped curve (see Figure 4.1), while FC can be represented by a
straight line parallel to the output axis, the two representations being summed to arrive
at the TC curve.

Figure 4.1 Fixed cost, variable cost and total cost in the short-run

C TC

VC

FC

X
Source: Koutsoyiannis (1979)

94
The S-shaped curve for VC translates into a U-shaped short-run average variable cost
(SAVC) curve (see Figure 4.2). The initial increase in productivity is expressed as a fall
in cost per output or average variable cost, until a minimum is reached at the trough in
the U shape at output Xi. From there, diminishing retums are experienced and so the
SAVC curve shows a positive gradient. Since TC is the sum of FC and VC the short-run
average total cost (SATC) is derived in the same way as SAVC and so similarly results
in a U-shaped curve. SATC, therefore, shows the costs of running the plant as a whole
for different levels of output. In studying these curves management are able to identify
the point of minimum costs per unit of production (X2 in Figure 4.2).

Figure 4.2 Plant cost in the short-run

SMC
C
SATC

SAVC

Pi

X, X
Source: Koutsoyiannis (1979)

The addition of average fixed costs (AFC) to SAVC has the effect of shifting the SATC
curve up and to the right of SAVC. Close to the origin, FC has a large impact on TC, so
SAVC and SATC are some distance apart. As output increases due to the greater use of
variable inputs, AFC falls and the two curves converge. The relationship of the two
curves to short-run marginal cost (SMC) is such that SMC intersects SAVC and SATC at
their lowest points, X] and X2 respectively. The point X2 shows where the firm is
operating at its optimum for the whole plant, i.e. the minimum of SATC. Parkin et al.
(1997, p.265) describe how the intersection of SMC and SATC represents a point of

95
equilibrium where the firm is capable of breaking even, or eaming “normal profit” as
the authors refer to it. In terms of revenue, below Pi variable costs are not being covered
so there is no rational reason for production. For any unit price above Pi and output Xi
the variable costs are covered and a contribution made to the fixed costs. At price P 2 and
output X2 all the costs, both fixed and variable, are covered and the plant earns “normal
profit”. On the SATC curve any output above or below the optimum denotes higher
average costs. In the long-run this may put the company at a cost disadvantage in
comparison to competitors so it will need to plan a resizing of the plant. This is covered
by the next section dealing with long-run costs.

4.2.2 Long-run costs

As has been noted previously, in the long-run all factors of production are variable and
this enables the planning of facilities for production in the future as the firm can look at
different plant sizes. According to Pratten (1971, p.4) the question facing the firm is as
follows:

“What would the effect of scale be on the average costs of production of a series of
altemative plants built at a point in time, each perfectly adapted to the required scale and
operated at that scale?”

The firm is able to consider different plant sizes, each with its own distinctive
production systems, based on the current technical knowledge. Rosegger (1986, p.72)
cautions that the long-run average cost curve (LAC) does not describe costs in the long-
run but reflects the cost structure of a long-term commitment and is therefore a planning
aid. The plants are theoretical at this stage but can be realistically costed as part of the
planning process. Figure 4.3 shows the SATC for three plant sizes: SATCi, SATC 2 and
SATC3 .

96
 Figure 4.3 Altemative plant sizes for long-run planning
SATC
C
SATQ

C,

LAC

X, X
Source: Koutsoyiannis (1979)

The LAC envelopes the SATC curves for the three plants under consideration. It is
drawn as a smooth line on the assumption that there are an infinite variety of plants, not
just the three shown. If this were not so and there were only three available plant sizes
then LAC would be scalloped as it traced around the outside of the three plant curves,
from one curve intersection to the next (Thompson and Formby, 1993, p.224). The
company would then select the plant size that was best suited to the appropriate level of
output in the long-mn. If output Xi is planned for then it is clear that the firm would
select system SATCi with average cost Ci. At output X2 the firm would need to decide if
the new output was for the short or long-run. It therefore has a choice between staying
with SATCi and accepting that average costs will rise in the short-run to C4, or taking the
long-term view and reconstructing around plant SATC2 and benefiting from the lower
average costs, C2 , in the long-run. The crucial decision depends on the future long-run
expectations of the firm. Taking the long view, the company uses LAC to select the plant
which, in the long-mn, will produce the company’s planned level of output at the
minimum unit cost, but can, in the short-run, cope with temporary increases or decreases
in output.

At output X3 the company would be aware that plant SATC3 would not only be most
suitable for the desired output but would also give access to all the economies of scale

97
available. Up to this point increases in plant size would enjoy increasing retums to scale
but beyond it there would be decreasing retums resulting in diseconomies of scale. The
minimum point for SATC3 coincides with the minimum of LAC to denote optimum
scale in both the short-run and the long-run. LAC is also in tangential contact with
SATCi and SATC2 , but not at the minimum points for each curve. Assuming that the
production systems are inflexible then for SATCi and SATC 2 the contact is on the
negative slope of the SATC curves, revealing increasing retums to scale. In Figure 4.3
this is shown by output X2 which is achievable, either by under-utilising plant size
SATC2 , or over-utilising plant size SATCi, though as already noted, plant size SATC 2
gives rise to a lower unit cost (Thompson and Formby, 1993, p.225). Those plants that
are larger than SATC3 are assumed to suffer diseconomies of scale, perhaps due to the
managerial diseconomies described by Caimcross (1966).

The development of long-run costs can also be expressed incrementally by investigating

marginal costs (MC). MC in the long-run (LMC) is not an envelope curve since it
connects the points on the SMC curves that signify the output where the LAC-SATC
curves are in tangential contact. This is clarified by Figure 4.4 showing SMC and LMC
intersections at outputs Xa and Xm, omitting SATC2 for clarity.

98
 Figure 4.4 Marginal cost in the long and short-run

LMC
SATC
C SMC

LAC

X
Source: Koutsoyiannis (1979)

Moreover, Xm marks the intersection of four curves in total such that:

SATC3 = SMC3 = LAC = LMC

The firm will choose whichever short-run cost curve gives it the lowest per unit cost in
relation to desired output. Xm, though, brings further advantages because it is the point
at which the lowest absolute per unit cost is achieved in the long-run, assuming that the
firm can justify the level of production. This, then, is the optimum plant size and is the
one that can exploit all available economies of scale. The smallest size of plant, or
lowest output, that still exploits all available economies of scale is said to have the
minimum efficient scale (MES) or be the minimum efficient plant size (MBPS). The
terms MES and MBPS tend to be interchangeable in the literature, although this thesis
will use MBPS when referring to specific plant sizes and MBS for the firm as a whole.
In Figure 4.4, MBS and MBPS are represented by output Xm for a single plant firm.

In practice, it may prove difficult to define the MBPS since the planning horizon given
by the LAC is based on existing technology and a cost minimising, profit maximising

99
firm would not seek to offer itself as an experiment in diseconomies of scale. For this
reason plant sizes beyond the MBPS tend to be speculative or exploratory in character,
rendering the shape of the LAC highly uncertain after the MBPS. Pratten (1971) does
not consider the MBPS to have progressed if a newly developed plant doubles the scale
but delivers a reduction in total average unit costs of less than 5% and a reduction in
value added per unit of less than 10%. The next section of this thesis will look at how
different shapes of LAC affect the long-term plans of companies

4.2.3 Alternative long-run average cost curves

The implication of the U-shaped LAC curve is that there is one size of plant that enjoys
all the economies of scale and thereafter increasing plant sizes are weighed down by
decreasing retums to scale. In other words, the larger plants suffer from increasing unit
costs. However, the danger of this occurring may cause companies to avoid exploring
the LAC curve beyond the established MBPS and so the precise shape of the curve
remains largely conjectural (Pratten, 1971).

Thompson and Formby (1993, p.229) provide three basic shapes of curve, and these are
illustrated in Figure 4.5 below. LACa shows the early onset of decreasing retums to
scale at Xa, which is in contrast to LACb which delays their onset until much later, at
Xb. LACc describes a wide range of outputs in a saucer-shaped curve, from Xc to Xd,
over which constant retums to scale are being exploited, before decreasing retums set in.
This means that firms with plants of different sizes can operate with the same level of
average cost. This should not be taken to mean that a large plant can be run at the same
output as a smaller plant for the same average cost, but it does mean that for planning
purposes there is a wider choice of outputs with each denoting an optimum of a
particular plant’s SATC curve that is equal to that of another plant of a different size.
The smallest size of plant that is still on the flat, lowest, section of the curve is the one
that has achieved the minimum efficient plant size (MBPS) and suffers no production
cost disadvantage against any other size of plant on the flat section. This is represented
by Xc on LACc.

100
 Figure 4.5 Altemative LAC curves

LACa

LACc

LACb

Xc Xb Xd X

Source: Thompson and Formby, 1993

The altemative LAC curve shapes, LACa and LACb, can cause particular planning
problems for multiproduct firms operating a number of different production processes.
This is because the different optimum production levels for each process need to be
matched, first to each other and then to the desired final output. This is considerably
eased if a curve like LACc is present because there is a wider choice of plant sizes
available for achieving such matching.

4.2.4 Integrating different production processes

At the plant level there may be supplementary processes which are indivisible. When
integrating different processes the optimum output then becomes the optimum for all the
processes in combination with one another. Maxcy and Silberston (1959, p.76) give the
simple illustration of two machines, A and B, Machine A producing 25,000 items a year
to feed Machine B, which has a capacity for 50,000 items a year. To achieve the 50,000
items, though, would mean duplicating Machine A, and, furthermore, an output

101
of 100,000 would mean operating four of Machine A and two of B. Complete processes
are often a compound of manufacturing activities like these, for example forging
involves not only moulding the parts but also machining them, subsequently followed
by assembly. There being no common technology between the processes, synchronising
the processes can be complex. Figure 4.6 illustrates the way in which machines A and B
relate to each other.

Figure 4.6 Matching machine outputs

SATC,

SATCb
'3

0
25 50 X (OOOs)

Source: Maxcy and Silberston, 1959

Figure 4.6 shows that to achieve the minimum short-run average cost for Machine A
(SATCa) an output of 25,000 units at cost Pi is appropriate, which is shown by the cost
curve for machine B (SATCb) to mean higher than minimum costs on that machine (P 3
compared to P2 ). Machine A can be doubled up so that both are operating at their
optimum of 25,000 to match the 50,000 optimum for Machine B at cost P 2 , but if
demand slackens and an output of only 25,000 is achievable then neither Machine B nor
the two Machines A are running at their optimum. At this level, the costs for two
Machines A are higher than if the company had planned for the lower level and installed
only the one Machine A. The principle can be extended to entire processes contained
within one plant. The optimum for a company is dependent upon the lowest

102
common multiple of the processes that make up the firm’s activities (Rhys, 1972,
p.289).

The consolidated figure for all the different outputs would be the optimum for a single
production transformation process contained within a plant. Firms do not necessarily
structure themselves around this one process, though, and indeed may encompass
technically separate plants providing the main components for assembly into a final,
complex product. At the planning stage, the multi-plant company would need to be
aware that the short-run cost curves for all the plants would need to be synchronised
according to the same principles as for individual machines within one process.
Moreover, the company would need to also refer to the relevant long-run cost curves for
each plant to arrive at a long-run structure for the company as a whole.

4.2.5 Cost curves for the multi-plant firm

The multi-plant firm is one that manufactures items using technically distinct processes.
If the product is kept constant while the plant sizes are varied, which might happen if
the plants are of different ages, then Thompson and Formby (1993, p.396) recommend
three steps to bring the plants into the most technically efficient alignment:

• Step 1 - calculate firms MC from the various plant MCs.

• Step 2 - calculate most profitable output where overall MC equals overall
marginal revenue (MR).
• Step 3 - allocate profit maximising output among the plants so the MCs of the
last products at each plant are equal.

This can be expressed by the formula:

M R = M C = M Cpi = M C p2 = .. .MCpn

where PI and P2 are various plant sizes up to plant Pn.

103
In the case of a multi-plant firm manufacturing component parts for a single final
product then the plants differ not simply in size but in the fundamental production
technology. Synchronising the output levels so that all are operating at their optimum is
eased if they have “reserve capacity” (Koutsoyiannis, 1979). This is illustrated by a flat
base to the short-run cost curve, a saucer-shaped curve (similar to LACc in Figure 4.5
above), and though plants may still have to be multiplied along the same principles as
for matching the output of different machines (Maxcy and Silberston, 1959) at least the
range of optimum values facilitates this process.

At the planning stage, the long-run costs under consideration by the firm are similarly
eased if there is a range of optimum values exhibited by the different plants. If plant A
has a very sharp, U-shaped LAC then the alignment with plant B’s LAC is more easily
conducted if that for B has a wide optimum range. This is distinct from the SAC curve:
a saucer-shaped LAC curve means that different sizes of plants can all be operating at
the same minimum cost. The firm is not, then, tied to one size of plant for this particular
process. While the company might establish one plant A and one plant B, from a
planning perspective it can look to expand with varying sizes of plant B without
encountering diseconomies of scale. Figure 4.7 below shows this for the two plants, A
and B, where LA Ca offers little in the way of altemative plant sizes while LAC b has a
wide choice of plant sizes. However, if LACe is still flat where the output of LACa is
doubled, shown by point 2X\, for two plants of A producing at output Xi to match the
output of just one plant of B producing at output 2X\, plant B would result in the lowest
unit cost. This introduces an additional degree of flexibility to long-term planning.

104
 Figure 4.7 LAC for two plants

LACb

X, 2X, X

The ability to combine plants effectively, such that each is operating at minimum cost, is
particularly relevant to industries involving multiple products. The plants are likely to
have none or few technical commonalities and so the only common link is in final
assembly. If each manufacturing process comes under the remit of a single company,
then it is the company’s responsibility to align the plant outputs with each other in order
that an overall MES for the company might be achieved. Finding the precise output
quantity of output necessary for MES can be difficult to calculate but it may be inferred
when certain sizes of firms have shown significant longevity. The method for doing this
is discussed in the next subsection concerning the analysis of firm survival.

105
4.2.6 Survivor analysis

As Pratten (1971) points out, quantifying economies of scale is fraught with difficulty
since it is not possible to inflict experimental research techniques on commercial
operations. Neither is it possible to simply assume that the firms that are ranked the
highest in terms of output have the greater access to economies of scale since ranks may
change without revealing the variation in output quantity that brought about the new
rank (Hymer and Pashigian, 1962). Instead, Rogers (1993) suggests that the three
possible approaches to determining the relationship between plant size and production
cost are econometric, engineering and survivor analysis. The econometric cost curve
estimation appears highly quantifiable but suffers from difficulties in measuring certain
costs, such as normal profit and cost of capital. The engineering approach looks at the
costs of the inputs but the cost structure can be skewed by the relative availability of
particular factor inputs, thus precluding direct comparisons between plants in different
locations. The approach can explain where efficiencies have occurred but it is not
predictive in showing which plant will be the most efficient. In any case, such definition
of costs may not be necessary, Freeh and Ginsberg (1974) stating that it is more
important to be able to understand that those involved in the activity will strive to
maximise some utility fimction and will therefore alter the size of their organisation
should they observe one that is better able to maximise it. As Altman (1999) asserts, it is
only necessary that a firm behaves in a manner that is consistent with its survival.
George Stigler (1958, p.54) criticised the econometric and engineering approaches for
failing to take into account the effectiveness of the inputs:

“...as if one were trying to measure the nutritive value of goods without knowing
whether the consumers who ate them continued to live.”

He proposed an altemative, the survivor principle, which took a retrospective,

longitudinal view. This determined economies of scale based on the Darwinian
assumption that the companies that out-lasted the competition must enjoy the lowest
costs, or in more general terms, an advantage that was contingent upon the size of their

106
output. The theory proposed that the most efficient company size would take the largest
share of the market, and by extension the most efficient company would operate with
the most efficient plant size. Stigler used this technique to investigate the existence of
economies of scale in the US steel industry, each plant size representing a short-run cost
curve tracing out part of the long-run cost curve for the industry. The smallest and
largest plants lost market share over the two decades of the study, suggesting that they
operated in those parts of the LAC curve that brought decreasing retums to scale or
suffered diseconomies of scale. Medium sized plants tended to survive, suggesting that
they operated in the extended flat section at the bottom of the curve.

Stigler’s study showed that the survivor approach observes firms converging on an
optimal size MES, or MBPS in the case of a single plant firm, during which time they
take an increasing share of the market while their smaller rivals leave the industry or are
consolidated into larger groups. This implies that firms that have been in the industry for
longer will hold an advantage since they have had more time in which to reach the most
efficient size. Dunne and Hughes (1994) tested this amongst British companies and
found that the smaller, younger companies showed the greater tendency to grow faster.
This suggests that as firms converge on the optimal plant size the rate of increasing
retums to scale decelerates, implying a flattening to the LAC curve where growth
becomes progressively less important.

Klepper (1996) draws parallels with the product life cycle theories which argue that
once a new product has reached maturity it is the manufacturing process that comes to
prominence. However, this suggests the emergence of a dominant design for the
product, which Klepper considers to be an imprecise concept particularly where market
demand is diverse. Nevertheless, Klepper (2002) found that the larger incumbents held
the advantage because the R&D costs of innovation could be allocated over a larger
output. Over time Klepper states that R&D can be costlessly imitated, and it might be
suggested that this would be the point at which small firms could enter with commonly
held technology and so expand rapidly. The greater advantage, though, lies with entrant
firms that are in some way related to the incumbents, perhaps comprising staff

107
from those firms or being diversifications from related industries (Klepper, 2002b). The
Ford Motor Company, for example, was the third firm to be founded by Henry Ford and
the remnants of the company immediately preceding it even metamorphosed into the
Cadillac company under a different management.

The evolution of a firm is therefore better understood if the industry is taken as the unit
of analysis, rather than the product, indicating an industry life cycle. Cantner et al
(2006) found that the firm specific factors, such as capital and initial endowments, were
major determinants of a new entrant’s ability to survive in the long term. Although
intensity of competition reduced survivability rates as growth in output was directly
reflected in market share, if the market was expanding then this slowed the exit rates
from the industry as output could increase, or stabilise, even as market share fell.

Although survivor analysis is useful where market shares are clearly growing in a static
market, Koutsoyiannis (1979) points out that this intuitively attractive technique is
founded on assumptions that are not always realistic, principally that firms operate
under the same set of conditions, and Rogers (1993) points specifically to the different
uses made of capital and labour. Stigler (1958, p.57) was not unaware of this:

“Since various firms employ different kinds or qualities of resources, there will tend to
develop a frequency of distribution of optimum firm sizes.”

Although this suggests that survivor analysis can only result in general indications of
optimal output size there is the additional problem of technical advances, which Pratten
(1971) shows will result in plants that have different short-run cost curves. Shepherd
(1967) even suggested that industry entrants may employ smaller plants simply because
new technology permits them to be operated efficiently at lower outputs. Over an
extended period of time the LAC curve may become more an historical record of the
industry diversity rather than a scale curve to aid in production planning. It could be
argued that new production techniques are firm-specific and because they are not
available for planning by rivals they are not, therefore, part of the industry LAC.

108
Makadok (1999) found some basis for this but the effect diminished after a period of
around eight years as rivals leamt to imitate each other, referred to by Klepper (2002) as
costless imitation.

Taking this point further, in a globalised industry where the technical factors of
production are available to all participants, the principles underlying plants with firm-
specific advantages will soon become disseminated throughout the industry. This will
have the effect of equalising technical resources, leaving local labour resources as one
that might provide an advantage that can only be imitated by moving to the same
location, assuming that the labour resources are effectively unlimited. In such an
industry, the short-run cost curves that comprise the industry LAC might, in practice,
imply moving plants to different locations to exploit the marginal cost advantages.

Freeh and Mobley (1995) noted that a plant may operate in an environment where it
receives assistance from its government in some way, and it could be suggested that this
would include a requirement for incoming foreign direct investment (FDI) to engage in
joint ventures with local firms. The motivation for government policies of this kind is
that joint ventures are an effective way of transferring the advantages of the established
firms to the smaller entrants. However, if the industry is global then it is likely that
government support would only maintain smaller entrants at the local level while the
multinationals continue to enjoy a dominant share of the global market. Lueng et al
(2003) found that large foreign banks were compromised by regulations in Shanghai
that favoured domestic rivals, which maintained diversity and suppressed dominant
market share in the city. However, in terms of total market share, which in an
intemational environment would be global, survivor analysis would still have revealed
the overall advantage that the large foreign banks held. This may also explain the
finding by Gorg and Strobl (2003) that multinationals are quicker to leave local markets
than their domestic rivals when the commercial environment becomes more challenging
since they have other locations within the global market. This is not an option for the
government assisted local companies which only enjoy such assistance in their domestic
environment.

109
The degree of global market share that would signify the exploitation of economies of
scale, though, is difficult to determine. Shepherd (1967) cautions that the technique is
descriptive, rather than normative, and predictions based on the measured trends
therefore reveal only the current minimum efficient scale (MES), not necessarily the
technical optimum. Growth of market share is therefore observed to be converging on
an optimum that can only be inferred, the state of the industry being an indication of the
current MES for the firm, and by extension, MEPS for the plants that comprise it.
Indeed, Shepherd suggests that MEPS is the more reliable measure since the firm as a
whole, particularly one of a multinational nature, is subject to a multitude of complex
factors. It might be argued, for example, that in the study of foreign banks in Shanghai
that local market distortions mean that output there should not be included in the foreign
banks’ global market share (Lueng et al, 2003). So, while the total output of the banking
firm with the largest market share may not necessarily be exploiting all the available
economies of scale in the industry, it is quite valid to assume that its largest branch in a
market less encumbered by local distortion is operating at MEPS.

On this basis, the more that an industry is globally spread the more reliable the survivor
analysis since the various local anomalies will tend to equalise. Furthermore, the
survival trends will be more reliable if they are well established over time based on a
shared production technology, one where progressive developments are refinements of
the existing system rather than a distinct new system. This particularly applies to the
automobile industry as long as it is narrowly confined to the mass market passenger car
manufacturers using the unitary all-steel welded body for their products. There are other
technologies available, but it is this load-bearing body that has become the dominant
feature of the volume industry, defining both the product and the production system.
Since this foundation for the industry became established in the 1920s it can be assumed
that the high entry costs mean that plants are only constructed after rigorous economic
evaluations based on long experience have taken place. On this basis it would be
unnecessary for a research programme to investigate the minutiae of engineering or
econometric costs, the evolution of the industry for the preceding decades being
sufficient to establish reliable survival trends. The following sections of this thesis will

110
examine the market shares of the dominant manufacturers to derive the MES for the
industry and then the MBPS for the plants used in the automobile manufacturing
processes.

4.3 Econoniies of scale in the automobile industry

Recent discussion of productivity in the automobile industry has been focussed on the
rise of flexible manufacturing, even suggesting that this phenomenon now takes
prominence in defining the costs of production. However, flexible manufacturing does
not redefine the production technology but is instead a method for more fully utilising
it. Truett and Truett (2001) assert that the presence of economies of scale in the
automobile industry continues to be a significant factor for companies, recommending
that the Spanish industry, for example, would find cost advantages in expanding. The
expansion would not necessarily be uniform across all the production processes since
each process can have a distinct MBPS. There are four processes that comprise the
production of automobiles. Table 4.1 shows the main processes (as described by Rhys,
1972; Wells and Rawlinson, 1994; Wells and Nieuwenhuis, 2001) summarised into four
major fimctions:

111
 Table 4.1 Four major automobile functions

Major Function Processes

Design (R&D) Research and development (R&D)
Engineering
Pressing (BIW) Press shop
Powertrain Forging and machining
Foundry
Powertrain assembly
Final Assembly BIW
PaintShop
Sub-assembly
Final assembly

Sources: Rhys 1972

Wells and Rawlinson 1994
Wells and Nieuwenhuis 2001

Pressing and powertrain production use specialised manufacturing processes as dictated

by the nature of the product and so are capital intensive. Pressing is the most specific to
the automobile industry because it stamps out panels from sheets of steel which are
subsequently welded together to form the load bearing structural bodyshell, known in its
pre-assembly unpainted state as the body-in-white (BIW). Assembly is a logical
sequence of activities that has developed over time and tends to be more labour
intensive. Vehicle design is a knowledge-based activity encompassing all the research
and development (R&D) necessary to bring a new model to the point of production.
R&D is dependent upon human knowledge aided by physical prototypes and,
increasingly, information technology (IT).

As distinct activities, each process has its own economies of scale. Husan (1997) states
that MES has changed as the industry has developed, notably between the years 1958
and 1975. It is tempting to agree that manufacturing has become more capital intensive
but it does not necessarily follow that technical progress leads to greater economies of
scale. It may instead lead to a flattening of the LAC curve for the plant, allowing MES
to be gained at lower output levels than previously, particularly with the rise of flexible

112
manufacturing. Table 4.2 illustrates this by comparing estimates provided by Rhys
(1972; 1999). The figures presented seem to contradict Husan’s view that MES
increases over time, at least with respect to pressing and powertrain, though the MBPS
for final assembly does appear to have risen.

Table 4.2 Development of MES in the automobile industry

Process 1972 MES 1999

Pressing ,000 1,000
2 ,000 ,000
,000
Powertrain ,
1 000 250,000
2 00,0 0 0
Final assembly 250,000
,000
R&D 2,000

Sources: Rhys (1972; 1999)

The figures form Rhys (1999) would suggest an MES for the industry of 2 million units
a year, with the plants being multiplied as necessary to meet this common denominator.
In the next subsection this thesis we use survivor analysis to attempt to reveal the MES
for the current automobile industry.

4.3.1 Survivor analysis of the automobile industry

This survivor analysis of the global automobile industry takes the years 1975 to 2005 as
the period for investigation, sampling the data at five year intervals. In taking a three
decade period it is ten years longer than the study conducted by Stigler (1958). The
1975 to 2005 period commences with the coming to global prominence of the Japanese
automobile industry, the later rise and consolidation of the Korean industry and, lastly,
the recent emergence of the industry in China. All the production figures have been
sourced from the International Organization of Motor Vehicle Manufacturers (GIGA),
either directly (2000-2005) or indirectly (1975-1995) from Ward’s Automotive Year
Books. These published figures have been sourced, in turn, from the national

113
organisations that collect the data. The year 2006 has been omitted from the survivor
analysis since results figures less than a year old from emerging economies are often
subject to revision. Output figures from 2006 will, though, be used to illustrate the most
recent changes, particularly if they are considered to come from a reliable source.

The growth in the global production of automobiles shows no sign of slowing down,
although much of the most recent increase is due to the expansion of output in China.
Figure 4.3 shows how the top thirteen manufacturing countries in the world have
continued to grow a year on from the survivor analysis period, with all of them
producing above 1 million units a year in 2006 (unrevised figures), along with the rate
of growth over 2005.

Table 4.3 Top thirteen automobile producing nations

Country Production 2005-2006

Japan (units) Change (%)
9,756,515 +6.3%
Germany 5,398,508 +1.1
China 5,233,132 +25.9
US 4,366,220 -6 .0
South Korea 3,489,136 +3,8
France 2,723,196 -10.7
Brazil 2,092,029 +3.3
Spain 2,078,639 +0.9%
India 1,473,000 +24.2%
Canada 1,389,536 -4.3
UK 1,442,085 -8 .6
Russia 1,177,918 +11.6
Mexico 1,097,619 +22.4

Source: OICA

Although the countries that have dominated the automobile industry for the longest
period of time, such as Germany, France and the US, continue to have a substantial
presence it is Japan that now holds sway. It also continues to grow at a respectable rate.

114
 the 2006 output figure being an increase o f 17.7% over the 2000 result (JAMA, 2007b). Other
countries have shown must faster increases in output over 2005, 42.3% in the case o f the Czech
Republic, but it is China’s combination of high output and continued high rate o f growth that
indicates its possible future status as the industry leader. China even overshadows India which has
also shown remarkable increases in production.

It is because of locations like China and India that global production is rising with each successive
year. Figure 4.8 traces the change in total global production since 1975 in five year intervals. As

Cantner et al. (2006) found, the growth suggests that smaller firms will be able to maintain a
presence in the industry for longer than if the total global output were stable or declining, and it
may even encourage successive entrants to come forward.

Figure 4.8 Total global production, 1975-2005

Output (m)

50 -

40 -

30 -

20 -

10 -

0 '
1975 1980 1985 1990 1995 2000 2005

Year

Source: OICA

In order to apply survivor analysis, the total output for each year has been broken down into the
contributing output figures from the individual automobile manufacturers. The range of output
extended from as low as 29,000 units on one occasion, by Uaz of Russia, to over 6 million units by
Toyota. The individual output results were allocated to output ranges and then the production share
o f the total output by those ranges was

115
calculated. Initially, these output ranges were taken in 1 million unit groups, except for output in
the zero to 1 million units which was divided into two ranges o f 500,000 each. This was in order
to enable us to examine the movement of smaller companies at the lower limit o f the industry. For
clarity, the groups have been labelled in output ranges o f whole millions (e.g. 1 million to 2
million), though in fact the upper figure in each range should strictly be one unit below that shown.
The results for this are shown in Figure 4.9 below.

Figure 4.9 Global production share of output seven ranges

50 -
40 - •5+
4tn-5m
30 - 3m-4m
Market Share (%) •2m-3m
20 - lm-2m
10 - # - 0 .5 m - lm
—^ 0 -0 .5 m
0
1975 1980 1985 1990 1995 2000 2005
Year Source: OICA

The level of aggregation of output data shown by Figure 4.9 appears too complex to reveal a single
overall trend. On closer inspection, the two lowest output ranges, from zero to 0.5 million and 0.5
million to 1 million, reveal a relatively consistent decline. From 1975 to 1985 the two ranges
alternate, suggesting movement of manufacturers between the two. From 1990 to 2005 both ranges
are in decline in terms of production share, indicating that survival rates below 1 million units a
year are deteriorating. Given that throughout this period the majority of manufacturers in this range
are Chinese it demonstrates that the government support through enforced joint ventures with
foreign companies is not enough to sustain these companies. No Chinese manufacturer has yet
broken into the 0.5 million to 1 million range. However, this also illustrates the weakness o f
survivor analysis in a growing market where production share may decline while output is stable or
even growing. For example Chery is making excellent progress

116
and SAIC is gradually expanding its presence as an independent manufacturer. Subaru
has also consistently grown since 1975, starting the period with output of just over
100,000 units and ending it on 500,000 units. These companies may also be joined by
other illustrious companies such as Mercedes-Benz and Chrysler as their groups
deconstruct.

The other output ranges in Figure 4.9 are even less clear. The 1 million to 2 million
range is in general decline though it has wide fluctuations that are mirrored in the
opposing variations of the 2 million to 3 million range, also in an apparent decline. The
3 million to 4 million range emerges inconsistently but appears to share some
synchronisation with the 4 million to 5 million range which is even more sporadic. The
range above 5 million units a year surfaces only towards the end of the period, as might
be expected given the growth rate of the global production total, and too little time has
elapsed for this to be identified as a significant trend.

To bring greater clarity to the data the three middle output ranges were re-categorised
into two, giving 1 million to 2.5 million and 2.5 million to 4 million, and the top range
reduced to 4 million units or more. The results of the five output ranges are illustrated in
Figure 4.10 below.

117
 Figure 4.10 Global production share of five output ranges

60

50
40 4nr+^
2.5m-4m
Market Share (%) 30 ln>2.5m
20 -M —0.5m -lm

10 —^1^— 0-0.5m

1975 1980 1985 1990 1995 2000 2005

Year

Source: OICA

Figure 4.10 shows the two lowest output ranges as before but it brings out the decline in the 1
million to 2.5 million unit range more clearly. The share of global production enjoyed by those
manufacturers in the 2.5 million to 4 million range has been on an upward trend since 1980,
though there is no data for 2000 since they appear to have moved into the range above judging by
the sudden rise by the 4m or more range. This is borne out by inspecting the production shares for
individual companies for the five highest ranked companies, the industry leaders. Figure 4.11 (see
below) shows this for Toyota, GM, Ford and VAG with data for Renault and Nissan as separate
organisations until their alliance began in 1999.

118
 Figure 4.11 Production share for the industry leaders, 1975-2005

Market Share (%)

25 n
•GM
20 -
• Toyota
Ford
VAG
)K Nissan
# Renault
—h - Renault-Nissan

1975 1980 1985 1990 1995 2000 2005

Year

Source: OICA

Although Figure 4.11 appears to show a convergence in production share to around the 12% level,
excluding Ford’s falling share, extrapolating the trends would suggest that Toyota will continue to
accumulate production share at the expense of the others, intimating a future divergence o f the
trends. In addition, in an expanding market even a declining production share will translate into
output growth, and this is shown by Figure 4.12 below.

119
 Figure 4.12 Global production output for top five manufacturers, 1975-2005

Output (m)

7
6 ■GM
• Toyota
5 Ford
4 •VAG
■Nissan
3 Renault
2 • Renault-Nissan
■Average
1

1975 1980 1985 1990 1995 2000 2005

Year

Source: OICA

The five manufacturing groups shown in Figure 4.12 appear to be converging on each other but it
is not possible to estimate what the ultimate level of output might be for as long as the market
continues expanding. An average of the outputs for each year can be used to illustrate a trend line
over the period for these manufacturers and although this gives an output figure for 2005 of a little
over 5 million units the variance on either side of this is over 1 million units. Furthermore, the
gradient o f the average is clearly positive and will continue to be so as long as the market
continues to expand. According to forecasts by the automotive consultancy Global Insight, the
world production for automobiles will expand another 28% to over 60 million vehicles by 2010,
much of this due to further increases in China where production will be over 12 million vehicles
that year. If Toyota maintains the trend in production share growth shown in Figure 4.11 since 2000
then a production share of 15% in 2010 would give it a production output o f 9 million vehicles.
This is corroborated by Toyota itself which has declared as its target a 15% share o f the global car
market by 2010 (Toyota, 2007). A further difficulty is that as the market expands manufacturers
release more models.

120
Toyota, for example, lists 53 models in production in Japan alone (Toyota, 2007b),
although the number of vehicle platforms is much less.

Survivor analysis seems to show that output ranges less than 2.5 million units a year
have a poor record of longevity, either because manufacturers move into a higher range
or drop down into a lower one. Table 4.4 shows the top twenty manufacturers for 2005
and only eight manufacturers were above this supposed threshold. Of those that are
under the threshold there are some that can call on support from equity holders, such as
Suzuki (Toyota), Daihatsu (Toyota), and Mazda (Ford). This may not in itself be
relevant since equity holders seek a retum on their investment and so it would be
expected that these companies should strive to exploit economies of scale just like
independent companies. Altematively, BMW has no extemal support but is able to
practice cost recovery due to its premium brand image.

121
 Table 4.4 Top twenty manufacturers, 2005

Rank Company Output (m)

1 Toyota 6.2
2 GM 5.7
3 VAG 5.0
4 Renault 4.9
5 Ford 3.5
6 Honda 3.3
7 PSA 3.0
8 Hyundai 2.7
9 DC 2.0
10 Suzuki 1.7
II Fiat 1.5
12 BMW 1.3
13 Mazda 1.1
14 Mitsubishi 1.0
15 Daihatsu 0.8
16 AvtoVAZ 0.7
17 Fuji (Subaru) 0.5
18 Chery 0.2
19 Tata 0.2
20 SAIC 0.1

Source: OICA

Although the survivor analysis has been effective in identifying higher output ranges
comprising companies that might be considered the industry leaders, it has not
conclusively demonstrated that the remaining output ranges, those with falling output
shares, comprises manufacturers that are in danger of being eliminated from the
industry. Indeed, the growth of the largest manufacturers seems to be a fairly recent
phenomenon while the smaller companies, those below 1 million units a year, comprise
new entrants together with some of the oldest names in the industry. It should also be
noted that in 2007 this range was joined by Mercedes-Benz and Chrysler due to their
demerger. Since MES is, by definition, the minimum size that captures all the available
economies of scale, the survivor analysis seems to indicate three conclusions:

122
 1. In the very lowest output range the firms are mostly new entrants to the industry.

2. In the 0.5m to Im output range there are firms that are long-term survivors in the
industry.

3. Above output of 2.5m firms have an additional advantage that grows in

proportion to size.

To investigate MES more closely, Shepherd (1967) advised the analysis of individual
plants to uncover the MBPS first, from which the MES could be derived. This assumes
that all companies will attempt to achieve MBPS whatever their overall size, and as
they expand they will do so with plants that exhibit the same cost advantages. Thus
companies will progress towards MES in units of MBPS. The following subsections
will evaluate MBPS taken from the dominant and longest surviving companies in the
industry for each of the four processes that make up automobile manufacturing.

4.3.2 Body construction

The distinguishing feature of automobile manufacturing is the characteristic and

restrictive manner in which the steel bodies are formed. From the time the automobile
was bom in 1876 in Baden-Wuerttemberg, where Nikolaus Otto produced the first high
revving, lightweight intemal combustion engine (Maxton and Wormald 1995), until
1915, car production was a craft activity. This depended on the skills of carpenters to
constmct and fit an unstressed timber-frame body to a metal chassis. Even in 1920,
twelve years after the introduction of the Model T, car bodies were still 85% wooden.
Six years later car bodies were 70% steel.

The production revolution took place in 1915 when Dodge lifted output from 5,000 to
50,000 due to the production innovation of Edward G Budd and his Austrian partner,
Joseph Ledwinka (Maxton and Wormald, 1995). The process has been termed “the

123
Budd Paradigm” by Nieuwenhuis and Wells (1997) because it is one that distinguishes
automobile production from all other industrial activity. It encompasses the pressing of
steel sheets into body panels and subsequent welding together into complete load-
bearing car bodies which give the final vehicle its shape and practical purpose. As befits
a paradigm, ‘Buddism’ is restrictive and self-referential since it refers only to the
pressing and welding of steel bodies, it is not possible to substitute synthetic materials
such as glass reinforced plastic (GRP) and even alternative metals, such as aluminium,
can only be used with significant modifications. As a defined technology it also
predetermines the available econoniies of scale. The following subsection will
demonstrate the nature of this paradigm.

4.3.2.1 The Budd Paradigm

The paradigm has its roots in the New England clock and lock industry of 1860
(Nieuwenhuis and Wells, 1997; 2007). To make the new technology fit the automotive
industry Budd and Ledwinka had to redefine the basic elements of the automobile: the
body-in-white (BIW) was formed from the cowl, roof, side and rear panels onto which
body panels could be hung. The basic method of forming the panels has not changed
since that time: sheets of steel are inserted into high precision dies mounted on presses
and the panel shapes stamped out. Those panels that are being attached together are held
in place by a jig for welding. The resultant BIW forms the basic structure of the vehicle
and can be sent forward for painting, powertrain installation and fitting out in the final
assembly function.

The BIW process is capital intensive and highly amenable to automation. Indeed, the
costs of the huge step-up in volume that Dodge were forced to undertake eventually led
them to look to Chrysler for financial rescue (DaimlerChrysler, 2004). The fixed body
structure of the BIW placed physical limitations on the finished model, thus ending the
craft era capability of customising each model to the customer’s taste. The Budd
Paradigm is therefore central to the mass production of standardised automobiles. This
was further enhanced by another major advantage of using steel alone which was that

124
enamel paint could be used and baked at 450°F (230°C), reducing paint drying times
from between three and eight weeks to one or two hours (DaimlerChrysler, 2004).

In the early 1920s, Citroen used the same technology to raise its production from 30 or
50 per day up to 400 to 500 per day. In 1925, the technology reached the UK when
Pressed Steel was set up by Budd and William Morris. The Budd Paradigm was thus the
technological driver that reduced the price of the final product and so created the mass
market (Church, 1994). Dodge continued to develop it with the 1928 Victory Six as a
‘monopiece’ construction of inner and outer panels, followed by the Chrysler Airflow
and then the Lincoln Zephyr. Budd perfected the process with the Nash 600
monocoque, which weighed around 600 pounds less than its predecessor and 300
pounds less than the body proposed by Briggs (DaimlerChrysler, 2004). At this point
the three main advantages of steel-body production were in place. These were:

1. Ease of manufacturing - production automation, faster paint drying times, suited

to high-volume low-cost production.
2. Body strength - obviates need for separate chassis, improves passenger safety.
3. Weight reduction - improves vehicle performance and fuel economy.

The Budd Paradigm also comes at a high capital cost. Table 4.5 shows the typical
investment required for such a plant in the early Century, including the paintshop.

Table 4.5 Typical Budd plant investment

Process Investment (£m)

100
Press shop
Press tooling per model 20-65
BIW construction 50-100
Paintshop 200-250
Total 370-515

Source: Nieuwenhuis and Wells 2002

125
The technological basis to the Budd Paradigm processes means that it should be more amenable to

quantification of MBPS, as will be discussed in the next subsection.

4.3.2.2 Economies of scale in body pressings

The initial part of the BIW process is the shaping of the steel panels and for panel stamping alone
Rhys (1972, p.289) states that the maximum usage that can be expected from a press results in 2
million panels a year, since body presses appear to have an optimum speed of 2 million stampings
a year. The dies can last for 7 million stampings, so the optimum number of pressings is 2 million
a year over 3.5 years. Since panel production is necessarily capital intensive, the presses cannot be
replaced by labour except at the very lowest output levels o f craft production, suggesting a single
plant size with a short-run minimum average cost equal to the long-run minimum average cost for
the industry. Figure 4.13 shows this with increasing returns as production rises to the 2 million
mark, but a rapid onset of decreasing returns as the machinery is pushed beyond its design limits.

Figure 4.13 Theoretical economies of scale for panel stamping

Cost

LAC
MBPS

^ Output (million pressings)

However, BIW is also more than simply the stamping of panels, it also involves welding in jigs

and paint preparation processes. The MBPS for the entire BIW process

126
will therefore differ from that of panel stamping alone. Furthermore, BIW production
tends to use flexible manufacturing to feed multiple models directly into the final
assembly plant, more so than powertrain which can serve multiple models assembled in
different plants. It is therefore difficult to use survivor analysis of the main companies
in the industry to evaluate BIW in isolation from final assembly. Nevertheless, it is
unlikely that BIW, being almost as highly capital intensive as panel stamping, would be
allowed to operate with substantial cost disadvantages simply to suit final assembly.
Taking this into consideration, it can be suggested that where BIW and final assembly
operate together then a sense of the econoniies of scale for BIW can be inferred if
necessary.

Rhys (1999) bases the 2 million output figure for panel pressing on the technical
capability of the machinery. There is little empirical evidence of this occurring, for
example Ford’s Dearbom Stamping plant presses panels for both the F-150 and
Mustang sports cars with combined production in excess of 740,000 in 2005
(Automotive News, 2006a). The Harbour Report 2007 assesses productivity for
production plants across North America (Reliable Plant, 2007). For 2006, it concluded
that the most productive for stamping was the Honda plant in Marysville and total
production capacity for the plant amounted to 440,000 units per year (Honda, 2006).
Toyota, Georgetown was ranked second in the report for productivity with a declared
production capacity of 500,000 units a year (JAMA, 2005). This suggests that the
MBPS for BIW is around 500,000 units a year.

To explore the shape of the LAC curve it is necessary to investigate the viability of
lower output rates. The UK Toyota plant at Bumaston had output in 2005 nearly half that
of the Georgetown plant, at 264,000 units (SMMT, 2006b). In Australia, the Ford
Geelong plant recorded total BIW production of 101,014 in 2005, with capacity set at
120,000 units a year (Automotive News, 2006a). Due to the location of the plant the
output probably represents the priority of local assembly over economies of scale, partly
due to reasons of tariff barriers, and Silberston (1972, p.377) states that, in such a case,
scale has a relative value:

127
“If the market is too small to contain even one plant of the minimum optimum scale,
then it follows that any plant set up to produce for that market cannot be as efficient as
it is possible to be, because its scale will be too small.”

The difference between the US, UK and Australian output levels is explained by the
costs of transporting BIW to Australia and also the tariff structure that adds 10% to
imported vehicles. This makes it cheaper overall to produce BIW at Geelong, even at
this output level. Much below this and final assembly is usually fed by imports of BIW
in complete knock down (CKD) kit form to serve the local market. For example, from
2000 the Opel Corsa was assembled from CKD kits in China at the rate of 22,000 units
a year and sold as the Buick Sail (Sieradski, 2002). This suggests that a stamping plant
at this level of output would introduce cost disadvantages in excess of the costs of
transporting CKD kits from larger plants elsewhere. In view of the various levels of
output shown by different plants it would appear that the LAC curve for BIW
production has a shallow gradient to the increasing retums to scale side. Based on the
Harbour Report, the MBPS is around 500,000 units a year (e.g. Georgetown plant), with
a shallow gradient from 250,000 a year (e.g. Bumaston plant) and the onset of
diseconomies of scale beyond 750,000 units a year (e.g. Dearbom Stamping). The LAC
curve for BIW is therefore illustrated by Figure 4.14 below.

128
 Figure 4.14 Economies of scale for BIW production

Cost

MBPS LAC

Output (000s)
100 250 500 750

In Figure 4.14 the output for the Geejong plant is just above the 100,000 level, at the
point where the increasing returns to scale dictate that the curve begins to take on a
shallow slope. The Dearbom plant would be near the 750,000 output level, beyond
MBPS but the slope of the curve is shallow enough not to incur pernicious cost
disadvantages. Between the two would be the Honda, Marysville plant at 440,000 units
a year, and the Toyota, Georgetown plant at 500,000 units a year, judged by the Harbour
Report to be the most efficient amongst the competition and therefore representing
MBPS.

4.3.3 Powertrain

Powertrain production uses processes that are quite distinct from the pressing of shapes
into sheet steel that characterises the BIW function. Engines and transmissions are made
up of intricately shaped metals that are designed to perform under highly stressful
conditions to the extent that parts stamped from sheets of material would not be robust
enough. Powertrain parts are therefore fabricated by either casting or forging.

129
Casting involves producing the basic shape of the component by pouring liquid metal
into a mould and then machining away the surplus material to arrive at the final, desired
specification. This is commonly used for engine blocks, cylinder heads and gearbox
casings. Forging produces components by hammering them into shape, usually when
the material is hot enough to be malleable. The desired shape is arrived at by plastic
deformation of the material, which also introduces a degree of work hardening to
increase the component’s mechanical strength. This is applicable to components such as
crankshafts, camshafts and piston rods. Although the two processes are technically
unrelated there is a degree of substitutability. For example, gears can be cast as blanks
rather than forged before the precise shape is machined into them (Gear Product News,
2006).

Bhaskar (1979) suggests that since more than one engine variant would be required for
the complete model range the powertrain plant would need to have capacity for between
500,000 and Im units a year. This appears to be the current requirement in 2006, Ford
stating that its British plants in Bridgend and Dagenham will each target an output of Im
engines a year by 2009 (Ford, 2006). The Bridgend plant produces three types of engine
on two lines, four cylinder engines on one and in-line sixes and V 8 s sharing the other.
Toyota Deeside produced just under 500,000 engines in 2005, though more than half
were exported to other Toyota plants for final powertrain assembly (SMMT, 2006b). In
the US, the market is large enough to invite plant investments that explore the
economies of scale fiirther. Table 4.6 shows this for the greenfield investments by
Japanese firms.

130
 Table 4.6 Japanese powertrain capacity in the US 2004

Plant Output Capacity

Honda engines, Ohio 1,079,564 1,160,000
Honda transmissions, 1,003,289 1,000,000
Ohio
Nissan engines, 728,986 950,000
Tennessee 540,000
Toyota engines. West 456,231
Virginia
Toyota transmissions. 389,859 360,000
West Virginia 600,000 (2007
projected)

Source: JAMA 2005a

Table 4.4 indicates the range of outputs where economies of scale might apply: from
around 500,000 for Toyota up to around Im for Honda and Nissan. Since Toyota
powertrain capacity is the same for the UK and the US, this implies that Toyota has
settled on a size of plant that exploits the available economies of scale. Honda, though,
shows that scale of Im is possible and Ford demonstrates that worldwide distribution
creates the demand conditions where output of this level can be planned for, even when
manufacturing takes place in a market as small as the UK. Figure 4.15 below shows the
LAC for powertrain production where the SATC for Toyota plants would have their
minimum cost tangential with the LAC curve at 500,000 units a year while Ford and
Honda plants would have it at Im units a year. This LAC curve shows that Toyota is not
suffering significant cost disadvantages with such an output, despite its plant being half
the size of that of the competition, and so the LAC curve has a shallow slope until the
MBPS at Im units.

131
 Figure 4.15 Economies of scale for powertrain production

Cost

LAC

MBPS

Output
500,000 Im

4.3.4 Final assembly

Final assembly has come to symbolise the scale of car production more than any other
process yet, in practice, it is an uncomplicated procedure that was arrived at through
exploration of work practices rather than the development of a core technology. Henry
Ford introduced many of the flow production line techniques to the automobile industry
and, as a logical process, it has become the standard assembly method for the world’s
automobile industry, often referred to as ‘Fordism’ (Hounshell, 1984). Fordism, though,
was not a single idea but an approach that integrated and refined production processes
while restructuring employment practices. Dankbaar (1992, p.3) states that Fordism was
more than simply mass production, it extends into mass consumption since higher
wages paid for production translate into higher disposable incomes for consumption.

The principle of the moving assembly line is quite simple, the product being moved
along stations to be progressively assembled by stationary employees working from
parts delivered to them. Neither is the concept novel, its use predating the automobile
industry itself (Lewchuck, 1987; Hounshell, 1984). Even later developments such as the
Toyota Production System (TPS) and lean production amount merely to detailed

132
refinements of the overall process rather than a fundamental change in the underlying
principles. It is therefore relatively straightforward for final assembly plants to produce
more than one model, and in fact there are few single model plants in operation.

4.3.4.1 Final assembly economies of scale

For reasons of competitive advantage, the quantitative costs associated with any part of
production are not usually published so econometric and engineering evaluations of
scale are unsuitable, but by using survivor analysis the plant level strategies for
production can be analysed. Some additional information comes form the interviews
conducted by this research.

Table 4.7 Japanese assembly operations in the US 2004

Plant Output Capacity Utilisation Employees Investment

$m
Mitsubishi, 112,984 240,000 47% 1,786 1,514
Illinois
Nissan, 267,350 400,000 67% 4,100 1,430
Mississippi
Toyota, 374,048 300,000 125% 4,659 2,600
Indiana

Source: JAMA, 2005a

The results shown in Table 4.7 for capacity and utilisation reveal the problems in
making quantitative calculations. Toyota has the highest output yet seemingly lower
capacity than Nissan, while Mitsubishi is moribund with less than 50% capacity
utilisation. The problem here is that the capacity statement does not take account of the
installation investment. Toyota has by far the highest investment in its plant so higher
output is only to be expected, implying that the company has installed for the higher
output. Nissan may have the physical plant area size to offer higher capacity but the
investment suggests that the installation is for lower production. Not only is Nissan’s
investment markedly lower than Toyota’s but it also has a higher labour content.

133
indicating that the quoted capacity may be higher than the installation could achieve in
practice. Mitsubishi’s investment is slightly higher than Nissan’s but it has recently had
to reconfigure its installation to align with lower US sales. The company has reduced
the Illinois plant to a single working shift with an output capacity of 140,000 units a
year for 3 car models and one SUV (Automotive News, 2004). This accords with Truett
and Truett (2003) who found evidence that final assembly was competitive when output
was at least above the range of 150,000 to 200,000 units a year. This thesis learned from
Mr Sasaki of JAMA that assembly needs to be at least 200,000 units a year to exploit
technical efficiencies.

Mitsubishi’s output might indicate that 140,000 is the lowest limit for MBPS where the
local market is large enough to permit pursuit of industry standard economies of scale,
in contrast to Australian plants where they are constrained by the limited local market.
However, Toyota did not reveal a marked tendency towards the larger size of powertrain
plant (see section 4.3.3 above) so in final assembly, unless there are constant retums to
scale as far as Toyota’s output, it is likely that there is a shallow downward slope from
output of 150,000 units a year to the latest data on MBPS at around 350,000 units a
year. It is also reasonable to expect that the gradient of the curve will be particularly
shallow after the figure of 250,000 units a year, the approximate capacity of the Toyota
plant at Bumaston in the UK (SMMT, 2006b). This is shown in Figure 4.16, below.

134
 Figure 4.16 Economies of scale for final assembly

Cost

LAC

MBPS

Output (000s)
150 250 350

Due to the greater use of labour for smaller plants it is reasonable to expect that on the
increasing retums to scale side of the LAC curve the slope will be relatively shallow.
This is because it is relatively easy to design a production line with varying quantities of
labour, but only up to a certain point. For example, this thesis found that at Bentley
output of 3,500 units per year was viable with a workforce of around 1,000 but this was
for a product for which a cost recovery pricing strategy was possible. For mass market
production, below 150,000 units a year production would appear almost unviable and
beyond 350,000 units a year there will be an overcrowding effect amongst the labour.
The production capital will also reach its design limit, leading to a steep rise in
diseconomies of scale. Since flexible production techniques mean that final assembly
plants can produce a number of models this aids in maintaining production rates, for
example the Bentley line currently assembles three variants of the GT. However, if net
output capacity exceeds anticipated quantities of market demand then actual output will
necessarily suffer, Hyundai’s US plant being forced to suspend production for two
weeks due to the rise in production of the Santa Fe sports utility vehicle (SUV) being
insufficient to offset the drop in production of the Sonata saloon car (Automotive News,
2007c).

135
4.3.5 Research and development: economies of scale

Research and development (R&D), which includes all the design tasks necessary to
bring a product to the point of production, is the source of long-term sustainability for a
firm. The overall LAC curve for an industry implies that new products will become
available as necessary to fulfil the chosen capacity of plant. Some industries are more
R&D dependent than others, the top three highest spending being technological
hardware, pharmaceuticals and the automotive industry, with the vast majority of the
spending being conducted by companies in the US, Japan, Germany, France and the UK
(DTI, 2006).

Although absolute levels of R&D spending reveal the monetary amounts that are
necessary to compete in an industry, they do not show the strategic importance that the
activity has within a company. For this reason, R&D intensity is usually measured as a
percentage of sales (Rosegger, 1987). However, this does not show the relative
productivity of R&D, i.e. the efficiency in generating innovations. For this reason R&D
should also be related to the number of employees or the number of product lines in
order to express R&D efficiency, yet such data is usually considered commercially
sensitive.

The data that is available is that of total expenditure since it often has to be declared for
regulatory reasons. Even before investigating the data it is obvious that the higher levels
of R&D spending will be more affordable for larger companies. There may then be
economies of scope as larger firms can spread capabilities across a wider product range
and diversification of risk as they absorb instances of failure more easily. Larger
budgets may also provide protection from smaller entrants to the industry due to the
high barriers to entry created by the absolute R&D spending levels (Thompson and
Formby, 1993). Despite this, larger R&D budgets and economies of scope in the
product range do not mean that the firms are more efficient and can take firms into
diseconomies of scale in the R&D function. Parkin et al. (1997) argue that there may
even be a strategic pressure to increase R&D spending as companies compete to bring

136
new products to market. This suggests that even if the industry might be viable without
R&D expenditure, if one company invests then the others are compelled to match it in
order to maintain their strategic position. The altemative is the risk that they might
suffer disproportionately heavy financial losses should they miss out on a crucial
innovation. It is the asymmetry of risk that induces the companies to intensify their
R&D spending.

Although there may be sound strategic reasons for operating with large R&D teams this
does not negate the existence of a MBPS for R&D, one that could be met by using
smaller teams. Cooper (1964) puts forward three possible reasons to explain the greater
efficiency of smaller R&D departments:

1. Comparatively higher technical capability in the engineers.

2. Greater personal attention to cost control.
3. Improved communication and coordination.

The research by Cooper (1964) covered firms in a variety of industries. One executive,
who was responsible for a large department of over 1,200 engineers, reported that this
size of department was advantageous when dealing with large projects for which the
personnel could be divided into specialist teams. In so doing, this seemed only to
provide fiirther evidence that smaller teams were more effective, although the executive
also felt that his department held a breadth of experience that brought economies of
scope to larger projects. A distinction should therefore be drawn between R&D
functions that are efficient, in terms of achieving economies of scale, and those that are
effective in achieving their design briefs, thus benefiting from economies of scope. In
order to achieve higher economies of scope in the product range greater product
differentiation is necessary, entailing greater use of R&D resources. For R&D to be
effective in putting forward diverse products the function may have to grow even
beyond its efficient economic scale.

137
The minimum size of a team depends on the nature of the task and the divisibility of the
necessary inputs. It might be possible for a research team to be as small as two, one
conducting the experiments and the other writing the reports. For larger projects the
research team will have to expand, although coordination difficulties may introduce
diseconomies of scale. Since the activity is not machine paced there will be no sudden
rise in the cost curve as the equipment reaches its physical operating limits. It is likely,
therefore, that the cost curve will have shallow gradients. Cooper (1964) investigated
large firms in various industries with different sizes of R&D teams, ranging from 150
engineers up to 1,200 for a large department. Figure 4.17 illustrates these findings.

Figure 4.17 LAC curve for R&D, various industries

Cost

LAC
MBPS

150 1,200
Engineers
Derived from: Cooper, 1964

Figure 4.17 shows MBPS when the R&D team is relatively small, around 150
engineers, with costs rising gradually so that a team of 1,200 engineers has a cost
disadvantage, although it may be more effective if the task is large and there is a time
constraint. At some point beyond this level costs will rise to a degree that not even the
effectiveness of the team is enough to counter the costs and it will be necessary to
divide the task between separate teams, perhaps by engaging different companies. The
next subsection will evaluate this in a specific industry, that of automobile
manufacturing.

138
4.3.5.1 Vehicle research and development

R&D is one function in an automobile company that eludes decisive estimation of

efficiency. It has a complex interaction with the other functions as well as market
demand such that it needs to balance economies of scale with economies of scope; the
requirement for additional models may push the R&D function into diseconomies of
scale. The function includes product concept generation, product planning, product
engineering and production engineering (Clark et al., 1987). Development costs are
mostly determined in advance so, in theory, the higher the production the lower the
proportion development costs will take of the total cost. Dunnett (1980) found that
production figures supporting R&D in the late 1970s ranged from “significant” to 5m
units a year. Rhys (1999) puts the figure at 2m units a year.

In 2006 there were a number of manufacturers producing less than Im units a year in
total (see Appendix 13) but which maintained an R&D operation. The smallest of these
was probably Proton with production of 144,395 units, but with substantial government
assistance the company cannot be said to be independent. The Chinese entrant Chery
had output of 308,425 units for the year and courtesy of input from numerous extemal
consultancies maintained control of its own R&D. Again, though, this was also with
significant political assistance. Subaru is owned by Fuji Heavy Industries, with Toyota
as a minority partner, and has output of a little less than 600,000 units a year with its
own R&D capability. The smallest standalone producer of passenger vehicles is BMW,
all other smaller producers being members of parent groups, and it achieved output of
1.4m vehicles in 2006.

The problem with any output figure is that it is dependent on the demands of the market
and the length of the model life cycle. One plan has suggested that the MG-TF sports car
would be profitable with sales of 20,000 a year (Autocar, 2005), and with a similar ten-
year life cycle to the Mazda MX5 this suggests a total of 200,000 for this model. Of
course, with a life cycle of this length it is unlikely that there would be a dedicated R&D
team working on successive sports cars. Not only do such niche models benefit

139
from economies of scope through the production system and into retail distribution,
they also benefit from economies of scope in R&D where an engineering team can
transfer experience and skills from more mainstream projects. Since design teams move
from project to project it is preferable to consider the R&D spending over the entire
model range and relate it to sales. Table 4.8 shows this for a selection of global
manufacturers for the period June 2005 - June 2006 using data from the Department of
Trade and Industry (DTI) in order to show a consistent approach to the figures.

Table 4.8 Global R&D Expenditure 2005/6

Company R&D R&D R&D
Expenditure Expenditure % of Expenditure/employee
Ford £m sales £
4659.8 4.5 15,500
Toyota 3726.8 4.1 14,000
VW 2799.9 4.3 8,600
Honda 2308.4 5.4 16,700
BMW 2140.3 6.7 20,700
Subaru 256.8 3.6 9,500

Source: DTI, 2006

It is no surprise that the range in R&D expenditure is wide, but this may be explained
by the breadth of the model range. This is particularly the case for Ford which has a
large number of light trucks which are classed, and used, as passenger vehicles. Large
firms also tend to invest in long-term projects, for example in the case of the automobile
industry this would include fuel-cell research. Indeed, interviews conducted with Honda
for this thesis found that the company readily promoted its developments not only of
automobile related technology, but also robots and jet engines. Although the company
claimed that these projects were related to mobility there was no credible connection to
the automobiles industry. Other companies were also engaged in research that was
beyond the limits of automobile production and might have been indicative of their
search for a new manufacturing paradigm. To demonstrate how these additional projects
inflate overall R&D expenditure R&D intensity is usually found by relating it to the
number of sales it helps to capture, shown by expenditure as a percentage of sales
income. This finds that the companies shown in Table 4.8 are quite tightly grouped.

140
BMW being the heaviest investor and Subaru spending the least for its level of sales. It
is possible that Subaru receives a small degree of informal R&D support from its
parent, Fuji Heavy Industries, that has not been accounted for.

Table 4.8 follows the DTI surveys in showing R&D commitment by relating total R&D
to the entire workforce. This indicates the relative R&D for the different companies by
showing that the level of expenditure for each worker. For example, although Toyota’s
R&D expenditure is 20% lower than Ford’s it is only spending 10% less per employee,
implying that it is making a greater R&D commitment. BMW makes by far the biggest
commitment while VW spends a remarkably low £8,600 per employee.

The discovery that VW is a low R&D spender per employee is difficult to reconcile
with its status as a high spender of R&D overall and its dominant position in the retail
market. Assuming that the purpose of R&D activity is comparable throughout the
industry then by using Toyota as a benchmark, due to its known efficiency in other
fimctions, it might be reasonable to conclude that higher spending per employee denotes
a greater efficiency in its investment. Altematively, it might be the case that VW is able
to engage in R&D programmes with a lower expenditure for each R&D engineer.
However, VW does have a larger workforce, some 22% higher than Toyota (DTI, 2006)
and it is likely that this is weighted towards its production operations in emerging
regions. Rates of R&D spending per employee are therefore not a reliable measure of
R&D efficiency if it includes employees engaged in other activities.

Companies rarely publish figures for total R&D staff although it may be possible to
identify the workforce for specific R&D centres, useful if they have the capability to
take a full new model programme from proposal to production. Such centres in the US
range from 1300 personnel at Honda to 800 at Toyota, and even as low as 129 at Mazda.
Similarly, in the UK, Lotus Engineering, for example, claims to be able to deliver
complete vehicle engineering proposals using 1000 engineers, less than Honda employs
in the US (Lotus, 2006). This thesis found from interviews that Aston Martin had an
R&D strength of 170 personnel and Bentley one of 600, both claiming that this

141
gave them design authority with detailed assistance on request available from their
parent companies. Extrapolation from these figures suggest that a basic model range of
around three distinct models might indeed be supported by an R&D strength of around
1200 personnel.

The purpose of R&D is to propose new models so a further measure of R&D

effectiveness would be to relate expenditure to the number of models. A broader range
of vehicles possibly reflects economies of scope in both R&D and production. However,
there are difficulties in simply counting the number of models that each manufacturer
claims to produce. Many products share platforms but design differences allow them to
attack different market niches. On rare occasions one product range may use model
names to differentiate trim levels or brands with very little real difference in design;
more commonly one model may be sold under different names for different markets (e.g
Mitsubishi Pajero/Shogun). There will also be models that are designed for foreign
markets only, such as Honda’s joint venture involvement in the China-only Guangzhou
Honda model to be released in 2010. Table 4.9 counts the number of models sold
according to company reports and websites, although it may undercount where models
exist only as part of short production runs or to serve obscure markets. The R&D per
model has been calculated from the DTI figures shown in Table 4.8 above.

Table 4.9 R&D expenditure per model, 2005/6

Company Models R&D expenditure per model

£m
Ford 59 79.0
Toyota 81 46.0
VW 44 63.6
Honda 29 79.6
BMW 11 194.6
Subaru 9 28.5

Derived from: DTI, 2006

Company data

142
Due to the various problems with precisely defining and counting models Table 4.9 can
only be an estimate of the R&D spending per model but it does give some indication of
how the company strategies relate to each other. It should be noted that Table 4.9 does
not show the actual expenditure for particular models since in practice there will be one
core model and then a series of variants spun off from it. The figure shown is the
average across all those core models and variants. Subaru spends the least per model
with nine models based on four platforms, one of which is a light kei-car platform with
three variants. From Table 4.9 it would seem that it would cost Toyota, on average,
£46.Om to put a new model onto the market whereas it would cost BMW £ 194.6m.
Since Toyota enjoys far higher total sales than BMW (see Table 4.4) it is reasonable to
conclude that a low R&D spend per model is an acceptable analogy for an efficient
R&D operation. As in a previous estimate of R&D efficiency which related spending to
sales Subaru shows itself to be the most efficient.

This research estimates that the MBPS in R&D lies somewhere close to that of Subaru,
with the proviso that it might be receiving some assistance from its parent group. Using
production data from 2006, the year of the DTI survey used, Subaru appear to show the
following conditions for MBPS in R&D to be met: output of 600,000 units per year and
expenditure of £256m per year over four platforms and nine models. Since the kei-cars
are peculiar to Japan it could be suggested that Subaru has a core range of three
platforms and five models in order for it to be comparable as a conventional automobile
company. The company is not entirely self-sustaining since it can call on resources
within its parent company, Fuji Heavy Industries (FHI) and also Toyota which owns a
9% share of Subaru. Another 2% is owned by Suzuki but this is countered by FHI have a
1% stake in Suzuki. The value of these relationships is difficult to quantify but it would
probably offset the subtraction of the kei-car line from the production and R&D
expenditure data. This suggests that an automobile company with a similar range of
conventional products (i.e. three platforms, five models) would need output of around
600,000 units a year and an R&D investment of around £250m a year. Data from other
empirical sources, including interviews conducted for this thesis, suggest that the R&D
personnel should number around 1200 engineers.

