The Institutional Economics of Foreign Aid: Bertin Martens, With Uwe Mummert, Peter Murrell and Paul Seabright
The Institutional Economics of Foreign Aid: Bertin Martens, With Uwe Mummert, Peter Murrell and Paul Seabright
The Institutional Economics of Foreign Aid: Bertin Martens, With Uwe Mummert, Peter Murrell and Paul Seabright
of foreign aid
Bertin Martens, with Uwe Mummert, Peter Murrell and Paul Seabright
2001 authors
Table of contents :
Chapter 1 : Introduction
by Bertin Martens (European Commission)
2
3
Foreword
The idea to research the subject of this book originally arose in the foreign aid
Evaluation Unit of the European Commission (EC). In the summer of 1997, I was
working with a few colleagues and consultants on two global performance reports of EC-
financed Phare and Tacis institutional reform programmes in the transition economies of
Central and Eastern Europe. We had worked our way through a pile of monitoring and
evaluation reports on individual Phare and Tacis projects, interviewed many programme
managers and consultants and completed some fieldwork as well. While we were
compiling a synthesis report, we were struck by the convergence of views that emerged
out of this apparently chaotic pile of information. There were many similarities in project
design, implementation and outcomes, in behaviour of project managers and policy
decisions, despite a wide diversity of project circumstances, across countries, sectors and
types of projects. Evaluation reports on EC programmes in other regions of the world
revealed similar patterns. Surely, there had to be common factors and incentives in the
EC aid delivery process that generated these similarities in outcomes, out of a diversity of
situations.
To illustrate this point, let me just cite a few of these general findings. A Phare
evaluation report (European Commission, 1997, p 55-56) concluded that there is "a
tendency for the Commission to contract out expertise in the transition process and retain
in-house expertise in financial and administrative procedures", "strong emphasis on
financial and procedural control rather than substantive design and performance" and
contracts were "based on inputs and activities specifications rather than on outputs and
effective results". At the same time, the EC appeared to be reluctant to move towards
conditionality-driven Phare programmes as a means to enhance programme effectiveness.
An evaluation report on the EC’s Mediterranean aid (European Commission, 1998)
concluded that multiple objectives came without "a comprehensive analysis of the
linkages and interconnections between different policy objectives" and consequently "no
goal hierarchy, leaving the Commission with unclear guidance for aid management and
implementation". The frequent recurrence of these and similar findings indicated that
3
4
aid programme performance was not only determined by the particular circumstances of
individual project managers and recipient countries but also – and perhaps predominantly
- by the incentives embedded in the institutional environment of the aid agency and its
aid delivery process. This finding may not sound terribly original to persons who have
had any substantial involvement in aid delivery. However, the virtual absence of any
studies of incentive structures in aid delivery processes is even more striking. This book
is an attempt to fill that gap.
When this research project was launched in 1998, the objective was to study the
institutional incentives and constraints that affect the performance of foreign aid
organisations in general. We had in mind not only the problems that EC aid programmes
were confronted with, but more generally the institutional incentives that all types of aid
organisations face. A quick literature review revealed that very little research was done
in this domain. Most aid performance studies look either at global macro-economic
performance indicators or use game-theoretic approaches to examine the impact of
conditionality-driven aid programmes. Often, poor programme performance is ascribed
to institutional and incentive problems in the recipient countries; incentive problems in
donor organisations are rarely evoked or examined. This research project has transposed
some of the findings from modern institutional and organisational theory to the domain of
foreign aid and thereby, we hope, fills a gap in the analysis of foreign aid delivery
processes and performance assessments.
This book has many shortcomings, the principal one being that it proves nothing. It
puts forward possible interpretations of observed behaviour of foreign aid organisations,
based on models derived from modern organisation theory. These models are sufficiently
general to be applicable not only to EC aid but to all types of organisations involved in
the delivery of foreign aid, including multilateral and bilateral donors and NGOs.
Examples are implanted in the text for illustrative purposes only and there is no attempt
to empirically validate the models in a rigorous way. Doing so would require a
substantial amount of data collection on the behaviour of aid organisations and agents
within those organisations. Such an empirical exercise would go far beyond the presently
available economic aggregates on foreign aid. It is hoped that this book provides some
of the theoretical foundations that will stimulate such empirical research and applications.
4
5
Financial support from the EC Tacis programme for the studies by Paul Seabright,
Peter Murrell and Uwe Mummert is gratefully acknowledged.
Bertin Martens
5
6
Chapter 1 Introduction
1. THE SETTING
This book is about the institutions1 that guide the behaviour of persons involved in
the implementation of foreign aid programmes. Institutions are the formal and informal
rules of behaviour that constitute incentives for all agents involved in the aid delivery
process; they affect the performance of foreign aid programmes. It analyses how these
institutions affect outcomes of the aid delivery process. The analysis covers a variety of
organisations, from taxpayers-donors, politicians, lobby groups, donor agencies and
consultants in donor countries, to recipient organisations in beneficiary countries. It
diverges from more traditional approaches to aid performance because it seeks to explain
that performance in terms of incentives inside the aid delivery process, rather than
1 We stick to North’s (1990) definition of institutions as the formal and informal rules, regulations, laws,
contracts and agreements that guide people’s behaviour. Organisations are the groups of people that
adhere to a particular set of these rules.
