Mid-term Evaluation of the EU’s Generalised System
of Preferences:
Final Report submitted by:
Michael Gasiorek, CARIS, University of Sussex
with
Javier Lopez Gonzalez, CARIS
Peter Holmes, CARIS
Maximiliano Mendez Parra, CARIS
Jim Rollo, CARIS
ZhenKun Wang, CARIS
Maryla Maliszewska, CASE, Warsaw
Wojciech Paczynski, CASE, Warsaw
Xavier Cirera, IDS, Sussex
Dirk Willenbockel, IDS, Sussex
Sherman Robinson, IDS, Sussex
Kamala Dawar
Francesca Foliano, UCL, London
Marcello Olarreaga, University of Geneva
This report was commissioned and financed by the Commission of the European Communities. The
views expressed are those of the consultant and do not represent the official view of the Commission.
Personal data in this document have been redacted according to the General Data Protection Regulation 2016/679 and the European
Commission Internal Data Protection Regulation 2018/1725
2
Contents:
List of Tables ......................................................................................................................... 4 List of Figures ........................................................................................................................ 5 List of Abbreviations ............................................................................................................... 6 Executive Summary ............................................................................................................... 7 1 Introduction .................................................................................................................. 12
1.1 Motivation and objectives ........................................................................................ 12 1.1.2 Section Summary ................................................................................................ 15
1.2 Overview of the GSP............................................................................................... 16 1.2.1 Section Summary................................................................................................. 19
2 Preferential Access, Trade and Competitiveness ........................................................... 21
2.1 The structure of the EU‘s GSP regimes.................................................................... 22 2.1.2 Section Summary................................................................................................. 31
2.2 GSP and developing country exports ....................................................................... 32 2.2.1 EBA countries ...................................................................................................... 37 2.2.2 GSP+ countries:................................................................................................... 39 2.2.3 GSP countries:..................................................................................................... 43 2.2.4 Section Summary................................................................................................. 47
2.3 Impact of preference regimes on other LDCs ........................................................... 48 2.3.1 Similarities in export structures ............................................................................. 48 2.3.2 Relative Export Competitive Pressure Index .......................................................... 50 2.3.3 Section Summary................................................................................................. 54
2.4 GSP and LDC development needs .......................................................................... 55 2.4.1 Preferences and development .............................................................................. 55 2.4.2 Analysis of changes in intensive and extensive margin .......................................... 57 2.4.3 Section Summary................................................................................................. 65
2.5 Section 2: Conclusions ........................................................................................... 67
3 Utilisation Rates ........................................................................................................... 69
3.1 Descriptive stats on utilisation rates ......................................................................... 69 3.1.2 Section Summary................................................................................................. 73
3.2 The determinants of preference utilisation ................................................................ 74 3.2.1 Preference utilisation: econometric analysis .......................................................... 75 3.2.1 Section Summary................................................................................................. 83
3.3 Price margins – or who captures the preference rent? .............................................. 83 3.3.1 Methodology ........................................................................................................ 85 3.3.2 The Data ............................................................................................................. 86 3.3.3 Econometric Analysis ........................................................................................... 87 3.3.4 Section Summary................................................................................................. 95
3.4 Section 3: Conclusions ........................................................................................... 96
4 Gravity Modelling .......................................................................................................... 97
4.1 Aggregate modelling of trade and investment ........................................................... 97 4.1.1 The target model .................................................................................................. 97 4.1.2 Data .................................................................................................................. 100 4.1.3 Estimation and Results ....................................................................................... 101
3
4.1.4 Trade................................................................................................................. 103 4.1.5 FDI .................................................................................................................... 105 4.1.6 Section Summary............................................................................................... 106
4.2 Sectoral multilateral gravity modelling of trade........................................................ 108 4.2.1 Section Summary ......................................................................................... 108
4.3 The impact of preferences on trade flows at the product level ................................. 109 4.3.1 Section Summary ......................................................................................... 116
4.4 Section 4: Conclusions ......................................................................................... 117
5 Computable General Equilibrium Evaluation of GSP .................................................... 118
5.1 Introduction .......................................................................................................... 118 5.1.1 Section Summary ......................................................................................... 119
5.2 The GLOBE Model................................................................................................ 119 5.3 Patterns of Trade and Production in the Benchmark Equilibrium ............................. 123
5.3.1 Section Summary ......................................................................................... 132 5.4 Simulation Results ................................................................................................ 132
5.4.1 Change from 2004 to 2006 EU GSP – GSP06............................................... 133 5.4.2 A World without the EU GSP – MFN04/06 ..................................................... 134 5.4.3 Full Utilization of EU GSP Preferences – FULLGSP ...................................... 135 5.4.4 Further Reform of the EU GSP: The Extreme Case - ZEROTM ...................... 136 5.4.5 Section Summary ......................................................................................... 136
5.5 Section 5: Conclusions ......................................................................................... 150
6 Qualitative Assessment of the GSP+ ........................................................................... 151
6.1.1 Implementation & effects of international conventions .................................... 152 6.1.2 Lessons from the literature ........................................................................... 152 6.1.3 From ratification to implementation: legal analysis ......................................... 154 6.1.4 Challenges of implementation: lessons from case studies .............................. 161 6.1.5 Quantification of implementation effects ........................................................ 164 6.1.6 Section Summary ......................................................................................... 166
6.2 Costs and benefits of fostering sustainable development and good governance – GSP+ beneficiaries‘ perspective................................................................................................ 167
6.2.1 Section Summary ......................................................................................... 172 6.3 Selection criteria for GSP+ .................................................................................... 173
6.3.1 Vulnerability criteria ...................................................................................... 174 6.3.2 International conventions .............................................................................. 182 6.3.3 Section Summary ......................................................................................... 185
6.4 Section 6: Conclusions .......................................................................................... 187
7 Conclusions and policy recommendations ................................................................... 189
7.1 What do we learn from the analysis undertaken? ................................................... 189 7.2 Policy options ....................................................................................................... 191
7.2.1 Amending/improving existing schemes.......................................................... 191 7.2.2 Alternative policies ....................................................................................... 195
8 References:................................................................................................................ 199
4
List of Tables
Table 2.1: EU Imports by Preference Regime........................................................................ 22 Table 2.2: Coverage of EU Preferential Regimes 2008 .......................................................... 24 Table 2.3: Coverage of EU Preferential Regimes 2002-08 (share of tariff lines) ...................... 25 Table 2.4: Share of Tariff Lines by Regime and Size of Tariff (2008) ...................................... 25 Table 2.5: Average Tariff by Regime and TDC Sector (2002 and 2008) .................................. 26 Table 2.6: Preference Margins by TDC Sector Compared to MFN (2002 & 2008) ................... 27 Table 2.7: Preference Usage by Regime Type 2002-2008 ..................................................... 35 Table 2.8: Summary Results on Export Similarity .................................................................. 50 Table 2.9: Summary RECPI by Regime Type ........................................................................ 52 Table 2.10: Competitive Pressure by Country upon each Regime Type – all trade .................. 53 Table 2.11: Competitive Pressure by Country upon each Regime Type – MFN > 0 ................. 54 Table 2.12: Preferences and Development............................................................................ 57 Table 2.13: Annual Growth of Exports by Category 1991-2008 .............................................. 59 Table 2.14: Export Growth Decomposition by Varying Identification Procedures ..................... 61 Table 2.15: Differences between the Hypothetical MNF and Applied Tariffs ............................ 64 Table 2.16: Difference between the Hypothetical MFN & Hypothetical Applicable Tariffs ......... 65 Table 3.1: EBA Suitability ..................................................................................................... 71 Table 3.2: GSP Suitability ..................................................................................................... 72 Table 3.3: Determinants of Non Utilisation............................................................................. 82 Table 3.4: Export Price Ratio Specification ............................................................................ 89 Table 3.5: Multinomial Logit for Selection Utilisation .............................................................. 90 Table 3.6: Export Price Ratio Specification with Multinomial Selection (pref. margin) .............. 92 Table 3.7: Export Price Specification ..................................................................................... 94 Table 4.1: Percentage Change in Aggregate Trade ............................................................. 105 Table 4.2: Percentage Change in FDI ................................................................................. 106 Table 4.3: Percentage Change in Trade at Sectoral Level ................................................... 108 Table 4.4: Gravity Model at Product Level-Tariff Regime ..................................................... 111 Table 4.5: Gravity at HS4 with Selection ............................................................................. 114 Table 4.6: Gravity Model at HS4 with Wooldridge Panel Selection ....................................... 115 Table 5.1: Regional Aggregation of the Model ..................................................................... 122 Table 5.2: Commodity Aggregation of the Model ................................................................. 123 Table 5.3: Selected Benchmark Macro Indicators by Country............................................... 125 Table 5.4: Regional Origin Shares in Extra-EU Imports by Commodity (%) ........................... 126 Table 5.5: EU Share in Countries‘ Total Exports by Commodity (%) ..................................... 128 Table 5.6: Sectoral Shares in GDP by Country (%) .............................................................. 130 Table 5.7: Simulation Scenarios ......................................................................................... 132 Table 5.8: Percentage Changes in the Power of EU Import Tariffs by Scenario & Commodity Group ................................................................................................................................ 138 Table 5.9: Change in Real Absorption by Country and Scenario – ....................................... 139 Table 5.10: Change in Real Absorption by Country and Scenario – ..................................... 140 Table 5.11: Terms of Trade Change by Region and Scenario .............................................. 141 Table 5.12: Change in Aggregate Export Volume by Country and Scenario .......................... 142 Table 5.13: Change in Export Volume to the EU by Commodity – GSP06 ............................ 143 Table 5.14: Change in Export Volume to the EU by Commodity – MFN04 ............................ 144 Table 5.15: Change in Export Volume to the EU by Commodity - FULLGSP......................... 145 Table 5.16: Change in Export Volume to the EU by Commodity – ZEROTM ......................... 146 Table 5.17: Change in Real Output by Sector and Region – GSP06..................................... 147 Table 5.18: Change in Real Output by Sector and Region – MFN04 .................................... 148
5
Table 5.19: Change in Real Output by Sector and Region – FULLGSP ................................ 149 Table 6.1: Ratifications of 27 Conventions in Present and Past GSP+ Beneficiaries ............. 156 Table 6.2: Kendall Rank Correlation Coefficients between Export Concentration Ratios and Selected Economic and Geographical Characteristics ......................................................... 178 Table 7.1: Impact of changing the graduation threshold: ...................................................... 194
List of Figures
Figure 2.1: Incidence of Preference Margins at 10-digit level.................................................. 29 Figure 2.2: Distribution of countries by the share of the EU in total exports ............................. 33 Figure 2.3: EBA – Preference Margins .................................................................................. 38 Figure 2.4: EBA Countries: Share of MFN>0 Trade ............................................................... 39 Figure 2.5: GSP+: Tariffs and MFN Preference Margins ........................................................ 40 Figure 2.6: GSP+: GSP and EBA Margins............................................................................. 41 Figure 2.7: GSP+: Change in Share of Trade Eligible for Duty Free Access............................ 42 Figure 2.8: GSP+ Countries: Share of MFN>0 Trade ............................................................. 43 Figure 2.9: GSP Countries: Share of MFN>0 Trade ............................................................... 44 Figure 2.10: GSP Countries: Frequency Chart with Difference between MFN and GSP .......... 45 Figure 2.11: GSP Average Preference Margins ..................................................................... 46 Figure 2.12: Change in Share of Duty Free Eligible Exports under MFN & GSP+.................... 47 Figure 2.13: Changes in Exports by Type and Grouping (2002-2008) ..................................... 60 Figure 2.14: Correlation between Intensive and Extensive Margins and Applied Tariffs ........... 62 Figure 2.15: Correlation between Preference and Export Margins .......................................... 63 Figure 3.1: EBA Utilisation Rates .......................................................................................... 69 Figure 3.2: GSP Utilisation Rates.......................................................................................... 70 Figure 3.3: Correlation between Average Preference Margin and Utilisation Rates ................. 74 Figure 3.4: All GSP Regimes: Correlation between Preference Margin and Non-Utilisation of Preferences ......................................................................................................................... 75 Figure 3.5: Probability Distribution Function of Preference Non-utilisation Exports as a Share of Total Exports in 2007 – by Country ....................................................................................... 77 Figure 3.6: Probability Distribution Function of Preference Non-utilisation Exports as a Share of Total Exports in 2007 – by Product ....................................................................................... 77 Figure 3.7: Probability Distribution Function of Preference Non-utilisation Exports as a Share of Eligible Exports in 2007 – by Country .................................................................................... 78 Figure 3.8: Probability Distribution Function of Preference Non-utilisation Exports as a Share of Eligible Exports in 2007 – by Product .................................................................................... 79 Figure 3.9: Prices and the Preference Rent ........................................................................... 84 Figure 3.10: Kernel Estimate of pdf for Log Ratio of Prices .................................................... 87
6
List of Abbreviations
AGOA: African Growth and Opportunity Act
CAFTA: Central American Free Trade Agreement
CAP: Common Agriculture Policy
CES: Constant Elasticity of Substitution
CET: Constant Elasticity of Transformation
CGE: Computable General Equilibrium
CITES: Convention on International Trade in Endangered Species
CPIA: World Bank‘s Country Policy and Institutional Assessment
EBA: Everything But Arms
EU: European Union
FDI: Foreign Direct Investment
FK: Finger Kreinin index
GEI: Gender Equality Index
GNI: Gross National Income
GSP: General System of Preferences
HDI: Human Development Index
HPI: Human Poverty Index
ILO: International Labour Organisation
LDCs: Least Developed Countries
MFN: Most Favoured Nations
NAFTA: North American Free Trade Agreement
NGOs: Non-Governmental Organisations
NTMs: Non-Trade Measures
OECD: Organisation of Economic Cooperation and Development
OLS: Ordinary Least Squares
PTAs: Preferential Trade Arrangements
PPP: Purchasing Power Parity
RCA: Revealed Comparative Advantage
RECPI: Relative Export Competitive Pressure Index
RoO: Rules of Origin
RTAs: Regional Trade Agreements
TOT: Terms of Trade
WTO: World Trade Organisation
UNCTAD: United Nations Conference on Trade and Development
UNICEF: United Nations Children‘s Funds
7
Executive Summary
Overview:
1. This report considers the extent that the EU‘s GSP regimes meet the needs of developing
countries and puts forward recommendations for possible improvements.
2. The report is structured into 7 sections: (1) Introduction and overview of the GSP scheme;
(2) an analysis of the degree of preferential access, trade and competitiveness using
descriptive statistics; (3) an evaluation of utilisation rates and determinants of utilisation; (4)
assessing the impact of the GSP scheme through a gravity modelling framework at the
aggregate, sectoral and bilateral-product level; (5) a computable general equilibrium
analysis of the GSP scheme; (6) an assessment of the GSP+ scheme; (7) conclusions and
recommendations.
3. More precise information on preferential trade between the EU and its partner countries was
used in this study than in previous studies. Previously unavailable highly detailed data was
used for the analysis of GSP preferences. This 10-digit data on trade and tariffs for any
given product, country and year, distinguishes between the regime of entry into the EU. It
can be used to identify whether product ―x‖ is eligible for preferential access to the EU from
country ―y‖ together with the appropriate tariff; it can also be used to calculate how much
trade actually entered under that given regime, and how much trade for the same product,
country and year combination may have entered via a different regime.
4. Positive evidence of the effectiveness of the EU‘s GSP scheme was identified using this
data: there is some evidence that the EU‘s GSP preferences can be effective in increasing
LDC exports and welfare; that utilisation rates are typically high, that LDC exporters tend to
benefit from preference margins received, and that countries seeking GSP+ status attempt
to ratify the appropriate conventions.
5. However, there are also a number of important caveats when considering the policy
implications arising from this study. These caveats centre on structural features, such as the
generally low level of EU MFN tariffs and the structure of LDC trade, which inevitably
constrain the effectiveness of the GSP regime.