143
Figure 4.18 below illustrates a possible LAC curve for R&D based on the level of
production output that it both serves and is served by. Since R&D is dependent upon
human assets rather than being paced by machinery, the shape of the curve is much
shallower than it would be for capital intensive processes like BIW. Up to the optimum
of around 600,000 units a year there will be some increasing returns to scale as
teamwork effects and economies of scope bring additional benefits. Thereafter,
diminishing returns to scale will set in as a crowding-out effect appears. For production
output that is higher than Im units a year it is feasible that multiple R&D centres will be
employed. It should be noted that the minimum point appears to have shifted to the right
compared to that shown in Figure 4.17 above. This is because an automobile is a
complex product and so the R&D team will be divided into smaller groups depending
on each aspect of the project.

Figure 4.18 Economies of scale for R&D

Cost

MBPS
LAC

Production output
600,000

Having determined the LAC curves for the four functions at the core of the automobile
firm these will now be combined in order to reveal the MBS for an automobile
manufacturer.

144
4.4 Combining the LAC curves

From the estimates of economies of scale, both published in the literature and provided
by the research interviews, for the three main automobile manufacturing functions it is
possible to combine the LAC curves for each function to examine the overall LAC for a
car firm as a whole and reveal the MBS for the industry. Thus, from the evidence
presented in sections 4.3 and 4.4, it is suggested that MBPS for each functions is as
follows:

• BIW LAC(B): MBPS = 500,000 units a year

• Powertrain LAC(P): MBPS = Im units a year
• Final assembly LAC(A): MBPS = 350,000 units a year
• R&D LAC(R) = 600,000 units a year (production), 1200 personnel, 3 platforms,
5 models

To arrive at an MBS output figure for the ideal firm it is necessary to calculate it from
the lowest common denominator amongst the processes. This results in an output figure
of 21 million units, a result that is nearly half the total global production for 2005.
However, each process has a reasonably flat range alongside the MBPS output figure so
a visual analysis of the LAC cost curves can indicate an acceptable point for the MBS.
Figure 4.19 does this by combining the separate LACs for the complete, integrated firm.
LAC(B) has been positioned high relative to the cost axis since it is the most capital
intensive process, with LAC(P) slightly further down since it includes additional labour
in assembly work. Final assembly, LAC(A), where all the parts are brought together, has
been positioned lower down the cost axis because it is the most labour intensive. R&D
is the most dependent on human assets and so the LAC(R) curve is the flattest, revealing
a shallow slope of increasing returns to scale as team effects and economies of scope
bring benefits.

145
 Figure 4.19 LAC curves for a prototypical integrated automobile firm

Cost

LAC(P)

LAC(B) LAC(R)

LAC(A

V'

350 500 600 1,000 Output

(000s)

Visual inspection of Figure 4.19 seems to indicate that output at 600,000 units a year
results in an acceptable and achievable low level of costs for the combination of
functions. It may not represent the absolute optimum, but with only the final
assembly plant needed to be duplicated to match the output then it offers a pragmatic
solution for the industry. This also accords with the results of the survivor analysis
which indicated that the minimum output range for survival was 0.5m to Im units a
year. This results in a company structure as shown in Figure 4.20 below.

146
 Figure 4.20 Automobile company exploiting acceptable economies of scale

R&D

BIW X 1plant

Powertrain x 1 plant

Final Assembly x 2 plants

Total output
600,000 units a year

In lieu of any manufacturer achieving 21 million units a year, then an output of

600,000 units a year represents, to a reasonable and pragmatic degree, the prototypical
MBS for the industry, known in this thesis as MESp. A similar output for MBS was
also mentioned in an interview for this thesis with automotive analyst Mr Yamaguchi,
who put the figure between 500,000 and 700,000 units a year.

Making use of the horizontal tendency in the LAC curves for each process it is possible to
have plants working at their full designated capacity, albeit one that is not at absolute
measures of MBPS. In this way the schematic shown in Figure 4.20 would represent a
prototypical model of an automobile firm with an R&D function at the optimum, a BIW
function being slightly beyond the optimum and the Powertrain and Final Assembly
functions being operated slightly sub-optimally relative to absolute measures of MBPS.
Any cost penalties would be minimised by planning for these outputs, it being common,
for example, to find final assembly plants operating at around 250,000 units a year.
However, the greater cost penalties emerge when the installed production systems are
themselves underused, resulting in a relatively rapid rise in costs on the prototypical firm’s
SATC curve (SATCp). Figure 4.21 shows how a prototypical firm that planned

147
for this level of output for achieving MESp would be vulnerable to fluctuations in
demand in the short-term.

Figure 4.21 SATC curve for a prototypical integrated automobile firm

Cost
SATC,

MES,

350 500 600 1,000 Output

(000s)

Figure 4.21 implies that firms would incur the greater costs penalties by operating
above the minimum, but this can be managed by controlling demand in the form of an
order bank or a customer waiting list. Sub-optimal production presents more
difficulties in the short-term and demand would be managed using various marketing
tactics such as price incentives, a cost that would then have to be balanced with the
cost advantage of raising production back towards the MESp output.

The sensitivity of MESp to output changes has been foreshadowed in the survivor
analysis of the world automobile industry which found that firms producing below 1
million units a year were in a weak position, and indeed this could be extended to
manufacturers producing below 2.5 million units a year. At the same time it was
difficult to reconcile this with the continuing survival of certain firms within those
output ranges, Subaru being a notable example. This is clarified by the sensitivity of
MESp to output changes which would motivate a firm to diversify risk by finding
higher levels of output and wider models ranges. The survivor analysis found Toyota to
be the industry leader but with output that is nearly thirteen times that of Subaru and

148
with nine times the number of models. If a Subaru model, such as the Impreza, were to
suffer a catastrophic failure in the market such that sales fell by 50% then the 60,000
unit shortfall would reduce total production of the company by 10%. If Toyota were to
suffer the failure of a model with the same loss in production then the reduction in
output would be barely 0.008% of the total. Furthermore, with 81 model variants on
offer it is likely that much of the demand could be transferred to a similar Toyota
model rather than being lost to a rival manufacturer

Should such a shortfall occur, Toyota could absorb it by reducing output at one plant
while leaving the rest operating at full capacity. This is because Toyota does not
manufacture from plants that are thirteen times the size of Subaru’s, instead it has a
multitude of plants of a comparable size and up to 52 manufacturing companies
overseas, even if not all are engaged in automobile manufacturing (Toyota, 2007).
Survivor analysis has demonstrated the expansion of production at Toyota, as well as
other industry leading manufacturers, and as each new plant is brought on line the
company has advanced in steps that can be measured in units of MBPS. In other
words, a new assembly plant that is constructed in accordance with MBPS will bring
an increased capacity of 350,000 units. For a company like Subaru this would
represent an increase in capacity greater than 50% but for Toyota it amounts to just a
0.05% increase in capacity.

Since a new plant must be served by other functions the firm will expand in units of MBS,
and in the case of MBS? this was found to be around 600,000 units. This may not be
strictly necessary if the firm has access to different factor endowments such that it is able to
accept different levels of output. For example, this research found Toyota engine plants had
output of 500,000 engines a year and final assembly plants had output of 250,000 units a
year. The firm could thus expand in steps of 500,000 production capacity units, although
actual rises in output may not be this dramatic as it takes time to bring a new plant up to
full working capacity. Nevertheless, this clearly demonstrates that increases in capacity for
larger firms are much less risky than for smaller firms.

149
If firms in the automobile industry are expanding in units of MEPS to reduce risk then
as they expand output the SATC curve will flatten out in recognition of the reduced
variation in cost with changes in output. However, section 4.3.5.1 of this thesis has
shown that R&D costs rise significantly as firms gain in size. Figure 4.22 therefore
shows the SATC curves for three types of automobile firms:

• SATCp: prototypical firm, optimum output at MES? of 0.6m units

• SATCi: intermediate firm, optimum output at 1.2m units
• SA TC l: industry leading firm, optimum output at 5m units

The LAC curve is scalloped around the three firm sizes although it should be noted
that other intermediate sizes of firm are available according to the multiples of SATCp.
It should also be noted that firms smaller than SATCp have not been shown but it is
assumed the LAC curve rises steeply for these lower output quantities, thus they suffer
unsustainable cost disadvantages in production compared to firms at, or above, MESp.

The LAC curve shows rising costs after MESp on SATCp because R&D increasingly
experiences diseconomies of scale. In this example, each firm is shown to suffer the same
amount of production shortfall from its minimum cost position, the reduced output denoted
by points Xi, X2 and X3 . Since the SATC curves become increasingly flat with the growth
of the firm, the costs disadvantages that relate to the shortfall become less marked. The
costs associated with the MESp are shown at Cp, but for output Xi on SATCp the costs rise
to Ci. For output X2 on SATCi, the costs are shown by C 2 , meaning that the production
shortfall has had less of an effect than it did on SATCp. The least affected is the production
shortfall on SATCl, shown at X 3 , which incurs costs C3 . Not only are these costs barely
distinguishable from what they might have been at the minimum cost output for this curve
of 5m units, but they are very close to the costs associated with the MESp for the industry.
This shows that although larger firms incur cost penalties in R&D and, moreover, are
unable to achieve their minimum cost output because it would be almost impossible to
have a multitude of factories all operating at full capacity at the same time, the shallow
slope of the SATC curve means

150
that they are able to remain close to the optimum for the industry despite variations in
output.

Figure 4.22 SATC and LAC curves for automobile industry firms

Cost

SATCp
SATC,

MESp

LAC

Output (m)
X, 0.6

It is this ability of larger firms to remain reliably close to the minimum costs for the
industry that allows them to benefit from an influx of revenue. For example, although
Subaru is a profitable firm, its net operating income for the 2007 financial year
(ending March 2007) amounted to £44.4m, whereas for Toyota in the same period, net
operating income amounted to £7.0 billion (source: company reports). Expressed as
net operating income per unit, Toyota earned £909 per unit and Subaru earned £74 per
unit. Of course, this is a single sample and other years might show less of a difference,
nevertheless the difference is wide enough to be indicative. It should also be noted
that there are cases where large firms, despite the diversification of risk, suffer a
multiple failure which incurs heavy production and financial losses which can only be
addressed by multiple plant closures to realign the company on a new SATC curve;
the restructuring that has taken place at companies like Ford and GM would be a case
in point.

151
4.5 Conclusion

This chapter has described how the presence of economies of scale for the four main
functions in automobile manufacturing is expressed as MEPS for plants and MBS
figures for integrated companies. Survivor analysis did not conclusively identify an
overall MES for the industry within a growing market, although it did suggest a
minimum output range for survival and reveal a trend by the largest manufacturers
towards ever higher outputs and dominant markets shares. Instead, established
companies were investigated for the MEPS related to individual processes. Since each
function uses a distinct production technology it would be unlikely that each MEPS
output figure could easily be matched with that of the other functions to arrive at an
accessible MES output for the firm as a whole. Instead, the flat ranges in each of the
function LAC curves allow a rather more pragmatic MES output to emerge, that of
MESp. This was found to occur at around 600,000 units a year. However, companies
operating at such a level of output are sensitive to changes in output which lead to
rapidly increasing costs. Larger companies are less susceptible to the onset of these
increased costs since they can diversify the risk across larger numbers of products and
plants. Output higher than MESp is therefore a strategic move to reduce risk rather than
a search for further economies of scale. Indeed, rising R&D costs beyond MESp
indicate that larger firms may be suffering diseconomies of scale but ones that are
countered by the flattening of the SATC curve as output capacity increases.

There is nothing in the foregoing to suggest that a company is compelled to combine

all four of the automobile manufacturing functions. In attempting to bring together the
LAC curves for each function a company might consider instead focusing only on
those functions where it felt it had the advantage. Generally, though, this does not
happen and only the smallest of manufacturers operate with only three or less
functions. Since this cannot be explained by economies of scale, which are concemed
with technical efficiencies, the next chapter of this thesis will investigate the reasons
behind vertical integration of the four functions.

152
 Chapter 5;
Structuring the Automobile Industry Paradigm

When a function is internalised the relevant economies of scale then become the
responsibility of the parent company but this does not have a material effect on the
quantification of the cost benefits of scale. For the purposes of the generic model being
constructed by this research the argument is clarified by keeping the two concepts
separate. The preceding chapter set down the principles of scale and demonstrated how
these determine the size of automobile industry. In particular, that for a given production
process, one using a specified production technology, only a predefined production
output range can make the most efficient use of the factors of production. Chapter 4
demonstrated these output ranges for each function based on data gathered from the
literature, industry sources and fieldwork interviews.

In order to demonstrate how automobile companies are structured this chapter is divided
into two parts. In Part I, transaction cost analysis will be investigated as a mechanism
for explaining the inclination of firms to vertically integrate. This will include the cost
advantages of internalising functions and how they might be managed within a
governance structure. In Part II, the automobile industry will be used as an example of
how structural considerations can lead to a standard prototypical organisational form of
vertical integration for a car firm. Chapter 4 can then be drawn on to provide the sizes of
plant that are included within each internalised function. Since the considerations of size
and structure prescribe a basic type of automobile company, one with defined
limitations and potential, this thesis terms it an automobile industry paradigm. This
chapter will conclude with empirical evidence gathered during the fieldwork phase of
this research in order to contextualise the theoretical arguments.

153
 Part I: Company Structure

In Part I, transaction cost analysis (TCA) theory will be examined as a structuring

approach to the economics of business relations. This concerns the “make-or-buy”
decision, whether to produce in-house as part of a firm’s hierarchical structure, or else to
buy from another firm through the mechanism of the market. The market relationship is
governed by bilateral contracts formulated ex ante but which cannot account for all
future contingencies. Consequently, the market relationship may suffer from ex post
opportunism or hold-ups when benefits accrue disproportionately to one of the parties.
The economic costs of this can be likened to the friction between the parties,
specifically economic fnction in this situation. In order to overcome this fnction one
party may vertically integrate the other into its hierarchical structure. In a mature and
global industry, where these business relationships may be a common feature,
transaction cost analysis suggests that there is an optimal structure for firms that are
using the same production technology.

A firm’s structure is designed to minimise the costs of company functions working with
each other. The purpose is for a firm to achieve the same, or better, minimum costs as
its rivals. As David and Han (2004, p.40) summarise:

“The theory’s central claim is that transactions will be handled in such a way as to
minimize the costs involved in carrying them out.”

Cost is therefore the deciding factor behind ‘make-or-buy’. Harrigan (1986, p.536)
suggests that there may yet be a third motivator to organise, not just ‘make-or-buy’ but
‘diversify’. The author characterises diversifying vertical integration as being one that
takes the company beyond its core capabilities. This, though, seems to be more a
problem of semantics than organisational theory. If integration is to be vertical then it
must take the firm outside its established area. When it is part of a diversification
strategy benefiting from managerial economies of scope then, at best, it is horizontal
integration and at worst it does not involve transactions at all. In order to define when
transaction costs are relevant David and Han (2004) list three ways in which transaction
costs are manifested:

154
 • Asset specificity - the degree to which assets may be redeployed.
• Uncertainty - the predictability of ex ante outcomes.
• Frequency - how often the relationship needs to be renegotiated.

This chapter will show that by an historical process of consolidation these

considerations may culminate in vertically integrated firms based around their main
functions. Since the degree of vertical integration has unlimited potential the term ‘fully
vertically integrated’ is a nebulous one. In order to confine the discussion to the defining
functions the comprehensively vertically integrated firm will be said to be structured in
accordance with the forthcoming proposal, the full-fiinction model. This is to obviate
comparisons between firms operating at different levels of scale which have the
opportunity to integrate with a multitude of peripheral firms. Vertical integration here is
not about absorbing quantities of extemal resources but about intemalising the core
functions that define an industry.

5.1 Vertical integration and the full-function company

When a company takes more responsibility for production processes by intemalising the
supply chain it is said to be increasingly vertically integrated. A vertically integrated
manufacturer is one that has intemalised all the major functions of production, although
sub-components and raw materials may still be supplied from extemal sources.
Horizontal integration is concemed with economies of scale, such as the scope of
products and the quantity of output, while vertical integration is concemed with the
number of production stages included within the firm (Cabral,
2000).

Vertical integration is a relative term because it depends on the degree to which

processes have been intemalised and so complete vertical integration would be difficult
to achieve since it would include even the processing of raw materials. The minutiae of
vertical integration are not the focus here, though the exposure to the concomitant
economies of scale is relevant. For this reason, companies that operate the main
manufacturing functions, and are therefore exposed to the attendant economies of scale,
will here be termed ‘full-function’ rather than vertically

155
integrated. In effect, the production process is being treated as interconnected modules
of discrete production activity, each with its own inherent economies of scale, as
described in the previous chapter.

Figure 5.1 illustrates the company boundary of a typical full-function company with
four modules of production. The product originates in the Design function and is put
together in the Assembly function. The product is made up of two components
manufactured in two technically unrelated processes. Process 1 and Process 2, though a
more complex product would require more processes. Each process is analogous to a
production plant, although in practice the plants may be multiplied. The final products
are released to the market which is, of course, extemal to the functional operations of
the company, though information from it is fed back to the company’s Design function
by way of a marketing capability.

Figure 5.1 Typical full-function producer

Production
company
Design

Process 1 Process 2

Final
Assembly

Market

A fumiture factory, for example, might have a process for manufacturing wooden frames
and another for the soft fabrics, there being no technical commonality between the two.
The manner in which the major functions relate to each other is a key element in the
stmcturing of the company, initially as to whether the functions should be intemalised
and then the level of output for each. Conceming intemalisation, transaction cost
analysis is an instmctive method for comprehending the resultant

156
vertical integration structure of the company. Outside of the company’s boundary lie the
suppliers with which the company has a market-based relationship.

5.1.1 Component supply industry and functional relationships

A multi-stage production system involves inputs from a myriad of component suppliers.

Even for a full-function firm operating in multiple core areas there will be component
suppliers contributing at the sub-functionary level. These range from small individual
parts to larger sub-assemblies that are built up from a collection of components. The
completed sub-unit is then fed into the main assembly line.

The firm is in a vertical relationship with a supplier if their contribution is at a prior

stage in the sequence of the production transformation process, thereby enabling the
firm to complete its work. The part being supplied would therefore have to make a
unique contribution to the production process, if it were duplicating a current activity at
the firm the relationship would be one of ‘tapered integration’ (Cabral, 2000). Between
different suppliers the relationship is horizontal if they are producing broadly the same
product. In terms of integration, it is vertical integration when a producer internalises a
supplier of complementary parts and it is horizontal integration when the absorbed
company is supplementing a current capability. Horizontal integration is particularly
pertinent in international growth when a foreign supplier is absorbed in order to secure
component supply at the new location in place of components sent from the home base.

The varying levels of integration between suppliers and producers are illuminated by the
range of relevant factors stated by Odaka et al. (1988). This includes technological
distinctiveness of the part, production technology, demand and economies of scale.
Components may use a unique technology or may be an item that is common to other
industries. Walker and Weber (1987) then list three costs that arise in the buyer-supplier
relationship:

157
 • Switching costs incurred by the buyer when changing supplier.
• Adjustment costs incurred by the supplier when changing the product or output.

• Transaction costs due to opportunistic behaviour by the supplier.

For generic components the firm has purchasing power and can switch suppliers for a
relatively low cost. This might be termed a low dependency relationship and one that
can be controlled by the mechanisms of the open market. High adjustment costs occur
when the supplier has specialised to the buyer, thereby introducing specificity in the
transaction, but the supplier can also use this position to hold the buyer hostage to the
relationship. For example, in fuel cell vehicle development Honda first purchased fuel
cell stacks from Ballard of Canada for the FCX VI in 1999, but began developing its
own independent technology at the same time for the FCX V2 vehicle. Perhaps one of
the most famous cases, though, was that between GM and Fisher Body during the
period in the 1920s when the car industry moved into the Budd Paradigm style of
unitary body production. This will be discussed more fully later in this chapter.

Not only can opportunistic behaviour and adjustment be seen as categories of

transaction costs but so too can switching costs. Transaction cost analysis is concemed
with all the costs that arise in managing the relationship between buyer and supplier,
culminating in the ‘make-or-buy’ decision. The transactions between buyer and supplier
are govemed by contracts, but these cannot be written in such a way that all the
contingencies are accounted for. Transaction costs therefore arise due to the element of
risk and uncertainty that create economic drag or fiiction, expressed as costs.

5.1.2 Transactions as mechanical friction

The transaction cost model involves the transfer of knowledge and resources under a
system of ownership controls. Simply put, the firm will expand the scope of its activities
to a point where the cost of an activity carried out within the organisation equals the cost
of conducting it with an extemal organisation. The purpose is to minimise costs across
separate technologies either by integrating activities or

158
contracting with extemal organisations (Williamson, 1998). These costs are not
necessarily quantifiable:

“Unlike production costs, transaction costs are very difficult to measure because they
represent the potential consequences of altemative decisions.” (Klein et al., 1990,
p.197.)

Dealing with another organisation brings with it additional costs, such as the
information search to find the appropriate partner, bargaining costs in arranging the
relationship and finally administration costs in the course of the relationship (Viitanen,
2002). The relationship is govemed by mutually agreed contracts and the extemalities
can be summarised as follows:

• Search costs
• Contracting costs
• Coordination costs

Although information technology (IT) improvements have reduced relative costs of

extemal resources this is a contributory factor and will not necessarily eliminate
altogether the costs of administering an extemal relationship.

Transaction cost analysis focuses on the links between different functional activities.
The point of delivery represents the interface between the supplier’s and the producer’s
assembly process, regulated by formal statements of the bilateral relationship. When the
supplier delivers the components right up to the assembly stations this brings the point
of delivery into closer proximity but it does not change the fundamentals of the supplier-
producer relationship. Should the assembler find that by looking at the costs of this
relationship that it could achieve lower costs by absorbing the component production
within the company then this would suggest the need for increased integration. Levy
(1985) was of the opinion that integration was a logical outcome of working in close
proximity because it realised flow economies through a highly transaction specific
relationship. However, this does not necessarily change the physical nature of the
component production and supply since JIT (just-in

159
time) supply is not reliant on intemalisation, although it may open possibilities to
subsequent innovation. When transaction cost analysis leads to increased integration it
is done to reduce the costs of contracting with an extemal organisation, not to change
the nature of the processes.

Choosing to integrate the processes within the firm does not mean that there has
occurred a market imperfection, the resources are still available extemally, but it does
represent a failure of the market to provide the most cost effective stmcture. In essence,
the industry and the product remain unchanged. Williamson (1981) made the point that
transaction costs are the economic analogy of mechanical fiiction between moving
parts, they are the drag that occurs when transferring between activities, and this has
been reiterated by subsequent researchers such as David and Han (2004). The activities
are technologically separable goods and services, the transaction being the transfer
process. The point could also be made that technological separation implies that
different economies of scale are available for these activities. However, the interest here
is not with the intemal costs of each process but the costs arising between processes.
Like David and Han (2004), Williamson (1987) puts transaction costs into three
dimensions:

• Frequency
• Uncertainty
• Specificity

Williamson places the emphasis on the last two. Contracts serve to control the transfer
between the principal and agent, for example between the assembler and the component
supplier. Uncertainties result from “human nature as we know it” and the risk is
heightened by the degree of commitment, or specificity, of the assets (Williamson,
1981, p.553). While the market offers the freedom to find business relationships,
contracts are needed to control the form of the relationship since Williamson assumes
that human nature will play a central role.

160
5.1.3 Contracts and the market

The behavioural assumptions Williamson (1981) has in mind are the bounded rationality
of agents acting within available knowledge, and the tendency towards opportunism
when new knowledge emerges after the contract has been agreed. If all relevant
information were known then complete bilateral contracts could be written ex ante, but
this not being the case no contract can be formulated to cover all possible contingencies.
This is exacerbated by the human desire to exploit subsequent advantages when they
accrue asymmetrically to one of the parties, giving rise to an ex post opportunist
situation where one of the partners can exert power not accounted for in the original
contract. One form of opportunism is the hold-up, where the owner of the relevant assets
threatens to withhold supply in order to extract terms more favourable to its operations.
For Williamson, opportunism is the main reason for attempting to regulate business
through contracts or vertical integration.

“But for opportunism, most forms of complex contracting and hierarchy vanish.”
(Williamson, 1993, p.97.)

Williamson is perhaps displaying an overly negative opinion of human nature, and

concedes that transacting parties are usually “well-socialized” enough to economise on
transaction costs to a degree. Furthermore, an incomplete contract is also one that has
an inbuilt flexibility that might yield unforeseen innovations. A rigid contract can bind
two firms together when economic efficiency would be better served if they dissolved
their relationship (Tirole, 1989). This may be unwittingly entered into, as Williamson
(1993) suggests:

“...opportunism is more often suppressed unknowingly or selectively and that, once

done, the ramifications are rarely assessed.” (Williamson, 1993, p. 100.)

For Klein (2000) there is another, implicit, way in which the incomplete contract can be
confined. If the benefits of reneging on the contract are offset by damage done to the
offender’s good name, what can be termed its ‘reputational capital’, then the contract is
costlessly self-enforcing. This is an attractive proposition for a company because
internalisation also has costs. When a company absorbs a supplier it is, by

161
definition of the relationship, expanding into an industry in which it previously had
little or no experience and may push the firm into diseconomies of scope. Although the
information search costs in seeking suppliers are being reduced by such a move there
then appears an opportunity cost in forgoing engagement of suppliers in the future that
might have been more suitable. In a sense, the producer becomes locked-in to a supplier
that may not provide the greatest benefits. This may be accentuated during a time of
fast moving innovation, the period of uncertainty described by Williamson, when
another supplier may unveil a development of considerable impact.

The foundation of this dichotomous decision making, integration of resources or

finding those resources in the open market, is the nature of the market itself. However,
there are difficulties in defining precisely what shape the market takes. Ankarloo and
Palermo (2004) restate Williamson’s assumption on the primordial existence of the
market, with zero transaction costs acting as a reference point. Since such a utopian
state can only be aspired to, due to the human frailties described by Williamson, then
contractual relationships introduce transaction costs. The concept of the firm is then
defined by its boundary of contracts linking it to the market. This is suggestive of a
sense of evolution from the pure market to one that deals with inevitable, if
unpredictable, market failures. Ankarloo and Palermo dispute the dichotomy, pointing
out that the market basis does not explain the existence of economic institutions like
firms since Pareto efficiency can be achieved by a centralised and formal allocative
model that a government controlled economy would provide. This is a problem for
Williamson’s theory since he refers his theory of the firm, and transaction costs, to the
failures of the market. If an alternative to the market exists then there are reasons for
firms to exist other than to deal with transaction cost problems and market failures.

Ankarloo and Palermo’s Marxist polemic does not actually rule out a role for
transaction costs but they introduce a third element to refute the market-firm dichotomy.
This is the regulatory function of the state to legislate a framework for markets and the
manner in which companies should be permitted to operate within them, yet the authors
seem unaware that this is simply an institutionalised version of the contract. Moreover,
since the functions of the state need to be paid for, any state promulgated regulations
bring with them their own costs, thus reinforcing the idea of

162
market failure and transaction costs. Klein (2000, p. 127) illustrates this when he states
that referral of a contract in dispute to a third party introduces a time lag and the need to
communicate the minutiae of the contract results in “increased noise”. These costs are
avoidable when a self-enforcement mechanism exists, such as when the renegade firm
risks injury to its reputation or “transactor’s reputational capital”. Klein notes that this
is particularly effective with large contracts.

To take the contracting process out of the hands of the contracting parties and utilise the
services of a regulatory intermediary is simply to politicise the existing process, it does
not create a third alternative to the market and the firm. It is a fundamental point that for
a government to regulate a system of contracts, which is in itself already a form of
regulation, is either to add another layer of regulation, or at least create the forum where
the contracts are formulated, but it does not provide an alternative to the necessity of
making contracts. It hardly matters whether the primordial market ever existed or is an
“expositional convenience” used by Williamson (1975, p.20). Conversely, it would be
quite impossible for the concept of the firm to predate the market unless all economic
activity had begun with one universally integrated firm. Williamson states that it would
even be possible to start from a theoretical position of primordial central planning, the
point being made is that economic friction caused by bounded rationality manifests
itself as transaction costs. The purpose of the firm, then, is to internalise the transactions
and so reduce the costs, thereon expanding until the costs of the internal transactions are
equal to or exceed the costs of external transactions between firms. The market is
simply the external relationship between firms, not an institution in itself. Moreover, the
market is not defined by the medium of the transaction, although money tends to reduce
transaction costs in comparison to barter since it is usually mutually acceptable
(Williamson and Wright, 1994).

5.1.4 The imperfect contract

Williamson (1975) looks at the market as a theoretical starting point but cannot define
its exact nature because, as the antithesis of a contractually controlled system, the
market is unstructured and unpredictable. If the market were predictable then contracts
could be perfectly formulated to cover all contingencies. The choice companies face is
between internalisation of resources into an integrated hierarchical

163
firm or finding those resources in the market and regulating them by contracts. The
transaction cost approach is not derived from some poorly defined notion of the market
but has varied antecedents, namely organisation theory and contract law, in addition to
economics (Williamson, 1981, p.550). These three are indivisibly interrelated, a
contract being written by participating organisations in order to gain some control over
the economic fiiction between them.

Concerning contract law, Klein (1980, p.356) underlines the basic problem of contracts,
that:

“.. .complete, fully contingent, costlessly enforceable contracts are not possible”.

This is stated without direct reference to the market because contracts are made as
security, however limited, against the other party’s failure to act in accordance with
expectations. Slater and Spencer (2000) discuss the difference between uncertainty,
which is unknowable, and risk, which may be calculated for by the transacting parties.
Yet the insurance industry is able to calculate premiums without having to make a fine
distinction between the two concepts, so it could be suggested that from the perspective
of transaction costs theory the difference is mainly semantic. In any case, Klein (1980,
p.357) does not express a need to distinguish the two:

“Contracts can be usefully thought to refer to anticipated rather than stated

performance.”

Two reasons are given for no contract being complete: firstly that not all contingencies
can be accounted for; secondly that performance is difficult to predict. Specific
reference is even made to Williamson’s warning about “opportunistic behaviour”. This
is a reminder that transaction costs and contracts are formulated with reference to the
behavioural issues of economics and organisations, not the conditions of a theoretical
original market. Klein (1988, p.202) takes this further, showing that long-term contracts,
while they are partially enforced by reputational capital, a large enough shift in the
demand or supply characteristics will move the “contractual arrangement outside of the
self-enforcing range”. Although investments specific to the business relationship require
long-term contracts for security, far from enforcing

164
the relationship this actually creates the conditions for a potential hold-up, albeit one of
unknowable source ex ante. It is the nature of this specificity that is relevant to the
necessity for vertical integration.

5.1.5 Specificity

Klein (1980) particularly focuses on the risks involved when firm-specific investments
are made since the firm making them can be held hostage by its contractual partner,
which can then exploit its advantage by appropriating the quasi rent stream. The
investing firm has most to lose due to the difficulty in transferring the investment,
whereas the renegade firm has most to gain, at least in the short-term. Williamson
(1981) also stresses this point, preferring the term “asset specificity”, but supporting the
view that increased specificity brings higher risks and therefore more significant
transaction costs.

According to Williamson (1981, p.555) specificity can occur in three ways:

• Site specificity
• Physical asset specificity
• Human asset specificity.

The specificity requires commitment since it is derived from the value of an investment
being higher for the relationship in question over any other. For example, the fast-food
chain McDonald’s operates restaurants with franchise arrangements. The main physical
assets are the restaurant buildings with brand specific architecture and interior design.
Since these are important in projecting the appropriate corporate image the company
retains control of them by owning them. This power is partially countered by the
franchisee’s power to withhold service standards set by McDonald’s, the franchisee
having the potential to profit from cost savings that accrue to their part of the business
but which are damaging to the corporate business as a whole (Milgrom and Roberts,
1992). McDonald’s is therefore careful to put in place a system of incentives and
penalties to support the contractual relationship. Lieberman (1991) found strong
empirical support for integration by firms that carried significant sunk

165
costs due to asset specificity and anxiety about being “locked-in” to a business
relationship. Lyons (1995) also found asset specificity to be predictive of vertical
integration in UK engineering firms.

Additional specificities that are derived from the three mentioned above have also been
put forward. Williamson (1983) included dedicated assets which are generic in nature
but specific to the particular transaction and might have been put in place in expectation
of new business. This being the case then, as far as the transaction in question is
concerned, they have the same risks for the supplier as physical asset specificity, should
the deal fall through the assets have less value when applied elsewhere. Joskow (2003)
puts forward the idea of intangible asset specificity as would be contained within a
brand image, giving the example of McDonald’s fast food chain. Again this is
derivative and aspects of it appear in human asset specificity (service) and physical
asset specificity (restaurant decor). It is also covered by the reputational capital
described by Klein (2000).

Masten et al (1991) add temporal or time specificity where a supplier of a time

sensitive product gains progressively greater power over the buyer as the deadline
looms. It is notable that defence contracts in the UK often detail penalties to counter
this hazard, the most recent example being the overruns on the Nimrod reconnaissance
aircraft giving rise to a £500m penalty against manufacturer BAE Systems by the
British government (Sunday Times, 2006). High temporal specificity also occurs on a
sequential production line where a stoppage at one place causes the entire line to halt.
Joskow (2003) believes this to be a sub-category of site specificity. Indeed, specificity
as a concept is most comprehensively covered by the site, physical and human
specificities originally set out by Williamson (1981), the derivations being useful in
explaining the nuances of particular instances.

5.1.6 Site asset specificity

Site specificity anchors an industrial activity in a particular location due to the pre
existing characteristics of the site. This can be taken to mean the natural resources as a
factor of production, but as Joskow (1985, p.38) puts it:

166
 . .buyer and seller are in a ‘cheek-by-jowl’ relation with one another”.

Historically, industries were usually tied to a particular location due to the reliance on
natural resources and poorly developed transportation. Sheffield rose as a steel town
due to the proximity of iron, coal and local rivers for water power. Once the industry
was established it took on Joskow’s more sophisticated definition of site specificity
based on external economies of scale, attracting developments in steel making due to
the industrial infrastructure as much as the natural resources:

“...reflecting ex-ante decisions to minimize inventory and transportation costs”

(Joskow, 2003, p. 14).

Nuclear power stations also need a readily available supply of water but they tend to be
located in less populated areas. This avoidance of a particular factor, rather than
attraction to it, might be termed ‘negative site specificity’. Once operating, though, a
second nuclear power station can be built in juxtaposition, yet the two power generators
are not in a buyer-seller transactional relationship. The proximity of the second station
is due more to reasons of economies of scale and political issues, including the original
negative site specificity.

With advancements in transportation links political pressures can often be the deciding
factor in site specificity. UK government policies have influenced choice of location,
from intervention by way of the Industrial Development Certificate of the 1960s
(Adeney, 1988), to inducement by way of financial incentives later in the 1980s and
beyond (Garrahan and Stewart, 1991). In effect, this altered the price of factor inputs.
However the policy of the 1960s showed that while government policy could be
instrumental in locating manufacturers it was not the sole motivator and the long term
success was not one that could be reliably predicted. Jaguar avoided having to move by
taking over Daimler’s Browns Lane plant in 1952 (Underwood, 1989). This plant
continued in production for a further fifty-three years before it was found to be
uneconomic in comparison to the other Jaguar sites at Halewood and Castle Bromwich,
at which point it was closed as a production facility.

167
5.1.7 Physical asset specificity

Physical asset specificity concerns the degree to which capital investments are firm
specific. Williamson (1981) tells us that asset specificity is the most important element
in transaction costs, the less specialised the assets the fewer the hazards. When
specialised facilities are installed then the relationship between the buyer and seller
becomes “locked in” with both parties interested in the longevity of the contract. In a
survey of hospitals, Coles and Hesterly (1998) concluded that specialised equipment
that could not be transferred to other hospitals, because it was configured for a service
unique to one hospital, was defined as a specific asset. Physical asset specificity can
therefore be contingent on the service being offered, it being possible to offer a reduced
standard of service with standardised capital equipment. However, Williamson’s
perspective on this is that once the more specialised medical equipment is installed it is
then in the interests of the surgeons to remain with the equipment, based on the rational
assumption that this would result in a higher quality of service.

Coase was not so impressed by the impact of asset specificity, noting that businessmen
he spoke to in 1932 did not attach great importance to it since they could arrange
contracts that took care of their concerns. He claims that:

“...contractual arrangements were able to handle the asset specificity problem in a

satisfactory manner” (Coase, 2000, p. 18).

The views of Coase and Williamson are not necessarily irreconcilable. Williamson is
taking a holistic perspective, ex ante to ex post. Entering into an asset specific
transaction carries higher risks than if standardised, transferable assets were used. Early
on in the relationship hold-ups can be enacted as the holder of the assets is able to
exploit their position to extract concessions from the purchaser. Once the relationship is
stabilised, Williamson believes in the longevity of the relationship, a phenomenon that
Coase observed ex post. Specific physical assets appear to create high entrance and exit
costs which have to be assessed ex ante, but this also brings stability to the relationship
once all costs have been accounted for, and as Williamson states (1993, p. 105):

168
 . .the wise prince is one who seeks to both give and receive credible commitments”.

This also acts as a reminder that transactions and contracts are aspects of human
relationships, indicating that human asset specificity might have a crucial role.

5.1.8 Human asset specificity

The unique knowledge and skill that is dedicated to the transaction is of human origin,
hence human specificity. Coles and Hesterly (1998) draw a distinction between human
asset specificity which links the person to a unique service, and general human capital
which is transferable between organisations. The example they provide is of assistants
who provide a service tailored to the particular needs of the surgeons at that hospital, a
unique service that might not suit surgeons at another location. Human asset specificity
is therefore inherent to the supplier and the buyer can exploit this if there is a credible
alternative source of supply. The supplier would then find that the knowledge
accumulated could not be transferred to another buyer and obtain the same value.

Since the asset itself resides within the worker they are also in a position to cause a
hold-up. Joskow (2003) gives the example of the special skills that aircraft designers
accrue through experience, but it is quite feasible that these designers could withhold
their labour if they believed they were not being adequately recompensed (Williamson,
1981). This represents an opportunity for the “appropriation of quasi rents” (Klein,
1988, p. 199). Klein believes that this situation is peculiar to human assets, indeed that
the entire ‘make-or-buy’ decision is actually a choice between the physical making of
products and the buying of knowledge, not between internal integration and
outsourcing. A firm does not need to own its physical assets to use them for production,
and Klein gives the example of a firm not owning the offices in which it works.
Furthermore, the entire production transformation process is replete with physical assets
any one of which has some potential to be the source of a hold up. However, whatever
the extent of ownership of the physical assets, the human assets that operate them can
never be owned in the same way. It is solely through owning the labour contracts that a
firm can have a claim on the human assets, and the

169
contracts detail the extent of those claims. This definition of output by contract is also
applicable to physical assets, which is the basis of the outsourcing agreement, but the
owner of the physical assets has additional discretionary powers over the asset.

Klein (1988) asserts that in a free world the human assets are at liberty to seek
employment elsewhere since humans retain discretionary powers over themselves,
notwithstanding the details of the labour contract. This might threaten the sustainability
of the firm but as long as key personnel remain then the knowledge embodied in the
human assets will perpetuate within the organisation:

“The employees come and go but the organization maintains the memory of past trials
and the knowledge of how to best do something.” (Klein, 1988, p.208.)

Klein describes this as ownership of the firm’s organisational capital, which is

embedded in the human assets. The human asset specificity lies with the organisational
capital and so vertical integration internalises this organisational capital even if the
number of labour contracts compared to the market remains constant. The great
advance for the firm is that it is no longer bound to ex post consequences of the
contracts because any advantages that accrue disproportionately to the human assets
also accrue to the firm through its ownership of the organisation capital. Klein (1988)
believes that even if some employees can threaten a hold-up by resigning, the
organisational capital remains in place with the key personnel that stay.

Like all transaction cost analysis this is a ‘make-or-buy’ dichotomy, internal flexibility
still foregoes the flexibility of accessing more widely spread human assets in the
market, so the choice is based on relative costs. Nevertheless, within existing labour
regulations the integrated company can make adjustments to plant locations, production
output and other tangible assets because it owns the labour contracts of the workers that
operate those assets. The transaction costs that the integrated firm saves are in
negotiating and renegotiating these contracts at the same time as avoiding one
overarching, yet rigid, contract between two separate firms. This concords with the
views of Hart and Moore (1990) concerning the property rights theory of the firm,
which is more specifically based upon the ownership of physical assets, where the
authors conclude that an integrated firm can shed workers selectively, but when

170
contracting with another firm it is the whole firm that must be “fired”. However, by
taking a strictly physical asset view of the firm, it is difficult for the property rights
theory to account for the opportunism of human frailty:

“Surely integration does not give a boss direct control over workers’ human capital, in
the absence of slavery?” (Hart and Moore, 1990, p.l 150.)

Klein’s insight into the ownership of the organisational capital through the employment
contracts deals with this control issue and has crucial importance for knowledge
intensive processes such as design work. Where the knowledge is at the core of the
firm’s operations then it needs to secure the services of those human assets that carry
the knowledge, at least in high enough numbers that the knowledge is perpetuated
within the organisation. If the value of this knowledge is high enough it can push other
cost considerations, such as economies of scale, into a distant second place. Klein also
argues that the same principle can be applied to the physical assets since knowledge is
required to operate these too. While this may be relevant when the production
technology is in flux during a time of rapid development, where the technology is
mature but capital intensive the switching costs of the physical assets come to the fore.
The implication of this is that companies made up of processes that are, separately,
knowledge intensive and capital intensive, will vertically integrate those processes for
different reasons: the human assets to secure knowledge, the physical assets to secure
production.

As Klein points out, the securing of human assets cannot be done by direct contract and
so additional incentives become the key for securing access to them. Holmstrom and
Milgrom (1994) explored the complexity of incentive systems within firms. Since the
monitoring of output from human assets is “complex and costly” recourse is taken to
monitoring the output from physical assets and basing incentive schemes on that. In this
way workers can be rewarded for production output and so it is in the interests of the
firm owner to secure a better bargaining position through being the owner of the
physical assets. This becomes more problematic when the output is dependent solely
upon the human assets, such as in design work. As Poppo and Zenger (1998, p.859)
show, the difficulties in achieving this through a market mechanism encourage firms to
internalise knowledge-based activities:

171
“Markets, by contrast, lack the capacity for such managerial intervention: when
measurement is highly problematic, markets simply fail.”

Internalisation is, therefore, a system for gaining control by the management through
ownership of labour contracts. This internalises knowledge generation into the firm’s
hierarchical structure and so it gains some measure of control over it. Any unforeseen
benefits that emerge, such as new product innovations, accrue to the firm through its
ownership of the labour contracts. Although this has to be balanced with the loss of
access to innovations occurring externally, and by definition innovations are
unpredictable and of unknowable source, internalisation of human assets means that
they can be treated more like physical assets. Output, though, is more difficult to
quantify for human assets than physical assets and so economies of scale for human
assets are analysed differently. This, in turn, suggests that the manner in which company
functions are integrated depends on the preponderance of either human or physical
assets within that function. The next section of this thesis will investigate how the
functions can be brought together in a prescribed governance format.

5.1.9 Unitary and multidivisional governance structures

Transaction cost analysis highlights the make-or-buy dichotomy in terms of the decision
of whether to internalise an activity into a vertically integrated functional structure or to
transact for it in the market. Chandler (1990) views this as a strategy since it is an
allocation of resources necessary for carrying out long term goals. Once the strategy has
been determined and the functions internalised, then a corporate governance structure
needs to be imposed on those functions that have been brought within the company
boundary. At the simplest level. Chandler (1977) observed that before the arrival of the
railway in the US those firms that were involved in the transportation of goods tended to
leave regulation of the business environment to government legislation. However,
railway transportation necessitated careful coordination of traffic and this was
heightened by the technological advances in the speed of the locomotives. It was
therefore necessary to implement a form of management that was controlled from the
centre.

172
The strategic purpose of the first railways engendered a hierarchical and simple form of
structure, the unitary or U-Form structure. Generally, the simplest strategy is likely to
be one pursued by firms that are small, or singular in their purpose, and so provide
fertile ground for this kind of structure. This may begin with the most senior manager,
often the owner, taking responsibility for all the major decisions which are then passed
directly to those who will implement them. The centralised format of the U-Form
provides a hierarchical structure for administering the functions of the organisation
(Ferguson and Ferguson, 2000). It is designed to reduce opportunistic behaviour by
individual employees through the establishment of collective bargaining that consigns
workers to recognised groups, thereby removing the possibility of acting to their own
advantage. Personal advancement is then offered in the form of promotion, with new
entry being restricted to the most basic levels of job definition (Moschandreas, 1994).
The basic structure of such a firm is shown in Figure 5.2

Figure 5.2 U-Form functional governance structure

Chief Executive Officer

(CEO)

Production Marketing Finance Human

Resources

Source: Moschandreas, 1994

In the U-Form structure the functions have functional managers who gather information
which they present to the CEO and await instructions. The CEO is therefore responsible
for the daily operations of the company as well as its strategic direction. This means
that managers specialise according to their function and communication is conducted
functionally, meaning that information passes up the hierarchy to the CEO but not
across to other functions, except through the CEO. The system assumes that the CEO
has enough experience of the functions to be able to make informed operational
decisions, but it also implies that the functions will work together under the direction of
the CEO. Moschandreas (1994) argues that in practice

173
the U-Form is vulnerable to “empire building” where functional managers attempt to
cultivate as much influence with the CEO as possible in order to further the interests of
their departments or even their own personal interests at the expense of profit
maximisation.

The U-Form is considered applicable as long as the numbers of functions under the
CEO are not too diverse and each function can discharge its responsibilities. However,
should a company internalise a greater variety of functions then it becomes
progressively more difficult to coordinate them as if they were facets of a single
business. For example, a company might have two manufacturing functions, one
producing pharmaceuticals and the other cosmetics. The marketing requirements for
each might be so distinct that a single marketing function would not be suitable and a
way would have to be found to divide its operations between the two manufacturing
functions (Ricketts, 1987). In this way diversification brings with it complexity and the
need to coordinate the different functions, meaning that the CEO has less time for
strategic decisions. One solution is to add layers of bureaucracy but this can reduce the
flexibility of the enterprise and encourage opportunism (Moschandreas, 1994). Hill
(1985) also puts forward the concept of the H-Form, a holding company with
centralised profit collection but decentralised strategic, financial and operational
control. However, this does not seem to offer an organisational structure as such and
might be better conceptualised as an investment vehicle.

Chandler (1977) observed that diversified, multiunit enterprises developed a middle

management to monitor functional performance and coordinate material flows while the
senior management evaluated information flows from the middle management and
allocated strategic resources accordingly. This supplanted the role of the market
between independent organisations such that:

“...the visible hand of management carried on the functions hitherto performed by

market mechanisms in American industry.” (Chandler, 1977, p.377)

This form of structure was the multidivisional M-Form, a decomposition of the U-Form.
At its most basic it comprises a general office to monitor the operating divisions and
implement strategic allocation of resources. The M-Form is a

174
multidivisional organisational structure of bureaucracy that separates strategic and
tactical planning, delegating operational control to internal divisions while a central
headquarters retains strategic and overall financial control. This prevents divisional
managers from diverting funds opportunistically to their own operations and frees
senior managers from making detailed operational decisions for which they lack
information (Freeland, 1996a; 1996b). As Klein (1988) and Hoskisson et al (1993) also
noted, ownership of labour contracts means that any errant staff can be more easily
dealt with than through the mechanism of the market.

With the headquarters having intimate knowledge of the financial affairs of each
division it can allocate and monitor capital more closely than could shareholders
through the external capital market. This means that it is more efficient at allocating
resources, although if economies of scope existed between the functions when they
were in the U-Form structure then they will be lost in the M-Form. Furthermore, when
administrative functions such as human resources and accounts are distributed amongst
the divisions they may lose access to the economies of scale that existed when they
were unified in the U-Form. The M-Form therefore has the advantages of being able to
allocate resources, inhibit opportunism and encourage innovation within functions but it
may also increase the costs of operating those functions. The M-Form of organisational
structure is illustrated in Figure 5.3 below with examples of the functions that are
included with the divisions.

175
 Figure 5.3 M-Form multidivisional governance structure

General Office
Senior management

Division A Division B Division C

Production Marketing Human

Resources

Source: Moschandreas, 1994

The M-Form originally came to prominence as a system of bureaucracy used by Alfred

P Sloan to control company divisions. Sloan had made his reputation with the Hyatt
Roller Bearing company, purchasing it in 1899 for $5000 and selling it to William
Durant in 1916 for $ 13.5m. The absorption into GM put Sloan in control of the
disparate United Motors component supplier subsidiary, which he then reorganised into
semi-autonomous business units. With the near financial collapse of GM in November
1920, GM’s president, Pierre du Pont, invited Sloan to set about a similar strategy of
reorganisation for GM as a whole (Chandler, 1990).

Sloan identified the main problem at GM as being lack of control, a classic transaction
cost issue, with high levels of inventory and a confused product line at a time of
economic recession (Sloan, 1986, p.42). This was tackled by drawing together the
separate divisions under the command of a centralised administrative structure to ensure
more effective co-ordination and control. Conversely, rival automotive company Ford
had taken a highly integrative approach that was supported by a rigid system of controls
throughout the company which precluded a unified central control. It also exhibited an
overdependence on existing structures to the degree that it remained inflexible in the
face of GM’s rise (Kuhn, 1986). Henry Ford had originally shown great innovation in
creating his company but in fashioning it in the image of production flows had failed to
apply a strong corporate governance structure.

176
The form of loose vertical integration at GM was based on three principles (Sloan,
1986, p.50):

1. A degree of independence for each division so that it had a sense of its own
responsibility and contribution.
2. Appropriate measure of return on investment for each division to reveal the
level of its efficiency.
3. Centralised decision making on additional investments with regard to their
utility to the group as a whole.

The second two items are concerned with financial controls while the first recognises
the advantages of the human element, described by Williamson as being at the root of
opportunism, but here it is presented as having the potential for a positive contribution.
In other words, opportunism is seen as a positive attribute to be nurtured and so
illustrates that the human element is a symmetrical risk with both negative and positive
outcomes possible. Williamson (1975) puts forward three strategic control systems
inherent to the M-Form:

1. Internal incentives and control to align personal interests with those of the firm.

2. Internal monitoring system reporting to senior management that is more

comprehensive and prompt than an external audit for the benefit of
shareholders.
3. Resource allocation based on objective evaluation of profitability

In practice, the division of responsibility in an M-Form corporation may not be as

distinct as the theory sets out. Assuming that the personnel who are stationed within the
headquarters have been promoted up the corporate hierarchy, rather being trained purely
as strategists, then it might be expected that they would wish to exploit their experience
of operational matters. As Moschandreas (1994) points out, senior managers may
become involved in operational matters rather than restricting themselves purely to
strategy. Furthermore, they may not be interested in profit maximisation for the benefit
of shareholders but simply a satisfactory return. There is also the possibility of
opportunism within an internal incentive system amongst

177
divisional managers who are eager to protect their own interests. Finally, the divisions
themselves may become so big that they take on the same disadvantages as large U-
Form firms, necessitating further divisionalisation.

In order to test the empirical evidence for the M-Form researchers have looked at how
it may be applied in new and different situations. Chang and Choi (1988) took a
transaction cost approach to analyse company structure with regard to the state of the
market. This research focused on Korean companies operating in the poorly developed
domestic market, one that can be assumed to contain more market imperfections than
would those in more developed countries at that time. Korean business groups, known
as chaebols, exhibit vertical integration and diversification, enabling them to pursue
both economies of scale and economies of scope. Although business groups are not
confined to Korea they have been remarkably conspicuous in that country, taking
40.7% of total manufacturing shipments by value six years before the study took place.
Donaldson (2001) argues that the M-Form reduces uncertainty by internalising markets,
and this can be observed in Korean business groups. However, this uncertainty is based
upon the poor technical development of the resource market rather than uncertainty
created by high rates of innovation. Korean firms can be seen to diversify in order to
create a market rather than internalise one that already existed. Added to this is the fact,
conceded by the authors, that the Korean groups at the time of the study enjoyed
substantial external government support.

Hill (1988) suggests two types of diversification: related diversification and unrelated
diversification. If the diversification is related, i.e. the divisions work with each other in
the same industry, then the organisation is structured to both differentiate the functions
so that they may specialise, while also internalise them to reduce transaction costs. Such
a multidivisional form would need more operational control of the functions from the
centre, an organisational structure Hill terms the CM-Form. Alternatively, if the
diversification is unrelated then the purpose of the M-Form is to exploit the economic
benefits of an internalised capital market. According to Teece (1981) control of capital
in unrelated diversification is effective because the senior management is privy to
financial information to an extent that external shareholders are not, and is able to
monitor and act on investments more swiftly and cheaply:

178
“Furthermore the M-Form creates its own internal miniature capital market which
replicates external capital market functions and economises on transactions cost.”
(Teece, 1981, p. 175)

U-Form integrated firm also internalises the functions of the market, although it does
not attempt to replicate the capital market to the same degree. The M-Form, by
separating the divisions, creates an environment where the divisions must appeal for
funds based on their investment potential. It might be argued that the same processes
could be achieved within a U-Form structure, perhaps with the CEO personally
gathering and evaluating the data of different functions rather than expecting proforma
information to be passed up the command chain, but this is not a capability that is
inherent to the formal definition of the U-Form structure. However, it does suggest that
the distinction between the M-Form and U-Form is characterised by the formal
reporting of data, rather than the separation of operational and strategic decision
making.

Taking the operational perspective, other literary sources considered how production
was operated within the two basic types of governance structure. Burton and Obel
(1980) found that the M-Form was suitable for parallel processes, when they share the
same resources, but the U-Form was suited for sequential processes that were working
on different stages of one production system. Furthermore, relatively small companies
were found to be better served by the U-Form alone whereas larger companies have a
choice between the M-Form and the U-Form. For these larger firms the evidence
suggested that the M-Form was preferable under all circumstances but most especially
when diversified and undergoing high rates of growth. It is at times such as these that
clarity needs to be brought to the reporting of data, but in a fast moving environment
there is also the difficulty in separating operational and strategic decision making
processes.

Donaldson (2001) draws a similar comparison between the M-Form and the U-Form
structures. The research was conducted within contingency theory which seeks to align
the appropriate organisational structure with the three factors of environment, company
size and strategy. For example, the functional structure of the U-Form has

179
cost advantages in simplicity and specialisation, which then places a high priority on
the pursuit of economies of scale. Alternatively, the M-form decentralises control of the
functions and allows them to operate as semi-autonomous business units, a strategy
which is particularly suited to highly diversified product ranges.

Focusing on the three factors, Donaldson (2001) defines environment as being the
stability of the firm’s situation, a vertically integrated hierarchical U-Form structure
having the advantage when there is little technological change while a more flexible
“organic” M-Form structure is better suited when the industry is in flux. This is in line
with the findings of Burton and Obel (1980) that the M-Form was well suited to
accommodating change. Considering the factor of size, Donaldson (2001) argues that
employment of larger numbers of staff mean that tasks are more repetitive and therefore
amenable to the setting of policies by a distant M-Form headquarters, while smaller
staff numbers tend to be engaged in less routine tasks and senior management can be
more personally involved using a U-Form approach. It might also be argued that routine
tasks need less operational control, suggesting that some M-Form headquarters might
have actually been U-Form but shorn of the need for regular operational involvement,
thus leaving the management to focus on strategic issues.

Donaldson’s last factor, strategy, concerns the range of products being made available, a
diversified M-Form organisational structure being better suited to a diversified product
range. Donaldson claims that contingency theory permits a wider range of corporate
structures depending on the level of certainty and indeed, if Chandler is correct that
structure follows strategy, then there should be as many structures as there are strategies.
There may, though, be limits to the degree of change the M-Form can contain and some
theorists have suggested that the business landscape has experienced a quantum change
due to technological developments and globalisation of markets, leading to a new form
of organisational structure beyond the M-Form. Doz and Prahalad (1991) argue that the
M-Form of structure was inadequate for accommodating the wide disparity between
countries, functions and approaches to business when the company is a diversified
multinational corporation (DMNC). Although the M-Form relieves senior managers of
making quotidian operational decisions they are still reliant on a flow of information in
order to make judgements about resource allocation. Although the information should be
of an objective type it

180
still requires a degree of interpretation and this can be difficult when the data is from an
unfamiliar source, perhaps due to geographic or technical differences, possibly
undermining the validity of the resource allocation decisions.

Bartlett and Ghoshal (1993) redefined the M-Form as a systematic devolution of assets
and responsibility from corporate to divisional levels, suggesting that a new bottom-up
approach would allow greater entrepreneurial and strategic control of the divisions. This
recognises that an organisation is fundamentally a social structure and argues that a
focus on functional structures obscures the roles of the human actors. As a consequence,
the three levels of management (top, middle and front-line) no longer amend
information from strategic to operational as it descends the corporate hierarchy, instead
the front-line becomes the source of entrepreneurial activity, the middle coordinates the
divisions and the top management creates the overall purpose while challenging the
status quo. This has particular ramifications for globally spread companies where the
lines of communication may be too long for effective strategic control from the centre
to the dispersed divisions. As Freeland (1996a) points out, the sanctions and incentives
by which the M-Form headquarters controls the divisions is dependent on the quality of
the information passing between them. Bartlett and Ghoshal’s “managerial theory of the
firm” allows local divisions of DMNCs to exploit local resources and opportunities
within the framework created by the top management. The economies of scope of the
DMNC are thereby retained while economies of scale are effectively localised.

Hedlund (1994) criticises any structural form for being a crude categorisation of
hierarchy that did not account for the detailed changes that take place within companies.
This is particularly relevant when managing knowledge creation at different levels of
aggregation, from the individual to the corporation as a whole. Where a hierarchical
structure such as the M-Form would create institutional divisions between the
knowledge holders, Hedlund suggests that an N-Form structure would be more
appropriate. This is a heterarchical structure that allows each division to behave
entrepreneurially (Ferguson and Ferguson, 2000). Hedlund defined the N-Form as an
attempt to recompose the divisions of the M-Form to allow them to work together
through lateral communication at middle management level and temporary groupings of
joint personnel. Senior management then operates more like that in the

181
U-Form where it becomes involved in operation processes in order to promote cross
links between functions. The N-Form then allows the growth of knowledge in the
corporation, Hedlund referring to it as economies of depth as experience is
accumulated, but this can lead to the firm being introspective and therefore vulnerable
to exogenous developments. Indeed, just as Chandler states that structure should follow
strategy so Hedlund concedes that the choice between the M-Form and the N-Form
depends on the nature of the firm’s environment and that a compromise between the
two structures may be more appropriate.

A compromise between organisational structures might be achieved by having two in

place concurrently. For example, a hybrid of the M-Form and U-Form shows the M-
Form decomposition of divisions in order to separate strategic and operational decision
making. As before, this preserves the system of internal controls based on vertical flows
of information in order to conduct efficient allocation of resources. Into this is
introduced some of the functional character seen in the U-Form, such as marketing or
human resources, which might, depending on the organisation, be better able to serve
the company as a whole (Moschandreas, 1994). Such a hybrid structure is shown in
Figure 5.4 below.

Figure 5.4 Hybrid U-Form and M-Form governance structures

Senior
Management

Human
Resources

Division A Division B Division C

Source: Moschandreas, 1994

If the divisions need to overcome some kind of separation problem, be it spatial or

technical, then a matrix approach can be implemented. When disparate functions

182
would benefit from being coordinated with each other the matrix permits two lines of
communication: vertically through the hierarchy, and horizontally between the
functions. It is not an alternative form of corporate structure since it can exist within an
M-Form, for example, but it does have implications for the degree of decentralisation
and the role of managers (Ferguson and Ferguson, 2000). It does not have a preset
format and Kolodny (1979) considered the matrix to be a flexible coordination device,
the level at which it was applied in the enterprise depending upon the nature of the
information processing demands. The most well known application of the matrix theory
is the aircraft industry where the range of technical processes requires a multidivisional
M-Form, but since the divisions are working towards unified projects targets, often in
an exploratory or research environment, it is necessary to establish close
communication between the divisions. In application the matrix system is often
transitory, tending to be shaped according to the nature of the project, with the divisions
reporting to both the divisional management and the project management as necessary.
Costs occur, though, when simultaneous projects duplicate efforts being made
elsewhere or the command structure becomes too complex.

It is feasible, then, that the matrix system could even be applied to a U-Form enterprise.
This would then bring the functions into closer contact, particularly useful if the final
product involves different production technologies. In essence, the choice of
organisational structure then comes down to the U-Form or the M-Form, the M-Form
being the one that seems to attract the most attention. Yet if the purpose of management
is to control the activities of a company, then it could be argued that the U-Form simply
describes the system of command, each level holding responsibility according to its
position in the system. Thus for a large company, lower management would tend to have
more operational control while higher management would tend to have more strategic
control. Conversely, the purpose of the M-Form is to formalise the transfer and
evaluation of data within a large or complex organisation. The fact that senior
management then use the data to make strategic decisions is due to their elevation in the
command system rather than the official definition of their role. Indeed, the need to keep
strategy and operations in alignment suggests that senior management would have a
close interest in operational matters, albeit through the proxy of lower managers.

183
The current automobile industry presents particular challenges to the M-Form structure
since it is made up of technically distinct functions which are required to work closely
on joint projects almost as complex as in the aircraft industry, yet they may also be
geographically spread. The following section of this chapter will investigate how these
considerations impact on the corporate governance of automobile manufacturers.

184
 Part II: Constructing the Automobile Industry Paradigm

Part II will draw on the theoretical basis of paradigms discussed in Chapter 2 (see 2.2.3
above) and use it as a conceptual framework for the automobile industry. The resultant
model provides an organisational structure of vertical integration and output size for the
prototypical automobile company. Chapter 2 described the role of paradigms in science
and the rest of this chapter will construct a paradigmatic view of firms in the automobile
industry. Industries can be considered to be paradigmatic when they have developed
within confines as much as scientific paradigms. The parameters are set by the
organisational vertical integration and the production technology, this giving rise to the
economies of scale determining the present limitations and future potential of the
technology. Technical progress is then demonstrated by a process of puzzle solving,
exploring the limits of the technology, which brings the manufacturers closer to the
theoretical output optimums. Organisational structure will arise by a similar iterative
process as companies converge on a commonly held degree of vertical integration.

Since car manufacturing is constrained by parameters of scale and integration it is

proposed that this can be viewed as a paradigm. If this is the case then development of
the industry has not been driven by progressive linear innovation for its own sake but by
a process of internal problem solving. The paradigmatic view of the automobile industry
demonstrates how companies must adhere to the strictures of the paradigm in terms of
scale and vertical integration in order to operate at their optimal levels. This assumes
that there is no quantum change in the underlying technology to cause a paradigm
revolution, which in any case would have no effect on the internal characteristics of the
current paradigm. The approach to be taken here will begin by defining the structuring
influences on contemporary automobile firms.

185
5.2 Constructing the automobile industry paradigm

Economies of scale are most closely associated with physical assets since they are often
based on capital investments designed for volume production. Human assets are
constrained in their output and may rapidly exhibit diseconomies of scale, particularly
when related to management factors due to the fixed nature of managerial capital (Levy,
1985). In either case, as has been seen in the previous chapter, there is a point of
equilibrium where economies of scale are being exploited to the full. A firm can access
these cost benefits by owning the physical assets of that process or buying from a
company that does so. Lyons (1995) found that UK engineering firms tended to access
economies of scale through the market mechanism, yet this would also imply additional
costs of making transactions. It would only be to the firm’s advantage, therefore, if the
marginal benefit of accessing economies of scale through the market exceeded the
marginal cost of transacting in the market. Furthermore, it might be presumed that
transacting in this way will always provide some access to the supplier’s economies of
scale.

D’Aveni and Ravenscraft (1994) investigated the effect that vertical integration had on
costs when economies of scale are available internally and found that additional
economies were uncovered even after controlling for the effects of economies of scale.
However, the authors also found that internalisation insulated the company from market
pressures and so a slackening of incentives to minimise costs allowed inefficiencies to
emerge. McAfee and McMillan (1995) concluded that much of this effect was due to the
chance for opportunism as the company grew larger and that the hierarchical structure
creates a distance between the management and the workers that have the knowledge.
This results in a principal-agent stand-off:

“The holder of the information exploits the bargaining power the information gives him,
earning rents; in anticipation of this, the principal manipulates the outcome.” (McAfee
and McMillan, 1995, p.406.)

The authors see these rents as a necessary evil, the internal “lubrication” to the
mechanical fiiction of transactions. This, then, sets an upper size limit on the firm and

186
reverses the conventional view that an industry is competitive when the firms within it
are relatively small, the authors arguing:

“Thus firms are small because the industry is competitive.” (McAfee and McMillan,
1995,p.402.)

This suggests that integration is not a simple matter of internalising economies of scale
enjoyed by an external party because in the process the advantage will be countered by
rising costs elsewhere. The decision whether to ‘make-or-buy’ depending on the
transaction costs is what limits the vertical boundaries of the firm, but implicit within
this is the trade-off between economies of scale in production and diseconomies of scale
in the hierarchy. This points to a standard firm size and organisation when three
conditions are satisfied:

• The production technology underlying the quantification of economies of scale

is accessible throughout the industry.
• The transaction cost considerations are common to all participating firms.
• All the human assets behave in largely the same rational manner within the same
organisational structure.

The previous chapter has outlined how economies of scale influence firms within the
automobile industry, providing a reference point towards which companies converge.
The first part of this chapter has shown that it is the rational behaviour of the human
assets that induces an additional cost, a transaction cost, which then forces the make-or-
buy decision. This is driven by a vulnerability to specific assets which may have the
potential for ex post advantages that accrue asymmetrically to the owner of the assets,
offering them an opportunity to create a hold-up. Since no contract can account for all
contingencies, if the costs of such opportunistic behaviour are relatively high then the
firm will integrate the activity into its hierarchical structure. Since opportunism has a
human source then integration is primarily concerned with obtaining control of human
assets through labour contracts, although this may then result in a principle-agent
conflict instead.