6
7
recipient country policy performance. The proposed approach also differs from game-
theoretic models that attempt to explain aid performance in terms of outcomes of
strategic interactions between donors and recipients at macro-institutional level. This
book goes down to micro-level decision-making processes and behaviour by agents
working in donor agencies, subcontractors hired by these agencies and officials working
in beneficiary country organisations. It will show that individual agents’ incentives and
constraints can divert significantly from those of the organisation that they work for and
thus lead to very different behavioural outcomes from those predicted by macro-
institutional approaches. On the other hand, the models and analysis presented in this
book are sufficiently general to be applicable not only to conditionality-based
institutional and policy reform programmes but also to non-conditional aid and more
traditional investment projects.
It is also a book about the use of foreign aid to achieve institutional and policy
reform in recipient countries. In our view, the performance of donor-induced reform
programmes is correlated with the institutional set-up of donor agencies and aid
programmes to deliver these reforms. Donors cannot realistically and successfully tackle
institutional reform questions in recipient countries if they do not have an appropriate
institutional set-up in place to deliver the required types of aid in credible manner. For
instance, for a given institutional setting in a donor agency, a project-based approach to
institutional reforms is more exposed to moral hazard problems than a conditionality-
based approach; project approaches allocate aid to inputs while conditionality approaches
pay for results. For instance, evaluation studies show that the institutional reform
performance of EC-financed Phare programmes in Central and Eastern European
countries is negatively affected by the absence of conditionality (Martens, 2001).
Alternatively, recipient countries cannot expect to receive appropriate technical and
political support for their reform plans from donors that do not have the institutional
technology to deliver their contributions in a credible way. This book examines how the
switch from traditional investment projects to institutional reform programmes affects
performance incentives for individual agents involved in aid implementation as well as
the overall performance of a donor agency.
7
8
aid performance studies published during the 1990s already paved the way. These
studies aimed to find a correspondence between changes in aid volumes and changes in
macro-economic variables in the beneficiary countries. See for instance, White (1992)
and Boone (1994). Generally, they conclude that foreign aid has no tangible impact on
economic growth in the recipient countries. Burnside & Dollar’s (1996/2000) landmark
study came to similar conclusions. However, the authors qualified this finding with the
remark that aid does indeed have a positive impact on growth when the policy
environment in the recipient country is conducive to growth. It also raised the question
of the direction of causality: does aid cause good policies, or do good policies induce aid
flows? In other words, does aid have an active or a passive role to play in policy and
institutional reforms in beneficiary countries?
While Burnside & Dollar (1996) present some econometric evidence in favour of
the active hypothesis, Dollar & Svensson (1998) are more cautious about the direction of
causality. An analysis of a sample of World Bank policy-based loans revealed that donor
efforts have no significant impact on recipient country policy performance. Policy
outcomes are basically generated by domestic political environments, not by donor
influence. This suggests a more passive role for donor agencies: selecting genuine
reformers and using aid programmes as a commitment device to ensure that the reformers
are not derailed from their mission. It puts responsibility for the success (or failure) of
aid programmes more squarely on the shoulders of the recipients.
However, Dollar & Svensson (1998) also note that about a third of all World Bank
structural adjustment loans fail to meet their policy targets but are nevertheless paid out
in most cases. This happens because the Bank does not only care altruistically about
economic development but also more egoistically about a country’s financial situation to
service previous Bank loans and pressure from other donors and creditors to do so. If this
happens regularly, borrowers will not fail to notice the creditor’s lack of commitment to
his own commitment devices. Clearly, moral hazard can occur on the donor side too.
This demonstrates that the donor’s policy stance and own internal incentives do indeed
matter for the outcome of aid programmes and would lead to a more balanced conclusion
whereby donors and recipients share responsibility for the success or failure of aid
programmes.
8
9
Over the years, the debate on the respective roles and responsibilities of donor
agencies and recipient governments has moved away from desired policy and institutional
reform outcomes to the processes of reform, including how incentives and constraints
influence outcomes. When an optimal outcome is not achieved, it is usually not because
of lack of knowledge about this outcome among decision-makers but rather because
optimising agents face incentives and constraints that deviate their behaviour from this
target (Ranis & Mahmood, 1992). More than a decade after the start of the great
economic transition wave, donor agencies have apparently not digested this message.
They continue to blame poor reform performance on obnoxious recipients, rather than
examine the weaknesses in incentives inside donor and recipient institutions.
Since this book is about institutions, it naturally applies the techniques and insights
of institutional economics. Modern institutional economics is a tree with many branches
and twigs. The terminology “institutional economics” covers a wide range of schools of
thought and methods. It includes several varieties of transaction costs economics, from
Coase (1937, 1960) to North (1990) and Williamson (1985), as well as various branches
9
10
of organisation theory, including property rights (Grossman & Hart, 1986) and
incomplete contracts theory (Tirole, 1999), and its analysis of organisational design
(Holmstrom & Milgrom, 1991; Aghion & Tirole, 1997). This is not the place to explain
the details of each of these schools. Interested readers are referred to more general
handbooks of institutional and organisational economics (for instance, Furubotn &
Richter, Masten & Williamson, Laffont & Tirole, 1993).
This book applies these insights to informational problems that may occur in the
various steps of the delivery process of foreign aid from the donor to the final
beneficiary. It examines how these informational problems induce biases in the
behaviour of agents involved in this process and how this affects the ultimate
performance of aid programmes. Fortunately, it also attempts to explain how some of
these problems can be tackled through clever institutional design.