6. The policy conclusions focus on measures to increase the effectiveness of the GSP
scheme, including issues such as product coverage, further tariff reductions, maximising
utilisation, rules of origin, and the role of graduation as well as general improvements to the
GSP+ scheme. We also consider alternative trade-based policies. These we argue are likely
to be important in focusing on the trade and development needs of those developing
countries most in need, such as aid for trade policies, policies for non-tariff measures and
EU import subsidies.
8
Conclusions from a consideration of the descriptive data:
7. The EBA has many more tariff free lines than GSP+, which in turn, has many more than
GSP. Under GSP there are 4781 additional duty free tariff lines, 9717 under the GSP+ and
under EBA 11053. The number of MFN greater than zero lines is similar across the GSP
and GSP+ regimes.
8. Over time there is an increased number of MFN zero lines, resulting in preference erosion
for those countries with preferences. Again, there are substantial differences between GSP,
GSP+ and EBA, both in numbers of tariff lines equal to zero and also in the levels of tariffs
applied.
9. The structure of the EU‘s preference regimes‘ average tariffs, tariff peaks and preference
margins means that the scope for offering significant preferential access to developing
countries is largely limited to a few sectors (live animals, vegetable products, processed
foodstuffs, textiles, and clothing).
10. The assessment of the importance of preferences by country groupings indicates that on
average a high proportion of GSP countries' trade enters under MFN=0. In 2008
64.45 percent of GSP countries exports to the EU entered the EU with a zero MFN tariff,
61.26 percent of GSP+ countries' exports, and 62.85 percent of EBA countries' exports.
11. The shares of trade paying a positive MFN tariff for the GSP, GSP+ and EBA countries in
2008 were 22.07 percent, 13.18 percent and 6.08 percent respectively. Overall, these
shares have been rising over time. This suggests there is more scope for improved access
to the EU, either by improving the preferences or by increasing their utilisation.
12. On average the preference regimes themselves do not, however, account for a substantial
amount of the relevant countries‘ trade with the EU. This is even more the case if we
consider their share of total trade, as opposed to solely their trade with the EU. In 2008, on
average just over 7 percent of GSP countries' exports used GSP preferences when
exporting to the EU. For the GSP+ and the EBA countries this was just over 24.5 percent
and 23.4 percent respectively. Both the GSP countries and the EBA countries exported
around 5 percent of their trade using other preference regimes. For GSP+ countries, the
share using other preferences was zero, while for those countries with other preferential
regimes it was just over 12 percent.
13. This suggests that with low MFN tariffs, relatively few tariff peaks, and the composition of
LDC exports, the extent to which bilateral preference regimes can help developing countries
is, in principle, structurally limited.
14. Analysis using the Finger-Kreinin index of export similarity and the relative export
competitive pressure index (RECPI) suggests that the greatest amount of competitive
pressure for EBA countries comes from GSP and MFN exporters. For GSP countries, the
principal source of competitive pressure comes from MFN exports, while for the GSP+
countries it comes from the GSP exporters.
15. There is little evidence that the EU‘s preference regimes have led to a diversification of
exports into new products.
9
16. The relationship between preference margins, utilisation rates, and different measures of
development does not suggest a high degree of correlation between countries‘ development
needs and the height of the preference margin, or the extent of preference utilisation.
17. Changing the graduation thresholds is likely to have some positive impact on EBA exports,
but at the expense of the GSP countries who graduate. In aggregate this would appear to be
a blunt way of helping those countries most in need. It is also worth noting that for any given
country, graduation tends to introduce distortions with respect the relative export prices.
Such distortions can lead to a misallocation of resources.
Conclusions from the econometric analyses:
18. Utilisation rates are typically high, though not for all countries, and are positively related to
the height of the tariff and the extent of the preference margin, and with mixed evidence
regarding rules of origin.
19. The rents from preference margins are not entirely absorbed by the importers, the evidence
suggests that exporting countries appropriate between a half to all of the implied rents.
20. The aggregate gravity modelling of trade suggests that trade between the EU and
developing countries is typically lower than that of non-developing countries. Once this
factor is controlled for, the growth of trade and investment with the EU in recent years has
been higher for GSP preference receiving countries than for non-beneficiary countries. The
increase in trade ranges from just over 10 percent for the Cotonou group of countries, to
nearly 30 percent for the GSP+ group of countries.
21. The aggregate gravity modelling of investment suggests a positive impact of the preference
schemes on FDI flow, although data constraints make a literal interpretation of the numbers
unwise.
22. The sectoral gravity modelling was undertaken for six sectors (vegetable products, prepared
foodstuffs, footwear, textiles, clothing, machinery). This resulted in a mixed picture on the
impact on trade, depending on the sector and on the regime of entry.
23. The bilateral gravity modelling exercise identified some evidence that preferences arising
from the EU‘s free trade agreements as well as those applied to the Cotonou countries had
a positive impact on trade with the EU, rather than EBA, GSP, or GSP+ arrangements.
Conclusions from the CGE analysis:
24. The incremental change in applied EU GSP tariff rates from the pre-2006 to the 2006-08
system generates only small aggregate welfare gains for GSP beneficiaries, except for a
sub-set of Latin American GSP+ countries.
25. Among the EBA regions in the model, Cambodia and Bangladesh benefit most from the EU
scheme, while the EBA Sub-Saharan Africa composite region gains very little overall
(however, due to data constraints not all EBA countries in sub-Saharan Africa are included
in this composite region). Among the GSP+ countries, the biggest gainers are Ecuador and
Costa Rica. Understandably, welfare gains are considerably smaller for the ordinary GSP
10
countries with moderate preference margins vis-à-vis MFN tariffs, with the exception of
North Africa, and Southern and Eastern Europe.
26. While there are some significant trade and output effects for a sub-set of agricultural
commodities and regions (notably fruits and vegetables in Ecuador, Costa Rica and
Argentina, sugar products in the Caribbean, North Africa and Sub-Sahara African EBA
beneficiaries, oils and fats in North Africa), the substantial expansionary impacts of the EU
GSP occur in the textile, apparel and leather goods industries within Southern and Eastern
Europe, North Africa, Cambodia and Pakistan.
27. Perhaps counter-intuitively, the underutilization of existing EU GSP preferences is not a
major factor reducing the potential gains from the existing GSP scheme in comparison to the
full utilization of existing preferences.
28. A hypothetical complete removal of all EU duties on imports from existing GSP leads to
large gains for a subset of the Latin American GSP+ countries and the standard GSP
countries Thailand, Argentina and Brazil. In contrast, all EBA regions in the model lose out in
this speculative borderline scenario – a clear-cut case of preference erosion.
29. In all the scenarios under consideration, the aggregate welfare impacts on the EU are
negligible.
Conclusions from the GSP+ analysis
30. It is too early to tell whether the GSP+ will become an effective mechanism promoting
sustainable development and good governance. Significant progress in these spheres tends
to take longer than the scheme‘s timeframe to date. One general conclusion from the
literature is that the design of the GSP+ is relatively robust in providing opportunities for
improvements in some countries or in some spheres, while the risk of negative effects is
very limited.
31. GSP+ appears to be effective in promoting ratifications of the 27 conventions. Case studies
and a literature review suggest that de jure implementation beyond ratification already faces
several constraints. We do not find evidence of any significant positive effects of GSP+ here.
32. De facto effects are yet more difficult to identify, measure and compare across countries and
time. We find some evidence suggesting positive effects in the sphere of gender equality. In
other spheres, such as corruption, civil liberties, etc., we find no effects. We do not identify
any negative effects of GSP+ on de facto implementation.
33. The costs of effective implementation of human rights conventions are mainly related to the
social and economic rights dimension, where the adequate provision of education and
health services is in practice very difficult in a number of developing countries. While these
costs are high, the literature suggests that benefits outweigh costs by a large margin.
34. Costs of implementation are an important factor in countries' decisions to adopt international
labour conventions. Case studies suggest that in some instances the costs of complying
with ILO conventions in practice can be identified with the costs of effective implementation
11
of the labour code. Overall, benefits are believed to outweigh costs, in some instances (e.g.
child labour) by a very large margin.
35. Most of the economic literature suggests potential significant gains from good governance,
particularly in the reduction of corruption, although this view is not uncontested. The
information from the case studies suggests that costs incurred have been small, largely due
to very limited implementation.
36. A cost-benefit analysis of environmental conventions is complex for several reasons. GSP+
countries have ratified several of the environmental conventions only fairly recently.
Progress with implementation somewhat limited, giving little information on actual costs. The
role of foreign aid is very important in financing the implementation efforts. It could be
argued that the GSP+ conventions have motivated donor resources that would otherwise
not have entered the countries. Given that many of the projects required under the
conventions (reporting, data collection, action plans, etc.) are costly, they would not have
been implemented without external support.
37. Our analysis indicates that the current vulnerability criteria are broadly consistent with the
selection of smaller, landlocked countries, prone to terms of trade shocks and with limited
export diversification, as measured at the product level. However, the criteria are not
strongly linked to income per capita levels. This is not particularly problematic given that
almost all of the poorest countries are classified as vulnerable. However, modification of the
criteria ensuring that countries below certain income per capita level are considered
vulnerable irrespective of their exports to the EU could be discussed.
38. To improve the stability and predictability of the vulnerability criteria, we recommend the
introduction of a three-year transitional period before a country loses its vulnerable status.
39. Another area where some modifications could be proposed concerns the selection of
conventions. However, we do not see a clear-cut case either for reducing the number of
conventions to avoid duplication of their mandates (e.g. the ILO Convention concerning the
Abolition of Forced Labour and the ILO Convention concerning Forced or Compulsory
Labour) or for introducing new ones. There are arguments in favour of both strategies and
more experience with the current scheme might be needed before a decision on
modifications is taken.
12
1 Introduction The EU‘s Generalised System of Preferences (GSP1) is a central element of the EU‘s strategy
towards developing countries. The EU aims to promote an understanding of sustainable
development that incorporates trade as an essential element facilitating economic and social
development. The GSP scheme is a core part of the EU‘s trade strategy towards developing
countries, alongside other policies such as the Economic Partnership Agreements (EPAs) and
other bilateral and regional trading agreements. The GSP scheme has evolved considerably
over the years, with substantial changes occurring in 2006. The most recent scheme is
applicable from 1st January 2009 to 31st December 2011.
The overall aim of this report is to consider the extent to which the GSP regimes corresponds to
the needs of developing countries, and in that context to put forward recommendations for
possible ways forward.
The report is divided into sections and is set out as follows:
1. Introduction and overview of the GSP
2. Preferential access, trade and competitiveness
3. Utilisation Rates
4. Gravity Modelling: aggregate, sectoral and bilateral
5. CGE analysis of the GSP scheme
6. Assessing the GSP+ scheme
7. Conclusions and policy recommendations.
The remainder of this introduction discusses some of the broader issues that underlie the
objectives of the study, it identifies the important issues that this research must address and
summarises the key elements of the EU‘s GSP scheme.
1.1 Motivation and objectives An important part of this study is to consider how the EU‘s GSP system could be reformed or
improved in order to better address the growth and development objectives through
encouraging the trade of developing countries, especially those most in need. The issue of
growth and development objectives clearly raises a set of wide-ranging and interlinked issues to
do with the domestic constraints and distortions within individual countries, as well as the
relationship between these and the external environment they face, their internal stance with
regard to trade policy, and more broadly the domestic policy agenda. In this light it needs to be
recognized that the external trading environment, such as the GSP system, can at best only be
a facilitator, albeit potentially a significant one, towards the meeting of the growth and
development objectives. It is therefore only likely to be successful when combined with an
appropriate domestic institutional environment and appropriate domestic policies. It is also worth
noting that even with regard to trade objectives, the extent to which the EU‘s GSP scheme could
1 In this document unless it is explicitly stated where we refer to the ―EU‘s GSP scheme‖ we take
this to include GSP and GSP+.
13
impact on any given developing country will also depend on the importance of the EU in that
country‘s overall export markets.
The principal role that the GSP could play is to encourage greater growth of developing country
exports in existing products (the intensive margin), and through diversifying into new products
(the extensive margin), consequently contributing to the development process.
In this context GSP success could include (in no order of importance):
o Greater impact on those developing countries most in need – the most vulnerable,
those with the lowest income levels, small islands, landlocked, etc.
o Higher economic growth, as a result of higher exports and greater integration in the
world economy.
o More regional trade, which may in turn be influenced by possibilities of regional
cumulation in the underlying rules of origin.
o A positive impact on ―sustainable‖ development, in the context principally of areas
such as labour standards, environment etc.
o Reduction in poverty.
o Diversification.
o A positive impact on investment flows.
It is worth underlining that a successful GSP scheme can have these effects because it offers
developing countries preferable or preferential access relative to other suppliers into the EU
market. The extent of its success therefore must depend on the extent of that preference
margin, as well as on the relationship between that margin, and the incentive for firms and
countries to utilize the preferences on offer.
The core mechanism transmitting these beneficial effects is preferential access to markets,
which may lead to higher levels of exports and consequently imports. This can enable countries
to develop better and/or more industries, leading to increases in productivity, competitiveness
and possibly diversification. It may also encourage more investment. This may be related to the
stability and time frame of the preferential regime, which are also related productivity and
diversification issues.
Each of the positive impacts noted above may enable the economy to become more productive
and increase levels of growth, thus increasing aggregate income per capita. The relationship
between this transmission mechanism, poverty and sustainable development is therefore highly
complex. For example, even where increased exports may lead to higher growth rates, this may
not necessarily lead to a reduction in poverty as the impact of trade on poverty depends on the
availability of relevant transmission mechanisms (see McCulloch, Winters and Cirera (2002)).
This is because changes in trade can impact on consumption choices, on relative prices
therefore inducing sectoral reallocation with consequent distributional effects, and on revenue
from trade taxes. The greater engagement in international trade also raises issues of
diversification versus specialisation, which are in turn often related to vulnerability, as well as
14
issues of the geographical concentration of economic activity (economic geography) and long-
run spillover effects.
The analysis of the GSP undertaken in this study is therefore intended to first evaluate the
existing operation of the GSP scheme and to ascertain the extent to which it appears to be
tailored towards those countries most in need. In assessing the impact and effectiveness of the
EU‘s GSP scheme it is important to identify as precisely as possible the role of the GSP scheme
itself, as opposed to the impact of other changes in trade policy either within the countries
themselves, or indeed with regard to other trading partners. Empirically, as is well known, this is
a difficult task. In order to do so it is important to have some variation either across time, across
countries or across sectors with regard to the GSP regime faced, which is then not highly
correlated with some other policy change.2 For this study we have been given access to
extremely rich and detailed trade and tariff data which allows us to identify the actual use of
preferences by country and HS 10-digit trade category. This enables us to consider the role of
the GSP scheme much more precisely than previous work in this field.
A second important set of issues addressed in this study concern the policy recommendations
that might arise. In part these policy recommendations are likely to stem from the analysis
evaluating the current system – from examining the relative effectiveness of the different
regimes – GSP, GSP+ and EBA, and its application to those most in need. Here it is worth
noting that preferential access is likely to give countries a comparative advantage in the EU
market which they otherwise would not have had. This can lead to trade being diverted away
from other developing countries – hence while the preferences in a given sector may impact
positively on one country they may have a negative impact on third countries. This in turn is
likely to depend on the speed and costs of adjustment in the third country and the nature of
competitive interaction. Trade diversion and its opposite, trade reorientation, are therefore likely
to be a feature of the differences in the preference schemes, of graduation and de-graduation,
and of any change in MFN tariffs. This will need to be borne in mine together with the
possibilities for trade creation.
Consideration of the policy options will also result from a consideration of the literature on GSP
schemes. Broadly speaking however, there are two policy approaches available which are not
necessarily mutually exclusive. The first approach is based on reforming elements of the
existing system. For example, this could be in relation to the product coverage of the GSP or
GSP+ schemes, or it could be in relation to the underlying rules of origin and their operation.
Similarly, the issue of graduation will be important to consider. Would amending the current
graduation thresholds help those countries most in need? How does graduation impact both on
those countries who have graduated and also on third countries? Here again, the ex post
analysis undertaken in the main body of the study will be able to consider these issues.