187
There is, though, the possibility that the human actors may not act as rationally as
economic theory might suppose. Williamson (1973) conceded this point in listing the
three human factors that influence transaction costs. Bounded rationality for the human
limits of knowledge and opportunism for the limits of human morality, but then
“atmosphere” acts as a sociological limit to acting upon information. While the human
actors may become aware of information that points to an economically advantageous
action, there may exist social mores that inhibit them from exploiting the advantage. As
Williamson (1973, p.317) states:

“Individuals who value independence highly may favour markets over hierarchy, while
others may favour internal organization because of associational satisfactions which
they derive.”

This creates a problem when proposing a model for an industry since it must
accommodate those individuals who might wish to base their selection of organisational
structure not on economic factors alone but also as an expression of their individuality.
Yet for economies of atmosphere to be crucial in determining organisational structure
they must be powerful enough to overcome economic factors, not just influence them.
Whipp and Clark (1986, p.33) noted that US corporations exhibited a predilection
towards productivity over innovation, but also conceded that this was “well suited to the
American market”. Williamson (1996) gave some indication of the extent of atmosphere
by sub-dividing “self-interestedness” into self-serving opportunism and frailty of motive
and reason. This frailty affects only quotidian operations:

“If they slip, it is a normal fiiction and often a matter of bemusement.” (Williamson,
1996, p.49)

Such frailty is hardly of a magnitude that would seriously threaten the economic
structuring of an organisation. This thesis views atmosphere as being of a comparable
intensity, springing from personal preference. Should a participant in a business be
swayed by such personal concerns over economic factors one might even wonder if this
will inevitably inhibit the sustainability of the firm, given that the financial viability
ultimately rests upon the management of economic costs rather than personal

188
preference. This might even suggest that economies of atmosphere are more relevant to
corporate failure, particularly if they involve cases of management that is reluctant to
wholly embrace the economic argument. In any case, since individual preferences are
highly variable, in sum they may be inconclusive, at least in larger firms. This is not the
case for the other major human factor, opportunism, which will only vary in degree but
will always have the same directional vector (i.e. a temptation to exploit the business
relationship for personal gain). In proposing a general model for an industry, therefore,
it is valid to focus on the technical and organisational cost factors that are generally
prevalent.

Assuming that firms in an industry are faced with the same technical and organisation
costs then a common type of economic foundation should become apparent. Output is
prescribed by the presence of economies of scale, while vertical integration is
prescribed by the presence of transaction costs resulting in a full-function structure (see
Figure 5.1 above). The next sub-section of this thesis will investigate how the two
forces of transaction cost economics and economies of scale create a paradigmatic
structure for firms in the automobile industry.

5.2.1 Paradigmatic view of the automobile industry

The paradigmatic approach can be applied to any of the major functions of automobile
manufacturing depending on the distinct nature of the underlying technology.
Nieuwenhuis and Wells (1997) termed one process in automobile production, that of
body panel pressing, as the Budd Paradigm and believed that as the process that defined
the final product and the economies of scale for the industry it therefore structured the
industry (see Chapter 4). Later developments in powertrain, particularly in the field of
electric vehicles, show that the Budd Paradigm is not one that rules over the production
of vehicle propulsion systems so it cannot be considered to encompass the entire
production process, only that connected to body fabrication and painting. However, the
Budd Paradigm predominates since it is this that sets down the MBS for the firm, other
processes being multiplied to equalise with the production of car bodies. This implies a
structure for an automobile company but it does not explicitly address the issue of
vertical integration which as has been

189
suggested is driven by transaction cost analysis. Thus, the Budd Paradigm does not of
itself imply a size and shape for a prototypical automobile firm.

When the different functions of automobile production are defined by a common form
of technology then the fixed costs are effectively common throughout the industry. If
markets for sourcing the factors of production are generally equal for all participant
companies, as they would be for a capital intensive industry such as this, then the
transaction costs in procuring those resources will also be common to all companies.
This means that there is an optimal structure to the extent of vertical integration for the
firm just as much as there is an optimal output for prescribed economies of scale.
Considerations of scale and vertical integration therefore arrive at a fixed size and shape
for a prototypical automobile manufacturer. Structural differences would only arise
where there are differences in the fundamental technology, which then changes the
capital costs, or else changes in the factor endowments. Since labour affects mainly final
assembly it does not have an effect on the production technology in the body or
powertrain fabrication processes. R&D is also more concerned with the knowledge
contained within the human assets rather than labour costs per se. Labour costs therefore
have the greatest influence on the internal structuring of the final assembly process.

Developments in technology have the potential to change the paradigm by affecting the
economies of scale and the transaction costs. It has been suggested that new powertrain
systems might change the industry but so far this has been confined to hybrid engines
which use electric motors only in an auxiliary role. Were there to be a wholesale switch
to electric forms of propulsion then, although it would not necessarily affect body
production, it could change the manner of the integration with powertrain production.
Car firms might find that there were cost advantages to be gained from divesting
themselves of their new electric powertrain production units and instead sourcing
production externally.

The Budd Paradigm, as applied to BIW production, would be unaffected by changes in

powertrain production technology, though a new MBPS for powertrain would have an
effect on the lowest common multiple when matching the scale of the different
functions. For the automobile paradigm being described in this chapter, which

190
includes scale and the full-function model of organisation, outsourcing of any function
would have an impact. Indeed, this would represent a move to a new automobile
paradigm, the paradigm revolution described by Kuhn. Essentially competitive
paradigms can remain side by side since both provide acceptable solutions to their
practitioners in respect to the problems they define. As the new paradigm proves itself
more adept at solving a wider range of problems it gathers greater support. Kuhn claims
of those that adhere to the superseded paradigm, they:

“.. .are simply read out of the profession, which thereafter ignores their work” (Kuhn,
1970, p.l9).

For automobile production, in the early twentieth century the current paradigm
superseded the craft production paradigm where coachbuilt wooden bodies were
attached to separate steel chassis. It is not that craft production has disappeared
altogether, for low volume production or individual customisation it is still an option, but
it has been bypassed as an effective means of manufacturing standardised mass market
vehicles on a large scale. Indeed, puzzle solving can carry on in the preceding paradigm,
even if not with quite the same intensity as in the new paradigm. As Kuhn states:

“Mopping-up operations are what engage most scientists throughout their careers.”
(Kuhn, 1970, p.24).

A paradigm is successful while the adherents continue the “puzzle solving” activities.
This is the fruitful part of a paradigm and the existence of, as yet, unexplained
phenomena are viewed, not as weaknesses, but simply anomalies awaiting further
research. The conditions for a revolutionary change to a new paradigm are set when the
adherents of the existing paradigm find that it no longer adequately matches nature and
so lose faith in it (Chalmers, 1983). For industry, these problems arrive in the form of
exogenous shocks but not even dramatic rises in oil prices have done anything other than
shift market preferences for cars. Not only did the oil price increases have no impact on
the fundamentals of the narrowly focused Budd Paradigm, but also no impact on the
automobile paradigm put forward in this chapter because the powertrain function is
essentially unchanged. In the meantime, these

191
exogenous shocks have served to further motivate internal problem solving. It was one
such exogenous shock that motivated Honda to design the CVCC lean-bum engine in
order to meet emission regulations in the US (Demizu, 2003), and this represented the
development of existing technology rather than a technical revolution.

Since paradigms are viewed as sociological constmcts their characteristics can only be
indicated, not proven. Moreover, those within the paradigm may be unwilling to
recognize its existence when it might suggest their research is thereby constrained.
However, paradigms aid comprehension of areas of activity since a conceptual
framework can be applied, along with the methodological and predictive powers of the
paradigm. Applied to an industry, this would prescribe a future stmcture that could be
superseded only by a paradigm revolution based on a change in the fundamental
technology.

Such a paradigmatic perspective on the automobile industry is only valid if the

underlying technology allows it. It can then be demonstrated how this is organised
according to an established approach to arrive at a prescribed stmcture for a
paradigmatic automobile company. This implies that any car company operating with
the same fundamental technology would be govemed by the same paradigm and so
would be drawn to the same ultimate stmcture. Any company within the paradigm that
did not achieve the prescribed scale and stmcture would be deemed to be operating at a
sub-optimal level (i.e. operating at above minimum cost).

5.2.2 Vertical integration of the automobile paradigm

When a company takes more responsibility for production processes by intemalising the
supply chain it is said to be increasingly vertically integrated. A vertically integrated car
manufacturer is one that has intemalised all the major functions of automobile
production. Car manufacturers have often explored the limits of vertical integration, for
example, Henry Ford built a steel mill for his new factory at River Rouge (Rae, 1984).
However, the purpose of this thesis is not to include every last detail of automobile
manufacturing but to investigate the main stmcturing areas. The perspective taken of car
firms in their generic form is therefore in accordance with the full-function model
previously put forward (see section 5.2).

192
The company boundary of a typical full-function automobile company with its four
modules of production is shown in Figure 5.5 below. This is a functional schematic of
the vertical integration of the firm; it does not imply actual plants or their geographic
positioning. The Design function instructs the Powertrain and Body functions, the
output from these feeding into the Final Assembly function. The final products are
released to the market, which is external to the functional operations of the company,
and information from it is fed back to the company’s Design function by way of a
marketing capability which is part of the R&D activity.

Figure 5.5 Typical full-function automobile producer

Car company .............................................................

Design

Body Powertrain

/
Final
Assembly

Market

In theory, any of the major functions can be external, including Design and Final
Assembly. However, from an historical perspective, the default position for a car
company before it vertically integrates has been to commence with design and final
assembly functions. In this way the vehicles are conceived and put together by the new
firm, sourcing all components externally. Rare examples of production existing before
the R&D capability include Tata of India, which had its Indica small car designed by
IDEA in Italy, although Tata had control over the parameters of the design process (SID,
2006). However, as will be demonstrated, most mass market firms start with the Design
and Final Assembly functions at least, and then integrate Powertrain and Body as the
business develops.

193
Since such an integrated structure takes time to develop and become established it is
necessary to research archive data. The fundamental technicalities of automobile
production were first introduced at the beginning of the twentieth century with the first
experiments with flow production systems and the rise of mass production. This vastly
improved assembly methods and brought efficiency benefits to the existing forging and
casting techniques used for heavy components like powertrain. Finally, body production
was revolutionised by the Budd system of manufacturing load bearing welded steel
bodies that brought clear constraints to the style of vehicles that could be produced and
indicated what the economies of scale might be. Once the technicalities of production
were defined it was possible for a stable organisational structure to emerge. The process
by which vertical integration occurs is not a static one and it is as possible for firms to
divest themselves of their functions as it was to integrate them in the first place. By this
reasoning, if functions remain intemalised then, rationally, the transaction costs must
continue to be minimised. The purpose of this section of the thesis is to build a
representation of the core functions of automobile production as they have become
established over time, whilst also noting how they might have been adjusted in detail to
take account of the contemporary conditions.

5.2.3 Defining the functions

As a multi-stage process, automobile production involves inputs from a myriad of

component suppliers. Wells and Rawlinson (1994) concede that there are difficulties in
identifying the limits of the supply industry, the numbers of firms or even whether they
can be considered automotive or not. However, for the purposes of this thesis, it is
important to define how these firms relate to the automobile assemblers so that their role
outside the vertically integrated full-function structure can be delineated.

Even for a full-function firm operating in the four core areas there will be component
suppliers contributing at the sub-functionary level. These range from producers of small
individual parts to larger sub-assemblies that are built up from a collection of
components before the completed unit is fed into the main assembly line. The
illustration of a full-function automobile company (Figure 5.2) has omitted the
substantial role played by this supply industry. Firms in the supply industry can be

194
categorised into three types according to the nature of their output and its use (Odaka et
al., 1988):

1. OE (original equipment) for installation on new products.

2. RP (replacement parts) or SP (spare parts) to replace worn out parts.
3. Reconditioned parts, as low cost substitutes for RPs and SPs.

Since OEs are made to the design or requirements of the vehicle assembler, unlike the
other two categories, only supplies of OEs are of direct interest here. A noticeable
feature of this ancillary industry is its tiered structure of primary and secondary
suppliers, the primary or Tier 1 suppliers often being sub-assemblers using parts
manufactured by the secondary or Tier 2 suppliers, although some secondary parts may
be procured directly by the vehicle assembler. There is a tendency in Japan for suppliers
to be tied exclusively to one producer, as exemplified by the keiretsu structure. This may
also include a degree of supplier selection within an obligation model of the supplier-
buyer relationship wherein a preferred supplier is groomed for a high profile role. The
supplier retains semi-autonomous control over production while the buyer can influence
product specification, R&D and investment decisions (Morris and Imrie, 1992). This is
analogous to the M-Form of intemalised organisational stmcture in that the supplier has
operational control while the buyer sets the overall strategy and capital investment
commitments.

In Europe the relationship tends to be more polarised, suppliers being either majority
owned or independent of the car producer (Wells and Rawlinson, 1994). When the
supplier is independent this can lead to an adversarial relationship with cost reduction as
the battleground. This operates through a tendering system that promotes a divided
supply industry that is unable to make long term commitments to R&D or economies of
scale in production. Smitka (1991) notes that Chrysler operated a policy of short term
contracts with suppliers that kept costs down through intense competition, but inhibited
long-term investment in the supply industry and so stifled development. Despite this,
Carr (1993) found little correlation between UK supplier profitability and production
volume, instead sourcing the decline in the supply industry to the 40% rise in the
currency exchange rate in 1979 which forced domestic suppliers to hold prices for four
years in order to resist foreign competition. This left little for

195
reinvestment, although the study found that GKN was able to maintain world leading
R&D expenditure up to 1983. This suggests that the connection between price tendering
and low investment is not a direct one. Moreover, a competitive market is not described
by the price mechanism alone but also that of product features so it is unlikely that price
tendering represented one-dimensional contracts but was more likely to be one aspect of
multidimensional contracts. It would seem therefore, that the problem with the
adversarial relationship is that during periods of change (such as exogenous economic
shocks) low switching costs mean that there is no reason for the buyer to support the
supplier, thereby ending the relationship. It is notable that part of Nissan’s recovery has
been predicated upon an imposed 20% reduction in purchasing costs spread over three
years and a 50% reduction in the number of suppliers (Donnelly et al., 2005).

The varying levels of integration between suppliers and car producers is illuminated by
the range of relevant factors stated by Odaka et al (1988). This includes technological
distinctiveness of the part, production technology, demand and economies of scale. The
core functions of the full-function car producer are technologically separate but closely
tied in terms of the production system. For instance, the body panels are used solely by
that assembler. This is not necessarily the case with other components, which may use
an unrelated technology and may be shared with other car producers. Rhys (1972)
reported that Lucas had a virtual monopoly in the UK for lamps, windscreen motors,
dynamos and ignition coils. Toyota also recognised that some components were of a
generic technology that could be purchased from a variety of manufacturers, while
other components were of a specialised nature and therefore required Toyota
specification and “teaching”, secured by a close capital or financial relationship. The
procurement category of Special Factory Purchasing requires ‘distinctively special
facilities’, leading to possible financial ties in the future (Lamming, 1993, p.24).

The main consideration in this relationship between supplier and vehicle assembler is
govemed by transaction cost analysis. For generic components the car company has
purchasing power and can switch suppliers for a variety of reasons, be they cost,
delivery, quality and so on. As Odaka et al (1988) have pointed out, the supplier is
operating at a different technical, productive and scale level to the car producer. For

196
this reason, a fuel system supplier is sub-functional to the powertrain production which
a gearbox or engine supplier is not. For each function it has been seen that there is a
critical MBPS and the relationship with the supplier is functionary when the supplier is
responsible for this scale of output. This is a proxy for asset specificity since although
physical assets may not be specialised their workload is to the degree that without the
contract to supply they would be redundant. Monteverde and Teece (1982) also discuss
the crucial role played by the human assets where the supplier is engaged in research in
support of the car manufacturers own R&D, arguing, like Klein (1988), that this will
lead to vertical integration. As will be seen in the following subsections, when suppliers
occupy a complete function such as powertrain production then they tend to become
intemalised into the automobile manufacturers stmcture.

5.2.4 Integration with powertrain

Extemal supply of powertrains was common in the early years of the automobile
industry. Thoms and Donnelly (1985) focussed on the cradle of the British automotive
industry, Coventry. White and Poppe was the main engine supplier in the city and in
1906 the company was supplying 15 car manufacturers. In 1913 engine output reached
2,000 units but it progressively lost business to car companies seeking to vertically
integrate powertrain production. Daimler, for example, produced its own engines and in
1908 planned for additional output of 1,000 engines, of which 400 to 600 were destined
for Rover. White and Poppe was forced to invest in specialised capital as technology
progressed but this created an asset specificity problem with Singer extracting credit
terms from the company in 1912. The directors at Standard were aware of the power
they had over White and Poppe, agreeing that if:

“...conditions warrant it, deliveries can be suspended at any time according to

requirements.” (Thoms and Donnelly, 1985, p.88)

Engine maker Hotchkiss and Cie was closely involved in supplying Morris and in the
two years after being vertically integrated into Morris output quadmpled (Thoms and
Donnelly, 1985). In the US, since a sufficient supply base had yet to emerge in the new
industry, Henry Ford founded the Ford Manufacturing Company in 1905 to

197
ensure reliable supply of engines to required specifications and prices, then
consolidated with the Ford Motor Company a short time later in 1907 (Hounshell, 1984,
p.220).

Since most of the vertical integration with powertrain suppliers occurred in the early
years of the industry there are few instances of it occurring in the modem era, even if
that is taken back to 1945. This research is not aware of any significant automobile firms
divesting themselves of their powertrain functions. Later entrants, such as Honda, began
as an engine manufacturer in 1946 and then branched out into motorcycles in 1949 and
then cars in 1967 (Honda, 2006). In Korea, Kia began engine production in 1973, a year
before it began production of its first car (Kia, 2007). The Chinese manufacturer Chery
was established in 1997 and began engine R&D at the same time, designing the
ACTECO engine family in partnership with engineering consultants AVL of Austria;
annual engine production capacity is 400,000 (Chery, 2007).

Extemal engine suppliers operate now on a contractual basis with limited output, VM
Motori of Italy supplying diesel engines to three major groups (DaimlerChrysler, LDV
and BMC) with licensed manufacturing granted to Hyundai in Korea (VM Motori,
2006). It is quite common, however, for powertrains to be supplied from one division to
another within a single group, such as VW’s W12 engine (in its basic form) to Bentley
Motors.

5.2.5 Integrating body production - the case of GM and Fisher Body

Body production is an interesting example of integration because the coachbuilding

industry preceded the emergence of the automobile industry and so the early automobile
manufacturers were able to source bodies extemally from an established market.
Subsequent innovations in body production that led to the load bearing steel body
created a revolutionary upheaval in the industry which was addressed by intemalising
the body production function into the automobile manufacturers. In the case of GM and
its body supplier, Fisher Body, this has become a celebrated example of intemalisation
and resulted in the application of the M-Form of organisational stmcture. It also brings a
critical perspective to concepts of economies of atmosphere

198
and geographic proximity of manufacturing facilities. Subsequently, most automobile
manufacturers have retained this function internally, although there has been some
limited growth in contract manufacturing of steel bodies.

In the craft era of automobile production the vehicle bodies were commonly
manufactured by large numbers of extemal suppliers able to produce in small volumes
to customers’ specific requirements. Automobile firms were, in the main, assemblers of
outsourced components, the ultimate shape and purpose of the final vehicle being
defined by the coachbuilder. Even when the automobile firms reached mass production
levels of output the separate body could be manufactured extemally in a plethora of
different body styles, albeit to automobile firm specifications rather than that of the
coachbuilder.

The emergence of the Budd Paradigm changed that because the system resulted in
continuous high volume production of standardised bodies which defined both the shape
and purpose of the vehicle. According to transaction cost theory, the high asset
specificity and high uncertainty during the changeover to the Budd system would have
culminated in vertical integration (David and Han, 2004). Accordingly, the absorption of
Fisher Body by GM from 1919 to 1926 was one of the most notable cases of vertical
integration in automobile industry and an example of Chandler’s view that stmcture
followed strategy. Although this took place during the early years of GM’s formation
into an M-Form style organisation, the essentials of the Budd system were being put in
place and they have not fundamentally changed since that time. The steel panels were
stamped out and welded in a manner that cannot change due to the nature of the
material, although there has been some refinement of the process. For this reason the
fundamentals of the story still shed light on the industry today, Freeland (2000)
describing the case as “paradigmatic” and Klein (2000, p. 107) states that:

“...we can leam a great deal about the economics of contractual arrangements by
studying the conditions under which the Fisher-GM contract failed.”

The situation in 1919 was that the vast majority of automobile bodies made for GM
were of the open type, only 10% were closed body styles, yet by 1927 the proportion

199
of the latter type had risen to 85% (Sloan, 1986, p. 152) due to the enabling technology
of the Budd Paradigm. This was pioneered by the Essex division which reduced closed
body prices such that between 1924 and 1926 prices of the “Essex coach” converged
with, and then undercut, the open version (Georgano, 2000; Sloan, 1986, p. 159). As
Sloan points out, with Ford pursuing a static policy of minimal product development
with the open body Model T the demand in the market for the new closed automobile
was taken up by competitor automobile producers, Chevrolet’s share of this new
market rising from 40% in 1924 to 73% in 1926.

Fisher Body was a dominant force within the coachbuilding industry and GM was
vulnerable to its supply. GM had tried to secure the relationship with the contractual
approach, taking the precaution of obtaining a 60% holding in Fisher Body. This was
enough to preclude the Fisher brothers, as owners of the voting stock, from merging
the company with other manufacturers, particularly Willys-Overland in Cleveland.
However, it did not give GM operational or strategic control of Fisher Body and supply
of bodies to other vehicle manufacturers continued (Freeland, 2000). GM had also
entered into a ten year contractual agreement sourcing bodies from Fisher Body for a
payment of cost plus 17.6%. According to Klein (2000) this removed from Fisher Body
the motivation to reduce transport costs, since any additional costs were covered by the
contract, and indeed attracted increased profits. This was particularly relevant to GM’s
plant at Flint which was some 57 miles distant from Fisher Body’s Detroit plant.
However, the Fisher brothers were reluctant to invest in a plant in close proximity to
Flint and this constituted an apparent hold-up. Yet Klein (1988) argued that specific
physical assets do not have to be owned by the supplier, the buyer can purchase them
and employ the supplier to use them. Coase (2000, p.29) takes a similar line, pointing
to the funding by GM of a previous Fisher Body facility in order to supply a Chevrolet
assembly plant:

“It is not without interest that it was Fred Fisher who suggested that the body plants be
built on Chevrolet property.”

Freeland (2000) found that the Fisher brothers were reluctant to set up a new site at
Flint, not because they were averse to making the investment but because it would have
made production at their Detroit plant less economically viable. It was much the

200
largest Fisher Body plant, twice as big as the next in size, and enjoyed considerable
economies of scale as it supplied important customers in the immediate locality. Even if
GM had made the investment in a Fisher Body plant at Flint, it would have forced the
Fisher brothers to restructure the Detroit plant to reduce its capacity and render
production there less economic.

This positioning of the plant also demonstrates that geography was already a significant
factor in the industry. Indeed, with the technical principles of the industry established,
proximity of certain facilities would be a constant throughout the future development of
the industry. Although the absolute costs of transport have changed over the period
since the Fisher Body episode, as a cost relative to production costs and competitor
costs it is as relevant now as it was then. GM’s plant was much further from Detroit
than its rivals and with the rising popularity of the closed steel body transport costs
would have come to dominate total body production costs. This might have been
alleviated if bodies were generic automotive components since GM would not have
been dependent on the distant plant. Unfortunately for GM, load bearing bodies are the
least generic because they define the vehicle type. Although other body suppliers were
available there would still have been switching costs due to the specialised nature of the
bodies. It was also not yet clear how far the market would shift towards closed steel
bodies and, with the future of GM in the automobile market at stake, it was vital for
GM to be able to control its supply. As Walker and Weber (1987, p.590) state:

“...as uncertainty rises regarding a buyer’s future requirements and, correspondingly,

regarding potential adjustment costs for suppliers, contracts become very difficult to
write. Producing a component in-house consequently becomes more attractive.”

If Fisher Body’s output had been wholly dependent upon GM then GM would not have
been in conflict with its supplier. Only by being in a symbiotic relationship would a
relatively costless, self-regulating relationship have been possible instead of vertical
integration. As it was, vertical integration was the short-term solution to the extemal
problems of body supply within the new Budd Paradigm. According to Klein, Fisher
Body had found itself in a position where it could hold up GM, forcing GM to

201
purchase the remaining 40% of the supplier at a premium of over 50% to the market
value (Klein, 2000, p. 124).

Klein speculates that the two companies had always been aware that their contract was
“incomplete” but were able to rely on the self-enforcement mechanism of their
reputational capital. This was changed by the unforeseen popularity of the closed body
which presented a critical opportunity for GM to overtake Ford, yet this new
development also gave unprecedented leverage to Fisher Body over GM. The original
contract had been formulated to prevent GM holding up Fisher Body but now GM’s
only escape before the expiration of the contract in 1929 would have been by voluntary
bankruptcy. The stability afforded by the 1919 arrangement was inflicting “rigidity
costs” that could only subsequently be avoided by vertical integration (Klein, 2000, p.
130).

Lieberman (1991) found empirical support for demand variability or uncertainty being a
determinant of vertical integration, although David and Han (2004) found less
convincing evidence. In a stable industry GM and Fisher Body might have sustained a
long standing relationship. Coase (2000) denies that a hold-up occurred because he
cannot conceive of a circumstance when it would have been in the Fisher’s interest to
cause one. The brothers were, in any case, employed by GM by this stage and so it
would have been injurious to their own careers to have acted for Fisher Body against
GM. However, a hold-up does not have to culminate in a complete cessation of supply; it
might involve resistance to fully comply with the wishes of the buyer. This would
certainly fit the predicament GM found itself in where rivals in Detroit were being
favoured by the close proximity of the local Fisher Body plant.

A possible alternative for GM might have been to renegotiate the contract, but this would
have perpetuated the problem of rigidity in an industry that, for GM and its annual
model change policy, required flexible responses until the scale demands of the new
technology had been settled. While economies of scale in body production remained an
externality to GM it would have had no control over it. Although Lieberman (1991)
found that the quantities of product flowing between the contracting parties had no
significant impact on the tendency to integrate this did not recognise the critical role
played by economies of scale. Once the quantities involved

202
in the deal are within striking distance of achieving significant economies of scale then
they become internal to the relationship and it is in the interests of the buyer to have
some control. Coles and Hesterly (1998, p.327) were of the opinion that economies of
scale were crucial to vertical integration when the firm could achieve these economies
internally just as much as a supplier could. Fisher Body was in a position where it could
enjoy economies of scale in the new production system by supplying a diverse number
of car assemblers. Yet this describes a typical supplier-buyer relationship, GM only
needing to internalise Fisher Body if it could better exploit the economies of scale. Since
forcing Fisher Body to invest in the Flint plant would have, in the short term at least,
reduced the access to economies of scale there was no reason on these grounds to
vertically integrate.

As owner of Fisher Body, GM could gain control of the new technology and avoid the
costs of renegotiating contracts at frequent intervals. Although Williamson (1975)
chooses not to emphasise it, frequency is one of the three dimensions of the transaction
cost approach. Klein agrees, concluding that the affair with Fisher Body was not
concentrated on asset specificity alone, but if there is a need for flexibility when asset
specificity is high and reputational capital is low then intemalisation protects the
manufacturer. However, Coase (2000) was of the opinion that the Fisher brothers would
not have wished to risk their reputation because they were supplying bodies widely
throughout the industry.

There is some possibility that Williamson’s concept of atmosphere might have some
relevance here. Freeland (2000, p.42) notes a reluctance on the part of the Fisher
brothers to relinquish control of their family firm:

“Accustomed to the autonomy of mnning their own firm, the Fishers made it clear that

they did not wish to sell their business to another company and to stay on as employees.”

It would be tempting to categorise this as personal preference, and therefore attributable

to atmosphere, but the brothers had started the company as an innovative manufacturer
of automobile bodies, without reference to the preceding generation of carriage makers,
and by 1916 the company was the largest body manufacturer in the

203
industry. The Fisher brothers had also started the Hinckley Motor Company around 1914
to manufacture engines, initially for trucks bound for the European war (Coase, 2000).
In 1919 the Fisher brothers were formulating a plan to enter the automobile production
business themselves, exploring a $10m joint venture with Willys-Overland. Far from
being held together by atmosphere, the siblings appeared motivated by a desire to exploit
the economies of scope in their managerial and entrepreneurial skills, culminating in an
automotive group to rival GM. Since GM lacked the knowledge the brothers had in body
production it was vital to secure their services during a period of uncertainty: it cost GM
$27.6m for the first 60% holding and another $136m in stock for the remaining 40%
(Freeland, 2000). Clearly, the brothers were not defending their family firm for reasons
of atmosphere but because they believed they had a successful business model, and one
with a quantifiable price.

The conclusion of the Fisher Body episode was part of the implementation of the M-
Form organisational structure at GM. The senior management at GM had needed
strategic control over Fisher Body and this then became a constituent part of GM’s M-
Form of structure. The critical component was not the cost of the physical assets at
Fisher Body but the costs of the uncertainty derived from the variability in the market.
Intemalising Fisher Body allowed GM to make labour contracts directly with the
management, initially comprising the Fisher brothers themselves, thereby permitting
greater flexibility at a time when the new production technology had yet to settle (Klein,
1988). Indeed, the day after Fisher Body was acquired by GM, a $5m investment in a
body production facility at Flint was announced while the Detroit Fisher Body plant was
subsequently closed. The strategy set the tone for the industry as a whole and GM’s
vertical integration introduced an industry wide trend (Klein,
2000).

5.2.6 Integration with design

The ability to research and design future products lies at the very heart of any
corporation. R&D is the source of all product innovation and so its vertical integration
within the firm allows it to come into contact with all parts of the firm (Armour and
Teece, 1980). It might be argued that these contacts could be contracted for, but
innovation is opportunism at its most unpredictable and fertile. Furthermore,

204
integrating it with existing production facilities provides a focus for the R&D activity.
Coles and Hersterly (1998) considered the possibility that research can make
unexpected leaps forward, rendering existing technology obsolete. Should the research
be conducted by an extemal organisation then it creates an asset specificity problem:

“New development in technology could render the methods used by the contractor
obsolete before the terms of the contract expire. The contractor, however, may be
unwilling to incur the additional costs of the new technology before the existing
contract expires.” (Coles and Hesterly, 1998, p.326.)

The Design aspect of the full-function model is the one least likely to be extemal in the
first place, therefore least likely to be a target of subsequent vertical integration. This is
because the Design function is the origin of the ultimate product, the other functions
serving to bring the product into being. For this reason, automobile companies that
operate without their own design capability tend to owe their existence to national and
political strategies rather than entrepreneurial innovation. Many developing nations
have attempted to industrialise by inviting foreign firms to enter their markets using the
mechanism of joint ventures with local firms in final assembly. The local automobile
firm may be wholly owned and a standalone assembly facility, but it is dependent on the
foreign firm for the product designs.

Countries where this has been commonly practised include Iran, India and China. Iran
Khodro, the national automotive champion, started out manufacturing British Hillman
Hunters as the Paykan from 1967 under licence from the Rootes Group, continuing
under Chrysler and then Peugeot from 1978. Iran Khodro’s attempts at developing the
Paykan were not far reaching, although it did manage to fit the Peugeot 405 body onto
the Paykan platform to create the 405 RD (rear-wheel-drive) for reasons not specified
(Iran Khodro, 2005). India went in a similar direction, manufacturing the 1948 Morris
Oxford as the Hindustan Ambassador virtually unchanged until the present day.
Ironically, the few that are now imported into the UK have to undergo extensive
modifications to meet current legislation.

205
China has used car assembly as a device for technology transfer. Foreign firms are
limited to maximum shares in joint ventures of 50%, the only foreign majority-owned
plant being a Honda factory manufacturing for export. In partnership with foreign firms
the largest Chinese firms have achieved high levels of output, 911,748 units in the case
of SAIC in 2005, but it has only been recently that indigenously designed and produced
cars have made a significant impact on the market. Even so, they have yet to reach
globally recognised standards, local firm Chery having to postpone entry into the US
market by two years until 2009 due to difficulties in overcoming design challenges
(Automotive News, 2006c). The company is also dependent on numerous extemal
consultancies, such as Lotus of the UK and AVL of Austria. This demonstrates that lack
of geographic proximity is no impediment to the supply of R&D, although companies
usually operate R&D branches in target countries.

There may be some notion that the economics of atmosphere has a role to play in the
promotion of R&D capability by national governments. This can include the elevation of
car firms to the status of national champions in order that they might be catalysts for
wider industrialisation. The implication of the political involvement in stmcturing the
industry in China and Iran is that car companies that do not have control of product
design are in a potentially unstable condition should that support fall away. The
underlying instability, though, and hence the fundamental stmcturing factors in these
cases, are based on the economic concepts that deal with the treatment of assets. R&D is
defined by human asset specificity and advantages resulting from innovation will accme
to the company that owns the labour contracts of the engineers. R&D suppliers therefore
work on a contract by contract basis in order to control the succession of innovations.
This increases transaction costs for the producers so governments compensate for this by
providing long-term support.

The efforts made by governments to promote R&D within domestic companies may be
an effective policy. Zenger (1994) noted that engineers not only bring knowledge with
them to their jobs but they also accumulate knowledge through work experience which
makes them attractive to new employers. In this way the knowledge can become
disseminated throughout the industry. This thesis has also found that R&D teams do not
need to be excessively large. It is difficult for any company to monitor and provide
incentives for knowledge but it costs less to do this in small teams

206
because managers are in closer contact with the engineers, they may even be among their
number, so have a better understanding of performance. It could also be suggested that
since achievement is more readily identified in small teams it is easier for engineers to
gain recognition and status. Zenger states:

“Small firms will attract individuals with superior talent and ideas, and will motivate
higher effort than large firms” (p.713).

Thus, smaller R&D departments will recruit better engineers and then compound the
advantage by having them work more efficiently, though this may mean the firm losing
some access to economies of scope. Large firms, as noted previously, can replicate the
small firm advantage by fragmenting into small teams, buying up small companies or
hiring small teams as outside consultants, while continuing to access economies of
scope. This presupposes that contracts can be written by large companies for those
engineers that are indistinguishable from the contracts that small companies could write,
yet such a condition must be entirely theoretical since, if the two contracts are
inseparable, then the firms must, de facto, be the same. If a firm is defined by its
contractual relationships, as put forward by the theory of transaction cost analysis, then
only a small firm can write a small firm employment contract.

Chapter 4 demonstrated that economies of scale for all the functions in automobile
production are high, but not necessarily inaccessible. The output of the functions can be
brought within reasonable synchronisation of each other, and this chapter has
demonstrated how the functions can be intemalised. The next subsection will detail the
model of an automobile company according to the considerations of scale and stmcture.

207
5.3 Resultant size and shape

Having established the organisational structure of the full-function model through the
study of transaction costs it is now possible to quantify the dimensions from the earlier
investigation of the economies of scale for each function, or productive unit (Whipp and
Clark, 1986). This then results in the proposed prototypical full-function automobile
company model which is at the basis of the automobile industry paradigm. Figure 5.6
shows the four functions with the MBPS for each as in the case of the prototypical
automobile company outlined in Chapter 4 (see section 4.5 above). Continuing from
Figure 5.1, the design function is termed R&D and comprises 1200 personnel working
on 5 model programmes based on 3 platforms with intended output of 600,000 units a
year. Body production is now expressed as body-in-white (BIW), in deference to the
Budd Paradigm, with MBPS of 500,000 units a year. The Powertrain function is shown
with a MBPS of Im units a year while the Final Assembly function has been allotted a
MBPS of 350,000 units a year.

Figure 5.6 Full-function model quantified

Car company ...............................................

R&D
600,000 units
1200 personnel

BIW Powertrain
500,000 units Im units

\ /
Final
Assembly
350,000 units

Market

Figure 5.6 shows that there would be a great deal of mismatch in output capacity
between the various functions if the firm were constructed in this way. The capacity

208
limit in Final Assembly creates a constriction in the production flow that forces the
preceding functions to operate well below the scale shown. However, as the LAC
curves for each function depicted in Figure 4.19 above show, there is a range of plant
sizes which may not be operating at the optimum but they are within the relatively flat
region of the LAC curve. Taking the MBPS of one process as the anchor point the other
processes can be adjusted accordingly. If this were BIW, for example, R&D would be
reduced slightly in size (perhaps 1000 personnel working on 2 platforms and 4 models).
Powertrain would be half the size (500,000 units a year) while Final Assembly would
be reduced to a capacity of 250,000 units a year but with two plants. As subsections
4.3.2 to 4.3.5.1 above show, there are examples of these plant sizes within the industry.

This thesis takes the view, however, that R&D MBPS is the anchor point for the full-
function automobile manufacturer because beyond that point it is inflicted with
diseconomies of scale. For the prototypical automobile company, MBSp is considered
to occur at around 600,000 units of output a year, which is aligned with MBPS for R&D
but will require adjustment in the other functions to match it. This would involve BIW
producing above MBPS, Powertrain well below it (but within the flat range of its LAC
curve) and Final assembly made up of two plants with output of 300,000 units a year.
As is now standard in the industry, each of these functions makes use of flexible
production to manufacture multiple models, usually variants of one another.

209
 Figure 5.7 Automobile industry paradigm

Car company .................................................

R&D
600,000 units
1200 personnel

BIW Powertrain
600,000 units 600,000 units

Final
Assembly x 2
600,000 units

Market

Since the Final Assembly process has been duplicated it is possible that the output split
between the duplicate plants could occur anywhere within the flat range of the relevant
LAC, thus the total output could comprise Plant A with annual output of 350.000 units
and Plant B with 250,000 units, or even three plants with output of 200.000 units a year.
The extent that capacity would be allocated between the plants would depend on
company specific factors, for example the larger plant might be a later addition to the
company. The company might also have to calculate the degree to which higher unit
costs at the smaller plant could be compensated for by lower unit costs at the larger
plant.

The simplicity of the paradigm suggests that a simple form of governance structure
could be applied, this being the U-Form. Historically most automobile companies have
started with this structure, Henry Ford’s autocratic control over the company he founded
being a particular example. GM under Alfred Sloan then developed the M-Form
structure in order to manage the multidivisional strategy. The focus of this thesis is
directed more at the strategic matters of size, in terms of access to economies of scale,
and structure, in terms of functional intemalisation. Nevertheless,

210
governance structure is relevant in demonstrating how a paradigmatic automobile firm
can manage its functions in a dynamic commercial environment. M-Form and U-Form
governance structures will therefore be evaluated where appropriate. The following
section of this chapter will investigate the empirical evidence collected during the
fieldwork phase of this study from elite interview information on the structuring
considerations within the industry. This will then feed into the next chapter of this
thesis which will examine the various ways that sub-optimal firms can approximate to
the paradigm.

5.4 Empirical evidence of vertical integration

The data on vertical integration W2is gathered as part of the elite interview schedule
conducted with senior managers in the automotive industry and its related institutions.
Two themes emerged from the collected interview data:

• Degree of functional control

• Extent of vertical integration

Functional control was concerned with the degree to which the companies had control
over their core functions. All companies involved were insistent that it was this that
gave them a right to a place in the industry. R&D was seen as particularly crucial, and
companies were keen to show that where inputs came from external parties this was in
a supportive role and not one that impinged on the design authority. MG Rover came to
the same conclusion, but only after a process of trial and error. The company had made
use of some external engineering consultants but found that the loss of control in the
R&D function proved to be a handicap. MG Rover representatives concluded that the
company needed to exercise more control over its R&D function in particular and that
it benefited from control of its other functions.

Vertical integration showed the extent of the companies’ control, i.e. the functions that
were necessary in order to compete effectively in the industry and market. Prime
amongst these functions was R&D, without which a company could not shape the

211
final product, but this also extended into control over the production capabilities. Here,
where the dependence is on physical assets, vertical integration was related more to the
control of output. Honda was notable in extending its corporate structure well beyond
what would be predicted by transaction cost analysis, and there was some possibility
that this was due to the company culture or atmosphere.

5.4.1 Vertical integration - contextual sources

Interviews conducted at Honda found representatives claiming that the company was a
provider of transport solutions and that it was vertically integrated around this purpose.
This meant that not only could it be defined as a full-function manufacturer in
automobiles, as a group it also took in the original motorcycle operations, a robot
research programme and a new jet engine facility. Although those functions that were
not related to automobile production were beyond the scope of this study, it was a
puzzle as to how they were related to each other. Table 5.1 summarises the various ways
in which the exercise of control was conceived by the interviewees to manifest itself at
Honda and the implications this was thought to have for the company.

Table 5.1 Honda - areas of control and derived benefits

Data Source Area of Control Benefit

Mr D Blume (Jaguar) Company philosophy Technical innovation
Mr T Sumita Research Internal resources
Mr T lida Intellectual property rights Available for sale
Company philosophy Creativity
Mr K Nakamura Semi-autonomous regional Market closeness
divisions

Table 5.1 shows that the basis of the business was its company philosophy. Honda’s
culture was emphasised in many interviews, both inside and outside the company (e.g.
Mr lida, Mr Blume), with the company’s philosophical stance seen as a source of
technical creativity. It was therefore vital for the company to have control over research
and the subsequent horde of intellectual property rights which represented the future
potential of the product range.

212
Honda’s view was that product differentiation was bom out of its control over
technology and this was fundamental in a market where consumers made choices for
personal reasons. David Blume, president of Jaguar, Japan, felt sure that part of
Honda’s success lay in its ability to pursue its own programmes. He felt that this
created the conditions for the company to develop the highly innovative CVCC engine
technology a decade before its alliance with Rover. This put it in a risky, but fertile
position:

“... even forcing it to come up with ideas.” (David Blume, Jaguar.)

It was clear that Honda placed a high priority on control of its technical assets, the
latest example of this being the company’s intemalisation of fuel cell technology after
the early partnership with Ballard of Canada. Mr Sumita described the fuel cell
research and how it progressed from a joint venture with Ballard to a sole Honda
project. Mr Yoshino, ex-Honda CEO, denied that there was any preset sequence to the
relationship, only that in its ultimate form it should be wholly under Honda control. He
took up on the financial perspective, namely that if Honda could bring in enough
revenue to sustain itself then it could develop wholly owned technologies for the future
and further strengthen the potential of the company. The distinction was made by
Tatsuya lida, that Honda had no objection to working with other companies but that
sole control:

“...allows us to be creative and innovative, do the things that we think should be done.”
(Tatsuya lida, Honda.)

It was quite apparent to this research that Honda maintained a strong integrated
stmcture, despite being spread across the continents. Mr Rogers argued that a core
competitive advantage for Honda was its integration of company functions, paying
constant attention to designing for production. He contrasted this with engineering
consultancies, like Ricardo, that housed “excellent” skills but did not have the
capability to see proposals through to production. In Japan, vertical integration even
extended into the sales network, though Mr Yoshino mentioned that this was unique to
Japan and that overseas it was better to have independent dealers who understood local
market conditions. It was therefore apparent that domestic integration, where the

213
company had the comprehensive knowledge, was more reminiscent of the U-Form
structure, but in foreign markets, where it lacked knowledge, it needed to call in
external resources. According to the theory of Bartlett and Ghoshal (1993), Honda
could have internalised this local knowledge but contrary to the theory it chose to use
local knowledge as an external resource to support its integrated structure. It can be
assumed that this brought benefits in transaction costs.

This reliance on its own capabilities was at the very core of the company and was said
to be the main strategic force. Kenji Nakamura said that when the senior management
set the general direction for the company this then allowed workers to have a degree of
autonomy within the framework. This seemed to suggest an M-Form of organisational
structure, and many commentators discussed the semi-autonomous nature of Honda’s
global divisions. Hiroyuki Yoshino went one step further and by claiming that it was the
policy of the company to encourage each major region to have full-functional capability
in “development, production, sales and marketing” he seemed to be alluding to the
managerial theory of Bartlett and Ghoshal (1993).

Table 5.2 shows the various ways in which integration is understood at Honda and the
benefits reported by the respondents. Some activities took it outside of the full-function
structure, such as motorcycle manufacturing, jet engines and robots. This might indicate
an M-Form structure, and since there was considered to be a derived technical benefit
for the more orthodox automobile side of the business, this hinted at Hill’s CM-Form of
organisational structure. Jet engines and robots were looked on as a testing ground for
product innovations, to be incorporated in automobiles at a later date. Mr Sumita and
Mr lida admitted that none of the technology from ASIMO was used intact in other
applications, but the research was fertile ground for additional ideas. In line with this
approach, Mr Nakamura stated that Honda encouraged engineers to transfer between
different areas of the company and gain wide experience. Mr Kobayashi, in his typically
candid manner, pointed out that the publicity generated by such projects was a highly
cost effective way of recruiting engineers of the best standard without having to offer
high salaries.

214
 Table 5.2 Honda - functional integration and its impact

Data Source Functional Integration Impact

Mr C Rogers Full vertical integration Design for production
NSX supercar project Mainstream innovation
Mr T Sumita Research: Jet engine, ASIMO Future applications
Automobiles-motorcycles Motorcycle production as
overseas pilot plant
Mr T lida Motorcycle racing project Mainstream innovation
Research: jet engine, ASIMO Innovation for current
Mr S Kobayashi Engineer-customer empathy products
Honda R&D

However, not all functions were treated equally and interviewees reported that ultimate
R&D authority lay in Japan even though overseas locations had R&D centres; Mr Ikeda
pointed out that the American head of R&D in the US had spent three years in Japan
learning the Honda methods. The production processes tended to have more autonomy
but it could be argued that they operated closely to Honda standards. Thus it would
appear that Honda’s organisational structure was a variation of the U-Form, with
foreign extensions being allowed operational flexibility, rather than an M-Form where
the foreign functions would have operational control. There was certainly no notion of
the managerial theory of the firm proposed by Bartlett and Ghoshal (1993) with regard
to the management of international operations.

There was some sense that a matrix format was also emerging at Honda to draw
different model programmes closer and exploit economies of scope. Mr lida stated that
as recently as 1992, the CRX model programme had used a dedicated design team that
did not share its development with other teams. Mr Rogers pointed out that by keeping
the NSX super car project in-house the technologies developed for it permeated
throughout the rest of the range, stating that advances in lightweight parts contributed
to the high standard of the new diesel engine. The company had achieved similar
economies of scope with this ‘trickle down’ effect before, Mr lida describing how
motorcycle racing had been used to set the technical goal of the company. In any case,
Mr Yoshino did not think that Honda was ambitious to move up to higher value models.
This indicated that the benefits of economies of scope from premium models were
probably quite limited.

215
Despite the increasingly close working relationship between projects there was a belief
that there was further to go. Mr Sumita believed that motorcycle and car divisions
needed to work more closely, perhaps to create a low cost entry level car to bridge the
gap between the two product types. He also believed that the economies of scope
brought an advantage in production when expanding abroad because it could pilot
production in a new location using motorcycle manufacturing. Although only America
seems to present any evidence for the motorcycle business being used as the corporate
spearhead, even at the time that the business was being established in 1979 the planning
for the automobile facility was already well advanced (Honda, 2006).

The opening of new plants may blur the distinction between M-Form and U-Form.
Using one production system to test the local environment before expanding into
another system suggests the M-Form, evaluating the data before sanctioning further
allocation of resources. However, since the automobile plant had been planned before
the motorcycle plant had been fully established it suggests a more centralised
governance structure. It seems that Honda was operating a domestic U-Form with
international extensions. These extensions were allowed limited autonomy only where it
was necessary to account for local differences, such as marketing and product
adaptation, but these adaptations were targeted responses and operational control would
be too general a description. As Mr Kobayashi pointed out, and shown in Table 5.2, the
company put R&D at the core of the company and in close contact with the needs of the
global markets, a strategy further emphasised by having all CEOs promoted from the
R&D function. This strategy supports the suggestion of a U-Form governance structure
with the functions working closely together and finds its equivalent in other contextual
firms. Daigo Umeki, for example, reported that his company, Toyota, had a similar
policy to independent development as did Honda:

“We want to make everything by ourselves, otherwise it is very difficult to handle

[innovations] and very difficult to find out the root cause [of problems].” (Daigo
Umeki, Toyota.)

He conceded that internally generated innovation is not always successful, and called
attention to Toyota’s poor record in Formula 1 racing with its wholly owned team. Yet if
Toyota had acquired another team it would have been dependent on that team’s

216
knowledge base, but by running its own team Toyota internalised the knowledge.
Purchase of an established team would have formally internalised it but there would
have been no links throughout the company, unless actively promoted.

Table 5.3 summarises the arguments concerning the manner in which divisions may
operate within a company structure. This reveals the extent of autonomy enjoyed by the
divisions and the kind of facilities they derived from larger parent groups.

Table 5.3 Automobile companies and dependency/autonomy

Data Source Company Dependency Autonomy

Mr D Blume Jaguar Short-term external synergy Long-term internal
Jaguar with Ford synergy
1 model 4 model range
30,000 units a year 150,000 units a year
Brand differentiation
Saab GM total control Brand defence
None
Ford PAG Ford resources Semi-autonomous
Mr T Watson Aston Martin Jaguar sports car platform Internal range of
Aston Martin (cancelled) platforms
Mr K Yamada Toyota None Internal knowledge for
Toyota sustainability

Although GM is famous for implementing the M-Form of organisation, it is clear that it

has full operational and strategic control of its Saab division. Conversely, Ford appears
to have embraced the M-Form in its approach to the brands within its Premium
Automotive Group (PAG) division comprising Jaguar, Volvo, Land Rover and Aston
Martin. However, there is no distinct demarcation between strategic and operational
issues, neither was it understood whether PAG was the division or if that status
belonged to each brand. Jaguar had been dependent on Ford while producing one core
model at a rate of 30,000 a year, but in the long-term Jaguar was intended to achieve
internal vertical integration with 4 models and output of around 150,000 a year. Mr
Blume advocated the view that a sense of independence was needed, even within a
larger group, to protect the brand and its future potential at a time when brands like
Jaguar were the very future of Ford:

217
“To protect the brand from its helpers, from being smothered, to maintain some
independence within a large corporate entity is very important... To have the
independence of the brand’s own management to defend what the brand stands for,
selfishly and stubbornly, to take the independent view, at the same time recognising that
it is part of a large corporation.” (David Blume, Jaguar.)

Mr Blume alluded to the situation at Saab where GM had run the operation down to the
extent that there were too few people to defend the brand and so the brand would lose
its differentiation:

“It starts with you can’t command the emotional value added, and therefore you can’t
charge for it so you become non-premium and a brand falls back into the swamp.”
(David Blume, Jaguar.)

Like Jaguar, Aston Martin seemed to have both operational and strategic control. It had
a dedicated range of vehicle platforms, separate this time even from stablemate Jaguar,
yet Mr Watson also quantified the overall beneficial influence that Ford had on Aston
Martin sales (see Appendix 6). Mr Watson described how Aston Martin’s semi-
autonomous status within Ford allowed it to react quickly to market conditions. The
example he gave was of proposed derivatives of the DB7, which was approaching the
end of its production run, the prototype Zagato version being ready for viewing by
potential customers within a year. This suggested a strategic ability to reallocate
resources to new model programmes as required. Mr Watson asserted that these were
examples of Aston Martin’s independence from its parent:

“We are not Ford Motor Company, we are Aston Martin.” (Tim Watson, Aston Martin.)

On closer inspection it was found that the governance structure for divisional
automobile companies was not that of an established type. Table 5.4 reveals that the
divisional companies under Ford and VW had elements of both strategic and operational
control such that their own governance was reminiscent of a unified U-Form structure.
All of the divisional companies listed show that control of entire product programmes is
crucial, the parent being used as a supplementary resource.

218
This is remarkable given that the parent companies Ford and VW were also vehicle
manufacturers, and were indeed the dominant producers in the group. The benefits of
the organisational structure appeared to favour the subsidiary company and indicated a
system of hidden subsidies. This might suggest the internal capital market of the M-
Form, yet so dependent are the divisions on these parental indulgences that it appears
that the funds are not being made available internally to overcome imperfections in the
external capital market but to mask the inability of the divisions to raise capital
independently.

Table 5.4 Divisional control of operations and strategy

Data Source Company Level of Control

Dr U Eichhom Bentley Strategic Operational
New product programmes HR: engineers, engineering
Mr A Hallmark Bentley Body design management
Engine assembly
Mr D Dickson Bentley Production technology Component supply

Mr T Watson Aston New product programmes Engine assembly

Martin Platform technology
Mr D Blume Jaguar New product programmes

The parent companies imposed some strategies on the divisional companies, such as
Jaguar obtaining its aluminium body technology from Ford even while Aston Martin
obtained its new platform technology independently. Bentley has responsibility for its
new product programmes but based its GT series on the existing VW Phaeton platform
and engines. At the same time, Bentley does not have operational control over body
production for the GT series since the bodies are supplied by VW.

Throughout the contextual sources all the interviewees were in agreement that control
over an internal R&D function was crucial, and this included the divisional companies.
This implied strategic control that even divisional companies enjoyed, though
supplemented by access to resources within the parent. Furthermore, where the parent
company provided finance to divisional companies it was not to correct for
imperfections in the external capital market, instead by way of hidden subsidies it
seemed to be investing in disregard of capital investment principles. Operational control
at divisional companies tended to extend as far as the production systems

219
which, for historical reasons, remained within the company structure. For example,
Bentley manufactured its old V8 engine internally but W12 engine production was more
dependent upon VW. At the same time, all the divisional companies had some claim to
being full-function organisations.

Although the divisional companies could not claim the complete full-function structure
of independent companies such as Honda and Toyota, they had enough to be able to
claim similar governance structures. Honda in particular had a centralised governance
that was more U-Form than M-Form, even if overseas functions often had a small
degree of autonomy. Divisional companies such as Jaguar and Bentley had far more
control over their strategy than any overseas Honda function, so although they were
dependent upon their parent companies for support there was an argument for describing
their governance structures as U-Form based on their strategic control. This is not to say
that the parent groups are not involved in divisional strategy, and Jaguar seems to
receive a significant amount of strategic involvement from Ford, but this never seems to
be consistent. Hedlund’s (1994) N-Form of governance structure might be applicable,
but there is little sense amongst the contextual companies of the heterarchical structure
that the theory proposes. Instead, the empirical evidence appears to suggest that the U-
Form has been adapted to the larger companies, perhaps facilitated by some kind of
matrix that created temporary networks of management according to the project,
whether it was a new car programme or an overseas production plant.

Since MG Rover was almost uniquely small as a full-function manufacturer it might be

expected that its strategy would call for a more defined governance structure than was
found amongst the contextual companies, anticipating in particular the U-Form. This
might then bring clarity to the governance structure of the larger firms.

5.4.2 Vertical integration - MG Rover

Whatever the view of industry commentators on the long-term sustainability of MG

Rover, at the time the interviews for this research were undertaken the general consensus
at the company was that it had the capabilities to survive as a standalone operation.
Although it was conceded that aid was needed in the form of funds and

220
temporary new products there were no misgivings concerning the integrated state of the
company. Table 5.5 collates the comments made by the participants related to MG
Rover.

Table 5.5 MG Rover - theoretical and practical considerations of autonomy

Data source Considerations

Theoretical Practical
Mr S McKee No third party
Head of International No shared profit
Media Relations No outside equity holders
Mr R Ball Core intellectual property Supporting partners
Head of HR Positive morale Complacency
Development Motivation
Dr C Millard Opportunism
Head of Strategy
Mr R Oldaker “Fleet of foot” Short-termism, tactical not
Chief Engineer strategic
Small team of experts “Difficult to manage”
Maintain core skills Supporting partner
Mr M Harbour Internal integration Supporting partners
MEP

Mr Harbour, a previous project manager at the company, raised the importance of

maintaining internal integration at MG Rover for its long-term survival. This was a
point raised on many occasions: that MG Rover had a distinct advantage in being
tightly integrated with all the main functions on one site at Longbridge. In fact, the only
facility off-site and not owned by MG Rover was the body pressing facility operated by
BMW. Since all other BIW processes were conducted at Longbridge, MG Rover could
be considered to be a full-function company.

As Mr Oldaker stated, functional integration meant that the company could be flexible
and fast reacting based around small teams of industry experts who embodied the core
skills, even if like a Formula 1 team such individuals could be hard to manage. This
view found agreement in Dr Millard who felt that people were motivated because they
understood that there was “no safety net”, contrasting this with Jaguar where he
believed that security of employment under Ford must be akin to a nationalised
company. He coupled the sense of danger with an awareness of being able to act on
opportunities, giving rise to a motivated team of workers. He was a strong advocate of

221
positive opportunism within an integrated company, and believed that MG Rover was
of a good size to exploit opportunities as they came up:

“We can move our business around very quickly to take advantage of opportunities that
may present themselves.” (Dr Millard, MG Rover.)

Mr Oldaker agreed that as an independent manufacturer MG Rover could be more “fleet

of foot”, though he added that MG Rover was a “short-term operation” and tended to
make tactical rather than strategic decisions. Dr Millard believed that the company had
the flexibility that was necessary in the current commercial climate, arguing that an
independent company could differentiate its products to compete in a fragmenting
market. Mr McKee also noted that the close proximity of MG Rover’s facilities created
very short lines of communication, and this was particularly illustrated by the
abbreviated period of time that it took to develop the SV supercar from the lacklustre
X80 proposal:

“.. .SV styling was changed during a big argument in the car park at Longbridge after
the first motor show.” (Mr McKee, MG Rover).

As with the contextual companies there was general agreement at MG Rover that R&D
was central to the company’s future. However, MG Rover was obliged to use external
consultants with variable rates of success. Mr Oldaker stated that the reason for MG
Rover approaching TWR as a contractor was that it had a more advanced digital design
capability than MG Rover as a result of its takeover of Daewoo’s design facility, though
MG Rover later upgraded its own facilities in order to achieve totally digital design.
Although TWR collapsed, and seriously undermined the new medium car programme.
Dr Millard trusted in “normal business diligence” to guard against this again. He stated
that MG Rover needed strategic control while looking to external design houses for
access to technical innovations design. He emphasised the importance of having an
integrated approach to car design, stating that he would be interested in adding specific
external skills but not in the “wholesale hiring” of another engineering team. The
company would thus be able to access specific capabilities without being vulnerable to
hold-ups.

222
Transaction costs came to the fore in contracting with another external consultant,
ProDrive, for the rear-wheel-drive large car platform where MG Rover made the
dominant input at the initial, defining stage and then the final, finishing stage. The
experience had taught the company that using outside contractors was not as cost
effective as anticipated: the contractors needed close monitoring by a small team of MG
Rover specialist engineers, and the project needed a tight definition at the start to ensure
that the contractor stayed within the desired parameters. The transaction costs were
therefore higher than had been originally anticipated. Concerning the V8 engine for the
project Mr Oldaker reported that MG Rover had very little involvement. Most of the
company’s work lay in adapting the existing car’s electronics to accommodate the Ford
sourced engine, while the engine was enhanced and supplied by Ford collaborators
Roush Technologies.

While MG Rover was active in defining the limits of strategic control in R&D there was
little of this in its other functions. There was no attempt to fill vacant production
capacity with assembly for other manufacturers, although Powertrain did supply engines
to various other manufacturers, and although the largest of these was Land Rover this
was a contract inherited from when MG Rover and Land Rover were part of Rover
Group. Strategic control in the functions concerned with production was part of the
company’s status quo and the status of the functions was not developed further. This
was not the case with R&D which the company found, by heuristic exploration of
external contracts, to be crucial to the strategic control of the company’s long-term
future. In particular, the company found that the physical assets of R&D could be
acquired relatively easily, but that the human assets were more difficult to access
externally and so it was fortunate that the company found it had considerable capability
internally. It was therefore able to retain sufficient strategic control over R&D while it
maintained strategic control over the other three functions almost by default in not
developing them further.

The governance structure that would appear to be appropriate for a firm of this size
would be the U-Form. Indeed, strong operational leadership from senior management
was a feature of the company and Mr Ball described how there was a general feeling
that CEO John Towers was “the messiah”. More frequently, though, this research found
that managing director Kevin Howe was instrumental in all the strategic and

223
operational decisions. This could have resulted in a company where operational issues
were dealt with at the expense of long-term strategic issues, a commonly cited
weakness of the U-Form, but it would appear that both these issues tended to suffer.
Chapter 7 of this thesis will investigate these matters in closer detail while the next
chapter will show what strategic options were open to MG Rover in attempting to
compete with firms that had achieved at least MBS? in the industry.

5.5 Conclusion

The investigation of transaction costs has shown that there are two main dimensions to
vertical integration: physical assets and human assets. Hitherto, physical asset
specificity has been considered to be the main driver of vertical integration,
intemalisation of these assets being a method of pre-empting ex post contractual
opportunism. Although there is empirical support for this view it fails to recognise that
human assets pervade all company activities, including control of the physical assets.
Indeed, opportunism is essentially a human problem, not a technical one. This is
because the decision to create a hold-up and appropriate ex post advantages is a human
one, which then uses the physical assets to extract the benefit. Klein, in particular, has
argued that vertical integration is a matter of intemalising labour contracts in order to
introduce greater flexibility into the production system.

Assuming that companies using the same production technology are operating in the
same economic environment, since they are therefore faced by the same transaction
costs they will converge on a single form of vertical integration. This is the full-function
model which determines the functions that should be intemalised within the company
boundary. For an automobile manufacturer this should include R&D, BIW, powertrain
and final assembly. It is an economic model of vertical integration and it does not denote
a particular governance structure or geographic location of function. However, by
juxtaposing theories of economies of scale with that of transaction costs it is possible to
provide for each function prescribed levels of scale. This results in a model of size and
vertical integration structure, but not governance structure.

224
Vertical integration brings with it control problems that different organisational
governance structures are intended to address. According to theory, the unitary U-Form
is effective for relatively small firms where the management can take a close
operational and strategic role in the integrated functions. Although small firms are
unable, by their nature, to exploit the available economies of scale, the simplicity of
their structure brings this problem to the fore. The theory of the multidivisional M-Form
isolates senior management so that it can specialise in strategic matters and leave
operational decisions to the various divisions. The structure has advantages over
transacting in the market because the relationship between headquarters and the
divisions is transparent, thereby facilitating efficient allocation of resources and capital.
This has particular utility when the industry is undergoing rapid development or a
company is particularly diverse. The divisions themselves may be single functions or
themselves have a U-Form style structure with a number of functions, although without
the overall strategic control. The M-Form can bring benefits from economies of scope in
the product range and management capability although this can overshadow the pursuit
of economies of scale. There are also doubts about whether the M-Form can stretch to
an international structure.

The data gathered by this research was very supportive of the need for an integrated
structure. This was even to the extent that there was more support for the U-Form than
the M-Form structure. This is in line with the notion that the parameters of the industry
are fairly stable and the industry has reached maturity, a level of stability that Burton
and Obel (1980) and Donaldson (2001) suggest is suited to the U-Form structure. Even
large, international companies, such as Honda, maintained a strong focal centre in the
domestic market which showed a U-Form structure from which international extensions
radiated. Although the foreign operations appeared to have operational control this was
strongly influenced from the centre. For those firms which were divisional companies,
such as Jaguar and Bentley, they too showed a U-Form integrated structure within a
larger group, but this time at a much lower level of scale. This might have implied a
holding company H-Form of governance except that the parent company had occasional
strategic and operational involvement in the divisional companies. At the same time, the
parent company had its own automobile production facilities that somewhat accorded to
the U-Form of organisational structure.

225
In this regard, MG Rover made a crucial case study. It clearly had a U-Form structure
with poor achievement of scale, like the previously mentioned divisional companies.
However, unlike them, it had to seek its external assistance in the market rather than
within a larger group. This resulted in increased transaction costs, such as when MG
Rover design projects were outsourced only to be become the victim of ex post
difficulties. On another occasion, that of V8 powertrain procurement, by surrendering
almost all responsibility to the external firm many of the transaction costs were avoided
while the company retained strategic control of the project as it related to MG Rover. It
would seem that MG Rover, due to the precarious commercial situation it was in, had
arrived at a well-defined and lean full-function structure, as prescribed by the full-
function model. However, as the previous chapter noted, MG Rover was unable to
access the economies of scale prescribed by the same model. Since its organisational
structure did not provide it with access to economies of scale, except by going outside
of the structure, only a new form of organisational structure would suffice.

The next chapter of this thesis will explore the methods by which a company that is
uncompetitive within the automobile industry paradigm can use devices to approximate
to the paradigm. In the subsequent chapter. Chapter 7, empirical evidence relating to
such approximations will be introduced. In particular, the chapter will focus on how
MG Rover had a full-function structure but lacked the scale to sustain itself. It will be
suggested that the company could have used its human asset endowments as the basis
for an organisational restructuring, which would then have allowed it to retain the
transaction cost advantages of a vertically integrated, full-function organisation while
obtaining access to the necessary economies of scale. The reasons for the failure of MG
Rover to achieve these benefits will also be examined in Chapter 7.

226
 Chapter 6;
Approximating to the Automobile Industry Paradigm

The automobile industry paradigm described up to this point suggests strict conditions
for the long-term sustainability of an automobile firm since it is constrained by
parameters set by both economies of scale and vertical integration. So far, this thesis has
put forward a prototypical size and structure for an automobile firm operating closest to
its optimum. This applies principles of economies of scale to indicate plant sizes which,
in combination, provide a size for the firm. Transaction cost analysis is applied to
indicate a full-function model of vertical integration. The full-function structure is
effectively independent of firm size since it is related to the economic friction that
exists between operations, thereby driving the make-or-buy decision.

This chapter of the thesis will first examine the cost disadvantages of falling below the
required levels of output and then demonstrate how these might be ameliorated
internally with different production strategies. Chapter 4 showed that if the MBPS of the
automobile company functions are brought into reasonable synchronisation then this
would result in the prototypical automobile manufacturer producing around 600,000
units a year at MESp. It has been found that beyond this optimum point diseconomies of
scale emerge, partly due to rising R&D costs. However, firms producing at MESp are
sensitive to changes in output that lead to large variations in costs. As a company grows
in output by adding new plants, each designed to produce at MBPS, this sensitivity is
reduced so the firm is able to absorb changes in output without suffering the large
variations in cost. Furthermore, large firms can diversify risk by offering extended
product ranges. These firms are the industry leaders.