10
11
The four studies in this book use a common methodological approach and apply the
techniques of only one branch of modern neo-institutional economics, namely principal-
agent or agency theory. Principal-agent theory starts from the simple observation that
modern organisations are usually hierarchically structured, with principals giving
instructions to agents. Principals in a company, a club or a public administration, can not
take all decisions and carry out all tasks themselves. They need to delegate at least part
of the work to agents. While the principal appropriates the benefits (and costs) of the
task, the agent receives a reward - a wage, a stock option, a promotion, etc. - in return for
carrying out the specified tasks. Delegation implies that the principal does not have full
information about the activities of the agent. If he would want to have full information
and monitor every aspect of the agent’s activities, he might as well carry out the
delegated tasks himself; there would be no gain from delegation. Delegation may result
in two types of problems. First, the agent may deviate from the instructions given by the
principal and carry out the delegated tasks in such a way that it advances his own
interests, rather than those of the principal. This is called moral hazard. Second, at the
time of reaching agreement with the principal, the agent may have access to information
inaccessible to the principal, and may manipulate this information in ways that run
against the principal’s interests (as when sellers of second-hand cars are more likely to
offer low-quality cars for sale, or when counterfeit money drives out good as classically
described in Gresham’s Law).. This is called adverse selection. Both problems lower the
return from the task for the principal, compared to the return under perfect information.
Agency theory has found a wide range of applications to virtually every aspect of
organisation and contracts (Laffont & Tirole, 1993). It is therefore somewhat surprising
to see that there are, as yet, very few traces of applications in the development economics
11
12
and foreign aid literature to the organisations involved in the implementation of foreign
aid.
Frey et.al. (1985) are more explicit about the difference between the macro- and
micro-institutional approach and suggest that there are two ways to model the behaviour
of (international) organisations. Either the organisation is treated as an aggregate unit
possessing well-defined preferences or the different interests of individual members of
the organisation are considered and the behaviour of the organisation as a whole is taken
to be the outcome of the individuals’ actions. Frey et.al. chose the first option for their
model of World Bank lending behaviour, partly because organisation-level data sets are
available that allow to test organisations’ behavioural assumptions. This book has
chosen the second option because our analysis of foreign aid organisations has led to the
12
13
conclusion that there is indeed strong divergence of interests between the various agents
involved in aid delivery, even though they may contractually be committed to the same
organisation and aid programme objectives.
The main disadvantage of our methodological option is the lack of empirical data
sets to test the models. Aid agencies, NGOs, aid services suppliers or other organisations
involved in foreign aid do not normally collect intra-organisational data on the behaviour
of individual employees, their motivation, incentives and effort spent on different tasks,
or their objectives and strategic behaviour in negotiations with agents of other
organisations. Collecting such data would require surveys, personal interviews, detailed
analysis of resource allocations and procedural complexity in aid organisations, etc. It is
not the purpose of this book to empirically test the models and assumptions presented.
Some of these are illustrated with ad hoc examples only. This book is meant to lay the
more theoretical foundations for empirical approaches to the institutional economics of
foreign aid. As such, it provides a basis for an empirical research agenda that will require
considerable data collection efforts.
This research project was not just an attempt to find new applications of agency
theory. It is used as an analytical tool to generate novel findings about the performance
of aid organisations, institutions and implementation processes that go beyond existing
development aid research, including research on aid conditionality. Before we look at the
findings of the individual studies in this book we first scrutinise – in the next section - the
existing principal-agent literature and examine to what extent some of its major findings
can be transplanted to the institutions and organisations of foreign aid and yield novel
insights concerning their performance.
Foreign aid grants are usually paid by citizens in a donor country, either as taxes
channelled through an official aid agency or as voluntary donations given to non-
governmental organisations (NGOs). Aid is normally not collected and dispensed by
13
14
private for-profit enterprises, unless they have set up a separate charitable organisation.
However, private agents and enterprises may act as subcontractors in the execution of aid
projects. Alternatively, foreign aid loans channelled through development banks are
mobilised on international financial markets and passed on to recipient countries. Both
loans and grants can be handled through bilateral as well as multilateral aid agencies.
Both official and non-governmental agencies share some of the typical characteristics of
public administrations.
Economic analysis of public administrations started with Simon (1958) and Downs
(1967). However, agency theory and asymmetric information models to investigate the
behaviour of organisations emerged in the 1970's only. Most applications focused on
private organisations where contractual exchange plays an important role. Applications
to public administrations emerged fairly recently only (Martimort, 1991; Holmstrom &
Milgrom, 1991; Tirole, 1994). The informational characteristics that distinguish public
administrations from private enterprises can be summarised as follows:
b) While private enterprises have multiple principals2 (shareholders) who share the
same objective – profit - public administrations have multiple principals
(politicians, parliamentarians, etc.) who rarely share objectives. While one
parliamentarian prefers to allocate more resources to road construction in
developing countries (he has a construction company in his constituency),
another may want to prioritise research in aids prevention (his constituency
includes a medical research laboratory).
2 In agency theory, multiple principals situations are known as joint delegation of tasks.
14
15
c) While private companies can measure the opportunity cost of alternative options
in terms of profits, public administrations have no clearly defined or measurable
trade-off between their multiple options. This is likely to result in potential
inconsistencies and contradictions, and inefficient resource allocation. Political
principals (parliamentarians) may delegate an unclear or even inconsistent set of
instructions to an official aid agency; consequently, the agency can not perform
optimally.
15
16
towards a single goal, thereby neglecting some of the others goals and disrupting any
political consensus among multiple political principals on their multiple objectives. This
is an intrinsic predicament of public administrations. In official aid agencies, pay is
usually fixed and not linked to any performance indicator. Some multilateral
development banks link pay to the volume of loans approved. This may result in biased
performance incentives, steering loan officers away from wider development objectives
and undermining the financial viability of the bank.