The second approach is to consider whether there are any alternative policies which may be
worth pursuing. Here it will be important to consider the extent to which such policies fall within
2 See for example Evenett (2008).
15
the remit of the EU, or whether they might require international agreement, for example at the
WTO. Closely related to this is the question of trying to benchmark the GSP scheme against
alternative (and maybe first best) instruments. The issue here is whether there may be more
efficient alternatives in particular with regard to the integration of developing countries in the
world economy by impacting not only on access to third markets but also on domestic
incentives. For example Olarreaga and Limao (2005), put forward the suggestion of import
subsidies.
As preference erosion occurs with the decline in MFN tariffs, countries and sectors may lose the
comparative advantage afforded to them by the preferential access and thus and exports/growth
may decline. In the context of this study it will therefore be important to consider the evolution
not simply of preferential trade policy but also multilateral trade policy. For example, where the
current preferential arrangements appear to be subject to the impact of preference erosion,
which inevitably diminishes their effectiveness, import subsidies would not have the same
drawback. Similarly with the decline in MFN rates and the consequent preference erosion, it
may also be interesting to consider the possibilities for preferential treatment with regard to non-
tariff measures, such as in the area of SPS or TBTs, which can also serve to restrict access to
markets. To the extent also that preference erosion may in turn have complicated the process of
multilateral trade liberalisation, alternative preferential policies may help to ease the logjam.
It is also important to bear in mind that trade economists typically see welfare and
efficiency/productivity gains from trade coming primarily from domestic liberalisation and not
simply from increased access to export markets and increased exports. This therefore raises an
interesting question concerning the relationship between GSP schemes and domestic trade
policy. Here the insights of Baldwin suggests that it may be the case that increased exposure to
export markets changes the domestic political economy in favour of greater domestic
liberalisation (the ‗juggernaut effect‘). The research is not unanimous, however, for example,
Ozden and Reinhardt argue that countries that receive GSP tend to be more protectionist.
1.1.2 Section Summary
The EU‘s GSP scheme offers developing countries preferable access to EU markets relative to
other countries‘ suppliers with the aim of promoting sustainable development in poorer
countries. Its function is to encourage greater growth of developing country exports in existing
products and encourage diversification into new products. This can potentially contribute to
development by, for example, increasing productivity, poverty reduction, improving standards
and increasing foreign direct investment.
However, this section notes that there are limits to how much the EU‘s GSP scheme can
achieve on its own. It is likely to be more successful when combined with appropriate domestic
institutions and policies and when the EU is an important export market for a developing
country. Additionally, eligible developing countries need to utilize the preferences on offer.
16
The section therefore introduces the complex relationship between the GSP scheme, poverty
reduction and sustainable development because even where increased exports may lead to
higher growth rates, it may not necessarily lead to a reduction in poverty. This has important
consequences for formulating effective policy recommendations.
1.2 Overview of the GSP In 1968 UNCTAD recommended that developed countries adopt generalized systems of trade
preferences for exports from developing countries, and in 1971 the European Union became the
first to adopt such a preference scheme. Since its inception in 1971, the European Community
and its successor the European Union has intended to implement its GSP regime through ten-
year long programmes. However, formally single multi-year regulations, currently lasting three
years, were promulgated by the EU, in effect allowing the EU‘s GSP regime to change over
time. Changes, sometime substantial, in GSP provisions have occurred at interim reviews.
The current GSP scheme is distinctive from the previous GSP scheme prior to 2006 in terms of
predictability and simplicity. It runs three years relative to one year – GSP coverage and country
eligibility are no longer subject to annual revisions. It is composed of three rather than five
separate regimes. The three different preference programs under the current GSP are: (a) the
basic or general GSP for which all 176 developing countries and territories are eligible; (b)
GSP+ program which offers additional tariff reductions on top of the general GSP to a selected
group of developing countries that are vulnerable and are implementing specified core
international human, labour and environmental standards and with respect to good governance;
(c) the Everything-but-Arms program offers duty-free and quota-free market access to the 50
Least Developed Countries (LDCs).
Under the EU‘s GSP scheme imports by the EU from developing countries amounted to EURO
40 billion in 2004, compared with EURO 22 billion by the US under its GSP scheme, the second
most widely used. The value of EU imports under its GSP is also greater than the total value of
imports under the US, Canadian, and Japanese GSPs combined. The EU‘s imports from all
GSP eligible countries have increased steadily since 2004, EURO 46 billion in 2005, EURO 51
billion in 2006 and EURO 58.6 billion in 2007. Imports from EBA eligible countries also
increased by 35 percent in 2006. The GSP+ beneficiary countries exports to the EU increased
15 percent in 2006 and a further 10 percent in 2007.
However, evidence on the development impact of the GSP appears to be mixed; together with
the ongoing preference erosion resulting both from the decline in MFN tariffs and the
proliferation of regional trading arrangements, it is clear that the GSP needs careful evaluation.
Basic GSP: The European Union‘s basic GSP provides preferences for which all developing
countries are automatically eligible and is more favourable for some products than the EU‘s
17
MFN tariffs. The EU reports that of the 10,300 tariff lines in the EU‘s Common Customs Tariff,3
roughly 2,100 products have a MFN duty rate of zero and tariff preferences are not relevant for
these. Of the 8,200 products that are dutiable, GSP covers roughly 7,000, of which about 3,300
are classified as non-sensitive and 3,700 as sensitive. Of the rest of tariff lines not covered by
the GSP, a number of them fall into HS chapter 93, arms and ammunition. Non-sensitive
products have duty free access and sensitive products benefit from a tariff reduction. The
sensitivity of product is determined by whether or not it is produced in the EU and by how
competitive European producers are. The non-sensitive category covers most manufactured
products4 but excludes some labour intensive and processed primary products -- such as
textiles, clothing and footwear. In addition, agricultural products covered by the EU‘s Common
Agriculture Policy are deemed to be too sensitive to be granted duty-free market access from
any potentially large and competitive suppliers.
For the sensitive products, the tariff preference is a flat 3.5 percentage point reduction from the
corresponding ad valorem MFN tariff rates. For example, a reduction in a MFN rate of
14 percent by a flat 3.5 percentage points results in a preferential duty rate of 11.5 percent (the
reduction from a 14 percent to an 11.5 percent tariff is a 25 percent preferential margin, or a
25 percent reduction in the MFN duty). While if the MFN rate is 7 percent, a reduction by 3.5
percentage points results in a preferential duty rate of 3.5 percent (the reduction from 7 percent
to 3.5 percent is a 50 percent reduction of MFN tariff). The flat 3.5 percentage point reduction
does not apply to the textile and clothing sectors. For these sectors, the reduction is 20 percent
of the applicable MFN tariff rate.
There is a graduation clause in the basic GSP and GSP+ schemes. This clause does not
affect EBA eligible countries. Graduation is triggered when a country becomes competitive in
one or more product groups. Preferential access is withdrawn for exports of a given product
group (section of the custom code) for any country for which exports of the product group
exceed 15 percent of total EU imports of the same product group under the GSP over the past
three consecutive years. For textiles and clothing, the threshold for withdrawal of basic GSP
preferences is 12.5 percent of the EU‘s total imports of textiles and garments under the GSP.
For example, preferential access for Vietnamese exports of footwear, headgear, artificial flowers
are suspended due its success in these exports. Of course, the same principle is applied to the
de-graduation or re-establishment of preferences. (For example, preference access to Algeria
exports of mineral products, Indian exports of pearls, precious metal and stones to the EU
markets have been re-established). In terms of GSP terminology, covered imports refers to all
imports listed in the GSP regulation, whether or not a country is graduated out of any sectors;
3 European Commission: ―Generalized System of Preferences – user‘s guide to the European Union‘s
scheme of Generalized Tariff Preferences‖. The EU Common Custom Tariff is based on the Harmonized
System nomenclature and supplements it with its own subdivisions referred to as Combined
Nomenclature (CN) subheadings. Each CN has eight digit code number. The first six digits refer to the HS
headings and subheadings. The seventh and eighth digits represent CN subheadings. The EU reported
total number of approximate 10,300 tariff lines of the Common Custom Tariff. 4 HS chapters 25 to 99, excluding chapter 93, arms and ammunition. See the European Commission
website on trade – GSP.
18
eligible imports are then all the imports listed in the GSP regulation and for which the country
receives the GSP preference reduction. For the purposes of graduation calculations it is EU
covered imports which are used. Hence even if a country is currently graduated for most of its
imports under GSP, such as China for example, all that country‘s imports into the EU are
included when calculating the shares of EU imports accounted for by all other countries.
The GSP+ Program: The European Union also adopted a ―Special incentive arrangement for
sustainable development and good governance‖ (GSP+ program), which provides additional
preferences for those vulnerable non-LDCs that comply with a list of 16 international
conventions on human and labour rights, and 11 conventions on good governance and the
environment. The GSP+ tariff preferences are more attractive than the regular GSP
preferences.
The design of the GSP+ program was motivated in part by an unfavourable WTO ruling against
a previous EU scheme providing special preferences for selected developing countries that
were actively implementing anti-narcotics programs. The dispute panel‘s ruling states that it is
permissible to differentiate among non-LDCs as long as the distinctions among countries are
based on ―a widely-recognized development, financial, [or] trade need.‖ Accordingly, the
European Union‘s new GSP+ provides for greater preferences for vulnerable non-LDCs meeting
specific widely recognized criteria including ratification and implementation of international
conventions on human and labour rights, good governance and the environment.
The GSP+ program offers additional tariff reductions. It allows preferential access to the EU
market for imports from eligible developing countries for the same 7,000 products as the EU‘s
basic GSP scheme as well as a few other products that are excluded from basic GSP
preferences.5 But all products enter at zero rate ad valorem duty under the GSP+ program,
rather than some at a zero rate and some at a reduced rate from the MFN ad valorem tariffs as
under the basic GSP program. Note, however, when a tariff line is subject to both ad valorem
and a specific duty, only the ad valorem duty is waived.
In order to be eligible for the GSP+ program, a country must first be classified as ―vulnerable‖ by
satisfying the following two criteria: (a) a country cannot be classified as high income and its five
largest sections of its GSP-covered exports to the EU must account for over 75 percent of its
total GSP-covered exports; and (b) GSP-covered exports from the country must represent less
than 1 percent of total EU imports under the GSP.
Then to qualify for the additional preferences under the GSP+ program, a vulnerable country
must have ratified and effectively implemented twenty-seven of the most important international
conventions. In addition to ratification of these conventions, the country is required to provide
comprehensive information concerning the legislation and other measures to implement them.
5 Examples include natural honey, asparagus (uncooked or cooked by steaming or boiling in
water), frozen, or strawberries, raspberries, blackberries, mulberries, loganberries, black-, white and
redcurrants, and gooseberries – see footnote (3) to Annex II to the Council Regulation (EC) No 732/2008
of 22 July 2008, OJ L 211/1.
19
It must commit itself to accepting regular monitoring and reviewing of its implementation record.
Finally, the country must make a formal request to qualify for GSP+. 16 countries were granted
GSP+ preferences from January 2009, but in mid-2009 Venezuela was deleted from the list of
beneficiary countries.6
The GSP+ program has some limitations. First, like the basic GSP, the GSP+ program does not
cover 1,200 of the EU‘s tariff lines that have non-zero MFN tariff rates. Products deemed very
sensitive like beef and other meats, dairy products, some processed fruits and vegetables, oils
and processed sugar, are not covered by the GSP+ program. Second, like in the case of basic
GSP, graduation rules also apply to the GSP+ program. Third, there may be limitations related
to the application of rules of origin. Fourth, the implementation of some the international
conventions required for eligibility for GSP+ may not be an immediate development priority in
many low income countries and may distract attention and effort from other possibly higher
priority reforms needed to accelerate growth and poverty reduction.
Everything but Arms (EBA): The European Union provides special preferences to all LDCs
under its Everything but Arms (EBA) program adopted in March 2001. Under its EBA program,
the European Union has unilaterally granted to 50 least developed countries quota-free and
tariff-free access to its market for all products except arms without the LDCs‘ having to give
reciprocal preferential access to the EU in return. The EBA program is the most generous one
of the European Union‘s Generalized System of Preferences. It is compatible with the WTO‘s
enabling clause as it grants special preferences to a permissible grouping of developing
countries, the LDCs.
1.2.1 Section Summary
This section described the development of the EU‘s GSP system and describes the current
framework incorporating three separate regimes:
(a) the basic or general GSP for which all 176 developing countries and territories are
eligible;
(b) GSP+ program which offers additional tariff reductions on top of the general GSP to a
selected group of developing countries that are vulnerable and are implementing
specified core international human, labour and environmental standards and with
respect to good governance;
(c) the Everything-but-Arms program offers duty-free and quota-free market access to
the 50 Least Developed Countries (LDCs).
6 Commission Decision of 11 June 2009, OJ L 149/78.
20
Both the GSP and the GSP+ schemes incorporate a graduation clause which is triggered when
a country becomes competitive in a given product group which results in the suspension of
preferential access for these products.
Taken together, the EU‘s GSP schemes are significant. The value of EU imports under these
systems is greater than the total value of imports under the US, Canadian, and Japanese GSPs
combined.
21
2 Preferential Access, Trade and Competitiveness The underlying principle of the EU‘s GSP scheme is that preferential access can play an
important role in fostering sustainable development. This part of the study focuses on identifying
the de jure degree of preferential access granted to developing countries, their differences
across preference regimes (notably GSP, GSP+ and EBA), as well as on the relative amounts
of trade covered and the linkage between this and underlying competitiveness.
The analysis is based on extremely detailed (10-digit) trade and tariff data supplied by the
Commission services. The advantage of working with extremely detailed data as that it allows
for much more precise calculation and the results are not subject to possible aggregation bias.7
There are a number of detailed tables that underlie the discussion in this section of the report as
well as the subsequent section. Where relevant we include the tables in the main body of the
texts. Supplementary tables and other detailed tables providing country level information can be
found in the Appendices attached.
In this part of the report we cover four areas:
1. First, we look at the structure of the EU‘s GSP system, by examining the degree of
coverage and preferential access which is currently granted under the EU‘s existing
regimes. In the first instance this involves examining the differences in structure in
aggregate across the GSP, GSP+ and EBA regimes, and then examining the differences
by sector. In the second instance this involves looking at the differences in tariffs and
preference margins across sectors.
2. Secondly we address the issue of the suitability of the GSP regimes by considering the
extent to which the preferences offered by the EU align with the structure of developing
country exports, and thus we link the EU‘s preference structure with that of each
developing country's export profile.8This involves looking at the share of each country‘s
exports covered by the regimes and comparing this to the shares of MFN tariff-zero
trade.
The purpose of this analysis is to assess the significance of the GSP regime for
developing countries in terms of the preference margins which they entail, bearing in
mind the developing countries‘ exporting structures. We also consider the ―suitability‖ of
the preference structures by exploring what the change in the average tariff countries
face in the EU, or what the difference in exports might be for each country under the
different preference regimes. It is by looking at the difference across regimes by country
7 It should be noted that the dataset we work with derives from two sources: disaggregated data on trade
flows which in principle identifies the regime (eg. Preferential, MFN etc) under which the flow occurred;
and disaggregated tariff data which identifies the applicable tariff. Merging and cleaning the two datasets
is a substantial operation in its own right and has been an important part of this study. More details on this
can be found in the Appendices. 8 Though of course it must be recognized that the preferences are likely to impact on the structure of
trade and that therefore these are endogenous.
22
that we can provide prima facie evidence of the extent to which the different GSP
regimes can facilitate EU trade with developing countries, as well as to examine the
extent to which the GSP product coverage addresses the trade and through this
developmental needs of developing countries.
3. Thirdly, we consider the extent of competitiveness between developing countries, in the
context of the differential preference margins across the EU‘s GSP regimes.
Competitiveness could apply to individual countries becoming more efficient /productive
and therefore competitive over time; or it could apply to countries becoming more
competitive vis-à-vis other countries due to improved preferences. Here the analysis
focuses on the latter interpretation of competitiveness .
