Export Performance and Its Determinants: Supply and Demand Constraints

UNITED NATIONS CONFERENCE ON TRADE AND DEVELOPMENT
POLICY ISSUES IN INTERNATIONAL TRADE AND COMMODITIES

STUDY SERIES No. 26
EXPORT PERFORMANCE AND ITS DETERMINANTS:

SUPPLY AND DEMAND CONSTRAINTS
by
Marco Fugazza
Associate Economic Affairs Officer

Trade Analysis Branch
Division on International Trade in Goods and Services, and Commodities
United Nations Conference on Trade and Development
Geneva, Switzerland
UNITED NATIONS
New York and Geneva, 2004
NOTE
The views expressed in this study are those of the author and do not necessarily reflect the views
of the United Nations.
The designations employed and the presentation of the material do not imply the expression of
any opinion whatsoever on the part of the United Nations Secretariat concerning the legal status of
any country, territory, city or area, or of its authorities, or concerning the delimitation of its frontiers
or boundaries.
Material in this publication may be freely quoted or reprinted, but acknowledgement is

requested, together with a reference to the document number. It would be appreciated if a copy of
the publication containing the quotation or reprint were sent to the UNCTAD secretariat:
Chief
Trade Analysis Branch
Division on International Trade in Goods and Services, and Commodities
Palais des Nations
CH-1211 Geneva
UNCTAD/ITCD/TAB/27
UNITED NATIONS PUBLICATION

Sales No. E.04.II.D.20
ISBN 92-1-112627-4
ISSN 1607-8291
© Copyright United Nations 2004

All rights reserved
ii
ABSTRACT
What are the major determinants of export performance? Does the importance of these
determinants vary with export performance itself? Using quantile regression techniques this study
investigates the contribution towards the performance of the external sector of linkages to
international markets relative to internal supply-side conditions. Results indicate that, while trade
barriers continue to be of concern, poor supply-side conditions have often been the more important
constraint on export performance in various regions, in particular in Africa and the Middle East,
despite a generalized deepening of international trade integration. Beside strong linkages to
international markets, good transport infrastructures, macroeconomic soundness and good quality
institutions appear to be major determinants in the development process of the external sector.
iii
ACKNOWLEDGEMENTS
I am grateful to Sam Laird, who has improved significantly this study with
extremely valuable comments and suggestions. I have benefited from discussions with
Lucian Cernat, Jörg Mayer and Frédéric Robert-Nicoud. I also thank my colleagues at
UNCTAD and seminar participants at the University of Geneva for their helpful
comments.
iv
CONTENTS
NON-TECHNICAL SUMMARY .......................................................................................1

INTRODUCTION .............................................................................................................3
I. THE THEORETICAL CONTEXT.........................................................................7
II. THE COMPONENTS OF GROWTH ...................................................................9
(a) The Dataset ....................................................................................................... 9

(b) Estimation Strategy............................................................................................ 9
(i) Foreign market Access.......................................................................... 12
(ii) Supply Capacity ................................................................................... 19
(iii) Export Constraints ............................................................................... 20
(c) Sensitivity Analysis .......................................................................................... 24
III. EXPORT PERFORMANCE AND ITS DETERMINANTS .................................25
(a) An Extended Theoretical Framework............................................................... 25

(b) The Data ......................................................................................................... 25
(c) Estimation ....................................................................................................... 26
(d) Results and Interpretation................................................................................ 27
(i) General Considerations ........................................................................ 32
(ii) GDP and Population ........................................................................... 32
(iii) Internal Transport Frictions ................................................................. 32
(iv) The Macroeconomic Environment....................................................... 33
(v) Foreign Direct Investment ................................................................... 33
(vi) Institutions........................................................................................... 34
(vii) Foreign Market Access ......................................................................... 39
(e) Sensitivity Analysis........................................................................................... 40
IV. POLICY IMPLICATIONS ....................................................................................41
V. POSSIBLE FURTHER RESEARCH .....................................................................43
REFERENCES...................................................................................................................44
APPENDIX ......................................................................................................................46
v
Tables
1. Bilateral trade equation estimation ............................................................................ 10

2. Bilateral trade equation estimation with intra-regional trade dummies....................... 11
3. Components of export growth and regional concentration
of foreign market access............................................................................................. 14
4. Components of regional exports growth .................................................................... 17
5. Geographical composition of regional foreign market access growth.......................... 17
6. OLS regressions......................................................................................................... 28
7. Quantile regressions with institutions and paved roads .............................................. 29
8. Quantile regressions with Lab institutions and paved roads ....................................... 30
9. Quantile regressions with institutions, Lab institutions and paved roads.................... 31
10. Countries exports real value position in sample distribution ...................................... 35
11. Countries position in sample exports real value distribution ...................................... 37
Figures
1. Foreign market access and supply capacity................................................................. 13

2. Regional origin of foreign market access .................................................................... 18
Graph 1: Benchmarked export performance and components .............................................. 21
vi
NON-TECHNICAL SUMMARY
A strong and performing external sector is found in most country experiences to be the
companion of a growing economy. The most striking and well-known example is the East Asian
countries experience. Identifying the elements that significantly affect export performance should
facilitate the design of policies to improve performance and ultimately overall economic growth.
Such policies may also help to contain the negative effects on the trade balance that often occur
immediately following trade liberalization.
Determinants of export performance can be split into internal and external components.
External factors are related to market access conditions and other factors affecting import demand.
Apart from trade barriers and competition factors foreign market access is also determined by
transportation costs, which include geography and physical infrastructures. Internal factors refer
to supply-side conditions. Supply capacity is also affected by location-related elements, which
may for example, affect access to raw materials and other resources. It also depends upon factor
costs such aslabour and capital. Beside resource endowment, factor costs are essentially the outcome
of economic policy and the institutional environment. Access to technology, which is likely to
affect the productivity of the external sector, may also be an important determinant.
In order to examine these issues, an econometric model of bilateral trade flows using gravity
techniques, was constructed. This model was tested using data series representing foreign market
access and supply capacity for a sample of 84 countries. It is thus possible to decompose export
performance and identify the extent to which it has been constrained by its components.
The main findings are as follows: first, in the aggregate, all regions have benefited from the
greater integration of the world economy in the 1985-99 period. Access to extra-regional markets
in particular has been a key factor explaining export performance. However, intra-regionally-
generated foreign market access has also been important in most regions, possibly underscoring
the increasing significance of regional trade agreements. However,this is not the case for the Sub-
Sahara African countries whose intra-regional trade declined in all periods but 1992-95. In addition,
African and Middle Eastern countries appear to have faced severe supply capacity constraints over
the last two decades, while their access to foreign markets has remained largely unchanged. East
Asian and Pacific countries’ export performance has been driven by equal improvements by both
supply capacity and foreign market access. The export growth of South Asian countries can mainly
be explained by an important increase in their supply capacities.
The impact on export performance of various supply capacity factors controlling for foreign
market access is investigated. This analysis used econometric techniques, namely quantile regression
techniques that allow the consideration of possible non-linearities in the relationship between
export performance, supply capacity factors and foreign market access. It is thus possible to observe
relationships between export performance and its components that vary with the level of export
performance. It isalso possible to place the analysis in a development process framework, although
limited, by considering three successive periods.
1
It was found that limitations on foreign market access are major contributors to poor
export performance. However, good performers in the second half of the 1990s also faced higher
external constraints but were able to overcome them.
There is also evidence that exports can be expected to respond less than proportionally to
a variation in import demand from abroad, although this not always true. In general theoretically,
a rise in exports would tend to increase factors of production prices, which contain exports
expansion.
Internal transport infrastructure captured by the percentage of paved roads is an important

supply capacity element and is found to have a significant and positive impact in raising performance,
as does a good macroeconomic environment. The contribution of foreign direct investment to
capital formation is included in order to include a technology-related element, possibly linked to
the structure of the external sector. The finding is that FDI is significant and has a positive impact
on export performance at all levels.
The empirical results also suggest that foreign market access and the structure of the external
sector interact. So as the external sector expands and diversifies, domestic producers make an
effort to overcome supply capacity constraints and increase their capacity to meet new market
opportunities. The evolving structure of the external sector also makes a difference at all stages of
development and could dominate the role of international linkages at an intermediate stage of
structural change. However, once a sector has reached structural “maturity”, as it seemed to be the
case for the best export performers among developed countries in the late 1990s, the significance
of foreign market access logically increases. In the present study, the institutional framework is of
much less importance than has been suggested in other recent studies in the empirical growth
literature.
The general policy implication is that foreign market access and supply capacity have to be
considered equally important along the development process of the external sector. Simultaneous
efforts to improve both supply capacity and foreign market access enhances the performance of
and the structural deepening of the external sector. Important elements of supply capacity at the
early stage of development of the external sector are transport infrastructure and macroeconomic
stability. FDI is a significant determinant at all levels of export performance.
2
INTRODUCTION
Despite the worldwide fall in trade barriers that has occurred in the last two decades,
export performance has varied substantially across countries. World exports increased by almost
220 per cent in twenty years. The figure jumps to 720 per cent for East Asian and Pacific countries
and falls to 80 per cent for Sub-Saharan countries. The exports of “best performers”, such as the
Republic of Korea, China, Cambodia and Viet Nam, have grown by more than 15 per cent annually
over the whole period. “Worst performers”, mostly African and Latin American countries, have
negative annual growth rate records in at least one decade.
As a result of various trade negotiations and autonomous reforms, access to international

markets has improved in the last twenty years. Nonetheless, it is likely that there is still much to
gain from further improvements in market access conditions. Concerns have also been raised
about the necessity to improve supply conditions. Supply conditions are fundamental in defining
the export potential of an economy. For a given level of access to international markets, countries
with better supply conditions are expected to export more.
This study investigates factors possibly explaining divergence in export performance.

Particular attention is devoted to factors affecting supply conditions after controllingfor access to
international markets.
The relevance of such an exercise rests also on the fact that no clear policy implications
emerge from economic literature1 which looks at the relationship between open trade and economic
growth. The positive correlation between output growth and export performance is a strongly
asserted empirical observation. Thus, a better knowledge of the determinants of export performance
should contribute towards a better qualification of the relationship between open trade and growth.
Determinants of export performance can be split into external and internal components.
External components include market access/entry conditions and a country’s location vis à vis
international markets. Internal components are related to supply-side conditions.
Foreign demand is influenced by various elements. Firstly , it is strongly linked to geography

(the structural component). Typically, countries at the centre of a fast growing region are more
likely to benefit, ceteris paribus, than countries situated outside that region. Second, it is likely to
be related to competition and trade policy (the market access/entry component), which could
have, in principle, a similar impact on trade than geography. Finally, both the quantity and quality
of physical infrastructures (the development component) are expected to play important roles.
1
See Harrison (1996) for a review, Yanikkaya (2003) for a comprehensive review and set of estimates and Rodriguez
and Rodrik (2000) for a re-examination of the relationship between trade policy and economic growth and a
critical review of the literature.
3
Various elements are expected to affect supply-side conditions significantly. First of all,
supply conditions are likely to be strongly related to location and the policy variables. The size of
the country, which also determines the size of the internal market, together with the internal
geography of the country are the structural variables that could have an effect on the supply
capacity of a country. Economic policy could also be expected to affect supply capacity by affecting
factor prices. Development variables also have to be taken into consideration. They generally
correspond to stock variables that are most often the outcome of previously implemented policies,
such as public investment in transport infrastructures. Any measure related to general institutional
development could be relevant. Technology could also be seen as part of the development variables
set. In fact any variable able to capture prevailing technological conditions must be considered.
Recently Redding and Venables (2004a) investigated the relative contribution towards export
performance of international linkages relative to internal geographical factors. They find that
most of the differential in export performance of various countries and regions over the last three
decades can be due to differences in the evolution of external components. Nevertheless, they find
that internal components related to supply capacity such as internal geography and institutional
quality also have played a significant role in explaining the observed differential in export
performance.
This study builds on the work of Redding and Venables. It uses the same theoretical model
of bilateral trade flows and adopt a similar empirical strategy. The latter initially consists of building
data series to capture the external components of export performance and, the whole the foreign
market access component, using gravity techniques. Then, these series are used to investigate the
importance of foreign market access relative to supply capacity components. In other words, the
exercise is to identify the possibl main determinants of the supply-side conditions after having
controlled for the external elements. However, this study has a different econometric approach
from that used by Venables and Redding. Econometric techniques are used to control for
unobservable country heterogeneity possibly affecting the real values of countries’ exports.
Accounting for unobservable heterogeneity should allow the identification of any differences in
the effect of and importance of export performance components, which are linked to the degree of
development of the external sector itself. In other words, the techniques used here allow for the
testing for non-linearities in the relationship between export performance and its components.
While dynamic panel techniques would seem to be the most desirable approach, data availability
is likely to restrict their implementation. In this context, cross-sectional analysis proves to be a
viable alternative. Regression techniques which are able to account for unobserved heterogeneity
across countries, namely quantile regressions, are used. Moreover, more emphasis is put on the
determination and impact assessment of variables related to supply conditions. This is done with
the aim of determining as clearly as possible what are policy implications.2
The study reveals important differences across countries and regions when looking at their
respective determinants of export performance. External and internal components prove to have
played an equal role in determining export performance for Asian countries. Their improvement
in the South East and Pacific region appears to be high relative to that observed in any other
region. Sub - Saharan African countries owe their export performance to the evolution of external
components. The latter were strong enough to more than offset a relative deterioration of their
internal production conditions.
2
In his comments on Redding and Venables (2004a), Maskus (2003) insists on the necessity to better identify
supply conditions variables in order to retrieve specific policy implications.
4
Further investigation also indicates that good internal conditions are necessary to obtain
good export performance. Particular attention should be paid to the macroeconomic dimension.
Good infrastructures and non-stringent institutions are also necessary to put the export sector on
a durable development path. In addition, there is scope for promoting a dynamic process of
diversification across and within sectors. Constant efforts to support diversification are particularly
relevant for commodities exporters when a secular downward trend is observed in volatile
commodities prices.
The next chapter presents the theoretical context. The empirical strategy adopted in order
to estimate exports components is presented in chapter II. Chapter III contains the empirical
approach used to assess export performance. Chapter IV presents policy implications and
conclusions.
5
6
I. THE THEORETICAL CONTEXT
Recently developed models of trade provide possible support for investigating the role of
supply capacity in determining the export performance of a country. In particular, the Krugman
and Venables (1995) model identifies an empirically assessable decomposition of bilateral trade
into market access and supply capacity. The theoretical framework is essentially a standard new
trade theory model based on product differentiation derived from a constant elasticity of substitution
demand structure.
The economy consists of a number N of countries. Only the manufacturing sector is

considered. Firms in that sector operate under increasing returns to scale and produce symmetric
differentiated goods, which are used in consumption. Preferences are represented by a CES utility
function in which the elasticity of substitution σ between any pair of products is the same. The
representative utility function of country j is given by
σ / (σ −1)
 N 
Uj = ∑ ni xij(σ −1) / σ  σ>1 (1)
 i 
where ni is the set of varieties produced in country i , and x ij is the consumption in country j of a
single product variety from this set.
In that framework, the demand in country j for each variety produced in country i, is a
function of country’s j total expenditure on differentiated products Ej, the price of the good pij and
the price index Pj defined over the prices of individual varieties produced in i and sold in j. Total
expenditure is assumed to be exogenously given. Demand for each variety writes as
(2)
xij = pij−σ E j Pj(σ −1)
Where
Pj = [∑ n p ] i
1−σ
ij
1 / (1−σ ) (3)
The elasticity of demand is identical across varieties and equal to σ. E j Pj(σ −1) is a scale
factor that indicates the position of the demand curve in market j. The producer price p i is assumed
to be the same for all varieties produced in country i. Transport frictions, which reflect the cost of
getting a good from country i to country j, are set proportional to producers price. This cost is
composed by three elements: the cost of getting the product to and from the border in countries
i and j (ti and tj respectively) and the cost of getting the product across the border (Tij). Intra-
country cost would reflect internal geography and infrastructure. Inter-country cost would reflect
external geography and policy barriers. Thus price pij = p i t i Tij t j and the value of total exports of
country i to country j is given by
ni p i xij = ni pi1−σ (t i Tij t j ) E j Pj(σ −1)