16
17
council, this often results in a stand-off between net contributors and net recipients with
the latter representing a majority in terms of countries. In the case of the European
Commission, only donor countries are members of the decision-making council.
However, the unanimity rule in EU decision making makes it harder to play off members
against each other.
The above cases of multiple principals and objectives have been analysed in detail in
the relevant agency theory literature and are easily applicable to various types of aid
agencies. However, a unique and most striking characteristic of foreign aid is that the
people for whose benefit aid agencies work are not the same as those from whom their
revenues are obtained; they actually live in different countries and different political
constituencies. This geographical and political separation between beneficiaries and
taxpayers blocks the normal performance feedback process: beneficiaries may be able to
observe performance but can not modulate payments (rewards to the agents) in function
of performance. Although donors are typically interested in ensuring that their funds are
well spent, it is extremely difficult for them to do so, since there is frequently no obvious
mechanism for transmitting the beneficiaries’ view to the sponsors. Even if there would
be such a mechanism, beneficiaries’ views are likely to be biased because (a) they do not
pay for their benefits and (b) their preferences and objectives are unlikely to fully
coincide with those of the donors. In addition, ascertaining the beneficiaries' views in
developing countries – often with low levels of democracy and literacy – may not be
easy, or even possible, in many cases. Instead, the sponsors must rely on various indirect
indicators of programme performance.
This typical characteristic of foreign aid delivery processes stands at odds with the
established neo-classical view of the efficiency of public institutions, as originally
formulated by Becker (1983) and recently restated by Whitman (1995). According to that
view, competition between politicians to get (re-)elected makes them exercise pressure
on public institutions to do what the beneficiaries – whom they represent – want.
Dissatisfied taxpayers and beneficiaries can lean on their political representatives to
17
18
This broken feedback loop induces stronger incentive biases in foreign aid –
compared to domestic wealth transfer programmes - diverting it from its original
purposes. It explains, for instance, why the interests of domestic suppliers of aid goods
and services – consultancy companies, experts, suppliers of goods – dominate decision
making: they are the direct beneficiaries of aid (they receive the contractually agreed
reward) and have direct leverage on domestic political decision-makers. They have first-
hand information on the outcomes of the aid programmes in beneficiary countries and
they are part of the constituency of the decision-makers in the donor country. Their
informational advantage may make them the ultimate beneficiaries of foreign aid. The
intended beneficiaries’ interests are geographically and politically too remote to
overcome the direct beneficiaries’ leverage. This shift away from a neo-classical perfect
information feedback loop to a broken information feedback loop typical in foreign aid is
probably one of the most important contributions of this book to the debate on the
performance of foreign aid.
18
19
This shift in the composition of foreign aid started in the 1980s, when foreign aid
programmes shifted away from a pure investment project focus towards policy and
institutional reform. This shift occurred because of (a) growing awareness that recipient
countries lacked an appropriate institutional and policy environment to make aid work in
a competitive market environment and (b) donor and creditor concerns about the
deteriorating financial situation of many developing countries in the early 1980s. To
meet these concerns, a fast-disbursing non-project financial instrument was required in
the form of structural adjustment programmes that targeted financial aid on overall
balance of payments and budgetary support, not on specific investment projects.
Disbursement was usually conditional on economic policy changes and institutional
reforms. It accelerated in the early 1990s when the so-called transition economies
emerged on the foreign aid scene. The wave of transitions from centrally planned to
market economies further emphasised the need for institutional and policy reform3
programmes.
In the case of EU foreign aid, for instance, the shift towards institutional reform is
quite obvious. In the early 1990s, the Phare and Tacis technical assistance programmes
for the transition economies in Eastern Europe emerged. They focused predominantly on
institutional reform. By 1999, they represented nearly a third of all EC-managed foreign
aid (EC, 2000). In the meantime, EC aid to Sub-Saharan Africa, the Mediterranean, Asia
and Latin America also shifted attention to institution building. Last but not least, the
growing volume of institutional economics studies of development issues has no doubt
served as a more theoretical platform in support of this trend. Many studies have
demonstrated the strong positive correlation between the quality of institutions and the
level of economic development (for instance, World Bank, 1997; Clague, 1998).
This shift away from traditional investment projects towards institutional reform
programmes has implications in terms of performance incentives in principal-agent
relations in foreign aid implementation, as well as for the overall performance of foreign
aid programmes. Traditional investment projects produce tangible outputs that are fairly
3 “Institutional reform” and “policy reform” are considered as virtually synonymous expressions in this
text: all government economic policies, and changes therein, aim to redistribute wealth, either directly
through fiscal redistribution or indirectly through changes in property rights systems. All these policies
are implemented through changes in formal institutions or legal systems.
19
20
easily verifiable and measurable, such as roads, schools and hospitals, institutional
reform programmes produce less tangible outputs that are much harder to verify. Reform
projects produce intangible outputs, such as draft laws, organisational reform plans,
policy advice and trained staff. Their ultimate impact is more diffuse and hard to verify
and consequently more easily subject to post-contractual uncertainties. To the extent that
aid aims to achieve institutional reforms in the recipient country, monitoring of outcomes
and impact becomes more difficult compared to traditional investment projects. Increased
difficulty and cost of monitoring facilitates moral hazard and adverse selection in foreign
aid programmes, and makes it easier to diverge programmes from their original purposes.