4. Fourthly we consider whether there is any prima facie evidence of the extent that the
EU‘s GSP regimes are well directed to countries‘ development needs and if there is any
relationship between preference margins, utilisation rates and selected indicators of
development.
2.1 The structure of the EU‘s GSP regimes In this section we consider the structure of the tariffs preferences in aggregate, across sector
and by country, under the different EU preferential schemes. While the main focus is on the role
of preferences for developing countries, it is interesting and important to first consider the
relative importance of preferential trade for the EU. Table 2.1 below provides a summary of EU
imports by preferential regime. This table is based on actual trade into the EU, using the
underlying 10-digit trade data. Table A1 in Appendix 4 provides similar information but this time
broken down by TDC category (the TDC level is a 21 sector aggregation of the HS
nomenclature).
Table 2.1: EU Imports by Preference Regime
MF
N=
0
MF
N>
0
GS
P=0
GS
P>0
GS
P+0
GS
P+
>0
EB
A=0
EB
A>0
Oth
er
pre
f=0
Oth
er
pre
f>0
Un
kno
wn
Tra
de u
nde
r
ze
ro ta
riff
s
2002 53.06 23.14 2.92 2.12 0.27 0.05 0.28 - 16.82 0.42 0.93 73.75
2003 52.65 23.26 2.86 2.01 0.23 0.05 0.29 0.00 17.36 0.47 0.81 73.39
2004 58.22 22.96 1.75 1.80 0.21 0.06 0.34 0.00 10.99 0.42 3.25 71.51
2005 61.70 23.14 1.59 1.89 0.29 0.05 0.33 0.00 8.47 0.32 2.21 72.38
2006 62.25 24.08 1.48 1.90 0.31 0.04 0.38 0.00 7.33 0.27 1.97 71.75
2007 61.21 24.20 1.79 1.95 0.34 0.04 0.35 0.00 8.18 0.23 1.71 71.87
2008 62.67 23.34 2.09 2.09 0.41 0.05 0.46 0.00 7.71 0.22 0.97 73.34
Source: own calculations based on TARIC data supplied by the European Commission
From Table 2.1 we can see that the importance of "preferences" in total EU imports is low. In
2008 86.01 percent of EU imports entered under MFN arrangements and of this just over
60 percent entered duty free. GSP, GSP+ and EBA account for 4.18 percent, 0.46 percent and
23
0.46 percent of total EU imports respectively. The remaining imports into the EU therefore enter
either via other preferential arrangements such as RTAs, or cannot be classified.
If we look at this by sector (see Table A1 Appendix 4) there are four sectors where trade
entering under all the GSP regimes constitutes more than 20 percent of total EU imports. These
are Footwear (28.43 percent), animal or veg. fats (27.9 percent), live animals (22.81 percent),
raw hides (20.47 percent). There are 4 sectors where they account for more than 10 percent of
EU imports, which are clothing (19.62 percent), plastics (14.22 percent), prepared foodstuff
(12.96 percent), textiles (11.14 percent). EBA preferences also play an important role in Section
XIb, where 6.96 percent of total imports entered under the EBA regime. Note that this is the
share of trade entering under the ―GSP‖ regimes, not the share of trade accounted for by ―GSP‖
countries – as many of these also export under MFN. By adding the columns where tariff are
equal to zero (MFN=0, GSP=0, GSP+=0, EBA=0 and Pref=0) it is also possible to have the
share of trade that enters the EU paying a zero tariff, present in the last column of the table.
We can also see that Sections III and XIb have the smallest share of trade under zero tariffs,
while Sections V, X and XXI are those with the largest share. Finally, it can be seen that other
preferences have an important role in different sectors than the GSP regime. For example, in
Agriculture and Food products (Sections I, II, III and IV) the GSP regime has little importance
while other preferences constitute the main preferential channel of import. This can probably be
explained by the fact that these preferential regimes are more comprehensive, implying that the
EU is liberalising more of their sensitive products, while in the GSP (with the exception of the
EBA) regime there are still several products (particularly in agriculture) protected by the EU.
When taken together with the earlier bullet point on preference margins, this suggests that other
than plastics, trade using ―GSP‖ preferences tends to have a bigger share of the EU market in
those TDC sectors where the preference margin is higher. This could either be because the
preference margin is effective in giving the countries access to the EU, or it could be that these
are sectors in which these countries have an underlying comparative advantage. Further,
where there is ―GSP‖ access at a zero tariff there is no further scope for additional preference
reductions – though there may be scope for facilitating use of preferences, for example through
less restrictive RoRs or simplified administrative procedures. In this context it is worth noting
that for 10 of the 22 TDC sectors, across all the ―GSP‖ regimes, more than 50 percent of
imports paid either an MFN or ―GSP‖ tariff, and for two sectors - footwear, and clothing - over
75 percent of imports paid a tariff. These are therefore sectors where there is potentially scope
for improving the degree of preferential access.
In Table 2.2 below we examine the preferences by the number of tariff lines across the EU‘s
preferential regimes under the enabling clause. In the table we focus on the key differences
between the MFN regime and the GSP, GSP+ and EBA regimes. The table details the level and
type of access by tariff line (at 10 digits) for each of the preferential regimes.
Not surprisingly, the difference between the GSP and GSP+ regimes is smaller than that
between the GSP and EBA regime. Under GSP there are 4781 additional duty free tariff lines,
24
under GSP+ there are 9717, and under EBA 11053. The number of MFN greater than zero lines
is similar between the GSP and GSP+ regimes, but much less so with regard to EBA. On closer
examination of the differences between the GSP and GSP+ many of these differences occur in
textiles and clothing products. This could have an important impact on some developing
countries for which these sectors represent an important share of their total exports. (e.g. Sri
Lanka and Pakistan). This suggests that a country that is highly concentrated in the textiles and
clothing industries is likely to benefit considerably more from GSP+ preferences than from GSP
preferences.
Table 2.2: Coverage of EU Preferential Regimes 2008
GSP GSP+ EBA GSP GSP+ EBA
MFN = 0 3152 3152 3152 22.1 percent 22.1 percent 22.1 percent
MFN > 0 1187 1089 49 8.3 percent 7.6 percent 0.3 percent
Duty Free 4781 9717 11053 33.5 percent 68.1 percent 77.5 percent
Positive pref. tariff 5139 301 5 36.0 percent 2.1 percent 0.0 percent
Total 14259 14259 14259 100.0 percent 100.0 percent 100.0 percent
Source: own calculations at 10-digits from TARIC
*these lines are preferential specific tariffs for sugar
It is also interesting to see how the preferential regimes have evolved in time. In Table 2.3 we
can identify this by looking at the share of tariff lines across the different regimes and tariff types
for the years 2002, 2005 and 2008. In covering these years, we take into consideration the
different revisions of the GSP regimes.9 From this table we see how the share of tariff lines
falling under the category of duty free MFN has increased from 2002 to 2008 by 5.6 percentage
points. Looking at the GSP regime in particular, the share of tariff lines that faced a positive
MFN tariff has decreased by 4 percentage points during the same period where there seems to
have been a move towards more preferential duty free treatment. Where the GSP+ is
concerned we see a milder reduction in lines facing a positive MFN and an increase in those
being granted duty free access under this regime. The EBA regime remains largely unchanged,
but it is worthwhile noting that the increase of tariff lines falling under the MFN zero category
since 2002 comes at a loss of preferential margins for EBA countries.
9 Note that we work with shares in this table as opposed to tariff counts as there is a heterogenous
amount of tariff lines reported in the EU‘s tariff database. Hence in 2002, the total amount of tariff lines is
14,155, whilst in 2005 this number decreases to 13,990, only to increase again in 2008 to 14,259.
25
Table 2.3: Coverage of EU Preferential Regimes 2002-08 (share of tariff lines)
2002 2005 2008
GSP GSP+** EBA GSP GSP+ EBA GSP GSP+ EBA
MFN = 0 16.45 16.45 16.45 22.07 22.07 22.07 22.11 22.11 22.11
MFN > 0 12.34 8.12 0.23 11.89 7.78 0.21 8.32 7.64 0.34
Pref. Duty Free 37.11 72.14 83.32 32.57 67.32 77.71 33.53 68.15 77.52
Positive pref. tariff 34.10 3.29 0.00 33.47 2.84 0.01 36.04 2.11 0.04
Source: own calculations at 10-digits from TARIC
** GSP+ here refers to the special arrangement for drug trafficking prevention
Where Table 2.1 identifies the number of tariff lines by type of access across the different
regimes, Table 2.4 distinguishes between the different regimes by the height of the tariff faced,
where once again we compare 2002 with 2008. This allows us to consider the difference
between the actual degrees of preferences granted under the different regimes.
Table 2.4: Share of Tariff Lines by Regime and Size of Tariff (2008)
2002 2008 Change
MFN GSP GSP+ EBA MFN GSP GSP+ EBA MFN GSP GSP+ EBA
Tariff = 0 16.45 53.56 88.58 99.77 22.21 55.73 90.28 99.62 5.77 2.17 1.70 -0.14
Tariff 0<t≤5 34.37 17.27 2.60 0.18 28.21 18.52 2.01 0.23 -6.16 1.25 -0.59 0.05
Tariff 5<t≤10 26.65 14.40 1.12 0.00 29.11 13.38 0.72 0.06 2.47 -1.02 -0.40 0.06
Tariff 10<t≤15 9.38 4.90 0.71 0.01 8.56 4.09 0.61 0.03 -0.82 -0.82 -0.10 0.02
Tariff 15<t 9.04 5.86 3.13 0.04 7.64 4.27 2.52 0.06 -1.40 -1.60 -0.61 0.01
specific non ad
valorised 4.12 4.00 3.86 0.01 4.27 4.01 3.86 0.01 0.15 0.01 0.00 0.00
Source: own calculations based on TARIC data supplied by the European Commission
The table shows the distribution of tariffs across the EU‘s current preferential regimes, where at
the 10-digit level we count the number of tariff lines that are: zero; between 0 and 5, between 5
and 10; between 10 and 15; and above 15. In each case, we provide the share of total tariff
lines that are in each identified category. In so doing the table summarises the different degrees
of preference accorded by the different regimes in 2008, and thus helps to identify the potential
importance of improved access. In 2008, over 22 percent of tariff lines under the MFN regime
were duty free whilst the majority of tariff lines (just below 50 percent) were within the range of
zero to 5 percent. Comparing this to the GSP regime in that year we see that there are 33
percentage points more tariff lines awarded duty free access under the GSP regime. Where we
compare the latter to the GSP+, the table shows us that a further 34 percentage points
separate the GSP from the GSP+ duty free concession with the differences between the GSP+
and the EBA regimes being much lower (a further 9 percentage points). The table shows us
the marked differences between the GSP and the GSP+ regimes in terms of duty free treatment
and height of tariff. Turning to the evolution of tariffs, we see that MFN and GSP tariffs have
seen the larger reductions over time, while the GSP+ and EBA regimes have changed litt le.
26
Tables 2.1 and 2.4 provide an overall assessment of differences between the three preferential
regimes, and between them and MFN access. In Tables 2.5 and 2.6 we explore this in more
detail by looking at both sectoral differences in average tariffs and in preference margins.
Table 2.5 looks at the average tariff by regime according to the TDC sector classification in the
years 2002 and 2008. In general, it can be seen a reduction between 2002 and 2008 on the
average tariffs in every sector. The decrease in tariffs is observed, in general, in all tariffs
regimes. However, in some sectors (sectors III and IV) the MFN tariffs, on average, have
increased with the correspondent co-movement of the respective preferential regimes (GSP and
GSP+). Nevertheless, the preferential tariffs do not need to have the same proportional change
than the MFN tariff. Therefore, the difference between the MFN tariff and the tariff applied in
each preferential regime will not remain constant.
Table 2.5: Average Tariff by Regime and TDC Sector (2002 and 2008)
2002 2008
TDC Description MFN GSP GSP+ EBA MFN GSP GSP+ EBA
I Live animals; animal products 20.6 19.1 14.5 0.0 17.3 14.8 10.6 0.0
II Vegetable products 12.4 10.0 7.4 0.2 9.7 7.7 4.7 0.3
III Animal or vegetable fats and oils 7.3 4.3 1.4 0.0 8.6 5.3 2.1 0.0
IV Prepared foodstuffs; 16.1 12.6 2.2 0.3 17.3 11.8 2.5 0.3
V Mineral products 0.7 0.1 0.1 0.0 0.7 0.0 0.0 0.0
VI Products of the chem.& allied inds 5.0 0.9 0.3 0.0 5.1 0.9 0.2 0.0
VII Plastics and Articles thereof 5.9 1.4 0.0 0.0 5.5 1.1 0.0 0.0
VIII Raw hides and skins, leather, furskins 2.9 0.8 0.2 0.0 3.0 0.9 0.2 0.0
IX Wood and articles of wood 2.8 0.9 0.0 0.0 2.4 0.6 0.0 0.0
X Pulp of wood or other fibrous... 1.5 0.0 0.0 0.0 0.0 0.0 0.0 0.0
Xia Textiles 6.7 5.4 0.0 0.0 6.2 5.0 0.0 0.0
XIb Textile articles (clothing) 11.5 9.2 0.0 0.0 11.2 9.0 0.0 0.0
XII Footwear, headgear, umbrellas... 8.3 4.6 0.0 0.0 7.6 4.0 0.0 0.0
XIII Articles of stone, plaster, cement... 4.0 1.3 0.0 0.0 4.0 1.3 0.0 0.0
XIV Pearls, precious, semi-precious stones 0.8 0.0 0.0 0.0 0.7 0.0 0.0 0.0
XV Base metals and articles of base metal 2.4 0.5 0.1 0.0 2.0 0.5 0.1 0.0
XVI Machinery and mechanical appliances 2.4 0.4 0.0 0.0 2.3 0.3 0.0 0.0
XVII Vehicles, aircraft, vessels, transport 5.1 2.1 0.0 0.0 4.6 1.7 0.0 0.0
XVIII Optical, photographic... Instruments 2.4 0.2 0.0 0.0 2.3 0.2 0.0 0.0
XIX Arms and ammunition; 2.3 2.3 2.3 2.3 2.2 2.2 2.2 2.2
XX Miscellaneous manufactured articles 2.6 0.1 0.0 0.0 2.5 0.1 0.0 0.0
XXI Works of Art, collectors' piece... 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
Source: own calculations based on TARIC data supplied by the European Commission
The extent of the preference margins by sector can then be seen in Table 2.6. Here we report
on the preference margin as compared to MFN tariffs for 2002 and 2008, as well as giving the
change over time. Since the EBA regime is an integral part of the GSP regime, the requirements
to be met (apart of the development conditions) are similar for both regimes. Therefore, if a
particular country cannot meet the administrative requirements, for example, of the EBA regime,
it is likely that it cannot meet the GSP requirements; furthermore, in the absence of any other
preferential regime, such as ACP preferences or a bilateral trade agreement, the appropriate
27
comparator should be the MFN tariff since it will be the only alternative regime available. This
analysis therefore identifies the relative importance of preferential access across different
sectors.