1−σ
(4)
7
Equation (4) is taken as the theoretical support for estimation of a gravity trade model. It
can be rewritten as
[
ni p i xij = ni ( p i t i )
1−σ
](T )[E (P / t )( ) ]
1−σ
ij j i i
σ −1 (4’)
The first term reflects the supply capacity of the exporting country, thereafter denoted b y
sci. It is the product of the number of varieties and their price competitiveness, which is measured
by the product of the producer price and internal transport costs. The last term measures the
trans-border transport costs component. The first term into brackets is the market capacity of
country j, thereafter denoted by mj. It depends positively on total expenditures in j, on country j
internal transport costs and, on the number of competing varieties and their prices expressed in
the price index.
At the country level, that is looking at the total value of exports of country i, the following
is obtained
X i = ni pi ∑ j ≠i xij = sci ∑ (Tij ) m j

1−σ (5)
j ≠i
The term ∑ (Tij ) m j represents country i foreign market access FMAi or equivalently
1−σ
j ≠i
country i market potential which refers to the concept developed by Harris (1954). It corresponds
to the sum of the market capacity of all country i exports destination countries, weighted by
bilateral trade costs. Then, the product of supply capacity and foreign market access gives the total
value of exports of country.
The relative importance and evolution of these components are investigated empirically
in the next chapter.
8
II. THE COMPONENTS OF EXPORT GROWTH
The model presented above postulates that the effect of a rise in expenditure on traded
goods in a given country would benefit relatively more than those of its trading partners that are
relatively closer (the demand pecuniary effect). In this context, distance has to be interpreted not
only as a pure geographical element but also as any element that possibly represents a barrier to
trade, such as tariffs, non-tariff barriers, anti-competitive barriers, etc.
The model also suggests that in order to capture fully the demand pecuniary effect just
described, favourable supply conditions are expected to play an essential role. In addition, equation
(5) shows that access to foreign markets may be reduced by poor supply capacity.
(a) The Dataset
Bilateral trade flows for 84 countries are obtained from the UN COMTRADE database.
Data are deflated by the United States GDP deflator in order to obtain real values. The base year
for the deflator is 1995. Data on trade flows are combined with geographical characteristics and
data on GDP. Sources are detailed in Appendix A.
Some countries do not report all of their trade flows. In that situation, information is
completed by using mirror data, that is, data declared by the trade partner. This is likely to be
imprecise and, as a consequence it increases possible measurement error. To account for the latter,
data are weighted by the product of trade partners’ GDP in all regressions based on bilateral trade
flows.
As bilateral trade flows are usually characterized by high year-to-year fluctuations and this
study is essentially concerned with medium to long-term determinants of export performance,
they are averaged over four year periods. It examines export performance over the 1980-99 time
spell, which gives five periods of analysis.
Some sensitivity analysis based on different period spells and country samples is presented
in section II(c).
(b) Estimation Strategy
In determining the export performance of a given country, it is first necessary to quantify

the respective roles of foreign market access and supply capacity. Total export growth can be
decomposed into supply capacity and foreign market access growth. The approach consists of
estimating a gravity model equation where the dependent variable is exports (logarithm) from
country i to country j and the dependent variables are bilateral distance (logarithm), an indicator
of the existence of a common border, exporter-country dummies and importer-partner dummies.
To account for region specific trade frictions, dummies that indicate whether trade partners belong
or not to the same geographical region, are introduced.
9
ln(X ij ) =α + λpartn j + βcounti
543 52423 5
242 3
Value of Exports Export partner Market Capacity Supply Capacity
from i to j (2)
+ γ 1 ln(distij ) + γ 2bord ij + γ 3 regionrr + uij
522222 24222222 3 1
Bilateral Trade Costs Stochastic error
Bilateral distance distij and the border dummy bordij are assumed to capture geographical
bilateral trade costs. Namely, geographical bilateral costs or a monotonic transformation of them
are estimated through dist ijγˆ1 exp(γˆ 2 bord ij )∏ exp(γˆ3 regionrr ) .
r
Exporters and importer partners fixed effects, counti and partn j respectively, are introduced
to control for supplier capacity and market capacity. They can also be expected to control for
institutions and policy related bilateral trade costs.
The results excluding and including region specific trade friction dummies, are reported
in Tables 1 and 2 respectively. Tobit estimation is used to account for zero bilateral trade values. In
addition, in order to allow for measurement error in bilateral trade flows that is correlated with
the volume of trade, observations are weighted by the product of country and partner GDP.
As expected, the coefficient on “distance” is negative and the one on “border” is positive. A
negative coefficient on “distance” indicates that the further away two countries are from each
other the less they would tend to trade. The positive coefficient on “border” indicates that the
immediate proximity of two countries increases their tendency to trade. Moreover, the introduction
of dummies for region specific trade costs logically lowers the impact of distance and amplifies
that of the existence of a common border. The within-region trade dummies are jointly statistically
significant at the 1 per cent level in all periods. Coefficients on within-region dummies reflect the
effect of any element that facilitates trade within the region. Thus, it could coherently be the case
that coefficients are negatively signed, the important element being their evolution over time.
Table 1
Bilateral trade equation estimation
Period 1980-83 1984-87 1988-91 1992-95 1996-99

Obs 6972 6972 6972 6972 6972
Ln(X ij ) 1 2 3 4 5
Ln(dist ij ) -1.013 -1.007 -0.968 -0.946 -0.941
0.021 0.022 0.016 0.014 0.014
bord ij 0.366 0.364 0.46 0.638 0.673
0.061 0.066 0.048 0.043 0.045
Country dummies Yes yes Yes yes Yes
Partner dummies Yes yes Yes yes Yes
Estimation Tobit Tobit Tobit Tobit Tobit
Log Like. -18882 -18959 -17652 -17756 -18271
LR Chi2 10922 10808 14378 15780 15603
Prob > chi2 0 0 0 0 0
Pseudo R2 0.2243 0.2218 0.2894 0.3077 0.2992
10
Table 2
Bilateral trade equation estimation with intra-regional trade dummies
Period 1980-83 1984-87 1988-91 1992-95 1996-99

Obs 6972 6972 6972 6972 6972
Uncensored 5943 5841 6062 6357 6455
Ln(Xij) 1 2 3 4 5
Ln(distij) -0.715 -0.800* -0.764* -0.783* -0.840*
0.038 0.040 0.029 0.025 0.026
bordij 0.554 0.464* 0.486* 0.619* 0.616*
0.065 0.071 0.050 0.045 0.049
Country Yes Yes Yes Yes Yes
dummies
Partner Yes Yes Yes Yes Yes
dummies
Weap 0.588* 0.114 -0.035 -0.107*** -0.231*
0.106 0.113 0.076 0.065 0.067
Wlac -0.010 0.121 0.436* 0.473* 0.448*
0.195 0.219 0.162 0.138 0.144
Weca 2.319* 2.054* 1.817* 1.466* 1.191**
0.784 0.805 0.615 0.595 0.610
Wmena -5.409* -7.137* -5.598* -3.467* -5.292*
0.917 0.966 0.685 0.568 0.582
Wna -0.544* -0.492* -0.303* -0.221** -0.182***
0.144 0.152 0.109 0.097 0.102
Wsa -0.228 -0.472 -0.463 -0.677 -0.298
1.292 1.259 0.812 0.656 0.638
Wssa -6.746* -8.871* -11.018* -0.621 0.631
1378 1.621 1.187 1.030 1.074
Wwe 0.898* 0.809* 0.875* 0.796* 0.666*
0.078 0.082 0.059 0.052 0.055
Estimation Tobit Tobit Tobit Tobit Tobit
Log Like. -18743 -18836 -17396 -17544 -18086
LR Chi2 11199 11055 14889 16203 15971
Prob > chi2 0.000 0.000 0.000 0.000 0.000
Pseudo R2 0.23 0.2269 0.2997 0.3159 0.3063
Prob> F 0.000 0.000 0.000 0.000 0.000
(dummies)
In the case of North America the coefficient on the regional dummy is negative. Nevertheless,
the negative impact decreases sharply over the 1988-1995 period suggesting that trade has been
eased. This is likely to capture the trade facilitation impact of NAFTA.
The coefficient on the Eastern Asia and Pacific intra-regional trade dummy is initially
positive and then becomes negative. This is not an indication of falling intra-regional trade or of
a rise in intra-regional trade barriers. Rather, it reflects the impact of rapid growth of trade with
partners outside the region.
The coefficient on the Western Europe regional dummy is positive for the all periods
considered but shows a general tendency to decrease. More precisely, it first decreases over the
1980-87, period increases over the 1988-91 period and, decreases sharply over the 1992-99 period.
The overall fall in the coefficient can be seen as the consequence of on one hand the stabilization
of intra-regional trade due to a finalized integration process of the EU countries and, on the other
11
hand of the growing relative importance of trade with countries outside the region. Part of the fall
in the value of the coefficient observed for the 1992-99 period could also be explained by the
apparent fall in intra-regional trade due to the introduction in January 1993 of a new system for
collecting statistics on trade between EU member States INTRASTAT.3
The same downward tendency is observed for Eastern European countries with the sharpest
decrease in the coefficient during the period that includes the fall of the Berlin wall. Regressions
have also been run with a regional dummy that includes all Western and Eastern European countries.
The coefficient on the dummy and its significance both increase, although moderately, over time.
This is likely to capture the process of European enlargement and integration and confirms the
arguments just mentioned.
Coefficients on the sub-Saharan region dummy are negative and large compared to other
regions’ coefficients when significant. This very much reflects poor infrastructures and poor
geographical factors. For the last two periods, the coefficient turns to be non-significant. This
result might reflect an improvement in the trade conditions compared to those prevailing in the
1980s. However, it could also reflect the fact that trade integration has occurred more at a sub-
regional level than at a regional level. Trade has intensified radically between countries belonging
to the Southern African Development Community over the 1980-99 period compared to other
regional economic communities. The share of intra-trade in regional trade jumped from 54 per
cent in 1980 to 79 per cent in 1999.4 An upward trend is observed for most of the regional
country groups, which again could indicate that trade has become more sub-region specific than
region specific.
The Middle East and North Africa dummy has a negative coefficient, which increases and
decreases alternatively. Together with its relatively high value, this is likely to reflect the negative
impact of conflicts within the region and the volatility of prices of oil exports to countries outside
the region.
(i) Foreign market Access
Following Redding and Venables (2004b) estimates obtained in the first stage of the analysis
are used to construct supply capacity and foreign market access series. Because intra-regional trade
dummies are not always significant, series for estimations do not include them. However, results
obtained with series including intra regional trade dummies are discussed below.
The supply capacity estimate is given by the exponential of exporter country dummy times
its coefficient. That is
(
SCi = exp βˆcount i ) (3)
Foreign market access estimate takes the form5

( )
FMAi = ∑i ≠ j exp λˆpartn j distijγˆ1 exp(γˆ2bordij ) (3’)
3
See the GATT annual report 1994 for a brief description.
4
See UNCTAD (2002), in particular Table 1.4.
5
The version with intra regional dummies is
( )
FMAi = ∑i ≠ j exp λˆimp j distijγˆ1 exp(γˆ2bord ij )∏ exp(γˆ3 regionrr )
r
12
Equations (3) and (3’) allow us to decompose the sources of export growth over the last
decades, and the results for the 84 countries are summarized in Table 3. Results aggregated to the
regional level are graphed in Figure 1. Graphs in Figure 1 report the evolution of foreign market
access and supply capacity with respect to the initial level prevailing in each region. Table 4 contains
exports and its components growth rates at the regional level. The last row presents the respective
growth rates for the whole sample of countries.
Figure 1
Foreign market access and supply capacity
EAP ECA NA SA
LAC MENA SSA WE
3.23148 2.71426
Foreign Market Access
Foreign Market Access
.976126 .978753
1 5 1 5
period period
EAP ECA NA SA
LAC MENA SSA WE
1.63145 1.76416
Supply Capacity
Supply Capacity
.388274 .459202
1 5 1 5
period period
Before turning to the interpretation of these results a further step in the decomposition
analysis is taken. Following Redding and Venables (2004a) the geographical sources of the export
growth components are examined.
According to equation (3) foreign market access of country i located in region Rn is given
by the sum of the access to markets in each region. That is
FMAi = FMAiR1 + FMAiR2 + ... + FMAiRn
where FMAiRn denotes country i market access derived from region n. The change in the
market access of country i can be expressed as the sum of the contribution of each region. We have
∆FMAi FMAiR1 ∆FMAiR1 FMAiRn ∆FMAiRn

= + ... + (4)
FMAi FMAi FMAiR1 FMAi FMAiRn
13
14
Table 3
Components of export growth and regional concentration of foreign market access
Exports Growth (%) FMA Growth (%) Sup. Cap. Growth (%) FMA Growth IN (%) FMA Growth OUT(%)
Country region 8087 8491 8895 9299 8087 8491 8895 9299 8087 8491 8895 9299 8087 8491 8895 9299 8087 8491 8895 9299
Austria we 29 61 -10 45 -3 33 82 -8 41 23 -50 62 -5 37 77 -9 3 22 102 -5

Belgium-Luxembourg we 10 48 -6 4 -2 29 74 -8 17 15 -44 12 -5 31 59 -10 3 25 100 -5
Switzerland we 24 49 -10 28 -3 34 78 -8 28 16 -51 48 -4 37 74 -8 6 20 96 -6
Cyprus we -4 18 -22 -14 -3 26 83 -8 75 -18 -58 13 -6 35 62 -9 0 18 102 -6
Germany we 25 45 -15 35 -3 32 83 -7 36 11 -50 51 -5 36 77 -8 4 22 99 -5
Denmark we 17 40 -7 33 -2 32 80 -9 35 0 -46 35 -4 34 76 -10 7 21 96 -6
Spain we 42 66 5 70 -1 29 75 -7 37 12 -42 71 -5 35 66 -8 6 17 94 -6
Finland we 19 42 -2 48 -1 28 74 -8 38 15 -39 73 -5 30 63 -9 8 23 97 -6
France we 13 51 -8 38 -2 32 79 -10 24 19 -49 54 -4 35 76 -11 8 19 95 -6
United kingdom we 5 36 -10 46 -3 25 66 -9 8 16 -47 63 -6 28 57 -11 9 19 95 -6
Republic of Ireland we 52 59 5 106 -1 28 78 -7 31 0 -50 35 -4 34 59 -9 3 21 103 -5
Italy we 28 50 -8 42 0 29 78 -5 77 24 -32 146 -4 34 73 -5 10 18 94 -6
Netherlands we 12 27 -10 32 -3 31 76 -8 39 18 -47 55 -6 37 67 -9 4 20 97 -6
Norway we 10 31 -12 47 -2 30 75 -11 23 -2 -46 51 -4 33 71 -12 8 19 95 -6
Sweden we 20 34 -15 46 -1 27 71 -8 21 19 -47 71 -4 30 61 -8 9 21 95 -6
Greece we 30 24 -11 29 1 30 76 -6 41 23 -50 52 -4 40 65 -6 8 17 94 -6
Portugal we 60 83 -6 50 -1 28 75 -7 31 7 -49 66 -5 31 64 -8 8 22 96 -6
Burkina Faso ssa 5 27 0 38 -9 26 86 -7 40 70 -54 133 -63 0 124 -13 3 28 84 -7