Let us now move from these general institutional and informational aspects of
foreign aid to the more specific findings of the studies in this book. Each study focuses
on a different set of institutions and agents in the aid delivery process and examines their
specific informational problems. In a standard official bilateral aid setting, the chain of
principal-agent relationships starts with taxpayers as principals, who wish to transfer part
of their income to recipients in other countries. They delegate the implementation of this
transfer programme to their representatives (parliamentarians, politicians) who become
their agents. These agents, in turn, become the political principals to an aid agency in
charge of implementation of aid programmes. Within the aid agency, a hierarchical
command chain creates a further series of principal-agent relationships. When actual
implementation is subcontracted to a private consultant or aid services supplier company,
the task manager in the aid agency becomes a principal to the contractor, who is his
agent. Depending on the contract he gets, the contractor may also be an agent to the
recipient agency or counterpart administrator in the beneficiary country. He may end up
being an agent to two principals, putting him in a typical joint delegation situation. The
recipient agent, in turn, is an agent to his political principals and the beneficiary
population at large. Each of these principal-agent interfaces in the long chain of
command creates a potential for incentive misalignment and moral hazard. As such, the
20
21
final outcome of the aid process may be quite different from the original objective
envisaged by the taxpayer, the aid agency or the recipients.
Many variations on this standard chain of command are possible. Aid agencies may
chose to implement a project directly, without passing through a private contractor; or
they may provide non-project structural adjustment aid that does not require
implementation through technical assistance. Alternatively, donor country citizens may
prefer to allocate aid through an NGO. This skips the political representation part of the
scheme but decision making needs to be done anyway, by the (un)elected Board members
of the NGO. Otherwise, NGO aid goes through similar delivery steps. Another
alternative is to channel aid through multilateral organisations. In that case, several
donor country governments delegate implementation responsibility to a multilateral
organisation; again a case of joint delegation. Last but not least, donor country
governments may set up a multilateral development bank that mobilises financial
resources on international capital markets rather than using taxpayers’ money.
Each of the four papers in this book looks at a specific part of the principal-agent
chain. Paul Seabright examines performance incentives inside donor agencies. Peter
Murrell analyses the incentives in the interactions between donors, contractors and
recipients. Uwe Mummert investigates what happens inside the recipient country once
the recipient government has decided to formally adopt a donor’s institutional reform
proposal. My own study covers the evaluation feedback loop between the contractor’s
performance and the donor’s aid objectives. The studies emphasise institutional and
policy reform programmes and not more traditional investment projects, though many of
their findings may be applicable to the latter type of aid as well. The main difference
between institutional reform and investment projects is that the outputs of the latter type
are mostly physical and therefore more readily observable and measurable, making them
less easily subject to moral hazard and adverse selection. The trend towards more
institutional reform aid thus increases the probability of moral hazard in aid.
21
22
The first paper, by Paul Seabright, examines the incentives and biases in the
behaviour of foreign aid agencies. He focuses on two informational problems in aid
agency behaviour, namely inputs bias as a result of the broken feedback loop and
problems caused by multiple principals and objectives.
Two ideas are important here. First, when agents’ salaries are not directly linked to
ostensible performance, demonstration of their abilities and career concerns dominate
their behaviour. Second, when an agent faces multiple tasks that compete for his time, he
will focus on those that are more likely to satisfy his career concerns. This may motivate
him to focus on tasks that are more easily monitorable by his superiors. However, to the
extent that easily (e.g. input-related) and less easily (e.g. output- or results-related)
monitorable tasks are complementary and equally important for the overall performance
of aid programmes, management will have to ensure that incentives for easily monitored
tasks are less high-powered in order to avoid agents diverting effort away from less easily
monitored but still important tasks. Reality is often different however. Careers are often
build on demonstrating good performance in more easily monitorable tasks, such as
“committing and spending budgets”. This may provide an explanation for the “inputs
bias” that is so often observed in aid agency behaviour (for instance, EC 1997).
22
23
On the other hand, organisations may rationally place ''too much'' emphasis on input
tasks, provided these reveal information about talents that may be valuable in output
tasks, and in spite of the fact that the incentives to perform the output tasks well will
thereby be blunted. Separating the input tasks from the output tasks would be too costly
in terms of failing to exploit the links between these skills. Bundling the tasks together
may be the lesser of two evils even though it inevitably leads to a misallocation of agents'
efforts towards the inputs. Separation of tasks is important for relatively junior members
of an agency, because of the greater information about their skills that such separation
yields. Too much generalism among junior staff makes it harder to allocate them
subsequently to responsible positions in the organisation. Empirical studies of
bureaucracy have often failed to distinguish between those aspects of bureaucratic
behaviour that reflect poor organisational design and those that are unavoidable
consequences of the kind of activity the organisation is obliged to undertake. Some
features of organisations, though perhaps regrettable, are the inevitable result of the fact
that individuals' behaviour cannot be precisely monitored.
Multilateral aid agencies, like the World Bank or the European Commission, are
cases of joint delegation from multiple principals or member states. On the one hand,
joint delegation may induce credible competition between members, enhance the
credibility of the agency and allow it to resist pressure from individual members. For
instance, it may enable the agency to commit itself to procedures that would not be easy
to implement for a bilateral donor, such as transparent and competitive procedures for
tendering and procurement, and a commitment to avoid linking aid to narrow
considerations of market access or the fortunes of particular political and economic
interest groups. However, this commitment may be easier to deliver on the side of input
procedures (tendering and contracting) than on the side of the preparation and selection
of projects. This latter fact further reinforces the input bias already described. Joint
delegation may also achieve economies of scale and scope. For instance, a single aid
accounting system or a single set of procedures can be shared by different aid
instruments. Multilateralism can also enable the exploitation of economies of scale and
scope that are beyond the capacity of bilateral donors.