Those firms that fall below the MBS output are deficient in structure, scale or both. This
chapter will investigate some of the strategies by which a sub-optimal company might
approximate to the cost advantages of MBS described by the automobile industry
paradigm.

227
6.1 Introduction

Firms that fall below the output required for MESp suffer debilitating cost
disadvantages, relative both to the prototypical model and to the industry leaders. Such
firms will be deficient in at least one area: either the scale of output is too low, or they
do not operate with a full-function organisational structure. They then have to solve
their predicament by approximating to the precepts of the paradigm to put them on a
convergent trajectory. This is possible using internal or external solutions. Internal
solutions could involve more intensive use of limited production facilities through
flexible manufacturing, which then reduce the company’s exposure to changes in
market demand assuming it has the necessarily vdde model range. Alternatively, the
company might extend the use of the facilities by perpetuating the output of products
with unchanged specification for longer periods of time.

In another approach, the automobile firm might seek external solutions by contracting
with other firms. Clearly, in the make-or-buy dichotomy, this can mean seeking
resources in the open market. However, this is not an approximation of the paradigm
but the antithesis of it. Since the full-function model demonstrates cost advantages in its
integrated organisational structure, any firm that sources externally as an alternative to
the model will suffer cost disadvantages. In order to approximate to the full-function
model a firm must find a form of inter-company relationship that represents an
alternative to the make-or-buy dichotomy.

Reflecting the two forces that structure an automobile company, there are two ways in
which this external partnership can come about. The first is size, where the partners are
uncompetitive in terms of quantity. A full-function manufacturer that is uncompetitive
in this regard is unsustainable only by relative degree: the further behind the standard
set by the industry leaders, the less sustainable it is. This can be corrected through a
pooling of joint resources with another company, although some of the advantages may
be offset by additional transaction costs. For consistency, this thesis will refer to them as
horizontal joint ventures as it involves the enhancement of existing capabilities through
sharing with a partner. Naturally, it is a strategy that is also of incremental value to
industry leading producers, further improving their relative competitive positions.

228
A second form of partnership is structural. Here, the company has some functions at the
desired size but it is not a full-function organisation. Such a firm is more than simply
uncompetitive within the paradigm, it has no prospect of sustainability as long as it is
lacking functions; these must be sourced externally. It is possible to do this
contractually in the open market, but the full-function model demonstrates that this
brings transactional cost disadvantages. To approximate to the paradigm there is the
need to develop a vertical joint venture (VJV), a complementary relationship where the
two parties bring distinct capabilities to the alliance. An additional strategy may be
necessary if scale is still lacking, but if the parties already enjoy competitive economies
of scale in their existing functions, then the resultant VJV will be converging upon the
prototypical model and so have access to most of the available economies of scale.
VJVs have less value for firms that already comprise full-function structures because
the strategy implies relinquishing one or more of the functions.

This chapter will show that for firms to come together in a VJV each must have a
different structural basis in order to have arrived at their contrasting structures. In
particular, the distinction between human assets and physical assets in the automobile
industry paradigm implies a demarcation line within the full-function model that might
be exploited for the VJV. Specifically, human assets prevail in R&D while physical
assets prevail in production. This distinction between asset types may be heightened by
the possibilities raised by globalisation. New locations offer a range of different factor
endowments, as described by the Heckscher-Ohlin (HO) model, and elaborated on by
Dunning’s eclectic paradigm; some of the objections to globalisation will also be noted.
A firm that is already operating at, or above, MESp will receive incremental benefits
from the different factor endowments to enhance its position relative to its peers.
However, for a firm that is at a sub-optimal level of output a restructuring around
different factor endowments may raise it to a more competitive position. Thus, not only
does an international vertical joint venture (IVJV) have the potential to replicate the
organisational structure, and even scale, of a prototypical automobile company, but it
can also bring additional benefits due to specialisation.

Throughout the discussion examples will be given of cases that illuminate the
theoretical approach so far described. These will develop the potential for joint

229
ventures as forms of organisation beyond the make-or-buy dichotomy. Globalisation
will be examined as a potential source of cost benefits due to factor endowment
differentiation, particularly with regard to the division between human and physical
assets. This will be applied to the joint venture structure and then further refined to give
rise to the IVJV. The following chapter will then examine the empirical research data
based on the theoretical principles established in this chapter.

6.1.1 Sub-scale production

The technical advantages of scale production are well documented and Pratten (1971)
attributes further benefits to companies that achieve very large volumes. Firstly, R&D
costs can be spread over a greater number of units thus encouraging greater research.
Secondly, large firms can expand incrementally in plant sizes that each achieves MBPS,
depending on the process. Thirdly, a large firm can save costs on designing new plants
due to learning effects from previous projects. Finally, larger firms can negotiate
discounts on larger procurement contracts of components by extending the benefits of
scale into the supply industry. Many of these points were covered in Chapter 4, although
R&D was actually found to rise as firms grew in size.

Exploitation of economies of scale may be a fundamental aim of the industry but, as

Pratten (1971) has shovm, it is not a condition that can be arrived at immediately. Even
with a given level of technology learning effects will influence actual output, so firms
joining the industry now are at a greater disadvantage than if they had begun with the
start of the industry. The Ford company that exists today was actually Henry Ford’s third
attempt at manufacturing. The vast domestic US market had already proved fertile
ground for Olds with the first mass produced automobile and even the famous moving
production line was not new (Lewchuck, 1987). Yet, though Ford was aiming to increase
efficiency, at every turn it is not clear that the concept of an optimum output was
understood. Ford advanced the technology and techniques for production flows to
improve production synchronisation with the levelling of production times, yet this
strategy is unrelated to the concept of an optimum output that would exploit the
economies of scale for individual processes (Tolliday, 1998).

230
The Ford company attributed its success to its production methods but did not
consistently quantify this advantage. The Rouge River plant was designed to be a highly
vertically integrated production facility with its own blast furnaces and glass making
facilities. Thus, along with the latest company labour practices, the new plant was
intended to be the ultimate expression of Fordism. This formula was repeated for the
Dagenham plant in the UK. However, it proved beyond the capabilities of the company
to match the minimum cost outputs of the different facilities within the plants so that all
production activities could be operating simultaneously at MBPS. The Rouge River
plant never reached more than 50% of its capacity utilisation (Williams et a l, 1994) and
where once 100,000 workers produced 1,200 cars a day there are now 3,000 devoted to
producing 800 Mustangs a day (Economist, 2002a). Fiat copied the old Ford Highland
plant when it built its production facility in Turin, but this simply imported all the
original associated problems along with the unsuitable architecture.

In the short-run, it is important that companies operate plants at minimum cost output
and that the resultant cost structure is comparable with that of competitors. In the long-
run, companies are at liberty to choose larger plants, on condition that the actual output
accords to that indicated by the industry LAC. According to Maxcy and Silberston
(1959, p.93) the main benefits of a move towards an optimum scale are yielded in the
initial stages, asshown below:

1000 units rising to 5000 units= 40% drop in costs

50.000 to 100,000 = 15% drop in costs
100.000 to 200,000 = 10% drop in costs
200.000 to 1,000,000= 5% drop in costs

Of course, the MBS? put forward by this thesis is in the middle of the highest range
quoted by Maxcy and Silberston. Pratten (1971) also suggests that the largest benefits in
economies of scale are gained in the 100,000 to 250,000 range for a firm producing one
model, economies available above that range being those attributed by Pratten to body
pressings. Cusumano (1985, p.216) was another who found that Nissan achieved its
greatest savings with the first expansion in output, finding a 31 per cent decrease by
raising production from 23,000 in 1955 to 130,000 in 1960. However, as the company
grew so did the model range and by 1999 the company had 24 global

231
platforms, five of them taking 62% of output and only one totalling over 300,000 units a
year, for a total output of around 2.5m (Automotive Design and Production, 2004). This
thesis has found that beyond production capacity of 600,000 units a year the cost
advantages can be attributed to diversification of risk, of which a burgeoning product
range is a crucial part of the strategy. However, Nissan’s example shows that this too
can lead to rising costs if it is not controlled.

If the major gains from pursuing economies of scale are found by smaller companies
that are expanding, it is also possible to mitigate, to a certain extent, the scale
requirements of the Budd BIW process by using stamping dies designed for economic
production at low volumes. The short-lived revival of Jensen was predicated partly on
utilising soft resin tooling which reduced die cost from £40m to £lm, suiting planned
production starting at 300 units a year (Financial Times, 2001a). This is only a small
saving since all the other associated costs, such as the presses, remain the same. If the
presses are capable of being run at 2m operations a year then there will be an output
mismatch if low volume dies are fitted.

There was another option available to Jensen that might have ameliorated the effects of
sub-scale manufacturing. The marketing solution would have been the cost recovery
tactic where the final product is priced to cover the additional costs of sub-optimal
manufacturing (Wells and Nieuwenhuis, 2001). With a published price of £42,650 in
2001 the Jensen SV8 was considerably more expensive than the comparatively high
volume Alfa Romeo sports car (GTV V6: £25,240) but similarly priced to other low
volume sports cars (e.g. Noble M12 GTO: £44,950) (Autocar, 2001). Although Pratten
(1971, p. 147) stated that “a higher price can be charged for a non-competing article” it is
unlikely that there would be no competition whatsoever, even if it is limited to another
sports car or a substitute product; the existence of any alternative would mean that costs
still have a crucial role. In any case, price recovery serves only to support high costs of
production, it does not address the underlying cost problem itself. This may be achieved,
though, by approximating to the automobile paradigm using internal or external
resources.

232
6.2 Internal approximations to the automobile paradigm

Automobile firms will commonly exhibit a full-function structure including MEPS for
particular plants and processes. With MESp offered at a relatively accessible capacity of
600,000 units a year firms are able to expand capacity in units of MEPS. However, if a
company finds that its scale of output is short of MES?, it must find an approximation to
it. As a full-function firm there are two internally generated solutions available: either to
extend the use of the production facilities over time, thus reaching for the ultimate output
levels over the long-term but not the rate of annual use, or exploiting economies of scope
by using the facilities more intensively with flexible manufacturing techniques and a
broader product range. Although neither tactic achieves quite the optimums prescribed
by the theoretical economies of scale they do at least represent a pragmatic
approximation.

6.2.1 Extending use of the production system

Extending model life cycles can approximate to achieving maximum panel press usage
by attaining ultimate levels of output over time, for example the figure of seven million
pressings for the life of a die (see section 4.3.2.2), could be achieved by keeping the
same model in production for a number of years. The most famous example of this
would be the VW Type 1 (Beetle) which can claim production from 1938 to 2003. It
achieved an annual peak of 1.3m imits in 1971, well short of the 2m annual optimum for
pressing rates alone (as opposed to the BIW process as a whole), but it made up for this
with a total production run of nearly 22m that somewhat exceeded the extended life
strategy. Other examples would be vehicles such as the Jaguar XJS, which remained in
production largely unchanged for 21 years even if total production only amounted to
112,000 (Buckley and Rees, 2002).

Maxcy and Silberston (1959) caution that no matter how much the life of the dies is
extended this cannot correct for the disadvantage of running heavy press equipment at
low rates. In addition, it does nothing for the operating rates of other functions, although
powertrain production is usually boosted by the fact that powertrains have the potential
to be shared with other models. Recently, though, it has become increasingly difficult to
extend engine production due to the progressive tightening of

233
engine exhaust emission regulations. Model life cycle extension is therefore a
decreasingly appropriate strategy.

6.2.2 Economies of scope and scale in flexible production

Economies of scope elsewhere can have a direct impact on production and the
exploitation of economies of scale, the enabling technology being flexible manufacturing
systems. Flexible techniques have become a feature of all three of the manufacturing
functions since multiple products diversify the risk of output fluctuations. For example, a
wide product range means that a high volume manufacturer, like Honda or Toyota, can
use one production facility to make relatively low volume, niche products. In final
assembly this may be relatively easy to achieve as long as the different vehicles can
physically join the production line and the routine nature of vehicle assembly means that
the same workers can work on different vehicles with little additional training.
Economies of scope can also be found in the body pressing stage because, although the
dies for external panels are dedicated to a particular model, internal panels may be
common to a range of models. Powertrain production is also amenable to flexible
production of multiple products, for example engines tend to be designed and made in
“families” that offer a range of sizes. However, since these techniques depend on the
manufacturer having a wide range of products on offer then it tends to be a strategy
better suited to larger manufacturers, those operating at MES? or higher.

Economies of scope in common components has the result of raising the utilisation of the
production system, but as Clark (1987) noted, this only serves to improve access to
economies of scale if it employs production capacity that might have been used
elsewhere. Few manufacturers achieve more than 50% usage of their body presses
(Maxcy, 1981), so the ability to change dies for panel stamping will raise the capacity
utilisation rates of the presses. This does not quantitatively change the economies of scale
that are available in panel stamping but it does make them more accessible. This is
further improved by minimising the time necessary to change the dies, often possible in
just a few minutes (Cusumano, 1985, p.285). These producers also benefit from having a
flatter total cost (TC) slope due to higher mechanisation and lower variable costs,
reducing the break even point and leading to a rapid increase in profits

234
(TR - TC = Profit), though also rapid increases in losses if the strategy fails (Rhys,
1972, p.276).

Although flexible production is an opportunity to access economies of scale, it also

introduces costs of its own. Even if presses are run more frequently by alternating dies
for different models, the time taken to switch dies implies additional costs and does not
address the cost of the dies themselves. Nieuwenhuis and Wells (2002) put a total BIW
investment cost at between £370m and £515m for one model, including £20m-£65m for
the dies (see section 4.3.2.1), so with the simple addition of a new set of dies for a
different model, using otherwise the same facilities, the BIW process would be in the
range of £390m to £5 80m for the two unrelated models. Since some of the unseen
panels may be shared economies of scope will reduce the total costs but mixed
production can never exploit all the available economies of scale, if only because both
sets of dies cannot be used simultaneously. Nevertheless, this does at least represent a
significant improvement over running plants below the MEPS.

One company, Fumia Design, has even proposed the use of symmetrical outer panels,
for example matching front and rear doors, thereby reducing the number of dies
necessary per model (Fumia Design, 2006). GM found further advantage in its C-Flex
system for welding the panels together. The company uses welding robots that can be
simply reprogrammed for different models instead of having to be replaced, resulting in
cost savings of $100m per vehicle launch and a plant area reduction of 100,000 square
feet (AutoTech Daily, 2002).

Mehrahi et al (2000) suggest a novel form of flexible manufacturing, the reconfigurable

manufacturing systems (RMS). These authors see RMS as a quantum advance over
flexible manufacturing because it includes the possibility of adjusting production
capacity to fit market demand. This implies access to economies of scale for a wide
range of outputs rather than achieving a single optimal output with a mix of models. This
would be of interest to all manufacturers, but particularly a prototypical manufacturer
seeking to reduce the risk of fluctuating output and those below MESp seeking to reduce
the optimum level of output.

235
RMS is achieved by modularisation of the production process in terms of both
hardware and software. As consumer demand changes so one system can be
reconfigured into another by the expedient of replacing, deleting or modifying selected
modules. However, for such an approach to be effective, the operators would need a
resource pool that they could draw on, or deposit redundant modules into, without
incurring additional costs. Essentially, RMS reduces production capacity by removing
machines but the authors fail to explain how this can be done without leaving those
machines idle, and therefore incurring costs. Furthermore, it is difficult to understand
how the remaining machines could be employed in a reduced production process unless
they were inherently flexible. Under the Mehrabi et al (2000) system this would not be
possible since modularisation is contrary to generalism, it only applies to specialist
machines.

From the description by Williams et al (1994) of the Toyota Production System (TPS),
developed by Taiichi Ohno in the 1950s, it is possible to visualise the underlying logic
of RMS in the creation of U-shaped sub-assembly cells, essentially spurs feeding into
the main production line. However, Williams et al (1994) point out that this led to
duplication of equipment. Furthermore, the purpose of the cells was not to bring
flexibility to production but to achieve uninterrupted production flows, bringing the rate
of sub-assembly work up to the same speed as the main line. This would then inhibit, or
eliminate, the buffer stocks that build up between machines that operate at different
output levels, as described by Ho et al (1983). The same situation that is being
controlled by buffer stocks can also cause blockages when the buffer is not large
enough (Gershwin, 1987) or stoppages if the buffer becomes exhausted. TPS sub-
assembly cells constituted an organisational innovation but they had no effect on the
design of the main line which continued with linear production flows dependent upon
automatic transfer machines (ATM).

Since final assembly operations are characterised by the sequence of activities rather
than a specific technology they have attracted many innovations to improve the
production flow. Womack et al (1990) have argued that Toyota has shown superior use
of the production system by developing lean, or fragile, production. The most well
known development is that of Just-In-Time (JIT) whereby inventories at the assembler
are kept at the lowest possible levels, necessitating deliveries from

236
suppliers at time critical intervals (Krafcik, 1988). This involves close relationships with
suppliers, the primary ones being termed Tier 1. At the Honda plant in Ohio, for
example, 92% of Tier 1 suppliers provide JIT and design contributions, with Bellamar
producing seats in parallel with the vehicle for which they are intended and supplying
them direct to the production line (Mair, 1994).

The JIT system depends on a self-regulating pull-through production system govemed

by “kanban” cards and “andon” lights, with the machines and subassemblies organized
to minimise labour. However, Williams et al. (1994) take the same view of JIT as they do
of TPS in general: indeed, they term it the “romance of Toyota”. Instead, the authors put
the spotlight on Shotaro Kamiya and his Toyota Motor Sales (TMS) division, created
using selling techniques learnt in the United States. To keep the production system
running it was necessary to maintain the stability of demand and this, according to
Williams et al. (1994), was the crucial contribution to production made by TMS to TPS,
rather than the innovations within the plant itself. By 1975, Toyota had overtaken Nissan
and continues to hold this position, taking 40% of the Japanese car market in 2005 by
consolidated sales (Automotive News, 2006). Despite declarations of “genbashugi”, the
primacy of production, according to Williams et al. it is the success of the marketing
strategy that enables production to operate at high levels of efficiency. This is echoed in
Chapter 4 (see section 4.5) which showed that a major source of cost was the fluctuation
in output from that which was planned.

TPS, therefore, represents a refinement of the existing production method: the

production line is still a linear sequence, there is large scale use of ATMs and the kanban
system is similar to the system of human chasers used by Ford at the Highland plant.
Williams et al. calculate that final assembly by Toyota makes up only 14% of the total
cost of production anyway, the rest being borne by suppliers using unrelated methods of
production. Williams et al. assert that Taiichi Ohno understood this primacy of
maintaining a smooth production flow by production levelling, flexible manufacturing
also helping to achieve this by assembling multiple models. Refinements such as this
have led to the progressive improvements in output reported by Rhys (1972, 1999), but
not the quantum change that would signal a paradigm revolution in car production. The
Nissan plant at Sunderland is recognised as the most

237
efficient in Europe, yet it is neither quantitatively nor qualitatively unique (Times 100,
2004) and so MBPS is largely unaffected. This indicates that the continuing
development of production systems has increased the ability to exploit the economies of
scale, but not necessarily redefined the overall MES? for an automobile company.

If a company finds that internal solutions of extending or intensifying production are not
available then it can try external solutions. In the make-or-buy dichotomy the simple
option is to make contractually regulated transactions. The problem with this is that it
introduces the very transaction costs that the full-function organisational structure avoids
and thereby permits those costs to persist in a structural form. Since the company is
attempting to get as close as possible to the full-function model expressed within the
automobile industry paradigm then an external solution should comprise more than
simply transacting for resources in the market. An alternative way by which it can
approximate to the automobile industry paradigm is through alliances with other
automobile manufacturers. The nature of such potential alliances will be discussed in the
next section.

6.3 External approximations to the automobile paradigm

Extemal relationships that are contracted for in the market are part of the make-or-buy
decision process and do not represent a novel form of relationship that might
approximate to the automobile paradigm. This describes the supply industry but it can
also extend even to vehicle assembly under contractual conditions. For example, the
contract assembler Magna Steyr assembles eight separate models for thee different
automobile companies at its plant in Graz, Austria. These models are variants
supplementing production of mainstream models made by the automobile manufacturers
themselves. There is a degree of asset specificity since the production facilities are
specialised, but the bilateral relationship is balanced by the intellectual property rights
being retained by the client automobile companies. Despite an output of 250,000 units a
year the plant is fed by six body plants and two paint plants when one each would
normally suffice. The firm is not, therefore, benefiting from access to economies of scale
and cannot do so as long as its clients insist on the separation of processes by product
range (Automotive News, 2006d). The clients are not exploiting

238
economies of scale through Magna Steyr either, though presumably there are cost
advantages over producing at their own plants. Furthermore, the relationship does not
represent a novel form of organisational structure since the plant supplements client-
owned assembly plants elsewhere. In essence, Magna Steyr is little different to the Tier
1 sub-assembly suppliers, albeit resulting in fully assembled variants of mainstream
models. The company is thereby supporting the automobile industry paradigm but it is
not operating within it.

When two companies form an alliance they are side-stepping the dichotomy of the
make-or-buy decision by entering into a conjoined activity or joint venture. The
structure of the alliance depends on the extent that the activity can be divided between
the partners. If the activity is not divisible then the partners will pool their resources
into one jointly held facility. This is a horizontal, or lateral, alliance where both parties
are able to expand their common capabilities in unison. In this situation, both
companies are comprised of integrated functions but are seeking to share one or more of
them in order to increase output, such as when entering a new market. For example, the
NUMMI assembly plant in California started in 1984 (JAMA, 2006), operated as a joint
venture between Toyota and GM, and allowed the Japanese company to share in the
scale benefits of the facility as it developed manufacturing in the US. The preceding
section demonstrated how an internal problem of scale could be overcome with an
internally generated solution, such as flexible production. An alliance or joint venture,
however, provides an extemal solution in the form of a partner.

However, if the activity is divisible then it opens the possibility of the partners taking
complementary positions, each company retaining its unique capability, in order to form
a vertical alliance. In such a situation, the partners are in search of a capability that they
had not previously intemalised but had perhaps been supplied to them extemally
through transactions. The vertical partnership might be intra-functional, each partner
contributing subsidiary skills within a function, and so each company gains access to a
new facet of the range of related skills. The two companies are individually full-
function but are seeking augmentation within one or more functions. Altematively,
when neither company has a full-function stmcture the partners might specialise in
complete, self-contained functions within a multi-product production system. When a
vertical alliance or joint venture of this nature occurs it is inter

239
functional. The partnership offers the possibility of the two companies sharing in a
combined full-function structure that grants access to complementary functions that are
entirely novel to each of them respectively.

6.3.1 Hybrid organisational structure

Williamson (1991) considered the possibility of an interim form of firm organisation

which neither made its transactions on the market nor regulated the transactions through
an intemalised hierarchy. This intermediate mode he termed the hybrid and it is a
stmctural form made up of an eclectic mix of the two polar opposites, market and
hierarchy. However, it is not at all clear how this renders the hybrid stmcture a distinct
form of organisational stmcture in its own right. The author gives, as an example, the
case of the contract between the Nevada Power Company and its fuel supplier, the
Northwest Trading Company. This is described as a neo-classical contract because it
anticipated ex post opportunities, in effect it recognised that opportunism could occur
and so enshrined mechanisms for dealing with it when it did so. At such a point the
“excuse doctrine” could be invoked, relieving the parties from strict legal enforcement
of the contract terms. In this way it escaped the rigidity of a classical contract which
fixed an ex ante agreement which was then enforced by law.

There is not, though, a clear demarcation between classical and neo-classical contracts.
If no contract can be complete, and all are open to some ex post exploitation, then
contracts of varying cormptibility should be compared as if on a continuum, not as
distinct options. Artz and Bmsh (2000, p.338) found in their research that, instead of a
choice between discrete forms of organisational stmctures, there was a continuous range.
This being so, then the so-called neo-classical contracts are simply contracts that are
more complete than others because they cover a greater number of contingencies. If
items in the contract carry no legal status then they are not binding and only the risk to
reputation remains. This may offer an opportunity for arbitration, yet this would exist
anyway, whether it was specified in the contract or not. In economic terms the pertinent
issue is whether there is a legal contract between the parties, or whether the transaction
has been intemalised into an integrated firm where:

240
“...hierarchy is its own court of appeal.” (Williamson, 1991, p.274.)

When Williamson puts forward franchise relationships between companies as being

hybrid forms he describes how the two parties have greater autonomy than in a
hierarchy, but this then entails greater use of rules than in a market in order to protect the
interests of the parties. The relationship between a franchisor and a franchisee appears to
be unique because of the complexity of what is being supplied. This includes product
branding, quality, pricing and even employment policies. A franchise agreement is
therefore one that regulates a multitude of transactions between the parties by bundling
them into one contract. Yet the essence of the contractual relationship remains the same,
it is only that in this situation the contract has been written in recognition of a complex
situation. It is therefore unlikely that the hybrid structure is one based on a new kind of
contract, all contracts being written to regulate specific relationships anyway.

6.3.2 Co-operative organisational structure

Many commentators have suggested that firms can work together in a manner that is less
dependent upon the mechanisms of the market and more dependent upon a spirit of co-
operation and a sense of shared destiny. This concept of working together to mutual
benefit may represent an alternative to the choice between internalisation or market-
based transactions and it has been approached from various stand points.

If a company’s planned output is located in that part of the LAC curve that exhibits
increasing returns to scale then, rather than produce by itself, it can elect to form a joint
venture in order to produce in a larger plant, one that approaches more closely the MBPS
output level. Kogut (1991) points out that joint ventures are useful in mature industries
where the LAC is already well known. Excess capacity can be eliminated at the planning
stage by combining production capacity with a partner so that together economies of
scale can be exploited.

Cleeve (1997) looked at Japanese firms entering the UK. Due to the indivisibility of
certain production technology the incoming firm might find itself having to invest in
more capacity than its intended output required. Naturally this would put the firm in a

241
position where increasing returns to scale existed, and hence higher unit costs than at
MEPS, a situation which could be enhanced by partnering production with another
firm. Hennart and Reddy (1997) also found that Japanese firms tended towards joint
ventures when economies of scale were larger than intended output, particularly if the
incumbent firms were already producing near the optimum,

Kakabadse and Kakabadse (2000, p. 114) looked at firm service activities which, being
generic in nature, were amenable to joint ventures or consortia:

“The key objective of such arrangements, therefore, is to achieve economies of scale

through a beneficial local source delivery to its consortium members.”

Weijian (1991) noted that additional benefits came in other areas, such as corporate
learning, which could have a significant impact on shifting the SAC curve as personnel
become more adept at operating the existing plant. Learning between partners indicates
that co-operation brings its own benefits. Wilding and Humphries (2006) putting
forward the concept of co-operation as an alternative form of organisational structure
that might minimise the need for contracts or vertical integration. This would then
support the possibility of approximating to a full-function model in a maimer that
replicated internalisation and did not involve transacting extemally. They found that
market-based negotiations brought increased costs when they became adversarial in
nature. The authors advocated behaviour, explained as co-operation using fewer
suppliers over a longer term, co-ordination of knowledge exchange and collaboration in
planning.

Like Williamson’s hybrid form, seems to underestimate the role that contracts
have in the relationship between two firms. Although the benefits of behaviour are
quite apparent, it is not clear how it can be put in place ex ante between new partners
without some kind of enforcement. Instead, behaviour is the ex post benefit of
aligning transactions with the appropriate organisational stmcture. In a previous paper,
Humphries and Wilding (2004, p .lll9 ), acknowledged that this condition of empathy
between organisations could not be arrived at immediately, tmst had to be nurtured until
the full benefits of the relationship could be realised:

242
“Organisations should attempt small, simple, co-operative projects that improve
efficiency because these are perceived as being non-threatening: discussions about costs
should be left until some maturity has been achieved.”

This seems to concede that co-operation can be developed if the incremental steps are
low risk, and thus the transaction costs are also low. However, this serves the purpose of
maintaining the equanimity of the partnership such that one partner cannot assume a
significant advantage over the other. Additionally, there is no reason why each “simple”
step should not be enforced by a “simple” contract, thus spreading the transaction costs
of specifying the ultimate contract over the relationship development period. At the
other extreme trust can have a malevolent aspect, as when one partner has a
disproportionate amount of power which it can be expected, or ‘trusted’, to use to its
own advantage. Blois (1972, p.268) argued that when a supplier depends to a large
extent on one customer then it is quasi-intemalised because of the customer’s
bargaining power:

“In some cases large customers are prepared to indicate that they regard their suppliers
as being extensions of themselves.”

The concept of regulating company relationships by a spirit of trust in place of either

contracts or internalisation has often been applied to Japanese companies. Dyer (1996)
used the principles of transaction cost analysis (TCA, see Chapter 5) to look at the
keiretsu form of alliance structure seen in Japan, particularly in application to the
automotive industry. These are vertical buyer-supplier relationships so it is described by
the make-or-buy decision of standard TCA but it results in an alliance form of
organisational structure. The author found that around 38% of component costs in Japan
could be sourced to the keiretsu alliance, as opposed to the United States which was
much more polarised with 10% attributable to partnerships and 48% produced
internally, the remainder being sourced in the market.

One explanation put forward for this difference is that American companies can call on
a legal system that is more closely adapted to the characteristics of market transactions
while Japanese companies are culturally orientated towards a trusting relationship.
However, this places too little emphasis on the keiretsu structure. Gilson

243
and Roe (1993) point out that a keiretsu is a form of organisation that anchors a network
of companies to a common core, often a bank, and each company is a part owner of the
network through a system of cross-holdings. This means that suppliers and buyers in the
network are also equity owners, creating a degree of integration that lowers transaction
costs and permits knowledge transfers. In a subsequent publication. Dyer (1997, p.548)
reiterated the role of trust but acknowledged that:

“A stock ownership position held by an automaker acts as a financial hostage which

encourages the supplier to make partner-specific investments.”

The asset specificity is therefore founded on a sense of joint ownership and acts as a
self-enforcing influence on the relationship:

“Self-enforcing safeguards can control opportunism over an indefinite time horizon.

Conversely, contracts control opportunism for only a finite time horizon. (Dyer, 1997,
p.548, italics in original.)

An altemative form of joint ownership might involve retaining specific assets within a
jointly operated production system, particularly if the technology is readily divisible.
Pires (1998) looked at the relationship between the vehicle manufacturer and suppliers at
the VW truck factory in Resende, Brazil. This plant is based in a consortium approach
where production is divided into distinct sub-assembly modules. Seven dedicated
suppliers have sole responsibility for production within each technologically separable
module and must guarantee delivery to the final assembly line. There were hopes that the
model could be extended to automobile plants but there are two crucial differences.
Firstly, the techniques used in truck production lend themselves to sub-division and,
secondly, truck manufacturing is a relatively new activity for VW. Pires also reported
that the company had found it more difficult to institute such a style of production at its
new Brazilian engine plant in Sao Carlos. Despite this, the model does illustrate that
asset specificity is one that can secure a bilateral agreement when the partners are co-
dependents using distinct technologies.

Mariti and Smiley (1983) recognise that firms may share specific activities through
horizontal co-operative agreements. This is not a make-or-buy decision because this is

244
not a supplier-buyer relationship. The two parties have already internalised the activity
and wish to share it in order to exploit further benefits. In particular, this allows them to
access economic advantages:

“...through co-operative agreements, firms can take advantage of economies of scale in

one or more of their production processes while remaining separate entities.” (p.439.)

Although the authors argue that this is not simply a market transaction since the
agreement excludes other firms, and therefore changes the competitive landscape, this is
also the purpose of contracts negotiated in the market. Even when a co-operative
horizontal transaction is non-monetary, often involving an exchange of knowledge or
pooling of expertise in a joint venture, it remains a transaction even if it does not include
physical movements of cash. For example, in 1998 Ford and PSA (Peugeot-Citroen)
agreed to work together on the new range of small diesel engines, the DLD family
(Diesel Progress, 1999). Both companies had existing diesel engine design capabilities
but in combining their experience they were able to realise economies of scope. This is
often the reason for engaging extemal engineering consultancies since it widens the
breadth of skills that are available for the project and the Ford-PSA relationship might be
seen to be in the same vein.

If the co-operative modes of inter-firm relationship so far described fail to offer an

altemative to the make-or-buy dichotomy, Dussauge et al (2004) elaborated on the
technological basis of the dichotomy and provide grounds for a new type of
organisational stmcture. The authors categorise collaborative relationships into scale and
link alliances. Scale alliances combine similar resources in order to seek economies of
scale in a horizontal joint venture. Link alliances, by contrast, are orientated towards
knowledge acquisition and combine complementary skills to allow the partners to access
skills they have not developed for themselves. This opens access to economies of scope
but it has ramifications for the longevity of the partnership because Dussauge et al found
that link alliances tend to be more volatile and result in one partner acquiring the other in
order to secure long-term access to the overall knowledge base. This is because, the
authors argue, link alliances are based on the ex ante uncertainty of potential knowledge
which, when gained, accrues to each

245
partner separately and gives rise to the possibility of asymmetric ex post opportunities.
Naturally, this is the kind of behaviour Williamson warned about in transaction costs
and reveals the central role played by human assets in link alliances. Altematively,
scale alliances involve joint learning and continuing efficiency considerations which
would be lost if the partnership were dissolved.

In practice the distinction might not be so clear cut. For example, one partner may have
the necessary skills but be lacking scale and so from that firm’s perspective the alliance
is about finding scale through a pooling of resources. On the other side of the
relationship the partner may be seeking economies of scope by access to its opposite
number’s wider knowledge base so will view it as a link alliance. This can be the
driving force behind the complementary partnership of a VJV. When the alliance is a
link-scale compound form it suggests that the manufacturing strong partner will prefer
to break off the alliance once it has found the knowledge it was seeking since it will not
need to intemalise its partner’s production capability. This will then leave the
knowledge strong partner marooned without access to scale production. In sum, the
link aspect deals with a specific problem which is then solved through the alliance,
while the scale aspect deals with a problem that is endemic to the system and is only
solved by its continuing operation. Although link alliances may be, by their nature,
short-term, subsequent sections of this thesis will show that a link-scale alliance in a
VJV can be sustainable depending upon the stmcture. The next sub-section of this
thesis will demonstrate the cmcial role that human assets play in business relationships,
with implications for link alliances.

6.3.3 Human assets in link alliances

Klein (1988) revisited the relationship between Fisher Body and GM which had
become a centre piece in the transaction cost argument (see section 5.3.5 above). The
author tackles the idea that contractual imperfections lead to opportunism and hold ups
that can only be resolved by vertically integrating the two parties into one firm. This is
particularly relevant to link alliances where the asymmetric development of the
relationship can lead to one partner holding an ex post advantage over the other.

246
The original contract between Fisher Body and GM had secured the relationship by
guaranteeing Fisher Body exclusive dealings with GM and thereby reduced Fisher
Body’s asset specificity risk. GM was in turn protected from price hold-ups by a price
protection clause in the contract. Joskow (1985) found similar contracts concerning
coal supplies for power generation but there would have been far less likelihood of
technical innovation changing the nature of the industry. The equanimity of the Fisher
Body-GM agreement was disturbed by the exponential increase in demand for closed
bodies which Fisher Body was in a position to exploit for its own gain.

Klein asserts that the long-term contract actually caused the problem by constricting
the relationship during a time of change. The essential feature of the relationship,
though, was not the physical assets owned by Fisher Body but the human assets, which
by their nature cannot be owned. The human assets can only be controlled by
employment contracts and it was at this level that vertical integration occurred,
according to Klein (1985, p.207):

“It is in this sense of owning a firm’s set of interdependent labour contracts and the
firm-specific knowledge embodied in the organization’s team of employees that an
owner of a firm can own the firm’s human capital.”

Vertical integration therefore has little impact on the number of contracts but since it
involves absorption of the organisational capital of the other company the nature of the
relationship is changed. It was no longer necessary for GM to pre-specify how physical
assets should be managed, such as production locations and output, and although the
requirement for their use remained without the binding force of legal constraint there
was far greater flexibility. Similarly, a link alliance constrains the main driver of
innovation, knowledge, which is held by the human assets. If a firm is seeking specific
knowledge then it will quit the link alliance when it has acquired it sufficiently.
Altematively, if it has a continuing need to develop the link alliance more than its
partner then it will intemalise the activity. Although the Fisher Body-GM relationship
might appear to be a scale alliance enforced by a contract, in fact it had the character of
a link alliance providing GM with access to steel body technology. Since GM’s future
depended on being able to direct the human assets involved it had the greater need to
intemalise the operation.

247
In cases where one partner is seeking specific knowledge then opportunism manifests
itself in the form of early termination of the relationship once the knowledge has been
gained. For example, Hill et al (1990, p. 119) emphasise the risk that is attached to
entering foreign markets by the mechanism of licensed production. The authors cite the
example of RCA having its colour TV technology expropriated by Japanese licensees.
The picture is more complex when the relationship is a link-scale compound. When
Toyota commenced production in the US it did so with the NUMMI joint venture plant
with GM. Although this provided both companies with scale it also permitted them to
leam about each other’s manufacturing techniques. As predicted by Dussauge et al
(2004), the link alliance aspect was not extended as Toyota went on to establish wholly
owned production sites, but the scale aspect was strong enough to keep the joint venture
plant in continuing operation.

A compound link-scale alliance that seemed to hold long-term potential in the UK was
that between Rover Group and Honda. In 1979 Honda was still a young company,
having begun automobile manufacturing in 1963 and exports to the US in 1970. There
had been an early attempt at a European production plant in 1963, manufacturing
mopeds in Belgium, but this had proven to be more of a challenge than the company
could manage:

“The Belgian experience was one of hardship in every aspect of work, including
production, sales, development and management.” (Honda, 1999, p. 137)

The venture exposed the lack of market knowledge and poor adaptation of the product
to European tastes. The experience seems to have impressed on Honda its lack of
experience in the region. Indeed, its comprehensive lack of economies of scope in
management, R&D and product range was quite conclusive. The subsequent
introduction of the N600 small car into the US market appears to have been rather
tentative, being initially released into Hawaii. The contiguous US was entered with
greater commitment with the introduction of the Civic in 1973 and then the Accord in
1975 (Honda, 2004).

248
After the apparent debacle in Belgium, the automobile market in Europe was also
approached with caution. Honda lacked economies of scope in manufacturing
knowledge and marketing to which it could gain access via a link alliance. At the same
time, the financial damage caused by the Belgium production site made a scale alliance
in a shared production facility equally attractive until sufficient demand justified a
wholly owned operation. For Rover Group production capacity was not an issue but it
was in need of vehicle designs. Although the company probably had the strategic
resources in R&D at the time to design its own vehicles, the vicissitudes of the previous
few years had conspired to leave it with a tactical need for a new model to maintain its
production and market presence in the short-term. For Honda it was therefore both a link
and a scale relationship, for Rover Group it was purely a link alliance.

The link-scale theory of Dussauge et al (2004) would have anticipated that the link
alliance aspect would last only as long as the need for knowledge persisted while the
scale aspect would endure. As will be discussed in greater detail in Chapter 7, Honda
soon gained economies of scope from its experience of European manufacturing and
markets and the creation of its own assembly operation by 1989 gave it greater access to
economies of scale in addition to sharing production with Rover Group (JAMA, 2006),
just as had Toyota and GM. This was not reciprocated by Rover Group which revealed a
chronic lack of economies of scope in R&D, being capable of developing its own
designs for some vehicles, particularly at Land Rover, but barely able to transfer these
skills to its mainstream models.

As Dussauge et al (2004) assert, a link alliance becomes destabilised once just one party
has gained the knowledge it was seeking. It seems that the Honda-Rover alliance was
anchored by the physical asset specificity of the scale aspect in joint manufacturing, and
this secured the continuity of the relationship in order to maintain the human assets of
the link aspect. As Chapter 7 will show, the alliance was ended by the acquisition of
Rover Group by BMW in 1994, which broke the scale aspect of the Honda-Rover
relationship and the Japanese company was able to extricate itself from the link aspect.
Even so, from that point Rover Group, and its successor MG Rover, retained the Honda
sourced technology until the company’s demise in 2005.

249
However, without access to subsequent Honda developments via the link alliance the
product technology gradually fell behind the competition.

The complexity of compound link-scale alliances suggests that firms will work together
for a variety of reasons, many of which they may not share in common. Firms are not
simply motivated by scale or knowledge, there are other factors involved such as
differences in the production environment, particularly when the companies operate in
spatially diverse regions with inherent factor endowments. Globalisation has increased
access to these factor endowments and so suggests that the IVJV may encompass link-
scale considerations while offering the possibility of a new form of organisational
structure outside of the hierarchical-transactional dichotomy. The possibilities raised by
globalisation will be discussed in the next section.

6.4 Opportunities raised by globalisation

Globalisation of trade has brought economic benefits by allowing countries to exploit

their comparative advantages and make better use of their factor endovraients. The
mechanism by which this is achieved is specialisation which allows nations to focus on
their strengths and trade the surplus in return for those goods in which their trading
partners specialise. Although this is a simplified view of international trade, for example
trade will also involve exchanges of goods which are unavailable to the importing
country, this section of the thesis will show how international trade has accelerated over
the past few decades. However, this trend has also brought with it a great deal of
criticism.

The development of international trade based on the principle of free markets in turn
offers new opportunities to companies. Instead of being part of an international
exchange system, companies can become enmeshed in the economies of other nations
through foreign direct investment (FDI). For this to occur, the investing firm needs to
hold some advantage that the domestic firms do not possess. At the same time, the
foreign location must present an opportunity to the foreign firm that exceeds the risks
involved in operating in an alien environment.

250
Automobile firms have been early explorers of international locations but their heuristic
evolution has highlighted the complexity of engaging in FDI. Companies that have
attempted to internationalise find that they must account for the benefits of the foreign
location as well as the advantages they bring to it. Since automobile firms comprise
different production systems a single mode of market entry for the firm is not
applicable. International factor endowment differentiation can mean that the firm needs
to approach the new market in a selective fashion, with perhaps only certain processes
being amenable to FDI.

Automobile firms may also find that globalisation presents opportunities that outweigh
disadvantages that it had been suffering in its home market. It might allow a firm to
access factor endowments that may favour certain functions over others but which were
not readily available in the home base. In this way, globalisation opens the possibility of
geographically dividing the full-function integrated structure according to the
availability of factors of production. For a company that is operating beyond the
minimum output (MESp) prescribed by the automobile paradigm such a strategy would
be of incremental benefit unless the company restructured the existing full-function
structure around the new possibility. For such a firm, at or beyond MESp, FDI is used to
enhance advantages it already possesses. The possibilities for a firm operating well
short of the optimum prescribed by the paradigm, though, are much greater because of
the opportunity to restructure around a new location. Furthermore, because the
opportunities are applied only to discrete processes, depending on whether they are
capital or labour intensive, it creates the greater potential for vertical relationships rather
than horizontal. In this section the nature of globalisation and the opportunities it offers
for automobile companies will be discussed, showing that the IVJV is the organisational
structure that best approximates to the automobile paradigm.

251
6.4.1 Globalisation of trade

One of the most powerful arguments for the gains to be made from the international
movement of goods is the Heckscher-Ohlin (HO) model. This uses differences in
national factor endowments to explain how countries are biased towards either capital or
labour intensive production, assuming that the underlying technology is commonly
accessible. Although this should lead to specialisation by countries in production to suit
their factor endowments this is inhibited by diminishing returns. For a country endowed
with labour the progressive attraction of that factor to production will lead to the factor
price of labour being bid higher. Conversely, the factor price of labour will fall in a
country endowed with capital just as the factor price of capital is bid higher. The two
national trends will converge until factor prices have equalised and the conditions for
international specialisation have been removed. Schott (2003, p.686) argues that this
implies that international industrial development has an end point for each technical
process:

“This single cone version of the model has all countries of the world producing all
goods, so that both Japan and the Philippines, for example, are assumed to produce
identical electronics and apparel goods using the same techniques.”

Based on empirical evidence, Schott (2003) takes the view that industry develops in a
dynamic manner but always originating from its factor endowments. Convergence is
therefore a tendency, not an absolute. Schott’s “multiple-cone equilibrium” would have
the Philippines tending towards production of labour intensive clothing and Japan
tending towards production of capital intensive high-technology goods. This is even as
factor price convergence pulls in the opposite direction with a tendency towards
equilibrium in labour and capital. Although countries, or regions, rich in labour will
attract capital to a degree, as new industrial developments emerge countries will revisit
their persisting factor endovmients.

Deepening the HO model in this way brings greater insight to the subtle distinction
between the long established practice of internationalisation and the recent rise of
globalisation. International trade has a long history and Peter Dicken (2003) describes
how this has encompassed many commodities, such as spices, but adds that

252
production processes used to be organised within national barriers. Trade took place as a
result of the factor endowments described by the HO model with a flow of raw materials
from the periphery to the core and a return flow of manufactured goods (see Figure 6.1).
It is by the mechanism of international trade that companies can intemationalise,
geographically spreading operations to take advantage of the different factor
endowments.

Figure 6.1 Pre-global trade

Core Periphery
Production of Source of raw materials and
manufactured goods. foodstuffs. Market for
manufactured goods.

Source: Dicken (2003)

Berger (2000) points out that there has existed a common understanding since the 1990s
that the international economy has changed such that it is now embodied by a single
market in goods, services, capital and labour. As predicted by the HO model, the
dichotomy caused by the factor endowment differential has been replaced by a blending
of the two sides such that the periphery no longer describes a location, the site of raw
materials also hosting a manufacturing base. Although internationalisation worked
through the mechanism of international trade, globalisation removes the distinction
between locations. As Sturgeon (2000, p.3) states:

“The underlying hypothesis is that the more that national economic systems come to
resemble one another, the fewer barriers will exist to the flow of resources to their most
efficient use, and the further the world economy will become integrated, or globalized.”

Yet Sturgeon and Florida (2000) anticipate Dicken (2003) in finding that the ultimate
state of globalisation is, in reality, a continuing process and that it has yet to supplant
intemationalisation. Furthermore, Sturgeon (2000) points out that the degree to which
industries can exploit globalisation is uneven and the author finds evidence of
globalisation in capital markets, regionalisation in supranational clusters of industry

253
and of intemationalisation in national trade policies. Dicken (2003) also distinguishes
between intemationalising processes, which are characterised by quantitative increases
in economic activity between nations, and qualitative globalising processes, which
include the integration of economic activities across nations. On this basis, while global
markets are measured quantitatively the globalisation of production is essentially
qualitative (Sturgeon, 2000).

Applying this to Schott (2003), quantitative increases in intemational trade will not only
bring national industries to a degree of convergence, the increase in opportunities will
encourage further innovations that will again differentiate between relative factor
endowments. Schott enriches this view of factor endowments by including factors
beyond capital and labour to take in factor efficiency, demand, political policies and so
on. Yet this complexity is not always well understood by the companies that seek to
exploit the advantages, nor the governments that seek to control it. Concern has been
voiced on the changes that globalisation is bringing and how companies might use it to
challenge existing political authority.

6.4.2 Critics of globalisation

There is little doubt that the globalisation of intemational trade has brought many
economic opportunities. Hill (2003) considered the influence of capital flows, as
manifested by foreign direct investment (FDI), quoting the United Nations figures of
$60 billion in 1980, $210 billion in 1990 and $732 billion by 1998. Hill infers that this
shows the rise of intemational production systems and the spread by organisations into
the markets of their foreign competitors. The caveat to this is that it involves only a
select number of countries, with ten developed countries receiving up to 70% of the
world’s FDI in 1999. In the same year, Africa received a meagre 1% of the world’s FDI
(Hill, 2002, p.6). Rugman and Hodgetts (2001) also show that while the majority of
intemational trade is between the triad of the North American Free Trade Agreement
(NAFTA), the European Union (EU) and Asia, an even greater proportion of the trade is
within the individual elements of the triad. Figure 6.2 illustrates this using direction of
trade proportions for 1997.

254
 Figure 6.2 Direction of trade 1997 - inter-triad and intra-triad trade movement
proportions
NAFTA
Inter-trade: 50.9%
Intra-trade: 49.1 %

Asia EU
Inter-trade: 46.9% Inter-trade: 39.4%
Intra-trade: 53.1% Intra-trade: 60.6%

Source: Rugman and Hodgetts, 2001

Cowling and Sugden (1998) suggest that the inequality of trade patterns show how
trade is not so much free, more that corporations are free to act across intemational
borders. In this context, freedom means the ability to implement strategic decisions.
Since these firms are controlled only by those who have a direct vested interest in it, the
strategic decisions that drive intemational trade are made by the senior management of
the corporations. Cowling and Sugden (1998, p.347) conclude that:

“...freedom in free trade is the freedom of an elite to manage intemational trade in their
interests, despite the objections of others.”

Cowling and Tomlinson (2005) state that the concentration of power in a few
transnational corporations puts developing nations at a disadvantage. One such
objection came from the Malaysian Prime Minister, Dr Mahathir Bin Mohamed, who
laid the blame for the East Asian crisis of 1997-8 on the huge new intemational
corporations:

“Their strategy is simple. Become so big that no one else can compete. The small must
either allow themselves to be swallowed or suffer failure very quickly.” (World
Economic Fomm, 2002.)

Some commentators believe that the perceived iniquities engendered by globalisation

will lead to retaliation. Polanyi (1963) argues against the commoditisation of three

255
elements of production: labour, land and money. Essentially, the author sees these
elements as being fictitious commodities: land, because it is not in production, and
labour and money because they are not produced for ultimate sale. Indeed, they are
fundamentally social activities, and so society rebels against their commercialisation in
a self-regulating market (Silver, 2003). Furthermore, taking labour out of the market
will bring about a transformation as great as the original establishment of the
competitive labour market (Polanyi, 1963). Munck (2002) points out that a worldwide
expansion of capital will also lead to an expansion in the proletariat, increasing the
importance of labour. Even if the workforce has not been globalised yet:

“...there is the tendency towards the creation of a global labour market” (Munck, 2002,
p. 11).

Countering this are social forces that would retain the differences in factor
endowments. Trade unions such as Britain’s Transport and General Workers Union
(TGWU) are sensitive to the risk of jobs being moved overseas as a result of the
opportunities presented by intemational recmitment (Benady/TGWU, 2003). Cowling
and Tomlinson (2005) note the existence of intemational institutions, such as the WTO,
and the rise of consumer groups that seek to force a “corporate responsibility”
programme on corporations, but nevertheless conclude that the power of MNEs to
conduct their own strategies has merely been mitigated, not banished. Although Bartlett
and Ghoshal (1993) suggested that the M-Form of corporate structure was unsuited to
coping with making resource allocation judgements for globally dispersed operations,
the example of Ford seems to show that control from the strategic headquarters can be
highly effective. The company’s strategic management compared information gathered
at individual plants as the basis of threats to the local employees for moving production
elsewhere (Beynon, 1984). The company was also able to use the information to put
pressure on political authorities, one example being the 1977 engine plant in Bridgend,
South Wales. Of the £180m required £115m came from the British government.

According to Bamet and Miiller (1974, p.41) the strategic management of a global
company, as seen in the M-Form structure, is able to coordinate all the factors of
production, whether mobile or immobile:

256
“It is this ability of World Managers to weigh all these factors and to co-ordinate
decisions on pricing, financial flows, marketing, tax minimisation, research and
development, and political intelligence on a global level that gives the world
corporation its peculiar advantage and extraordinary power.”

Hymer (1972) claimed that this represented a unification of world capital and labour
which reduced the independence of nation-states. This view is based on Hymer’s theory
concerning the manner in which MNEs can use the advantages they have developed in
their ovm markets to exploit advantages in new markets, despite the fact that they are
operating in an alien environment. This will be discussed in the next subsection.

6.4.3 Foreign direct investment

Although the comparative advantage theory of intemational trade can trace its lineage
back to David Ricardo in the 18^^ Century, the manner in which firms can use it as an
enabler for their own intemational stmcturing has only recently been articulated. Until
Hymer proposed a new approach to foreign direct investment (FDI) in 1960 the
accepted view of such foreign commitments was that they were a form of portfolio
investment (Rayome and Baker, 1995). Portfolio investment committed funds which
exploited differences in interest rates but without securing operational control (Hymer,
1960). Hymer took the view that this did not sufficiently explain FDI since funds for the
foreign operation tended to be raised locally and, in any case, the resultant distribution
of FDI did not follow the availability of low interest rate opportunities but industrial
ones. Instead, Hymer concluded that firms would wish to have some control over the
foreign operation in order to reduce competition and secure the full financial benefits of
their capabilities. The funds that were raised locally to fund the investment were only
benefiting from interest rate discrepancy by coincidence; the main purpose was to obtain
sufficient funds to achieve at least minimum control of the project. Indeed, by exploiting
a low interest rate environment for FDI the flow of capital would be the reverse of that
predicted by the theory of intemational capital movements (Hymer, 1960).

257
Hymer’s theory arose from the imperfections in the intemational market caused by the
lack of integration between national markets. This meant that firms could leverage the
advantages they enjoyed in their home market, perhaps because they had developed
there earlier, and apply them in new foreign markets where companies had yet to evolve
to the same state or did not have access to the same resources. The intemationalising
firm would thereby find additional advantage in the foreign market over its home
location (Yamin, 1991). However, this advantage could only be exploited by drawing on
resources in both countries and the advantage would have to overcome the
disadvantages of operating a business in an unfamiliar environment (Kindleberger,
1984).

In situations where an existing local firm is, or has the potential to be, fully competitive
with the intemationalising firm then Hymer contended that FDI would eliminate
competition, perhaps through merger, acquisition or some other form of collusion. This
occurs when the firms are in an oligopolistic situation such that a coordinated
relationship would bring increased rewards. Hymer’s theory was further refined by
Kindleberger in order to articulate the nature of the market imperfections, culminating in
the Hymer-Kindleberger theory of FDI (Rayome and Baker, 1995). These imperfections
included:

• Departures from perfect competition in goods, for example product differentiation

and pricing strategies.
• Departures from perfect competition in factor markets, for example patent laws,
differential access to capital and management skills.
• Economies of scale.
• Political intervention

However, the Hymer-Kindleberger theory has been criticised for focusing on the
endowments of the investing organisation and for not including costs that might apply to
such factors as the planning of the investment or the financial burden of the acquisition.
The theory was therefore not predictive of the precise form of resultant corporate
structure or how it came into being (Rayome and Baker, 1995). This was because Hymer
appeared to be unaware of Coase’s theory of the role of transaction

258
costs in structuring firms, a criticism that is not shared by Horaguchi and Toyne (1990).
They point out that Hymer was not simply describing FDI in terms of the proactive
exercise of market power, but also reactive in attempting to minimise costs through
internalisation. Hymer (1970) noted that this influences the govemance structure of the
multinational firm. In its local operations it needs to adapt to local conditions using
decentralised decision making, while in its intemational operations it needs centralised
decision making in order to coordinate the flow of information.

“They must therefore develop an organizational stmcture to balance the need to

coordinate and integrate operations with the need to adapt to a patchwork quilt of
languages, laws, and customs.” (Hymer, 1970, p.445)

This seems to suggest that the M-Form of decentralised govemance would suit domestic
firms while the U-Form would provide greater control over the disparate foreign
fimctions. The empirical evidence that was discussed in Chapter 5 of this thesis
similarly found that intemational companies tended to use a U-Form with extensions
into overseas markets.

Caves (1996) took a new perspective on Hymer’s theory when observing that FDI
tended to occur in a select number of industries. He noted that as the national firm
considers FDI it needs to confi”ont the costs of the expansion. Transaction costs imply
that the firm will exploit opportunities in the domestic market until such a point that the
marginal retums to its expansion begin to decline. The empirical data reviewed by Caves
indicates that firm size is the most reliable indicator of the move abroad, yet firm
specific factors will have enough impact to render the decision details largely
unpredictable. Since the intemationalising company carries disadvantages of being alien
to the new environment it will seek to minimise the information costs by approaching
the most familiar foreign markets first. At the same time, it has already gained an
information advantage, perhaps conceming production technology, in its home market
which can be used to offset the other disadvantages of operating in an alien environment
(Calvet, 1981).

Buckley and Casson (1998) consider the theory by Caves to be static since it focuses on
firm-specific competitive advantage, choice of location and determination of firm

259
boundaries. Although this is not to say that Caves is in error, the theory is incomplete
by not taking into account market volatility, the role of uncertainty and managerial
capabilities. Flexibility is the key to the actions of the company within an intemational
environment that is in a state of flux, so the accompanying theories of FDI need to
reflect the dynamic developments.

6.4.4 Dynamic theories of FDI

Vemon (1966) takes the dynamic view of global operations in his theory of product
cycles. The cycle follows the product from its initial development and introduction to
product standardisation and maturity. The theory considers matters of economies of
scale, the timing of innovation and the distribution of knowledge. The argument is that
entrepreneurs in a location will have the best knowledge on products to release and thus
innovation in that product will be initiated in its home market.

“All of these considerations tend to argue for a location in which communication

between the market and the executives directly concemed with the new product is swift
and easy, and in which a wide variety of potential types of input that might be needed
by the production imit are easily come by.” (Vemon, 1966, p. 195)

As the product evolves during its life-span it passes through three main stages (Jones
and Wren, 2006):

• Product development process - product specification and production process

uncertain, production located close to intended market
• Maturing product - established product and production process, economies of
scale take precedence over location
• Standardised product - fixed product and production process, low pricing strategy
takes precedence

As an example, Vemon shows that high labour costs in the US have meant a growth in
such innovations as automatic washing machines and drip-dry fabrics. During the first
stage in the product’s development in the market place there are strong locational

260
factors, such as a need for close control over the evolving product and communication
with the market demand. As the product gains in popularity it matures into a
standardised product which is generally acceptable, price then becoming the
differentiating feature. Consequently, economies of scale come to the fore and
production location may eventually move to a low labour cost region. Thus, as the
industry grows, supply shifts from home production for the home market to export for
foreign markets and then imports back to the home market from foreign production
sites. Should the popularity of the product decline then companies will begin to exit the
industry while the remainder restructure. Gal vet (1981) sees the shift in emphasis of
location as being a response to the loss of competitiveness as the uniqueness of the
product is diffused, international expansion being the attempt to capture the remaining
rent from the product’s development.

Antras (2005) takes Vernon’s point on communication and analyses this from the
transaction cost perspective, communication being expressed as friction and therefore
incurring transaction costs. Whilst these costs are high, during the period when the
product is still poorly defined and market demand has not yet coalesced, R&D and
production are internalised. Even when the product has become standardised and
production has moved abroad then Antras argues that the move into the new market
creates uncertainty due to the lack of knowledge and this is then alleviated by FDI
which maintains the internal functions of the firm. Vernon (1979, p.256) similarly took
the view that firms would reduce risk that was derived from their lack of knowledge:

“...firms are acutely myopic; their managers tend to be stimulated by the needs and
opportunities of the market closest at hand, the home market.”

Vemon (1979) emphasised that R&D tends to reflect the resources and market
characteristics in which it is located. Yet even when there are cost advantages to
producing in a foreign location the firm may still not have enough information to make
the decision to restructure. This may require a “trigger”, a point at which a threat
becomes clearly quantifiable. This threat may have varying relevance depending on
where the firm is on the product cycle. Almor et al. (2006) divided the firm’s activities
into three parts: R&D, marketing and production. They see R&D as

261
determining the firm’s competitive advantage during the early introductory phase,
supplanted by the importance of marketing during the growth phase and finally
production during the product’s maturity phase. The locational advantages remain
constant throughout but become relevant to different firm activities as they come to
prominence. Early R&D, as Vemon noted, has a strong bias towards the home market
but as the product output grows marketing takes an intemational perspective followed
by the intemationalisation of production.

Almor et al (2006) note that with three factors driving the intemational growth of the
product, intemationalisation of the company cannot be measured by the location of
production alone. Indeed, a significant problem for Vemon’s product cycles is when it
concems a class of product, such as electronic devices or automobiles, which are
conceptually standardised yet being continually updated. Klepper (1996), instead,
prefers to take the industry as the unit of analysis and refers to the industry life cycle
where innovation in product and process is developed concurrently. In such a
circumstance, technical R&D is likely to be retained in its original home location while
market oriented R&D would be distributed throughout its target areas in order to
develop local variants. Production might drift towards low labour cost locations but
continuous R&D innovation in processes would still require a substantial presence in the
home market.

FDI is therefore not a simple matter of systematically moving production at

predetermined points in the product’s cycle, although this may indeed be one of the
factors to be taken into consideration. Buckley and Casson (1981) investigated the
factors that need to be accounted for when deciding on FDI and found that in addition to
the costs of servicing the foreign market it was also important to evaluate the demand
conditions and future growth potential. Porter (1986) gives four factors that lead a
company to seek competitive advantage through global expansion. These are:

• Economies of scale
• Co-ordination advantages
• Proprietary teaming curve
• Comparative advantage of the location

262
However, these do not necessarily apply equally to all globalisation situations. The first
brings competitive advantage if it enhances the ability of the company to approach or
achieve economies of scale but would inhibit an intemational move if the new
production site was a substitute for existing output. Having coordination advantages is
relevant only if it is applicable in the new location to a greater degree than the
incumbents.

The proprietary teaming curve is gained in the home location but this can be disturbed
by an intemational move that introduces new elements. For example, in an assembly
operation the position on the teaming curve will have been achieved after experience has
been accumulated by the workforce, and although much of the technical teaming can be
transferred, the workforce in the new location will, nevertheless, have their own
operational teaming curve to climb. Under these conditions, a firm would expect to use
its established home base as a foundation for exploring intemational expansion
incrementally. This suggests that a firm would retain its existing govemance stmcture as
far as possible, such as the U-Form or M-Form, depending on the complexity of the
organisation, rather than implementing a new govemance stmcture that was flexible
enough to accommodate regional differences. Yet as Vemon (1979) observed with the
“myopic” management, the status quo of the current industry leaders may inhibit them
from taking the risk of restmcturing and so destabilising their competitive positions.
This would involve the firm progressing in stages, establishing or leaming at each stage
before going on to the next. The Uppsala Intemationalisation Model takes one such
behavioural approach, the firm deepening its foreign involvement due to the
management push to leam and the pull of market demand (Buckley, 2002). It is made up
of four stages. These can be summarised as follows (Barkema et al., 1996):

Stage 1 - no regular export

Stage 2 - exports through agents
Stage 3 - offshore sales subsidiaries
Stage 4 - overseas production

Clearly, Stage 1 shows no significant intemationalisation even if no firm is entirely

insulated from the greater commercial world. Stage 2 takes a product into the
intemational market via agents but has little effect on the operation at home

263
(Andersen, 1993; Viitanen, 2002). Stage 3 brings a commitment to a specific overseas
market, possibly with some custom product design, and the company might establish a
degree of physical presence through franchising, licensing, joint ventures and strategic
alliances (Hill et al, 1990). The progressive intemationalisation culminates in Stage 4
where the firm is operating in the foreign market as an incumbent. At this point, the
firm might claim to be globalised if the operations are integrated with those in the home
base. Conversely, if the existing govemance stmcture is retained, foreign operations
might also constitute a simple extension of the domestic operation.

It should be emphasised that stage models describe the company ratcheting-up its
exposure to foreign conditions and this implies that it is going through different levels
of commitment, yet it would be facile to state that globalisation is reached
incrementally. Any corporate activity is incremental if the opportunities are fertile; to
state that globalising companies expand by stages is simply to describe the growth of
success. The only types of firm that can avoid any kind of incrementalism and
instantaneously globalise are those of the ‘bom global’ kind. Bell et al (2001) have
described such firms as small, entrepreneurial organisations that can draw on a
sophisticated knowledge-based infrastmcture but this hardly describes the large scale,
technically mature automobile company described by the automobile industry
paradigm. Aw and Batra (1998) in a study of Taiwanese manufacturers found that small
and medium firms used geographical diversity in place of the product diversity enjoyed
by large firms. This indicates that FDI is not a strategy open only to those firms with a
strong, home base but also those that are seeking to counter disadvantages at home or
whose opportunities are global by nature.

Johanson and Vahlne concede that there are exceptional circumstances when stages can
be omitted because the leaming is inconsequential or has been gained elsewhere. The
three exceptions cover instances when company resources are large enough to render
the risks of intermediate steps relatively containable, situations where the market is
homogenous and occasions when the market is similar enough to other markets for
experience there to be relied upon. Each of these exceptions seems to describe large
automobile manufacturers that can expand in units of MBPS (i.e. constmcting new
plants that have all the size benefits of MBPS) within a reasonably

264
globalised market. Since the stage model theory focuses on the role of the company one
might also add the opportunity to leam from other companies, or first movers.

Johanson and Vahlne (1990) defend the stage model theory because leaming effects
account for a reverse flow of information which then influences the progress of
intemationalisation towards globalisation. This two-way process is part of the
integration that characterises the degree of integration necessary for globalisation.
Indeed, it may be a problem for any theory of intemationalisation that the nature of
globalisation is progressing and reducing the alien nature of foreign locations. This does
not mean, however, that there is no differentiation between home and foreign locations,
but it does mean that it has become more complex and subtle than Hymer originally
envisaged. The eclectic theory, or paradigm, has arisen to draw together the different
theories of FDI by applying them to different perspectives. This is discussed in the next
sub-section.

6.4.5 The eclectic paradigm

Although company leaming must influence the progressive intemationalisation of a

firm, the stage model theories do not specify what knowledge is necessary in order to
embark on the next step. Not only are the theories not normative but, in allowing
exceptions, they also fail to be completely descriptive. They therefore connect poorly
with the HO model since they do not emphasise the complex differentiation of factor
endowments.