Cote d’Ivoire ssa 12 -18 -17 35 -2 27 86 -7 11 -53 -44 20 -59 11 156 -11 4 28 84 -7
Ghana ssa -16 29 18 -15 -4 26 86 -7 -26 50 2 -41 -58 -8 125 -16 4 28 84 -7
Kenya ssa -12 -3 24 16 -1 31 91 -8 -17 -11 -44 5 -47 5 170 -15 3 32 88 -8
Madagascar ssa -10 -15 -10 -28 -1 33 95 -8 25 4 -52 -28 -43 9 164 -14 4 34 92 -8
Mali ssa -9 82 -38 34 0 26 84 -7 0 82 69 -25 -54 2 138 -15 4 27 82 -7
Mauritania ssa 8 16 -25 33 2 25 82 -7 115 91 -15 73 -50 7 126 -13 5 26 81 -6
Mauritius ssa 46 74 -7 36 0 35 96 -8 76 86 -27 51 -45 7 169 -14 4 36 94 -8
Niger ssa -24 -4 -67 2 -7 26 83 -7 -25 10 -60 -68 -63 -2 116 -13 3 28 82 -7
Nigeria ssa -43 7 -23 4 1 28 85 -7 -74 22 -4 -35 -40 15 174 -14 3 29 83 -7
Rwanda ssa 36 -17 -68 -34 -2 30 90 -8 -23 17 -90 -98 -45 8 150 -16 3 32 87 -7
Senegal ssa -6 22 -37 -34 2 25 83 -6 2 27 -53 -77 -55 5 147 -14 5 25 82 -6
Tanzania ssa -33 -1 28 -11 -1 32 92 -8 -61 -2 30 -32 -42 13 150 -15 3 33 89 -8
Uganda ssa -13 -41 19 19 -2 30 90 -8 -19 -26 4 -54 -42 9 140 -14 3 32 88 -7
South Africa ssa -5 18 -5 19 2 32 91 -7 -2 -10 -41 44 -49 21 91 -14 4 32 91 -7
Zambia ssa -53 29 8 -44 -3 30 95 -8 -76 -11 -16 -31 -40 9 162 -12 4 32 89 -7
/...
Table 3
Components of export growth and regional concentration of foreign market access (cont’d.)
Country region 8087 8491 8895 9299 8087 8491 8895 9299 8087 8491 8895 9299 8087 8491 8895 9299 8087 8491 8895 9299
Bangladesh sa 61 39 28 109 -3 27 84 -7 94 32 -28 129 3 5 85 -15 -3 27 84 -7
India sa 22 47 38 50 3 38 98 -9 23 11 -21 61 -2 14 98 -31 4 39 98 -8
Sri Lanka sa 16 26 51 51 3 39 98 -9 38 -2 -5 58 3 -1 75 -14 3 41 99 -8
Nepal sa 47 24 40 43 5 35 101 -8 -21 92 8 34 3 2 101 -9 5 42 101 -8
Pakistan sa 14 59 2 34 2 32 97 -8 32 17 -39 41 4 1 107 -6 2 38 96 -9
Canada na 24 22 3 47 24 8 103 -1 -6 6 -50 35 29 4 105 0 -8 43 89 -8

Mexico na 23 -8 80 120 18 18 105 -2 19 -34 -47 106 -34 12 175 -1 22 18 102 -2
United States na 1 49 0 46 -1 35 88 -7 -2 7 -48 51 19 20 73 -6 -10 44 95 -7
Algeria mena -25 -8 -30 41 -1 28 72 -8 -53 10 -57 41 -26 7 119 -15 0 29 71 -7

Egypt, Arab Rep. mena -28 2 3 11 -3 29 89 -8 -42 17 -31 19 -31 10 177 -10 2 31 81 -7
Iran, Islamic Rep. mena -13 -1 -15 -54 1 30 87 -8 -37 -33 -60 -62 -26 -14 103 -15 4 33 86 -7
Israel mena 26 49 6 63 -5 24 85 -8 -12 47 -25 62 -31 -12 118 -10 2 31 81 -7
Jordan mena 87 23 -29 67 -5 25 83 -8 1145 99 -25 113 -34 -16 105 -19 2 31 81 -7
Kuwait mena -64 -96 535 -68 -3 31 86 -9 -61 -97 447 -78 -40 11 108 -28 3 33 85 -8
Morocco mena 2 45 -18 61 0 27 75 -7 7 35 -45 63 -32 -1 87 -24 2 28 74 -7
Syrian Arab Republic mena -37 26 19 19 -4 26 95 -6 -77 396 17 -58 -26 1 160 -6 3 32 83 -6
Tunisia mena -12 43 4 44 -3 27 71 -8 -14 -13 -28 61 -31 -2 85 -22 0 29 70 -8
Argentina lac -4 38 24 66 -4 26 111 -3 -15 4 -43 55 -39 35 215 5 11 24 84 -7

Bolivia lac -32 7 -3 24 -2 21 107 -3 -71 53 11 -7 -40 21 224 4 14 21 82 -6
Brazil lac 21 17 -6 32 1 22 96 -6 18 -11 -53 34 -46 22 228 -4 11 22 82 -6
Chile lac -9 74 11 44 -5 22 107 -5 -8 57 -44 58 -49 12 247 2 12 24 84 -7
Colombia lac 22 28 2 47 5 15 95 -4 18 3 -46 39 -33 10 180 0 17 16 79 -6
Costa Rica lac 1 23 19 120 8 14 90 -5 -11 27 -27 186 -31 12 164 0 18 14 79 -6
Domican Republic lac 13 80 -19 54 12 11 83 -5 -1 172 -54 101 -35 11 145 -3 19 11 77 -5
Ecuador lac 8 -1 10 45 4 16 95 -5 -20 5 -11 71 -35 12 183 -4 16 17 80 -6
Guatemala lac -23 -6 18 51 8 16 92 -3 -30 -13 -36 44 -33 37 167 6 19 13 79 -5
Honduras lac -6 -31 -5 38 10 14 88 -4 -32 -2 -65 70 -33 24 156 2 19 12 78 -5
Jamaica lac -40 36 13 0 13 10 83 -5 -77 151 44 -40 -34 15 149 -1 20 10 77 -5
Nicaragua lac -47 -17 51 27 9 14 88 -4 -35 -81 158 13 -31 21 155 2 19 13 79 -5
/...
15
16
Table 3
Components of export growth and regional concentration of foreign market access (concluded)
Country region 8087 8491 8895 9299 8087 8491 8895 9299 8087 8491 8895 9299 8087 8491 8895 9299 8087 8491 8895 9299
Panama lac -8 -3 29 21 8 15 91 -5 -19 115 -66 197 -34 17 173 -4 18 14 79 -6
Peru lac -24 12 22 25 3 20 99 -4 -30 -7 -33 30 -36 24 201 1 15 20 81 -6
Paraguay lac 2 125 -26 39 -8 22 116 -2 -55 129 -53 -30 -45 19 238 6 12 23 83 -6
El Salvador lac -4 -36 58 35 8 15 89 -4 -58 -21 33 1 -33 29 155 1 19 13 79 -5
Trinidad and Tobago lac -45 -20 -5 24 6 14 86 -5 0 -35 -53 -27 -37 7 142 -4 16 15 79 -6
Uruguay lac -9 35 -4 50 -31 19 160 -2 -4 5 -57 42 -59 9 316 2 11 24 84 -7
Venezuela lac -30 14 -4 33 9 15 91 -4 -27 -2 -61 0 -33 18 196 2 17 14 79 -6
Bulgaria eca -24 6 36 12 5 40 100 -9 -17 -22 -47 24 -2 38 117 10 5 40 100 -9

Hungary eca 1 31 3 122 -2 28 72 -9 32 23 -55 103 -28 34 114 5 -2 28 71 -9
Poland eca -17 43 16 56 -2 32 81 -9 -42 64 -46 35 -8 23 119 6 -2 32 80 -9
Romania eca 8 -21 -15 42 -2 28 76 -8 23 -21 -68 42 0 28 118 7 -2 28 74 -9
Turkey eca 97 42 -2 65 -5 29 82 -9 106 48 -48 60 -46 36 122 -3 -4 29 81 -10
Australia eap 3 53 -7 37 6 38 98 -8 3 18 -56 45 10 64 111 -11 3 21 87 -5

China eap 58 59 46 83 8 49 96 -9 70 -1 28 115 13 72 104 -11 3 21 82 -6
Hong Kong, China eap 44 30 -28 1 8 53 103 -8 26 -2 -65 6 13 79 115 -9 3 21 81 -6
Indonesia eap -11 13 16 48 6 45 115 -8 5 12 -25 75 10 63 134 -8 1 23 83 -6
Japan eap 49 31 -2 22 10 45 104 -7 46 5 -54 31 19 88 131 -7 5 20 81 -6
Republic of Korea eap 96 63 3 60 10 40 87 -11 26 143 -47 70 15 51 90 -12 4 21 80 -6
Malaysia eap 28 51 49 59 8 53 135 -7 15 27 -13 65 11 66 152 -8 1 22 82 -6
New Zealand eap 10 31 -2 25 5 32 92 -8 20 -7 -47 36 7 49 96 -11 4 20 88 -5
Philippines eap -8 38 25 131 9 50 100 -10 -23 16 -32 111 14 70 110 -11 3 22 82 -6
Singapore eap 31 79 32 53 -3 46 143 -7 56 35 -35 61 -5 63 174 -7 1 22 83 -6
Thailand eap 39 122 42 51 5 44 112 -8 62 90 -25 66 8 64 133 -8 1 21 81 -7
Taiwan eap 87 56 -4 49 9 41 97 -9 62 190 -27 67 12 52 104 -11 3 21 81 -6
(Province of China)
Equation (4) indicates that the contribution to country i foreign market access growth of
a given region is larger the larger the share of this region in country i foreign market access or the
larger the increase in market demand in the partner’s region.
Graphs in Figure 2 show the evolution of foreign market access related to trade within the
region and across regions. Table 3 contains growth rates at the country level. Table 5 reports the
latter at the regional level. Benchmark values are presented in the last row.
As to the interpretation of these results, generally speaking, all regions have faced the same
upward trend since the beginning of the 1980s in their access to foreign markets as can be observed
in the upper graphs of Figure 1, although this trend is partially reversed in the last period. This is
equally true for all regions, as indicated in Table 4. Nevertheless, there is an upward overall tendency
which denotes the increasing degree of integration in the world trade system.
Table 4
Components of regional exports growth
Exports Growth FMA Growth Supply Capacity Growth

region 8087 8491 8895 9299 8087 8491 8895 9299 8087 8491 8895 9299
eap 44 44 8 43 7 46 108 -8 42 34 -88 53

eca 9 23 4 66 -2 31 80 -9 7 34 -90 48
lac -6 21 2 41 3 16 96 -4 -4 3 -95 38
mena -26 -1 -6 17 -3 28 81 -8 -36 -1 -66 32
na 8 37 5 52 20 13 101 -2 -14 17 -102 43
sa 22 47 30 51 2 34 96 -8 30 19 -48 55
ssa -23 10 -12 11 -2 29 89 -7 -25 -7 -64 16
we 19 44 -10 38 -2 30 77 -8 26 17 -85 50
Benchmark 17 40 -2 42 1 28 87 -7 20 21 -86 49
Table 5
Geographical composition of regional foreign market access growth
Foreign Market Access FMA Growth Within the FMA Growth Outside the
Growth Region Region
region 8087 8491 8895 9299 8087 8491 8895 9299 8087 8491 8895 9299
eap 7 46 108 -8 10 64 122 -9 3 21 82 -6

eca -2 31 80 -9 -17 30 117 5 -1 31 79 -9
Lac 3 16 96 -4 -40 18 196 1 16 16 80 -6
mena -3 28 81 -8 -31 -3 127 -14 1 30 78 -7
na 20 13 101 -2 27 5 104 0 7 28 98 -4
sa 2 34 96 -8 3 3 100 -12 2 37 95 -8
ssa -2 29 89 -7 -51 6 156 -14 4 30 86 -7
we -2 30 77 -8 -5 34 70 -9 6 20 97 -6
Benchmark 1 28 87 -7 -6 33 91 -8 6 25 85 -7
17
Figure 2
Regional origin of foreign market access
EAP ECA NA SA
LAC MENA SSA WE
4.02662 2.71225
Foreign Market Access IN
Foreign Market Access IN