23
24
Joint delegation may on the other hand result in confusion over objectives, or in
agencies pursuing their own interests. For instance, sharing of tasks may generate
distortionary incentives and divided loyalties to different principals. Another example is
that of a multilateral aid agency that reports on the “needs” of beneficiary countries and
may have an incentive to exaggerate these needs in order to boost the importance of the
agency. Similarly, it may bias reporting on the beneficial impact of aid programmes to
justify its activities. Other beneficiary countries and member states may be aware of this
over-reporting but not take any action because it benefits them too when their turn comes.
The potential for wide variations in performance in cases of joint delegation indicates that
a multilateral agency that fails to exploit the benefits that come from the presence of
multiple principals is actually likely to perform worse than would bilateral donors. If it
cannot find ways to do so it cannot really justify its existence.
Peter Murrell’s paper moves away from donor agencies and into the domain of
interactions between donors, their (sub-) contractors and the recipients of the aid. He
also focuses more on the specific incentive problems posed by institutional reform aid
and identifies the following actors in a principal-agent model: the donor principal (the
political superiors in donor agencies), the donor agent (the direct administrator of a
programme, a task manager), the contractor (a profit-seeking consultancy or a non-profit
NGO) who has a contractual relationship with the donor agent, the recipient principal
(political superiors in the recipient agency), the recipient agent (the institution that is the
formal beneficiary of the project). Given the inherently incomplete contract between
donor agent and contractor, there is a margin for manoeuvre in project implementation.
The equilibrium outcome will depend on who is in control, the contractor, the recipient,
or both. Each of these situations is likely to push the actual project outcome away from
the intended outcome. The model allows us to examine two sources of information
problems that are at the basis of this performance bias:
24
25
produce less tangible outputs which leaves more freedom to the contractor, whose reward
is not dependent on achievement of targets; he usually receives a fixed lump sum
payment only. As a result, he may collide with the recipient to bend the project to their
common interest. Traditional inputs-driven project-based approaches to institutional
reform, as practised by most technical assistance programmes, such as the European
Commission’s Phare and Tacis programmes, are therefore unlikely to yield the intended
results.
This brings us to the second issue that goes to the core of the institutional reform
debate: the desirable extent of foreign donor leverage in domestic reforms and the
representativeness of the recipient organisation. In the “interest group” case, the
recipient agent acts on behalf of some narrow interest group, rather than in the interests of
a broader political spectrum. In the ideal – though perhaps somewhat utopian –
“embedded” case, when the recipient principal and his agent(s) represent the same
broader interests, the benefits derived from the reform depend on the fit between the
proposed reform and the wider interest of the recipient country. If embeddedness is the
case, than the donor has an interest in facilitating – rather than trying to prevent – post-
contractual adjustment in the outputs of the project; that will only increase the benefits
derived from the project by the beneficiary country. In the interest group case, the
reverse is true and stricter monitoring will be required.
A peculiar situation arises in the case of EU aid for institutional reforms in its
candidate member states in Central and Eastern Europe. There, the EU is not so much
concerned with the welfare of the recipient country but rather the strict implementation of
25
26
the Acquis Communautaire4. It may therefore seek an alliance with recipient country
interest groups that share the EU’s interest. Also, the contractor can be a non-profit
organisation with a specific agenda, an interest group in its own right, that is willing to
forgo some of the contractual rents in order to achieve its own agenda. This could be the
case in so-called twinning projects under the EC-financed Phare technical assistance
programme in candidate member states, where technical assistance is not normally
provided by commercial companies but by government administrations from EU member
states. The paper shows that the presence of interest groups and embeddedness
multiplies the difficulties that donors have in ensuring that the independent contractors
carry out their activities productively.
The paper by Uwe Mummert deals exclusively with institutional reform aid and the
informational problems that it induces in principal-agent relationships in the recipient
country. Whereas Murrell’s paper examined how donor-financed contractors could
achieve formal de jure reforms in recipient countries, Mummert examines what happens
de facto, once the de jure reforms are approved.
4 The set of EU rules, regulations and directives which all member states have to adhere to.
26
27
institutions have to be not only de jure (formally) integrated into the formal institutions
of the recipient country but also de facto embedded into the informal socio-economic
institutions of the recipient society.
Effective institutional reform implies that de jure changes affect the de facto choices
of the actors to whom the rules apply. Thus, it focuses on the ‘embeddedness’ of reforms
in the wider socio-economic environment, the ties between society and government
agents that are in charge of implementing reformed institutions, the power of informal
institutions in social networks and how they might provide obstacles to effective
institutional reforms.
The model developed in the paper distinguishes between two levels of principal-
agent relationships in the recipient country. Political principals, who adopted the donor-
induced institutional reform, delegate the task of implementation and verification to
government agents, in return for a reward. Since political principals are unable to
observe all the actions of government agents, moral hazard may occur. These agents, in
turn, give instructions to citizens to modify their behaviour in accordance with the new
institutions. However, government agents do not operate in a social vacuum: they are
embedded in social networks that include the citizens to whom they give instructions but
who may also exercise various types of social and economic pressure on them. Citizens
may informally influence the government enforcement agent through bribes or social
sanctions. As a result, collusion of interests between enforcers and citizens and deviations
from the intended outcome of the reform may occur. In order for legal de jure reform to
be effective, legal sanctions must be strong enough to penetrate the inhibitive layer of
informal non-legal sanctions. The model describes the mechanics of resistance to
reforms and the cost-benefit calculations implied by these mechanics. Particular types
and sources of resistance to reforms are explored: the tension between proscriptive and
prescriptive content, cooperation-defection differentials and the extent of fragmentation
in society.