Table 2.6: Preference Margins by TDC Sector Compared to MFN (2002 & 2008)
2002 2008 change in pref margin
TDC Description GSP GSP+ EBA GSP GSP+ EBA GSP GSP+ EBA
I Live animals; animal products 1.44 6.12 20.57 2.55 6.71 17.32 1.11 0.59 -3.25
II Vegetable products 2.33 5.00 12.22 2.01 5.01 9.4 -0.33 0.02 -2.82
III Animal or vegetable fats and oils 2.94 5.85 7.29 3.3 6.5 8.64 0.36 0.65 1.35
IV Prepared foodstuffs; 3.53 13.92 15.82 5.56 14.82 17.01 2.03 0.91 1.18
V Mineral products 0.67 0.67 0.73 0.69 0.69 0.74 0.02 0.02 0.01
VI Products of the chem... & allied inds 4.05 4.70 4.99 4.21 4.89 5.14 0.16 0.2 0.15
VII Plastics and Articles thereof 4.53 5.93 5.93 4.36 5.49 5.49 -0.17 -0.44 -0.44
VIII Raw hides and skins, leather, furskins 2.1 2.74 2.94 2.16 2.83 3.04 0.06 0.09 0.1
IX Wood and articles of wood 1.94 2.81 2.81 1.83 2.43 2.43 -0.11 -0.38 -0.38
X Pulp of wood or other fibrous... 1.48 1.48 1.48 0 0 0 -1.48 -1.48 -1.48
Xia Textiles 1.38 6.73 6.74 1.25 6.22 6.24 -0.13 -0.51 -0.51
XIb Textile articles (clothing) 2.31 11.49 11.49 2.24 11.2 11.2 -0.07 -0.29 -0.29
XII Footwear, headgear, umbrellas... 3.71 8.31 8.31 3.56 7.59 7.59 -0.15 -0.72 -0.72
XIII Articles of stone, plaster, cement,... 2.65 4.00 4.00 2.62 3.95 3.95 -0.03 -0.04 -0.04
XIV Pearls, precious, semi-precious stones 0.84 0.84 0.84 0.74 0.74 0.74 -0.1 -0.1 -0.1
XV Base metals and articles of base metal 1.94 2.31 2.43 1.51 1.9 2.02 -0.43 -0.41 -0.41
XVI Machinery and mechanical appliances 2.01 2.4 2.4 1.95 2.27 2.27 -0.06 -0.13 -0.13
XVII Vehicles, aircraft, vessels, transport 2.93 5.05 5.05 2.88 4.62 4.62 -0.05 -0.43 -0.43
XVIII Optical, photographic,... Instruments 2.23 2.42 2.42 2.09 2.27 2.27 -0.14 -0.15 -0.15
XIX Arms and ammunition; 0 0 0 0 0 0 0 0 0
XX Miscellaneous manufactured articles 2.51 2.61 2.61 2.38 2.49 2.49 -0.13 -0.12 -0.12
XXI Works of Art, collectors' piece... 0 0 0 0 0 0 0 0 0
Perhaps unsurprisingly, tariffs are highest in agriculture and foodstuffs (TDC sectors I – IV),
followed by textiles, clothing and footwear (TDC sectors XI and XII). In all other sectors average
tariffs are low (on average less than 5 percent). Using data on exports to the EU at the 10-digit
level, but then aggregating up to the TDC level, we see that for GSP preferences the average
un-weighted preference margin is less than 5 percent for all TDC sectors except prepared
foodstuffs (5.56 percent). This is low and therefore on average a priori one might not expect
GSP preferences to have a big impact on trade. It is important to remark that the overall
preferential margin for sensitive products under the GSP regime is 3.5 percentage points off the
MFN regime10, and this drives these small preference margins. For GSP+ countries, relative to
the GSP regime the biggest preference margins are in live animals (4.16 percent), prepared
10
Some exceptions apply. Particularly for non-ad valorem tariffs there is a particular treatment that could
yield slightly different preference margins.
28
foodstuffs (9.27 percent), textiles (4.97 percent) and clothing (8.95 percent); for EBA countries
relative to GSP the sectors with the biggest preference margins are: live animals
(14.77 percent), vegetable products (7.39 percent), animal or veg fats (5.34 percent),
prepared foodstuffs (11.45 percent), textiles (4.97 percent) and clothing (8.95 percent). On
average, it is really only on agriculture and processed foods, and textiles and clothing that there
is much scope for improved preferential access, and by and large this really only applies to the
GSP countries, as these preferences are already being offered to the GSP+ and EBA countries.
It is also worth noting that in most sectors there has been a decline in preference margins -
especially for live animals and vegetable products for the EBA countries, explained mainly by
the reduction in the MFN tariff. This is important because it makes very clear that unless there
are high tariff peaks in other sectors, the scope for offering significant preference margins is
limited to these specific sectors.
Figure 2.1 below explores this issue of tariff peaks and their possible significance consider. The
information is analogous to that in Table 2.6 above, but here we consider the incidence of
preference margins at the 10-digit level across different preference margin boundaries by TDC
sector. Hence, if you take the top left panel ―live animals, animal products‖, here we show for
how many 10-digit products is the preference margin in comparison to the MFN tariff between 0-
5 percent, 5-10 percent, 10-15 percent, 15-20 percent and greater than 20 percent. We do this
for each of the three regimes: GSP, GSP+ and EBA.
For this sector we therefore see that for countries entering under the GSP regime the
preference margin is less than 5 percent in just over 500 cases, and between 5 percent and
10 percent in just under 100 cases. In contrast, the preference margins are much more
significant for the EBA regime. There are only 67 10-digit products with a margin of less than
5 percent, 255 with a margin of between 5 percent and 10 percent, 278 with a margin between
10 percent and 15 percent, and then 104 and 173 for the remaining two categories.
The information contained here is useful in two regards. First, for any given regime – GSP,
GSP+ and EBA – it shows how significant the margins are as well as identifying sectors where
there are tariff peaks which may be playing a role. Secondly, as the EBA regime offers tariff free
access on almost all products this is a useful benchmark or counterfactual against which to
compare the level of preferential access offered by the GSP and GSP+ regimes. Again, taking
the example of ―live animals, animal products‖, currently under the GSP regime there are almost
no products for whom the preference margin is greater than 10 percent; whereas under the
EBA regime there are 173 products with a margin of greater than 20 percent. It would therefore
be possible to improve the degree of preferential access offered to the GSP countries in these
products, though of course it has to be recognised that this could then be at the expense of
exporters in the EBA countries.
29
Figure 2.1: Incidence of Preference Margins at 10-digit level
30
It is then interesting to consider the difference across sectors. What emerges quite clearly is that
higher tariffs / tariff peaks are primarily significant again in TDC sectors 1-IV, and to some extent
31
in clothing (XIb) and footwear (XII). In many sectors the preference margin is almost entirely
less than 5 percent, and in Chemical products (VI), Plastics (VII), and textiles, there are
significant numbers of products with preference margins of between 5 percent-10 percent. This
therefore serves to reinforce the message given above. The structure of the EU‘s preference
regimes is such that both with regard to average tariffs and with regard to tariffs peaks and
preference margins, the scope for offering significant preferential access for developing
countries is largely limited to a few sectors.
This in turn may have important implications for the incentives in given developing countries
with regard to the orientation of the structure of their exports. If preferences are effective in
impacting positively on countries‘ trade then the current preference structure may encourage
countries to develop industries within the sectors where preference margins are more
significant, and/or discourage countries to develop industries with low preference margins.
Either way, it is hard to see how the GSP regimes could impact significantly on the development
of exporting capacity in all those sectors where the margins are low, and where there are no
important tariff peaks.
2.1.2 Section Summary
This section identified the de jure degree of preferential access granted to developing countries,
their differences across the GSP preference regimes, the relative amounts of trade covered
and the impact this could have on competitiveness.
The section examined the structure of the different EU tariff preferences in aggregate by country
and sector, using detailed 10-digit trade data. It highlights that the importance of "preferences"
in total EU imports is low. In 2008, most EU imports (86.01 percent) entered under MFN
arrangements or duty free (60 percent). GSP, GSP+ and EBA account for 4.18 percent,
0.46 percent and 0.46 percent of total EU imports respectively.
There are four sectors where trade entering under all the GSP regimes constitutes more than 20
percent of total EU imports: footwear (28.43 percent), animal or veg. fats (27.9 percent), live
animals (22.81 percent) and raw hides (20.47 percent). On average, it is really only on
agriculture and processed foods, and textiles and clothing that there is much scope for improved
preferential access, and by and large this really only applies to the GSP countries, as these
preferences are already being offered to the GSP+ and EBA countries. Where ―GSP‖ access
has a zero tariff there is no further scope for additional preference reductions. However, there
may be scope for facilitating use of preferences, for example through less restrictive rules of
origin or simplified administrative procedures.
The section concludes that it is unlikely that the GSP regimes could impact significantly on the
development of exporting capacity in those sectors where the margins are low, and where there
are no important tariff peaks. This suggests that the scope for offering significant preferential
access is largely limited to a few sectors.
32
2.2 GSP and developing country exports In this section of the report we consider the relative importance of the GSP regimes for
developing country exports. To do this we consider the information on the structure of the
preferential regimes and link this to developing countries trade with the EU. However, prior to
drawing the connection between the regimes and trade with the EU, it is important to put into
perspective the relative importance of trade with the EU for the developing countries. This can
be seen in Figure 2.2, which shows for each of the three regimes – GSP, GSP+ and EBA, the
relative importance of the EU in total trade. Hence for each country grouping we show, for how
many countries is the EU‘s share in their total trade less than 10 percent, 25 percent,
50 percent, 75 percent and 100 percent respectively.
From this figure we can see that it is only for 9 EBA countries that the EU as a destination
market comprises more than 50 percent of their total exports. For 20 countries the EU
comprises less than 25 percent. For all the GSP+ countries the EU comprises less than
50 percent of their total exports, while for 9 of the 14 countries the EU comprises less than
10 percent. For the GSP the distribution is more even. Nevertheless for 80 countries the EU
comprises less than 50 percent of their total exports, and for 28 countries, less than 10 percent.
This is important because it indicates that in total the EU comprises more than 50 percent of
total exports for only 42 of the 175 countries. Of course, this does not mean that the EU is
therefore an unimportant destination; it is likely to be the principal destination for more countries
than this. Nevertheless, for many countries most of their trade is not with the EU. This puts into
perspective the extent to which preferential trade policy by the EU could impact on trade and
development more generally.
33
Figure 2.2: Distribution of Countries by the Share of the EU in Total Exports
14
6
19
8
1
0
5
10
15
20
<10 <25 <50 <75 <100
EBA
EBA
28
20
32
23
10
0
10
20
30
40
<10 <25 <50 <75 <100
GSP
GSP
9
23
0 00
5
10
<10 <25 <50 <75 <100
GSP+
GSP+
Table 2.7 below provides summary information by grouping countries by regime and looking at
the usage of preferences by these grouping. Table A2 in Appendix 4, then provides analogous
information but this time by country by detailing for each country their share of exports to the EU
under the various regimes. In these tables we also delimit the share of trade that enters via duty
free concessions per regime.
If we look at the importance of preferences by country groupings we see that on average a high
proportion of GSP countries' trade enters under MFN=0. In 2008 64.45 percent of GSP
countries exports to the EU entered the EU with a zero MFN tariff, 61.26 percent of GSP+
countries' exports, and 62.85 percent of EBA countries' exports. It is interesting that we see a
34
big rise in the EBA share of MFN zero trade between 2007 and 2008 from 51.16 percent to
62.85 percent and a corresponding decline in the ―other preferences share‖ from 12.82 percent
to 5.95 percent. While there has been no change over this time period in the MFN=0 number of
tariff lines, this switch most probably connected to the ending of Cotonou related preferences,
and thus to changes in the regime of entry applied for, although a priori, one would expect there
to be a decline in the other preferences share as a result of the ending of Cotonou, but with a
rise in the EBA=0 share, as opposed to a rise in the MFN=0 share. It is also interesting to see
that the total share of exports to the EU which enter duty free is almost identical across
preference regimes.
We also see that the shares of trade paying a positive MFN tariff for the GSP, GSP+ and EBA
countries were 22.07 percent, 13.18 percent and 6.08 percent respectively. By and large these
shares have been rising over time. This suggest that it is here that there is potentially more
scope for improved access to the EU, either in terms of improving the preferences or to the
extent that this reflects non-utilisation, the take up of these preferences. It is interesting that
while in principle almost all EBA countries‘ trade could be duty free, tariffs are in fact paid on
over 6 percent of this trade. This is unlikely to be driven by the few exceptions to the EBA
regime and suggests that there are some issues of non-utilisation here. On average only just
over 7 percent of GSP countries' exports used GSP preferences in exporting to the EU. For the
GSP+ and the EBA countries this was just over 24.5 percent and 23.4 percent respectively.
Both the GSP countries and the EBA countries also exported just over 5 percent of their trade
using other preference regimes.
35
Table 2.7: Preference Usage by Regime Type 2002-2008
Year type MF
N=
0
MF
N>
0
GS
P=
0
GS
P>
0
GS
P+
=0
GS
P+
>0
EB
A=
0
EB
A>
0
Oth
er
pre
f=0
Oth
er
pre
f>0
Un
kn
ow
n
Total
imports (in
millions of
Euros)
2002
EBA
50.93
14.71
-
-
-
-
18.16
-
14.51
0.03
1.66
13,468.01
GSP
56.31
21.99
8.15
5.91
-
-
-
-
6.13
0.50
1.01
317,896.09
GSP+
60.01
18.28
-
-
17.37
2.99
-
-
0.00
0.00
1.35
13,542.93
OTHER
51.04
24.14
-
-
-
-
-
-
23.58
0.39
0.85
542,758.35
2003
EBA
45.61
20.46
-
-
-
-
20.54
0.00
13.03
0.02
0.33
12,480.80
GSP
57.22
22.36
7.58
5.34
-
-
-
-
5.81
0.53
1.17
339,056.78
GSP+
57.44
21.13
-
-
16.77
3.97
-
-
-
0.01
0.69
12,318.77
OTHER
49.84
23.99
-
-
-
-
-
-
25.11
0.46
0.60
533,657.94
2004
EBA
42.72
15.36
-
-
-
-
23.32
0.00
10.00
0.03
8.57
13,498.60
GSP
61.36
21.77
4.07
4.17
-
-
-
-
4.37
0.52
3.76
396,041.13
GSP+
59.74
19.74
-
-
13.47
4.13
-
-
-
0.00
2.92
14,089.07
OTHER
56.01
24.27
-
-
-
-
-
-
16.64
0.37
2.71
494,520.27
2005
EBA
50.88
11.42
-
-
-
-
21.15
0.00
10.03
0.03
6.49
15,623.96
GSP
64.01
22.76
3.27
3.88
-
-
-
-
3.76
0.42
1.90
492,109.15
GSP+
61.77
14.59
-
-
17.64
3.21
-
-
-
-
2.79
16,837.18
OTHER
59.72
24.21
-
-
-
-
-
-
13.48
0.25
2.34
487,302.58
2006
EBA
43.97
10.45
-
-
-
-
26.11
0.00
11.92
0.02
7.54
16,850.28
GSP
65.13
21.82
2.96
3.80
-
-
-
-
3.92
0.33
2.04
581,915.14
GSP+
65.25
12.32
-
-
17.27
2.32
-
-
-
-
2.84
21,148.34
OTHER
59.53
27.46
-
-
-
-
-
-
11.13
0.22
1.67
542,617.20
2007
EBA
51.16
8.39
-
-
-
-
22.24
0.00
12.82
0.02
5.35
19,098.54
GSP
63.00
23.38
3.42
3.74
-
-
-
-
4.53
0.27
1.65
628,619.71
GSP+
62.14
13.91
-
-
19.00
2.12
-
-
-
-
2.84
21,565.48
OTHER
36
Year type MF
N=
0
MF
N>
0
GS
P=
0
GS
P>
0
GS
P+
=0
GS
P+
>0
EB
A=
0
EB
A>
0
Oth
er
pre
f=0
Oth
er
pre
f>0
Un
kn
ow
n
Total
imports (in
millions of
Euros)
59.38 26.23 - - - - - - 12.61 0.19 1.60 533,822.34
2008
EBA
62.85
6.08
-
-
-
-
23.40
0.01
5.95
0.00
1.71
24,342.41
GSP
64.45
22.07
3.84
3.86
-
-
-
-
4.71
0.27
0.80
679,585.68
GSP+
61.26
13.18
-
-
22.16
2.42
-
-
-
-
0.99
23,344.49
OTHER
60.31
26.34
-
-
-
-
-
-
12.04
0.17
1.15
523,975.63
Source: own calculations based on TARIC data supplied by the European Commission
All this indicates that on average the preference regimes do not appear to account for a lot of
the relevant countries trade with the EU. Once again this would suggest that, on average, the
structure of the GSP regimes may not be well directed towards the export needs of developing
countries. This is even more the case if we consider their share of total trade, as opposed to
solely their trade with the EU. Of course, this may also suggest that with low MFN tariffs and
relatively few tariff peaks, the extent to which bilateral preference regimes can help developing
countries is, in principle, limited.