Cheng and Kwan (2000) studied FDI in China and found that local market size,
infrastmcture conditions, wage costs and political policies were all positive factors for
inviting foreign investment into the country. Furthermore, FDI seemed to be self
reinforcing in that once the process had started it was easier to encourage more,
suggesting that the FDI companies had their own intemal factors for making the
investment. This level of complexity cannot be contained within the foregoing theories
of FDI but instead is better served by the eclectic theory which seeks to reconcile the
different approaches (Dunning, 1979). The eclectic paradigm is more able to encompass
such complexity since Dunning (1979; 1981) formulated it to include factor
endowments in three forms, known as the OLI parameters (Ownership-

265
Location-Intemalisation). These encompass the constituent approaches as follows
(Dunning, 1979):

1. Ownership advantages - industrial organisation and ownership theories

2. Locational advantages - theories of intemational trade
3. Intemalisation advantages - theories of vertical integration

This takes a broad view of factor endowments and includes various kinds of assets.
Although the parameters are customarily referred to in terms of OLI, Dunning (1979)
states that for FDI to take place they are conditions which should be satisfied in a
sequence better described by ownership-intemalisation-location:

1. The firm has possession of net ownership advantages, in the form of intangible
assets, when compared to other firms in the same market.
2. It is then more beneficial for the firm to exploit these advantages intemally
rather than pass them on, by sale or lease, to an extemal source.
3. The preceding two advantages are better utilised in conjunction with a local
factor input in the foreign market (e.g. natural resources)

Dunning (1980) summarises the OLI elements as the ability to call on resources that the
competition cannot, the extent that the company’s resources are better retained within
its stmcture rather than being obtained from an extemal source and the profitability of
exploiting these resources in conjunction with the indigenous resources of a foreign
location. Dunning (1979) wams that the theory does not indicate which firms will
engage in FDI or where it will take place, although Dunning (2000) identifies four main
types of firm which can then be linked to other theoretical approaches:

1. Market seeking - Uppsala Stage 2

2. Resource seeking - HO model factor endowments
3. Efficiency seeking - HO model specialisation (Schott, 2003)
4. Strategic asset seeking - enhancement of ownership advantage (Erdener and
Shapiro, 2005; Bailey et al, 1994)

266
It is notable that the theory acknowledges that advantages are not evenly distributed,
opening the possibility for more dynamic approaches to FDI, such as Schott’s (2003)
theory of intemational trade. The eclectic paradigm shows production processes can be
separated into constituent parts and transferred globally in order to exploit differentials
in factor endowments. As Schott demonstrates, this does not conclude the process of
globalisation since new developments will change the relevance of factor endowments,
even within Dunning’s OLI categories. This has already been observed within Vemon’s
product cycle, but it is tempered by the TCA approach which would inhibit the infinite
division and dispersal of production processes, instead restricting this to technically
separable processes.

This needs to be bome in mind since Dunning’s approach, by the fact that it is eclectic,
offers a myriad of possibilities. Williams (1997) emphasises that ownership deals with
intangible assets, such as the access to parent facilities at a cost below the market rate in
the new setting or the benefits of existing economies of scale. Dunning (1979) points out
that a country might improve its stock of intangible assets by promoting a higher than
average expenditure on R&D. This would have the consequence of raising its revealed
comparative advantage. Ownership is also a control issue, comparable to Porter’s (1986)
co-ordination factor, and introduces some sense of human asset specificity. Locational
advantages are more closely related to the factor endowments described by the HO
model, though Williams (1997) explains that these include, amongst others, trade
barriers and tax regimes. For example, UK government policies have influenced choice
of location, from intervention by way of the Industrial Development Certificate of the
1960s (Adeney, 1988), to inducement by way of financial incentives later in the 1980s
and beyond (Garrahan and Stewart, 1991). Dunning (1979) explains that this means
locational advantages are dependent on the fit between the characteristics of the location
and the nature of the investing firm. Dunning (2000, p. 178) seems to anticipate Schott
(2003) in noting a shift from the attraction of natural resources, or factor endowments of
the basic HO model, towards:

“...a distinctive and non-imitatible set of location-bound created assets.”

267
Even these can change. For example, Ford of Canada being established to supply the
nations of the British Commonwealth from inside the tariff regime (Wilkins, 1979).
Subsequently, Anastakis (2004) observed the shift in ownership of Ford of Canada from
a semi-autonomous operation to an intemal division of the Ford Motor Company once
the trade barriers between the US and Canada had been dismantled. Locational
advantages can also include the existence of local industry clusters offering economies
of concentration that the incoming firm can exploit using its existing ownership
advantages (Buckley and Ghauri, 2004).

Site specificity takes a narrower view of plant location advantages, including local
rather than national characteristics. Table 6.1 contains a short summary of three
automobile plants in the UK and the reasons given for the selection of site. The
common features of this list are the existence of good transport communications, a large
labour pool and the availability of an existing industrial site. Nissan and Honda
constmcted new factories over the sites of old airports but with excellent road
communications. Ford sought out port access and built on a greenfield site, more
accurately described as a marsh, paying the price by having to build the foundations on
22,000 concrete piles and exceeding the budget by £2m (Collins and Stratton, 1993).

Table 6.1 Plant site selection criteria

Plant Start of Production Reasons for Site

Ford, Dagenham 1931 Logistical communications

Labour pool
Nissan, 1986 Existing site
Washington Local labour pool
Government incentive
Honda, Swindon 1992 Existing site
Logistical communications
Labour pool

Sources:
Dagenham, Washington: Georgano, 2000
Swindon: Mair, 1994

The intemalisation element of OLI is directly related to TCA and the avoidance of
market imperfections. This is pertinent when transacting across national borders

268
where alien legal structures may raise the risk of opportunism and thus transactional
costs. The make-or-buy decision facing the firm is solved by a range of solutions, from
hierarchical, wholly-owned FDI to market-sourced licensing and franchise agreements
(Dunning, 2000). The investing company would particularly need to protect its
knowledge base, suggesting a tendency towards FDI (Erdener and Shapiro, 2005).
Bailey et al (1994), suggest that this may be part of monopoly capitalism, emphasising a
strategic defence against oligopolistic rivals.

6.4.6 Strategic FDI

The eclectic approach provides an overview of a company’s FDI activity within the
context of the industry, but firms will also act in their own strategic interests. For
example, although Hymer had rejected the portfolio approach to foreign extension as
inadequate. Miller and Reuer (1998) found that diversification of exchange rate risk was
a driving factor behind FDI since it provided significant hedging advantages over
exporting. Cowling and Tomlinson (2005) mention the ability of multinationals to
exercise strategic control over production, employment, investment and advertising as a
route to higher profits. For example, VW consolidated its two underperforming North
American plants on its operations in Mexico so it could operate that one plant more
efficiently (UNCTAD, 2002). The attraction of low labour cost may also be weaker than
previously thought. Erdener and Shapiro (2005) found that low labour rates are not
crucial to the FDI decision and instead the motivation is towards the establishment of
strategic assets, locating plants in order to more effectively service the local market
rather than exploit cost differentials.

To be fair to Hymer (1960), the less well known FDI motivator in his theory was the
desire to eliminate the competition. Buckley (1990) suggests that MNEs can use their
intemational coverage to raise entry barriers to the industry, cross-subsidising those
markets that are most vulnerable to new competition. It is entirely reasonable that a
company should have control over its functions, or else it could hardly lay claim to
being a company at all. Yet if the purpose is for higher profits then it might be valid to
suggest that strategic control is used to change the emphasis of the economic decisions
rather than to take their place. This being the case then strategic decision

269
making only has value to the degree that it brings benefits that exceed those lost
through implementing a Pareto inefficient plant.

Should another firm find greater cost advantages within the OLI theory, perhaps by
retaining a low labour cost location, then it could find itself at an advantage to the firm
that had elected to invest in a higher labour cost location for a strategic purpose. In
other words, if the strategic risk is miscalculated then the strategic decision will be
rendered a disadvantage. Buckley (1990) pointed out that in a static approach to FDI,
for which the OLI theory has been criticised, organisational structures are likely to
converge to an equilibrium, so it can be further argued that FDI based on strategic
decision making is likely to have a more variable element to it. Buckley argues that
intemalisation brings long-term advantages because the benefits are relative to the
market, but competitive advantage is more transitory since it is relative to other firms.
Strategic decision making might therefore be concemed with exploiting temporary
competitive advantage. Since risk aversion depends on the decision makers within the
firm it can be suggested that strategic decision making is a characteristic of individual
firms and not a part of the underlying stmcture of the industry.

Strategy may also take precedence over labour costs as a technique for dealing with
labour organisation. Cowling and Sugden (1998, p.76) discuss the strategic “divide and
mle” concept whereby a company will eschew the cost advantages of maintaining a
single plant and instead extend into foreign plants in order to prevent labour from
unifying itself in opposition to the company. Whether this actually happens depends on
the risk of labour being imified in opposition to the company and the degree to which
the company is averse to this happening. There may even be an element of Williamson’s
concept of “atmosphere”. Yet if this decision is a balance of strategic advantage and
economic cost then it can only be made with knowledge of the cost considerations
contained within the OLI theory. It is not an altemative decision making process since
any strategic decision implies a prior evaluation of the economic benefits of the
altemative. Indeed, Table 6.1 shows that even if there were a strategic reason for the
firms concemed extending into the UK as a production location, at the same time those
sites had an economic foundation.

270
If a firm does engage in FDI for strategic reasons then it is entirely reasonable to expect
a strategic response from rivals, a possibility that the OLI model of FDI would not
account for since it is not concemed with strategy. The response may result in a
defensive merger between local rivals or an aggressive extension into the FDI firm’s
home market (Jones and Wren, 2006). This seems to argue that firms may make a
strategic decision to invest abroad based on reasons that are in defiance of the
advantages set out by the OLI theory. They may even engage in FDI to match the
strategies of rival companies, Kuemmerle (1999) observing industry trends in the
progressive intemationalisation of R&D. This opens up a multitude of possible
strategies, such as strategic alliances that access economies of scale, partner specific
expertise or allow a low risk entry into a new market. Buckley and Casson (1998b)
added cultural factors, tmst and psychic distance to the more quantifiable factors of
location and intemalisation seen within the OLI theory. Yet even with the additional
sophistication of their proposed model, the authors conceded that it assigned a passive
role to national governments which may have their own, strategic, reasons for
encouraging FDI. On this level alone FDI can become highly complex, with
governments setting defined limits to permitted FDI while the investing companies may
perceive heightened risk in some countries over others.

Obtaining information is therefore cmcial to reducing risk, but it is not unique to the
strategic theories of FDI. Johanson and Vahlne (1990) criticise the eclectic model for
assuming that managers have access to perfect information. It is the impossibility of
achieving this that leads firms to be biased towards ownership advantages, about which
it has accumulated knowledge from its own operations. Strategic FDI may therefore be
a method for dealing with information deficiency. In order to preserve the advantages a
firm has in an alien environment it can enter in stages, not in the systematic approach
prescribed by stage models, but to progressively leam about the OLI factors.

Jones and Wren (2006) argue that the area of greatest strategic importance is that of
knowledge, with particular reference to R&D. Borensztein et al. (1998) found that FDI
was effective in the transfer of technology and led to greater economic growth,
demonstrating why national governments tend to be enthusiastic supporters of FDI. A
caveat to this, though, is that the benefits were only realised if the host country had

271
sufficient technological resources of its own to integrate with the incoming advances.
Where this is not the case the FDI is left isolated from its immediate industrial
environment and the nation obtains fewer benefits.

This suggests a crucial role for intemational alliances and Dunning (1995) concedes that
the OLI paradigm needs adjustment in order to explain how joint ventures can open an
avenue to additional knowledge by forming a mechanism for extending the firm’s
ownership advantages beyond its formal boundaries. Alliances permit access to new
information but leaming is a dynamic process, interactive with the environment, and
alliances will only stabilise if they are stmctured to account for the destabilising
influence of new knowledge. Since knowledge is contained within human assets this
represents the keystone of an alliance. Kuemmerle (1999) found that FDI in R&D sites
was on the rise, but that the purpose was to extend existing R&D from the home-base. If
the foreign R&D operation was intended to augment existing knowledge then it tended
to be tied to institutions such as universities, whereas if it was intended to exploit
existing knowledge then it tended to be tied to a production facility.

Beamish and Makino (1998) expanded the typology of JV stmctures to reveal the
complexity of intemational relationships beyond the customary view of a JV between a
foreign firm and a domestic incumbent. The study found that these “traditional” joint
ventures had the potential for superior performance but had a high failure rate. For
Dussauge et al. (2004) this style of JV, dependent upon exchanging knowledge, might be
classified as a link alliance and the increased possibility of opportunism would impose a
short lifespan on the agreement. A stabilising factor is suggested by HO and OLI theories
when differentiated factor endowments provide the circumstances for specialisation
within the full-fimction stmcture. Promising though this approach is, it implies a radical
restmcturing, even deconstmction, that the strategic theories of FDI may not anticipate.
Much of the reason for this is that firms are orientated to using strategies to exploit their
existing capabilities, not to restmcture around new ones. For example, this research
concluded from survivor analysis that as firms expanded their product range and plant
output the risk from fluctuating demand was diversified. For such firms additional
overseas plants represented gains to the firm as a whole, rather than gains specific to the
new plant. Conversely, smaller firms may use FDI to restmcture as the antidote to a
weakness in

272
their position in their home market, perhaps because they are operating below MBS?
overall or MBPS in particular processes.

To put FDI into an empirical context, the next sub-section of this thesis will investigate
the progress Ford made in its early years as it learnt to tackle the complexity of
evolving an intemational stmcture with different govemance stmctures that reflected the
opportunities.

6.4.7 Ford: a failure of global structure

As Burton and Obel (1980) stated, the U-Form of govemance stmcture is suited to
companies that are relatively small, a condition most companies are in as they first
become established. Ford retained this stmcture as it grew from its Mid-West base, the
production expansion that extended across the home US market, and then Burope from
1911 (Georgano, 2000), involved assembly operations building up components shipped
in from the main Detroit factories. As Tolliday (1998, p.59) points out, the Highland
Park and subsequent River Rouge plants were designed as predominantly component
manufacturing plants supplying dispersed branch assembly operations. In 1921, of the
total built-up vehicle output only 9.5 per cent originated from Highland Park, the rest
coming out of the various assembly facilities. Ford opened thirty six branch plants
between 1912 and 1925 in order to assemble cars close to their local markets. Two of
these, Chicago and Minneapolis, could assemble up to 200,000 a year, but many were at
the 40,000 level and utilised the earlier stationary assembly methods.

As Ford developed its intemational expansion it would have been an opportunity to

implement an M-Form stmcture so that local managers could be at liberty to procure
components locally and adapt products to the local market. From 1912, local production
of certain components was permitted for the UK plant at Trafford Park but at the same
time Ford rigidly imposed its “American system” of work practices, such as fixed day
rates of pay, and this brought the company into conflict with British work practices at
that time (Lewchuck, 1987). British managers attempted to ameliorate the American
influence by adapting work practices to suit local conditions, and even after Ford in
Detroit culled the local UK management in 1924 and installed a new team, it

273
too found it had to change American methods to suit local conditions (Tolliday, 1998).
Yet Ford of the US persisted in its authoritarian rule from the centre. European sales
were collapsing and Ford fell from first to fourth in British rankings, behind Austin,
Morris and Singer, mainly due to Ford’s failure to give the Model T an engine that suited
the local tax regime, leading to the product being priced out of the market (Wood, 1988).
Most extraordinary of all, after the Great War, the British arm was even compelled to
produce only left hand drive vehicles in common with the US (Tolliday, 1998).

It would appear that a weakness of Ford’s style of U-Form govemance stmcture at this
time was that it was inadequate at accommodating the dynamism of foreign markets.
Vemon’s product cycle would have forecast that the Model T would find a receptive
market in the UK since it had set the standard for the industry elsewhere. In the event,
neither the product nor the production system, both now icons of automotive history,
were entirely applicable to an alien environment. As suggested earlier, the product cycle
theory needs to allow for the standardisation of the concept after which continuous
adaptation and development takes place. In Ford’s case the company missed its chance to
adapt the Model T, but it did address this in 1932 when the company released the Model
Y for the British market, a new product originally designed in Detroit. Its popularity
induced the British arm to surreptitiously conduct its own limited R&D to adapt the
product. This was comparable to the emerging M-Form stmcture then taking root at GM
but at Ford it was instead met with a definitive restatement of control from the central
management in the US. As leading Ford manager Charles Sorenson testily telegraphed to
the chief engineer at the Ford Dagenham plant, A R Smith:

“SMITH. ARE YOU AWARE THAT WE ARE CONTROLLING DESIGN OVER

HERE. SORENSON” (Quoted in Tolliday, 1998, p.67).

With the U-Form stmcture held largely in place it was not until 1937 that Dagenham was
allowed, despite continuing American resistance, to take on greater local powers and
initiate a replacement for the Model Y of British origin. This became the Prefect, later
joined by the Anglia variant, and Ford moved back up to third place in the UK rankings
for car production (Wood, 1988, p.66). Up until then, the company had been

274
conducting operations in the UK, quite overtly, as an integral part of the U-Form
structure in place within the United States and with little reference to UK national
differences. For example, local information such as conditions of tax regimes, labour
practices and market demand had been largely disregarded. The more conspicuous
information had been transparent and the company recognised the need to minimise
transport costs exactly as it had in the home market, and also to evade punitive local
tariffs by assembling locally from supplied complete knock-dovm (CKD) kits. The
unified, authoritarian approach of the U-Form had not been sufficiently developed to
cope with the local complexities, only the more conspicuous matters of entering an
alien environment, and it was only when Ford stretched to a more multi-divisional M-
Form that it was able to exploit opportunities specific to the UK market.

The move towards the M-Form allowed the British arm of the company to develop
within the local environment while benefiting from central strategic support. In
particular this might have been the economies of scope that can become available under
the M-Form. Considering the state of development within the industry at the time this
change in govemance stmcture brought Ford into line with GM. However, this research
programme has found that in the current industry the U-Form appears to have reasserted
itself amongst manufacturers such as Honda and Toyota, albeit with sufficient
adaptation to local conditions. This suggests that although the move to the M-Form by
Ford brought benefits based on divisional empowerment the greater benefits would have
come fi*om adapting the U-Form to its intemational extensions. For example, the
Model T began to fail on the UK market because it lacked the correct engine and
steering configuration, not because it was inherently unsuitable. The Model Y seemed to
correct the problem, being an American design adapted to local conditions, but
thereafter Ford abandoned centralised control and gave the British division both
operational and strategic control based on its own R&D capability. From this point the
British operation seemed to take on a domestic U-Form of govemance appropriate to its
strategy of being a largely insular UK based automobile manufacturer.

In the process of this evolution, no attempt was made to account for UK factor
differences in the wider sense proposed by Schott (2003). It might be conceded that the
economies of the US and the UK were closely aligned at the time so the

275
differences would have been slight, nevertheless, there was apparently no indication
that the Ford Motor Company was seeking to restructure around the opportunities raised
in the British industrial environment. This is demonstrated by the lack of influence by
the UK branch on the US operations as part of a dynamic interchange between the two
branches. This does not mean that Ford was not aware of the principles behind the OLI
theory, indeed the three factors seem to have been very much part of the FDI plan. The
weakness was that the corporate govemance stmcture was ineffective in fully evaluating
the value of each factor, particularly that of location. Instead, the company seemed to
act more in accordance with the strategic theories of FDI in taking a manufacturing
position within the UK market which was not strong enough to offset the weakness in
the underlying economic considerations. As was suggested in the preceding sub-section,
strategic FDI cannot be implemented in ignorance of economic factors.

If strategic approaches to FDI have a role it is when the industry is standardised and the
market globalised, at which point the economic factors are embedded. For example, a
company might establish a plant in a location to match a rival’s FDI strategy. Since the
production technology and degree of local R&D will by now be part of the industry’s
status quo the strategy will be concemed with competitive advantage (Buckley, 1990)
and the benefits will be incremental. The M-Form of govemance stmcture might be
sufficient for FDI of this level, but as Bartlett and Ghoshal (1993) suggest, it would be
insufficient for accommodating the full range of opportunities offered by intemational
environments, such as in the dynamic form espoused by Schott (2003). This research
concluded in the preceding chapter of this thesis that the centralised style of the U-Form
could be adapted by large firms to the demands of operating in overseas environments.
An altemative approach, though, applicable to sub-optimal firms below MESp, would
be to restmcture around the opportunities suggested by Schott. The next subsection of
this thesis will show how a firm can be restmctured in recognition of these intemational
opportimities.

276
6.5 Reassessing the full-function structure

The full-function model contains the four main functions for a prototypical car firm and
since the underlying cost factors are common throughout the industry the full-function
structure has become a feature of the industry (see Chapter 5). On the basis that
integration means internalising the responsibility for economies of scale then in
principle the options for structuring are polarised: a function is either integrated or it is
not. Alliances that take place at the sub-functional level, whether of the scale or link
type, introduce additional transactional costs. Scale alliances seem to be more stable but
their benefits are static and the alliance lasts as long as the specific plant. Link alliances
are apparently more volatile due to uncertainty in knowledge development leading to ex
post opportunism. However, in a VJV where the division of the link alliance is at a
functional level, then the benefits of knowledge development accrue to the partner that
has sole responsibility for that function. There is less possibility of opportunism and so
the partnership remains bilateral. This in turn indicates that a VJV structured
functionally is likely to be stable in the long-term.

Although the theory of the VJV, and the IVJV version, has been laid out it is by the
nature of the alliance that empirical examples are difficult to source. For the VJV to be
formed both partners would have to be in one of two conditions:

1. Have complementary functions that are deficient in scale which they are
prepared to abandon in the interests of restructuring around a VJV
2. Deficient in the full function structure to a complementary degree (i.e. the two
organisational structures are able to dovetail in a VJV)

In the first condition a firm would need to have the courage of its convictions, and the
trust in its partner, to restructure solely for the benefit of the VJV. In the second
condition both potential partners would have to be operating without functions that
competitors deemed crucial to long-term sustainability. They would therefore be in an
unstable condition and dependent upon some kind of support not accessible to rivals
(e.g. government funding, consumption of intemal funds for survival). The opportunity
to form a VJV in the second condition would therefore only last as long as the support
remained in place. The next sub-sections of this thesis will examine

277
 examples of vertical relationships in the automobile industry. The first, between GM
and A O Smith, is used to demonstrate the stability of a vertical relationship. The
second, between Rootes Group and Iran Khodro, shows the possibilities in an IVJV.

6.5.1 Vertical joint venture at GM

General Motors was organised with a novel form of corporate govemance stmcture, the
so-called M-Form (Williams et al, 1994), a bureaucratic stmcture for dealing with the
management of intemalised fimctions. Under William Durant, the company had been
constmcted from a collection of unconnected assemblers and suppliers for the same
reason that Henry Ford had brought functions in-house, but the economies of
concentration in GM’s home base of Flint meant the company did not have to
intemalise processes to the same extent as Ford (Chandler, 1990).

This was illustrated by the previously discussed occasion (see section 5.3.5) of GM
purchasing Fisher Body in 1926 (Coase, 2000; Klein, 2000) which may be contrasted
with the following long-term contractual relationship with the functional supplier, A O
Smith, a chassis supplier to GM in 1932. The two companies had been in a close, almost
symbiotic, relationship for years and R H Coase (2000) found that even the production
method was tailored to GM’s needs, exhibiting the characteristics of de facto vertical
integration. In a co-dependent, long-term relationship the interests of the companies are
served best through the maintenance of the stability of the relationship rather than
jostling for a position of dominance. Williamson argues that when asset specificity is
high:

“...buyer and seller will make special efforts to design an exchange that has good
continuity properties” (1981, p.555).

The example of A O Smith shows that high specificity can enforce a bilateral
relationship (Williamson, 1998). Furthermore, because each company occupied a
discrete function there was a reduced possibility of opportunism since any ex post
advantages would accrue to the partner responsible for the function. Clearly this was not
vertical integration as long as the two companies were under separate ownership,
though the relationship was indeed a vertical one since A O Smith’s responsibilities

278
occupied an upstream position in the production process. Etemad et al (2001) found
that when the supplier specialises enough a co-dependency arises and the authors’
proposed “interdependence paradigm” states that this relationship can be stable over the
long-term.

The elegance of the solution to the intemalisation question by A O Smith and GM is in

the manner in which operational functions could be intemalised to the relationship. As
independent companies they were not extending beyond their established competencies.
They were also at liberty to take advantage of innovations when they occurred and
compete within the industry for additional business. Later, when the all-steel unitary
body was changing the industry, GM was free from the costs of restmcturing a chassis
production unit, this being an extemality to GM that A O Smith had to deal with. At the
same time, asset specificity was a mutual problem, A O Smith specialising in chassis
production and GM having no such capability. This was the basis of the
interdependency that enforced the quasi-intemalisation of chassis production. From this
flowed the benefits of intemalisation, such as a reduction in principal-agent costs of
information searches, uncertainty, and contractual imperfections. With chassis output
tailored to GM’s assembly output the production processes of the two manufacturers
would have converged, thus the economies of scale for chassis production were
accessible to GM.

The relationship between the two companies predates the Budd all-steel body
paradigm, the chassis being manufactured using distinct technology, so chassis
production was a separate function from body production. A O Smith was therefore a
functionary supplier, i.e. it had responsibility for the function of chassis production.
Given the diverse range of activities that were held under GM’s M-Form umbrella of
govemance it is interesting to note that A O Smith was not intemalised. If economies of
scope existed in the M-Form then it would be expected that they would be effectively
exploited by such intemalisation. The fact that this did not take place demonstrates the
value in the stmcture of the relationship in the form that it developed. This put A O
Smith in a vertical position within the production process and in a close working
position with GM, so although the relationship was not termed a joint venture it is
reasonable to characterise it as such. Figure 6.3 illustrates the relationship between A O
Smith and GM as a VJV. This has been adapted from the

279
full-function model to show a pre-Budd era production system. It demonstrates A O
Smith’s critical position within the production process, the supplier’s hold-up power
being balanced by GM’s status as sole buyer.

Figure 6.3 Vertical joint venture between A O Smith and GM

r GM A O Smith
i
Research and
Design Chassis
Production

Powertrain

Fmal
Assembly

7
s/
Market

A VJV depends on the supplying partner having responsibility for the complete
production process that defines the specific function. With the two being thus
interdependent, the relationship is self-regulating and stable in the long-term, A O Smith
continuing as a chassis supplier to GM until 1990, finally moving out of the industry
altogether in 1997 (Assembly, 2002). However, this strategic relationship does not seem
to fit any explicit form of govemance stmcture. Although this thesis has cast doubt on
the efficacy of the M-Form of govemance, particularly due to the difficulty in separating
strategic management from operational management, in the case of GM and A O Smith
this division appears to be more distinct. While A O Smith would have retained
operational control, it is quite feasible to view strategic control as lying with GM since it
determined vehicle design and therefore chassis design as well as product demand. At
the same time, the cmcial feature of the M-Form, which is the transfer of objective data
from operational units to headquarters for strategic analysis, would have been missing.
From this one example a definitive judgement on the govemance stmcture for a VJV
cannot be made, but this may become clearer by investigating further aspects of the VJV
relationship. Since it would appear that a VJV can be stable if the assets are divisible, it
would be

280
instructive to investigate how else an organisational structure, usually expected to be
vertically integrated, might be divided. The following sub-sections will do this from the
perspective of human assets and intemational opportunities.

6.5.2 Human assets and the full-function model

Chapter 4 of this thesis compared the output of the three manufacturing functions (BIW,
powertrain and final assembly) with the R&D function, showing that diseconomies of
scale set in early with R&D. Research has suggested that much of this is due to the
rising staffing levels of larger departments. Clark et al. (1987, p.766) stated that
Japanese companies were able to counter this and find a relative advantage through
project management:

“In the best of the Japanese projects, a heavyweight project manager leads a
multifunctional team, in which problem-solving cycles are overlapped and closely
linked through intensive dialogue.”

Although the presence of a “heavyweight” project manager implies a large team, it is

noticeable that the multifunction stmcture indicates that the team is made up of smaller
groups. As Cooper (1964) pointed out, a multi-group large team cannot entirely replicate
the efficiency of small teams and he noted that larger R&D departments have
disproportionately higher spending levels (see section 4.3.5). This results in decreasing
retums to R&D investment and reveals the existence of diseconomies of scale. Within a
full-function company the economies of scale captured elsewhere must compensate for
the diseconomies in the design function or else the model would not be sustainable.
Smaller companies, those with sub-scale production output, suffer the reverse effects.
They might have small design functions that exploit economies of scale while the
manufacturing functions are hampered by sub-optimal production.

The difference between R&D and the manufacturing functions is the relative importance
of human assets. Klein (2000) has shown that vertical integration is essentially
concemed with securing human assets within a hierarchical company stmcture. Since
R&D is concemed with the flow of new products to differentiate the

281
firm R&D must be crucial to the long-term prospects of the firm. It is therefore vital for
the firm to control the R&D assets by intemalisation of this function and the human
assets that comprise it. The volatility of link alliances shows the uncertainty involved in
attempting to share this function with another firm.

The three manufacturing functions also make use of human assets but to a much lesser
degree. As Klein (2000) demonstrated, intemalisation of physical assets is concemed
with gaining hierarchical control over the human assets that operate those physical
assets. Thus, the physical assets of production are integrated for reasons of managerial
control; the human assets of R&D are integrated for reasons of access to knowledge. In
combination they form one discrete production system: the full-function model. Factor
endowment differentiation offered by globalisation indicates that advantage might be
found in dividing the integrated stmcture according to the distinction between human
and physical assets. The following sub-section provides an example of an IVJV that
demonstrates some of these attributes, and suggests a possible form of such a VJV that
could have been beneficial to MG Rover.

6.5.3 International vertical joint venture: Rootes Group

In the 1960s, Iran was seeking to industrialise and had identified the car industry as a
potential catalyst. The country was endowed with a large labour pool as well as natural
resources, most particularly oil, which provided it with the funds to import capital.
However, it had little existing industry and so was weak in engineering skills, the
human assets of technology. It was therefore necessary for the company to import
foreign vehicle designs for local production.

In 1967, the British company Rootes Group licensed the design, and initially supplied
CKD kits, of the Hillman Hunter to Iran Khodro. In Iran, this model was known as the
Paykan, Persian for Arrow, which had been the Rootes Group project name for the car.
Rootes Group changed ownership over the next few years, in 1967 being acquired by
Chrysler and then subsequently by Peugeot in 1979 when Hunter production in the UK
was brought to a close. However, CKD kits were supplied until 1985 and then
production continued with local component supplies. By 1993, the Iranian operation

282
became 98 per cent self-sufficient with output capacity of 120,000 units per year (Iran
Khodro, 2004).

Up to this point, the relationship between Rootes Group and Iran Khodro had been
asymmetric in character since there was a difference in the significance to the two
parties. Under the terminology of the theory put forward by Dussauge et al (2004) the
relationship had aspects of both link and scale alliances. Access to the Iranian market
provided useful additional output for Rootes Group, while for Iran Khodro it was
principally a link alliance, providing access to vehicle design. As the theory of
Dussauge et al predicts, the scale aspect of the partnership was relatively long lasting
with body panels being pressed in the UK even beyond the end of the Hunter’s
production life. The link side of the relationship was very short lived, constituting
simply the supply of the original vehicle design.

From Iran Khodro’s standpoint the relationship could be characterised as a VJV, and
this is illustrated by Figure 6.4 below. This shows Rootes Group supplying the R&D in
the form of a finished product design, while Iran Khodro was engaged in the majority of
the production. It is not, though, how the model would be depicted from the Rootes
Group perspective since the output for Iran was supplementary to all other functions and
so would be described as a scale alliance. The mismatch in perspectives is partly
explained by the HO model of relative factor endowments between the two countries
and Schott’s adaptation of it. Without vehicle designs Iran Khodro would have found it
very difficult to conceive its ovm products, so accessing the engineering human assets at
Rootes Group in the heart of the British car industry also suggested some locational
advantages. For Rootes Group, though, the situation is complicated by the fact that
production in Iran was simply an addition to the UK plants which were, at the time,
fairly competitive within the domestic UK industry. Furthermore, as Schott (2003)
suggests, factor endowment differentiation is complex and political factors had a great
bearing on the partnership for the British operation. Government support in Iran of the
nascent domestic industry would have made the abandonment of production in the UK
by Rootes Group highly risky. Similarly, the eclectic paradigm shows that the locational
advantages of production in Iran would have been outweighed by the risks of
transferring production facilities to an uncertain political environment.

283
 Figure 6.4 VJV between Rootes Group and Iran Khodro: Iranian standpoint,
1993
Rootes
Group
Research and
Design

r ■ Iran Khodro

Powertrain BIW

Final
Assembly

Iran
Market

The role of the Rootes Group in the partnership was to solve the specific problem of
Iran Khodro’s need for a vehicle design. Once the design had been supplied, the R&D
side of the Rootes Group had no more contributions to make, at least until the problem
of a subsequent vehicle design arose. In the meantime, the Paykan was developed
further by Iran Khodro using later Peugeot technology, the ageing model being refreshed
by building the old body style on a Peugeot 504 platform. The partnership was deepened
when Iran Khodro designed its own vehicle, the Samand, on a Peugeot 405 platform
(Iran Khodro, 2005). Although the partnership with Rootes Group as a long-term scale
alliance and a short-term link alliance had been in accordance with the theory of
Dussauge et a l , given even more time in partnership with Peugeot the link aspect
reasserted itself. Since Iran Khodro at least had the ability to develop the Paykan further,
the need to replace it did not occur until sometime in the 1990s. It was then able to do
this in a new link alliance with Peugeot, one unrelated to its link alliance with the
Rootes Group. This indicates that in a link-scale combination alliance, the scale aspect
secures the partnership in the long-term

284
and so allows the link aspect to continue from one project to the next, which Dussauge
et al. show would not be the case for link only alliances.

The scale alliance aspect continued because the problem of scale is endemic to the
manufacturing system, while link alliances tend to provide solutions to specific
problems. A scale alliance is therefore predicated on the nature of the physical assets,
while link alliances are predicated on the knowledge contained within the human assets.
In combination, link-scale alliances can exploit the divisibility of a firm into fimctions
that favour either physical or human assets. Iran Khodro could readily act on this since
it was lacking the R&D capability, but for the Rootes Group it would have been far
riskier to abandon a production capability which was, at the time, still commercially
viable. For this reason, the relationship can only be depicted as a VJV from the Iran
Khodro point of view, for the Rootes Group it was a horizontal joint venture in
production only. It cannot be said to have been a true VJV between the two partners
because the Rootes Group never relinquished its production capability in favour of Iran
Khodro.

However, as it was, it did provide some additional perspective on the divisibility of

hitherto integrated company structures. The spatial division of human and physical
assets, according to factor endowment differentials, can also be viewed as a type of
link-scale alliance when it brings together R&D and production capabilities which are
geographically separate. It might be considered a hypothetical question whether the M-
Form of structure could have coped with internalising Iran Khodro into Rootes Group,
yet the reality is that the M-Form cannot be applied when national interests regulate
against it. It is not simply that Rootes Group were prohibited by the Iranian government
the opportunity to consider the question, the prohibition was part of that intemalisation
calculation since it raised costs to an impossible level. It has been discussed previously
that the M-Form is a bureaucratic device for accommodating diversity as long as it is
limited in scope, failing to account for the full range of opportunities that exist in
intemational locations.

This then raises the question of the kind of govemance stmcture that would be suitable
for an IVJV. This thesis has argued that the U-Form has the most suitable control
mechanisms for an intemational enterprise, but it is not generally possible to

285
have a centralised decision making process within a JV, Indeed, an IVJV may require a
particular governance structure of its own to manage the strategic structure of the
combined operation. For further investigation, this thesis proposes that for the main
case study company, MG Rover, a link-scale combination alliance could have been
structured along functional lines in an IVJV, each partner maintaining control of
discrete functions such that any ex post advantages would accrue to that partner. The
empirical evidence for the potential of such an IVJV will be discussed in the next
chapter of this thesis.

6.6 Conclusion

This chapter, building on the analysis conducted in the previous chapter, has described
the implications for an automobile company of not achieving the structure and size put
forward by the automobile industry paradigm. The chapter has examined various ways
in which a company, uncompetitive within the automobile industry paradigm, can
approximate to the paradigm. Intemal approximations include extending production
over time or intensifying production through flexible production techniques. However,
neither strategy deals directly with the cost disadvantage of sub-optimal output,
although they may ameliorate the effects.

Extemal approximations to the paradigm involve making alliances with other firms.
Although intermediate forms of governance have been suggested as sufficient
compromises in the make-or-buy dichotomy they do not represent novel forms of
govemance. This is because they are not in accordance with the possibility of
divisibility in the full function model and thus are simply adjuncts to it.

This chapter has sought to demonstrate that integrated companies are divisible in two
dimensions. The first is the distinction between human and physical assets, as illustrated
by the link and scale alliance structures. Scale alliances tend to be secured by physical
asset specificity which has long-term applicability, whereas link alliances are generally
seen to be subject to ex post opportunism and so only last for relatively short-term
periods. If the link alliance is made between complete, discrete functions, however, then
the advantages of new developments accrue to the partner responsible for the function
and so opportunism does not arise. This indicates that VJVs involving

286
complete and complementary functions may be the basis of long-term sustainable
partnerships. The second dimension is the distinction created by factor endowment
differentiation between different international regions. This indicates that factor
endowment differentiation may be enduring and provide an international division to an
alliance. In combination, the IVJV has the potential to competitively replicate the
structure and size prescribed by the automobile industry paradigm.

Against the background of the automobile industry paradigm proposed by this thesis,
the next chapter will examine the activities and subsequent collapse of the main case
study company, MG Rover. In particular, from the perspective of approximating to the
automobile industry paradigm, this will focus on the company’s attempts to secure its
future, ultimately unsuccessful, by way of international joint ventures.

287
 Chapter 7;
A Sustainable Model for MG Rover

This thesis has set out the conditions for an automobile industry paradigm. This defines
the theoretical structure for a prototypical automobile manufacturer in terms of scale and
vertical integration. These two structuring forces have been assessed in terms of
economies of scale and transaction costs, respectively. The resulting model has extended
the Budd paradigm to show how an optimal firm would integrate the four main
processes (R&D, BIW, powertrain and final assembly) into a full-fimction structure,
each function exploiting the available economies of scale.

Survivor analysis has found that although the industry leading automobile
manufacturers benefit from strategic advantages, the MES for the industry is likely to be
somewhat lower. A company that is close to the MES for the industry represents the
prototypical form of company, and its output is signified by MES? at around 600,000
units a year. The onus is then on those companies that are sub-optimal to MESp to
formulate a strategic response that enables them to approximate to the advantages of the
paradigm. As was examined in Chapter 6, this can be done intemally, using the
company’s own resources, or externally, by accessing those of another company. Internal
approximations tend to focus on the utilisation of the physical assets by either stretching
the production life of the product or exploiting economies of scope and using flexible
manufacturing to intensify the production system. However, neither approach excuses
the company from the cost implications related to minimum efficient plant size (MEPS)
or overall minimum efficient scale (MESp).

Extemal approximations to the paradigm do not include transacting with companies in

the market since this is diametrically opposed to the full-fimction feature of the
automobile paradigm, not an approximation of it. In order to replicate the full-fimction
stmcture by using an extemal mechanism the company must partner itself with another
company in a manner that offers an altemative to the make-or-buy dichotomy. Alliances
have been put forward as a quasi-intemalised organisational

288
structure and this can involve sharing of facilities in a horizontal joint venture. However,
it has been suggested in the previous chapter that joint ventures are most stable when
they coincide with the divisible company activities. Since companies are most readily
divisible along functionary demarcations this thesis has put forward the concept of the
vertical joint venture (VJV). This replicates the full-function model so that partner
companies take responsibility for one or more of the complementary functions.

Intemational factor differentiation also encourages divisibility by offering cost

advantages when structuring around human and physical assets. Allied to the VJV
structure, the resulting intemational vertical joint venture (IVJV) allows companies to
continue to specialise in the human and physical factor endowments of their existing
functions to their mutual benefit. If the functions are also competitive in terms of scale
then there will be sufficient approximation to the automobile industry paradigm for the
combined companies to be competitive.

Thus far this thesis has presented arguments from scholarly sources conceming the
stmcture and govemance of automobile companies. At this juncture it is therefore
necessary to include primary fieldwork data to directly reveal the views of the
interviewees. This will bring a new perspective to the challenges of automobile
companies attempting to maximise the opportunities that face them. It will also clarify
some of the mechanisms of govemance stmctures, in particular the validity of the M-
Form of govemance for managing automobile firms.

This chapter has been divided into two sections and Part I will deal with the contextual
companies. This will include some information that is also related to MG Rover’s
predecessors (e.g. BL, Rover Group) since these had organisational stmctures quite
distinct from that of MG Rover and the data might more properly belong to the
contextual industry. It should also be noted that since the UK industry used to be
dominated by the predecessors of MG Rover, many of the examples given by the
contextual industry sources included references to those firms. Part II will then look at
the historical antecedents to MG Rover, its five years of operation and then suggest a
stmcture that might have prevented its ultimate demise. Reasons for this proposed

289
option not proving to be feasible will be discussed and compared to the final,
posthumous structure of the company’s assets.

290
 Part I: The Contextual Automobile Industry

If a ftill-function company lacks the scale of output necessary to be competitive with its
peers within the automobile industry paradigm, then it can obtain benefits by the use of
its existing resources. There are two options for making intemal approximations to the
paradigm: one is to extend production mns, the other is to intensify the use of production
facilities. This section will examine how the companies that make up the contextual
industry applied the two approaches.

7.1 Internal approximations to the automobile industry paradigm

Although MESp represents the minimum sustainable size for an automobile company
there are incremental advantages to higher output ranges, mainly due to the
diversification of risk. The techniques for intensifying or extending the use of production
capacity were therefore of utility to those companies above MESp to compound their
advantage. It was also likely that automobile companies with output below MESp would
try to use the techniques to approximate to the requirements of scale production. For a
particular process this meant attempting to approximate to MEPS in the relevant
function. Chapter 6 has already cast doubt on the feasibility of this approach by
manufacturers (see section 6.2 above), nevertheless the fieldwork sought elite views to
provide a more empirical context.

Conceming product life cycle extension, Mr Anthony Millington of ACEA gave as an

example the Rover Mini, popular in Japan despite a thirty year production run, but
added that Rover Group had regularly refreshed the product with special editions. This
was sufficient to keep a highly differentiated model in production but Professor Daniel
Jones cautioned that, in general, there is little opportunity to extend production mns
because of the capricious influence of fashion:

“Honda sells a hot product, then it goes and they have to find another hot product.”

Mr David Blume of Jaguar was more specific:

“If Honda makes an interesting move, Toyota is ready to cover it.”

291
Professor Jones felt that within this strategy it was only by modular design that vehicles
could be updated piecemeal, allowing certain parts of the vehicle a longer life-cycle.
There was no evidence that this was being enacted upon and, at Honda, Mr Takeshi
Sumita added that the company policy now was to change a model in all markets and
production sites simultaneously. Honda would not selectively extend production runs
according to the market: for the 2002 Accord, only China was slightly behind by three to
four months. This clearly revealed a strong centralised govemance stmcture coordinating
global production and component supply and simplifying the investment in new
technology. Mr Sadanori Ito at METI pointed out that environmental concems were
pushing for greater technical innovation, hybrid engines being an example, and by
implication this prevented long production runs. This indicates that extended production
runs are the last resort of a firm and more difficult than they were in the past.

Intensification of production based on economies of scope was more fmitful as a device

for improving output efficiency of products. Many of the interviewees referenced the
Toyota Production System (TPS) as the system for achieving this and Mr Koji Yamada
of Toyota clarified that TPS had been a refinement of the mass production system
developed by Ford. Mr Daigo Umeki of Toyota explained that just-in-time (JIT)
component deliveries are not strictly speaking about having low stocks but minimum
stocks, the actual quantity dependent on what the company believes is sufficient. He
asserted that TPS is not lean production but production in a “dangerous condition”,
meaning that a weakness in the system will be exposed by a stoppage and immediately
identified. This required careful centralised planning to establish the system for a plant,
implying that the opportunity for local operational control was extremely limited.

It is on this basis that TPS is applicable to other production systems with specialist
manufacturers organising production on a similar basis. According to Mr Douglas
Dickson, Bentley operates a version of TPS, particularly thanks to the role of Mr Dave
Hudson, an ex-Toyota production engineer, who designed the Bentley system. Mr
Adrian Hallmark described Bentley’s flexible production system as being efficient at
annual production of 15,000 a year from three shifts and capable of producing three
variants of the Continental model. Although the older Amage model was observed to

292
have its own, much shorter assembly line this too could cope with variants of the same
basic model. Dr Ulrich Eicchom, chief engineer, described how the Bentley production
system is modem in essence, being U-shaped with JIT delivery, but with craft
production of certain components, such as leather seats, taking place in sub-assembly
stations. Again, the system seemed to be designed to be almost self-supporting, with
very little need for operational involvement by divisional managers. Even Morgan’s
craft style of manufacturing did not preclude the use of a TPS style production flow
with kanban component supply controls (Mark Aston, 12 July 2002).

Flexible production was a cmcial method for maintaining output near to production
capacity and products were often designed for flexible production from the outset.
Professor Hiromi Shioji described how US manufacturers are able to benefit from
greater flexibility in the way that they design their light tmcks. Although the separate
chassis is usually thought of as being from the craft era of production, it also permits
greater flexibility in body shapes than the unitary style favoured by Japanese
manufacturers. At Aston Martin, Mr Tim Watson showed an updated version of the
approach, the company’s production strategy depending on the single aluminium VH
vehicle platform which could be adapted to various different models for assembly on
one line. Powertrain production could benefit in a similar way. Jaguar’s David Blume
stating that the Jaguar V6 engine is made in the same factory as that for Lincoln
engines, though with a custom configuration, and this allows the factory to operate at a
far higher scale of output.

Some respondents reported technical limitations to flexible manufacturing. Mr Daigo

Umeki of Toyota said that the main restriction to the number of models that could be
made on one line was storage space for the components. Another restriction is
powertrain configuration because a large machine is needed to lift and install the
powertrain, so to make it available for all types the machine would have to be even
bigger. According to Mr Takeshi Sumita, Honda lines are capable of assembling eight
different model types, but this is not always necessary and the US Accord line was fully
occupied with the one model. There were, though, still advantages to flexible production
even then because Mr Chris Rogers added that flexibility also allows faster new model
introduction.

293
The general consensus amongst the contextual sources was that the intemal techniques
available for approximating to the automobile paradigm mostly amounted to flexible
production methods for maximising the available production capacity. This would then
allow ranges of products to be fabricated in single plants that were constmcted in
accordance with MEPS for that function and diversify the risk of fluctuations in demand
for particular models. Before investigating how firms might make use of extemal
resources to improve their output relative to MESp the research looked at the
opportunities that would be available in the global industry.

7.2 Structure and opportunities of globalisation - institutions

The research interviews raised the subject of globalisation to uncover the competitive
advantages of entering foreign locations. The previous chapter of this thesis has
discussed globalisation as the integration of processes between countries, which would
then largely eliminate intemational differentiation. In the first instance, the interviewees
were asked to comment on the way that globalisation was regulated to find if factor
endowment differentiation could be attributed, in part, to variations in political regimes.
Rather than the supra-national institutions of globalisation, like the WTO, World Bank
and so on, the comments were concemed almost entirely with the next unit of analysis
down, national governments and their agencies. Three main areas emerged from the
interviews:

• Mandate of the institution

• Benefit to industry
• Cost to industry

These issues are all considered in this section and the contextual interviews covered a
wide range of institutions. Table 7.1 lists the institutions mentioned by the respondents
and their mandates to influence the industry along with the resultant costs and benefits.

294
 Table 7.1 Institutions of globalisation - effectiveness

Data Institution M andate B enefits Costs

Source
Mr M WTO Free trade Single market
Harbour US government Domestic Ineffective
METI MITI protection protection
Domestic Guidance Export control
Chinese promotion
Domestic Corruption
government protection
ACEA ACEA European Guidance
representation
TGWU National Domestic Incentives
governments protection
Global industry Distorts global
Unions Job protection Maintain competition

national skill
DTI Industry base
Mediation Powerless
Government guidance
Jaguar Regulations Development
SMMT SMMT UK industry Unify UK costs

representation industry

There was a marked lack of enthusiasm for the institutions of globalisation, the overall
impression being that they are simply part of the environment that companies operate
within. The WTO was seen as a promoter of free trade but the emphasis amongst
interviewees was on the more variegated influences of national governments. This meant
that not only was there little guidance on seeking intemational differences, such as factor
endowments, countries did not seem to promote these differences either. METI and
ACEA looked for a positive government influence in the form of guidance for the
national industries, and the SMMT liked to think that as a trade body it was an institution
that could unify and represent the interests of the UK industry to the domestic
government and intemationally. The TGWU was more defensive, looking to the UK
government as an institution to support domestic industry and maintain the national skill
base. From there on, opinions were overwhelmingly negative: domestic protection,
where it was effective, was seen to distort the industry and encourage corruption. Since
the benefits of factor endowment differentiation are accessed through intemational trade,
the effects of protectionism are seen here to blunt the attractiveness of foreign locations.
For example, the view at Jaguar was that government policies tended to result in higher
development costs to meet the various

295
regulations. Where institutions did provide some advantage it was through a supporting
role as a guide or skill provider, a factor that Dunning (1979) noted in the ability for a
country to improve upon its factor endowments. The TGWU did put forward one way in
which factor endowments are temporarily improved, this being the provision of financial
incentives for firms to invest in specified locations.

The intemalisation of markets within large corporations tended to divert attention away
from intemational differences. Honda showed that large companies have the intemal
resources to be self-supporting and therefore rarely need to interact with any institutions,
global or national. Contrary to critics of globalisation, Honda simply acquiesces rather
than confronting the institutional authorities, though its dependence on the US market
does have a significant impact on its operations. Table 7.2 illustrates this point.

Table 7.2 Institutions of globalisation - effect on Honda

Data Institution M andate Advantages D isadvantages

Source US government Market Costs o f special
MrH
Yoshino conditions models
Dealership Supplementary
isochrones Raises technical dealer network
Mr T lida US government EPA regulations
standards Equity
MrT China FDI control
Sumita government restrictions

Again, it seemed that intemational differences tended to increase costs rather than offer
opportunities to reduce them. The higher costs seemed to be because the automobile
industries have economies of scale that favour volume production of standardised
products, which then need to be adapted at additional expense for individual markets.
They might also incur stmctural costs in being obliged to adjust their infrastmctures.
Honda ex-CEO, Mr Hiroyuki Yoshino, noted the increased costs in the US of adapting
models for the local market and having to establish two dealer networks due to
prescribed catchment areas. Mr Sumita also pointed to the restrictions on investment in
China by the government there, foreign direct investment (FDI) being limited to equal
joint ventures with local companies. However, such market distortions would apply to
all incoming automobile firms. Indeed, Mr lida felt that the technical requirements
introduced by the Environmental Protection Agency (EPA) in

296
the US had the advantage of raising technical standards which had application
elsewhere, Honda being ready to exceed those standards. In sum, Honda did not declare
any particular hardship as a result of the existence of global and national institutions.

7.3 Globalisation and the contextual automobile industry

Although political institutions seemed to be attempting to promote globalisation by

reducing intemational differences, for example by handicapping the advantages of
foreign investing firms, the interviewees were asked whether they had found differences
that could be exploited. The purpose of this was to discover if factor endowment
differentiation was still in evidence, as theorised by Schott (2003), or if globalisation
meant that intemational differences had been generally eradicated. It also investigated
some of the OLI advantages put forward in the eclectic theory (Dunning, 1981). The
following categories of data analysis emerged from the interviews across these two areas
of research:

• Level of globalisation achieved

• Drivers of globalisation
• Inhibitors of globalisation

The interviewees were questioned for their thoughts on how intemational exposure had
changed the stmcture of the automobile industry. Table 7.3 summarises the views of the
interviewees as to the different global levels to which the world’s automobile industry
has been stmctured, and whether the companies had a global perspective or were
domestically orientated.

297
 Table 7.3 Global automobile industry and levels of intemational integration

Data Source Level of Analysis Level of International

Professor H Shioji Honda strategy Integration
Domestic
GM strategy Global
Professor D Jones GM strategy Global
MRI Japanese industry Domestic
METI Japanese industry Domestic
Toyota Technology Global
Toyota company structure Domestic
Supply network Domestic

It can be seen that Professor Shioji divided global strategies into two basic formats,
epitomised by Honda and GM. Honda’s strategy was centred on Japan and, according to
the other respondents, it had this in common with Toyota and the Japanese automotive
industry in general. This was, therefore, in agreement with this thesis that such
companies operated a U-Form of govemance stmcture. Conversely, Professors Shioji
and Jones considered that GM showed a global perspective, perhaps alluding to the M-
Form. Toyota’s assertion that technology, a generic factor of the automotive industry, is a
global resource was notable when the company stmcture is seen as being domestic. If the
technology is equated with physical assets then the company appeared to view them as
part of globalisation, retaining its human assets within its home base and implying that
this is where its ownership advantages lay. This view finds support in Klein (2000) that
control of an organisations assets originates in the ownership of the labour contracts. In
this way, the physical assets of production can be globally spread but control of them is
retained by having the core human assets under centralised control.

Respondents were asked to comment on the forces behind intemational stmcturing and a
shortage of human assets seemed to be a common problem and one quoted by several
respondents. MRI felt that HR was an inhibitor of intemational stmcturing in the case of
Japanese manufacturers due to the lack of Japanese managers who could oversee foreign
operations. This further emphasises the centralised govemance stmcture of Japanese
companies. ACEA also noted a strong social contract among Japanese manufacturers to
support the home base, METI adding that this was manifested in an investment policy
explicitly centred on Japan. The home base was

298
also seen as a source of technical innovation by both METI and MRI. This ignored the
possibility of technical innovation in other locations, a perverse conclusion given the
intemational evolution of the industry.

Honda was a good example of how companies intemationalise their activities only when
obliged to do so, as opposed to seeking new opportunities. It appeared that foreign
markets were beneficial only in exploiting economies of scale, while design variants and
economies of scope were a means to achieve this. Table 7.4 shows how Honda is
selectively globalised. There was support for core R&D being conducted in Japan, most
cmcially Mr Yoshino asserted the location advantage of maintaining 90% Japanese input,
but there was overwhelming support for a global application of the company’s culture.
Local R&D was seen to be part of this but the location advantage for it was minimal,
however, and Mr Yutaka Ikeda (28 February 2004) described how even the local R&D
was controlled from Japan, the project leader for the Accord US design team spending
three years in Japan. The purpose, as Mr Sumita articulated, was to exploit the location
advantages of producing as close to the target market as possible with a locally adapted
product. Conversely, Mr lida pointed out that it was possible for production to be shifted
around the world to maintain efficient production schedules, but supplying the American
market with production from the UK is, of course, the very opposite of producing close
to the target market. It would appear that although the location advantages of producing
close to the market are present, exploitation of economies of scale is again the primary
concem, coordinated through centralised control mechanisms.

Table 7.4 Domestic and global structures of Honda

Data Source Integration of Resource

Japan Global
Mr H Yoshino Company philosophy
Regional autonomy
Core R&D Local R&D
Niche R&D (China)
Mr T lida Company philosophy
Local management
Mr T Sumita Regional autonomy
Core R&D Local R&D
Market based production
Prof. T Demizu Senior management

299
Despite Honda’s so-called “glocalisation” policy, a globalised corporation with
delegated local responsibility, it seemed that the company had retained its unified
domestic structure. The global spread was claimed to be about servicing intemational
markets with local production and some ability for locally designed product variants.
However, when a plant was installed for a specified output it then became paramount to
maintain the output even if it meant serving multiple markets. First amongst all plants
would be those established in the home base. Honda representatives stated that the
variants in foreign markets were prepared at the local level, although it was clear that
Japanese central control remained strong. Including R&D facilities, it would appear that
ownership advantages were dominant over location advantages. This was reflected in a
govemance stmcture that was more reminiscent of the U-Form than the M-Form.

With markets seen as an opportunity to raise production levels there was a reluctance to
adapt products to local market conditions anymore than was necessary. Table 7.5
summarises the degrees to which products are adapted to overseas markets, as reported
by the respondents.

Table 7.5 Degrees of product adaptation to overseas markets

Data Source Unit of Analysis Strategic Response

Adaptation No Adaptation
Toyota Ferrari Japan Rain ingress
Mr M Harbour Nissan Europe No high
performance diesel
Product brands Option packs Core brand DNA
SMMT Jaguar No diesel
Honda Europe No diesel
Saab New SUV niche Hatchback only
Porsche
Mr P Woods Rover Mini Japan
(coincidental) Profitable niches
ACEA European sales in
Japan

At Toyota there was an amusing story about Ferrari’s failure to rain-proof their vehicles
against Japan’s rainy season, illustrating the manner in which Toyotas are more
conducive to foreign market adaptation. The easiest method of adaptation is to offer
special option packs, first mentioned in interview by Malcolm Harbour and expanded on
by Mr Woods. However, when it came to technical changes Mr Harbour

300
 and the SMMT noted that Japanese manufacturers tended to be unwilling to develop
diesel engine technology, Honda not releasing its first diesel engine until 2001. Similarly
with the body structure constraints set by the Budd Paradigm, there are limits to how far
a vehicle can be adapted without incurring expensive product redesigns, SMMT’s
mention of Saab’s hatchback being an example: sales of a saloon variant might not have
justified the development costs. This does not mean that new products are not designed
for foreign markets but only, as in the example put forward by SMMT of the Porsche
Cayenne, if that represents the core market for that product. In such a case, there may be
limited adaptation to the company’s home market.

At Honda there was a much more conclusive view of global markets: well established
intemational sales volumes meant that there was little in the way of new strategic market
challenges. Mr Yoshino was clearly proud of the level of intemational sales, specifically
quoting motorcycle successes. Mr Sumita mentioned that the product needed to be
matched to the market and Honda has sufficient economies of scope in the product range
for any market to be receptive to at least one model. Professor Demizu mentioned that
low price models were produced in Thailand, yet these were models already established
in the Honda canon. This could be further supported by the ability to schedule production
globally, as Mr lida boasted, yet again this is not a strategy that can be chosen a priori;
Honda has established worldwide production and so it can call on it when necessary, but
this is not the same as planning worldwide schedules and then installing the plants. It was
clear that as Honda grew in size and product range it became increasingly easy for it to
enter new markets with the minimum of adaptation.

Overall, the contextual interviews did not reveal a consensus on the opportunities raised
by globalisation, and much of the reason for this lies with the companies being industry
leaders within the automobile paradigm. All the Japanese manufacturers kept a strong
domestic structure from which activities were extended to global locations. This was
done strategically based on existing economies of scope in the product range, production
methods and management rather than to exploit factor endowment differentials. Where
there were cost benefits to new locations, from low labour costs to local government
incentives, these were incidental to the overriding strategy of supporting existing
production capacity. The degree of product adaptation practised

301
by the companies was confined by the technical limitations of the Budd paradigm, which
restricts the number of variations possible for a given vehicle body structure, but
corporate control also held back local autonomy in design. For example, although US
manufacturers use a separate chassis as the basis for their SUVs, and thus enjoyed
greater design flexibility, almost none of the Japanese vehicles in the US have used the
same system. Japanese manufacturers, therefore, remain relatively weak in SUV sales
while US manufacturers can offer the market product diversity.

Intemal approximations to the automobile paradigm were generally viewed in terms of

maximising the use of existing production facilities and this was supportive of the
precepts of the automobile industry paradigm. Industry leaders received incremental
benefits from such approximations since it diversified risk so there was no motivation
for fundamental restmcturing. Extemal approximations, in the form of inter-company
collaboration, were next examined for evidence of restmcturing that could provide more
than just incremental advantages.

7.4 Collaboration

If the automobile industry paradigm defines the lower limit of sustainable production
(MESp) then it behoves the sub-optimal automobile company to find some stmcture that
replicates the cost advantages of the paradigm. Intemal methods such as product life
extensions and flexible manufacturing help to maintain production levels but do not
correct for uncompetitive scale. Extemal methods involve some kind of partnership with
another company in order to share resources, although a partnership can take various
forms. Grossman and Hart (1986) define a firm as having control of its assets, but not
necessarily ownership. This being the case, for the purposes of comparison, the
relationship between parent and subsidiary can be viewed as being equivalent to a
partnership between independent firms.

This is not merely a research device; companies like Bentley, Aston Martin and Jaguar
have to make business cases for funding as much as any independent company, even if
the source of that funding is within the parent group. While this semi-

302
autonomy should not be confused with alliances between full independents it is
instructive on the nature of collaboration.

Two themes emerged from the collected interview data:

• Evaluation of potential partners

• Nature of joint ventures

Chapter 6 discussed how the divisional companies such as Jaguar and Bentley seemed to
hold a semi-autonomous status within their parent groups, with a certain amount of ad
hoc involvement on both operational and strategic matters by the parent companies.
Platform sharing might be expected to have long-term strategic importance due to its
similarity to scale alliance concept of Dussauge et al. (2004). However, these
relationships were often short-lived, sustaining the recipient until it could develop its
own platform design. This was also the case for link alliances between independent
firms, where Mr Yamaguchi saw frequent culture clashes. Numerous examples are
available in the literature. BMW tried a joint venture in engines with Chrysler, but this
lasted only from 2000 to 2006 (Just-Auto, 2007).

Apart from the exception of the relationship with Rover Group, joint ventures at Honda
did not involve equity holdings. These joint ventures were generally short-term and
compensated for a temporary weakness, a feature of link alliances. Table 7.6 sets out
how the participants at Honda viewed the role of joint ventures at the firm, the impact of
the Honda-Rover Group joint venture on MG Rover being further considered in Part II
of this chapter.

303
 Table 7.6 Honda - Impact of joint ventures

Data Partnership Im pact

Source Im m ediate C ontinuing
MRI Peugeot/Rover/Isuzu Diesel engine Ended
MrT Rover Introduction to Europe
Sumita Helped Rover
MrK Rover Joint products Life cycle problems
Coordination difficulties
Watanabe BMW - Rover Establish common ground
MrC Wrecked progress with Prevented Rover
Rogers Honda - any partner Honda sustainability
MrH “Three legged race”
Yoshino

With diesel engines Honda seems to have gone through at least three different
partnerships before developing its own units. This supports Dussauge et al.'s assertion
that link alliances are short-lived. The point is further emphasised by the joint venture
with Ballard to research fuel cell technology, Honda leaving the alliance when it had
learnt enough to conduct its own research. Most stark of all is Mr Yoshino’s assertion
that a joint venture is like running a “three legged race”, and Mr Kunihasa Watanabe
lists some of the coordination difficulties with Rover Group. Although Mr Watanabe
stated that these problems were solved with time, Mr Sumita commented that over the
long-term there were problems in having model life cycles that were not synchronised.
Despite this, Mr Sumita felt that Honda had gained from Rover Group in easing access
to the European market, just as his colleague Mr Chris Rogers believed that Honda was
a stable partner for Rover Group.

In the course of the interviews there was a high degree of support for the theories of
Dussauge et al (2004). Link alliances were expedient in the case of platform sharing
within Ford’s PAG, while Honda went through a series of link alliances while seeking to
enhance and subsequently protect its R&D. For scale alliances the interviews revealed
more longevity, Toyota’s joint venture with GM at NUMMI being the classic example.
The common feature of link and scale alliances amongst car manufacturers, that are
industry leaders within the automobile paradigm, is that the allinaces support existing
activities. In this sense they can be seen as bringing incremental improvement to the
competitive position. Like the intemal

304
approximations to the automobile industry paradigm, extemal approximations are not,
therefore, achieved through novel restructuring but adaptation of the existing
stmcture.

7.5 Conclusion - adaptation to the automobile industry paradigm by contextual

companies

Production extensions were unanimously rejected by the interviewees representing the

contextual automobile industry. Global markets mean that the older products would be in
direct competition with both the replacement model and the latest generation from
another manufacturer. This may change in the future as manufacturers find that retaining
product platforms can enhance economies of scope, particularly with regard to launching
low cost models such as the Renault Logan. In more general terms, production extension
was only applicable at the functional level with a marked tendency amongst the Japanese
companies to retain use of existing facilities in the home market. Production
intensification in the form of flexibility, however, was widely installed, even where it
was only to facilitate faster new model changeovers. This greatly facilitated production
scheduling and reduced the risk of variation in production output. It did not, though,
represent a restructuring but an adaptation of existing facilities to maintain production
capacity utilisation.

Similarly, the contextual companies, all of them multinational enterprises (MNEs) or

their divisional companies, used the apparent advantages of globalisation to support their
established structures rather than to restructure. In terms of the OLI model, the
advantages seemed to be on the ownership side with companies extending their existing
structures into new regions rather than undergoing fundamental restructuring themselves.
Locational advantages were therefore due to the size and structure of the market rather
than its inherent quality as a production site. There also appeared to be intemalisation
advantages which inhibited the formation of partnerships, most notably with link
alliances where ex post opportunities would tend to accrue to one partner over the other.
Scale alliances showed greater longevity, as predicted by Dussauge et al (2004), but they
did not lead to expansion of the alliance to other facilities. The contextual sources
provided a great deal of data showing that although a desire to

305
suppress costs was a major driver of intemational growth, this was countered by a
strategic desire to defend ownership advantages and maintain the integrity of the
domestic operations. This strategy was then managed by a centralised govemance
stmcture.

The presence of intemational factor endowment differentiation did not result in a

systematic strategic response. Many respondents noted that the ability to serve global
markets is complicated by local differences, either due to the policies of the authorities
or the market. There were many instances of companies failing to offer appropriate
products, particularly diesel engine derivatives. In Japan it was noted that European
manufacturers preferred to occupy profitable niches with low volumes of products
barely differentiated from the European versions. It was often stated that some locations
offered lower cost production, yet global production schedules could override them.

For those companies that were industry leaders within the automobile industry paradigm
the main concem seemed to be with maintaining the status quo of the company within
its industrial environment. As such, the intemationalisation strategy seemed to be
targeting at reducing risk, particularly in production variation. Since MG Rover was
sub-optimal within the paradigm, nor had a larger parent that was, it was examined as a
case study to see if it could throw light on the potential ways in which an uncompetitive,
full-function car producer might overcome its cost disadvantage. The empirical data for
this will be examined in Part II of this chapter.