.598834 .487027
1 5 1 5
period period
EAP ECA NA SA
LAC MENA SSA WE
2.41692 2.72625
Foreign Market Access OUT
Foreign Market Access OUT
.986391 1
1 5 1 5
period period
The results indicate that the North American countries (in this case the United States
Canada and Mexico) saw their foreign market access measure increase exponentially over the
1984-95 period. As reported in Table 4, foreign market access grew by 101 per cent over the
1988-95 period. As can be seen from the upper right graph of Figure 2 and Table 5, this is due to
the important relative increase in the intra-regional market access component. This appears to be
a clear indication of the positive impact the NAFTA has generated. Nevertheless, North American
countries also enjoyed a large increase in foreign market access outside the region. As reported in
Table 3, among the three NAFTA countries, Mexico has the highest growth –175 per cent – in
foreign market access generated within the region over the 1988-95 period. The corresponding
figure is 73 per cent and 103 per cent respectively for the United States and Canada.
East Asian and Pacific countries also benefited from a large increase in foreign market
access. As indicated in Table 4, it has been always above the benchmark figures. The latter has
been driven essentially by a rise in foreign market access within the region. Those countries that
on average enjoyed the highest growth rates are Singapore and Malaysia, as reported in Table 3.
Contrary to intra-regional foreign market access, the relative evolution of foreign market access
generated outside the region is less than that experienced by other regions. Figures in Table 3 show
a very high degree of homogeneity in the evolution of the latter component of foreign market
access, which tends to reinforce the view that regional integration has been particularly strong for
East Asian and Pacific countries.
Middle Eastern and African countries initially experienced a fall in foreign market access
generated within their regions. However, this trend was strongly reversed in the 1990s. Table 5
indicates that over the 1988-95 period foreign market access within the region grew by almost 160
18
per cent for sub-Saharan countries and 130 per cent for Middle Eastern and North African countries.
The highest growth rates are found for East African countries, which are also the best performers
in terms of overall foreign market access growth. However, this general tendency was subsequently
reversed.
A similar scenario holds for Latin American and Eastern European countries. Intra regional
foreign market access grew by almost 200 per cent in Latin America over the 1988-95 period. As
reported in Table 3, the higher rates of foreign market growth are found for countries belonging to
the MERCOSUR that was effectively launched at the beginning of the 1990s. The positive impact
of this regional trade integration process is captured by above sample average growth rates of the
intra-regional market access. The best performer in all foreign market access dimensions is Uruguay.
Table 3 also shows that these countries also benefited from the high growth of market access from
outside their region.
Foreign market access in South Asia is driven by both sources of access, although it appeared
to be driven principally by extra regional market access in the second half of the 1980s. Intra-
regional trade progression has also been positive although it tended to fall in the second half of the
1990s as suggested by Table 5.
Foreign market access progression for Western European countries remains among the
lowest in the period under observation. This reinforces the argument presented above that those
countries were already well integrated at the beginning of this period and just followed the general
trend. As shown in Table 5, growth rates for the region correspond almost exactly to those of the
whole sample. Moreover, Table 3 indicates that this result holds for all countries of the region.
Such an homogeneous pattern is likely to reflect a strong degree of integration. Estimates also
show fast growth of market access generated outside the region over the 1988-95 period. This
probably reflects the EU’s growing web of regional agreements with Central and Eastern Europe,
the Baltic States and the Euro-Med agreements.
Generally speaking, there is evidence of a positive impact of Regional Trade Agreements

on trade between partner countries in particular over the 1984-95 period. Nevertheless, there is
also evidence of a strong positive contribution to foreign market access improvements and of
better access to non-neighbouring markets as indicated in the two lower panels of Figure 2,
particularly in the 1984-95 period. For instance, the measure of foreign market access generated
outside the region for Eastern European countries has increased considerably since the mid-1980s
reflecting essentially the increasing integration of those countries to the Western European market.
For all regions, an increasing upward trend in the mid-1980s can be observed, which corresponds
to the beginning of an era of trade openness marked by the extensive unilateral liberalism under
World Bank/International Monetary Fund programmes, the implementation of the results of the
Tokyo Round and, the growth of Regional Trade Agreements.
(ii) Supply capacity
The relative evolution of supply capacity is slightly more differentiated than that of foreign
market access. There is no clear overall trend. However, all regions faced a sharp relative decrease
over the first half of the 1990s.
Asian countries show the largest relative increase in their supply capacity in the 1980s and
the lowest relative fall at the beginning of the 1990s. The best performers over the two decades
19
were Taiwan (Province of China) and Singapore. Figures reported in Table 3 indicate that the
bulk of the growth in supply capacity occurred in the 1980s. The Chinese and the Philippines’
supply capacities grew outstandingly in the 1992-99 period. Asian countries were also the best
performers in relative terms over the two decades. There is an almost symmetric evolution between
East Asian foreign market access and supply capacity evolution. In particular, the fall in supply
capacity in the first half of the 1990s was offset by an upward shift in foreign market access.
The African and the Middle Eastern countries mostly experienced low or even negative
growth in their supply capacities over the whole of the1990s as shown in Table 3. As a whole,
growth rates turned positive only in the second half of the 1990s as shown in Table 4. This may
reflect to a large extent the negative impact of conflicts on infrastructure and related investment.
Table 4 indicates that North and Latin American countries experienced the largest relative
fall in supply capacity over the 1988-95 period. Surprisingly enough, Table 3 reports that the
largest fall in supply capacity among North American countries was in the United States. However,
such observations are theoretically coherent in the context of strong regional integration. As shown
in Table 3, a decline in supply capacity was also experienced by most Latin American countries up
to the first half of the 1990s. Export performance, if not negative, remained very low in that
period, most likely as a result of the impact of economic turmoil that characterized the region.
This is also true to some extent for Western European countries. They faced a severe fall
in their supply capacity at the beginning of the 1990s after a decade of improvement. Nonetheless,
the trend was reversed again in the second half of the 1990s. Table 3 indicates that except for
Cyprus and Norway in the 1984-1991 period all European countries moved together. Together
with negative export growth, the fall in supply capacity observed for the 1988-95 period could
reflect the negative impact that German reunification had on European economies.
(iii) Export Constraints
In order to identify export performance constraints more closely and to qualify the above
arguments, it is necessary to look at the evolution of performance and its components with
respect to the country sample values to be able to elicit the policy implications. Ratios of regional
values over sample values are computed for each period and then normalized to the ratio prevailing
in the first period.6 This makes it possible to qualify the evolution of export performance for each
region across periods and with respect to world export performance for each period. Export
performance has been defined theoretically as the product of foreign market access and supply
capacity. That is, the exports ratio is equal to the product of the foreign market access and the
supply capacity ratios up to an error term related to estimation.
The Asia and Pacific regions are the only regions that have improved their export
performance relative to the whole sample of countries in all periods (Graph 1). Both regions, and
in particular South Asia and the Pacific, have experienced a relative improvement in their foreign
market access across periods. However, their export performance is driven by an outstanding
relative improvement in their supply capacity. This is likely to reflect a policy orientation aiming
6
For instance the bar plotted in graph 1 for Exports 84-87/80-83 in region eap correspond to
(Exportseap 84-87/ Exportssample 84-87)/ (Exportseap 80-83/ Exportssample 80-83).
20
Graph 1
Benchmarked export performance and components
Exports FMA Sup.Cap.

1.4
1.2
0.8
0.6
0.4
0.2
0
eap eca lac mena na sa ssa we
84-87/80-83
1.4
1.2
0.8
0.6
0.4
0.2
0
88-91/80-83
/...
21
Graph 1 (concluded)
Benchmarked export performance and components
Exports FMA Sup.Cap.

1.6
1.4
1.2
0.8
0.6
0.4
0.2
0
92-95/80-83
1.8
1.6
1.4
1.2
0.8
0.6
0.4
0.2
0
96-99/80-83
(Benchmark: ratio to 80-83 World Export Performance and Components)
22
to support and stimulate exporting firms’ productive capacities. This policy consisted not only in
levelling the playing field for exporters, but in boosting it in their favour by employing
interventionist policies7 such as the coordination of investment plans, directed credits and infant
industry protection.
Western European countries behaved in a similar manner as the country sample figures.
Their export performance is led to some extent by supply capacity despite its relative deterioration
over the 1992-1995 period. In fact, foreign market access fell relatively in the last two periods.
This deterioration with respect to sample levels indicates the strong degree of integration of the
region. This is confirmed by the fact that intra-regional trade has been constantly growing and
today represents almost three quarters of the total trade of European countries.
The experience of Eastern European countries was similar to that of their Western
counterparts, although supply capacity became a binding constraint element in the aftermath of
the fall of the Berlin wall. All in all, their relative position remained stable over the two decades.
The experience of North American countries is to some extent puzzling, as it might have
been expected to be quite similar to that of Western European countries. Instead, their relative
situation tended to deteriorate across periods, although only slightly, because of a relative
deterioration of supply capacity conditions. There is clear evidence that their foreign market access
position improved, but their performance was constrained by a poor evolution of supply capacity
conditions. On the other hand, theoretical insights predict that a negative relationship is likely to
appear between the two dimensions. This is what is found when looking at regional growth rates.
Supply capacity tends to decrease while foreign market access tends to increase.
With the advent of the NAFTA, some production has been shifted from the United States
to northern Mexican regions. This could explain the fall in supply capacity in the United States
over the 1988-95 period. Simultaneously, this new productive area bid up salaries, essentially
those of skilled workers in Mexico, which may have reduced supply capacity in the country, as
predicted by the model previously presented.
The relative situations of the African and Middle Eastern countries’ tend to deteriorate
over time. There is clear evidence that export performance is led by foreign market access. The
latter appears to be stable with respect to sample levels, which indicates that those countries are
likely to face constraints in the development of supply capacity.
Regarding Latin American and Caribbean countries, foreign market access drives export
performance. However, the latter fell in relative terms over the first two periods and stabilized
afterwards. Supply capacity tended to deteriorate over the whole period. As in the case of African
countries this can be taken as an indication of the existence of supply capacity constraints.
Overall, a careful examination of the results derived from the estimation procedures outlined
above indicates that supply capacity constraints seem to represent a significant barrier to the
development of the export sector. This is true for both developed and developing countries.
7
See The World Bank (1993) for a comprehensive argumentation.
23
Over the two decades under consideration, export performance has been led essentially by
an overall increase in both market access and supply capacity in Asian and Pacific countries. This
is true also for European countries although on a much lower scale. North American countries’
exports strongly benefited from NAFTA despite a poor evolution in their supply capacity, which
can be seen as a by-product of strong production integration. Export performance in Latin American,
Caribbean, African and Middle Eastern countries was mainly influenced by foreign market access
evolution. There is also evidence of an important impact of regional trade agreements within a
general increasing world trade integration tendency.
(c) Sensitivity Analysis
In order to evaluate the robustness of thr results, foreign market access components were
estimated using a sample of 149 countries. While the main conclusions are unchanged, there are
few differences in the estimates of growth rates but, not enough to change the conclusions.
Five-year periods were also considered. Again, qualitative results remained similar.
Further, using estimates that include dummies for intra-regional trade, the results are very
similar both in quantitative and qualitative terms. However, due to the loss of significance of the
dummy for the sub-Saharan region in the last two periods, estimates for countries of that region
lose some coherence.
Finally, a different estimation strategy was adopted, based on a more structural approach.
More precisely, the study estimates the following gravity model
ln(X ij ) = α + λ ln(GDPi ) + β ln(GDPj ) + γ 1 ln (distij ) + γ 2bord ij +
γ 3llocki + γ 4llock j + γ 5isli + γ 6isl j + γ 7 openi + γ 8open j + uij
where GDPi and GDPj stand for country and partner GDP, respectively. They represent country
i supply capacity and partner j demand capacity respectively. Trade costs estimates are augmented
by a series of indicators. Dummies llocki and llockj indicate whether the country and the partner
respectively are landlocked. Dummies isli and islj indicate whether the country and the partner
respectively are islands. Indicators openi and openj correspond to the Sachs and Warner (1995)
composite index.8
In this case the results remain similar from a qualitative point of view on aggregate, but
show some large differences across time. Foreign market access series obtained with the two
estimation strategies are highly correlated. However, R-squared values are lower in the second
approach than in the first. This implies that estimated values of exports could appear to be
significantly different from real ones. As a consequence, the estimated series for foreign market
access may not be fully consistent with real export series.
8
The latter establishes according to five criteria – tariffs, quotas coverage, black market premium, social organization
and the existence of export marketing boards – whether a country runs or not an open trade policy.
24
III. EXPORT PERFORMANCE AND ITS DETERMINANTS
This chapter attempts to clarify further the key components of supply capacity. It also
attempts to account for non-linearities in the process of development of the external sector. For
that reason econometric techniques able to deal with unobservable heterogeneity are used.
(a) An Extended Theoretical Framework
In the theoretical framework presented previously, supply capacity, as indicated in equation

(4’), is a function of the size of the export sector measured by the number of varieties produced,
producer prices and internal transport costs. A country’s GDP and its population are measures of
country size. Country size reflects the home market and is likely to be linked to the size of the
external sector and export prices. The latter directly reflect comparative costs of exporting which
are also linked for instance to institutions or real exchange rates.
Supply capacity is also expected to depend on foreign market access. Better access to
international markets would imply higher expected returns from export activities. As a consequence,
the external sector would tend to expand with some impact on supply capacity. The relation between
foreign market access and supply capacity is thus made endogenous. In order to qualify the
consequences of such a relationship, some general equilibrium features needs to be added to the
theoretical framework.
Redding and Venables (2004a) consider a production possibility frontier between exports
and other goods. In that context the model predicts a negative relationship between foreign market
access and supply capacity. High levels of foreign market access are expected to be associated with
a less than proportional increase in exports and a lower level of supply capacity. An expansion of
the export sector increases the cost of factors by increasing demand pressure and thus leads to
higher producer prices, which are negatively related to supply capacity. However, the sign of the
relationship could be arguable. Better foreign market access could also draw production resources
from abroad via foreign direct investment or labour migration. In that case, factor demand pressure
could be eased and the sign of the relationship could become uncertain at least to a certain extent.
Empirically, if the first effect (factor prices) dominates the second (factors supply) an estimate
of the elasticity of export performance with respect to foreign market access, which is less than one
would be obtained. In other words, export performance would be expected to grow less than
proportionally than foreign market access. On the contrary, if the elasticity of export performance
with respect to foreign market access is greater than one, then exports would growth proportionally
more than foreign market access.
(b) The Data
Sources of data on the variables described in the next chapter are presented in Appendix A.
Data availability is a major constraint and in order to keep analytical relevance and statistical
coherence, empirical investigations are run for the three 4-year periods covering 1988-1999. In
25
the core set of regressions, information remains available for all countries belonging to the sample
used for foreign market access empirical estimations. Chapter III(d) contains only the most
statistically significant results. Some sensitivity analysis contemplating various sets of possible
dependent variables is presented in chapter III(e).
(c) Estimation
The empirical counterpart of the extended framework takes the form:
ln( X i ) = α + λ ln(GDPi ) + β ln (POPU i ) + γ ln (FMAi ) + δ ln (ti ) + χCOMPi + ui
where POPi is population, ti is internal transport costs and related features and COMPi is a variable,
or set of variables affecting export sector competitiveness, either directly or indirectly.
It is usually argued that, in such empirical models, GDP is likely to be endogenous. This
issue is dealt with by considering one-period lagged GDP values.
Variables used to control for the competitiveness environment are assumed to be related to
the institutional and macroeconomic frameworks. Two indicators are used for institutional quality.
The first is the widely used index from the International Country Risk Guide database. It measures
the risk of expropriation, which is associated with institutional quality. A higher value of the index
is associated with better institutional quality. The second indicator is specific to labour market
institutions and as such is expected to reflect more precisely the labour cost dimension. Nevertheless,
it remains a qualitative measure. The indicator is built using Forteza and Rama (2001) data and
methodology. The index compiles four dimensions of the labour market: the ratio of minimum
wages to average labour costs in large manufacturing firms, the percentage of salaries that employers
and employees have to contribute to the social security administration, membership of the labour
movement measured as a percentage of the labour force and, employment in the general government
as a fraction of the labour force. A higher value of the indicator corresponds to a less flexible
market. Macroeconomic conditions are proxied by the real exchange rate.
The technological environment is captured possibly, by the contribution of FDI, in