Informal institutions are not inflexible and may also evolve as a result of reforms in
formal institutions. As long as the net benefit from cooperation exceeds the benefit from
defection, the new institution will be accepted. Assuming that cooperation-defection
differentials are variable across a population, a sufficient condition for successful reform
27
28
The extent of segregation in society and entry barriers to segregated groups may also
affect the effectiveness of reform. In highly fragmented societies, it is very difficult for
economic actors to change groups in order to escape restraining informal institutions
within their group. Fragmentation determines whether informal institutions that resist
reform will be perceived as subordinate to the formal institutions, as well as the intensity
of the impact of informal institutions on the dynamics of the market processes.
Empirical research has shown that ethnic, linguistic and social fragmentation in society is
indeed a significant determinant of economic development.
Because of the broken feedback loop between the beneficiaries and the donor
country, the risk of moral hazard by aid services suppliers is very high. This can only be
overcome by the introduction of an explicit information feedback mechanism: formal
evaluation of aid programmes. For a given aid budget, there is an optimal share of that
28
29
budget that should be spent on evaluation studies in order to maximize the impact of that
budget. That optimal share strikes a balance between spending more money on aid
projects to achieve objectives and spending more money on suppliers’ supervision so as
to ensure adequate performance for every unit of money spent on projects.
However, since evaluation is usually handled by the aid agency itself, it is subject to
the politician’s interest in keeping the middle ground between the opposing objectives of
taxpayers and suppliers. Consequently, evaluation will be manipulated in function of
these interests. Aid agencies and their political ‘owners’ have several instrumental
variables at their disposal to do so. First, they can reduce the share of the aid budget
spent on evaluation. This eases performance pressure on suppliers and enhances their
profit margins. On the other hand, it does of course reduce programme performance and
thus consumer surplus for the taxpayers. Second, they can vary the quality of evaluations
by manipulating the ratio of budget spent on evaluation data gathering to budget spent on
projects. Lower-quality evaluation studies are defined as having larger standard
deviations for the observations on project performance variables. With large standard
deviations, confidence intervals become wider and it becomes more difficult to prove that
a performance variable is off target. The final decision on budget share and quality of
evaluations is determined by the composition of the political constituency that elected the
politician. If taxpayers are predominant in that constituency, political principals will be
inclined to increase spending on evaluation; if suppliers are predominant, the reverse will
be true. Manipulation of the quality of evaluations allows politicians to drive a wedge
between the interests of opposing constituencies, for instance in a coalition government.
Less reliable reports satisfy suppliers because they make it more difficult to credibly
criticise their performance; at the same time, it satisfies taxpayers’ wishes to have
performance feedback. Coalitions are not only held together by the glue of transfers but
also by information smokescreens that veil opposing views and contradictory
information.
In short, because of the broken ‘natural’ feedback loop in foreign aid, inserting an
explicit evaluation function in foreign aid programmes is necessary to overcome moral
hazard on behalf of aid services suppliers. But it is not a miracle solution to eliminate
performance problems. Evaluation itself is subject to moral hazard, induced by the same
institutional and political incentives that affect aid projects performance.
29
30
***************
What lessons can we learn from these four studies? They show that the nature of
foreign aid – with a broken information feedback loop – combined with the nature of
public administrations (including aid agencies) in general – with multiple hard-to-
measure objectives and often multiple principals too – put a number of inherent
constraints on the performance of foreign aid programmes. All these constraints are due
to imperfect information flows in the aid delivery process. Informed institutional design
(of contracts, aid agreements, budget allocation mechanisms) can mitigate the impact of
some of these informational constraints. However, most aid agencies and programmes
work with fairly standardised institutional set-ups that leave little room for variation.
Any deviations from the standard set-up are time-consuming, complex and costly to
implement, unless there is a strong political will to go ahead. The range of foreign aid
performance objectives that each type of aid agency can address with reasonable chances
of success is limited. Since most aid agencies intend to pursue a wide range of objectives
and address an ever larger set of problems in strongly variable environments, they could
usefully apply the findings of modern institutional economics and organisation theory so
as to improve their performance and chances of success in meeting their objectives.
Similarly, they would also benefit from accepting the – often political – limits to their
range of effective actions. Making these limits more explicit would be an important step
in that direction.
This research project has been successful in terms of its own objective to help
explain the causes of the persistent problems and behavioural patterns in EC foreign aid,
identified earlier on in the foreword to this book. It points out two factors: multiple
political principals and multiple objectives. A national minister in charge of a bilateral
donor agency can take political ownership of his development policies and strategies;
s/he defends them in parliament where a coalition or political majority reduces the
number of political principals from many to one. By contrast, an EC commissioner in
charge of foreign aid has to deal with fifteen member states and hundreds of European
30
31
Fortunately, this research has also pointed out means to mitigate at least some of the
consequences of the biased performance incentives that underlay the EC's institutional
set-up and in particular those of its foreign aid programmes. Ideally, the number of
political principals overseeing EC foreign aid could be reduced to one if the Commission
had a political majority in Council and Parliament and if majority voting were introduced.