306
 Part II: MG Rover

7.6 The rise and fall of British Leyland/Rover Group

This section of the thesis will present the historical antecedents to the independence of
MG Rover in 2000. This commences with a brief history of the two brands, MG and
Rover, describing how they came together. The section will then establish the roots of
the problems facing British Leyland (BL), and then Rover, in particular their failure to
restructure, leading to a vicious cycle of lack of funds preventing new product
development and thence further declining revenues. BL first attempted to counter this by
improving its intemal capabilities with injections of government funding. When this
proved insufficient, it then tumed to extemal assistance in the form of a joint venture
with Honda, followed by ownership and funding by BMW.

7.6.1 The consolidation of MG and Rover

The MG Rover brand was comprised of two marques, MG and Rover, plus a variety of
moribund brands. MG was founded sometime in the early 1920s under the guidance of
Cecil Kimber, the exact date being unknovm. It was owned by William Morris as part of
his Morris Garages business until it was merged into the Morris Motors car
manufacturing business in 1935. Morris Motors, as part of the Nuffield Organisation
(including Wolseley and Riley), then went through a series of mergers itself:

• 1952 merger with Austin to form the British Motor Corporation (BMC)
• 1966 BMC merged with Pressed Steel and Jaguar to form British Motor
Holdings (BMH)
• 1968 BMH merged with Leyland to form British Leyland Motor Corporation
(BLMC)
• 1975 British government took a majority equity holding in BLMC and effectively
nationalised the company.

307
Throughout this period the bloodline of the MG marque became increasingly diluted.
The marque has always been a ‘badge engineering’ marketing tool of its parent and has
never had its own dedicated R&D or powertrain operations. It did have its own factory
though, at Abingdon, from 1929 to 1980. After the factory closed the nadir was
purportedly reached when the MG brand was attached to the poorly received Metro,
Maestro and Montego variants (Williams et al 1987; Sharratt, 2000). MG regained its
credibility in 1998 with the release of the MG-F sports car, and still more from 2000
under MG Rover when attachment of the badge to sportier versions of Rover models
was greeted more positively than previous ‘badge engineering’ strategies.

Rover has an altogether weightier presence within the history of the British automotive
industry, as documented by Graham Robson (1988). James Starley and Josiah Turner
began The Coventry Sewing Machine Company in 1861 but the latest innovation in
personal transport, the bicycle, prompted a change in direction in 1869. Starley then left
to start his own company with cyclist William Sutton, using the name Rover, and
producing the first safety bicycle in 1885. The company moved progressively into car
manufacturing during the early years of the twentieth century and concentrated solely on
cars from 1924. With the arrival of the Wilkes brothers in 1929 the products gained a
reputation for solid reliability. However, the company also showed moments of
innovation, such as the Land Rover off-road vehicle in 1947, the initial designs of which
were based unofficially on the Willys Jeep, and the first gas turbine road car, JET 1, in
1950.

Like MG, the Rover marque was subsumed into successive mergers:

• 1967 merged with Leyland Motor Corporation

• 1968 Leyland merged with BMH to form British Leyland Motor Corporation
(BLMC, renamed BL from 1977).
• 1976 Rover SDl established intemal merger between Rover and Triumph.

After the launch of the SDl, the Rover brand became part of British Leyland’s marketing
strategy, being applied to a variety of models, some of which had not

308
started out as Rovers. By 1986 the Rover brand was chosen for the car-making group,
changing from BL to Rover Group at the same time as Land Rover was spun off as an
autonomous business unit.

7.6.2 Internal rescue strategies - 1975 to 1978

In 1975, BLMC had come to dominate the domestic automotive industry in terms of
output but it lacked intemal resources to fund the replacement of ageing and unprofitable
product lines while still consolidating its operations after decades of mergers (Pilkington,
1999). The combined operation comprised almost every aspect of the automotive
industry in the UK, from tmcks and buses to cars, but without restmcturing it was simply
the same set of companies under a single corporate banner. In the words of Michael
Edwardes, chairman of BL 1977-1982:

“BL represents a microcosm of the issues affecting the British industry as a whole.”
(Edwardes, 1983, p.9.)

It was not a single full-fimction firm exploiting available economies of scale, as

prescribed by the automobile industry paradigm, but an affiliation of small full-function
firms, all of them suffering the same scale disadvantages they had before the mergers. In
the mid 1970s, BLMC was comprised of eight companies and fifty marques.
Manufacturing capabilities included Longbridge (Austin) and Cowley (Morris) plant
complexes, plus another seven final assembly plants handling production for Jaguar,
Rover and Triumph. Each of these plants had remained attached to its historical brands,
creating inflexibility in production (Bhaskar, 1979). When flexible production was
introduced in the 1980s with more automated manufacturing systems the low level of
sales simply exacerbated the problem of low capacity utilisation with the burden of the
higher fixed costs of the new capital investment (Williams et al., 1994).

The merger that created BLMC had not come about through commercial necessity but
political motivation, a government creation pushed through by Anthony Wedgewood
Benn via the mechanism of the Industrial Reorganisation Corporation in 1968 (Church,
1994). The government had been concemed about the inroads being made

309
by u s manufacturers, such as Chrysler’s purchase of Rootes, and the national balance of
payments. As the elements of the company were brought together it became apparent
that there were not the managerial resources to cope with the new structure (Bhaskar,
1979). The inclusion of the Leyland truck company was intended by the government to
improve financial controls in the car manufacturing division. Pricing of cars at BMH,
for example, was made without reference to the innovations on which the competitive
advantage was founded: Ford calculated that the technically advanced Mini, then priced
at £496, lost £30 on each one sold (Church, 1994). Investment funds for the future were
further dissipated by the company policy of continuing to pay dividends, even until
1975.

Some fault was found with the technicalities of production, Lewchuck (1987) criticising
the company for failing to adopt the Ford style of mass-production while retaining old
systems of payment and control. This view does not find favour with Williams et al
(1994, p. 137), pointing out that production engineer Frank Woollard had a good
understanding of the need for Automatic Transfer Machines (ATM) and used pre-
cellular flow in a pull-through production style reminiscent of Toyota’s TPS production
system. This resulted in car assembly at Longbridge CABl (car assembly building 1) of
12.5 finished cars per worker. However, while Woollard understood the global
innovations in production, output capacity did not find support in sales as it had at
Toyota. Foreign opportunities were reduced by the effects of national protectionism, the
company having little presence within the EEC while the home market was maturing. In
1975 imports took 33% of the UK market, overtaking BL on 31% for the first time, and
the situation was exacerbated by the growing influence of North Sea oil revenues on the
strength of the British currency. A crucial factor at this time would have been effective
cost recovery to fund future investment, but instead £103 was being lost per vehicle in
1975, deepening to £662 by 1980 (Williams et al, 1994). The company produced 1.26m
vehicles in 1975, dovm from 1.7m in 1973 and lost £76m (Church, 1994).

Government intervention continued throughout the 1970s. Three reports were published
during 1974-75: the Ryder Committee, the Trade and Industry Sub-Committee and the
Central Policy Review Staff. Most prominent was the 1975 Ryder Report, stating that
the company suffered from outdated facilities, a weak

310
organisational structure and poor planning in consolidating its disparate operations
(Alvarez Gil and Gonzalez de la Fe, 1999). The report was devised under the
chairmanship of Don Ryder of the publishers Reed International (Williams et al, 1994, p.
151). Emphasis was placed on economies of scale with the stated intention being to
expand output to fill available capacity, rather than rationalising the internal capabilities
of the company. Bhaskar (1979) noted that BL was in short supply of engineers of all
types, quantifying this deficit in thousands rather than hundreds. The Ryder Report
suggested consolidating group R&D at one location, but this was resisted by Jaguar
which attempted to protect its independent spirit.

The production funding entailed an investment of £2bn in new models and capital
equipment with a planned rise in production from 605,000 in 1975 to 961,000 by 1985,
closer to the theoretical production capacity limit of 1.2m (Williams et al, 1987). The
Ryder Plan was criticised for being expansionist without addressing internal problems of
a complex corporate structure, inadequate technical resources and poor industrial
relations (IR) of the company, which lost it 252,000 units in 1977 alone (MIRU, 1988).
The rise in production capacity might have been appropriate if the mass market had not
shifted in favour of the business user, company cars having become established as an
employment incentive during a period of pay restraint. Ford had come to dominate the
business market since the 1960s, taking 40% of the segment by 1975 (Church, 1994).
BL traded on its product innovation which appealed to the private purchaser, but just as
this segment was shrinking the sales were squeezed further by the rise in popularity of
imports (Williams et al., 1987).

Bhaskar (1979) concluded that the Ryder plan had failed partly because it was ambitious
beyond the competence of those enacting it, but the most important failing was in
linking funding to improvements in IR and output. When stoppages brought down
production, the funding that was available was spent on operational needs, instead of
investment. From 1975 to 1977, BL continued to suffer falling production and market
share, the car division alone dropping market share from 32% in 1975 to 24% in 1977
(MIRU, 1988). In 1977, Michael Edwardes was brought in from the National Enterprise
Board (NEB) having turned round the battery maker. Chloride. Internal restructuring in
1978 meant that BL was reorganised with the cars division being divided into Austin-
Morris and Jaguar-Rover-Triumph and each accorded

311
decentralised decision making. The Corporate Recovery Programme was presented to
the government in July 1979 with support from the workers. It was based on a reduction
in production capacity, from the 1.2m theoretical limit to a planned 750,000 units a year,
and a reassertion of the management’s right to manage the firm. Michael Edwardes
publicly stated that this strategy was:

“...shrinking the business to a level beyond which we believed it was not safe to go”
(Williams et ai, 1987, p.95).

Plants, such as Speke, were closed and car production was concentrated on plants at
Cowley and Longbridge. Government funding was not contingent on results and was
made available for investment in future programmes. However, the company found that
redundancies still lagged behind falling sales. Sales had halved to 315,000 between
1972 and 1980 but the work force only halved to 55,000 in the period 1979 to 1983
(Williams et al, 1994). Moreover, the company continued to be dogged by poor
economic conditions compounded by a lack of new models. With the company’s intemal
capabilities still uncompetitive, Michael Edwardes decided that the salvation of BL cars
lay externally, and so the search began for a joint venture (JV).

7.6.3 External rescue strategy - link and scale alliances

Michael Edwardes stated that the partner had to be in a position of commercial

coincidence; similar but complementary needs and aspirations within the same segments
as BL (Edwardes, 1983). The purpose of the JV was to improve BL’s intemal difficulties
with new product development, the intention being to retum the company to health as an
independent and competitive automobile manufacturer. In the terms of this research the
need was for a JV which would retum BL to a competitive status within the automobile
industry paradigm.

The fear was that the company would lose its full-fimction status and become an
assembler of vehicles, thus losing the political support that sustained it. Potential
problems were identified as cultural prejudice against foreigners, language barriers,
geographical distance and an insecure future for the company if the partner firm
dominated. Renault was looked at in 1978 and then rejected on fears that being larger

312
and state-owned it had the resources to ultimately overwhelm BL. In the same year, Bob
Price, managing director of Vauxhall, proposed that he could solve BL’s problems by his
company providing the mid-range replacement for the Marina, Maxi, Princess and
Allegro and receiving in retum the small car design that Vauxhall lacked. However, the
Gemini project being offered by Vauxhall was then cancelled by its owners, GM
(Edwardes, 1983).

A proposal by Edwardes, codenamed “Dovetail”, to solve the problem of poor R&D

engineering resources at BL was to merge with Chrysler UK, close two of the Chrysler
plants and absorb its R&D facilities. It should be noted that it was the inability of the
government to let Chrysler UK fail that had undermined attempts by BL to achieve
market growth (Church, 1994), the British government having a policy of supporting the
subsidiaries of US firms as if they were indigenous even though they operated without
reference to the United Kingdom:

“Their corporate strategies are intemational, their links are within the group rather than
within the country, the major decision loci are abroad, they can survive without Britain.”
(Wilks, 1990, p. 175.)

Edwardes seemed to implicitly recognise that developing a link alliance in R&D would
not have been stable and so a complete merger with Chrysler’s R&D was the preferred
option. The deal would have secured access to new engineering capability and one
dedicated to new BL models, rather than one distracted by its own model range; this had
been the very problem that had dogged the mergers that comprised BL and its
antecedents. However, the merger was overtaken by the purchase of Chrysler UK by the
Peugeot in 1978. Other plans were to offload Ley land trucks onto Fiat/Iveco, yet
another to have John Delorean take control of Range Rover sales in the US, but none of
these came to fruition (Edwardes, 1983).

In 1978, ex-ambassador to Japan, Sir Fred Wamer, met Honda President Kiyoshi
Kawashima. At that time, although the two companies were superficially competitive
within the automobile industry paradigm, the two had distinct differences. BL was still
handicapped by its multi-company heritage of 16 model families, 30 factories, 8 final
assembly plants and had come under state control. Honda was a tightly

313
structured private company with 2 model families and 2 assembly plants (see Table 7.7).
There was little commonality in engine technology, Honda having promoted the CVCC
(Compound Vortex Controlled Combustion) technology while BL had inherited whole
families of power units. However, there were also the commercial coincidences that
Edwardes was seeking. Production levels were broadly similar, British Leyland
producing 800,000 cars and trucks to Honda’s 740,000 (Mair, 1994), figures that include
SUVs, and they both specialised in front-wheel drive technology on cars in comparable
market segments. This suggested that the JV would be complementary in engine
technology, where BL could gain from Honda, but with commercial coincidence in the
basic vehicle configuration which would permit the companies to share in vehicle
development. With equivalent levels of output, and Honda having no government
funding to draw on, there would be little danger of the Japanese company absorbing BL.

Table 7.7 Honda and BL comparison

Organisation facet Honda BL

Models 2 16
Plants 2 30
Engineering strength Powertrain Interior design,
suspension,
packaging
Model strength Small cars Executive cars
Other capabilities Motorcycles Trucks, buses
Management culture Engineering Bureaucratic
Source: Pilkington 1996

The two companies also complemented each other in the US market, Honda being
present with a small car and BL having the large Jaguar sedan. Initially, though, the first
priority of the British company was to find a stop-gap product to keep the production
lines running between the launch of the Metro supermini in 1981 and the medium-size
Maestro in 1983. In 1979, the “Co-operation for business” deal was signed at the same
time as the Corporate Recovery Programme reduced British Leyland to 13 factories of
all types and shed 25,000 jobs. The Metro began the “product led recovery” in 1981 and
in 1982 the Honda Ballade emerged from Longbridge as the Triumph Acclaim, built up
from complete knock-down (CKD) kits.

314
While it was not the first relationship between Japanese and British manufacturers
(Nissan had built Austins under licence in the 1930s and 1950s) it was, perhaps, the
moment the Japanese industry came of age. It signalled the swansong of Triumph and
the arrival of the Japanese as potential saviours of the British car industry.

Honda was still a relatively new car manufacturer in 1978. It had begun manufacturing
cars in 1962 with the S360 sports car, but did not break into the mass market until 1969
when the Civic was designed for the US market. This was crucial to the future growth of
the company:

“We all believed that if the project failed Honda would have to give up its plan of
becoming a full-fledged carmaker.” (Hiroshi Kazawa, engineer, Honda, 1999, p. 109.)

The larger Accord built on the early sales penetration of the US market by the Civic and
eventually went on to become one of the biggest selling passenger cars in the market.
The company was also experienced in foreign production, particularly in motorcycles
where it had been manufacturing in Belgium since 1963 (Honda, 2006). However, in
Europe the company had no automobile production and little sales exposure, though the
products were conceptually similar to European models (Mair, 1994). Honda was
therefore looking for two benefits from a relationship with BL:

1. Market penetration
2. Production facility

Categorising the relationship according to the typology of Dussauge et al. (2004), from
Honda’s point of view it was predominantly a scale alliance, offering the company
access to BL’s final assembly facility. In the process, it could gain market knowledge and
build contacts with European suppliers, though this could be achieved independently of
BL. Figure 7.1 shows how the JV for the Triumph Acclaim brought together the
assembly functions of the two companies structures. Since the relationship involved
assembly from Honda supplied CKD kits. Figure 7.1 demonstrates that BL’s capabilities
in BIW, R&D and powertrain were largely irrelevant, although they were being
employed elsewhere within BL. Furthermore, the

315
market intelligence from the UK operation related to this model accrued mostly to the
benefit of Honda in Japan.

Figure 7.1 Honda and BL joint venture 1981 - Triumph Acclaim

Honda - BL ------------------------------ 1

Powertrain Powertrain

Final Assembly
(Japan and UK)

Japanese UK Market
Market

For BL, Figure 7.1 shows that the relationship was a scale alliance in final assembly,
allowing the company to continue operating an existing assembly facility until the new
models came on stream from 1983 with the Maestro and then the Montego. Pilkington
(1999) found little of financial or technical benefit to BL in the first stage of the
relationship, its other functions being effectively isolated, but at least it improved the
company’s reputation in the market. Pilkington also states that it was from this point that
BL recognised the potential for a link alliance, learning from Honda in order to improve
BL’s own R&D capability. However, Pilkington did not find any evidence that sufficient
knowledge was transferred during the period of the Triumph Acclaim project.

The link alliance aspect of the relationship seems to have developed in the volatile
manner that Dussauge et al describe. Pilkington (1999) describes five major

316
programmes and the degree of input by BL (see Table 7.8). The first, the Triumph
Acclaim, was a lightly reworked Honda Ballade but it created the cooperative
environment for a little more input, the Rover 200 of 1984 offering one BL powertrain
system.

Table 7.8 Honda-Rover alliance - BL/Rover share of vehicle programmes

Model Year of BL/Rover share

Release R&D Production
Triumph 1981 Cosmetic Final assembly
Acclaim 1984
Rover 200 Cosmetic, one Final assembly
Rover 800 1987 powertrain
30% of design, one BIW, final assembly
1990 powertrain
Rover 200/400 40% of design, two BIW, final assembly
1994 powertrains
Rover 600 Cosmetic, one BIW, final assembly
powertrain

Source: Pilkington, 1999

The Rover 200 was still heavily based on an existing Honda, the new Ballade, but its
release in 1984 was actually preceded by the start of a link alliance between the two
companies. In 1981, Project XX was announced, a new departure for both companies
since it was the first large car for Honda and the first front-wheel drive large car for BL.
By agreeing on common design references (or “hard points”) the two firms could offer
their own styling and powertrain variants. The final Honda product, the Legend, hid any
British influence very comprehensively and no British engines were offered. This
revealed Honda to be the dominant partner, as noted by Pilkington (1999). Indeed, the
Japanese seemed to specify the design to the degree that the only large engine that
would fit the range was the Honda V6. However, the Rover 800 range was considerably
more expansive, having four-cylinder British engines and a hatchback version in
addition to the luxury saloon with the Honda engine.

317
 Figure 7.2 shows the state of the alliance between BL and Honda during Project XX,
with a link alliance in R&D and scale alliances in final assembly (both brands being
assembled in both countries) but with only Honda providing engines to both versions. It
also shows that Honda continued to benefit in the form of market intelligence from
production exposure to the UK market.

Figure 7.2 Honda and BL link and scale alliance 1987 - Project XX

Honda r BL -------- 1

R&D
(Japan and UK)

BIW Powertrain BIW Powertrain

Final Assembly
(Japan and UK)
I

Japanese UK Market
Market

The independence of the two companies on final execution of the vehicles was
underlined when Honda were able to unveil their version of Project XX in October 1985,
a full nine months before BL showed theirs (Robson, 1988). From the disjointed
chronology it seems clear that Honda gained early from the link alliance and was then
able to pursue its own model agenda. In the sense of being able to get to the market
sooner the advantages accrued asymmetrically to Honda. However, despite the cost
savings in sharing the design stage it was still a prolonged gestation period, the later
commencing Accord, a wholly Honda design, reaching the market before the alliance
generated Legend (MIRU, 1988). There were also plans for joint manufacture but Honda
was reluctant to subject itself to BL's quality control, a policy vindicated

318
in the statement by Williams et al (1994) that even in 1991 the Rover 800 was averaging
£325 of warranty work per vehicle, four times that of the Legend.

In addition to forming a link alliance in R&D, BL needed to reform its structure and its
IR policies (Pilkington, 1999). With Honda’s skills in manufacturing it might be
assumed that this would have been achieved through use of the Japanese production
technology, but in fact only 10-15% of BL’s subsequent improvement in productivity
was due to technological innovation (Scarborough and Terry, 1996). Mair (1994) even
asserts that Nissan was more influential because it set the industry standard, though BL
would later gain more by easy access to Honda’s experience starting the Ohio plant,
rather than by Honda directly instructing BL in the UK. At the time of the
commencement of the alliance the reputation of Japanese manufacturing was low, the
Japanese advantages believed to be derived from the factor differentiation of low-cost
labour, exploitation of suppliers and copying of Western technology. Initially there was
organisational resistance within BL to learning from Honda and British visitors to Japan
were only learning by observation, the language barrier making the culture impenetrable
to them. The Honda experience in Ohio should have been an invaluable opportunity for
BL, but it was not until 1991 that the British started their first formal studies of the
Honda culture. As described by a BL manufacturing manager:

“We had our share of problems with Honda to start with. Then, after a while, it all sank
in. Why all the aggravation? They’re right, we’re wrong. All of a sudden the concept
changes, and things fall into place” (quoted in Mair, 1994, p.273).

BL did not adopt Honda’s method of production in batches, a US market orientated

build-to-stock system requiring extended planning (Pilkington, 1999). Rather, BL had its
own Minimum Inventory Control/Strategy (MIC/MIS) system using a version of JIT but
with large central warehouses and a version of kanban which was, though, not as
flexible (Pilkington, 1996; MIRU, 1988). Furthermore, the production system was
purchased from Honda and thus did nothing to improve BL’s production engineering
capability. Although the joint venture was intended to be a link alliance, transferring
knowledge to BL, by accepting Honda’s control over product and production
engineering BL’s own capabilities were being emaciated:

319
“It is all too clear that this attraction [to accept Honda systems] diverted the firm away
from implementing a learning-based strategy.” (Pilkington, 1999, p.467)

BL was in a position, though, to use the period of stability provided by the alliance to
conduct some reorganisation. It was also fortunate that the fundamentally perilous state
of the company had created an understanding by the unions of the need for
reorganisation. The company asserted the right to manage which found expression in the
direct communication with the labour force, often from Michael Edwardes himself. This
marginalised the unions, but also built trust and co-operation with the workforce based
on the particular circumstances of the company at that time. It was not, therefore,
something that could simply have been imported from Honda. Indeed, as Mair (1994)
stated, BL was learning from Nissan as much as Honda. The reorganisation of IR was
started before the relationship with Honda was promulgated, commencing with the
teamwork system for the all-British Metro, although Smith (1991) claims that this was a
case of workers coping and looking out for each other. In 1987 a TQ (total quality)
initiative was introduced and further refined for “Rover Tomorrow - The New Deal” in
1991, and this was formulated by senior management, being only indirectly influenced
by experience with Honda (Scarborough and Terry, 1996). This recognised the poor
existing economies of scope in management:

“The ‘enemy’ isn’t the Japanese or the Germans; it is the limitations of our managerial
‘culture’.” (Pascale and Athos, 1986, p.201)

From 1982 BL looked to capitalise on the gains made from the link alliance with Honda
by restructuring the company under Sir Austin Bide into the Cars Group and Land
Rover-Leyland Group. Jaguar was privatised in 1984 and in 1986 an attempt was made
to sell BL to Ford, but this was considered politically unacceptable in the wake of the
fallout from the scandals involving Westland Helicopters. In the same year a record loss
of £539.2m was announced and BL was renamed Rover Group when Graham Day was
installed as the new chairman. In 1987 the group was further broken up with the
privatisation of Unipart, Leyland and Freight Rover. Recognising that the remaining car
company. Rover Group, was still not ready for independence. Day called for recovery in
three phases, involving restructuring, then restoration of Rover Group to commercial
viability and finally privatisation (MIRU, 1988).

320
However, the four alternative recovery strategies that were under consideration seemed
to implicitly recognise that the company could not make a case for independence as a
mass market manufacturer. The four options were as follows:

1. Selling Rover to another manufacturer.

2. Selling the group to Honda.
3. A move to low volume specialist production with premium pricing.
4. Sale of the group to a non-automotive manufacturer.

Since the first three options imply that Rover Group needed a partner within the
industry, the fourth option would appear to have had little logic. Although this was
indeed the option that came to pass, it was only effected with the cooperation of Honda.

7.6.4 BAe purchases Rover Group

British Aerospace (BAe) first showed an interest in purchasing Land Rover in November
1987 to provide an add-on to sales of military aircraft. Instead, in 1988, Rover Group in
its entirety was sold to BAe for £150m with £520m in government aid and tax
allowances (Church, 1994; MIRU, 1988). The automobile alliance was secured by
equity cross holdings, Honda taking a 20 per cent share in Rover Group while Rover
Group took a 20 per cent share in Honda UK Manufacturing (HUM) in 1989 (Mair,
1994). This indicated that despite the fluctuations in the degree of collaboration for
successive projects, Honda was prepared to take a long-term view of the joint venture.

This politically acceptable solution provided long-term stability for Rover with minimal
disruption to suppliers and Honda. It was mooted that the two British companies would
be able to use a common computer aided design (CAD) system and merge administrative
operations in order to benefit from synergy. As it was, the relationship with Honda had
been continuing with the release of the Rover 200/Honda Ballade in 1984 providing
valuable assembly work for Rover. The Ballades made by Rover numbered only 30-
50,000 per year, but this was a significant utilisation of production capacity and with
50% of the value-added being British it made a further

321
contribution to the domestic industry as a whole (Williams et a l, 1987). This scale
aspect of the alliance lasted for the duration of the model (Pilkington, 1999). However,
Rover was still unable to extend the link aspect of the relationship into making an
entrance into the US market under Honda guidance, a move that was further undermined
by Day’s insistence on withdrawing from markets that did not provide acceptable
financial returns. The company was caught in a vicious cycle of not investing in markets
where there were poor sales, and the lack of investment then leading to a further
deterioration in market share. Despite the scale alliance aspect of the relationship with
Honda, the continuing lack of export sales meant that the three assembly plants in the
UK were not being fully utilised (see Table 7.9).

Table 7.9 Rover plant utilisation - 1987

Plant Capacity Output

Longbridge 400,000 279,000
Cowley South 250,000 132,000
Cowley North 130,000 58,000

Source: Williams et al., 1987

Throughout the period under BAe ownership Rover Group benefited very little from its
parent. Financial stability allowed the alliance with Honda to progress but then Rover
Group had to forgo its involvement in successive projects when a lack of funds forced it
to instead extend production of existing models, albeit with refreshed styling. Indeed,
Rover Group was at the mercy of BAe’s financial status and the economic recession
during the ownership period (1988-1994) reduced the funding available to the car
operation (Pilkington, 1999). By this stage in the link alliance Honda had learned as
much as it required from Rover Group about UK production and had expanded its
assembly plant in Swindon. Rover Group, however, had been fhistrated in its
requirement for knowledge transfer from Honda. Each of the five projects shown in
Table 7.13 had been to Honda’s specification, leaving Rover Group to make adaptations.
The pinnacle of the relationship was Project XX, but Rover Group lacked the funds to
join Honda in its replacement in 1991 and so was forced to extend production of the
Rover 800 version, an unsatisfactory strategy in a globalised industry (see section 6.2.1
above). The final joint project, resulting in the Rover 600,

322
had very little Rover involvement and the company was even reduced to seeking
assistance from external agencies in adapting its production facilities to the new model
(Pilkington, 1999).

However, although Rover Group was not gaining as much from direct involvement with
Honda as critics suggested it should, it was an opportunity to concentrate what resources
it had on its own projects. The company designed and installed its new range of petrol
engines, the award winning K Series, comprising the full range of small in-line fours to
larger V6 engines, available from 1988. The company had also retained its ability to
conduct new model programmes by itself, fundamentally re engineering the Metro into
the Rover 100 and delivering on the Land Rover Discovery and Freelander (Pilkington,
1999). The Freelander was an entirely new venture for the company, combining the
rugged capability of Land Rover with the Budd Paradigm BIW production techniques of
the passenger car division. The vehicle also, therefore, proved that Rover was still
capable of production engineering. As Alvarez Gil and Gonzalez de la Fe (1999, p.395)
stated concerning Rover’s R&D capability during the period:

“It has maintained the core skills needed to develop new vehicles, like the Rover 100,
but it does not have the financial resources to maintain vehicle development across the
full range.”

The situation in 1994 was that Rover Group had consolidated around three assembly
plants (see Table 7.14), had its own powertrain and BIW facilities, as well as a capable
R&D operation. It was therefore a full-function company handicapped by poor levels of
output which reduced its ability to make funds available for investment. Having already
exhausted its internally generated attempts to reorganise sufficiently during the Ryder
plan and Edwardes eras, the company had a continuing need for another external
relationship, particularly if this came in the form of funding.

323
7.6.5 External assistance for internal improvement —BMW

By 1994, from Honda’s point of view, the link alliance was reaching maturity (Carr,
1999). The company had its own manufacturing facility at Swindon, producing engines
since 1989 and complete automobiles from 1992 (Mair, 1994). The company also
handled its own European sales distribution and marketing and had established its
regional headquarters in Slough in 1989 (Honda, 2006). Although it was not full-
function in the UK, all R&D taking place in Japan, it was able to operate quite
independently of Rover. This may have contributed to a sense of detachment from Rover
Group, and despite BAe suffering financial problems with its regional aircraft division
the decision to sell Rover Group in 1993 appears to have taken Honda by surprise,
despite it being a known fact that BAe’s agreement with the UK government not to sell
Rover Group before 1993 was coming to an end (Parliamentary Report,
2000).

Honda suggested raising its equity share to 47.5% if BAe retained an equal share and
Rover Group employees were granted the remaining 5%. According to Brady and
Lorenz (2001) this would have valued the company at £65Om, so would have required
an investment from Honda of around £179m for the 27.5% to add to the 20% it already
owned. However, demonstrating that BAe gained little of operational value from Rover
Group, the aircraft manufacturer was interested only in a complete sale to raise all
available funds. This left two more options: a management buyout (MBO) with Honda
retaining the 20% share, or acquisition by another company. Brady and Lorenz discount
the MBO option because Rover needed $lbn in working capital, which financiers were
not prepared to lend.

The third option might have created a three-way partnership, Honda and another
company both supporting Rover Group through a 20:80 equity partnership. According to
Brady and Lorenz (2001), Honda was looking to enhance its leadership in joint R&D
projects by expanding its managerial influence over Rover Group and protect the flow of
royalties paid by Rover Group for Honda designs. This it could not have done if its co-
owner was another car maker with operational plans for Rover Group. When BMW was
revealed as the interested party it therefore brought Honda’s plans for Rover Group to a
halt. BAe sold its 80% share in Rover Group to the German

324
executive car manufacturer for £800m (Parliamentary Research Paper, 2000) and a
month later Honda swapped its 20% holding in Rover Group for Rover Group’s 20%
holding in HUM. All future collaboration between Rover and Honda was abruptly
cancelled, though this did not affect existing models which continued in production, two
even outliving the period of BMW ownership. In retrospect, Honda seems to have been
relieved at extracting itself from the alliance with Rover Group (Economist,
2000).

BMW considered that it lacked sufficient volume and saw the purchase of Rover Group
as a route into the mass market without risking the premium image of its own brand. The
alternative was to create a new brand of its ovm but Brady and Lorenz (2001) point out
that such a development can be an expensive strategy, quoting a figure of $5bn for
Toyota to establish the Lexus brand. For Rover Group, the strategy of protecting the
BMW brand by keeping Rover at distance had substantial benefits for its engineering
infrastructure:

“This rendered Rover substantial autonomy, but compromised integration and scale
economies, and the strategy then depended heavily on some revived basic engineering
capability” (Carr, 1999, p.417).

If “integration” is considered to be analogous to a link alliance, then without that close

connection or the exploitation of economies of scale either, it is difficult to conceptualise
BMW’s involvement as little other than a source of investment funds. There was
speculation that BMW’s main interest had been to use the advantages of ownership to
secure access to Land Rover’s off-road technology. To a certain extent this seems to have
been effective, providing the German company with an opportunity to compare
technologies and later BMW off-roaders featuring the same type of ‘hill descent’
mechanism as Land Rover. However, it is not clear to what degree this was a Land Rover
development, and it is even available on Jeep models that have no connection to BMW
or Land Rover. Furthermore, the BMW X5 that emerged after the company divested
itself of Rover owed more to BMW’s own 5 Series platform and used a development of
the four-wheel drive system that had been in the company repertoire since before the
purchase of Rover Group. Concerning the new MINI, the

325
vehicle was proposed by BMW, not its Rover subsidiary, even though the style clearly
paid homage to the British original.

Rover itself needed to design a new range of vehicles in place of those that presumably
would have come out of the alliance with Honda and this involved a complete
restructuring of the Rover Group’s R&D capability. This necessarily resulted in a
substantial delay and the only new Rover model to emerge during BMW’s tenure was the
Rover 75, released in 2000, and having little in common with any BMW model. Some
commentators believe that the front-wheel-drive 75 was based on a rear-wheel-drive
BMW 5 Series platform, mainly due to the existence of a transmission tunnel down the
centre of the Rover’s floorpan, though MG Rover later found when developing the rear-
wheel-drive version of the 75 that the rear of the floorpan had to be adapted just to make
room for a differential. At least the electrical system for the 75 did have a BMW base,
which then caused problems for MG Rover when attempting to have it interface with the
V8 engine sourced from Ford. There were indications that BMW was planning to
increasingly integrate Rover into the BMW infrastructure, the new Range Rover being
well advanced by the time Land Rover was sold to Ford in 2000 for £2bn (Brady and
Lorenz, 2001) and on release in 2003 it was installed with BMW engines. If BMW had
further subsumed the mainstream models of Rover Group strategically in this way it
could conceivably have led to further delays in model releases, perhaps for the BMW
branded vehicles as well.

Although BMW provided the investment funds that Rover Group’s low output denied it,
this still did nothing for output during the interim. From 1994-1999, Rover Group’s share
(including Land Rover) of the passenger car market in the UK fell as the products that
had come out of the JV with Honda steadily aged (see Table 7.10):

326
 Table 7.10 Rover Group’s share of UK market

Year Market share %

1994 12.8
1995 12.3
1996 10.9
1997 10.0
1998 8.6
1999 6.5

Source: SMMT/Parliament Report, 2000

Financial losses deepened in parallel with the loss in sales (see Table 7.11), although
Rover Group management disputed the scale of the losses and even claimed a notional
profit (Brady and Lorenz, 2001).

Table 7.11 Rover Group financial losses

Year Financial loss (£m)

1995 158
1996 119
1997 92
1998 642
1999 750

Source: Parliament Report, 2000

The strength of the British currency at the time suppressed Rover Group’s exports while
the company was slow to capitalise on the relative reduction in cost of imported parts,
gradually raising the proportion of imported parts in the Rover 75 from 15% to 25-30%
(Brady and Lorenz, 2001). In 1999, BMW sought government aid for its British
investments and was promised £152m in public funds on condition that it invested
£3.3bn to secure future UK production. Suspecting that this contravened European
regulations on state aid, the EU Commission announced an investigation that would last
for several months into 2000 (Parliament Report, 2000). When, in mid-March 2000,
BMW finally decided to cut its losses and divest itself of Rover and Land Rover it
immediately felt the benefit as its share price rose 30% on the news

327
 (Business Week, 2000). After decades of mergers and consolidation, Rover once again
found itself independent.

7.7 MG Rover - autonomy and sustainability

During the disposal process Rover became the subject of a short bidding war between
Alchemy Partners, led by John Moulton, and Phoenix Venture Holdings (PVH), owned
by John Towers, Peter Beale, John Edwards and Nick Stephenson through their financial
vehicle, Techtronic (2000). Alchemy Partners considered that Rover was not capable of
supporting a three-model range of saloon cars, particularly as two were already outdated,
and intended to build on the MG sports car with the Longbridge workforce reduced by
half to 4,500 (Holweg and Oliver, 2005). This created some political resistance in the
UK while BMW was unsettled by the possibility that under the Insolvency Act (1986) a
deal with Alchemy would leave it exposed to Rover redundancy payments should the
company fold within two years (Brady and Lorenz, 2001). The PVH bid then came to the
fore, comprising;

• Purchase price of £10 (never actually paid)

• £500m in loans and stock from BMW (including £427m loan), repayable by 2049
or if Rover was sold (£275m immediately, £225m within 3 years)
• Production of 200,000 units a year
• Redundancies of 2,000 personnel (from 8,800) at Longbridge

Under the private ownership of PVH, MG Rover consolidated all functions at

Longbridge. Rover’s R&D facilities were split with Land Rover though it was claimed
that around 600 engineers were retained by MG Rover (Automotive News, 2001). The
company inherited the new Rover 75 and the Honda-derived 25 and 45 while MG was
launched as the sports variant of the existing saloon range in addition to the MG-F sports
car. Apart from body panel stamping, which stayed with BMW and so was sourced from
the company’s facility at Swindon, MG Rover had the basis of a full-function structure all
situated at Longbridge. However, scale was another matter and in this respect the
company was substantially uncompetitive within the definition of the prototypical
automobile company which prescribed output of

328

Ik
600.000 units a year and an R&D strength of around 1200 engineers (see section 5.5
above). In comparison to MESp, therefore, MG Rover had R&D at half the necessary
strength and production capacity at a third that deemed necessary by the paradigm. The
company therefore needed to establish a strategy for approximating to the paradigm,
either internally or externally, in a manner sufficiently to ensure its sustainability.

7.7.1 MG Rover internal approximations - literary sources

Intemal approximations to the automobile industry paradigm include two possible

approaches. The first is to extend the model production runs and so achieve high levels
of capital use over a longer period (see section 6.2.1 above). As has been shown (see
section 7.6.3 above), lack of funds had previously forced the company to extend
production runs and so miss out on joining Honda in successive new model
programmes. On independence, MG Rover had no complete new models ready for
release, except for the estate/stationwagon version of the Rover 75 that had been
developed, and cancelled, under BMW. Costs of complete new model development has
been estimated to be up to £lbn (Observer, 2004), although this thesis has argued that
annual R&D costs of continuing new model programmes can be as low as £250m a year
(see section 4.3.5.1 above). Even the lower figure was beyond what the company could
budget for so it was obliged to continue with the models it inherited until it could find a
strategy to introduce models of its own. The contextual sources (see section 7.1 above)
were unanimous in agreeing that extending model production runs is not an appropriate
strategy in a globalised industry where manufacturers are in unfettered competition with
one another, except perhaps when an established platform can be used as the basis for a
new low cost variant (e.g. Renault Logan).

Intensification of the production capacity at Longbridge offered the second alternative

for approximating to the automobile industry paradigm. This would have meant
introducing model variants to serve market niches, raising total output towards the target
of 200,000 a year. By 2001 the company had a full line-up of MG variants of the three
Rover models, although the rugged looking “soft-roader” version of the Rover 25,
Streetwise, did not appear until 2003. In that same year two new models were released,
the MG SV and the City Rover, but since both of these were

329
manufactured elsewhere (Italy and India, respectively) they did nothing to intensify the
use of Longbridge production facilities. Furthermore, despite the introduction of the MG
variants, the slide in production output first experienced under BMW ownership
continued with PVH (2000-2005). Once again in its history the company suffered the
effects of a strong currency and lack of global purchasing power (Financial Times
2002b), offset slightly by home market sales buoyed by a strong domestic economy.
Table 7.12 shows the slide in production for complete years imder both ownership
regimes.

Table 7.12 Longbridge production output 1999-2004

Year Production output

1997 364,350
1998 303,800
1999 227,743
2000 174,885
2001 163,144
2002 142,928
2003 136,065
2004 107,583

Source: Holweg and Oliver, 2005

Although data shows that the Longbridge installation was capable of producing the
targeted output of 200,000 units a year (Automotive News, 2001), the fall in production
was so relentless that it is difficult to see how the trend could have been reversed without
entirely new models. Empirical data gathered by this research reveals some debate
within the company concerning the potential for continuing with the existing product
range.

330
 7.7.2 MG Rover internal approximations - contemporary views

Emphasising the age of MG Rover’s models, Mr Takeshi Sumita of erstwhile partner

Honda mentioned that the Rover 25 and 45 were based on the Honda Civic of 1990-98
vintage. Dr Chris Millard (6 January 2004), head of strategy for MG Rover, believed
that obsolete products could find a new lease of life in new markets because they would
appear to be fashionably novel. In any case, he believed that the underlying technology
hardly changes:

“So the ability to move your product around the world and prolong the life of the
vehicles way beyond what you’d normally expect and generate much longer returns”.

However, Mr John Tweedy, head of intemational markets for MG Rover, did not echo
this view. His experience was that foreign markets were becoming more sophisticated:

“We could have said that sort of thing about China a few years ago, that they were
happy to take cast-offs, but the market is evolving, they know what the latest models
are. In many cases the latest generation are all built in China.”

Mr Tweedy put greater emphasis on organising production to minimise costs, claiming

that with 6,000 employees the company had the same output as a so-called lean operation
with 60,000 employees. Mr Rob Oldaker, in a letter (24 November 2004), similarly
claimed that the company had achieved remarkable efficiency in production although in
preparing this thesis little independent evidence was found to support these views.
However, in an interview with Automotive Engineer (2004), Nigel Taylor, programme
manager large cars, described how production of the new V8 vehicle platform was
included in the existing large car manufacturing system, without any need for batch
building, by the simple provision of sub-assembly stations. In an interesting comment to
this research Mr Rob Oldaker (6 January 2004) noted that flexibility can also be a
problem, encouraging a short term perspective. The company’s own V8 platform might
be a good example of this, being a new platform but having a planned production of only
1000 a year and only for as long as the original Rover 75 platform remained in
production.

331
This thesis has noted that flexible production allows a company to diversify the risk of
production variations by rescheduling production to different models. Although MG
Rover had a flexible production system, with all of its models ageing in the market there
was no possibility of switching production to new or more popular models to
compensate. Evidence relating to contextual companies has shown that flexible
manufacturing is only of use when the model range in sum can fill the available
production capacity, but it does nothing to correct for a net decline in sales. Furthermore,
MG Rover was not in a position to extend the product life cycle due to the largely
undifferentiated nature of the models, increasing sophistication of foreign markets and
the need to develop products in line with new legislation. This research could find little
in the company’s production facility that would compensate for its lack of competitivity
within the automobile paradigm. Possible external sources of advantage were therefore
investigated, firstly with regards to globalisation.

7.7.3 MG Rover and globalisation - empirical data

Chapter 6 of this thesis discussed the opportunities presented by globalisation that might
be available to an automobile manufacturer. However, results from interviews conducted
within the contextual industry, and discussed earlier in this chapter, show that full-
function firms that are competitive within the automobile industry paradigm (i.e.
production output above MESp) only exploit global opportunities for incremental
benefit by way of extensions from an established structure, thus maintaining the
competitive status quo within the industry. MG Rover was sub-optimal within the
automobile paradigm and therefore required to make a quantum change in the way it
exploited opportunities. In particular, this thesis argued that intemational factor
endowment differentials represented potential for an intemational restmcturing of the
company.

MG Rover made an interesting case with regards to global stmcturing because its
intemational development had been put into reverse when it was relinquished by BMW
in 2000. Many of the respondents had experience of the company’s global stretch in the
days of BL and Rover Group so were able to draw contrasting examples with MG Rover.
Table 7.13 summarises how a global structuring brought advantages and disadvantages
to selected functions. For example, Mr John Bacchus and Mr Peter

332
Woods observed that Rover Group had found superior R&D and production facilities in
Japan. Mr Woods also pointed out that exposure to Japanese methods created pressure to
improve on the home-based British operations, and this can be related to the inclusion of
factor efficiency by Schott (2003) in factor endowment differentiation.

However, as noted by intemational theorists such as Hymer, there are costs involved in
dealing with foreign markets and Mr Kiff emphasised this view (see Table 7.13). Mr
Bacchus noted the increase in product complexity in trying to serve those new markets.
He believed that when the production facilities were wholly owned it created general
overcapacity by duplicating existing facilities elsewhere, most notably at home. All the
respondents noted problems with coordination at a global level and Mr Woods pointed to
the long delivery times of components. It was therefore Mr Tweedy’s view that interim
facilities were inadequate, the best long-term solution being a fully committed
production facility as part of a coordinated global stmcture. However, as this thesis has
argued, building new plants that met the requirement of MBPS meant expanding in units
of MEPS, and this is a strategy that tends to only be accessible to larger firms producing
higher than MESp.

Table 7.13 Elements of global industry - strengths and weaknesses

Data Source Unit of Analysis Strength Weakness

Mr J Kiff Japanese production Transferable
Overseas production Market proximity Hidden costs

Mr P Woods Rover production in Farmer in situ Long delivery time

Japan Assembly problems
Pressure to improve High cost
Japanese market Product complexity
Mr J Bacchus Intemational sales
Overseas production Overcapacity
Mr J Tweedy Overseas production High volume only

When BMW sold MG Rover in 2000 the British company lost its intemational stmcture.
As this was being rebuilt it provided the research with a unique insight into how
intemational operations can be started without antecedents, perhaps indicating how a
novel approach might arise. Sources at MG Rover were asked to give more detail on the
rebuilding of intemational sales networks. Table 7.14 illustrates the

333
various options open to the company for meeting the challenges of global markets,
including some historical data from pre-independence.

Table 7.14 MG Rover- market characteristics and strategic responses

Data Source Market Characteristics Strategic Options and Responses

Mr J Tweedy Loss of BMW sales network
Multifranchise
Low volume sales Niche product
New market search Largest markets first
Closed market Local partner
Mr J Kiff Market opportunism Current models only
Market entry Good product
Dealer volume Effective marketing
Comprehensive dealer coverage
New small car
Market knowledge Balance customer research and
Dr C Millard Captive import (CityRover) management whim
Dealer reputation more important
image
Market opportunism Prolong obsolete model life cycle
Global diversity Profitable niches
Mr J Bacchus Overseas market resistance Adapt product
(Rover SD l)
Mr R Oldaker Low volume (Rover V8) Niche only
Large brand portfolio Strategic confusion

MG Rover representatives seemed to recognise and act upon the wide array of marketing
options available, though this multifaceted response did not seem to result in
intemational sales of a sufficient volume in the time available. Mr Tweedy, as the
executive responsible for intemational markets outside Westem Europe, was a cmcial
source of information. With low sales volumes and the loss of access to BMW’s dealer
network, MG was able to maintain a profitable presence with low sales of minimally
adapted products, just as the contextual sources had described previously regarding Saab.
To an extent this provided a response to Mr John K iffs declaration that new market entry
required large scale commitment to marketing campaigns and comprehensive dealer
coverage. Mr Bacchus showed how the Rover SDl had failed due to its hatchback design
but, like Saab, it can be inferred that it might have enjoyed good retums as a niche
product. Clearly the last MG Rover products, such as the new V8, fell into this category.
This also indicated how firms with output at MESp, much lower than the industry
leaders, could nevertheless find a profitable market share.

334
The problem with MG Rover’s approach to intemational markets was that there appeared
to be little in the way of a coordinated strategy. As Mr Tweedy declared, it was a case of
“going for the low hanging fruit”. While this might equate to the Stage 1 of the Uppsala
intemationalisation model there was no suggestion that this was part of a planned series
of steps. As Mr Oldaker stated in reference to the Indian sourced CityRover, short-term
opportunism seemed to be the guiding principle. This thesis has looked at intemational
factor endowment differentiation as an opportunity for a restmcturing and in order to
reveal where factor endowment differentiation might occur the interviewees were asked
to categorise the level of intemationalisation appropriate for specific company functions.
A summary of the results is shown in Table 7.15. For the respondents at MG Rover,
reduction of production costs represented the main attraction for operating foreign
facilities, although Dr Millard mentioned that specific technologies might also be
sourced intemationally. None of the respondents detailed exactly where the competencies
they mentioned might be found, indicating that they were not ready to propose an
intemational restructuring of the company.

Table 7.15 Differentiation of global resources

Data Source Source of Competency

UK Base Global
Mr M Harbour Design Low cost production
Brand sensitive Low value products
Dr C Millard High value products
Design skills Production skills
Marketing
Export production for niches
Knowledge Learning
Equipment
Technology
Mr J Tweedy Export production for niches High volume production
Mr J Kiff Powertrain

Instead of seeking new opportunities that would have a strategic impact MG Rover
seemed to be attempting to exploit its existing stmcture. This indicated that the company
had a similar strategic view of global opportunities as the industry leaders. However, MG
Rover was short of MESp and so it was cmcial that the company should conduct a
fundamental review of its stmcture and how this related to global opportunities.
Commentators inside and outside the firm generally agreed that MG Rover needed a
partner in order to find long-term sustainability. Although OLI

335
paradigm is less powerful as a tool for formulating a strategy it can be used to describe
capabilities and how these might link with external resources. The next subsection of
this thesis will therefore evaluate how MG Rover might have accessed additional
capabilities through a collaborative relationship.

7.7.5 Collaboration —MG Rover literary sources

According to Bailey (2003), under the ownership of BMW the factories at Cowley
(Rover 75, from 2000 the MINI) and Solihull (Land Rover) received investment in
preference to the Longbridge plant, which was also manufacturing the original Mini.
Longbridge facilities improved slightly when the Rover 75 production system was
moved to the Longbridge site and the old Mini phased out in 2000. However, Bailey
concluded that the MG Rover models were ageing, requiring new model development,
either independently by MG Rover or with a partner. The alternative was a reduction in
output to small-scale, niche sports car production.

“For this reason, many industry experts do not believe that MG Rover has a viable long-
term future” (Bailey, 2003, p.73).

The most pressing need, therefore, was to invest in R&D for a new model range and the
only funds available were the £427m interest free loan from BMW, repayable by 2049,
along with the estimated £3 50m value of the plant and stock (DTI/Auto Industry, 2007).
This research suggests that this might have been enough to inaugurate an R&D
programme for a few years but not enough to replace the entire product range (see
section 7.7.1 above).

Upon independence in 2000, MG Rover had embarked on an immediate search for an

international partner. John Towers, the new CEO, openly discussed Chrysler and Honda
as possibilities (Automotive News, 2000), though these suggestions never rose above
mere speculation. As production and break-even targets were missed the company
announced a collaboration with a division of China Brilliance in March 2002 to develop
a new small car. Originally billed as central to the future of the company, by the time the
deal started to go awry just months later the criticality of the deal was being
downplayed. Concurrent deals were being sought with Tata in India

336
for a possible small car and the Daewoo production facility in Poland as a possible
bridgehead for Rover’s entry into Eastern Europe and production base for export
(Automotive Engineer, 2004). Professional Engineering (2003) reported optimistically
on the release of the CityRover, based on Tata’s Indica, and the Streetwise, based on the
Rover 25 (p.27):

“So if MG Rover is serious about wanting to grow, it may eventually have to attract and
cede its autonomy to a major player. For the moment, it’s all to play for.”

Automotive Engineer (2004b) identified a joint venture with Proton as a possible

candidate, but like the Tata relationship and the initial collaboration with Honda, this
involved the supply of an existing Proton model (Wira) and using it as the basis of a new
Rover. An attractive characteristic of the proposed alliance for MG Rover was that the
two companies operated in complementary markets. The collaboration would have
replaced the failed proposal for a deal with China Brilliance, which ended in 2003 with
the imprisonment of the Chinese company’s sponsor of the deal, but at least it did raise
some consultancy fees for MG Rover (DTI/Auto Industry, 2007). It was also mooted
that since Proton was the parent company of Lotus, the sports car manufacturer and
engineering consultant, that the two British companies could work together in the kind
of link alliance described by Dussauge et al. (2004).

Another proposition known at the time (Automotive Engineer, 2004b) was for MG
Rover to take over the ex-Daewoo factory in Poland and install it with obsolete Rover 45
production hardware for sale of the model in East Europe. This also fell through in 2003
when MG Rover’s bid was rejected (Professional Engineering, 2003), and in any case,
there was no reason to expect the Polish market to be receptive to an ageing design.
Although the proposal suggested that the firm understood it needed to rapidly increase
output capacity, and another plant could have taken it close to MESp, a jump in output
volume of this magnitude would have been very risky for a firm of this size. This is a
problem inherent to small companies: they need to expand output to achieve MESp, but
have to do so in steps that are each below MEPS. Other rumoured joint ventures
included licensing of engine production in India, advising Sonalika on the design of an
off roader and the possibility of licensed assembly of the Rover 75 in a remote comer of
Iran. None of these were of any significance, and indeed the Iranian

337
deal appeared to be nothing more than self-promotion by the politician who oversaw the
region.

The final possibility for an alliance came with Shanghai Automotive Industry
Corporation (SAIC) in 2004, along with Nanjing Automobile Corporation (NAC) in a
minority role. SAIC was in production JVs with VW and GM in China but had no
vehicle designs of its own, although it had just acquired the Korean SUV manufacturer
Ssangyong (DTI/Auto Industry, 2007). The Chinese company paid MG Rover £67m for
the property rights to the Rover 25 and 75 along with the K Series engines (Holweg and
Oliver, 2005) but was reportedly annoyed by speculation that it was about to invest £lbn
in MG Rover (DTI/Auto Industry, 2007), a figure that concords with some estimated
costs of new model development and this thesis argues would have sustained R&D for
four years. The proposed deal foundered in March 2005 concurrent with the financial
collapse of MG Rover. Holweg and Oliver concluded that SAIC’s intention had been to
gain access to MG Rover’s intellectual property rights (IPR) and production equipment
for use in its Chinese plants, which it would be better able to do if it bought them
outright. However, against global competition, SAIC would have had the same pressing
need to replace the model range as did MG Rover before it.

Holweg and Oliver (2005) concluded that MG Rover started out in 2000 with an ageing
product line that was suffering declining sales in a shrinking market segment and so
lacked the incoming funds to invest in new models. This made the need for a JV partner
critical to its long-term survival, and due to the production overcapacity in the West only
partners in rapidly growing developing markets, such as China and India, would be
suitable. However, it is the view of this thesis that Holweg and Oliver, like many others,
take too narrow a view of MG Rover as a source for partners of automobile models.
They demonstrate MG Rover’s need for a partner to bring salvation but do this so
comprehensively that they leave no reason why a partner should be interested. With MG
Rover apparently inadequate in every major area, from the model line-up to R&D
investment, there did not appear to be anything that MG Rover could offer an interested
party. On this basis the company was only worth the value of the land it stood on, much
of which had been sold a year before, and the value of its more general purpose
production equipment. Certainly, after a pessimistic

338
assessment of MG Rover’s product range, it is difficult to see how any prospect would
be attracted to the company’s IPR. It is the view of this thesis that the company did
indeed have value as a contributing partner in an automobile manufacturing alliance, but
only if it drastically reduced the scope of its activities.

Lack of scale in all its functions meant that MG Rover had little to offer in terms of
production capacity, and indeed in the one partnership that it did form, with Tata,
production took place in India even while Longbridge was operating well below
capacity. Furthermore, the available capacity at Longbridge was relatively low at a
declared 200,000 units a year, competitively close to MEPS for one plant but a third of
MESp, so a scale alliance would have necessitated investment in new facilities and a
tripling in market demand. However, R&D was only half of MEPS for the process and
this could have been more rapidly expanded than could production output since the core
team of engineers was already in place. A link alliance, therefore, would have shown
more promise, but only related to the transfer of knowledge conceming new models, not
the existing range. A link alliance in R&D for MG Rover would have served the purpose
of intensifying its R&D activity and thus improving output in that function.

In the case of MG Rover it is not possible to make a direct comparison with the figures
for R&D since it did not release any new core models, although it did release a number
of interesting variants of the range it inherited from Rover Group when it was sold by
BMW in 2000. In line with Table 4.8 shown previously. Table 7.16 presents the R&D
results for the company as surveyed by the DTI for the periods 2002/3, 2003/4 and
2004/5 (DTI 2003; 2004; 2005). The peak period, 2003/4, was boosted by additional
funding due to the evolving alliance with SAIC of China, while 2004/5 was cut short by
the financial collapse at MG Rover. Each of the three years is therefore a special case in
its own right; nevertheless the disparity between R&D expenditure at the MG Rover and
the rest of the industry is of a magnitude so great that comparisons can still be drawn.

339
 Table 7.16 MG Rover R&D expenditure, 2003-5
Year R&D R&D R&D

Expenditure Expenditure % of Expenditure/employee

£m sales £
2002/3 45.5 2.7 6,800
2003/4 90.5 5.2 13,700
2004/5 23.9 1.4 3,700

Sources: DTI, 2003; 2004;2005

If 2003/4 is taken as indicative of a period during which new model R&D was
progressing at an acceptable pace then it is instructive to compare this with other car
firms at the same time. Table 7.17 does this for the same companies that were shown in
Table 4.8 (period 2005/6). While it shows that MG Rover R&D as related to sales and
employee numbers was within the industry standards it only shows the rate of
investment; it is the measure of the fixed cost requirement for new model programmes
that is more interesting. As was previously shown in Table 4.8 there is a great deal of
disparity in absolute levels of R&D spending. Ford and Toyota’s R&D budget for that
year was up to forty times larger than MG Rover’s. These are the industry leaders in
terms of total output and have much wider product ranges to develop, as well as a certain
amount of pure research outside of the automobile paradigm as it is defined here (e.g.
fuel cell research is unrelated to the Powertrain function in the paradigm). Subaru was
identified in Chapter 4 as indicative of MEPS for the R&D function and it can be seen
that MG Rover was investing about a third of Subaru’s financial commitment for the
same year. Although this was well below the minimum required, around £250m
according to this research, in absolute terms this amounted to a requirement for around
another £160m; compared to other firms in the industry, this was not a large sum.

340
 Table 7.17 Global R&D Expenditure 2003/4
Company R&D R&D R&D

Expenditure Expenditure % of Expenditure/employee

Ford £m sales £
4189.7 4.5 12,800
Toyota 3484.0 4.3 13,200
Subaru 313.3 4.4 11,400
MG Rover 90.5 5.2 13,700

Source: DTI, 2004

Although by industry norms MG Rover’s R&D staffing level compared poorly with
output (see Table 7.18 below) the company enjoyed heritage benefits from the
investment made by BMW and had also shown some ability to explore new projects
(see Table 4.14 above). The proposed joint venture with SAIC would have resulted in an
effective increase in output by gaining access to the Chinese plant and market
(DTl/Auto Industry, 2007) supported by an increase in R&D engineering staff. The staff
numbers would have then been sufficient for MEPS in R&D, while production output of
Im units a year would have exceeded that which this thesis has argued would be
necessary for MESp.

Table 7.18 MG Rover R&D productivity

Company R&D Staff Output Ratio
(millions) Staff: Output
MG Rover (target 500 0.20 1 :400
output) 500 0.11 1 :216
MG Rover (2004
actual output) 700 1.00 1 :1429
MG Rover SAIC
(combined target
output)

Source: Financial Times 2005g

Automotive News 2006b
JAMA 2005a
These speculative figures taken from media sources in Table 7.18 for MG Rover
indicate the kind of productivity that was available in R&D. In such a scenario, MG

341
Rover would use its product development function to support the brand and would look
for a partner that was devoid of a product brand but could offer substantial production
capabilities. This would have resulted in a link style alliance, but one where any ex post
advantages in R&D would only accrue directly to MG Rover if it had primary
responsibility in that function. This would only be possible if the partner had little or no
R&D capability of its own but was competitive within the industry in terms of
production scale, a scenario that describes a number of emerging economy
manufacturers. However, it also implies that significant production at Longbridge would
have had to cease in favour of production facilities operated by the partner. With
advantages in the functions accruing to the partner responsible, long-term stability could
have been achieved. Furthermore, each partner would continue to exploit the
international factor differentiation advantages of its home region. Figure 7.3 depicts
how the resultant IVJV might have looked with SAIC.

Figure 7.3 Proposed IVJV between MG Rover and SAIC

MG Rover
R&D
(UK)

r SAIC

BIW Powertrain

Final
Assembly

Global
Market

The suggested closure of Longbridge as a production site recognises that an

organisational joint venture structure of this kind is not sustainable between companies
that operate existing full-function structures. This precludes producers

342
such as Proton or Tata that had attractive production capacity but also had product
strategies of their own (Automotive Engineer, 2004). It is in accordance with this view
that the purported alliances with these companies failed to develop.

Although the IVJV seems to have a solid theoretical basis a significant problem
concems a suitable governance structure. This thesis has found that a development of the
U-Form of governance structure has tended to prevail over the M-Form, particularly
amongst Japanese firms, but such a centralised form of management control would not
be applicable between independent parties. The example of GM and A O Smith (see
6.5.1 above) suggested that the M-Form might be more appropriate since it describes the
separation of strategy from operations. However, the IVJV is missing two crucial
features: the transfer of objective data for strategic analysis, and the existence of a
designated headquarters. Neither of these factors can be taken for granted between
independent parties without being explicitly implemented. The following sub-section
will investigate the empirical evidence for the potential of the IVJV relationship,
including data gathered from Chinese manufacturers NAC and SAIC after the collapse
of MG Rover, in order to gain a closer understanding of the organisational and
governance structures.

7.7.6 Collaboration —MG Rover empirical data

Although MG Rover exhibited the organisational structure of the full-function model it

lacked the scale in production signified by MBS? to be competitive within the precepts
of the paradigm. Other manufacturers in a similar predicament, such as Aston Martin and
Bentley, had larger parent groups to mask the cost disadvantages. MG Rover was
therefore exposed to all the cost disadvantages of being sub-optimal within the paradigm.
The company attempted to exploit its economies of scope in management by operating
an additional production site in Poland in the former Daewoo plant, as well as other
rumoured collaborations, but it lacked the resources to make the required strategic
commitment. In any case, even increasing the economies of scope in the product range
with the creation of a parallel MG brand was insufficient to fill production capacity at its
one existing plant in Longbridge. The remaining option was to find an externally based
approximation of the automobile paradigm and this meant collaborating with one or
more other manufacturers in a

343
suitable alliance. Although this thesis has put forward the IVJV concept this does not
mean that MG Rover had reached the same conclusion. Table 7.19 summarises the
search for a partner from the perspective of MG Rover, including the assets that were
being looked for and the outcome.

Table 7.19 MG Rover - search for a partner

Data Partner Assets Future Potential

Source MG Rover Partner
Dr Various Powertrain Technical
Millard R&D efficiency Financial resources dependency
MrR Large Long-term
Brand development Common components
Oldaker group New product
Tata Technology Premature end
Large R&D efficiency Production capacity
group
Mr J Various Product IPR Premature end
Tweedy Market access
Uncompetitive size

The table shows a great deal of speculation on the nature of the partnership that MG
Rover required and the competencies it could offer. Dr Millard, head of strategy for MG
Rover, asserted that a problem with looking to other companies for input, such as
powertrain supplies, was that it was necessary to accept their specifications and policies.
Dr Millard felt MG Rover lacked only financial resources, a difficulty that has been
commented on elsewhere, but claimed that the quantity of financial investment
necessary in 2004 was not as great as it had been previously. This seemed to imply that
cost recovery was not as onerous as many critics of the company believed, although of
course the company was at no point close to achieving its output target. It also suggested
a lack of urgency to the search for a partner. In the event this proved to be an
inappropriate position, although the company did have capabilities to offer a potential
partner.

There was some evidence that the company was effective in its R&D function and Dr
Millard noted that the costs of independent design had come down significantly. He gave
the example of crash test analysis using Linux clusters, which two years previously
would have taken up to three days, but could now be completed in three hours using
software costing less than £100,000. Dr Millard felt that the experience of

344
the engineers was more important than the software system. This human asset was
difficult to purchase yet accessible to MG Rover because previous partnerships had
sustained the knowledge base of the company. This resulted in what Dr Millard believed
was a small, elite R&D team at MG Rover:

“You could not hire 300 engineers and start to do what we do out there [in the R&D
centre]; our inventory goes home at night. If we didn’t carry the breadth of skill that
we’ve got, people who have delivered programmes with Honda, people who have
delivered programmes with BMW... [then we would not be] able to continue to iterate
and move those programmes on as ever they are.”

This view gave empirical support to the findings of this research that suggested R&D
costs for a basic model range are not as high as many industry commentators believe.
Mr Oldaker envisaged that R&D should be arranged like a Formula 1 team, a small
team of experts, “difficult people to manage” but dedicated to the end result. Although
he was not as confident as other respondents on the long-term viability of MG Rover as
an independent manufacturer, he believed that the company needed a larger partner that
would acknowledge MG Rover’s potential in R&D. He seemed to think that the
company could sustain autonomy in that area at least.

The last partnership entered into by MG Rover before the financial collapse in April
2005, that with Tata of India, illustrates the inability of the company to build on its
strengths or correct its weaknesses. It might have been expected that the alliance with
Tata would have either exploited MG Rover’s R&D capabilities, or filled vacant
production capacity at Longbridge. Instead, it brought a new product to Rover that was
neither designed nor manufactured with MG Rover as a partner. As Mr Rob Ball stated,
nothing was transferred to Tata although MG Rover did provide some advice in order to
“Roverise” the car for the UK market. Dr Chris Millard also described the relationship as
“piggy-backing” on Tata’s operation. Although the additional volume was of incremental
benefit to Tata it had no bearing on its decision to manufacture, the vehicle having been
in production as the Indica since 1998 and being taken up by MG Rover only in 2003. It
cannot, therefore, be considered a link or a scale alliance under the theory of Dussauge
et al. (2004) and was more reminiscent of a franchise type agreement, albeit without the
Tata brand.

345
Despite the inappropriate structuring of the relationship with Tata, according to Mr
Tweedy there was no shortage of potential partners. In order to discover what MG
Rover believed it could offer potential partners interviewees were asked to describe the
range of capabilities at MG Rover. Dr Millard reiterated MG Rover’s attraction as a
non-competing partner:

“It wouldn’t worry me if a [Rover] 75 was built on the other side of the world by
someone who shared in its development in the future. It wouldn’t really be damaging to
our market position in this country and the territories we compete in.”