percentage, to capital formation. Equivalently, the latter could represent a measure of production
development in the export sector. As argued and illustrated for instance in UNCTAD (1995,
2002b), FDI can be expected to contribute to the enhancing of a country’s competitiveness on
international markets by increasing the technological content of exports.9 FDI could be directed
towards either higher-value-added activities in newly targeted industries or higher-productivity,
higher-technology and knowledge-based activities within already targeted industries.
To account for possible endogeneity issues similar to those encountered with GDPe both
current and lagged values are used. Estimation results revealed no significant difference and results
are presented with current values.
9
See UNCTAD (2002c) for a discussion of the role of FDI in a sustained development context.
26
Internal transport frictions are introduced via the percentage of paved roads in total road
networks. Transport structure variables are likely to capture internal transport frictions more precisely
than exclusively geographical variables. In order to deal with possible endogeneity issues (the
development of infrastructures could be fostered by higher export growth) again current and lagged
values are used. In the event, results are similar in both cases, and only regressions using current
values are presented.
Taking into consideration the fact that unobserved heterogeneity might play an important
role in determining export performance, but the set of available variables and indicators does not
allow to control for it, quantile regression techniques are used,10 following the seminal work of
Koenker and Bassett (1978). Quantile techniques permit the study to allow for the existence of
unobservable heterogeneity not only through differences in the constant term, as is the case when
introducing dummies, but also through differences in coefficients. The strength of the estimation
relies very much on the fact that these differences are obtained within the same sample and not
across samples. Quantile regression allows the characterization of an entire conditional distribution
rather than only the mean of that distribution as in the case of standard OLS. In that sense,
quantile regression is robust even in the presence of outliers, which is not the case for traditional
conditional mean estimation procedures.
In the present context, the distribution of the real value of countries’ total exports are dealt
with. Quantile regression allows the identification of different responses of exports value to
FMA and supply capacity variables associated with different points on exports value conditional
distribution. Nonetheless, quantile regression coefficients measure the determinants of export
performance for under and over performing countries only in terms of export performance. As
estimation could modify the position of a given country, it may become hazardous to attribute
export performance determinants to over and under performing countries per se. Quantile regression
results represent the basis for policy-oriented experiments aiming at qualifying possible export
performance constraints.
(d) Results and Interpretation
Table 6 reports OLS regressions estimates and Tables 7 to 9 quantile regressions estimates.
Standard errors reported in italics are computed from 1000 bootstrap samples.11 Five conditional
quantiles are considered: 10 th (Q10), 25th (Q25), the median (Q50), 75th (Q75), and 90th (Q90).
The first observation is that some form of error heteroskedasticity seems to exist since the
estimated coefficients vary with the quantiles. This confirms inference from OLS residuals plotting.
Thus, quantile techniques seem to be appropriate for identifying the determinants of export
performance at different levels of the latter. Again, the usefulness of quantile regressions rests first
on the fact that they allow the identificationof possible elements that explain over- and under-
performance with respect to mean. Second, put in a time perspective, it may give useful insights
into the very process of export sector development.
10
A brief technical presentation of the quantile approach and its estimation techniques are presented in appendix C.
Buchinsky (1998) provides a survey and general discussion.
11
Almost identical results both in terms of coefficient values and standard errors values are obtained with 500
bootstrap samples.
27
Table 6
OLS regressions
Dependent Variable : Ln(X) # Observations :84

1988-91 1992-95 1996-99
Ln(GDPt-1) 0.838* 0.115 0.975* 0.085 0.967* 0.104
Ln(POPU) 0.043 0.137 -0.101 0.103 -0.024 0.132
Ln(FMA) 0.622* 0.211 0.546* 0.207 1.010* 0.210
Institutions -0.016 0.058 0.092*** 0.049 0.003 0.057
FDI in capital formation % 4.535* 0.972 3.643* 0.949 2.573* 0.748
Paved roads % 0.696** 0.324 0.737** 0.332 0.533 0.376
ln(Real Exchange rate) 0.055** 0.024 0.033 0.022 0.03 0.024
Constant -12.530* 4.744 -12.751* 5.040 -22.871* 5.038
Regions Dummies Yes Yes Yes

R-squared 0.9398 0.9457 0.9377
Prob>F 0 0 0
Ln(GDPt-1) 0.876* 0.077 0.916* 0.079 0.991* 0.082

Ln(POPU) 0.028 0.103 -0.009 0.086 -0.032 0.097
Ln(FMA) 0.636* 0.207 0.647* 0.209 1.027* 0.208
Lab. Institutions -0.87 0.645 -0.529 0.445 -0.775 0.607
Paved roads % 0.619*** 0.334 0.600*** 0.343 0.476 0.377
Constant -13.215* 4.791 -14.418* 5.075 -23.429* 5.013

R-squared 0.9419 0.9433 0.9372
Prob>F 0 0 0
ln(GDPt-1) 0.857* 0.100 0.982* 0.086 0.990* 0.093

ln(POPU) 0.046 0.127 -0.1 0.103 -0.031 0.122
ln(FMA) 0.646* 0.213 0.560* 0.205 1.028* 0.211
Lab. Institutions -0.889 0.656 -0.379 0.437 -0.776 0.621
Institutions -0.022 0.056 0.087*** 0.049 -0.001 0.056
Paved roads % 0.055** 0.023 0.033 0.022 0.032 0.024
ln(Real Exchange rate) 0.605*** 0.321 0.688** 0.342 0.474 0.372
Constant -13.139* 4.888 -13.090* 5.029 -23.440* 5.099

R-squared 0.9421 0.946 0.9392
Prob>F 0 0 0
28
Table 7
Quantile regressions with institutions and paved roads
Dependent Variable: Ln(X) # Observations :84

10 1988-91 1992-95 1996-99
Ln(GDPt-1) 0.743* 0.175 0.751* 0.203 0.785* 0.190
Ln(POPU) 0.175 0.180 0.205 0.230 0.167 0.195
Ln(FMA) 0.845** 0.366 0.802** 0.392 0.936** 0.470
Institutions 0.103 0.093 -0.054 0.108 -0.016 0.093
FDI in capital formation % 5.361* 1.991 3.525** 1.619 2.894* 1.196
Paved roads % 0.799 0.566 0.645 0.602 1.209*** 0.707
ln(Real Exchange rate) 0.075*** 0.044 0.059 0.039 0.116** 0.049
Constant -17.143** 8.100 -17.987** 8.924 -21.143** 10.215
25
Ln(GDPt-1) 0.798* 0.162 1.015* 0.156 0.846* 0.163
Ln(POPU) 0.093 0.175 -0.076 0.172 0.118 0.181
Ln(FMA) 0.850** 0.358 0.773** 0.402 0.817*** 0.427
Institutions -0.035 0.086 0.048 0.080 -0.034 0.085
FDI in capital formation % 5.135* 1.775 3.337** 1.493 1.744*** 1.016
Paved roads % 0.925*** 0.500 0.156 0.566 0.46 0.579
Constant -17.624** 8.022 -18.898** 9.443 -18.132*** 9.981
50
ln(GDPt-1) 0.821* 0.137 1.074* 0.157 0.832* 0.162
ln(POPU) 0.044 0.161 -0.208 0.177 0.013 0.183
ln(FMA) 0.551 0.389 0.699**** 0.404 0.612 0.445
Institutions 0.002 0.065 0.115 0.085 -0.003 0.092
FDI in capital formation % 4.490* 1.640 3.102*** 1.622 2.547* 1.006
Paved roads % 0.720*** 0.453 0.185 0.536 0.165 0.592
ln(Real Exchange rate) 0.048 0.033 0.037 0.038 0.027 0.043
Constant -10.723 8.925 -16.680*** 9.641 -11.401 10.484
75
ln(GDPt-1) 0.774* 0.140 0.936* 0.147 0.977* 0.149
ln(POPU) 0.054 0.169 -0.162 0.165 -0.18 0.175
ln(FMA) 0.177 0.408 0.305 0.420 0.951** 0.448
Institutions 0.021 0.066 0.109 0.073 0.036 0.081
FDI in capital formation % 4.143** 2.028 3.590** 1.614 1.939*** 1.082
Paved roads % 0.992*** 0.524 0.859*** 0.446 0.482 0.533
Constant -2.114 9.092 -5.444 9.414 -19.059*** 10.442
90
ln(GDPt-1) 0.798* 0.154 0.970* 0.144 0.983* 0.162
ln(POPU) -0.044 0.189 -0.178 0.172 -0.128 0.177
ln(FMA) 0.21 0.365 0.165 0.439 1.138** 0.454
Institutions 0.063 0.073 0.154** 0.070 0.01 0.085
FDI in capital formation % 5.117** 2.164 5.080* 1.561 1.63 1.129
Paved roads % 0.4 0.575 0.964** 0.466 0.238 0.528
ln(Real Exchange rate) 0.042 0.039 0.048 0.039 0 0.043
Constant -1.551 8.174 -3.148 9.636 -23.673* 10.364
.1 Pseudo R2=.7868 .1 Pseudo R2=.7883 .1 Pseudo R2=.7893

29
Table 8
Quantile regressions with Lab institutions and paved roads

10 1988-91 1992-95 1996-99
ln(GDPt-1) 0.913* 0.138 0.784* 0.154 0.887* 0.185
ln(POPU) 0.058 0.144 0.217 0.165 0.133 0.181
ln(FMA) 1.109* 0.344 0.839** 0.382 0.983** 0.437
Lab Institutions -1.690** 0.877 -0.783 0.929 -1.223 0.976
Paved roads % 0.81 0.653 0.746 0.592 1.217*** 0.702
ln(Real Exchange rate) 0.072*** 0.039 0.051 0.039 0.114** 0.049
Constant -24.659* 7.670 -19.500** 8.922 -23.767** 9.812
25
ln(GDPt-1) 0.829* 0.134 0.898* 0.144 0.908* 0.165
ln(POPU) 0.097 0.149 0.101 0.149 0.113 0.173
ln(FMA) 0.786** 0.353 0.815** 0.374 0.777*** 0.432
Lab Institutions -0.674 0.915 -1.048 0.745 -1.138 0.748
FDI in capital formation % 5.376* 1.550 2.818 1.516 2.800* 0.998
Paved roads % 1.042*** 0.569 0.382 0.516 0.718 0.564
Constant -16.967** 7.981 -19.417** 9.066 -18.820*** 9.887
50
ln(GDPt-1) 0.867* 0.120 0.879* 0.147 0.898* 0.148
ln(POPU) -0.013 0.137 0.021 0.145 0.044 0.163
ln(FMA) 0.571 0.381 0.674*** 0.409 0.658 0.428
FDI in capital formation % 4.101* 1.547 2.885*** 1.772 2.553* 0.963
Paved roads % 0.493 0.486 0.511 0.493 0.19 0.602
Constant -10.879 8.784 -14.24 9.855 -14.232 10.069
75
ln(GDPt-1) 0.817* 0.122 0.911* 0.132 0.960* 0.134
ln(POPU) -0.021 0.145 -0.103 0.136 -0.158 0.158
ln(FMA) 0.297 0.382 0.546 0.418 0.993** 0.453
Lab Institutions -0.527 0.685 -0.608 0.761 0.175 0.969
FDI in capital formation % 3.511*** 1.911 4.044* 1.610 2.174*** 1.130
Paved roads % 0.846 0.547 0.465 0.505 0.502 0.531
Constant -3.991 8.429 -10.233 9.662 -19.934*** 10.635
90
ln(GDPt-1) 0.762* 0.150 0.972* 0.142 1.064* 0.146
ln(POPU) 0.01 0.171 -0.146 0.157 -0.176 0.158
ln(FMA) 0.104 0.356 0.411 0.434 1.099** 0.467
Lab Institutions -0.086 0.846 -1.500*** 0.839 0.1 0.955
FDI in capital formation % 5.672* 1.967 3.136*** 1.638 1.424 1.162
Paved roads % 0.375 0.595 0.054 0.581 0.104 0.536
ln(Real Exchange rate) 0.055 0.038 0.022 0.042 -0.023 0.043
Constant 1.084 7.812 -7.246 9.925 -23.885 10.813

30
Table 9
Quantile regressions with institutions, Lab institutions and paved roads

10 1988-91 1992-95 1996-99
ln(GDPt-1) 0.754* 0.155 0.755* 0.193 0.869* 0.201
ln(POPU) 0.228 0.163 0.27 0.226 0.154 0.198
ln(FMA) 0.985* 0.341 0.848** 0.402 0.983*** 0.526
Lab Institutions -1.709** 0.884 -0.886 0.936 -1.306 0.897
Institutions 0.136*** 0.082 -0.033 0.106 -0.024 0.090
FDI in capital formation % 4.130** 1.909 3.154*** 1.674 3.269* 1.298
ln(Real Exchange rate) 0.043 0.044 0.045 0.040 0.129* 0.050
Paved roads % 0.886 0.634 0.673 0.604 1.214*** 0.710
Constant -20.274* 7.723 -19.638** 9.094 -23.721** 11.881
25
ln(GDPt-1) 0.831* 0.155 0.904* 0.161 0.930* 0.160
ln(POPU) 0.097 0.174 0.108 0.178 0.081 0.166
ln(FMA) 0.807** 0.359 0.862** 0.372 0.762*** 0.450
Institutions -0.007 0.076 -0.013 0.085 -0.035 0.078
FDI in capital formation % 5.359* 1.708 2.857 1.541 2.672* 1.036
ln(Real Exchange rate) 0.086** 0.038 0.029 0.038 0.066*** 0.041
Paved roads % 1.004*** 0.571 0.391 0.561 0.603 0.657
Constant -17.369** 8.147 -20.704** 8.988 -18.410*** 10.405
50
ln(GDPt-1) 0.865* 0.138 1.014* 0.155 0.870* 0.152
ln(POPU) -0.013 0.166 -0.141 0.174 0.078 0.186
ln(FMA) 0.573 0.408 0.6 0.395 0.66 0.440
Institutions 0.001 0.068 0.097 0.082 -0.036 0.090
Paved roads % 0.483 0.521 0.321 0.540 0.295 0.625
Constant -10.901 9.323 -13.856 9.302 -14.098 10.286
75
ln(GDPt-1) 0.825* 0.138 0.975* 0.140 0.977* 0.156
ln(POPU) -0.023 0.165 -0.192 0.161 -0.181 0.192
ln(FMA) 0.241 0.403 0.436 0.421 0.980** 0.433
Institutions 0.016 0.068 0.114 0.078 0.033 0.085
FDI in capital formation % 4.094** 2.095 3.533** 1.505 1.973*** 1.106
Paved roads % 0.879*** 0.538 0.771 0.499 0.511 0.524
Constant -3.133 8.893 -8.593 9.325 -19.720** 10.110
90
ln(GDPt-1) 0.784* 0.152 1.020* 0.136 0.996* 0.161
ln(POPU) -0.021 0.186 -0.224 0.165 -0.149 0.186
ln(FMA) 0.325 0.362 0.207 0.424 0.774*** 0.440
Institutions 0.073 0.075 0.122*** 0.073 0.042 0.09
FDI in capital formation % 4.768** 2.144 4.508* 1.339 1.697 1.147
Paved roads % 0.049 0.04 0.041 0.039 0.001 0.044
Constant -3.999 7.98 -3.89 9.162 -16.088 10.038