This could be either because preferences which are on offer are not being utilised for some
reason, or that the countries‘ export structures is such that they get access to the EU with
MFN=zero tariffs anyway. If it is the former then this suggests this could be because the
preference margins are too small to take advantage of and/or that the costs of complying with
the preference regime (e.g. rules of origin) are too high. If is the latter, then this would suggest
that the preference regimes are not well targeted to the export profiles, or to the underlying
comparative advantages, of the ―GSP‖ countries.
The figures given above were averages across all GSP countries and are likely to mask
considerable differences between countries. Hence, it is important to unpick this. Table A2 in
Appendix 4 provides analogous information to that in Table 2.7 above by country, where the
interested reader can examine the information for individual countries. Table A3 in Appendix 4
follows the same structure but where we consider the importance of each countries trade with
the EU by category, but this time as a share of their total trade.11 For example, for Afghanistan it
can be see that a little more than 20 percent of its exports are destined for the EU; and
19 percent of its total exports in 2008 paid a zero MFN tariff to the EU.
11
Table A3 indicates the importance of the different types of EU preferential regimes in total exports Here it s important to note total exports were obtained from a different source (UN Comtrade), and therefore there could be some inconsistencies with the data provided by the EC. Data in Comtrade is expressed in US dollars and we have made an adjustment by using the Euro/US dollar 2008 average exchange rate. We have also used mirror flows (the imports declared from every country in the world from each country), which might generate some underestimation of exports.
37
In contrast, the EU accounts for nearly 46 percent of the total trade of the Maldives. If we look at
the GSP+ countries we see that, with the notable exception of Sri Lanka, little more than
5 percent of these countries‘ exports entered the EU using the GSP+ preference. Pakistan is
also interesting. It can be seen that around 21 percent of its total exports have the EU as
destination and have entered paying a positive GSP tariff; higher than any of the other GSP
countries. This is explained by the importance of textiles and garments in Pakistan trade pattern
that under the GSP regime, have a positive GSP tariff.
The next section provides a more detailed consideration of each regime – GSP, GSP+ and
EBA.
2.2.1 EBA countries For 16 of the 49 EBA countries over 90 percent of their exports to the EU enter under MFN=012;
and for a further seven13 this applies to over 75 percent of their exports to the EU (see Table
A2). For these countries it seems less likely that the EBA preferences they have available to
them can make a substantial difference to their trade given that much of what they export is duty
free.
Of the remaining 26 countries, for 714 of them over 75 percent of their exports to the EU are
exported under EBA zero tariffs, and for a further one (Mozambique) if we also include zero
tariffs from some other preferential arrangement (EPA in this case). This also raises the issue of
why countries are using other preferences as opposed to EBA. This could be related to
administrative procedures or issues to do with rules of origin. As these 8 countries appear to be
using preferences to a substantial degree, it suggests that these preferences matter to these
countries, to some degree at least.
However, this leaves 18 countries. For nine of them15 given the structure of their actual trade at
the 10 digit level, the (hypothetical) average weighted MFN that these countries would have
paid if all their exports had entered via the MFN regime would have been less than 5 percent
(see Table A2 in Appendix 4). It is also interesting to note that three of these - Niger, Samoa,
and Tuvulu - pay MFN tariffs for over 50 percent of their trade with the EU. This might suggests
that given these nine countries‘ export structures, the preference margin afforded by the EBA
regime may be insufficiently important for the countries to make use of that margin.
This leaves nine countries. Five of these paid MFN tariffs on more than 10 percent of their
exports to the EU: Bhutan (60.5 percent), Cambodia (21.56 percent), Djibouti (12.51 percent),
Haiti (17.23 percent), Malawi (12.39 percent) – which of course again raises questions as to
12
Myanmar is a least developed country but which is excluded from the list of EBA since 1997 beneficiaries on grounds of poor human rights. This includes Cape Verde, which has graduated from EBA but with a 3 year transition period. 13
Guinea Bissau, Kiribati, Mali, Mauritania, Sudan, Togo and Zambia. 14
Bangladesh, Cape Verde, Laos, Maldives, Nepal, Vanuatu and Yemen 15
Benin, Eritrea, Ethiopia, Gambia, Comoros, Niger, Tuvalu, Uganda and Samoa.
38
why the EBA preferences are not being utilised. Of the remaining four countries there is either a
substantial amount of trade where the regime of entry is unknown (Soloman Islands) or they
fairly extensively use ―other preferences‖ (Madagascar, Tanzania).
In order to understand this further we consider the information we have on the average
preference margin across all the EBA countries (Appendix 4 Table A4). This can be seen in
Figure 2.3 below, where we take the underlying 10 digit export data and on that basis compute
the average tariff that would apply for each country if they were exporting on an MFN basis, and
then we compare this to the equivalent EBA tariff that would apply on that trade.
Figure 2.3: EBA – Preference Margins
25
75
65
1
0
5
10
15
20
25
30
<2.5% <5% <7.5% <10% <15% >15%
No
of
Co
un
trie
s
This tells us that out of the 49 countries for 25 of them, given their exporting structure, the
preference margin affored to them by the EBA regime is less than 2.5 percent, and for a further
7 countries the margin is less than 5 percent. Thus, given their exporting structures, the
preference margin is greater than 5 percent for only 17 countries, and greater than 10 percent
for only six countries. Once again this would suggest that the EBA regime provides significant
preference margins only to a relatively small subset of countries.
Consider also Figure 2.4 below, which gives the number of EBA countries for whom the share of
MFN>0 trade with the EU is less than 10 percent, between 10 percent and 25 percent;
25 percent and 50 percent, and so on. We see that for 38 of the EBA countries they pay MFN
tariffs on less than 10 percent of their trade with the EU; and there are only four countries where
MFN tariffs are paid for over 75 percent of their trade.
39
Figure 2.4: EBA Countries: Share of MFN>0 Trade
2.2.2 GSP+ countries:
For two of the GSP+ countries (Mongolia, Venezuela) over 75 percent of their exports entered
the EU under MFN duty free. Of the remaining countries, only one (Sri Lanka) had more than
50 percent of their exports to the EU exported under GSP+ zero tariffs; if we include exports
which paid a positive GSP+ tariff then there is only one more country (Ecuador) which exported
more than 50 percent of its exports to the EU under the GSP+ regime.
For five of the remaining countries (Bolivia, Georgia, Guatamala, Honduras and Peru) less than
10 percent of their exports paid positive MFN tariffs. This leaves five countries that pay positive
MFN tariffs on more than 10 percent of their exports, and where their GSP+ exports are less
than 50 percent: Columbia, Costa Rica, El Salvador, Nicaragua and Panama. The case of the
Central American countries is quite interesting here. They typically have a relatively well
developed textile industry which could potentially make use of GSP+ preferences. In reality they
primarily export to the US under the CAFTA and typically have a low share of exports going to
the EU.
If we link this to the extent of the preference margins we see that for all of these (except El
Salvador), that the average preference margin is less than 2 percent. On the one hand this
suggests that there may be little benefit from these countries utilising the preferences on offer.
However, on the other hand as discussed later in the report we show that preferences, even
where the margins are low are nevertheless still utilised. If the preferences are utilised, this is
presumably because they are useful for the countries concerned. This would therefore seem to
suggest, that the preferences are useful (as evidenced by their utilisation), but the lower the
margin the lower the benefit to the countries concerned. The large number of cases with a low
margin implies a lower net benefit.
40
The information on preference margins can be seen in Figure 2.5 below, which is once again
based on each country‘s trade at the 10 digit level for 2008. For each country there are three
columns. In the first column we give the actual weighted average tariff that each country paid on
their exports to the EU. In the next column (Hyp-margin) we compute the difference between
the average tariff that would apply for each country if they were exporting on an MFN basis and
the equivalent GSP+ tariff that would apply on that trade. While this is based on actual trade the
GSP+ tariff here is calculated assuming that the GSP+ regime applies to all of the exports, and
in that sense is a hypothetical tariff as in reality there may be some exclusions. The third column
(Act-margin) then gives the difference between the MFN tariff and the tariff that each country
actually paid, and thus gives a sense of the actual preference margin.
Figure 2.5: GSP+: Tariffs and MFN Preference Margins
0
2
4
6
8
10
12
Actual HYP-margin Act-margin
The figure is interesting on several counts. First, we see that the actual tariff paid is for all but
three countries (Columbia, Costa Rica, and Ecuador) less than 5 percent. Secondly we see a
divergence between the actual and the hypothetical tariff margins. This could occur for two
reasons – partly because of non-utilisation, and partly because some countries have additional
preferences / exemptions. Finally we learn that, except for Ecuador, El Salvador and Sri Lanka,
the hypothetical and actual preference margin for all countries is extremely low – around
3 percent or less. This suggests that for the majority of the GSP+ countries the benefits of the
regime in comparison to the MFN regime, in terms of the margin of preference, are relatively
low, though this does not preclude relatively high rates of utilisation.
The appropriate comparison however, is not necessarily with the MFN exporters, but with the
alternative regime these countries could be under. Figure 2.6 below therefore considers the
hypothetical margin for each country with respect to both the GSP regime and the EBA regime.
41
The first column therefore gives an indication of the additional preference the GSP+ countries
receive in comparison to what their tariffs would be if they exported under the GSP regime.
Once again we see that as with the MFN comparison for all but three countries (Ecuador, El
Salvador, and Sri Lanka) the average preference margin, and by extension the additional
benefit is very low. It is then interesting to consider, again given the GSP+ countries‘ export
structures, the extent to which the EBA regime offers better market access. Here we see that in
most cases once again there is little difference, except for Columbia, Costa Rica and Ecuador.
Figure 2.6: GSP+: GSP and EBA Margins
0
2
4
6
8
10
12
GSP-margin EBA-margin
Where the preceding figure compared the tariffs across the different regimes for the GSP+
countries, Figure 2.7 below compares the shares of trade under the different regimes (Appendix
4, Table A5). What we do here is to take the existing structure of trade for each of the countries
at the 10 digit level, and explore how much more (or less) trade would be duty free across the
different trade regimes. These are hypothetical calculations in the sense that we take the
existing structure of each country‘s trade but then assume that all trade that is eligible for duty
free access enters under each of the respective regimes – MFN, GSP, GSP+ and EBA. Not
surprisingly we see that any improvement in duty free access as given to the EBA countries
would increase the share of eligible trade but in only 5 cases (Colombia, Costa Rica, Ecuador,
Nicaragua and Panama) by more than 15 percent. We also see that in comparison to GSP duty
free access, GSP+ gives eight countries a more than 15 percent increase in the share of trade
eligible for duty free access; and for all but two countries (Panama and Venezuela) a more than
15 percent increase in comparison to MFN access.
42
Figure 2.7: GSP+: Change in Share of Trade Eligible for Duty Free Access
-100
-80
-60
-40
-20
0
20
40
60
MFN EBA GSP
Finally Figure 2.8 gives number of GSP+ countries for whom the share of trade under MFN>0
with the EU is less than 10 percent, between 10 percent and 25 percent; 25 percent and
50 percent, and so on. We see that five of the GSP countries pay positive MFN tariffs on less
than 10 percent of their trade with the EU; there are eight countries paying positive MFN tariffs
on between 10-25 percent of their trade, and only one country (Ecuador) paying positive MFN
tariffs on more than 25 percent of its trade.
43
Figure 2.8: GSP+ Countries: Share of MFN>0 Trade
2.2.3 GSP countries:
There are 114 GSP countries in the data. Of these there are 20 countries where over 90 percent
of their exports entered the EU MFN duty free; and a further 13 where over 75 percent of their
exports entered the EU MFN duty free. Of the remaining GSP countries, there is only one
country (Armenia) where GSP duty free exports account for more than 50 percent of exports to
the EU; but if you add in other preferences (e.g. EPA, RTAs), then there are a further 22
countries. There are two further countries where their share of total GSP exports is greater than
50 percent and these are Pakistan and India. We also see that there are 15 countries where
less than 10 percent of their exports paid a positive MFN tariff. This leaves 44 countries that
paid a positive MFN tariff on more than 10 percent of their exports.
Another way of thinking about these calculations is to consider the share of trade which pays a
positive MFN tariff, and looking at the distribution across countries. This is done in the figure
below:
44
Figure 2.9: GSP Countries: Share of MFN>0 Trade
Figure 2.9 gives the number of GSP countries for whom the share of MFN>0 trade with the EU
is less than 10 percent, between 10 percent and 25 percent; 25 percent and 50 percent, and so
on. We see that for 64 of the GSP countries they pay positive MFN tariffs on less than
10 percent of their trade with the EU; there are 21 countries that pay MFN tariffs on between 10-
25 percent of their trade, and 30 countries that pay MFN tariffs on more than 25 percent of their
trade.
Once again we need to think about why it is that 51 GSP countries are paying MFN tariffs on
more than 10 percent of their trade. Is this because they are not utilising preferences which are
in principle open to them because for example they are not attractive enough, or is it because of
failing to meet the administrative requirements? Consider Figure 2.10 below:
45
Figure 2.10: GSP Countries: Frequency Chart with Difference between MFN and GSP
This figure sets out the number of GSP countries for whom the difference between the MFN
tariff and the GSP tariff is less than 1 percent, between 1 percent and 2 percent and so on.16
Overwhelmingly, the difference between the GSP and MFN tariff is extremely small. For all but
four of the 51 countries the difference on average is more than 5 percent, and for 45 countries is
less than 3 percent. As discussed in Section 3.2.1, across all the GSP countries, 50 percent of
flow eligible for preferences use those preferences when margins are less than 6 percent and,
and 25 percent when margins are below 2.7 percent. This shows that even with low margins
countries utilised the preferences, but that when margins are low, e.g. below 2.7 percent, that
75 percent of eligible flows are not utilised.
This issue is explored in more detail in Figure 2.11 below. Consider first the difference in
(hypothetical) preference margins between the GSP regime, and both the MFN and the GSP+
regime in the figure below. Again we take the underlying 10 digit export data and on that basis
compute the average tariff that would apply for each country if they were exporting on an MFN
and GSP+ basis, and then we compare this to the equivalent GSP tariff that would apply on that
trade.
16
These differences in average tariffs are hypothetical differences
46
Figure 2.11: GSP Average Preference Margins
93
16
4 1 0 0
91
146
2 0 10
20
40
60
80
100
<2.5% <5% <7.5% <10% <15% >15%
No
of
Co
un
trie
s
MFN GSP+
Hence the first two columns show that for 93 of the 114 GSP countries the difference in the
weighted average MFN tariff, and the weighted average GSP tariff is less than 2.5 percent.
Similarly, the difference between the weighted average GSP+ tariff that would apply and the
weighted average GSP tariff is less than 2.5 percent for 91 of the countries. The difference
between the MFN and GSP tariff is greater than 5 percent for only five countries; and between
the GSP+ and the GSP tariff for nine countries.
This suggests two things. First of all that for the majority of GSP countries, given their export
structure, being offered GSP tariffs as opposed to MFN tariffs only gives these countries a small
preference margin. Secondly, even if all GSP countries were to be offered improved GSP+
tariffs, this would again not make a big difference on average for the majority of such countries.
Figure 2.12 below then considers what would be the (absolute) change in the share of exports
eligible for duty free access for the GSP countries, based on their existing 10-digit trade, but
they exported either entirely under the MFN regime or entirely under the GSP+ regime. If we
consider first the implications of the MFN regime, we see that for 62 countries the GSP regimes
increase their share of duty free access by less than 10 percent, for 52 countries by more than
10 percent. Analogously, if the GSP countries were offered improved preferences and suppose
these were equivalent to the EBA preferences then for 55 countries this would increase their
share of duty free trade by less than 10 percent, while for the remaining 59 countries, their
share of duty free trade would increase by more than 10 percent.