Mr Tweedy elaborated further MG Rover’s attraction to a potential partner: attractive

products and market access combined in a company of unthreatening size. He reported
that the company was receiving “plenty” of invitations from companies interesting in
collaborating and that MG Rover had a team, headed by Mr Dave Ruffell, specialising
in evaluating them:

“So, by a very logical process, we find that we are at the end of the telephone for a
number of people that are looking for a company that is not too big, not already
established and offer them a lot of potential.”

Despite this, the company did not seem to fully comprehend how an alliance might serve
its interests. Mr Tweedy categorically denied that MG Rover was taking a systematic
approach to collaborations, instead apparently entertaining offers as they came in. He
mentioned Iran as a large but closed market that would require a local partner in order to
penetrate it. He implicitly acknowledged the existence of an automobile industry
paradigm when he pointed out that most full-function compeinies, as defined by this
research, would be of a much larger size than MG Rover and already operating
internationally. They would, therefore, already be in competition with any potential
partners they might be looking for. This again highlighted the short-term character of
link alliances, ex post asymmetrical advantage being exploited at the expense of the
weaker partner. He did not rule out forming a network of alliances if the
“incompatibilities” could be worked out. The comments on the impact of joint ventures,
particularly pertaining to MG Rover, are summarised in Table 7.20 below.

346
 Table 7.20 MG Rover - Impact of joint ventures

Data Partnership Im pact

Source Immediate C ontinuing
Mr J Kiff MG Rover - Honda and Product development
BMW
DrC MG Rover - Honda and Established base for
Millard BMW Drive initial independence
MG Rover - future None: internal resources
partner improvement Premature end
MG R over-TW R New product
Rover - Honda Honda controlled
MrR MG Rover - Honda (Legend/800) specification
Digital design Development integration
Oldaker MG R over-TW R
MG Rover - ProDrive MG Rover initiate and
MG Rover - V8 conclude Not so cost effective
New engine
MrR partners None: learnt little from
Honda - Rover
Ball International sales Rover
Mr J MG Rover - BMW Loss o f network
Tweedy MG Rover - Tata New product Body pressings
Premature end
MG Rover - Suppliers Cost reduction Supplier innovation

The table shows that the immediate effect of various relationships tended to be financial
or to fill a gap in the product range. However, Dr Millard, head of strategy at MG Rover,
was explicit that in the long-term the core resources should be intemalised, particularly
with regards to coordinating and integrating vehicle development projects. MG Rover
had engaged external companies to work on new projects: TWR on the new medium car
(RDX60) and ProDrive on the rear-wheel drive Rover 75 V8. In both cases, MG Rover
had project control problems: firstly with TWR when the company collapsed, taking
much development work with it, and secondly with ProDrive in monitoring the progress
of the work. This is an oft mentioned weakness of link alliances, one partner taking
advantage at the expense of the other.

There was no mention of the company having an interest in a scale alliance where it
shared output with another manufacturer. The general consensus at MG Rover was that
the company had core capabilities that should be defended as an independent concern
and, although assistance was needed in the form of fimds and temporary new

347
products, the vertical relationship with suppliers was the only likely long-term
partnership for the company as it was then structured. There was no apparent intention
to restructure the company, either globally or functionally, and in this regard it shared
the same strategic outlook as the contextual industry. Perhaps most extraordinarily, 18
months after the collapse of the company, Mr Stewart McKee (17 July 2006), ex-PR
representative of MG Rover, stated that the company’s chief executive, Mr Kevin Howe,
had managed to convince the leadership of PVH that the company could be independent
in the medium-term and so there was no immediate need to finalise a JV. This was also
the impression gained from many of the other interviews.

It can be inferred from Mr McKee’s comments that the senior management of MG

Rover were unaware of the unique capabilities at the firm and the opportunities that
existed externally. The company needed to bolster its production scale while exploiting
its economies of scope in R&D. This naturally indicated that it needed to find a partner
that could provide additional production output in exchange for MG Rover’s R&D
input. As the link-scale alliance theory has suggested, and the company found in its own
experience, simply forming a relationship with another company was not enough.

Perversely, the relationships it formed tended to pass the benefits to the partner; TWR
and ProDrive exploited their ovm capabilities in R&D while Tata gained scale benefits in
production, both of these while capacity in these functions at Longbridge was underused.
Furthermore, despite its own experience, the company did not structure the relationships
in such a way to head-off ex post problems, the demise of TWR leading to a defensive
internalisation of R&D rather than a restructuring along functional lines with a new
partner. If the senior management understood international factor endowment
differentiation at its most basic level (i.e. low labour cost locations) then its exploration
of production in Poland showed that there was no commitment to restructuring the
company around the opportunity. Sourcing production of the CityRover small car in
India was not reflected in a lower price in the UK retail market but instead the price was
maintained at UK market levels as a tactic for improving cost recovery. Since the
product’s provenance was well-known in the UK market the pricing strategy
contradicted consumer expectations of a relatively low-priced product and thus sales
were far below company projections.

348
Although the company was active in pursuing different opportunities without a
systematic approach these explorations could not result in a unified strategic direction.
The result was therefore of a company that was static and defensive when only a
dynamic response could have brought it into a position of sustainability. Burdened with
an uncompetitive cost structure MG Rover was forced to consume its internal resources
simply to sustain itself. Over the period of PVH ownership of Longbridge, 2000-2005,
with output falling the available cash was used for operational needs rather than future
investment.

A lack of cash is not the cause of a company’s collapse but the result, so in the case of
MG Rover the conclusion must be that, given its unsustainable form, time was of the
essence in the search for a joint venture partner. This thesis has argued that MG Rover’s
R&D alone demonstrated a relative advantage, albeit still requiring significant
additional resources, so the only real potential for the company was as the provider of
R&D knowledge in a VJV where its role would be protected from ex post appropriation
by the partner. Instead, though, the MG Rover management seemed to prevaricate on
what the terms of a JV should be. Part of this problem would have concemed
governance; effectively, which party would determine overall strategy. With SAIC both
wealthier than MG Rover and closer to the largest market it could be anticipated that the
Chinese company would dictate future model programmes.

Such considerations were ultimately irrelevant. The subsequent collapse of the company
laid out the assets of the company for interested parties to choose from almost without
hindrance, thus reconstructing on their own terms the form of their preferred JV.
Nevertheless, the logic of the VJV is designed to bring benefits to both parties so the
VJV structure may still emerge even when all negotiating power has passed to one party.
It does, though, suggest that the resultant structure will be a variant of the VJV structure
that would have occurred between two independent parties. The following sub-section
will evaluate these considerations using literary sources and empirical data.

349
7.8 MG Rover assets - reconstructing the international vertical joint venture

In April 2005, PricewaterhouseCoopers acted as official receivers to dispose of the

assets of MG Rover. By July that year the process was largely complete, with the main
physical assets and the MG brand being sold to NAC, previously the minor party in the
proposed JV with SAIC. The lack of production locational advantages in Longbridge
were confirmed when NAC proceeded to extract the production equipment and move it
to a new plant in China, re-starting production in March 2007. Although NAC had the
physical means of production it did not own the product designs, which had been
previously secured by SAIC during negotiations for its aborted IJV with MG Rover.
NAC therefore had to reverse engineer the products in order to derive the technical
designs. NAC found ownership advantages in the MG brand while SAIC attempted to
similarly obtain the Rover brand name but without success as Ford took up on option to
purchase the brand from BMW in September 2006.

The division of MG Rover between NAC and SAIC took in some of the elements of the
OLI paradigm. The total investment by NAC and SAIC in acquiring MG Rover’s assets
in the UK amounted to £110m (DTI/Auto Industry, 2007), so the companies clearly saw
value in MG Rover. As owner of the physical assets, NAC seemed to perceive few
locational advantages to the UK except for maintaining brand equity, which meant
retaining a UK connection in the form of limited sports car production in order to
provide some authenticity to the British brand image of MG (Times, 2007). At a press
conference and online conference interview attended during the course of this research,
NAC made the MG brand ownership advantages very clear but the locational advantages
of the UK were more equivocal. Like the contextual companies, NAC gave first priority
to achieving production scale by bringing its primary plant in China to full capacity
production. Paul Stowe, head of quality for NAC, confirmed in an internet based
interview in which the author of this research participated, that the panels for the TF
would be supplied from NAC in China. Stadco, the former panel supplier for that model,
had also announced its involvement (Stadco, 2006). The extent of Stadco’s involvement
is not clear, but it does suggest that the majority of the BIW production would take place
in China rather than Longbridge.

350
SAIC and NAC also made some moves to divide MG Rover along human and physical
asset lines. In buying the physical assets of MG Rover, NAC did not initially pursue the
human assets and a senior manager of the Chinese company described how the company
had acquired at Longbridge the physical R&D capability to develop entire new models
but without the manpower that had existed before. He confirmed that R&D would be
centred on China, the role of a diminished Longbridge R&D being to “add a European
flavour” and a sense of the MG heritage to future models. The continuing purpose of
R&D in the UK was to “complement the heritage of the MG brand” and “understand the
talents that are endemic in the UK industry” but without making a full commitment to
UK R&D.

Indeed, NAC’s strategy seemed to be to recreate the Longbridge operation,

encompassing R&D and production capacity, at a new site in China. This would have left
the original Longbridge plant as nothing more than the European branch of the
operations centred on China. The result would have been a domestic operation in
accordance with the full-flinction model of the automobile industry paradigm but with
the same problems that MG Rover had suffered at Longbridge. This meant a lack of
scale in production, an ageing product range and poor exploitation of intemational factor
endowment differentiation by relinquishing access to MG Rover’s R&D capability.
However, in a subsequent email (28 March 2007) the senior NAC executive confirmed
that the plan had changed and that the company had decided to elevate Longbridge R&D
capability so that it would be the source of all new model strategies. R&D in China
would then focus on adaptations for the local market. To enhance continuity with the
previous MG Rover R&D capability NAC had rehired Mr Rob Oldaker to reprise his
role as the MG Rover head of engineering, carrying the title of Global Product Director,
and he would then lead the expanded R&D team in a partial recreation of the original
MG Rover team.

In contrast, SAIC seem to have been much quicker in approximating to the automobile
industry paradigm by acknowledging earlier the divisibility of functions by asset type
and factor endowment. The company was interested in the human assets of MG Rover
and these seemed to have locational advantages for the company. This was at variance to
what had been expected of SAIC. After MG Rover collapsed in April 2005, Holweg and
Oliver (2005) speculated that SAIC would then be able to

351
acquire the assets that it chose, namely the physical means of production. This is not,
though, what happened.

Having already obtained the rights to various model and engine designs, the company
eschewed a bidding war with NAC for the physical assets and instead rehired many of
the human assets of R&D, i.e. MG Rover’s development engineers, forming the Ricardo
2010 R&D unit under the guidance of Ricardo UK Ltd (DTI/Auto Industry, 2007). Mr
Anthony Smith, an engineer of 21 years with Ricardo and lately the company’s media
consultant, provided an additional perspective for this thesis. He stated that Ricardo UK
Ltd offered the concept of the 2010 unit and SAIC accepted because it provided access
to the human capital behind the IPR (intellectual property rights) formerly existing
within MG Rover. Mr Smith emphasised that IPR is not just the product design but also
the systems and experience that lies behind it.

Mr Smith asserted that SAIC’s involvement in Ricardo 2010 was a low risk approach to
securing technology transfer in the long-term, contrasting this with consultancies that
provide knowledge only for the term of the contract. In its initial stage the Ricardo 2010
unit represented a formal JV between Ricardo and SAIC, and it exploited the
intemational factor endowment differential in R&D that MG Rover had originally
offered SAIC in an IVJV. The 2010 unit recreated most of the MG Rover design team in
order to preserve the organisational capital that resides within the human assets and so
replicate the proposed IVJV. However, a problem for SAIC was that the organisational
capital in fact resided with Ricardo UK Ltd, and as Ricardo UK Ltd is a dedicated R&D
company serving a multitude of automobile manufacturers it meant that Ricardo UK Ltd
could have gained ex post benefits at SAIC’s expense. SAIC was dependent upon
Ricardo 2010 for its R&D, but without the Ricardo UK Ltd in tum committing all its
R&D capabilities to SAIC the relationship could not be defined as a tme VJV. It was
therefore important for SAIC to impose a greater degree of control over the relationship.

The IVJV strategy suggested by this thesis could have been imposed on SAIC and
Ricardo 2010 by demerging Ricardo 2010 from Ricardo UK Ltd. This would have
established Ricardo 2010 as an independent party to the IVJV with SAIC. However, this
does not solve the problem of a suitable govemance stmcture for an IVJV that

352
would formalise strategic control. In this case the solution came in the form of SAIC’s
complete ownership of Ricardo 2010. As suggested by Klein (2000), by taking
ownership of the Ricardo 2010 R&D unit from Ricardo UK Ltd SAIC secured access to
all subsequent ex post benefits which might instead have accrued to Ricardo UK Ltd. It
also gave SAIC control over the R&D strategy, though leaving the division with
operational control. This seems to clearly reassert the M-Form of govemance stmcture,
yet this thesis argues that ownership is not a necessary condition for the M-Form in an
IVJV. Had MG Rover retained its status as an independent party to the JV, then the
question of govemance might have been answered differently. As it was, Ricardo 2010
could never have been independent because SAIC needed to retain links with Ricardo
UK Ltd; SAIC therefore needed complete ownership to offset Ricardo UK Ltd’s close
operational connection with Ricardo 2010. An announcement made in 2007 stated that
the advantages of ownership would still depend on the locational advantages of keeping
the R&D unit in place so the Ricardo 2010 unit would be preserved within the physical
boundary of Ricardo UK Ltd. According to a Ricardo UK Ltd press release (January
2007);

“The transferred business will continue both to be based at offices at the Ricardo
Midlands Technical and have strong operational links with Ricardo UK Ltd.”

Although the engineering team at Ricardo 2010 was continuing the work it had started at
MG Rover the engineers reported to Mr Smith that they had gained the one resource they
had previously lacked: funding. The unit’s task within SAIC was to conduct advanced
engineering for new models, with a workforce of around 150, effectively the same size
as the R&D department at Aston Martin (see section 4.4.1 above). The engineering team
includes not only some SAIC engineers from China but also representatives of Chinese
supplier companies. By coincidence, the unit was operating in the same building that
Ricardo UK Ltd had used in helping BMW finish the engineering on the new MINI,
BMW having lost much of the R&D team when it sold its Rover Group assets.

With production and production engineering taking place exclusively in China, it was
put to Mr Smith at the close of the interview that this replicated the IVJV suggested by
this research; he agreed with this point of view. Just as Grossman and Hart (1986)

353
assert that a firm is one that has operational control of its assets, not necessarily
ownership (see section 7.4 above), the spirit of a VJV relationship between SAIC and
Ricardo 2010 is embodied by maintaining Ricardo 2010 as operationally separate. This
is somewhat reminiscent of the M-Form of organisational structure discussed in Chapter
5 and extends to Ricardo 2010 a measure of operational autonomy. Since the Ricardo
2010 facility exploits intemational factor endowment differentiation Figure 7.4 shows
the Ricardo 2010 R&D unit as operationally separate from SAIC, and therefore in an
IVJV style relationship, even if the equity is owned by SAIC. This stmcture can also be
applied to NAC, with Longbridge R&D taking the place of Ricardo 2010 and NAC
replacing SAIC, although in addition the Longbridge plant would have its own limited
final assembly capability from CKD kits.

Figure 7.4 IVJV style relationship between Ricardo 2010 R&D and SAIC

Ricardo 2010
R&D
(UK)

r SAIC

BIW Powertrain

Final
Assembly

Chinese
Market

With this stmcture SAIC was able to exploit the division between human and physical
assets, as suggested by Dussauge et al. (2004), and an intemational division according to
the factor endowment theory of Schott (2003). NAC, by purchasing mainly the physical
assets and moving them to within close proximity of its R&D facility in China, initially
recreated the full-function stmcture of MG Rover but this time within the confines of the
larger Chinese market. Although this would provide NAC with an

354
entry into the local market, since the range of vehicles has been generally considered to
be uncompetitive in the global market the need to replace them was even more pressing
than it was for MG Rover. Unlike SAIC, NAC had avoided the risk of hold up when
using an external R&D team by simply internalising the capability. While SAIC can
build on the IPR it has purchased using the team that generated it in the first place, it
would have been mostly immaterial whether NAC attempted to update the product range
that was purchased (as indicated to this research by the senior manager) or design an
entire new range; in either scenario NAC would have had to conduct R&D largely by
itself.

Although NAC has recently begun to follow SAIC in replicating the advantages of the
IVJV structure by enhancing the role of Longbridge as a site for R&D, the most recent
events have indicated a consolidation of the two strategies. Under political pressure to
create a Chinese automobile group that is competitive within the global industry SAIC
and NAC (under the auspices of the parent group, Yuejin) have entered into negotiations
to explore collaboration. Although it is not yet clear whether this will constitute a formal
takeover by SAIC, it does suggest that NAC’s strategy of recreating the Longbridge
operation in China was no more sustainable than MG Rover’s original strategy.
Furthermore, ownership of the physical assets of production does not seem to have
conferred any particular advantage on NAC, which has entered the market later than
SAIC with its largely unchanged version of the Rover 75 and has not revealed any
significant product R&D programmes. Only one process was unique in its physical
assets, that of producing the advanced but ill-starred K-Series engine, and the physical
asset specificity in this area is probably attractive to SAIC as it puts its own version of
the engine family into production. Again, though, it is not necessary for SAIC to own the
assets since it could have contracted with NAC to supply the engines until its own engine
family had been brought to the market.

Should the consolidation of SAIC and NAC take place it would have no material effect
on the logic of the structure shown in Figure 7.4 above. The combination of the two
companies would bring benefits in scale as part of the overall trend in consolidation
witnessed in the industry over the decades of the Budd Paradigm but it does not affect
the advantages that are derived from exploiting intemational factor

355
endowment differentials. Furthermore, it would not weaken the advantages of
maintaining semi-autonomous functions within organisational structure, as could be
contained by an M-Form of govemance. In this way, the status of Ricardo 2010,
currently known as the UK Technical Centre, as an operationally separate R&D unit
should remain unaltered and the organisational structure of Figure 7.4 remain intact.

7.9 Conclusion

In Part I of this chapter it was found that the contextual companies were mostly
comprised of full-function multinational enterprises (MNEs) that were either industry
leaders within the automobile industry paradigm and producing above MESp, or were
part of larger groups that were industry leaders. This research has found that they
preserved their ownership advantages over their human assets by retaining the full-
function stmcture originally established within their home base. The govemance
stmcture for this strategy seemed to be provided by the U-Form, with certain adaptations
necessary to include extensions into overseas markets. Although locational advantages
and factor endowment differentiation was recognised, these were subservient to the need
to exploit economies of scale at existing facilities.

Part II of this chapter has shown that MG Rover had the full-function stmcture
prescribed by the automobile paradigm, but it was far from achieving MES? output. This
has been attributed to its ageing model line up leading to inadequate income being
available for investment in future models. MG Rover was therefore exposed to all the
cost disadvantages of being uncompetitive within the paradigm and only lasted as long as
it had funds for its operational needs. While MG Rover had advantages in its high level
of functional integration, as well as brand ownership and control of human assets in
R&D, its problems stemmed largely from disadvantages in production output. The
company had an assembly plant with output capacity that was within a range that many
other companies accepted as being reasonably close to MEPS, but to achieve MESp the
company needed at least one more assembly plant, if not two, and the one it did have
never reached full capacity.

356
Even given its limited advantages, there was no evidence that the company understood
its strengths or how globalisation in its industry offered complementary solutions.
Indeed, all representatives of MG Rover gave the impression that the company was
operating with the same assets and opportunities as the contextual MNEs in the industry
with output above MESp. Despite early statements conceming the search for a partner, it
was only shortly before its financial collapse five years later that a suitable candidate, in
the shape of SAIC, appears to have been identified. However, this was too late to save
MG Rover.

The relationship between MG Rover and SAIC that was being considered conforms to
the IVJV concept put forward in Chapter 6 as a means by which a firm operating
inefficiently could best approximate to the automobile industry paradigm. The
commercial logic of such a relationship carried through beyond the collapse of the
British company. SAIC gained access to the IPR and related human assets of MG Rover
through its involvement in, and subsequent ownership of, the UK based Ricardo 2010
R&D unit. This resulted in almost the same stmcture as the mooted IVJV, dividing the
company functions into human assets (R&D in the UK) and physical assets (production
in China), with additional benefits from factor endowment differentiation. Indeed, it was
only the wider activities of Ricardo UK Ltd as an engineering consultant that prevented
the emergence of a VJV between two fully committed partners, and thus the necessity for
SAIC to take control of Ricardo 2010 in order to safeguard its R&D knowledge base.
The final stmcture is therefore in accordance with the IVJV model but secured by equity
ovmership.

NAC took a more circuitous route than SAIC. As the purchaser of the physical assets of
MG Rover it seems to have initially attempted to recreate the company in a Chinese
setting. Access to a larger market would have brought greater production output and
capacity utilisation, but this was still far below MESp and there was no reason to expect
the production capacity of the physical assets to have increased. NAC put in motion a
plan to maintain Longbridge as an extension of the Chinese operations which would have
ignored any possible advantages in intemational factor endowment differentiation.
Subsequently, though, NAC took largely the same position as SAIC and elevated the role
of the R&D capability in the UK.

357
From this it can be seen that the structure of the IVJV has been largely put into effect
and an approximation made of the automobile industry paradigm, though not in the way
originally envisaged. Rather than MG Rover remaining in existence as an independent
part of the IVJV, the company collapsed, with the IVJV style of structure and benefits
being largely effected by both SAIC and NAC. This has resulted in a full-function
structure but one that continues to recognise the benefits of endowment differentiation
and the maintaining of a division between human and physical assets.

358
 8: Conclusions

So, the purpose of this thesis has been to construct a conceptual framework for the
automobile industry and thereby provide a systematic perspective on the organisational
size and structure of companies that operate within it. In this thesis the resultant model is
known as the automobile industry paradigm and it describes the conceptual limits of an
automobile firm in terms of its most efficient size and structure. This research was built
on a theoretical framework supported by evidence relating to the contextual automobile
industry with the UK’s only mass market automobile manufacturer, MG Rover, as a
specific case study. As a qualitative problem the data was gathered from contemporary
literature and in semi-structured elite interviews with senior actors within the industry.
This permitted comparisons between MG Rover and the contextual companies at each
point as the paradigm was developed in this thesis.

The automobile industry paradigm was constructed from two distinct perspectives: the
first conceming economies of scale, based on the assumption of a common technology
(see Chapter 4), the second conceming organisational stmcture, based on transaction cost
analysis (see Chapter 5). In separating these concepts the thesis was able to isolate
factors relevant to firm size from those relevant to firm stmcture. It was then possible to
estimate the minimum size and stmcture for a sustainable automobile manufacturer,
known in this thesis as a prototypical automobile company. Such a company should
attain minimum efficient scale, or MESp for a prototypical company, and minimum
organisational stmcture, these being the four functions (R&D, Powertrain, BIW and Final
Assembly) of the full-function model. The purpose of this chapter is not to reiterate the
conclusions that were dravm at the end of each section of the thesis but to discuss the
overall meaning of the research for understanding the current state of the industry. This
will lead into discussions on how the industry is likely to develop, implications for
national government policy, lessons that may be applied in other industries and
suggestions for continuation of this research.

359
8.1 Current state of the industry

In order to quantify the importance of scale in the automobile industry the research took
two perspectives: the first, using survivor analysis, looked at the growth of market share
for selected output ranges and the second calculated MESp from the estimates of
minimum plant sizes, MBPS. The first approach was rather inconclusive since even over
a thirty year period there were only indications that the largest output ranges were
continually expanding, it was not possible to determine how much further this output
share would grow. At the same time, survivor analysis showed a second trend in the
output ranges below Im units a year. Although the output share was falling, there were
manufacturers within this trend that were holding on tenaciously, even expanding, even
if at their rates of growth it could be a decade before they rose into the next output
category. Survivor analysis therefore indicated two types of firm, the industry leaders
with output above 2.5m, and then a select number of smaller firms that seemed to
survive at some minimum sustainable level.

The reason for the survivor analysis not being conclusive is that markets in emerging
economies such as China and India are driving automobile sales rapidly higher and so
market shares are still evolving. The alternative approach, calculating MBS? from MBPS
for each function, was conclusive but resulted in a remarkably low output figure. This
was particularly the case with R&D, for which the accepted view is that it can cost up to
one billion pounds for a new model programme, yet this research found that the
continuing costs of R&D could be as low as £250m a year. This figure was derived from
the latest DTI data from on Subaru (the automobile division of Fuji Heavy Industries),
and even if the average figure over a four-year period is taken (2002-2006) this only
rises to £287.5m, with a peak figure of £313.3m in 2003/4. Furthermore, the 2006 DTI
data for Proton in Malaysia, the only year available, showed R&D investment as low as
£53.5m (DTI, 2006), albeit with significant extemal assistance and a smaller model
range.

The relatively low scale necessary for MBSp is symptomatic of the fundamental state of
the industry. Since the production technology has been well established for the past few
decades, at least since Toyota’s TPS system refined manufacturing techniques in the
1970s, then the industry would be expected to have reached a mature stage in its

360
life cycle. This would imply that the basic specification of the products should be fairly
standardised, profit levels expected to have fallen and there would be less product
innovation. The strategy for these firms, then, would be to reduce R&D and production
costs to a minimum and defend market share through product differentiation, perhaps in
market niches. This appears to be Subaru’s position with a limited number of vehicle
platforms underpinning a narrowly defined product range targeted at market niches
(principally four-wheel-drive). The strategy is reflected in the financial results, Subaru
showing a net margin of 2.3% for 2007, not as high as Toyota with 6.8% but still ahead
of Mazda with a net margin of 1.8% and Daihatsu with 1.4% (all figures courtesy of
Global Insight). Without closely analysing the different measures of financial returns it
would seem, then, that operating close to MESp does not result in high profits but at
least the business is generating an acceptable return.

The problem for a prototypical company is in maintaining output of MESp in an

environment of regular models changes. On occasion the failure of one product can bring
the company down, the NSU Ro80 being a good example. Joint ventures can help, a
scale alliance providing access to a production facility that has achieved MEPS but
without the obligation of filling the capacity alone. Link alliances provide similar access
to shared technical innovations but, as this research has demonstrated, there are risks of
ex post opportunism and so these alliances tend to be short lived. The strategy is also
open to the industry leaders and they tend to engage in them on a project-by-project basis
in order to spread the risk of opportunism.

Although MESp represents the minimum cost for the industry it is difficult to stabilise
output at this level so the alternative strategy is to expand output in order to diversify the
risk in output variation. The industry leaders have thereby reduced actual production
costs as low as possible by having output at multiples of MESp. An extensive model
range combined with global markets and flexible production mean that larger companies
have proportionately more stable output. For an industry leader to suffer rising costs due
to output variation of the same magnitude as a prototypical firm it would have to be
inflicted by a multiple demand failure, a statistically less likely occurrence given the
breadth of the model range. Nevertheless, it is a predicament that has impacted most
recently on Ford, resulting in the closure of

361
multiple plants. Interestingly, the output quantity for MESp suggests that Ford could
retrench to its domestic market after divesting itself of Jaguar and Land Rover and
restore its fortunes in the market as a priority. Ford could particularly exploit local
advantage based on its understanding of US buying habits (e.g. US consumer preference
for SUVs).

For automobile manufacturers that are operating well below MESp their first task would
be to raise output to the threshold level. This thesis found that MG Rover was in this
disadvantaged output range and suggested that an IVJV would have created a division in
the full-function structure that avoided ex post opportunism by retuming all the benefits
of a new development to the partner responsible for the function. Opportunism would
therefore be expressed as innovation and of greatest benefit to the firm that developed it.
Additional benefits can also come from international separation such that the contrasting
factor endowments of different regions could be exploited to mutual benefit. It is
therefore not simply a device for salvaging MG Rover’s R&D capability but also a
strategy for raising an assembly partner to a competitive position by providing
immediate access to current product designs.

The weakness with the IVJV organisational structure was in formulating an appropriate
governance structure. Although this research had found little support for the M-Form of
multidivisional control, instead concluding that modem production systems favour the
more centralised U-Form of govemance structure that binds operational and strategic
concems together, the IVJV is founded upon a clear operational separation of functions.
SAIC, and to a lesser extent NAC, partially replicated the IVJV stmcture with the R&D
human assets of MG Rover but intemalised the function even while permitting it
operational control. This resurrects the M-Form with the specific purpose of exploiting
the local factor endowments, specifically the engineering resources available in the UK.
In accordance with M-Form of govemance stmcture, strategic control is separated and
resides with SAIC. The problem for this research is that this does not indicate where
strategic control would have resided in a tme IVJV between independent partners. This is
particularly relevant for automobile companies where R&D and strategy are closely tied.
A possible answer may be found in the kind of loose conglomerations exhibited by
Japanese keiretsu groups where cooperative relationships are secured by equity cross

362
holdings. A govemance stmcture appropriate to an IVJV is therefore an area where a
continuation of this research could usefully be focused.

8.2 Paradigm revolution for the automobile industry

Forecasts by Automotive News and by Global Insight, the automotive industry analysis
company, suggest that total global automobile output could grow to 60m units by 2011.
Figure 8.1 draws on three sets of data, two of which are concemed with forecasting.
Actual output data from OICA is given for 2000 to 2006. The forecast data for 2007 to
2012 is simply an average of the forecasts from Automotive News and Global Insight.
Finally, the trend from this average forecast is extended from 2013 until 2020.

Figure 8.1 Automobile production output and forecast 2000-2020

Output (m)

70 H ■Global Output
•Forecast Output

Year

Sources: OICA, Automotive News,

Global Insight

Figure 8.1 indicates that global production could reach 79m units by 2020, and according
to World Business Council for Sustainable Development (WBCSD)(2004) the Chinese
passenger car market will rival that of the United States by 2030, although forecasts
conceming the Chinese market often underestimate the growth in demand. In any case,
such forecasts may become obsolete if the automobile industry paradigm should undergo
a revolutionary paradigm shift due to radical changes in the underlying technical
considerations, as foreshadowed in some of the following developments.

363
The observation that the industry leaders are making greater commitments to R&D was
one of the more interesting results of this research. It was expected that R&D costs
would have been lower for larger firms since an increased number of model variants
suggested that R&D per vehicle would have been less. Instead, the industry leaders were
spending more per model. The fieldwork data had a valuable contribution to make here,
particularly at Honda, where interviewees discussed at great length projects such as jet
engine design, robotics and fuel cell research. While the interviewees attempted to
justify these programmes, even with a degree of validity, it should be emphasised that
these new technologies are not relevant to the current automobile paradigm.

Research that is being conducted on new technologies represents the next paradigm for
the industry. As far as vehicle motive power is concemed, it is well known that oil is
both a limited resource and a source of environmentally harmful exhaust emissions.
Hybrid technology, which combines an intemal combustion engine with an electric
generator, is often cited as an interim technology but since it retains an orthodox engine
it is adding cost to the production process. A joint study by UBS and Ricardo Consulting
found that a hybrid powertrain, as fitted to a Lexus RX400h, was twice the cost of an
equivalent fully compliant clean diesel and up to five times the cost of an equivalent
petrol engine (UBS/Ricardo, 2007). In effect, hybrid technology adds another function
to the full-function model, that of electric powertrain.

A new paradigm for the industry would involve a distinctly new technology. Fuel cell
technology may be this technology because it not only changes the fuel but also the
production system. Whereas a traditional engine is a bulky device that can only be
positioned in a limited number of ways within a vehicle, generally at the front or the
back, a fuel cell system can be positioned with more flexibility. While the fuel cell stack
at the heart of the system can be placed where the engine was customarily located, the
ancillary components, such as batteries, can be distributed throughout the vehicle
platform. GM designed the AUTOnomy concept as a chassis that contained all the
elements of a fuel cell system, allowing the body design to be entirely separate (Nash,
2002). Like the craft system of automobile production in the early years of the industry
this concept means that the vehicle body is not the main load-bearing structure so there is
greater flexibility in body design and use of materials. This in turn

364
makes the BIW process of the Budd Paradigm redundant with subsequent implications
for economies of scale in the industry.

Fuel cell technology is still at the experimental stage and it is not known when, or even
whether, it will reach mass production and what the costs of production will be. Some
manufacturers, such as Mercedes-Benz, suggest that production fuel cell vehicles could
be introduced by 2010 but these technology demonstrators will be intended to test the
market. Furthermore, there is no significant hydrogen fuel infrastructure in place and
according to Fuel Cells 2000 (Fuel Cells 2000, 2007) there is just one fuel station in the
UK. This may mean that the first fuel cell applications will be commercial vehicles with
centralised fuel supply, and that personal vehicles will be supplied from their ovm
domestic hydrogen producing devices. There is also the possibility that fuel cell vehicles
will produce their ovm hydrogen from other fuel sources carried on board.

Since the future of fuel cell technology is unknown there is no way of telling if the
research strategies of the automobile industry leaders are the appropriate ones. If it
transpires that economies of scale for fuel cells are very high, particularly in view of the
capital intensive nature of production, or that consumers consider them to be a generic
product, then automobile manufacturers might find that they need to restructure their
hitherto full-function integrated structures. It is feasible that some manufacturers might
even specialise in the production of fuel cells and relinquish their vehicle production
facilities. Alternatively, vehicle manufacturers may relinquish powertrain production in
favour of a specialised supply industry. If this were the case, and GM’s AUTOnomy
concept became the industry standard, then automobile firms could manufacture generic
vehicles chassis in a central location and ship them to small, final assembly plants that
served their local markets in the style of micro factory retailing (Nieuwenhuis and Wells,
2003). This, of course, would only be economically feasible if body production was
suitably adapted to a new MEPS.

The uncertainty surrounding the possible paradigm revolution means that automobile
firms that are currently operating at or near MESp will not necessarily suffer a
disadvantage. Indeed, should MEPS for any given function be significantly reduced,
particularly for body production, then these manufacturers may be as well placed as

365
any larger manufacturer. Furthermore, if it transpires that fuel cell technology is not
suitable for personal vehicles then it may be that the current paradigm will continue but
using a different fuel source. This would represent a paradigm shift if it was due to some
evolution in the technical basis of the paradigm. One innovation that would make this
possible is if hydrogen was consumed within an intemal combustion engine in place of
oil based fuel. Although BMW developed the first hydrogen fuelled car, the 750hL in
2000 (BMW, 2007), using a conventional engine, Mazda’s rotary engine technology is
more suited to the consumption of hydrogen fuel (Yamaguchi, 2003). A switch in the
industry to production of rotary engines might require an adjustment in the production
technology that shifted the present paradigm, although not enough to herald an entirely
new paradigm.

8.3 Implications for national policy

This research has found that globalisation was not an ultimate state but was a continuing
process of related development between distinct geographic regions based on their factor
endowment differentiation. A narrow view of factor endowment differentiation would
suggest that globalisation would eventually lead to a homogenous global industry. A
broader view, though, as espoused by Schott (2003) shows that factor endowment
differences are fundamentally constant and so globalisation is a dynamic process as
regions continue to develop their specific advantages.

Automobile industry leading companies have tended to ignore the opportunities offered
by globalisation because they find greater advantage in their centrally controlled intemal
markets. This has resulted in industry leaders taking a strategic view of opportunities,
locating plants as part of an overall global plan rather than to exploit specific local
advantages. It is something of an irony that Japanese automobile firms have selected the
UK for production plants when the UK’s comparative advantage is in engineering rather
than labour intensive assembly work. Since the rationale behind the plants is strategic
they may be moved to other locations should the strategy require it. Ford’s conversion of
the Dagenham final assembly plant into a centre of excellence for diesel engines is an
example of this, consequently all Ford

366
branded cars are now imported into the UK, principally from assembly plants elsewhere
in Europe.

Dunning (1979) suggested that factor endowments can be developed further by

investment in facilities and workforce training. Given that the definition of any
forthcoming paradigm revolution in the automobile industry is replete with uncertainty
it could be argued that research outside of the current paradigm should be the preserve
of national governments. This would then create a bedrock of capability that
manufacturers could draw on by cooperating in specific programmes or recruiting staff
that have experience of the latest developments. Government research would thereby
provide knock-on benefits for the national economy in a similar manner to defence
programmes or space research.

There is nothing new in this suggestion and governments have at times become involved
in company operations at times of crisis, but with only sporadic success rates. Indeed,
MG Rover and its predecessors were subject to government involvement in more ways
than perhaps any other company in the industry. Part of the reason for MG Rover’s
ultimate demise was that the UK government could not, and would not, rescue it again.
This is not, though, what the development of factor endowment should entail. If the
skills that were prevalent within MG Rover have been redeployed elsewhere in the
economy to greater net benefit then the government was wise not to intercede in the
company’s fate. Government policy should instead be directed at ensuring these
fundamental capabilities are progressively enhanced.

Of course, there is no certainty that a government is more effective at identifying which

factor endowments to support than it is at identifying which companies to rescue and the
fieldwork interviews found little gratitude in the industry for government involvement so
far. For example, Mr Oldaker of MG Rover registered his disappointment with the
quality of graduates emerging from British universities. National pride can also cause
obstructions, the Chinese government promoting the development of R&D capability in
the domestic automobile industry even though the nation’s comparative advantage lies in
production. If the myriad of new Chinese car manufacturers are found to lack R&D
capabilities with which to compete against the global industry leaders then they will fail
to reach MESp and will require continuing

367
government support, or else they will leave the industry. Instead, the Chinese
government should enhance the nation’s underlying production capabilities so that it
retains its current production advantages. This can be achieved by embracing the IVJV
concept that allows direct access to R&D capabilities elsewhere in the world. The
mechanism for this is free trade, allowing companies to seek out resources in an open
market. The role of the government, then, is to provide a secure legal framework for
companies in a free trade environment with a pool of fundamental resources to draw on.

8.4 Applicability to other industries

The IVJV strategy is only suitable in exceptional circumstances. For a sub-optimal firm
to exist at all suggests some form of support, perhaps through government subsidies,
company cross-subsidies or consumption of intemal company resources. When two
companies are suboptimal to the relevant industry paradigm in complementary ways
there is the potential for them to form an alliance that allows them to take sole
responsibility for their strengths but relinquish their weaknesses to their partner.
Furthermore, they should be located in regions with contrasting factor endowments so
that they can exploit the related comparative advantages. The final result, in order to
replicate to their industry paradigm, should be an industry standard set of functions (i.e.
full-function) and output that is competitively close to the MBS for the industry.

The ship building industry might have potential in this way. At one time shipbuilding
was a major industry in the UK and the country retains many of the skills relevant to
naval architecture. For example, Harland and Wolff in Belfast designed and built such
famous craft as The Titanic. While the company has a ship repair facility it no longer
builds ships, although it continues to the ship design capability (Harland and Wolff,
2007) that would permit it to link with a shipyard elsewhere in the world. To illustrate
this. Figure 8.2 shows an IVJV between Harland and Wolff and an overseas shipyard in
an emerging economy to exploit low costs in production. As in the automobile IVJV, the
market intelligence would retum to the company that initiated the product, Harland and
Wolff. Interdependency would be created by the Harland and Wolff s need to have access
to a shipyard while the shipbuilder would need to have a supply

368
of ship designs. The interdependency obviates the necessity for rigid, exclusive
contracts and it is likely that both could pursue their own strategies and therefore their
own separate govemance structures. A similar relationship seems to be evolving
between Camival Corporate Shipbuilding, designers of the Cunard cmise liners, and the
Italian shipyard, Fincantieri. For a complete IVJV, Fincantieri would have to relinquish
all cmise line design to Camival Corporate Shipbuilding. Other industries that might be
amenable to similar division are the aircraft and shipbuilding industries, each having an
R&D capability that can be readily separated from production.

Figure 8.2 Proposed IVJV between Harland and Wolff and a

shipbuilding company

Harland and Wolff

Naval
Architecture

[" Shipbuilding Company

Shipyard

Global Market

The armaments industry might present political problems, however, it being important to
balance the cost advantages of foreign production with the need to maintain a strategic
manufacturing facility in the home market. For this reason, countries strive to be self-
sufficient in the defence industry. This may be counterproductive: just as IVJV alliances
need an enlightened approach by national governments in recognising that there is
mutual benefit in promoting cross-border dependency, so similar cooperation in the
defence industry may actually reduce the need for it.

369
8.5 Suggestions for subsequent research

Perhaps one of the most surprising discoveries of this research was the relatively low
quantities for MESp. This was based on a calculation from the MEPS for each function,
but empirical support was found in the companies that produce near this level. Subaru
was often cited in this thesis, but AvtoVAZ is not much larger and they will soon be
joined by Chrysler and Mercedes-Benz, fresh from their demerger. Below these
manufacturers there are, of course, numerous Chinese manufacturers all attempting to
reach sustainability during the period that they are protected under Chinese law. Once
the Chinese market is fully liberalised then the domestic manufacturers will be exposed
to the global industry. There is therefore a sense of urgency for them in achieving MESp
before the government support falls away. For this reason it is important for other
studies to provide support for the measure of MESp found by this thesis.

This thesis has attempted to provide a cogent argument for IVJVs but there has been, so
far, little empirical evidence of them occurring in their pure form. There do seem to be
suggestions of the IVJV structure in other alliances, such as Rootes Group and Iran
Khodro, but this research did not find any such relationships in the definitive form. This
is partly because they are only possible in exceptional circumstances, but also because
there is political resistance to cross-border interdependencies, particularly in key
industries such as this. It might therefore be advisable to study cross-border alliances
between countries that are politically close.

Another reason why the IVJV has been slow in emerging is related to the particular
problem for this thesis in formulating an appropriate govemance stmcture for the IVJV.
In discussing various govemance stmctures (see 5.1.9 above) the research found little
empirical support for the M-Form yet there may be some possibility that it would be
suited to the IVJV in recognition of the clear separation of operational functions. This
then leaves the problem of strategic control. According to interviewees at MG Rover,
however, automobile markets are similar enough to accept common vehicle designs.
Evidence for this can be seen in the way that SAIC and NAC put the Rover 75/MG ZT
into production in China with few changes, apart from a slight increase in rear passenger
space. Equally, conducting R&D in the intended

370
market may not necessarily result in increased sales, the Honda Ridgeline truck being
designed in the US but capturing relatively few sales. This suggests that it may be
possible to leave R&D strategy with the R&D partner since they are as well placed as
any to judge the needs of the market.

8.5 Closing remarks

This research originally set out to demonstrate that MG Rover represented a viable
proposition as a self-sustaining automobile manufacturer. Many critics believed that this
was a hopeless task given the apparent general trend in the industry towards
consolidation. This thesis has shown that its planned level of production was well short
of that necessary to reach the industry standard of MESp, but not by the order of
magnitude that had been suggested. Furthermore, the company was in the process of
arranging an IVJV that this research has suggested held great promise. Ultimately,
however, events conspired against the company. Nevertheless, the purchase of MG
Rover’s assets by two ambitious Chinese automobile manufacturers shows that there
was value in MG Rover and that the successors to the company have the potential to
make a contribution to the global automobile industry far beyond the remit of the
original operation at Longbridge.

371
 References
ACEA (2007) Automotive Production (EU + EFTA Countries) Summary 2006.
ACEA, 10 May 2007, accessed 16 July 2007, website
<www.acea.be/files/PRODUCTION%202006.pdf>

Adeney M (1988) The motor makers: the turbulent history of Britain’s car industry.
London: Collins

Altman M (1999) The methodology of economics and the survival principle revisited
and revised: some welfare and public policy implications of modeling the economic
agent in Review o f Social Economy Vol. 57, No. 4, pp. 427-449

Altshuler A, Anderson M, Jones D, Roos D and Womak J (1986) The future of the
automobile: the report of MIT's International Automobile Program. Cambridge MA:
MIT Press

Almor T, Hashai N and Hirsch S (2006) The product cycle revisited: Knowledge
intensity and firm internationalization in Management International Review, Vol. 46,
No. 5, September 2006, pp. 507-626

Alvarez Gil M J and Gonzalez de la Fe P (1999) Strategic alliances, organisational

leaming and new product development: the cases of Rover and Seat in R&D
Management, Volume 29, Number 4, October 1999, pp. 423-426

Amicus (2004) Motor vehicles fact sheet. May 2004, released by Amicus: London

Anastakis D (2004) From independence to integration: the corporate evolution of the

Ford Motor Company of Canada. 1904-2004 in Business History Review 78 (Summer
2004) pp 213-253

Andersen O (1993) On the internationalization process of firms: a critical analysis in

Journal o f International Business Studies, 1993 24 (2) p209

Ankarloo D and Palermo G (2004) Anti-Williamson: a Marxian critique of New

Institutional Economics in Cambridge Journal o f Economics 2004, 28, pp413-429

Antras P (2005) Incomplete contracts and the product cycle in American Economic
Review, September 2005, Vol. 95, No. 4, pp. 1054-1073

Armour H O and Teece D J (1980) Vertical integration and technological innovation in

The Review o f Economics and Statistics, August 1980, Vol.62, No.3, pp.470-474

Armour H O and Teece D J (1978) Organizational structure and economic

performance: a test of the multidivisional hypothesis in The Bell Journal of
Economics, Vol.9, No.l (Spring, 1978), pp. 106-122

372
Artz K W and Brush T H (2000) Asset specificity, uncertainty and relational norms: an
examination of coordination costs in collaborative strategic alliances in Journal o f
Economic Behaviour and Organization, Vol.41, 2000, pp.337-362

Assembly (2002) Then and now: the mechanical marvel in Assembly, 1 February
2002, accessed March 2002, website
<http://www.assemblymag.com/CDA/ArticleInformation/features/BNP
Features Item/0,6493,98331 ,OO.html>

Autocar (2001) New car prices m Autocar, 12 December 2001

Autocar (2005) How £100m could save MG in Autocar, 19 April 2005, pp. 46-49

Automotive Design and Production (2004) Nissan: how to achieve shift from website
Automotive Design and Production July 2004, accessed 14 September 2006, website <
http://www.autofieldguide.eom/articles/070401 .html>

Automotive Engineer (2004) In the right mind in Automotive Engineer, January 2004,
pp. 60-61

Automotive Engineer (2004b) Rover seeks Proton alliance in survival battle in

Automotive Engineer, March 2004, pp. 4-5

Automotive News (2000) New Rover Cars owner ponders possible partners in
Automotive News, 15 May 2000, 74, 5874, p42

Automotive News (2001) MG Rover brings suppliers to Longbridge in Automotive

News, 29 January 2001, p.40

Automotive News (2004) Mitsubishi crisis is round 2 for Gilligan in Automotive News 2
August 2004p24

Automotive News (2004a) 2004 Market Data Book in Automotive News supplement,
28 June 2004

Automotive News (2004d) Guide to global automotive partnerships in Automotive

News, supplement, 4 October 2004

Automotive News (20051) 2005 Market Data Book in Automotive News supplement, 23
May 2005

Automotive News (2005h) Asian’s new R&D Tai Mahals are a sign of success in
Automotive News, 16 May 2005p i

Automotive News (2005i) Subaru plan finallv has one mission in Automotive News,
13 June 2005 p24
Automotive News (2005J) Guide to global automotive partnerships in Automotive
News, supplement, August 2005

373
Automotive News Europe (2006) 2006 Market Data Book in Automotive News
supplement, May 2006

Automotive News (2006a) 2006 Market Data Book in Automotive News supplement,
22 May 2006

Automotive News (2006b) 2006 Global Market Data Book in Automotive News
supplement, 26 June 2006

Automotive News (2006c) 2006 Guide to China’s Auto Market in Automotive News,
supplement, 1 May 2006

Automotive News (2006d) Magna Stevr: big changes needed to be comt)etitive in

Automotive News, 14 August 2006

Automotive News (2007) Even sold, Aston works with Ford from Automotive News,
19 March 2007

Automotive News (2007a) 2007 Global Market Data Book in Automotive News
supplement, 25 June 2007

Automotive News (2007b) 2007 Guide to China's Auto Market in Automotive News,
supplement, 14 May 2007

Automotive News (2007c) Hvundai halts production at Alabama plant in Automotive

News, 5 October 2007

AutoTech Daily (2002) GM touts flexible bodv shop tooling from AutoTech Daily, 16
December 2002, p. 1

Autoweek (1999) What Morgan for the millennium? from Autoweek, 1 November
1999, website accessed 23 August 2006
<http://www.gomog.com/articles/autoweek.html>

Aw B-Y and Batra G (1998) Firm size and the pattern of diversification in
International Journal o f Industrial Organization, Vol. 16, pp. 213-3 31

AWKnowledge (2003) Capacitv utilisation data. August/September 2003

Bailey D, Sugden R and Harte G (1994) Making Transnationals Accountable: a

significant step for Britain. London: Routledge
Bailey D (2003) Globalisation, regions and cluster policies: the case of the Rover
Task Force in Policy Studies, Vol.24, Nos.2/3, 2003, pp.67-85

Bailey D (2007) Globalization and restructuring in the auto industry: the impact on
the West Midlands automobile cluster in Strategic Change, Vol. 16, pp. 137-144

Barkema H G, Bell J H J and Pennings J M (1996) Foreign entry, cultural barriers and
leaming in Strategic Management Journal, Vol. 17(2) pp 151-166

374
Bamet R J and Mueller R E (1974) Global Reach: The Power of the Multinational
Corporations. New York: Simon & Schuster

Bartlett C A and Ghoshal S (1993) Beyond the M-Form: toward a managerial theory
of the firm in Strategic Management Journal, Vol. 14, pp. 23-46

Beamish P W and Makino S (1998) Performance and suryiyal of joint yentures with
non-conyentional ownership structures in Journal o f International Business Studies,
Vol. 29, pp. 7P7-818

Bell J, McNaughton R and Young S (2001) ‘Bom again’ global firms: an extension to
the 'bom global' phenomenon in Journal o f International Management, 7, 3, p p l-l 7.
2001

Benady D/TGWU (2003) Jobs relocation: a one-way street? TGWU publication,

issued 1 December 2003, accessed 5 April 2005, TGWU website
<http://www.tgwu.org.uk/Templates/Joumal.asp?NodeID=90352>

Benson J K (1983) A dialectical method for the study of organisations in Morgan G (ed.)
(1983) Beyond method: strategies for social research. Beyerly Hills, CA: Sage

Berger S (2000) Globalization and politics. MIT IPC Globalization Working Paper
00-005, in Annual Review o fPolitical Science (2002) Vol. 3 pp 43-62

Beynon H (1984) Working for Ford. Harmondsworth: Pelican

Bhaskar (1979) The future of the UK motor industry. London: Kogan Page

Blois K J (1972) Vertical quasi-integration in The Journal o f Industrial Economics,

Vol.20, No.3, July 1972, pp.253-272

BMW (2007) Company Facts - Sustainability and the Enyironment from company
website, accessed 13 October 2007, <http://www.bmweducation.co.uk/default.asp>

Borensztein E, De Gregorio J and Lee J-W (1998) How does foreign direct
inyestment affect economic growth? in Journal o fInternational Economics, Vol. 45,
pp. 115-135

Brady C and Lorenz A (2001) End of the Road: BMW and Royer - a brand too far.
London: Pearson Education Ltd

Brannen P (1987) Working on directors: some methodological issues in Moyser and

Wagstaff (eds.)(1987) Research methods for elite studies. London: Allen and Unwin

British Psychological Society (1992) Ethical principles for conducting research with
human participants from British Psychological Society, accessed 1 Noyember 2006,
website <http://vyww.bps.org.uk/the-society/ethics-mles-charter-code-of-
conduct/code-of-conduct/ethical-principles-for-conducting-research-with-human-
participants. cfm>

375
Bryman A (2001) Social research methods. Oxford: Oxford University Press

Bryman R (1989) Research methods and organization studies. London: Routledge

Buckley M and Rees C (2002) The World Encyclopaedia of Cars: The Definitive
Guide to Classic and Contemporary Cars from 1945 to the Present Dav. London:
Hermes House

Buckley P J (1990) Problems and developments in the core theory of international

business in Journal o fInternational Business Studies, Q4 1990, Vol.21, Iss.4, pp.657-
665

Buckley P J (2002) Is the international research agenda running out of steam? from
Journal o f International Business Studies, Vol. 3 3, No. 2, Q2 2002, pp. 365-373

Buckley P J and Casson M (1976) The future of the multinational enterprise. London:
MacMillan press

Buckley P J and Casson M (1981) The optimal timing of a foreign direct investment
in The Economic Journal, Vol.91, No.361, March 1981, pp. 75-87

Buckley P J and Casson M (1998) Models of the multinational enterprise in Journal o

f International Business Studies, Vol. 29, No.l, Q1 1998, pp.21-44

Buckley P J and Casson M C (1998b) Analyzing foreign market entry strategies:

extending the internalization approach in Journal o f International Business Studies,
Q3 1998, Vol.29, No.3

Buckley P J and Ghauri P N (2004) Globalisation, economic geography and the

strategy of multinational enterprises in Journal o f International Business Studies,
2004, Vol. 35, pp. 81-98

Burton R B and Obel B (1980) A computer simulation test of the M-Form hypothesis in
Administrative Science Quarterly, Vol. 25, No. 3, September 1980, pp. 457-466

Business Week (2000) Unloading Rover may not win the race in Business Week, April 3,
2000, p. 59

Business Week (2007) China’s Auto Industry: Roaring ahead in Business Week, 2
May 2007
Cabral L M B (2000) Introduction to industrial organization. Cambridge, MA: MIT
Press
Caimcross A (1966) Introduction to economics. London: Butterworths

Calvet A L (1981) A synthesis of foreign direct investment theories and theories of the
multinational firm in Journal o f International Business Studies, Spring 1981,
Vol.l2,pp.43-59

316
Cantner U, DreBler K and Kruger J J (2006) Firm survival in the German automobile
industry in Empirica (2006) Vol. 33, pp. 49-60

Carr C (1993) Global, national and resource-based strategies: an examination of

strategic choice and performance in the vehicle components industry in Strategic
Management Journal, Vol. 14, No. 7, October 1993, pp. 551-567

Carr C (1999) Globalisation, strategic alliances, acquisitions and technology transfer.

Lessons from ICL/Fuiitsu and Rover/Honda and BMW in R&D Management. Vol.29,
lss.4, 1999, pp.405-421

Caves R E (1996) Multinational enterprise and economic analysis. Cambridge:

Cambridge University Press

Central Policy Review Staff (1975) The future of the British car industry: a report.
London: Her Majesty’s Stationery Office

Chalmers A F (1983) What is this thing called science? Milton Keynes: Open
University Press

Chase J (1995) Historical analysis in psychological research in Breakwell G M,

Hammond S and Fife-Schaw C (eds.)(1995) Research methods in psychology.
London: Sage Publications

Chandler (1977) The visible hand: the managerial revolution in American Business.
London: Harvard Universty Press

Chandler A D (1990) Strategy and structure: chapters in the history of the American
industrial enterprise. Cambridge, MA: MIT Press

Change S J and Choi U (1988) Strategy, structure and performance of Korean business
groups: a transactions cost approach in The Journal o f Industrial Economics, Vol.37,
No.2, December 1988, pp. 141-158

Cheng L K and Kwan Y K (2000) What are the determinants of the location of
foreign direct investment? The Chinese experience in Journal o f International
Economics, Vol. 51, pp. 3 79-400

Chery (2007) Cherv Engine Company from Chery website, accessed 9 August 2007,
website <http://www.cheryacteco.eom/en/index.jsp>

Church R (1979) Herbert Austin: the British motor car industry to 1941. London:
Europa

Church R (1994) The rise and decline of the British motor industry. London:
Macmillan

377
Clark K B, Chew W B, T Fujimoto, Meyer J and Scherer F M (1987) Product
development in the world auto industry in Brookings Papers on Economic Activity,
Vol. 1987No.3, Special Issue on Microeconomics, pp. 729-781

Clarke R (1987) Conglomerate firms in Clarke R and Me Guiness A (1987)(eds.) The

economics of the firm. Oxford: Basil Blackwell

Cleeve E (1997) The motives for joint ventures: a transaction cost analysis of
Japanese MNEs in the UK in Scottish Journal o f Political Economy, February 1997,
Vol.44, No.l, pp31-43

Coase R H (2000) The acquisition of Fisher Bodv by General Motors in Journal of

Law and Economics 43 (2000) pp 15-31

Coles J W and Hesterly W S (1998) Transaction costs, quality, and economies of scale:
examining contracting choices in the hospital industry in Journal o f Corporate
Finance 4 (1998) pp 321-345

Collins P and Stratton M (1993) British car factories from 1896: a complete historical,
geographical, architectural and technological survey. Godmanstone, Dorset: Veloce
Publishing

Cooper A C (1964) R&D is more efficient in small companies in Harvard Business

Review, 42 (3) pp. 75-83

Corbetta P (2003) (translated by Bernard Patrick) Social research: theory, methods

and techniques. London: Sage

Cowling K and Sugden R (1998) Strategic trade policy reconsidered: national rivalry
vs free trade vs international cooperation in Kyklos, Vol.51, Iss.3, pp.339-357

Cowling K and Tomlinson P R (2005) Globalisation and corporate power in

Contributions to Political Economy, Vol. 24, pp. 33—54

Creswell J W (2003) Research design: qualitative, quantitative and mixed methods

approaches. London: Sage

DaimlerChrysler (2004) An engineering odvssev: from all-steel to true unit bodies 1914-
1960 from Walter P Chrysler Museum, accessed 7 April 2005, website:
< hth5://www.chryslerheritage.com/sec500__pdf/An%20Engineering%200dyssey.pdf>

Dankbaar B (1992) Economic crisis and institutional change: the crisis of Fordism
from the perspective of the automobile industry. Maastricht: UPM

D’Aveni R A and Ravenscraft D J (1994) Economies of integration versus

bureaucracy costs: does vertical integration improve performance? in Academy o f
Management Journal, 1994, Vol. 3 7, No. 5, pp. 1167-1206

David R J and Han S-K (2004) A systematic assessment of the empirical support for
transaction cost economics in Strategic Management Journal, 2004, 25, pp. 39-58

378
Demizu T (2003) Honda: its technology and management. Osaka: Union Press

Denscombe M (1998) The good research guide: for small-scale social research
projects. Buckingham: Open University Press

DenzinN K (1978) The research act: a theoretical introduction to sociological

methods (2"*^ ed.). New York: McGraw-Hill

De Vaus D (2001) Research design in social research. London: Sage

Dexter L A (1970) Elite and specialised interviewing. Evanston IL: Northwestem

University Press

Dey I (1993) Oualitative data analvsis: a user-friendlv guide for social scientists.
London: Routledge

Dicken P (2003) Global shift: transforming the world economv. London: SAGE

Diesel Progress (1999) Peugeot Citroen and Ford sign diesel engine agreement in
Diesel Progress, December 1999, website accessed 15 August 2007, website
<http://findarticles.eom/p/articles/mi_mOFZX/is_l 2_65/ai_5 8674994>
Donaldson L (2001) The Contingencv Theory of Organizations. London: SAGE

Donnelly T, Morris D and Donnelly T (2005) Renault-Nissan: a marriage of

necessity? in European Business Review, Vol. 17, Iss. 5, pp.428 - 440

Doz Y L and Prahalad C K (1991) Managing DMNCs: a search for a new paradigm in
Strategic Management Journal, Vol. 12, pp. 145-164

Drucker P F (1972) Concept of the corporation. New York: John Day

DTI (2003) The 2003 R&D Scoreboard. London: HMSO

DTI (2004) The 2004 R&D Scoreboard. London: HMSO

DTI (2005) The 2005 R&D Scoreboard. London: HMSO

DTI (2006) The 2006 R&D Scoreboard. London: HMSO

DTI/Auto Industry (2007) Background information on MG Rover from Auto Industry

website, accessed 8 February 2007
<http://www.autoindustry.co.uk/features/MG%20Rover/MG_Rover_Background>

Dunne P and Hughes A (1994) Age, size, growth and survival: UK companies in thg
1980s in The Journal o f Industrial Economics, Vol. 42, No. 2, pp. 115-140

379
Dunning J H (1979) Explaining changing patterns of international production: in
defence of the eclectic theory in Oxford Bulletin o f Economics and Statistics, Vol.41,
Iss.4, pp.269-295

Dunning J H (1980) Toward an eclectic theory of international production: some

empirical tests in Journal o f International Business Studies, Spring 1980, Vol. 11
pp.9-SI

Dunning J H (1981) International production and the multinational enterprise.

London: George Allen and Unwin

Dunning J H (1995) Reappraising the eclectic paradigm in an age of alliance

capitalism in Journal o f International Business Studies, 1995, Vol. 26, Issue 3, p46I-
491

Dunning J H (2000) The eclectic paradigm as an envelope for economic and business
theories of MNE activity in International Business Review, 9, 2000, pp. 163-190

Dussauge P, Garrette B and Mitchell W (2004) Asymmetric performance: the market

share impact of scale and link alliances in the global auto industry in Strategic
Management Journal, Vol. 25, Iss. 7, pp. 701-711

Dyer J H (1996) Does goyemance matter? Keiretsu alliances and asset specificity as
sources of Japanese competitiye adyantage in Organization Science, Vol. 7, No. 6,
November-December 1996, pp. 649-666

Dyer J H (1997) Effectiye interfirm collaboration: how firs minimize transaction costs
and maximise transaction yalue in Strategic Management Journal, Vol. 18, Iss. 7,
pp535-556

Economist (2000) Walking away from Longbridge from The Economist, 16 March
2000

Economist (2002a) Special Report: Car Manufacturing in The Economist, 23

February 2002

Economist (2005) The global car industry: extinction of the predator in The
Economist, 8 September 2005

Edwardes M (1983) Back from the brink: an apocalyptic experience. London: Collins

Erdener C and Shapiro D M (2005) The internationalisation of Chinese family

enterprises and Dunning’s eclectic MNE paradigm in Management and Organization
Review, Vol. I, No. 3, pp. 411 -436

Etemad H, Wright R W and Dana L P (2001) Symbiotic international business

networks: collaboration between small and large firms in Thunderbird International
Business Review, Vol. 43 (4), pp48I-499, July-August 2001

380
Etter J-F and Pemeger T V (2000) Snowball sampling bv mail: application to a survey
of smokers in the general population in International Journal o f Epidemiology, 2000,
29, pp. 43-48

Faugier J and Sargeant M (1997) Sampling hard to reach populations in Journal o f

Advanced Nursing, 1997, 26 (4), pp. 790-797

Fech H E and Ginsberg P B (1974) Optimal scale in medical practice: a survivor

analvsis in The Journal o f Business, Vol. 47, No. 1, pp. 23-36

Ferguson P R and Ferguson G J (2000) Organisations: a strategic perspective.

Basingstoke: Macmillan Press

Feyerabend P (1993) Against Method. London: Verso

Fife-Schaw C (1995) Survevs and sampling issues in Breakwell G M, Hammond S

and Fife-Schaw C (eds.)(1995) Research methods in psvchologv. London: Sage
Publications

Financial Times (200la) How Jensen got back on the road in The Financial Times, 23
August 2001, p i 7

Financial Times (2002) A driving ambition to do the impossible in The Financial

Times, 3 May 2002, p i 4

The Financial Times (2002b) Defiant car maker sees a positive route to the future in
The Financial Times, 27 April 2002, p3

Financial Times (2005g) MG Rover plans to recruit 200 engineers and development
staff in Financial Times, 1 March 2005, p2

Fitz J and Halpin D (1995) ‘Brief encounters: researching education policv-making in

elite settings in Salisbury J and Delamont S (eds.)(1995) Oualitative Studies in
Education. Aldershot: Avebury Ashgate Publishing

Flick U (2004) Triangulation in qualitative research in Flick U, von Kardoff E and

Steinke I (eds.)(translated Jenner B) A companion to qualitative research. London:
Sage

Ford (2004) Welcome to the home of the Ford Motor Companv and our familv of
breinds. Ford company website, accessed 6 November 2006
<http://www.ford.com/en/default.htm?source=topnav>

Ford (2006) Ford flexible engine production boosted bv investment in Britain. Ford
press release 06/098, 22 August 2006
Freeland R F (1996a) The Mvth of the M-Form? Govemance, Consent, and
Organizational Change in The American Journal o f Sociology^ Vol. 102, No. 2
September 1996, pp. 483-526

381
 Freeh H E and Mobley L R (1995) Resolving the impasse on hospital scale
economies: a nevy approach in Applied Economics. 1995, Vol27, No.3, 286-296

Freeland R F (1996b) The Struggle for Control of the Modem Corporation:

Organizational Change at General Motors. 1924-1958 in Business and Economic
History, Vol. 25, No.l, Fall 1996

Freeland R F (2000) Creating holdup through vertical integration: Fisher Bodv

revisited in Journal o f Law and Economics, vol. 43, April 2000, pp. 3 3-66

Fuel Cells 2000 (2007) Online Fuel Cell Information Resource website, accessed 13
October 2007, <http://www.fuelcells.org/>

Fumia Design (2006) Lancia J: tribute to Lancia century from Fumia Design press
release, October 2006, accessed 8^^ January 2007, website <
http://www.fumiadesign.com/sel_news.html>

Galaskiewicz J (1987) The studv of a business elite and corporate philanthropv in a

United States metropolitan area in Moyser and Wagstaff (eds.)(1987) Research
methods for elite studies. London: Allen and Unwin

Garrahan P and Stewart P (1991) Nothing new about Nissan? in Law C M (ed.)
(1991) Restructuring the global automobile industry: national and regional impacts.
London: Routledge

Gear Product News (2006) Cast gear blanks: what to consider when in Gear Product
News, June 2006pp. 34-39

Georgano N (2000) The Beaulieu encvclonaedia of the automobile. London: The

Stationery Office

Gershwin S B (1987) An efficient decomposition method for the approximate

evaluation of tandem queues with finite storage space and blocking in Operations
Research, Vol. 35, No. 2, March-April 1987, pp. 291-305

Gilson R J and Roe M J (1993) Understanding the Japanese keiretsu: overlaps

between corporate governance and industrial organization in Yale Law Journal Vol.
102, No.4 (Jan 1993), pp.871-906

Gorg H and Strobl E (2003) ‘Footloose’ multinationals in The Manchester School,

Vol. 71, No. l,pp .l-19
Grossman S J and Hart O D (1986) The costs and benefits of ownership: a theory of
vertical and lateral integration in Journal o f Political Economy, 1986, Vol.94, No.4,
pp.691-719
Hagel J and Singer M (1999) Unbundling the corporation in Harvard Business
Review. March-April, pp. 13 3-41

Hakim C (1987) Research design. London: Allen Unwin

382
Hamel J with Dufour S and Fortin D (1991) Case studv methods. London: Sage
Harrigan K R (1986) Matching vertical integration strategies to competitive situations
in Strategic Management Journal, 1986, Vol. 7, pp. 535-555

Harland and Wolff (2007) Engineering Design Resources from company website,
accessed 17 October 2007, < http://www.harland-wolff.com/facilitiesl.asp>

Hart O and Moore J (1990) Property rights and the nature of the firm in Journal o f
Political Economy, 1990, Vol.98, No.6

Harvey-Jones J with Masey A (1990) Trouble Shooter. London: BBC Books

Hedlund G (1994) A model of knowledge management and the N-form corporation in

Strategic Management Journal, Vol. 15, pp. 73-90

Hennart J-F and Reddy S (1997) The choice between mergers/acquisitions and joint
ventures: the case of Japanese investors in the United States in Strategic Management
Journal, 1997, Vol. 18, pp. 1-12

Hill C W L (1985) Intemal organization and enterprise performance: some UK

evidence in Managerial and Decision Economics, Vol. 6, Iss.4, pp. 210-216

Hill C W L (1988) Intemal capital market controls and financial performance in

multidivisional firms in The Journal o f Industrial Economics, Vol.37, No.l,
September 1988, pp. 67-83

Hill C W L (2003) Intemational business: competing in the global marketplace.