31
(i) General Considerations
OLS estimates can be used as a benchmark. All coefficients are signed accordingly with
theoretical anticipations. All variables but labour institutions and ln(Population) appear to be
significant in some period. A positive relationship is always obtained between larger domestic
output, better foreign market access and export performance. Transport infrastructures, measured
by the log of percentage of paved roads in the total road network, are likely to play a positive role
in determining export performance. A positive impact is also expected from FDI’s contribution to
capital formation. The log of the real exchange rate is found to have a positive impact in various
estimations. In other words, devaluation could be expected to stimulate exports. Better institutions
are also found to affect positively export performance. For the 1996-99 period only log of lagged
GDP values, foreign market access and FDI contribution to capital formation are significant,
although at the one per cent level.
In all estimation exercises using quantile techniques, the sign of coefficients when significant
are in line with theoretical expectations and OLS estimates. Namely, higher GDP, better access to
foreign markets, better institutions, a more competitive macroeconomic framework, less internal
transport frictions lead to higher export performance. Larger countries, with a larger population,
are found to be less open. It was also found that FDI contribution to capital formation has a
positive impact on export performance. This could first pinpoint the existence of technological
transfer from foreign companies and underline its positive impact on export performance.
(ii) GDP and Population
In all regressions, a recurrent result is the high significance of log GDP (lagged). This
remains true across periods and across quantiles. The coefficient on the log of GDP is found to be
less than one in a majority of cases. This reinforces the result that larger countries are less open.
That is, larger countries will tend to export relatively less, although large countries are also those
whose exports volumes tend to be larger. On the other hand, total population never proves to have
any significant effect on export performance. The quality of regressions does not vary dramatically
with the institutional indicator used. In addition, general results remain qualitatively identical.
However, coefficients on ln(GDP), ln(POPU) can vary importantly and generally not in a
monotonic fashion. In general, for a given institutional indicator, qualitative results remain identical.
(iii) Internal Transport Frictions
The importance and significance of internal transport frictions varies from period to period
and from quantile to quantile. It appears that internal transport costs had a significant negative
impact on export performance over the 1988-91 period among the weakest performers, that is, for
those in the first fourth of the distribution. Internal transport facilitation played an important role
across all quantiles in the 1992-95 time frame. Its significance appears to increase substantially for
those exporters that are above the median of the sample. This remains true to a large extent over
the 1996-99 period although the percentage of paved roads is found to be significant at the 10 per
cent level for the first quantile only.
32
In other words, transport infrastructures are likely to play an important role at the early
stage of export sector development.12 The percentage of paved roads appears to be more significant
and its impact to be larger in lower quantiles. As shown in Tables 10 and 11, most African countries,
which are characterized by poor transport infrastructure are found in all periods in the left part of
the distribution of export performance. This is likely to indicate that African countries could do
much to raise their supply capacity by investing in transport infrastructure (Limão and Venables,
2001) present some empirical analysis indicating that levels of trade flows observed for African
countries are relatively low essentially because of poor transport infrastructures). The fact that this
sort of investment has not occurred in a significant manner in the last two decades could explain
the very low upward mobility of African countries in export performance.
(iv) The Macroeconomic Environment
The real exchange rate proves to have a significant effect on export performance in the
lowest quantile. This is verified for all periods and on average a 1 per cent real depreciation could
increase exports by 6 to 10 per cent.
Results indicate that an overvalued currency is detrimental to export performance. An

overvalued currency translates into a direct loss of price competitiveness for exporting firms. This
is of particular importance for commodities and manufactured products that are labour intensive.
Both types of goods are essential components of under export performers explaining the significance
of the coefficient on the log of the real exchange rate at low quantiles. Good export performers, on
the other hand, may rely on more capital-intensive production relative to weak performers. The
former may suffer less than the latter from export price competitiveness, measured by the real
exchange rate, while exporters in more labour-intensive activities may suffer less from high capital
rents.
In other words, good export performers are more likely to have a strong position in more
capital-intensive products markets and may face less aggressive competitors than exporters in more
labour intensive product markets. As a consequence, their competitiveness would not rely too
much on the real exchange rate but on the technological content of their product and thus to a
large extent on capital. This is not likely to be the case for producers exporting low skill products,
which are highly substitutable and whose demand is very volatile and price sensitive. Real interest
ratesare found to affect significantly export performance of good performers with high rates
increasing producer costs and hence impacting negatively on export competitiveness.
(v) Foreign Direct Investment
The quantile regression results indicate that FDI is likely to affect export performance
positively. This is true for most levels of export performance and for every period under
consideration. The experience in a number of countries suggests that FDI strongly contributes to
the transformation of the composition of exports. For instance, it has been well documented that
FDI inflows into Singapore or, more recently China, have helped to increase significantly the
technological content of exports by supporting strongly the development of knowledge-based
industries.
12
A comparable argument is documented in Limão and Venables (2001).
33
Consistent with these experiences, a positive and significant relationship between export
performance and FDI contribution to capital formation is found at all levels of export performance
in this analysis. In all periods but 1988-91, the strongest impact is obtained at the lowest two
quantiles. In the first two periods, the impact of FDI contribution to capital formation is non-
linear. Coefficients indicate a U-shaped relationship between export performance and the FDI
variable.
Assuming that FDI does contribute to the technological upgrading and structural evolution
of the export sector, these results indicate that the structure of the sector is an important ingredient
of export performance both at the early stage of development of the export sector and, at its latest.
A possible qualification of the argument would be to say that export performance is positively
affected by inter-sectoral diversification at the bottom of the distribution and intra-sectoral
diversification at the top of the distribution. “Best”-performers, essentially industrialized countries
and some emerging economies as shown in Tables 10 and 11, are characterized by an already high
degree of inter-sectoral diversification.13 FDI could then be expected to be directed towards
innovative activities within an already existing sector and,in that sense to stimulate essentially
intra-sectoral rather than inter-sectoral diversification. Results for the last period only indicate a
decreasing pattern of the impact of FDI across quantiles. This could simply indicate that good
performers in earlier periods have a maturity turning point in intra-sectoral development.
As a consequence, supply capacity constraints could also be relaxed by improving the

technological content of the export sector as indicated by the positive influence of FDI contribution
to capital formation on export performance.
(vi) Institutions
Both measures considered in the estimations are only rarely strongly significant. It appears
that this is essentially the case in the 1992-95 period and, it is particularly true for the general
measure of the quality of the institutional framework. In that case, the significance increases slightly
with quantiles. In the other periods, the significance of the coefficient on the labour institutions
indicator is higher for low export performers. The reverse is true when the measure of institutional
quality is considered. This remains true when both institutions are taken into consideration for
estimation. In the latter case, it is observed that estimates are not affected in a monotic manner.
The results however, are only slightly affected and the general analysis remains valid.
A significant role for institutional quality could have been expected at an early stage of
export sector development, but in the results this is the case only for the 1988-91 period. This
might be due to the existence of some collinearity between FDI contribution to capital formation,
institutions and the macroeconomic environment. However, simple correlation analysis does not
show any strong relationship among the variables considered for estimation. Higher correlation
coefficients are only found when FDI net flows are considered. This result reinforces the relevance
of interpreting FDI contribution to capital formation as an indicator of technological and innovative
transfer.
13
See UNCTAD (2002) Table 8.2A.
34
Table 10
Countries exports real value position in sample distribution
region Quant8083 Quant8487 Quant8891 Quant9295 Quant9699

France we 6 6 6 6 6
United kingdom we 6 6 6 6 6
Norway we 6 6 6 6 6
Germany we 6 6 6 6 6
Netherlands we 6 6 6 6 6
Switzerland we 5 5 5 5 5
Denmark we 5 5 5 4 4
Benelux we 5 5 6 6 5
Spain we 5 5 5 5 5
Sweden we 5 5 5 4 4
Portugal we 5 5 5 5 5
Austria we 4 5 5 5 5
Italy we 4 4 4 4 5
Finland we 4 4 4 4 4
Greece we 4 4 4 4 4
Republic of Ireland we 3 4 4 3 4
Cyprus we 1 1 2 1 1
Nigeria ssa 5 4 4 4 4
South Africa ssa 4 4 4 4 4
Cote d'Ivoire ssa 3 3 3 3 3
Kenya ssa 3 3 2 3 3
Zambia ssa 3 3 3 3 2
Gana ssa 2 2 3 2 2
Mauritania ssa 2 2 2 2 2
Niger ssa 2 1 2 1 1
Senegal ssa 2 2 2 2 1
Tanzania ssa 2 2 2 2 2
Uganda ssa 2 2 1 1 2
Madagascar ssa 2 2 1 1 1
Mauritius ssa 2 2 3 2 3
Burkina Faso ssa 1 1 1 1 1
Mali ssa 1 1 1 1 1
Rwanda ssa 1 1 1 1 1
India sa 4 4 4 4 4
Pakistan sa 3 3 3 3 3
Sri Lanka sa 3 3 3 3 3
Bangladesh sa 2 3 3 3 3
Nepal sa 1 1 1 1 1
Canada na 6 6 6 6 6
United States na 6 6 6 6 6
Mexico na 5 5 4 5 5
Algeria mena 4 4 4 4 4
Israel mena 4 4 4 4 4
Iran, Islamic rep. mena 4 4 4 4 3
Kuwait mena 4 4 1 3 2
Tunisia mena 3 3 3 3 3
Egypt, Arab rep. mena 3 3 3 3 3
Morocco mena 3 3 3 3 3
Syrian Arab republic mena 3 3 3 3 3
Jordan mena 1 2 2 2 2
/...
35
Table 10 (concluded)
Countries exports real value position in sample distribution
region Quant8083 Quant8487 Quant8891 Quant9295 Quant9699

Brazil lac 5 5 5 5 5
Venezuela lac 4 4 4 4 4
Argentina lac 4 4 4 4 4
Chile lac 4 3 4 4 4
Costa Rica lac 3 3 3 3 3
Guatemala lac 3 3 3 3 3
Jamaica lac 3 2 2 2 3
Trinidad and Tobago lac 3 3 3 3 3
Colombia lac 3 3 3 4 4
Ecuador lac 3 3 3 3 3
Uruguay lac 3 3 3 3 3
Bolivia lac 3 2 2 2 2
Peru lac 3 3 3 3 3
Dominican Republic lac 2 3 3 3 3
Honduras lac 2 2 2 2 2
Nicaragua lac 2 1 1 2 2
El Salvador lac 2 2 2 2 2
Panama lac 1 1 2 2 2
Paraguay lac 1 2 2 2 2
Hungary eca 4 3 4 4 4
Poland eca 4 4 4 4 4
Romania eca 4 4 3 3 3
Turkey eca 3 4 4 4 4
Bulgaria eca 3 3 3 3 3
Japan eap 6 6 6 6 6
Republic of Korea eap 5 5 5 5 5
Singapore eap 5 5 5 5 5
Indonesia eap 5 5 4 4 4
Australia eap 5 4 5 5 4
Malaysia eap 4 4 5 5 5
China eap 4 5 5 5 6
Thailand eap 4 4 4 5 5
Philippines eap 4 4 4 4 4
New Zealand eap 4 4 4 4 4
36
Table 11
Countries position in sample exports real value distribution
1980-83 1984-87 1988-91 1992-95 1996-99

France France France France France
United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom
Norway Norway Norway Norway Norway
Germany Germany Germany Germany Germany
Q6
Netherlands Netherlands Netherlands Netherlands Netherlands

Canada Canada Canada Canada Canada
United States United States United States United States United States
Japan Japan Japan Japan Japan
Switzerland Switzerland Switzerland Switzerland Indonesia
Denmark Denmark Denmark Denmark Switzerland
Benelux Benelux Benelux Spain Denmark
Spain Spain Spain Portugal Spain
Sweden Sweden Sweden Brazil Portugal
Portugal Portugal Portugal Rep. of Korea Brazil
Q5
Nigeria Mexico Brazil Singapore Rep. of Korea

Mexico Brazil Rep. of Korea Austria Singapore
Brazil Rep. of Korea Singapore Indonesia Austria
Rep. of Korea Singapore Austria Malaysia China
Singapore Australia Indonesia China Mexico
Australia Austria Malaysia Mexico Thailand
Malaysia Indonesia China Thailand Italy
Austria Nigeria Mexico Benelux Malaysia
Italy Malaysia Australia Sweden Benelux
Finland Italy Nigeria Australia Sweden
Greece Finland Italy Nigeria Australia
South Africa Greece Finland Italy Nigeria
India South Africa Greece Finland Finland
Algeria India South Africa Greece Greece
Israel Algeria India South Africa South Africa
Iran, Islamic rep. Israel Algeria India India
Kuwait Iran, Islamic rep. Israel Algeria Algeria
Q4
Venezuela Kuwait Iran, Islamic rep. Israel Israel

Argentina Venezuela Venezuela Iran, Islamic rep. Venezuela
Chile Argentina Argentina Venezuela Argentina
Hungary Hungary Hungary Argentina Hungary
Romania Romania Thailand Hungary Philippines
Turkey China Philippines Philippines New Zealand
Indonesia Thailand New Zealand New Zealand Poland
China Philippines Ireland Poland Chile
Thailand New Zealand Poland Chile Turkey
Philippines Rep. of Ireland Chile Turkey Trinidad and Tobago
New Zealand Poland Turkey Trinidad and Tobago Ireland
/...
37
Table 11 (concluded)
Countries position in sample exports real value distribution
1980-83 1984-87 1988-91 1992-95 1996-99

Rep. of Ireland Chile Romania Rep. of Ireland Iran, Islamic rep.
Cote d Ivoire Turkey Cote d Ivoire Romania Romania
Kenya Cote d Ivoire Kenya Cote d Ivoire Cote d Ivoire
Zambia Kenya Pakistan Kenya Pakistan
Pakistan Zambia Sri Lanka Pakistan Sri Lanka
Sri Lanka Pakistan Tunisia Sri Lanka Tunisia
Tunisia Sri Lanka Egypt, Arab rep. Tunisia Egypt, Arab rep.
Egypt, Arab rep. Tunisia Morocco Egypt, Arab rep. Morocco
Morocco Egypt, Arab rep. Syrian Arab republic Morocco Syrian Arab republic
Syrian Arab republic Morocco Costa Rica Syrian Arab republic Costa Rica
Q3
Costa Rica Syrian Arab republic Guatemala Costa Rica Guatemala

Guatemala Costa Rica Jamaica Guatemala Jamaica
Jamaica Guatemala Trinidad and Tobago Jamaica Colombia
Trinidad and Tobago Jamaica Colombia Colombia Ecuador
Colombia Trinidad and Tobago Ecuador Ecuador Uruguay
Ecuador Colombia Uruguay Uruguay Bulgaria
Uruguay Ecuador Bulgaria Bulgaria Bangladesh
Peru Uruguay Bangladesh Bangladesh Bolivia
Domican Republic Bulgaria Bolivia Bolivia Zambia
Poland Bangladesh Ghana Zambia Mauritania
Bulgaria Bolivia Mauritania Kuwait Peru
Ghana Peru Zambia Ghana Kenya
Mauritania Domican Republic Peru Mauritania Kuwait
Niger Ghana Domican Republic Peru Ghana
Senegal Mauritania Niger Domican Republic Domican Republic
Tanzania Niger Senegal Niger Niger
Uganda Senegal Tanzania Senegal Tanzania
Q2
Madagascar Tanzania Honduras Tanzania Honduras

Mauritius Uganda Nicaragua Honduras Nicaragua
Bangladesh Madagascar Jordan Nicaragua Jordan
Bolivia Honduras El Salvador Jordan El Salvador
Honduras Nicaragua Mauritius El Salvador Panama
Nicaragua Jordan Cyprus Panama Paraguay
Paraguay El Salvador Panama Paraguay Uganda
Cyprus Mauritius Kuwait Mauritius Senegal
Burkina Faso Paraguay Uganda Cyprus Mauritius
Mali Cyprus Madagascar Uganda Cyprus
Q1
Rwanda Burkina Faso Paraguay Madagascar Madagascar