47
Figure 2.12: Change in Share of Duty Free Eligible Exports under MFN & GSP+
62
22 21
8
1
55
29
11 118
0
10
20
30
40
50
60
70
<10 <25 <50 <75 <100
No
of
Co
un
trie
s
MFN GSP+
These figures suggest that for over half of the GSP countries the additional duty free access
that the GSP scheme offers is unlikely to significantly impact on their trade flows; and the
reason for this is that a great deal of their trade already enters under MFN duty free; similarly for
over half of the GSP countries, even if they were offered additional preferences this would be
unlikely to make a significant impact. It is interesting to note that the composition of the
countries falling into each of the preceding categories is quite different, with approximately half
of the countries falling into both groups.
2.2.4 Section Summary
This section assessed the relative importance of the GSP regimes for developing country
exports through looking at the structure of the preferential regimes and developing countries
trade with the EU.
The analysis indicates that in total the EU comprises more than 50 percent of total exports for
only 42 of the 175 countries. This puts into perspective the extent to which preferential trade
policy by the EU could impact on trade and development more generally.
On average the preference regimes do not appear to account for a substantial amount of the
relevant countries trade with the EU. Once again this would suggest that, on average, the
structure of the GSP regimes may not be well directed towards the export needs of developing
countries.
In examining the GSP average preference margins, the analysis suggests that for the majority
of GSP countries, given their export structure, being offered GSP tariffs as opposed to MFN
tariffs only gives these countries a small preference margin. Secondly, even if all GSP countries
48
were to be offered improved GSP+ tariffs, this would again not make a big difference on
average for the majority of such countries. If the GSP countries were offered improved
preferences and suppose these were equivalent to the EBA preferences then for 55 countries
this would increase their share of duty free trade by less than 10 percent, while for the
remaining 59 countries, their share of duty free trade would increase by more than 10 percent.
2.3 Impact of preference regimes on other LDCs The impact of different preferences and changes in those preferences will not only depend on
the exporting structure of individual GSP beneficiary countries, but also on the structures of
other competing countries (GSP and non GSP). Hence, while offering improved preferences to
one set of countries may help their exports to the EU, this may well be (at least in part) at the
expense of third countries‘ exports to the EU – and these third countries could be other LDCs,
other preferential partners, or MFN exporters. In other words changing preferential regimes is
likely to lead to trade creation, trade diversion and trade reorientation.
The precise distribution of these effects is difficult to quantify without an explicit modelling
structure. This we undertake in Section 6 of the report. However, looking at the structure of each
country‘s trade and comparing it with that of other countries can shed light on this issue.
In this part of the report we do this in two ways. First, we consider the overall similarity of the
exports of each country, with that of each of the regime types – EBA, GSP and GSP+.
Secondly, we consider how much competitive pressure is exerted on each country by each of
the other trading regimes, and also how much competitive pressure each country exerts on
each of the different trading regimes.
2.3.1 Similarities in export structures
If we consider first the structure of each country‘s trade, a useful way of examining this is by
looking at the similarity of each country‘s trade with each of the preferential regimes. This can
be seen in Table A.11 of Appendix 4, where we compute the Finger-Kreinin index of export
similarity between each country and each of the preferential regimes.17
Take the first row of the table. Here we see that the FK index for Afghanistan with respect to
other EBA countries is 0.008. The Finger-Kreinin index ranges between 0 and 1. If the structure
17
The mathematical formula for FK index is as follows:
(7)
i
i
ij
ij
i
ih
ih
ihjX
X
X
XFK ,min
Where i refer to a specific sector (or product), h to the home country and j to the partner country or to the
RoW. Xih/ Σ Xih is the share of product‘s i export in country‘s h total exports, Xij/ Σ Xij is the share of
product‘s i export in country‘s j total exports. The FK indices report here have been calculated on the
basis of the 10-digit disaggregation of trade flow.
49
of Afghanistan‘s trade were identical to that of the average of the other EBA countries then the
index would be equal to 1. The figure of 0.008 shows that the structure of Afghanistan‘s trade is
very different indeed to that of the average of the other EBA countries, and similarly with regard
to the GSP, GSP+ countries‘ trade. Indeed, looking generally at this table, the lack of similarity
for each country with each of the trading regimes comes across strongly.
Table 2.8 provides some summary statistics based on this table. For the summary in the left
hand panel, we first identified for each country with which regime of trade flow – EBA, GSP,
GSP+ or MFN - is the most similar to its trade. For each type of country we then simply count for
how many cases their trade is most similar to either of these regimes. This gives a prima facie
indication of the type of trading regime, which each of the GSP country categories are likely to
be competing with most. Hence, if you consider the first row, we see that for the EBA countries,
for 26 of these, their trade was most similar to other EBA trade flows on average, for one of the
EBA countries its trade was more similar to the GSP trade flows on average, for 15 of these
most similar to the GSP+ countries, and so on. So what we see from this left hand panel is that
the EBA countries typically have the highest incidence of similarity with other EBA countries,
and similarly the GSP+ countries have the highest incidence of similarity with other GSP+
countries. Interestingly, the incidence of similarity across the different regimes for the GSP
countries is very uniformly spread out across the four trading regimes.
While, on the one hand, the preceding gives an indication of the likely trading regime which
each group of country is most likely to be in competition with, it is important to consider not
simply the incidence of similarity but also the extent of that similarity. Light is shed on this in
Table 2.8. In the right hand panel of this table, we consider a different summary statistic. Here,
we first take for each country the regime with which its exports are the most similar (and this
could be any of the four regimes of entry identified in the left hand panel of the table), and then
count for how many countries is this level of similarity greater than 0.1, 0.25, and 0.5
respectively. We see that for the majority of the EBA countries the degree of similarity is less
than 0.1, and with only four countries having a degree of similarity greater than 0.25. Similarly,
approximately half of the GSP countries have a degree of similarity of less than 0.1. For the
GSP+ countries, the majority of these (12) have a degree of similarity greater than 0.1.
Clearly there are differences between countries, but what this suggests is that overall for each
of the countries considered, the degree of similarity on average with other countries is typically
very low. From the point of view of changing preferences impacting on other developing
countries, this might suggest a comparatively low level of impact. However, it is important to not
only consider the similarity in the structure of trade but also to consider the value of trade. It is
entirely possible that country A has a highly different structure to country B but size differences
and changes in preferences given to country B may well have a big impact on the exports of
country A. In order to see this we turn to the Relative Export Competitive Pressure Index
(RECPI).
50
Table 2.8: Summary Results on Export Similarity
Number of cases where the greatest
similarity is with:
Number of cases where
similarity is greater than:
EBA GSP GSP+ MFN 0.1 0.25 0.5
EBA 26 1 15 8 10 4 0
GSP 33 26 24 31 46 24 2
GSP+ 3 1 10 0 12 3 0
Source: Authors‘ calculations based on TARIC data supplied by the European Commission
2.3.2 Relative Export Competitive Pressure Index
In considering the degree of competitive pressure between any pair of countries, or between
groups of countries, there are two closely related questions. First, how much competition does a
given country (e.g. Uruguay) face from other countries; and in the context of this study it is then
interesting to see the level of competition by regime type. Hence, we are interested in examining
the degree of competitive pressure that each country faces from MFN, GSP, GSP+ and EBA
countries. The second question concerns how much competitive pressure each country exerts
on other countries; here again we are interested in examining this by regime type. The answers
to both these questions will depend both on the structure and the value of the countries‘ trade.
These issues are explored more formally and at a more aggregate level in Section 5 of the
report where we provide a multi-country CGE modelling exercise. However, working with highly
disaggregated data we can also shed light on this issue by comparing the structure and value of
each country‘s exports to the EU, with that of the structure and value of the exports to the EU
under each of the preferential and non-preferential regimes. We do this via the RECPI index
(Caris, 2008).
Consider the first question. In the first instance we would expect that competitive pressure on
Uruguayan exports is more likely to occur in tariff lines where both Uruguay and other partners
trade heavily with the EU and where there may be a tariff wedge between Uruguay and other
preferential partners. The RECPI index allows us to compare the value of trade by regime type
(e.g. MFN, GSP, GSP+, EBA) which is in direct competition with Uruguay.
i pipi
i jipi
xs
xsRECPI
Where spi represents the share of Uruguayan exports to the EU in total exports to the EU for a
given product i and xj,i is the value of exports of regime (MFN, GSP etc) j to the EU in the same
product i. The numerator gives the value of trade by regime type that is in direct competition
with Uruguay, and divides this by the same calculation based on Uruguay‘s values and shares
of trade. If we take the EBA exports then, RECPI would therefore tells us how much bigger (or
smaller) are EBA exports relative to those of Uruguay, but weighted by how important each
sector is in Uruguay‘s total exports to the EU.
51
Suppose EBA exports to the EU were twice the value of Uruguay‘s exports, but that these
exports were entirely in sectors in which Uruguay did not export at all. The index would then be
equal to zero. In this case any change in preference given to either Uruguay or to other EBA
countries is unlikely to lead to any significant pressure on Uruguayan exports, and thus would
not lead to any significant trade diversion. Alternatively suppose EBA exports were ―n‖ times
bigger than those of Uruguay in every single sector, then the index would be equal to ―n‖. The
larger ―n‖ is, the bigger the competitive pressure on Uruguay. The index therefore controls for
the relative similarity and size of the each type of regime‘s trade with the EU, compared to that
of Uruguay‘s trade with the EU.
The larger the RECPI is, the more likely it is that any improvement in preferences given to third
countries will impact on Uruguay‘s trade and thus results in either diverting trade away from
Uruguay. This is where those preferences are superior to those offered to Uruguay, or reorient
trade towards third countries (where the preferences match those previously granted to
Uruguay. Trade diversion would occur if a country initially had the same preferences as
Uruguay, but as a result of improved preference exported more to the EU in place of Uruguay.
This could be the case, for example, if other GSP countries were given improved access to the
EU, e.g. via GSP+. Trade reorientation could occur where Uruguay initially had better
preferences than the third country, which then obtained the same preferences as Uruguay.
The RECPI calculation allows us to look at the possible competitive pressure each country
faces, where we have calculated in each case the competitive pressure faced from each country
by different category of trade – EBA, GSP, and GSP+, as well as from those exports which
entered under the MFN category to the EU. We have also done this for all trade, as well as for
trade where the MFN tariff is greater than zero. The former provides an indication of the overall
competitive pressure a given country faces from the different categories of trade groupings. The
latter narrows this down to all trade where preference margins could be given.
The detailed results of these calculations by country are given in Table A.12 of Appendix 4,
where the calculations are once again based on the 10-digit disaggregation of trade flows. Each
entry of that table indicates how much competitive pressure each country faces for each of the
categories of trade. Hence, the first row of Table A.12 shows that with regard to ―all trade‖
Afghanistan faces the most competitive pressure from MFN exports to the EU, and the least
competitive pressure from other EBA exporters. This is also true when we look at ―MFN>0‖
trade. However, we also see that with regard to MFN>0 trade, Afghanistan also faces
considerable competitive pressure from other GSP countries. This suggests that improvements
in preferences given under the GSP arrangements could potentially have a substantial impact
on Afghanistan.
In principle, this is an analysis which is best done country by country as this gives the more
appropriate level of detail; and Table 2.9 below provides a useful summary presentation of this
information. For each country we have taken the ratio of its RECPI calculated with respect to the
category of preferential trade it belongs to, with the RECPI calculated with each of the other
52
categories of trade, and then take the average across all countries by regime type. The upper
panel does this with respect to all trade, and the lower panel with respect to MFN>0 trade.
Hence if you take the first row of the table, we see that on average for the EBA countries, the
greatest amount of competitive pressure overall comes from MFN exports, a similar amount of
pressure comes from GSP exports, and significantly less pressure from GSP+ exports. If we
consider the same but with regard to MFN>0 trade, the overall pattern is the same but we see
that relative to other EBA exporters there is considerably more competitive pressure arising
from MFN and GSP exporters. This suggests that changing the preferences received either by
other GSP countries, or by lowering MFN tariffs, could have a significant impact on EBA exports
into the EU market.
If we consider the situation for GSP exporters we see that with regard to total trade, the ratio
relating to EBA, GSP+ and MFN exports is in all cases less than one. This suggests that the
primary source of competitive pressure for these countries is from other GSP exporters, as
opposed to from EBA, GSP+ or MFN exporters. When looking at MFN>0 trade, we now see that
the most competitive pressure is from MFN exports. This indicates, that improving MFN
preferences is likely to have the biggest impact, on average, on GSP exports. For the GSP+
countries, we see that the largest amount of pressure stems from other GSP countries, and this
is true both of all trade and MFN>0 trade.
Table 2.9: Summary RECPI by Regime Type
EBA GSP GSP+ MFN
EBA 1 27.97 1.71 34.24
GSP 0.07 1 0.13 0.87
GSP+ 0.87 11.31 1 3.23
EBA 1 406.09 17.32 674.29
GSP 0.02 1 0.92 2.45
GSP+ 0.49 22.79 1 9.69
Source: own calculations based on TARIC data supplied by the European Commission
In order to address the second question, which is how much competitive pressure each country
exerts on each of the regime types, the same indicator can be used, but where the comparator
base is now the regime type itself.
i EBAiEBAi
i piEBAi
xs
xsRECPI
The numerator gives the value of each country‘s exports that are in direct competition with a
given regime type and divides this by the same calculation based on the values and shares of
trade for that regime type. For example, if we take the EBA as the regime type and Uruguay as
the specific country, then this version of the RECPI tells us how much bigger (or smaller)
53
Uruguay‘s exports are relative to those of all the other EBA flows but weighted by how important
each sector is total EBA exports to the EU.
Suppose Uruguay‘s exports to the EU were twice the value of total EBA exports but that these
exports were entirely in sectors in which the EBA countries did not export at all. The index would
then be equal to zero. Alternatively suppose Uruguay‘s exports were ―n‖ times bigger than those
of the EBA countries in every single sector, then the index would be equal to ―n‖. The larger ―n‖
is, the bigger the competitive pressure that Uruguay exerts on the EBA countries.
The detailed results by country for total trade are given in Table A.13 of Appendix 4. In general
we see that, perhaps not surprisingly, each individual country typically exerts very little
competitive pressure on each of the aggregate groupings. Tables 2.10 and 2.11 below provides
a sub-set of results where we have taken the largest RECPI‘s by EBA and GSP country for total
trade and MFN>0 trade as before.
Table 2.10: Competitive Pressure by Country upon each Regime Type – all trade
EBA GSP GSP+ MFN
Angola EBA 0.65 0.04 1.04 0.09
Equatorial Guinea EBA 0.32 0.02 0.51 0.04
Bangladesh EBA 0.02 0.00 0.02 0.00
Chad EBA 0.00 0.00 0.01 0.00
Mozambique EBA 0.00 0.00 0.00 0.00
Russian Federation GSP 5.05 0.32 8.47 0.72
Libya GSP 2.42 0.15 3.91 0.34
Saudi Arabia GSP 1.37 0.09 2.24 0.19
Iran GSP 1.06 0.07 1.72 0.15
Nigeria GSP 0.97 0.06 1.57 0.13
Kazakhstan GSP 0.92 0.06 1.49 0.13
Iraq GSP 0.78 0.05 1.27 0.11
Azerbaijan GSP 0.76 0.05 1.23 0.11
Algeria GSP 0.64 0.04 1.03 0.12
Mexico GSP 0.27 0.02 0.44 0.04
Source: own calculations based on TARIC data supplied by the European Commission
From Table 2.10 we can see that with regard to total trade it is clear that the biggest competitive
pressure is primarily generated by the energy exporting countries and is clearly related to their
energy exports. If we consider MFN>0 trade a somewhat different picture emerges. We see
that the largest amount of competitive pressure is typically exerted by China and India (see
Table 2.11), and each of these countries has the biggest impact on EBA countries.