Boston: McGraw-Hill

Hill C W L, Hwang P and Kim W C (1990) An eclectic theory of the choice of

intemational entry mode in Strategic Management Journal (1990) Vol. 11, ppll7-128

Ho Y C, Eyler M A and Chien T T (1983) A new approach to determine parameter

sensitivities to transfer lines in Management Science, Vol.29, No.6, June 1983,
pp.700-714

Holmstrom B and Milgrom P (1994) The firm as an incentive system in The American
Economic Review, September 1994, Vol.84, No.4

Holweg M and Oliver N (2005) Who killed MG Rover? Special report from the
Cambridge-MIT Institute’s Centre for Competitiveness and Innovation, 25 April 2005

Honda (1999) A dynamic past, an exciting future. Tokyo: Honda

Honda (2002) 2002 FactBook. Tokyo: Honda

Honda (2004) Honda’s US’s automobile operations in Honda Annual Report 2004, p.
15

383
Honda (2006) 2006 FactBook. Tokyo: Honda

Horaguchi H and Toyne B (1990) Setting the record straight: Hvmer, internalization
theory and transaction cost economics in Journal o f International Business Studies,
Vol.21, Iss.3, pp.487-494

Hoskisson R E, Hill C W L and Kim H (1993) The multidivisional structure:

organizational fossil or source of value? in Journal o f Management, Vol. 19, No. 2,
pp. 269-298

Hounshell D A (1984) From the American system to mass production. 1800-1932.

London: Johns Hopkins University Press

Humphries A S and Wilding R D (2004) Long term collaborative business

relationships: the impact of trust and behaviour in Journal o f Marketing
Management 2004, Vol.20, pp. 1107-1122

Husan R (1997) The continuing importance of economies of scale in the automotive

industry in European Business Review, 1997, Vol.97, No.l

Hymer S H (1960) The intemational operations of national firms, a studv of direct

foreign investment. PhD thesis, Massachusetts Institute of Technology

Hymer S (1970) The efficiency (contradictions! of multinational corporations in The

American Economic Review, Vol.60, No.2, pp.441-448

Hymer S (1972) The internationalization of capital in Journal o f Economic Issues,

March 1972, Vol.6, Iss.l, pp.91-111

Hymer S and Pashigian P (1962) Turnover of firms as a measure of market behaviour in

The Review o f Economics and Statistics, Vol. 44, No. 1, pp. 82-87

Iran Khodro (2005) company website, Paykan history, accessed 20 June 2005,
website <http://www.ikco.com/products/paykan.asp>

JAMA (2005a) Growing in America from JAMA website, accessed 30^ August 2006,
website: < http://www.jama.org/library/factsheets/Contributions_2005.pdf>

JAMA (2006) 2006: the motor industry of Japan. Tokyo: Japan Automobile
Manufacturers Association

JAMA (2007) 2007: the motor industry of Japan. Tokyo: Japan Automobile
Manufacturers Association

JAMA (2007b) Japan’s motor vehicle statistics from JAMA website, accessed 24
September 2007, website <http://www.jama.org/statistics/index.htm>

Janesick V J (2000) The choreography of qualitative research design in Denzin N K

and Lincoln Y S (eds.)(2000) Handbook of qualitative research (2"^* ed.). London:
Sage

384
Jick T D (1979) Mixing qualitative and quantitative methods: triangulation in action
in Administrative Science Quarterly, December 1979, Vol. 24, Issue 4, pp. 602-611

Johansen J and Vahlne J-E (1990) The mechanism of internationalisation in

International Marketing Review, Volume 7 Issue 4 ppII-24

Jones J and Wren C (2006) Foreign direct investment and the regional economy.
Aldershot: Ashgate Publishing

Joskow P L (1985) Vertical integration and long-term contracts: the case of coal-
burning electric generating plants in Journal o f Law, Economics and Organization
Vol. 1 (1) Fall 1985 pp33-S0

Joskow P L (2003) Vertical integration in Menard C and Shirley M M (eds.) (2005)

Handbook of new institutional economics. Cambridge, MA: Springer

Just-Auto (2004a) General Motors to relocate Asia-Pacific HO to Shanghai and

expand local design centre from Just-Auto, 23 June 2004, accessed 20 September
2006, website < http://just-auto.com/news_detail.asp?art=44815&dm=yes>

Just-Auto (2007) New chief at BMW engine plant from Just-Auto, 1 February 2007,
accessed 1 February 2007, website < just-auto.com>

Kakabadse A and Kakabadse N (2000) Sourcing: new face to economies of scale and
the emergence of new organizational forms in Knowledge and Process Management,
2000, Vol.7, No.2,pp.l07-II8

Kia (2007) Kia History from Kia website, accessed 9 August 2007, website
<http://www.kiainfo.co.Uk/about.htm#Kia%20History>

Kindleberger C P (1984) Multinational excursions. Cambridge, MA: MIT Press

Kitay J and Callus R (1998) The role and challenge of case studv design in industrial
relations research in Whitfield K and Strauss G (eds.) (1998) Researching the world of
work: strategies and methods in studying industrial relations. Ithaca, NY: Cornell
University Press

Klein B (1980) Transaction cost determinants of ^‘unfair” contractual arrangements in

American Economic Review 1980 70 (May) pp356-362

Klein B (1988) Vertical integration as organizational ovmership: the Fisher Body-

General Motors relationship revisited in Journal o f Law, Economics, and
Organization, Vol.4, No.l, Spring 1988
Klein B (2000) Fisher-General Motors and the nature of the firm in Journal of Law
and Economics 43 (2000) pp 105-141

385
 Klein S, Frazier G L and Roth V J (1990) A transaction cost analysis model of
channel integration in international markets in Journal o f Marketing Research, May
1990, Vol.27, pp. 196-208

Klepper S (1996) Entry, exit, growth, and innovation over the product life cycle in
The American Economic Review, Vol. 86, No. 3, pp. 562-583

Klepper S (2002) Firm survival and the evolution of oligopoly in The RAND Journal
of Economics, Vol. 33, No. l,pp . 37-61

Klepper S (2002b) The capabilities of new firms and the evolution of the U.S.
automobile industry in Industrial and Corporate Change, Vol. 11, No.4, pp.645-666

Kogut (1991) Joint ventures and the option to expand and acquire in Management
Science, Vol.37, No.l, January 1991, pp. 19-33

Kolodny H F (1979) Evolution to a matrix organisation in Academy o f Management

Review, Vol.4, No.4, pp.543-553

Koutsoyiannis A (1979) Modem microeconomics (2"^* ed.). London: Macmillan

KPMG (2007) Momentum 2007 KPMG Global Auto Executive Survey from KPMG,
website accessed 17 July 2007, website <http://www.kpmg.co.uk/pubs/
ACFFCBC.pdL>

Krafcik J F (1988) Triumph of the lean production system in Sloan Management

Review Fall 1988, Vol. 30 (1) pp 41-52

Kuemmerle W (1999) Foreign direct investment in industrial research in the

pharmaceutical and electronics industries - results from a survey of multinational
firms in Research Policy, Vol. 28, pp. 179-193

Kuhn T S (1970) The structure of scientific revolutions. London: University of

Chicago Press

Lakatos I (1970) Methodology of scientific research programmes in Lakatos I and

Musgrave A (ed.s)(1970) Criticism and the growth of knowledge. Cambridge:
Cambridge University Press

Lamming R (1993) Bevond partnership: strategies for innovation and lean supply.
London: Prentice Hall

Lee T W (1999) Using qualitative methods in organizational research. London: Sage

Leung M K, Young T and Digby D (2003) Explaining the profitability of foreign

banks in Shanghai in Managerial and Decision Economics, Vol. 24, pp. 15-24

Levy D (1985) Transactions cost approach to vertical integration: an empirical

examination in The Review o f Economics and Statistics, Vol. 67, No. 3 (Aug., 1985),
pp. 438-445

386
Lewchuck W (1987) American technology and the British vehicle industry: a century
of production in Britain. Cambridge: Cambridge University Press

Lierberman M B (1991) Determinants of vertical integration: an empirical test in The

Journal o f Industrial Economics, September 1991, Vol. 39, No. 5

Lord R G and Hohenfeld J A (1979) Longitudinal field assessment of equity effects

on the performance of major league baseball plavers in Journal o fApplied
Psychology, February 1979, Vol. 64, Iss.l, pp. 19-26

Lotus (2006) Lotus Engineering - The Total Solution Provider from Lotus
Engineering website, accessed 1 September 2006, website <http://www.group
lotus.com/generic/generic.php?section= 17&page=21 &page_id=21>

Lyons B R (1995) Specific investment, economies of scale, and the make-or-buv

decision: a test of transaction cost theory in Journal o f Economic Behaviour and
Organization, 1995, Vol.26, pp.431-443

Mair A (1994) Honda’s global local corporation. Basingstoke: Macmillan Press

Makadok (1999) Interfirm differences in scale economies and the evolution of market
shares in Strategic Management Journal 1999, 20, pp. 935-952

Malmberg A (1991) Restructuring the Swedish manufacturing industry: the case of

the motor vehicle industry in Law C M (ed.) (1991) Restructuring the global
automobile industry: national and regional impacts. London: Routledge

Mariti P and Smiley R H (1983) Co-operative agreements and the organization of

industrv in The Journal o f Industrial Economics, Vol. 31, Iss.4, June 1983, pp.437-
451

Masten S E, Meehan J W and Snyder E A (1991) The costs of organization in Journal

ofLaw, Economics and Organization vol. 7.1 1991

Maxcy G (1981) The multinational motor industrv. London: Croom Helm

Maxcy G and Silberston A (1959) The motor industrv. London: Allen and Unwin

Maxton G P and Wormald J (1995) Driving over a cliff? Business lessons from the
world’s car industrv. Wokingham: EIU/Addison-Wesley

Mazda (2004) Mazda History. Company website, accessed 6 November 2006, website
<http://media.ford.com/mazda/article_display. cfm?article_id= 15147>

McAfee R P and McMillan J (1995) Organizational diseconomies of scale in Journal

ofEconomics and Management Strategy, Fall 1995, Vol.4, No.3, pp.399-426

387
 Mehrabi M G, Ulsoy A G and Koren Y (2000) Rcconfigurable manufacturing systems:
key to future manufacturing in Journal o fIntelligent Manufacturing, August
2000 Vol. 11 (4) J ^

Merriam S B (1988) Case study research in education: a qualitative approach.

London: Jossey-Bass

Miles M B (1979) Qualitative data as an attractive nuisance: the problem of analysis

in Administrative Science Quarterly, December 1979, Vol. 24, Issue 4, pp. 590-601

Miles M B and Huberman A M (1994)(2"‘*ed.) Qualitative data analysis: an expanded

sourcebook (2"^ editionV London: Sage

Milgrom P and Roberts J (1992) Economics, organisation and management.

Englewood Cliffs, NJ: Prentice Hall

Miller K D and Reuer J J (1998) Firm strategy and economic exposure to foreign
exchange rate movements in Journal o fInternational Business Studies, Vol 29
pp.493-513

MIRU (Motor Industry Research Unit) (1988) Rover: profile, progress and prospects.
An appraisal and forecast of the Rover Group. Norwich: Motor Industry Research
Unit

Monteverde K and Teece D J (1982) Supplier switching costs and vertical integration in
the automobile industrv in The Bell Journal o f Economics, Spring 1982, Vol. 13,
No.l,pp.206-213

Morris J and Imrie R (1992) Transforming buyer-supplier relations. London:

Macmillan

Moschandreas M (1994) Business Economics. London: Routledge

Moyser and Wagstaff (eds.)(1987) Research methods for elite studies. London: Allen
and Unwin

Munck R (2002) Globalisation and labour: the new “Great Transformations”. London:
Zed Books

Nash T (2002) GM’s futuristic ‘skateboard’ in Motor, April, 2002, accessed 15

October 2007 < http://findarticles.eom/p/articles/mi_qa3828/is_2002047pnum
=4&opg=n9055816>

Nieuwenhuis P and Wells P (1997) The death of motoring?: car making and
automobilitv in the 2 century. Chichester: Wiley

Nieuwenhuis P and Wells P (2003) The automotive industrv and the environment.
Cambridge: Woodhead Publishing

388
Nieuwenhuis P and Wells P (2007) The all-steel body as a cornerstone to the
foundations of the mass production car industrv in Industrial and Corporate Change
(advance access), April 9, 2007, pp. 1-29

Observer (2004) What’s Rover driving at? in The Observer, Business, 4 April 2004, p.
3

O’Callaghan T J O (2002) The aviation legacv of Henrv and Edsel Ford. Ann Arbor,
MA: Proctor Publications

Odaka K, Ono K and Adachi F (1988) The automobile industrv in Japan : a study of
ancillary firm development. Tokyo: Kinokuniya

OICA (2006a) The World’s Automotive Industrv: Some Key Figures. GIGA, 29^
November 2006, accessed 16^^ July 2007, website
<http://www.oica.net/htdocs/Main.htm>

OICA (2006) Production statistics 1998-2005 from Organisation Internationale des

Constructeurs d'Automobiles website, accessed 5 November 2006, website <
http://www.oica.net/htdocs/Main.htm>

Panzar J C and Willig R D (1981) Economies of scone in The American Economic

Review, Vol. 71, No. 2, May 1981, pp. 268-272

Parkin M (1996) Economics. Wokingham: Addison-Wesley

Parkin M, Powell M and Matthews K (1997) Economics (3*^^^ ed.). Harlow:

Addison-Wesley

Parliamentary Research Paper (2000) Rover: the story so far from Economic
Indicators, Research Paper 00/41, 3 April 2000

Pascale R T and Athos A G (1986) The art of Japanese management. Harmondsworth:

Penguin

Payne S L (1951) The art of asking questions. Princeton, N J: Princeton University

Press
Pilkington A (1999) Strategic alliance and dependency in design and manufacture: the
Rover-Honda case in International Journal o f Operations & Production Management
1999, Vol. 19, Iss. 5/6pp460-473

Pires S R I (1998) Managerial implications of the modular consortium model in a

Brailian automotive plant in International Journal o f Operations and Production
Management, Vol. 18, Iss. 3, pp. 221-232

Plutchik R (1983) Foundations of experimental research. London: Harper & Row

Polanyi K (1963) The great transformation: the political and economic origins of our
time. Boston, MA: Beacon Press

389
Poppo L and Zenger T (1 9 9 8 ) Testing alternative theories of the firm: transactinn cost,
knowledge-based, and measurement explanations for make-or-buv decisions in
information services in Strategic Management Journal 1998, Voll9, pp.853-877

Porter M E (1986) Competition in global industries: a conceptual framework in Porter

M E (ed.) Competition in global industries. Boston MA: Harvard Business School
Press

Pratten C F (1971) Economies of scale in the manufacturing industry. Cambridge:

Cambridge University Press

Professional Engineering (2003) Rover’s bumpy road to recovery in Professional

Engineering, 23 July 2003, pp. 26-27

Punch K F (2005)(2"^* ed.) Introduction to social research: quantitative and qualitative

approaches. London: Sage

PricewaterhouseCoopers (2005) Global automotive financial review, published by

PricewaterhouseCoopers, August 2005

Rae J B (1984) The American automobile industrv. Boston MA: Twayne Publishers

Rayome D and Baker J C (1995) Foreign direct investment: a review and analysis of
the literature in The International Trade Journal, Vol.9, No.l, Spring 1995, pp.3-37

Reliable Plant (2007) 2007 Harbour Report: Toyota leads in total productivity in
Reliable Plant, July-August 2007, accessed 1 August 2007, website
<http://www.reliableplant.com/article.asp?articleid=6572>

Rhys G (1972) Motor industrv: an economic survey. London: Butterworths

Ricketts M (1987) The economics of business enterprise: new approaches to the firm.
Brighton: Wheatsheaf Books

Rogers R P (1993) The minimum optimal steel plant and the survivor technique of
cost estimation'm Atlantic Economic Journal, Vol.21, No.3, pp.30-37

Rosegger G (1987) The economics of production and innovation. Oxford: Pergamon

Press
Rossman G B and Rallis S F (1998) Learning in the field: an introduction to
qualitative research. London: Sage

Rugman A and Hodgetts R (2001) The end of global strategy in European

Management Journal August 2001 Vol. 19, No 4, pp 333-343

Ryen A (2004) Ethical issues in Seal C, Gobo G, Gubrium J F and Silverman D

(eds.)(2004) Qualitative research practice. London: Sage

390
Schott P K (2003) One size fits all? Heckscher-Ohlin specialization in global
production in The American Economic Review, Vol 93, No.3, June 2003, pp.686-708

Sharratt B (2000) The men and motors of “The Austin”: the inside story of a century
of car making at Longbridge. Yeovil: Haynes Publishing

Shepherd W G (1967) What Does the Survivor Technique Show about Econom ies of
Scale? in Southern Economic Journal Vol 34, No. 1, pp. 113-122

Shimokawa K (2003) Production innovation at Ford of Europe’s Cologne Plant.

Nikkan Jidosha Shumbun December 13 2003, accessed 23 August 2006, website <
http://imvp.mit.edu/papers/0304/shimokawa_cologne.pdf>

SID (2006) R&D Centres: Tata Motors Ltd. Society for Innovation & Development,
Indian Institute of Science Campus, Bangalore. Website accessed 24 August 2006 <
http://sid.iisc.emet.in/tml.html>

Sieradski L (2002) Opel Corsa: the accidental world car. globalEDGE website <
http://globaledge.msu.edu/NewsAndViews/views/papers/opel_corsa_accidental_worl
d_car.pdf>, website accessed 27 November 2006

Silberston A (1972) Economies of scale in theorv and practice in Economic Journal,

Vol 82, No. 325, (Mar., 1972), pp. 369-391

Silver B (2003) Forces of labour: workers’ movements since 1870. Cambridge:

Cambridge University Press

Slater G and Spencer D A (2000) The uncertain foundations of transaction cost

economics in Journal o f Economic Issues, March 2000, Vol. 34, No. 1, pp. 61-87

Sloan A P (1986) Mv vears with General Motors. London: Penguin

Sloman J (2001) Essentials of Economics. Harlow: Pearson Education

Smitka M J (1991) Competitive ties: subcontracting in the Japanese automotive

industrv. New York: Columbia University Press

SMMT (2006) Motor Industrv Facts - 2005. London: SMMT

SMMT (2006b) Manufacturing plants profile: Toyota Motor Manufacturing (UK)

Ltd. SMMT press release, August 2006

Stadco (2006) Stadco signs contract with Nanjing Automobile Corporation (UK).
Press release, 16 October 2006, accessed 12 March 2007, website
< http://www.stadco.co.uk/news/2006-10-16>

Stake R E (1995) The art of case studv research. London: Sage

Stake R E (1998) Case studies in Denzin N K and Lincoln Y S (ed.s) (1998)
Strategies of qualitative inquiry. London: Sage

391
Stake R E (2000) Case studv research methods. University of Illinois, unpublished
lecture notes, accessed 13 April 2002 <http://www.ed.uiuc.edu/CIRCE/EDPSY399/>

Stigler G J (1958) Economies of scale in Journal o f Law and Economics, Vol. 1,

October, 1958, pp. 54-71

Sturgeon T J (2000) How do we define value chains and production networks?.

background paper prepared for Bellagio Value Chains Workshop, September 25 -
October 1, 2000, Bellagio, Italy

Sturgeon T and Florida R (2000) Globalization and jobs in the automotive industrv.
Final report to the Alfred P Sloan Foundation, New York

Sunday Times (2006) BAE to get Nimrod lift-off in The Sunday Times 16 July 2006
p3

Teece D J (1980) The diffusion of administrative innovation in Management Science,

Vol. 26, No. 5, May 1980, pp.464-470

Teece D J (1981) Internal organization and economic performance: an empirical

analvsis of the profitabilitv of principal firms in The Journal o f Industrial Economics,
Vol. 30, No. 2, December 1981, pp. 173-199

Tellis W (1997) Introduction to case studv in The Qualitative Report, Volume 3,

Number 2, July 1997

Thomas R J (1995) Interviewing important people in big companies in Hertz R and

Imber J B (1995)(eds.) Studving elites using qualitative methods. London: Sage

Thoms D and Donnelly T (1985) The motor car industrv in Coventry since the
1890’s. London: Croom Helm

Thompson A A and Formby J P (1993) Economics of the firm: theorv and practice.
London: Prentice-Hall International

Times 100 (2004) Nissan: planning for qualitv and productivity. The Times and MBA
Publishing Ltd, file accessed 18 September 2006, website < http://www.thetimes
100.co.uk/downloads/nissan/nissan_9_full.pdf>

Times (2007) Chinese deal further blow to Longbridge, accessed 10 February 2007,
website <http://business.timesonline.co.uk/tol^usiness/industry_sectors/
engineering/article 1303182.ece>

Tirole J (1989) The theorv of industrial organization. Cambridge, MA: MIT Press

Tolliday S (1998) The diffusion and transformation of Fordism: Britain and Japan
compared in Boyer R, Charron E, Jurgens U, and Tolliday S (eds.)(1998) Between
imitation and innovation: the transfer and hybridization of productive models in the
international automobile industrv. Oxford: Oxford University Press

392
Toyota (2007) Global Vision 2010. accessed 25 September 2007, website
<http://www.toyotatr.com/eng/gvision2010.asp>

Toyota (2007b) Toyota In The World 2007. accessed 25 September 2007, website
<http://www.toyota.co.jp/en/about_toyota/in_the_world/pdf2007/databook_2007.pdf

Truett L J and Truett D B (2001) The Spanish automotive industrv: scale economies and
input relationships in Applied Economics, Vol. 33, No. 12, October 2001, pp. 1503-1513
Truett L J and Truett D B (2003) The Italian automotive industrv and economies of
scale in Contemporary Economic Policy, Vol.21, No.3, July 2003, pp.329-337

UBS/Ricardo (2007) Is diesel set to boom in the US? from company website, accessed
15 October 2007, < http://www.ricardo.com/download/pdf/R119361S.pdf>

UNCTAD (2002) World investment report. 2002. New York and Geneva: United
Nations

Underwood J (1989) Will to win: John Egan and Jaguar. London: W H Allen

Vemon R (1966) International investment and international trade in the product cycle in
Quarterly Journal o f Economics, Vol.80, No. 2, pp. 190-207

Vemon R (1979) The product cycle hypothesis in a new intemational environment in

Oxford Bulletin o f Economics and Statistics, Vol.41, No.4, 255-267

Viitanen H (2002) Internationalization process and theories. Swedish School of

Economics and Business Administration, accessed 18 January 2007, website
<www.shh.fi/~polsa/inteory/>

VM Motori (2006) Current vehicles from VM Motori S.p.a. website, accessed 7

September 2006, website < http://www.vmmotori.it/en/01/01/index.jsp>

Volkswagen (2005) Aimual report 2005 from Volkswagen AG website, accessed 23

March 2007, website
<http://www.volkswagen.co.uk/assets/pdf/AnnualReport2005.pdf>

Walker G and Weber D (1987) Supplier competition, uncertainty, and make-or-buy

decisions in Academy o f Management Journal, 1987, Vol. 30, No.3, pp.589-595

World Business Council for Sustainable Development (WBCSD)(2004) Mobility

2030: Meeting the challenges to sustainability, from company website, accessed 15
October 2007, < http://www.wbcsd.org/web/publications/mobility/overview.pdf>

Weijian S (1991) Environmental risks and joint venture sharing arrangements in

Journal o f International Business Studies, 1991, Vol. 22, Issue 4

Wells P and Nieuwenhuis P (2001) The automotive industrv - a guide. London:

CAIR-BT

393
 Wells P and Rawlinson M (1994) The new European Automobile industrv.
Basingstoke: Macmillan

Whipp R and Clark P (1986) Innovation and the auto industrv: product, process and
work organization. London: Frances Pinter

Wilding R and Humphries A S (2006) Understanding collaborative suddIv chain

relationships through the application of the Williamson organisational failure
framework in International Journal o fPhysical Distribution and Logistics
Management, Vol.36, No.4, 2006, pp.309-29

Wilkins M (1979) The maturing of multinational enterprise: American business

abroad from 1914 to 1970. Cambridge, MA: Harvard University Press

Williams B (1997) Positive theories of multinational banking: eclectic theorv versus

intemalisation theorv in Journal o f Economic Surveys. March 1Q97^ Vol. 11 Iss 1
pp71-100

Williams K, Haslam C, Johal S and Williams J (1994) Cars: Analvsis. History. Cases.
Providence, RI: Berghahn Books

Williamson O E (1973) Markets and hierarchies: some elementary considerations in

The American Economic Review, Vol. 63, No. 2, May 1973, pp. 316-325

Williamson O E (1975) Markets and hierarchies : analvsis and antitrust implications.

A studv in the economics of internal organization. London: Collier Macmillan

Williamson O E (1981) The economics of organization: the transaction cost approach in

The American Journal o f Sociology 87, 3, November 1981 pp548-577

Williamson O E (1983) Credible commitments: using hostages to support exchange in

The American Economic Review 73.4 September 1983 pp519-540

Williamson O E (1987) Antitrust economics: mergers, contracting and strategic

behaviour. Oxford: Basil Blackwell

Williamson O E (1991) Comparative economic organization: the analvsis of discrete

structural alternatives in Administrative Science Quarterly, 1991, Vol. 36, pp.269-296

Williamson O E (1993) Opportunism and its critics in Managerial and Decision

Economics, Vol. 14 (1993), pp.97-107

Williamson O E (1996) Economic organization: the case for candor in Academy of

Management Review, 1996, Vol. 21, No. 1. 48-57

Williamson O E (1998) Transaction cost economics: how it works: where it is headed in

De Economist, 1998, Vol. 146, No. 1, pp. 23-58

Williamson S and Wright R (1994) Barter and monetarv exchange under private
informniion in American Economic Review, 1994, Vol.84, No.l, pp. 104-123

394
Womack J P, Jones D T, Ross D (1990) The Machine That Changed the World. New
York: Rawson Associates/ MacMillan

Wood J (1988) Wheels of misfortune: the rise and fall of the British motor industrv.
London: Sidgwick and Jackson

World Economic Forum (2002) Mahathir Bin Mohamad, Malaysian Prime Minister.
Speech to The World Economic Forum - East Asia Economic Summit 2002 reported in
the website for Prime Minister’s Office of Malaysia, accessed 15 June 2003, website
<http://www.smpke.jpm.my/website/webdb.nsf/70pendatabase>

Yamaguchi J K (2003) The Mazda RX-8: The World’s First 4-door. 4-seat Sports
Car. Kanagawa: Ring

Yamin M (1991) A reassessment of Hvmer’s contribution to the theorv of

transnational corporations in Pitelis C and Sugden R (eds.)(1991) The nature of the
transnational firm. London: Routledge

Yin R (1994) Case studv research: design and methods. London: Sage

Yin R (2006) Case studv methods in Green J L, Camilli G and Elmore P B (eds.)
Handbook of applied social research methods. London: Sage

Youn Kim H (1986) Economies of Scale and Economies of Scope in Multiproduct

Financial Institutions: Further Evidence from Credit Unions in Journal o fMoney,
Credit, and Banking, May 1986, Vol. 18, No. 2, pp. 220-226

Zenger T R (1994) Explaining organization diseconomies of scale in R&D: agency

problems and the allocation of engineering talent, ideas, and effort by firm size in
Management Science, Vol. 40, No. 6, June 1994pp. 708-729

395
 Appendix 1

Record of data collection

Table A l.l - MG Rover

Organisation Participants Collection Date Record

MGR PR Stewart McKee 5/3/03 Field notes

MGR PR Greg Allport 26/05/04 Field notes

17/07/06 Field notes
5/3/03 Field notes
Retired John Bacchus 10/6/03 Field notes
Executive/Consultant 5/08/03 Email
Head of Global Alistair Morton Transcription
19/06/03 Field notes
Training John Kiff 2/7/03 Transcription
Business School
Consultant 5/8/03 Transcription
Head, HR Rob Ball
Development
Director, HR Bob Beddow 29/08/03 Transcription
Retired Executive Peter Woods
Head of Strategy Dr Chris Millard 17/9/03 Transcription
(with Stewart 6/1/04 Transcription
McKee)
Dr Chris Millard
Director - Product Rob Oldaker 5/10/05 Letter
6/1/04 Transcription
Development 26/05/04 Field notes
Intemational Sales, John Tweedy 24/11/04 Letter
8/8/05 Email
21/04/04 Transcription
Head Paul Stowe 8/02/07 Online conference
NAG, Head of
Quality Confidential interview
NAC-MG senior 19/02/07 Transcription
manager Anthony Smith 28/03/07 Email
Ricardo PLC 27/02/07 Field notes
media relations

396
Table A1.2 - Contextual Industry

Organisation Participants Collection Record

Position Date
Morgan
Assistant Managing Mark Aston 12/7/02 Field notes
Director 7/03/05 Letter
Redwood Pat Devereux 5/3/03 Field notes
Analyst
TGWU
National Secretary David Osbome 13/3/03 Field notes
Amicus-AEEU
National Secretary Duncan Simpson
Rover and ME? Malcolm Harbour 25/7/03 Transcription
SMMT
Policy and Economics Matthew Croucher 28/08/03 Transcription
Kyoto University
Faculty of Economics Professor Hiromi Shioji 24/10/03 Transcription
Kyoto University
Graduate School of Yasuo Sugiyama 24/10/03 Not transcribed
Economics Associate professor
Jaguar Japan
President David Blume 27/10/03 Transcription
Mitsubishi Research
Institute Ryuichiro Inoue 28/10/03 Transcription
Research Director
JAMA
Deputy Director Shigeru Sasaki 29/10/03 Transcription
Intemational Analyst Takehisa Yamakawa
Honda Tokyo
Assistant Manager PR Tatsuya lida 29/10/03 Transcription
METI
Deputy Director Sadanori Ito 30/10/03 Transcription
Aston Martin
Director PR Tim Watson 3/12/03 Transcription
Bentley
Director, Sales Adrian Hallmark 20/1/04 Transcription
Bentley
Director, Manufacturing Doug Dickson 20/1/04 Transcription
Bentley
Director, Engineering Ulrich Eichhom 20/1/04 Transcription
Honda Europe Takeshi Sumita 6/2/04 Transcription
Director, PR
LERC Prof. Daniel Jones 13/02/04 Transcription
Honda, Japan
Deputy Director, Strategy Saburo Kobayashi 19/2/04 Transcription
Chief Engineer, Strategy Hiroshi Kojo
Chief Engineer, Strategy Kenji Nakamura
Manager, Strategy Hitoshi Moriguchi

397
 Organisation Participants Collection Record
ACEA, Japan Date

Director General Anthony Millington 20/2/04 Transcription

Honda, Japan
ex-CEO Hiroyuki Yoshino 26/2/04 Transcription
Osaka Sangyo University
Business Faculty Prof. Tsutomu Demizu 27/2/04 Transcription
Automotive Engineering 27/2/04
Intemational (SAE Jack K Yamaguchi Transcription
publication)
Honda, Motegi 28/2/04 Transcription
General Manager Yutaka Ikeda
Chief Engineer, B1 Kunihasa Watanabe
TGWU David Osbome 16/03/04 Transcription
National Secretary
Rolls-Royce Colin Hutchinson 26/03/04 Transcription
General Manager, Sales
Toyota Museum Koji Yamada 8/06/04 Transcription
Group Manager
Toyota Daigo Umeki 8/06/04 Transcription
Group Manager, PR

398
 Appendix 2

Guide to Global Automotive Partnerships: How the World’s Automakers are connected (Automotive News September 2004)

Parent Full Ownership Equity Stakes Vehicle Assembly Alliances Technical/Parts Alliances
Group
Owns Owned By JV Contract Assembly
DaimlerChrysler - small
BMW BMW AG; Quandt family - 46.6% BMW Brilliance Magna Steyr,
Mini Others - 53.4% Auto, China Austria gas engines, Brazil MGR
Rolls-Royce - diesel engines, UK

PSA - small gas engines

Toyota - diesel engines
BMW - small gas
Daimler Chrysler McLaren - 40% Deutsche Bank -12% Beijing Automotive Karmann, Germ.
engines, Brazil Magna
Chrysler Mercedes-Benz (incl. Mitsubishi - 24.7% Kuwait Investments - Industrial Holding Magna Steyr, Aus.
Steyr - powertrains,
100% of Maybach, 7% McLaren Group,
Austria Ssangyong -
Smart) Other shareholders - UK
81% engines
Dongfeng Dongfeng Motor Dongfeng Dongfeng Honda
Motor Corp Corp: Automobile Co - Automobile (Wuhan)
Dongfeng Liuzhou 70% Donfeng Motor
Motor Dongfeng Peugeot
Citroen
Dongfeng Yueda Kia Maggiora, Italy
GM-
Fiat Fiat SpA Fiat Auto - 90% Agnelli family - 30% Nanjing Fiat-Yuejin
powertrains/purchasing,
(Fiat Auto Owns Public - 70% Motor Pininfarina, Italy
Eur and S. America
100% of Lancia and PSA-SEVEL, Italy,
Alfa Romeo) France
Ferrari - 56% (incl. Renault-Teksid, Italy
100% of Maserati) Suzuki, Hungary
(2005)
Tofas, Turkey

399
First First Automobile FAW Car-64% FAW Hainan
Automobile Works (FAW): FAW Huali (Tianjin)
Works Changchun FAW FAW-VW
Group Fengyue Auto Tianjin FAW Toyota
Tianjin FAW Xiali
Ford Ford Motor Co: Mazda-33.4% Common Stock - 60% AutoAlliance, US, Ford Otosan, Ford-Otosan Engine, Tur
Aston Martin Class B (family) - 40% Thailand Turkey Getrag-transaxles, France
Jaguar Changan Ford, China Otokar, Turkey Getrag-transmissions,
Land Rover Jiangling Motors, Pininfarina, Italy Germany & UK
Volvo China Santana, Spain MGR - engines
Thai-Swedish Nemak-aluminium, Can
Assembly, Thai. PSA - diesel engines
Fuji Heavy Fuji Heavy Industries Suzuki Motor - G M -20%
Industries (Subaru) 1.1% Suzuki - 2.7%
Other shareholders - 77.3%
GM General Motors: Fiat Auto - 10% Shareholders -100% AvtoVaz, Russia AM General, US Fiat - powertrains,
Adam Opel Fuji H I-20% CAMI, Canada Bertone, Italy purchasing, Eur & S
Holden GM Daewoo Delta Motor, S Africa Heuliez, France America
Saab Automobile Auto & T ech - Isuzu-GM, Australia Industries Honda - gas engines
Vauxhall Motors 44.6% Jinbei GM, China Mecaniques Isuzu - powertrains
Isuzu- 12% NUMMI, US Maghrebines, Pan Asia Technical
Suzuki Motor - SAIC GM Wuling Tunisia Automotive Center
20.3% Shanghai GM Magna Steyr, Aus Shanghai GM Dong Yue
Shanghai GM Dong Automotive powertrains,
Yue China
GM- GM Daewoo Auto GM -44.6% GM Holden - engines
Daewoo and Technology Suzuki -14.9% Suzuki Motor VM Motori - engine
SAIC - 10.6% design
Creditors - 29.9%

400
Honda Honda Motor Japan Trustee Services Bangchan General GM - gas engines
Motor Co Company B a n k -7.1% Assembly, Thailand
Master Trust Bank of Dongfeng Honda
Japan - 4.9% Guangzhou Honda
Odier shareholders - Honda Atlas, Pak
88% Honda Auto, China
Hyundai Hyundai Motor Co Beijing Hyundai Hyundai Mobis - DC, Mitsubishi Motor -
Motor Co Motor - 50% 14.5% engines
Dongfeng Yueda Kia INI Steel-5.3% VM Motori - engine
Automobile - 50% Chung Mong Koo - design
(Kia) 5.2%
Kia Motors - 47.4% Others - 75%
Isuzu Isuzu Motors G M -12% GM - powertrains
Motors Japan Trustee Services Industries Mecaniques
Bank-7.5% Maghrebines, Tunisia
Other shareholders -
80.5%
Mazda Mazda Motor Corp Ford Motor - 33.4% AutoAlliance, US & Colombiana
Motor Japan Trustee Services Thailand Automotriz,
Bank - 9.6% Suzuki Motor Colombia
Others 57%
MG Rover MG Rover Group: Phoenix Venture Tata, India BMW - diesel engines
(2000- Qvale Holdings-100% Ford - engines
2005) (MGR dealers and Lotus - engines
employees - 60%; J SAIC - development
Towers, N Stephenson,
J Edwards, P Beale -
40%)

401
Mitsubishi Mitsubishi Motors: Phoenix Capital Assoc Beijing Automotive Pininfarina, Italy Asian Transmission,
Motors Netherlands Car -33.3% Industrial Holding Philippines
(NedCar) DaimlerChrysler - DC, China
24.7% Hunan Changfeng,
Mitsubishi HI - 9.9% China
Mitsubishi Corp - 3.5% Proton
Other shareholders -
28.6%
Nissan Nissan Motr Dongfeng Motor - Renault SA - 44.4%
Motor 50% Japan Trustee Services
Renault SA - 15% Bank-5.5%
Zhengzhou Nissan Nissan (self owned) -
Automobile - 30% 2.7%
Other shareholders -
47.4%
Porsche Porsche AG Porsche and Piech VW, Slovakia Valmet, Finland VW - engines
families - 50%
Institutions - 30%
Public - 20%
PSA PSA/Peugeot-Citroen Dongfeng Peugeot Peugeot family - 29.2% Fiat/SEVEL, Italy & SOMACA, BMW - small gas engines
Peugeot- Citroen Automobile - Others - 70.8% France Morocco Ford - small diesel
Citroen 32% Toyota Peugeot engines
Citroen, Czech Renault - engines and
Renault Renault SA: Avtoframos - 62% Public-61.2% Fiat/Teksid, Italy transimissions
Bursa, Turkey Bursa - powertrains, Tur
Dacia (99.3%) Nissan Motor - Nissan Motor -15% GM, Europe IDRO, Iran Cofal - LCV engines. Bra
44.4% Other shareholders - Nissan, Spain SOFASA, Isuzu - diesel engines PSA
Samsung-70.1% 23.8% Dongfeng, China Colombia - engines and
SOMACA, transmissions
Morocco

402
Shanghai Shanghai Automotive GM-Daewoo Auto & Jinbei GM MG Rover -
Automotive Industry Corp (SAIC) Technology - 10.6% SAIC GM Wuling development, UK
Industry Shanghai GM (Ricardo 2010 from 2005)
Corp Shanghai GM Dongyue
Shanghai VW
Suzuki Suzuki Motor Corp Fuji Heavy - 2.7% GM -20.3% CAMI, Canada
Motor GM-Daewoo Auto & Fuji Heavy - 1.1% Chongqing Changan
Technology 14.9% Chase Manhattan Suzuki, China
Maruti Udyog - B a n k -8.1% Fiat, Hungary (2005)
54.2% Other shareholders - Jiangxi Changhe
70.5% Suzuki, China
Mazda, Japan
Mitsubishi, Japan
Nissan, Japan
Santana, Spain
Toyota Toyota Motor Corp Daihatsu Motors - Master TB of Japan - Indus Motor, Pakistan BMW - diesel engines
51.2% 7.8% Kuozui Motors, Taiwan FAW Toyota Changchun
FHI - 8.7% (Oct Toyota (self owned) NUMMI, US Engine, China
2005) -7.7% Sichuan Toyota, China Guangqi Toyota Engine
Isuzu-5.9% (Nov Japan TSB - 7.5% Tianjin FAW Toyota Tianjin Fenjin Auto Parts
2006) Toyota Industries - Toyota Peugeot - components, China
5.4% Citroen, Czech Tianjin Toyota Motor
Others-71.6% Engine, China
VW VWAG: Institutional abroad - FAW-VW, China
Audi AG (99.1%) 34.1% Ford, Portugal
(Audi owns 100% Private shareholders Porsche, Slovakia
Lamborghini) -33.1% Shanghai VW, China
Bentley Motors State of Lower
Bugatti Saxony - 13.7%
Seat VW (self owned) -
Skoda 9.8%
Institutional Germany
-9.3%

403
 Appendix 3
15 November 2002

Dear Mr McKee,

I was recommended to contact you by Teresa Kelly in the Education Partnership department.
I am a PhD student at the Cardiff Business School where I am looking at globalisation in the
automotive industry. I have chosen MG Rover as my case study, partly because of my
personal enthusiasm for the company and partly because the company has to have the most
exciting potential of any firm in the industry today. However, my own energy alone is not
enough to carry the research through and I need to ask for MG Rover to indulge me by
granting permission to conduct interviews.

I have spoken to Teresa about this and provided her with a list of the kind of questions I
would like to ask; I would be delighted to have the opportunity to send a copy to you. I am
looking at how MG Rover can use the process of globalisation for its own ends and remain
successfully independent, instead of being swallowed by an anonymous multinational. I do
not believe that my research would be a risk to the company’s need for commercial secrecy
but I would of course respect all information as confidential. I recently spoke with Gavin
Thompson, who was tremendously helpful, and he is exactly the kind of person with whom I
would like to meet.

Ultimately, I very much hope that my research will prove useful to the company even if over
the course of the next two years of research the potential benefit may not be obvious. If you
require me to clarify any points please do not hesitate to contact me.

Yours faithfully,

Michael Wynn-Williams

Tel: 029 2087 4001

404
 Appendix 4

Ca r d i f f
UNIVERSITY

PRIFYSGOL

CAERDYgl

Tel: +44 (0) 29 2087 4001

1^^ September 2003

Dear Sir,

I am a PhD researcher in the Japanese Studies Centre at Cardiff Business School looking
at globalisation in the car industry. I have taken as my case studies MG Rover and
Honda, looking at how they have progressed since their collaboration in the 1980s. It is
well known that Honda have gone on to becoming a major global force in the industry
with operations all over the world and I am seeking to gain an overview of the factors
that have led to their intemational success. I believe that their strategy may come to
represent a new business model.

It is my consideration that a great deal of the company’s success is due to the specifics of
its Japanese origin. I am not simply referring to national cultural mores but the particular
circumstances from which Honda emerged. This includes supply and demand
characteristics, national and company skill bases as well as governmental factors. It is my
belief that while much of Honda’s success is translatable to its foreign operations it is
still important for the company to retain its core capabilities at its place of origin.

For this purpose I need to have a clear understanding of the context within which the
company operates, from the progressive intemationalisation of the Japanese automotive
industry to the attitudes of the government towards globalisation. It is my hope that you
will grant me the opportunity to conduct interviews so that I may benefit from the
experience of experts in the field. This will provide invaluable credibility for my research
and so will ultimately contribute to the future prosperity of both Japan and the United
Kingdom. Since I have lived in Japan for several years and come to love the country as a
second home it is my sincere wish that I be able to make a contribution to both nations.

I hope the accompanying information gives you some idea of the purpose of my research.
Naturally I would welcome the opportunity to discuss the finer details of my information
needs with you at your convenience. Please feel free to contact me, either by mail, email
or telephone.

I look forward to hearing from you.

Yours sincerely,

Michael S Wynn-Williams

405
 Appendix 5

Mr P Woods
August 8^ 2003

Dear Mr Woods,

Thank you for the telephone conversation today. As I mentioned you were recommended
to me by Malcolm Harbour who had kindly allowed me to interview him. I have also had
extensive contact with John Bacchus and various people at MG Rover and Honda.

I am a PhD researcher at Cardiff Business School looking at globalisation in the car

industry, working in the same department as Professor Garel Rhys. My research concems
MG Rover and Honda, specifically how they have faired within the global industry since
their original collaboration. The actual details of the alliance are not the focus of my
study, but the relationship gives me the opportunity to compare two independent
companies that at one time shared a common view of the global industry. I am interested
in how the two companies are working within the same global trends to maintain
independence and I believe that the evolving strategies may come to represent a new
business model.

John Bacchus has been, and continues to be, very helpful in illuminating the period up to
the alliance. Members of the companies have been talking to me about the current global
opportunities. However, I am also keen to discover the Japanese angle. For example I
find that Honda, despite having functionary departments spread around the world, retain
all their core capabilities within Japan. It would appear to me that they are still very much
Japan based and I would like to know how this influences their intemational competitive
position. Furthermore I believe that both the industry and the consumer benefit from
independent minded companies that retain their domestic orientation. I would like to find
that there is a room in the world for a Japanese Honda and a British MG Rover, indeed
that there is a necessity for them to remain. However, to do that I need to get to the roots
of these individual companies and the intemational context within which they operate.

Clearly you are the expert on the Japanese part of the equation and I am sure you will be
able to provide critical insights. I would be grateful of the opportunity to interview you,
for which I would be happy to provide a list of topics and questions in advance by way of
preparation. In the meantime, if there is any thing you wish me to clarify please do not
hesitate to contact me.

I look forward to hearing from you.

Yours faithfully,

Michael S Wynn-Williams

406
Appendix 6
Power of a parent (1^ The Power of
a Parent
 Fiat acquires Flat acquires
60% of Ferrari 90% of Ferrari
5000

3750

2500

1250

0
ooo)OT-cNcO'5ui<ot^ooa)OT-cNjco-5in<ot^ooo)OT-cNicoTrio(Dr^ooo)gT-(\co3^^(Dr^ooa)0^<Nco^iO(Dr;-oog)OT-
55ioiniou^inininininincocpcp^®<bcocD<D<£>h'^-r^h-N-t^t^r^r^r^ooooooooooooooooooooojO)0>gjOTO>0)gjo)222
a)0>0)0)0)0>0i0)0)0505C3505a505$0)0)OT050)05o0>c»cna)C50)0>0)0)050)C50J0)0)a)0)0)cnOTa)0>0)0)C»0)0)0)0)oo
t-l-T-'.-'^T-'.-T-T-'.-T-T-T-'r-T-T- T- T-T-1-T-T-fT-T-V-T-T-T-'.-T-T-T-''^T- .trT-T-T-T: T-T-T-T-T-T-T-T-T-T-T-T-CMCM

— AML Production — Ferrari Production

407
 Appendix 6
Power of a parent (2^
Aston Martin -
Production 73-’01
V12
1500
DB7 Vantage
Launched

DB7
1125 Launched
Ford Acquires
remaining 25% of
Aston Martin
Ford Acquires
750 76% of
Aston Martin

375

0
co'^Locoh
r v . f ^ f s ^ | s ^ l
-coojOT-cMco'^LOCDr^ooojo^ojco^tiGSQ^SSSj:^
s . f s . | s ^ C 3 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 ) C D 0 ) 0 ) 0 ^ 0 > 0 > C D 0 > 0 ) O O
0>0)0)0>0) 0 5 0 5 C 3 5 0 > C D O ) 0 5 0 ) 0 5 a ) 0 0 ) 0 ^ 0 ) 0 ) 0 5 0 ) 0 > 0 5 0 ) a ^ O ) 0 0 t- t- t- t- t- t- ' t- t- t- t- x- t- t-
t- t- t- t- t- t- t- t- t- t- t- t- t- t- C sI CN

408
 Appendix 7

Morgan Factory Visit

12 July 2002

Semi-structured Interview
With Mark Aston

Global Strategy

1. Does Morgan have a globalisation strategy?

2. How did Morgan develop its approach to globalisation? Who was involved?
3. How does Morgan see the future of globalisation?
4. What are the benefits of independence for Morgan? What is the future?
5. Has Morgan considered a merger with another manufacturer?
6. In what areas does Morgan most need to find a collaborative partner?

Politics

1. Does Morgan benefit from globalisation of standards/certifications?

2. Does Morgan receive govemment/EU grants?

Production Process

1. How much does Morgan manufacture in-house?

2. Has Morgan been influenced by intemational methods of production eg JIT?
3. Is Morgan capable of simultaneous engineering?
4. How flexible is the production line?
5. Would Morgan consider manufacturing overseas?

Supply

1. To what extent does Morgan work with suppliers? Is it free to choose suppliers?
2. Were BMW actively involved with the design of the Aero 8? Did it impose
restrictions? What did BMW gain?
3. What proportion of supply is of foreign origin?

Market Demand

1. Can Morgan support global sales?

2. Can Morgan use BMW sales/service outlets?
3. What proportion of production is exported?
4. What provision is made for exchange rate fluctuations?

X Factor

1. What is Morgan's main competitive advantage? Who/what are the competition?

2. What core skills does Morgan have/lack?

409
 Appendix 8

BMIHT June 2003

Interview Protocol
Interviewee: Mr John Bacchus Date: 10^ June 2003

Company: BMIHT/Rover Place: Gaydon

Position held: Head of relations with Honda

Contact details:

1975
Key Issues:
Ryder Report
CPRS
NEB
Points for consideration:

1. Had British Leyland failed to consolidate an intemational presence?

2. Had there been the possibility of an intemational joint venture?
3. Had the threat of foreign competition been fiilly appreciated by the reports?
4. What lessons could have been leamt from intemational competitors at this stage?
5. Did the Ryder Report take into consideration the importance of the global
industry?
6. Why was Ford consistently profitable? Did its intemational presence give it a
particular advantage?
7. What would consumers have lost had BL disappeared?
8. Did the failure of BL show that mergers necessarily involve a total loss of
independence?
9. Was the survival of BL important for the British economy as a whole? Was it
important for the global industry?
10. Should the government aid have been restricted purely to a financial safety net?
Was the rescue of Rolls Royce a better model to follow?

1980

Key Issue:
Collaboration with Honda

Points for consideration:

1. Were BL and Honda looking to gain from collaboration in the same way, or were
their intentions complementary?
2. Was there something uniquely Japanese about Honda that made them an ideal
partner? Did Honda need something xmiquely British from BL?

410
3. Was independence still a preferred state for BL?
4. Was decentralisation an admission by BL that a degree of autonomy within the
company was good for business? Can we say the same for joint ventures?
5. Were the products (e.g. new Metro) of international standing?
6. Was there a strategy for maintaining an international presence?

1985
Key Issue:
Memorandum of Understanding with Honda (one of many?)

1. Did the closer relationship represent a continuing loss of competency by Rover?

2. What did Rover gain? What did Honda gain?
3. Was Rover becoming too Japanese? Better Japanese than French or German?
4. Could the relationship have been extended to international locations e.g. joint
manufacturing in the US?
5. Would increased government support have been more beneficial to Rover, rather
than closer collaboration with a partner?
6. Was Rover competitive in international markets? Was this thanks to Honda?

1990
Key Issue:
Ownership by BAe (from 1988)

1. Did Rover gain If om being part of a company with international standing? Did it
increase its global presence?
2. Did Rover benefit from having a British parent, as opposed to being sold to a
foreign company?
3. Did Rover find synergies with its parent? How was Rover reshaped?
4. Did Rover intend to collaborate ever closer with Honda?
5. Did Rover intend to maintain independence?
6. Was the government still involved?
7. How was Honda being reshaped by the changing relationship? Did it learn lessons it
could use globally?

1995
Key Issue:
Ownership by BMW

1. What did Rover gain from being part of an international company?

2. Was Rover given too much independence?
3. Was Rover constrained by being owned by a foreign rival?
4. Was there something uniquely British about Rover that attracted BMW? Did they
obtain it?
5. What did BMW contribute that no other company could have done? Was it
because they were German?
6. Would continued collaboration with Honda have brought greater benefits instead?
7. How did BMW ownership reshape the company?
8. Was Rover becoming more international again?

411
2000
Key Issue:
Independence

1. What did MG Rover save by being sold? Was it too late?

2. What did BMW leam that was uniquely British? Did this change BMW?
3. Was the involvement by government to the advantage of MG Rover?
4. What would have been the future of MG Rover had BMW not sold it?
5. How has independence reshaped the company? Does it retain a full range of
competencies?
6. What should the global industry leam from the experiences of BL/Rover/MG
Rover?

2005
Key Issues:
Long term viability
Intemational partners
Intemational sales

1. Is it to MG Rover’s advantage to remain independent? What does independence

mean within the global industry?
2. What does it mean for a company to be “globalised”?
3. Do foreign manufacturers have unique advantages?
4. Does the global industry need MG Rover?
5. What will the future competitive pressures be?
6. What are the most pressing challenges/costs within the industry?
7. Is it possible for a company of MG Rover’s size to survive? What level of
production output and model range is necessary?
8. What are the core competencies necessary for MG Rover’s long term success?
9. What kind of collaboration would bring long term survival?
10. Should the UK government be more supportive?
11. Is it to the customer’s advantage for MG Rover to remain independent/British?
12. Should MG Rover remain fundamentally British? Does the UK industry have a
uniquely British contribution to make to the global industry?

412
 Appendix 9

Mr Rob Oldaker

Interview Protocol
Interviewee: Mr Rob Oldaker Date: 06/01/04
Company: MG Rover Place: Longbridge
Position held:
Contact details: 0121 482 3397

Globalisation

1. Is MG Rover at a good economic size with regards to:

i) R&D
ii) Production levels (eg break even levels)
iii) Marketing and distribution (eg resources to open new markets)
2. How can MG Rover “punch above its weight”?
3. Is globalisation the death of innovation? Does it destroy differentiation of
products?
4. Will consumers begin to react against globalisation?
5. What would MG Rover benefit from a globalised industry? Is the MG ZT 260 a
good example of this?
6. What should MG Rover avoid in a globalised industry?
7. Within a spectrum of high value to low value, which products benefit least from
globalisation?
8. Does the purchasing of technologies on the global market mean that MG Rover no
longer has the ability to be a market leader/maker?
9. To what extent can MG Rover move its products around the world to accommodate
changes in other markets (e.g. selling an ageing model into China to extend the
model life)?
10. Can MG Rover sustain the benefits it gained from the relationship with Honda and
BMW?
11. If you were to create the perfect car company by cherry picking aspects of existing
firms, what would you choose and from whom?
12. What proportion of components are imported?
13. Does MG Rover have the resources to enter a new market by itself?

UK Industry

1. Does the UK industry need MG Rover to maintain its core capabilities?

2. Should the UK Government give preferential treatment to MG Rover?

Indian Industry

1. Is the Indian market the next global opportunity, after China?

2. What can they leam from UK manufacturers?
3. What lessons can MG Rover leam from TATA that can be applied at home (cf.
Honda’s influence on production technology)?

413
 4. To what extent is the CityRover a pilot study for a deeper relationship?
5. Does MG Rover have reciprocal access to the Indian market?

Joint Ventures

1. What are the advantages of independence that are unique to MG Rover?

2. Does flexibility mean that MG Rover is a short-term planner?
3. The president of Jaguar Japan described the rescue by Ford thus: Jaguar was a raft
being tossed about in a storm, the good ship Ford rescued it and took it on board. In
the future Jaguar will be refloated and will become part of the Ford flotilla (i.e.
independent but part of a larger group). Would you like this to happen to MG Rover?
4. Do you have a team specifically looking for joint ventures? Do you use formal
processes?
5. Do joint ventures spread the management talent too thinly?
6. Can the TATA collaboration only be for low value products? Should MG Rover
develop other collaborations concurrently?
7. Regarding the following two kinds of joint venture:
i) A few comprehensive joint ventures or a network of specific projects
ii) Long term or short term relationships
Do these represent the choices facing MG Rover?
8. Chris Millard mentioned two types of relationship:
i) Roverising a product already being made elsewhere, adding to their existing
scale (eg. CityRover)
ii) Equal partners, where you both have to strive to achieve scale.
Which is most suitable for MG Rover?
9. What kinds of problems were evident from working with a partner? Were the
problems generic or specific to this partner?
10. Does MG Rover have purchasing power with regards to suppliers?
11. What proportion of new car projects are contracted out (cf. SV and RD60)?
12. Does MG Rover need to make itself attractive to potential partners/parents?
13.

The Brand

1. How important is authenticity (i.e. British made and designed)?

2. Who buys MG Rovers?
3. How do you manage the brand image in different markets (eg relative success in
Mexico and Germany)?
4. How close do you need to be to your customers (especially foreign markets)?
How do you communicate with your customers?
5. Are product features paramount, or can you extend the brand image over mediocre
products?
6. Should MG Rover develop more variants (eg 75 Coupe, 4WD etc)
7. How does the SV fit with MG Rover?

NB: Request Japanese contacts

414
 Appendix 10

Fieldwork Japan 2003

Interview Protocol
Interviewee: Date:
Company: Place:
Position held:
Contact details:

Globalisation

1. Is globalisation an inevitable process?

2. How far will globalisation extend?
3. Does company size matter?
4. Will globalisation cause the death of innovation?
5. Do the industry and the consumer need independent manufacturers?
6. Should manufacturers be locally based?
7. What does Honda benefit from a globalised industry?
8. What should Honda avoid in a globalised industry?
9. Within a spectrum of high value to low value, which products benefit least from
globalisation?
10. Is it a paradox that high value products need to be close to their customers but in
foreign markets this would dilute the core brand values (i.e. would a Honda
assembled in China not be considered an authentic Honda by the customers and so
weaken the Honda reputation)?

UK Industry

1. What does Honda expect to gain from a UK operating base?

2. What are the strengths and weaknesses of the UK industry now?
3. How can UK car manufacturers continue to benefit from Honda’s influence?

Japanese Industry

1. Does the Japanese industry represent a threat to the UK industry?

2. Does Honda consider Japan to be the home base?
3. What can Honda leam from UK manufacturers?
4. Will the Japanese market be as open to imports as the UK market?
5. Could UK manufacturers ever start manufacturing bases in Japan?
6. How will the Japanese industry change in the future?

415
Joint Ventures

1. Taking the following companies as examples of the opportunities and dangers of

globalisation, what would the implications be for Honda in evaluating the need to
stay independent
i) Saab under GM
ii) Alfa Romeo under FIAT
iii) VW's sophisticated badge engineering

2. What core capabilities should Honda defend in any collaborative deals?

3. How did the experience with Rover change Honda’s attitude to joint ventures?

Government

1. Should the Japanese Government maintain an active interest in Honda?

2. Should the EU take an active interest in Honda?
3. Is Honda a local European company in Europe?

416
 Appendix 11

Dr C Millard
Head of Product Development and Strategy
MG Rover Group
PC Box 41
Longbridge
Birmingham B31 2TB

August 7**’ 2003

Dear Dr Millard,

I am a PhD researcher at Cardiff Business School looking at globalisation in the car

industry, working in the same department as Professor Garel Rhys. Like him I have a
great interest in the fortimes of MG Rover and so it was a natural choice for me to take
the company as my case study. My research is focused on the period since the
collaboration with Honda in the 1980s. MG Rover provides unique insights into
globalisation over this period since it is the only manufacturer of any note that has shown
new and original approaches to the perennial problems of the industry. I am interested in
how MG Rover are using the same global trends to maintain independence and I believe
that the evolving unique strategy may come to represent a new business model. It goes
without saying that product development lies at the very centre of the company’s future.

Bob Beddow and Rob Hall of the Human Resources department were kind enough to
give me a great deal of guidance on how the company has been shaped by events since
the 1980s. I have also had extensive contact with Stewart McKee. However, as my
research concerns the very core of the company’s purpose, it is crucial that I be able to
obtain first hand information from you. Your input would provide the credibility that my
project requires to achieve recognition. Naturally I am happy to fit in with your busy
schedule, and I would provide you with details of discussion topics in advance.

As I have made clear to those I have interviewed previously I am quite willing to respect
commercial sensitivities by formally agreeing to any requirements for confidentiality and
anonymity. However, I am more concerned with the generalities of MG Rover’s approach
to global opportunities and threats rather than specific product plans. My main priority is
to complete this research so that it would also benefit the company in which I share such
an abiding interest. I realise that before making a commitment to helping me with this
project you may wish me to clarify any points so please do not hesitate to contact me. My
office telephone number is:

029 2087 4001

I look forward to hearing from you.

Yours faithfully,

Michael S Wynn-Williams

417
 Appendix 12

Codes

Topical Codes Modifier Codes Intensifier Codes

Globalisation Trade Positive: +
Globalisation: GL Trade: TR
Foreign Direct Investment: EDI Export: EX Negative: -
Local/Domestic: LC Import: IM
Protection: PR Query: ?
National Industry Regulation: RG
National industry: NI Representation: RP
Direction/Guidance: DR

Strategy
Strategy: ST
Government Future: F
Government: GV Learning: L
Joint venture: JV
Vertical joint venture: VJV
Horizontal joint venture: HJV
Potential partner: PP
Action outside joint venture: IJV
Merger: MG
Industry
Industry: I
Supply: SU
Transplant: TR
Kit Assembly: CKD
Research: RS
Human Resources: HR
Management: MT
Intellectual property right: IPR
Technical
Technology: TC
Design: D
Manufacturing: M
Flexible production: FL
Company Structure
Vertical Integration: VI Economic
Horizontal Integration: HI Economies of scale: ES
Independent: I Financial: FC
Multinational company: MNE Environmental issues: E
Market
Brand: B
Market: MK
Competition: CT
Regional
Africa: AF India: IN
Australasia: AU Iran: IR
China: C Russia: R
Europe: EU United Kingdom: UK
Japan: J United States: US
Culture
Culture: C
Company specific: CS

418
 Appendix 13

Global Manufacturers; Rank by 2006 Output

R an k C om pany O u tp u t
1 T oyota 9,018,000
2 GM 8,743,248
3 Ford 6,563,092
4 V o lk sw agen
5 Daim ierChrysler 5,659,578
6 Hyundai-Kia 4,589,100
7 H onda 3,778,166
8 P eu geot-C itroen 3,633,813
9 N issa n 3,357,000
10 Fiat 3,238,346
11 R enault
12 Suzuki 2,363,968
13 BMW 2,346,319
14 M itsubishi 2,342,192
15 M azda 1,366,838
16 A vtoV az 1,313,076
17 China FAW 1,285,320
18 Isuzu
19 Subaru 803,266
20 Chongqing Changan 690,227
21 D ongfeng 622,317
22 Tata 592,656
23 SAIC 463,503
24 Chery 462,733
25 G az
26 Hafei
454,345
27 Volvo 437,752
28 Zheiiang G eely 308,425
29 Navistar 237,016
30 Anhui 231,917
31 Paccar 221,726
32 Mahindra & Mahindra 207,149
33 Proton
34 Iran Khodro
173,150
171,483
166,800
148,213
144,935
131,674

Source: Automotive News, 2007a

419
 Appendix 14

Global Manufacturers 1975-2005

Production output (000s)

Company 1975 1980 1985 1990 1995 2000 2005

GM 4954 5605 6992 4998 5777 5247 5657

Toyota 1715 2303 2678 3445 3371 4656 6157

Ford 3177 2903 3924 3371 3753 4039 3514
VAG 1547 2228 2474 2165 3007 4859 4979
Honda 328 846 1121 1658 1576 2251 3324
PSA 1656 1439 1978 1694 2494 2983
Nissan 1533 1940 2028 2117 2158
Chrysler 1839 711 1484 1220 1050
Renault 1486 1817 1478 1317 1518 4144 4892
Hyundai 1078 1185 226 558 1007 2023 2727
Fiat 1486 1875 1831 2184 1540
Suzuki 51 88 272 512 789 1182 1723
DC 387 737 867 2043 1965
Mazda 1302 722 822 1092
Kia 217 444 222 458
BMW 330 500 595 835 1323
Mitsubishi 289 660 636 833 1085 1094 998
Daihatsu 92 156 163 373 328 803
AvtoVAZ 713 798 900 525 756 721
Fuji 109 202 262 319 377 500 508
Tata 163
FAW 59
Chery 186
Gaz 117 52
SAIC 132
Porsche 98
Great 68
Wall
Proton 157 181 188
Naniing 111
Uaz 29
Kamaz 37 185 501 758 30
Daewoo
Isuzu 65 107 213 202 52 36 29
MG Rover 605 396 466 465 475
Mercedes 350 439 552 574 600
Volvo 225 136 395 156 371
Saab 112 93 97
Alfa 190 220 162
Romeo

420
Company 1975 1980 1985 1990 1995 2000 2005
Jaguar 294 38 324

SEAT 332 201

Skoda 324 165 174
AMO
Citroen 624
Peugeot 578
Ijmach- 43
Avto 754 1188 785
Others
Total 24070 27817 33861 34840 35869 43006 48099

Source: OICA, Ward’s Automotive

Proton data courtesy of Global Insight

421