However, this will require more fundamental EU institutional reforms that go far beyond
the domain of foreign aid only. As a second-best solution, the creation of an EC foreign
aid agency may be a step in the right direction, provided that it gets clear mission orders,
reducing the number of objectives, stating them more clearly and limiting interference by
multiple principals. Finally, enhanced monitoring and evaluation of aid programme
outcomes could help to strengthen the information feedback loop between beneficiaries
and donors/taxpayers. But it is unlikely to be effective unless fully independent from
political principals who manage aid programmes. Last but not least, one can conclude
from this research that the usual suspect for poor performance of EC aid – the presumed
lack of human resources - is unlikely to be the real culprit. Human resources are bogged
down in unproductive and inefficient administrative procedures and inputs-related
approaches, drawing them away from more results-oriented tasks. Increasing the number
31
32
of staff without fundamental changes in the incentives structures would only worsen
input bias because it would create opportunities to pile up more ineffective layers of
procedures. The only real solution is a fundamental reform of the incentives that staff are
confronted with.
This research has not only contributed to our understanding of the predicament of
EC aid but of foreign aid programmes in general. The models, techniques and
interpretations developed here have wider application potential to all types of foreign aid
institutions and organisations. All official aid agencies are public administrations and
therefore subject to the same incentive biases as ordinary public administrations, caused
by multiple objectives and principals. Even NGOs may be subject to some of these
biases. That explains their weak leverage, procedural bias and, in general, the imperfect
information environment in which they operate. Moral hazard and adverse selection are
facts of life in such an environment. On top of that, foreign aid suffers from a broken
feedback loop between beneficiaries and decision-makers that induces stronger
performance biases compared to ordinary domestic public administrations and the
transfer programmes that they operate.
This research project has been successful in achieving its own objectives: it has
clarified performance problems in foreign aid institutions in general, and in the EC as an
aid agency in particular. It has shown that the EC's predicament is by no means
exceptional in the world of foreign aid, that it is due to institutional incentives structures
and that there are ways to improve this performance. It remains to be seen, of course, to
what extent the findings of this research will be taken seriously and contribute to policy
decisions.
32
33
This book is meant to be an original and innovative contribution to the debate on the
performance of foreign aid programmes in general, and in particular on the role of the
institutional set-up and incentives provided by the organisations that manage these
programmes. It is certainly not the definitive view on that subject but may provide an
encouragement for decision-makers in foreign aid to become more creative in the design
of the institutional aspect of programmes and take into account the contributions that
institutional economics can make to facilitate their mission. We also hope that this book
will have laid the theoretical foundations for more empirical research in this direction.
33
34
References:
Aghion, Ph. and Tirole, J. (1997) “Formal and real authority in organisations”, Journal of
political economy, vol 105(1), p. 1-27.
Becker, G., (1983) “A theory of competition among pressure groups for political
influence”, Quarterly Journal of economics, vol 98(3), p.
Boone, P. (1994) “The impact of foreign aid on savings and growth”, mimeo, London
School of Economics.
Burnside, C.; Dollar, D. (1996) “Aid, policies and growth“, Policy Research Working
Paper nr 1777, World Bank.
Burnside, C.; Dollar, D. (2000) “Aid, policies and growth”, American Economic Review,
vol 90(4), pp 847-68, September 2000.
Clague, Ch. (ed.) (1998) “Institutions and economic development”, Johns Hopkins
university press, Baltimore and London.
Coase, R. (1937) “The nature of the firm”, Economica, vol 4, p 386-405.
Coase, R. (1960) “The problem of social cost”, Journal of law and economics, vol 3, p. 1-
44.
Collier, P.; Guillaumont, P. and S.; Gunning, J.W. (1997) "Redesigning conditionality",
World Development, vol 25(9), September 1997, p. 1399-1407.
Devarajan, S.; Dollar D.; Holmgren, T. (1999) “Aid and Reform in Africa”, draft
summary report, World Bank Development Research Group working paper.
Dixit, A. (1997), “Power of incentives in private versus public organisations”, American
Economic Review, vol 87(2), p. 378-382.
Dollar, D. and Svensson, J. (1998) « What explains the success or failure of structural
adjustment programs? », World Bank, mimeo
Downs, A. (1967) "Inside bureaucracy", Rand Corporation research study, Little, Brown
Boston (Mass.), 4th print.
European Commission (1997) "Phare Interim evaluation report", EC Evaluation Unit,
July 1997. Available at http:/europa.eu.int/comm/scr/evaluation
European Commission (1998) "Evaluation of aspect of EU development aid to the
Mediterranean region", EC November 1998, by Cowi, NEI and Andante.
Available at http:/europa.eu.int/comm/scr/evaluation
European Commission (2000) “The European Community External cooperation
programmes: policies, management and distribution", EC, Brussels and
Luxemburg, in collaboration with Overseas Development Institute, London.
Available at http:/europa.eu.int/comm/scr/evaluation
Frey, B.; Horn, H.; Persson, T.; Schneider, F. (1985) “A formulation and test of a simple
model of World Bank behaviour”, Weltwirtschaftliches Archiv, vol 121(3), p.
438-447.
Furubotn, E. and Richter, R. (1998) “Institutions and economic theory: the contribution
of the new institutional economics”, University of Michigan Press.
Grossman, S. and Hart, O. (1986) “The costs and benefits of ownership: a theory of
lateral and vertical integration”, Journal of political economy, vol 94, p 691-719.
Haggard, S.; Webb, S. (eds) (1994) “Voting for reform: democracy, political
liberalisation and economic adjustment”, Oxford university press, New York.
Holmstrom, B. and Milgrom, P. (1991) “Multitask principal-agent analyses: incentive
contracts, asset ownership and job design”, Journal of law, economics and
organisation, vol 7, special issue, p 24-52.
34
35
35