Nepal Mali Burkina Faso Burkina Faso Burkina Faso
Jordan Rwanda Mali Mali Mali
El Salvador Nepal Rwanda Rwanda Rwanda
Panama Panama Nepal Nepal Nepal
38
However, institutions seem to matter more at a higher level of export performance. This
result suggests that what appears to be essential in the overall growth process as suggested by
recent research14 is only partially true for export performance. It might also suggest that institutions
and macroeconomic variables are substitutable along the export development process. While the
real exchange rate is an essential price competitiveness component for low-performers, once
macroeconomic stability has been achieved and the composition of exports is more oriented towards
capital -intensive goods, as is most likely the case for high-performers, the institutional framework
comes in as an essential competitiveness ingredient. Better institutions are expected to guarantee
better protection of property rights, which becomes essential as production becomes more and
more capital intensive. Better institutions are also likely to be associated with more efficient
administration and in particular regulation, which could prove to be important price components
in industrialized countries.
(vii) Foreign Market Access
Results suggest that foreign market access remains essential even after controlling for supply
capacity. They do not show a time recurrent pattern. However, coefficients on foreign market
access reveal that it is always a significant explanatory component for under performers in all
periods as it was the case for most of the supply capacity elements analyzed above. The 1988-91
period is characterized by a non-monotonic influence of FMA across quantiles. The coefficient
first increases then falls in the Q75 regression to increase for the last quantile. The coefficient is
significant at the 1 per cent level for all quantiles. For the 1992-95 period the impact and the
significance of the estimates decrease over the first three quantiles. Coefficients become insignificant
in the upper two quantiles regressions. As export performance improves, it becomes less and less
responsive to foreign market access. In the 1996-99 period the tendency appears to be reversed.
The level of the reactivity of exports to foreign market access increases with export performance.
Coefficients remain significant at the 1 per cent level for all quantiles. In some regressions increasing
returns to scale appear. The coefficient on the FMA (log) variable becomes larger than one in the
Q90 quantile. From a theoretical point of view this would correspond to a small elasticity of prices
in the export sector with respect to the quantity of resources used in the sector. In other words, at
some points of the external sector performance there may be some increasing returns in the domestic
supply response to better foreign market access.15
Foreign market access is explained by various elements that include essentially geography,
trade policy and partners’ characteristics, such as the size of their market and transport facilities. It
could be the case that a more integrated world implies a more uniform set of partners’ characteristics
across countries. As a consequence, changes in the magnitude of the estimated coefficients could
be associated with changes in policy. However, this would be the case only if one could control for
the nature and composition of trade. In this empirical context, partner characteristics can not be
fully separated from policy components, so that a possible interpretation of the increased significance
of foreign market access in the upper part of the distribution across periods can be based on the
evolution of the external sector structure. An associated result of that just presented is the decrease
in the magnitude of the coefficient on the FDI variable across periods. The two observations put
together suggest that linkages to international markets have become more important than the
structural development of the external sector assumed to be captured by FDI contribution to
14
See Rodrik et al. (2002) for empirical assessment and for a critical review of empirical work.
15
This is a form of the home market effect as presented in Krugman (1980).
39
capital formation. This could be due to the fact that the external sector in high performing countries
has completed its structural change, that is, its intra-sectoral development and diversification.
This is plausible only if the FDI variable does proxy such a process in the upper part of the
distribution. In that context, access to international markets becomes relatively more significant in
determining export performance. This interpretation supports the argument that better access to
international markets can contribute towards the expansion of the external sector at all stages of its
structural development.
(e) Sensitivity Analysis
Only the most relevant results from a statistical point of view have been presented. However,
a large set of variables and their combinations have been considered in estimations.
Various variables related to internal transport costs were introduced. To account for transport
costs related to internal geography the study used in turn the equivalent of the average distance
between two points in a circular country given by 0 . 33 (area / π ) and, the percentage of population
0 .5
living within 100 km of the coast and rivers in 1996. The former variable is never significant. The
latter appears to be significant and positive at high quantiles for the 1992-95 and 1996-99 periods.
A larger proportion of people living close to rivers imply lower transport costs. Nevertheless, the
measure seems not to be fully appropriate for the 1989-91 period regressions. Moreover, it could
place too much weight on water transport, which is likely to be questionable in a number of
countries in the sample. The extent of the internal road network simply given by the total number
of kilometres of roads whether paved or not, was also used. Essentially this is due to physical
structural transport costs. Likewise,the number of telephone lines per ‘000 people and the number
of telephone lines per employee was also considered. None of these variables turn out to be
significant, even though the coefficients appear to be correctly signed. The study also considered
various combinations of measures of geography-related and physical-infrastructure-related transport
costs. However, none appear to be relevant and, collinearity may be present as there are very high
cross-correlation rates.
The study controlled for the macroeconomic environment using a large set of variables.
First it introduced the real interest rate, which appeared to be significant for high quantiles in the
last two periods. Nevertheless, the data were available only for 74 countries, which might mitigate
the robustness of the results. The spread between the borrowing and lending rates was also
considered. This variable could be seen as an indicator of the functioning of the capital market. It
remained insignificant in all regressions and was only available for 73 countries. The real effective
exchange rate computed by the IMF was also considered. However, this was only available for a
few countries, and the coefficients never appeared to be significant. The same is true for the index
of overvaluation of the currency computed by the World Bank following Dollar (1992).
Export concentration and diversification indices of exported products constructed by the

UNCTAD are introduced to account for the structure of the export sector. None of these indices
appears to be significant.
All these variables were considered in various sets of explanatory variables. None generated
any result that contained a message strikingly different from those presented previously, nor were
any more relevant statistically.
40
IV. POLICY IMPLICATIONS
Chapter II sheds some light on the relative importance of export performance components
including to which extent that export performance has been constrained by poor supply capacity
conditions. The results outlined in chapter III, helped in the understanding of the importance of
certain elements affecting supply capacity elements in the process of export development.
The East Asian countries’ experience indicates that strong export performance is likely to
support strong economic growth. It is thus essential to identify what determines export performance.
The major components of the latter are access to international markets and supply capacity. This
study investigates the relative importance of these two components. It attempts to define whether
export performance in various regions has been constrained by either of them. Supply capacity
appears to have been a limiting element of export performance in African, Middle Eastern and
Latin American countries. By contrast, supply capacity has been the driving force behind the
export performance of South Asia. Empirical analysis indicates that major determinants of export
performance are also likely to vary with the level of performance itself. Poor transport infrastructures
and a weak macroeconomic environment are expected to explain poor performance (e.g. African
countries). The structure of the external sector is also expected to matter at all levels of export
performance. As might be expected, foreign market access is highly significant. This has been a
particularly significant factor in explaining poor export performance for a number of countries
whose exports are badly affected by trade barriers. However, it was also found that market access
also affected high performing countries in the last period under investigation. Nevertheless, they
were able to overcome this constraint. This result may suggest that the structure of the external
sector and foreign market access are complements at low levels of performance and tend to become
substitutes at higher levels. That is to say that as the external sector structure develops, access to
international markets becomes more significant in explaining export performance.
The above analysis suggests that policy intervention should have static and dynamic effects:
• Transport infrastructures appear to be an important determinant of exports for low-

performers. Since it takes a long time to recoup investment in infrastructure and the
yields tend to be low, this reinforces the argument in favour of promoting public
investment, either national or international, and international technical cooperation
in that domain.
• Macroeconomic stability is also an important determinant of export performance at

the early stages of development of the external sector.
• Inter-sectoral diversification should be promoted at the early stage of development of

the external sector. This could be done via the promotion of foreign direct investment,
as suggested by the empirical results. This could also support arguments in favour of
more neutral sectoral policies. However, lowering policy trade barriers on intermediate
inputs could also be useful as a step toward enhancing export performance. Thus
there is scope for pushing forward the liberalization of intermediate inputs more rapidly
than final goods.
41
• In a more dynamic context, diversification should also be promoted within sectors.
This could operate through the promotion of technological improvement.
Technological improvement cannot be efficiently obtained without the presence of
adequate human capital. This implies that public investment should also be devoted
to raising the level of human capital (e.g. through education) and the “technological
competency” of the labour force (e.g. through training). However,technological
improvement remains the output of technological innovation which could either be
imported via FDI or nationally generated via R&D. Then, in the process of the external
sector development, inter-sectoral diversification should be associated with the
accumulation of competencies that will be able to lead to intra-sectoral diversification,
which appears at latter stages of development.
• Foreign market access proves to be significant at all levels of export performance.

This indicates that there is scope for further lowering trade barriers at all stages of
development of the external sector. Beside facilitating access to international markets
it would also facilitate trade in intermediate outputs. Moreover, increasing returns in
terms of export performance could be obtained.
• Finally, the empirical results suggest that it is necessary for policy makers to care
about all dimensions of the process of development of the external sector. For instance,
fighting for better access to international markets while neglecting supply conditions
is likely to be unproductive in terms of export performance, as suggested by the African
and to some extent the Latin American experiences.
42
V. POSSIBLE FURTHER RESEARCH
While the study gives some useful guidance to policy makers, a number of points could be
clarified by further work, and this would help to give greater specificity to policy guidelines.
The analysis has been carried out at an aggregate level, and in future work it may be useful
to carry forward the analysis to account for sectoral specificity. 16 This should allow better
qualification of supply capacity constraints and the identification of those sectors that face them
the most severely. Analysis at the country level could then be run sector by sector as foreign market
access series would be obtained for each of them. This could further refine the framework for
policy-oriented exercises.
It might also be useful to carry forward the analysis using a structural approach in
determining export performance components. This approach would allow the decomposition of
foreign market access and supply capacity into well-identified elements. For instance, it would
become possible to quantify the stringency of trade barriers and their impact on foreign market
access. Likewise, it might also be able to better assess the impact of special trade agreements and
in particular of RTAs. This could be done by simply considering trade agreements partners grouping
on top of, or instead of geographical grouping. It could then be assessed how much such agreements
contributed to a particular country’s or region’s export performance evolution. Typically, a pre-
agreement period of time and a comparison of the underlying results with those obtained in a
post-agreement period of time, could be considered. Hence, the impact on supply capacity could
be interpreted as some sort of adjustment cost or, could serve to identify those elements that
possibly contain the latter.
The analysis in this study did not explicitly take account of time. It is possible to compare
results across time as long as data are available. However, it should also be possible to integrate
time into the cross-sectional analysis itself without making it a panel analysis. This approach was
followed by Dollar and Kraay (2002) who considered decadal changes in their dependent and
independent variables, that is the relationship between changes over time. The relevance of such
an approach is supported by the empirical fact that trade to GDP ratios are not strongly correlated
with themselves across decades. This could lead to a better understanding of the decomposition
components influence. Practically, the total exports real value would be put as the dependent
variables, its lag T years ago (T=10 for instance), and the averages over the T years period before
the current date of a set of control variables.
16
Data availability in COMTRADE permits the consideration of exports at various sectorial levels rather than
simply at the aggregate level. Elbadawi, Mengistae and Zeufack (2001) adopt a similar approach using even more
disaggregated data. The data used are from more than 1400 manufacturers of textile and garments in six Sub-Sahara
African countries, Morocco and India. They conclude that geography is as probably as important for African
manufacturers as are trade policy and institutional infrastructure.
43
REFERENCES
Alcalà F. and A. Ciccone, 2002, “Trade and Productivity”, CEPR Discussion Paper #3095.
Buchinski M., 1998, “Recent Advances in Quantile Regression Models: A Practical Guideline for
Empirical Research”, Journal of Human Resources, 33:1, pp. 88-126.
Dollar D., 1992, “Outward-Oriented Developing Countries Really Do Grow More Rapidly:
Evidence from 95 LDCs, 1976-85”, Economic Development and Cultural Change, April,
pp. 523-544.
Dollar D. and A. Kraay, 2002, “Institutions Trade and Growth”, forthcoming Canergie-Rochester
Conference Series on Public Policy.
Easterly W. and R. Levine, 2002, “Tropics, Germs, and Crops: How Endowments Influence
Economic Development”, NBER Working paper No. 9106.
Elbadawi I., T. Mengistae, and A. Zeufack, 2001, “Geography, Supplier Access, Foreign Market
Potential, and Manufacturing Exports in Developing countries: An Analysis of Firm Level
Data”, The World Bank, Unpublished Manuscript.
Forteza and Rama, 2001, “Labor Market “Rigidity” and the Success of Economic Reforms Across
more the One Hundred Countries”, The World Bank Working Paper #2521.
Frankel J. A. and D. Romer, 1999, “Does Trade Cause Growth? ”, American Economic Review,
89, 3, pp. 379-399.
Gallup J., J. Sachs, and A. Mellinger, 1998, “Geography and Economic Development”, Proceedings
of World bank Annual Conference on Development Economics, The World Bank:
Washington.
Hall R., and C. Jones, 1998, “Fundamental Determinants of Output per Workers across Countries”,
Quarterly Journal of Economics,114:1, pp.83-116.
Harrison A., 1996, “Openness and Growth: A Time-Series, Cross-Country Analysis for Developing
Countries”, Journal of Development Economics, 48, pp. 419-77.
Irwin D.A., and M. Tervio, 2002, “Does Trade Raise Income? Evidence from the Twentieth
Century”, Journal of International Economics, 58, pp. 1-18.
Koenker R. W. and, G. Bassett, 1978, “egression Quantiles”, Econometrica, 46:1, pp. 33-50.
Krugman, P., 1980, “Scale Economies, Product Differentiation and the Pattern of Trade ”, American
Economic Review, 70, pp. 950-959.
44
Krugman, P. and A. Venables, 1995, “Globalization and the Inequality of Nations”, Quarterly
Journal of Economics,110:4, pp. 857-880
Limão N. and T. Venables, 2001, “Infrastrucutre, Geographical Disadvantage, Transport Costs,

and Trade”, The World Bank Economic Review, 15:3, pp. 451-479.
Maksus, K. E., 2003, Comment on Redding and Venables (2003), in Challenges to Globalization:
Analyzing the Economics, R. E. Baldwin and A. L Winters, eds., The University of Chicago
Press.
Redding S. and A. Venables, 2004a, “Geography and Export Performance: External Market Access
and Internal Supply Capacity”, Chapter 3, in (eds) Baldwin, R and Winters, A, Challenges to
Globalization: Analyzing the Economics, CEPR-NBER conference volume, NBER and Chi-
cago University Press.
Redding S. and A. Venables, 2004b, “Economic Geography and International Inequality” (joint with
Anthony Venables), Journal of International Economics, 62(1), pp. 53-82.
Rodriguez F. and D. Rodrik, 2000, ” Trade Policy and Economic Growth: A Skeptic’s Guide to
Cross-Sectional Evidence”, Macroeconomics Annual 2000, B. Bernanke and K. Rogoff, eds.,
MIT Press for NBER.
Rodrik D., A. Subramanian and F. Trebbi, 2002, “Institutions Rule: the Primacy of Institutions
over Geography and Integration in Economic Development”, NBER Working Paper No.
9305.
Sachs J. and A. M. Warner, 1995, “Economic Convergence and Economic Policies”, NBER Working
Paper No. 5039.
UNCTAD, 1995, World Investment Report, United Nations, Geneva.
UNCTAD, 2002a, Handbook of Statistics, United Nations, Geneva.
UNCTAD, 2002b, World Investment Report, United Nations, Geneva.
UNCTAD, 2002c, Trade and Development Report, United Nations, Geneva.
The World Bank, (1993), “The East Asian Miracle: Economic Growth and Public Policy”, Oxford
University Press for The World Bank, New York.
Yanikkaya H., 2003, “Trade Openness and Economic Growth: A Cross-Copuntry Empirical
Investigation”, Journal of Development Economics, 72, pp. 57-89.
45
APPENDIX
A. Data Sources
Bilateral Trade Flows: United Nations COMTRADE database
GDP, Population, Geographical variables (data on bilateral distance, existence of a common border):
World Bank World Development Indicators
United States GDP deflator: IFS from the IMF.
Internal Geography: Gallup, Sachs and Mellinger (1998) / (www2.cid.harvard.edu/ciddata)
Miscellaneous: CIA World Fact Book
Infrastructures: World Bank WDI 2003
Trade Openness: Sachs and Warner (1995) / (www2.cid.harvard.edu/ciddata)
Labor Market Indicators: Forteza and Rama (2001)
Effective real Exchange rate Overvaluation Index: World Bank World Development Indicators (Dollar
(1992) methodology)
Exchange and Interest rates: World Bank WDI 2003
FDI contribution to capital formation: UNCTAD handbook of statistics
Institutions: Expropriation risk form International Country Risk Guide database / Hall and Jones
(1998) / www.standford.edu/~chadj/datasets.html
B. Geographical Country Groups
Western Europe: Austria, Belgium-Luxembourg, Switzerland, Cyprus, Germany, Denmark, Spain,

Finland, France, United Kingdom, Republic of Ireland, Italy, Netherlands, Norway, Sweden, Greece,
Portugal
Sub-Sahara: Burkina-Faso, Cote d’Ivoire, Ghana, Kenya, Madagascar, Mali, Mauritania, Mauritius,
Niger, Nigeria, Rwanda, Senegal, Tanzania, Uganda, South Africa, Zambia, Zimbabwe
South Asia: Bangladesh, India, Sri Lanka, Nepal, Pakistan
46
Middle East and North Africa: Algeria, Arab Republic of Egypt, Islamic Republic of Iran, Israel, Jordan,
Kuwait, Morocco, Syrian Arab Republic, Tunisia
Latin America and Caribbean: Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Domican
Republic, Ecuador, Guatemala, Honduras, Jamaica, Nicaragua, Panama, Peru, Paraguay, el Salvador,
Trinidad and Tobago, Uruguay, Venezuela
Eastern Europe and Central Asia: Bulgaria, Hungary, Poland, Romania, Turkey
East Asia and the Pacific: Australia, China, Hong Kong China, Indonesia, Japan, Republic of Korea,
Malaysia, New Zealand, Philippines, Singapore, Thailand, Taiwan (Province of China)
C. Quantile Regression
Quantile regression, developed by Koenker and Bassett (1978), is an extension of the classical least
squares estimation of the conditional mean to estimation of different conditional quantile functions.
The conditional mean function is estimated by minimizing the symmetrically weighted sum of absolute
errors, where the weight is equal to 0.5. Other quantile functions are estimated by minimizing an
asymmetrically weighted sum of absolute errors, where the weights are functions of the quantile of
interest. As a consequence, quantile regression is robust to the presence of outliers.
The general quantile regression model is
y i = xi' β θ + uθi
or alternatively
xi' βθ
θ= ∫ f y (s | xi )ds
−∞
where β θ is an unknown k x 1 vector of regression parameters associated with the θtth percentile, xi is a
k x 1 vector of independent variables, yi is the dependent variable of interest and uθι is an unknown error
term. The θtth conditional quantile of y given x is Quantθ (uθi | xi ) = xi' βθ . Its estimate is given by xi' β̂θ .
As θ increases continuously, the conditional distribution of y given x is traced out. Although many of
the empirical quantile regression papers assume that the errors are independently and identically
distributed, the only necessary assumption concerning errors is
Quantθ (uθi | xi ) = 0
47
that is, the conditional θtth quantile of the error term is equal to zero. Thus the quantile regression
approach allows the marginal effects to change for countires at different points in the conditional
distribution by estimating βθ using several values of θ, θ ∈ (0 ,1) . It is in this way that quantile regression
allows for parameter heterogeneity across different levels of export performance.
The quantile regression estimator is the solution to the following minimization problem:
 
βˆθ = arg min  ∑ θ y i − xi' β + ∑ (1 − θ ) y i − xi' β 
β
 
 i: yi > xi β i: yi < xi' β 
'
48
UNCTAD Study Series on
POLICY ISSUES IN INTERNATIONAL TRADE

AND COMMODITIES
No. 1 Erich Supper, Is there effectively a level playing field for developing country
exports?, 2001, 138 p. Sales No. E.00.II.D.22.
No. 2 Arvind Panagariya, E-commerce, WTO and developing countries, 2000, 24 p.

Sales No. E.00.II.D.23.
No. 3 Joseph Francois, Assessing the results of general equilibrium studies of multi-
lateral trade negotiations, 2000, 26 p. Sales No. E.00.II.D.24.
No. 4 John Whalley, What can the developing countries infer from the Uruguay
Round models for future negotiations?, 2000, 29 p. Sales No. E.00.II.D.25.
No. 5 Susan Teltscher, Tariffs, taxes and electronic commerce: Revenue implications
for developing countries, 2000, 57 p. Sales No. E.00.II.D.36.
No. 6 Bijit Bora, Peter J. Lloyd, Mari Pangestu, Industrial policy and the WTO, 2000,
47 p. Sales No. E.00.II.D.26.
No. 7 Emilio J. Medina-Smith, Is the export-led growth hypothesis valid for develop-
ing countries? A case study of Costa Rica, 2001, 49 p. Sales No. E.01.II.D.8.
No. 8 Christopher Findlay, Service sector reform and development strategies: Issues
and research priorities, 2001, 24 p. Sales No. E.01.II.D.7.
No. 9 Inge Nora Neufeld, Anti-dumping and countervailing procedures – Use or

abuse? Implications for developing countries, 2001, 33 p. Sales No. E.01.II.D.6.
No. 10 Robert Scollay, Regional trade agreements and developing countries: The case
of the Pacific Islands’ proposed free trade agreement, 2001, 45 p. Sales No.
E.01.II.D.16.
No. 11 Robert Scollay and John Gilbert, An integrated approach to agricultural trade
and development issues: Exploring the welfare and distribution issues, 2001,
43 p. Sales No. E.01.II.D.15.
No. 12 Marc Bacchetta and Bijit Bora, Post-Uruguay round market access barriers
for industrial products, 2001, 50 p. Sales No. E.01.II.D.23.
No. 13 Bijit Bora and Inge Nora Neufeld, Tariffs and the East Asian financial crisis,
2001, 30 p. Sales No. E.01.II.D.27.
49
No. 14 Bijit Bora, Lucian Cernat, Alessandro Turrini, Duty and Quota-Free Access for
LDCs: Further Evidence from CGE Modelling, 2002, 130 p. Sales No.
E.01.II.D.22.
No. 15 Bijit Bora, John Gilbert, Robert Scollay, Assessing regional trading arrange-
ments in the Asia-Pacific, 2001, 29 p. Sales No. E.01.II.D.21.
No. 16 Lucian Cernat, Assessing regional trade arrangements: Are South-South RTAs
more trade diverting?, 2001, 24 p. Sales No. E.01.II.D.32.
No. 17 Bijit Bora, Trade related investment measures and the WTO: 1995-2001, 2002.
No. 18 Bijit Bora, Aki Kuwahara, Sam Laird, Quantification of non-tariff measures,
2002, 42 p. Sales No. E.02.II.D.8.
No. 19 Greg McGuire, Trade in services – Market access opportunities and the ben-
efits of liberalization for developing economies, 2002, 45 p. Sales No. E.02.II.D.9.
No. 20 Alessandro Turrini, International trade and labour market performance: major
findings and open questions, 2002, 30 p. Sales No. E.02.II.D.10.
No. 21 Lucian Cernat, Assessing south-south regional integration: same issues, many
metrics, 2003, 32 p. Sales No. E.02.II.D.11.
No. 22 Kym Anderson, Agriculture, trade reform and poverty reduction: implications
for Sub-Saharan Africa, 2004, Sales No. E.04.II.D.5.
No. 23 Ralf Peters and David Vanzetti, Shifting sands: searching for a compromise in
the WTO negotiations on agriculture, 2004, Sales No. E.04.II.D.4.
No. 24 David Vanzetti, User manual and handbook on agricultural trade policy simu-
lation model (ATPSM), 2004, Sales No. E.04.II.D.3.
No. 25 Khalilur Rahman, Crawling out of snake pit: special and differential treat-
ment and post-Cancun imperatives, 2004, Sales No. E.04.II.D.8.
No. 26 Marco Fugazza, Export performance and its determinants: supply and de-
mand constraints, 2004, 59 p. Sales No. E.04.II.D.20.
50
United Nations publications may be obtained from bookstores and distributors throughout the world.
Please consult your bookstore or write to:
United Nations Publications
All orders from North America, Latin America, the Caribbean and Asia and the Pacific should
be sent to:
United Nations Publications
Room DC2-853, 2 UN Plaza
New York, NY 10017, USA
Telephone: (212) 963-8302, Toll Free 1-800-253-9646 (North America only)
Fax: (212) 963-3489
E-mail: [email protected]
Customers in Europe, Africa and the Middle East should send their orders to:
Section des Ventes et Commercialisation
Bureau E-4, CH-1211
Geneva 10, Switzerland
Telephone: 41 (22) 917-2613/2614
Fax: 41 (22) 917-0027
E-mail: [email protected]
51
52
QUESTIONNAIRE
UNCTAD Study Series on
POLICY ISSUES IN INTERNATIONAL TRADE
AND COMMODITIES
(Study series no. 26: Export Performance and Its Determinants:
Supply and Demand Constraints)
Readership Survey
Since 1999, the Trade Analysis Branch of the Division on International Trade in Goods and
Services, and Commodities of UNCTAD has been carrying out policy-oriented analytical work
aimed at improving the understanding of current and emerging issues in international trade of
concern to developing countries. In order to improve the quality of the work of the Branch, it
would be useful to receive the views of readers on this and other similar publications. It would
therefore be greatly appreciated if you could complete the following questionnaire and return to:
Jenifer Tacardon-Mercado
TAB/DITC, Rm. E-8076
Palais des Nations
CH-1211 Geneva 10, Switzerland
1. Name and address of respondent (optional):
2. Which of the following describes your area of work?
Government Public enterprise

Private enterprise institution Academic or research
International organization Media
Not-for-profit organization Other (specify) _________________
3. In which country do you work? _________________________________________
4. Did you find this publication Very useful Of some use Little use
to your work?
5. What is your assessment of the contents of this publication?

Excellent Good Adequate Poor
6. Other comments:
53

Export Performance and Its Determinants: Supply and Demand Constraints

Uploaded by

Copyright:

Available Formats

Export Performance and Its Determinants: Supply and Demand Constraints

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Export Performance and Its Determinants: Supply and Demand Constraints

Uploaded by

Copyright:

Available Formats

UNITED NATIONS CONFERENCE ON TRADE AND DEVELOPMENT

POLICY ISSUES IN INTERNATIONAL TRADE AND COMMODITIES

EXPORT PERFORMANCE AND ITS DETERMINANTS:

Associate Economic Affairs Officer

Material in this publication may be freely quoted or reprinted, but acknowledgement is

UNITED NATIONS PUBLICATION

© Copyright United Nations 2004

NON-TECHNICAL SUMMARY .......................................................................................1

I. THE THEORETICAL CONTEXT.........................................................................7

II. THE COMPONENTS OF GROWTH ...................................................................9

(a) The Dataset ....................................................................................................... 9

III. EXPORT PERFORMANCE AND ITS DETERMINANTS .................................25

(a) An Extended Theoretical Framework............................................................... 25

IV. POLICY IMPLICATIONS ....................................................................................41

V. POSSIBLE FURTHER RESEARCH .....................................................................43

1. Bilateral trade equation estimation ............................................................................ 10

1. Foreign market access and supply capacity................................................................. 13

Graph 1: Benchmarked export performance and components .............................................. 21

Internal transport infrastructure captured by the percentage of paved roads is an important

As a result of various trade negotiations and autonomous reforms, access to international

This study investigates factors possibly explaining divergence in export performance.

Foreign demand is influenced by various elements. Firstly , it is strongly linked to geography

The economy consists of a number N of countries. Only the manufacturing sector is

ni p i xij = ni pi1−σ (t i Tij t j ) E j Pj(σ −1)

X i = ni pi ∑ j ≠i xij = sci ∑ (Tij ) m j

(a) The Dataset

(b) Estimation Strategy

In determining the export performance of a given country, it is first necessary to quantify

Period 1980-83 1984-87 1988-91 1992-95 1996-99

Period 1980-83 1984-87 1988-91 1992-95 1996-99

(i) Foreign market Access

Foreign market access estimate takes the form5

Foreign Market Access

∆FMAi FMAiR1 ∆FMAiR1 FMAiRn ∆FMAiRn

Austria we 29 61 -10 45 -3 33 82 -8 41 23 -50 62 -5 37 77 -9 3 22 102 -5

Burkina Faso ssa 5 27 0 38 -9 26 86 -7 40 70 -54 133 -63 0 124 -13 3 28 84 -7

Canada na 24 22 3 47 24 8 103 -1 -6 6 -50 35 29 4 105 0 -8 43 89 -8

Algeria mena -25 -8 -30 41 -1 28 72 -8 -53 10 -57 41 -26 7 119 -15 0 29 71 -7

Argentina lac -4 38 24 66 -4 26 111 -3 -15 4 -43 55 -39 35 215 5 11 24 84 -7

Bulgaria eca -24 6 36 12 5 40 100 -9 -17 -22 -47 24 -2 38 117 10 5 40 100 -9

Australia eap 3 53 -7 37 6 38 98 -8 3 18 -56 45 10 64 111 -11 3 21 87 -5

Exports Growth FMA Growth Supply Capacity Growth

eap 44 44 8 43 7 46 108 -8 42 34 -88 53

eap 7 46 108 -8 10 64 122 -9 3 21 82 -6

Foreign Market Access IN

Foreign Market Access OUT

Generally speaking, there is evidence of a positive impact of Regional Trade Agreements

(ii) Supply capacity

(iii) Export Constraints

Exports FMA Sup.Cap.

Exports FMA Sup.Cap.

(Benchmark: ratio to 80-83 World Export Performance and Components)

(c) Sensitivity Analysis

(a) An Extended Theoretical Framework

In the theoretical framework presented previously, supply capacity, as indicated in equation

(b) The Data

The empirical counterpart of the extended framework takes the form:

ln( X i ) = α + λ ln(GDPi ) + β ln (POPU i ) + γ ln (FMAi ) + δ ln (ti ) + χCOMPi + ui

The technological environment is captured possibly, by the contribution of FDI, in