54
Table 2.11: Competitive Pressure by Country upon each Regime Type – MFN > 0
EBA GSP GSP+ MFN
Bangladesh EBA 0.75 0.04 0.07 0.02
Mozambique EBA 0.12 0.00 0.01 0.01
Cambodia EBA 0.07 0.00 0.00 0.00
Madagascar EBA 0.02 0.00 0.01 0.00
China GSP 1.05 0.29 0.11 0.14
India GSP 0.37 0.13 0.14 0.05
Tunisia GSP 0.12 0.02 0.01 0.01
Morocco GSP 0.12 0.02 0.01 0.01
Mauritius GSP 0.09 0.01 0.01 0.00
Indonesia GSP 0.07 0.02 0.03 0.01
Egypt GSP 0.07 0.01 0.01 0.01
Russian Federation GSP 0.07 0.02 0.02 0.01
Vietnam GSP 0.06 0.02 0.01 0.00
Pakistan GSP 0.06 0.01 0.00 0.00
Source: own calculations based on TARIC data supplied by the European Commission
2.3.3 Section Summary
This section examined how changing preferential regimes is likely to lead to trade creation,
trade diversion and trade reorientation. The overall similarity of the exports of each country was
examined with that of each of the regime types – EBA, GSP and GSP+. The level of
competitive pressure exerted on each country by each of the other trading regimes, was also
examined along with how much competitive pressure each country exerts on each of the
different trading regimes.
The analysis used the Finger-Kreinin index of export similarity between each country and each
of the preferential regimes. The results indicated that the EBA countries typically have the
highest incidence of similarity with other EBA countries, and similarly the GSP+ countries have
the highest incidence of similarity with other GSP+ countries. The incidence of similarity across
the different regimes for the GSP countries is uniformly spread out across the four trading
regimes. From the point of view of changing preferences impacting on other developing
countries, this result suggests a comparatively low level of impact.
The section also used the RECPI index to look at the possible competitive pressure each
country faces from each country by different category of trade – EBA, GSP, and GSP+, as well
as from those exports which entered under the MFN category to the EU. The analysis suggests
that changing the preferences received either by other GSP countries, or by lowering MFN
tariffs, could have a significant impact on EBA exports into the EU market.
55
The analysis suggests that improving MFN preferences is likely to have the biggest impact, on
average, on GSP exports. For the GSP+ countries, the largest amount of pressure stems from
other GSP countries, and this is true both of all trade and MFN>0 trade.
The analysis with regard to total trade indicates that the biggest competitive pressure is
primarily generated by the energy exporting countries and is related to their energy exports.
However, when considering MFN>0 trade, the largest amount of competitive pressure is
typically exerted by China and India, and each of these countries has the biggest impact on
EBA countries.
2.4 GSP and LDC development needs In this part of the report we try to shed some light on the relationship between the EU‘s GSP
regimes and the extent to which these regimes are addressed to meet the countries‘
development needs. It needs to be emphasised that the notion of development needs is
extremely challenging and broad, and that the needs of countries will vary enormously. Hence,
in this section we assess whether there is any evidence that either preference margins or
utilisation rates, etc are related to certain measures of development or development needs –
such as GDP per capita, or the UN‘s Human Development Index, or Human Poverty Index. At
best these correlates should be seen as broadly indicative but nevertheless useful. Secondly,
we assess the evidence on whether there have been any significant changes in trade in either
the extensive or intensive margin as arising from the EU‘s preference regimes.
2.4.1 Preferences and development
In this section we use the underlying detailed 10-digit information described earlier and combine
this with information about countries‘ level of development. Specifically we take five different
ways of summarising the extent to which countries utilise the EU‘s preferences on offer, and
correlate these to five different measures of development. The five summary measures are:
1. PM1 (Preference Margin 1): These are the preference margins calculated as the
difference between the actual average tariff paid by each country, and the hypothetical
MFN tariff they would have paid, on their actual exports had all trade entered under the
MFN regime.
2. PM2 (Preference Margin 2): These are the preference margins calculated as the
difference between the hypothetical tariff that each country would have paid had all their
trade entered under the GSP preferential regime which applies to that country, and the
hypothetical MFN tariff they would have paid, on their actual exports had all trade
entered under the MFN regime.
3. ETS (Eligible Trade Share): this is the share of trade which enters the EU using the
preferential regime which applies to that country
56
4. DEA (Difference between eligible and actual): This is the difference between the share
of trade eligible to enter using the preferential regime which applies to that country, and
the actual share of trade which entered using that regime.
5. UR (Utilisation Rate): This is defined as the share of MFN non-zero trade which entered
using preferential rates.
The four measures of development we correlate these with are: the UN‘s Human Development
Index, the UN‘s Human Poverty Index, the growth of real GDP between 2000-2008, the
annualised growth rate over the same time period, and GDP per capita in 2008. The results of
these correlations are given in Table 2.12 below, and where we consider these correlations for
all the developing countries, and then broken down by type of regime. The top panel of the table
gives the anticipated sign on the correlation coefficient. In the remaining panels we highlight
those cases where the sign of the coefficient matches the expected sign. In interpreting these
correlations it is important to bear in mind that these are simply correlates and that therefore
one has to be extremely careful in imputing causality.
The picture that emerges from this table is highly mixed. With regard to the Human
Development index there is some evidence of the expected correlations for the GSP+ group of
countries, but very little for the EBA or GSP countries, and a similar picture emerges for the HPI.
There is some evidence that higher rates of utilisation are associated with higher rates of growth
(although the correlation coefficients are very low), but little relationship between growth rates
and preference margins. However, there some evidence that preference margins tend to be
higher for countries with lower levels of GDP per capita, although once again the coefficients
are low.
57
Table 2.12: Preferences and Development
Regime PM1 PM2 ETS DEA UR
HDI - - - + -
HPI + + + - +
GDP
growth + + + - +
GDP /
capita - - - + -
HDI
Correlation
ALL -0.096 0.091 0.021 -0.053 0.196
EBA 0.498 0.452 0.514 -0.535 0.321
GSP 0.040 -0.075 0.097 -0.073 -0.276
GSP+ -0.181 -0.132 -0.308 0.314 0.344
HPI
ALL 0.081 -0.056 -0.051 0.075 -0.263
EBA -0.477 -0.495 -0.455 0.457 -0.462
GSP 0.011 0.138 -0.145 0.115 0.270
GSP+ 0.341 0.257 0.428 -0.441 -0.451
Real GDP
Growth
rate
ALL -0.012 -0.215 0.066 -0.091 0.057
EBA -0.477 0.085 -0.024 -0.066 0.175
GSP -0.231 -0.304 0.187 -0.186 -0.066
GSP+ -0.360 -0.372 -0.368 0.423 0.219
Annualised
Growth
Rate
ALL -0.020 -0.216 0.036 -0.069 0.030
EBA 0.018 0.003 -0.112 0.006 0.110
GSP -0.240 -0.315 0.201 -0.203 -0.050
GSP+ -0.238 -0.229 -0.205 0.263 0.305
GDP per
capita
ALL -0.169 0.032 -0.192 0.168 0.124
EBA 0.012 0.009 0.041 -0.054 0.113
GSP -0.182 0.009 -0.246 0.229 0.080
GSP+ -0.318 -0.267 -0.452 0.431 0.252
Source: own calculations
2.4.2 Analysis of changes in intensive and extensive margin
In this section, we decompose export growth to the EU into the intensive and extensive margins.
Our primary interest is to grasp the relation between GSP preferences and growth of exports to
the EU in new or existing products. To this end, we define the extensive margin as that which
captures exports of new products into the EU market, whilst defining the intensive margin as the
process of consolidation of existing export flows. This is a slightly different focus to that
sometimes found in the literature in that it does not take into account quality changes via unit
prices or changes in destination of exports.18 It is nonetheless interesting as it allows us to
18
Our approach differs to that of the literature because the nature of our detailed dataset does not allow
us to investigate destinations of exports other than the EU.
58
capture the evolution of export growth into the EU and ultimately the role of preferences in the
process of export diversification.
Although not a panacea, diversification is important as it reduces the risk of trade shocks. There
is also evidence suggesting that diversification is linked with economic growth, especially at
lower levels of development (Heiko (2006) and Imbs, & Wacziarg (2003)). In addition, there is a
nascent literature on the productivity enhancing effects of engaging in export markets (Melitz
2003), although this does not differentiate between changes in the intensive or extensive
margin.
As a first exercise, and before undertaking the export decomposition, we document the growth
of exports both to the EU and to the Rest of the World (RoW). Holding everything constant, we
would expect changes in trade policy to affect export growth rates.19 Table 2.13 shows annual
export growth rates of country aggregates for the period 1991 to 2008. The aggregates are
clustered countries according to the regime that was applicable to them in 2008.20 The first entry
of the table tells us that during the period 1991-2001, countries which in 2008 were classified as
EBA saw an annual growth of exports to the EU of 4.8 percent. The second entry tells us that
during the period 2002 to 2008 the annual growth rate of EBA countries (in 2008) was
18.2 percent. The lower panels then consider export growth to the rest of the world and to the
world, whilst the final panel reports the annual growth rate of world trade as a benchmark.
The first striking feature of the table is the difference in annual growth rates across the periods
where 2002-2008 saw a doubling of annual trade growth rates with respect to 1991-2001. This
increase in world trade was accompanied by important reductions in transport costs and rapid
trade liberalisation. The period was also characterised by large increases in GDP which
stimulated foreign demand. Bearing these in mind, and considering EBA countries first, we see
how annual export growth of EBA countries to the rest of the world (RoW) was almost double
that to the EU25. This is not totally unexpected given that the RoW grouping saw faster growth
than the EU25 and also saw faster relative liberalisation. That is, most EBA countries have
received near duty free access to the EU since the 70‘s (under the Lome and Cotonou); hence
the change in preferences in the EU market has been relatively small. The GSP and GSP+
groupings however show slightly higher annual export growth towards the EU market. Whilst
this cannot be directly attributed to the changes in the GSP regime, it is an important finding.
Given that the RoW group showed faster GDP growth and undertook faster liberalisation, we
19
Annual export growth rates can vary according to both time varying and time invariant factors. The time
varying factors include GDP, population and changes in trade policy such as the introduction of the GSP
regime. We expect that extending preferences to a set of countries will increased the annual growth of
exports of this country to the preferential destination, but there is also reason to suspect that unilateral
preferences can have effects beyond a single destination. There could be a learning by doing effect
where learning to export to one destination makes it less costly to access o other destinations. This is
beyond the scope of this section but is treated in more depth in the econometric section of this report. 20
That is, if a given country was an EBA country in 2008, which it appears to be in the whole of the
sample.
59
would continue to expect export growth to the RoW to be higher than to the EU25. Hence it
seems that there might be something driving these results.
Table 2.13: Annual Growth of Exports by Category 1991-2008
1991-2001 2002-2008 1991-2008
Annual growth to EU 25
EBA 4.8 18.2% 8.6%
GSP+ 3.8 21.4% 8.0%
GSP 7.4% 23.0% 12.0%
Annual growth to RoW
EBA 11.0% 30.5% 15.7%
GSP+ 10.6% 19.7% 11.1%
GSP 12.9% 21.0% 13.9%
Annual Growth to the World
EBA 8.5% 27.0% 13.3%
GSP+ 9.3% 20.0% 10.5%
GSP 11.5% 21.5% 13.4%
Growth of World Exports
World 11.3% 20.3% 13.3%
Source: Own calculations using COMTRADE (WITS)
We now turn to the decomposition of export growth into intensive and extensive margins. This
involves separating trade flows into new, old and existing products. We start off by using a very
stylised algorithm that divides the dataset into two periods (method 1), namely 2002 to 2004
(period 1) and 2005 to 2008 (period 2). We then define old products as those where there is
any trade in that product in period 1 but not in period 2. New products are those where there is
some trade in that product in period 2 but not in period 1, and existing products require there to
be trade in both periods.21 This allows us to decompose the change in trade between two
periods as follows:
Xi,EU = [Existing - Old] + [New]
Change in exports = [Intensive margin] + [Extensive margin]
The change in exports is then the sum of the intensive and extensive margins. The former
comprises changes in existing exports minus products that disappear whilst the latter captures
changes in new exported products. To identify these, we re-classify the trade dataset so that it
has a homogeneous nomenclature (HS-2002) for the entire period which requires us to
21
Note that a given product belongs to any one of the periods if there is trade during any of the years that
make up the period. Hence, for a given product, if there are exports in 2002 but not in 2003 or 2004, this
product still counts as a product belonging to the period 1 set. We however eliminate any trade values
that are below 1000 euros so as to avoid incorrectly assigning products that may be casual exports.
60
aggregate the data to 6-digits.22 Figure 2.13 then shows the change in exports as delimited by
changes in new, old and existing products. From this figure, we surmise that changes in exports
have almost entirely been driven by changes in existing products for all three groupings.
Figure 2.13: Changes in Exports by Type and Grouping (2002-2008)
Source: Own calculations
It is likely that these findings are very sensitive to the methodology used to identify the three
types of products; hence in Table 2.14 we carry out a sensitivity analysis using two alternative
identification procedures. The first (Method 2) also splits the data into two periods, but now only
takes the first year (period 1) and the last year (period 2) to identify new, old and existing
products. Hence if a product does not exist in 2002 but it does in 2008, then it is defined as a
new product. Similarly if a product exists in 2002, but not in 2008, then it is taken as an old
product. It then follows that a product which is present in both periods is defined as an existing
product. Method 3 then extends period 1 and period 2 by two years so that requirements are
that the product be present in any two years within one period. As can be seen from the table,
the overwhelming majority of exports to the EU continue to be explained by increases in existing
exports (the intensive margin).
22
Keeping a constant nomenclature across the time period is of crucial importance for the identification of
products according to our delimitations. To do this we have had to sacrifice aggregation for precision.
61
Table 2.14: Export Growth Decomposition by Varying Identification Procedures
INTENSIVE EXTENSIVE
Existing Old New
Meth
od 1
EBA 98.62% -2.12% 3.50%
GSP+ 99.32% -1.10% 1.77%
GSP 98.46% -0.29% 1.83
Meth
od 2
EBA 101.74 -8.85 7.11
GSP+ 99.76 -2.87 3.11
GSP 100.06 -6.08 6.02
Meth
od 3
EBA 100.68 -6.21 5.52
GSP+ 101.35 -4.31 2.96
GSP 101.15 -5.77 4.62
Source: Own calculations using EU database
The above table, a priori, suggests that there is a relatively stable pattern of export growth at the
intensive margin and that this seems to be robust to changes in the identification procedure.
However, the ultimate question being the link between preferences and growth of exports in the
intensive and extensive margins, we now turn to analysing if there is any correlation between
changes in applied tariffs and shares of intensive and extensive margins. Figure 2.14 provides a
scatter plot (where the identification procedure follows Method 3) with the share of the intensive
(extensive) margin in explaining export growth on the vertical axis, and the change in actual
applied tariff during the period 2002-2008. Each point represents a country in our dataset. Here
we clearly see that there appears to be no correlation between intensive (extensive) margins
and changes in applied tariffs.
62
Figure 2.14: Correlation between Intensive and Extensive Margins and Applied Tariffs
Source: Own calculations using EU database
It is also worth considering the correlation between the intensive and extensive margins and the
height of the preferential margin. In the figures below, we consider two types of preferential
margin, the first being the difference between the MFN tariff and the applied tariff (inclusive of
any preferences and termed PF1) and the second being the difference between the MFN tariff
and the tariff faced under the relevant GSP scheme for each country (PF2). Once again, Figure
2.15 demonstrates that there is no correlation between the height of the preference margin and
the degree of the export margin.
63
Figure 2.15: Correlation between Preference and Export Margins
Source: Own calculations using EU database
Whilst the above figure shows little overall correlation between changes in applied tariffs and the
margins of exports, we also need to consider how changes in the intensity of exports behaves at
different preference levels. In Tables 2.15 and 2.16 we identify new, old and existing products
by region and classify these according to the degree of the preference margin (in 2008) across
10 digit products. The difference between these tables is the calculation of the preference
margin which is as above. In Table 2.15 we calculate the preference margin as the difference
between the hypothetical MFN tariff and the effectively applied tariff.23 Whereas in Table 2.16
the preference margin is calculated as the difference between the hypothetical MFN tariff and
the hypothetical tariff in each countries applicable regime (i.e. if an EBA country then the
preference margin is calculated as the difference between the hypothetical MFN and the
hypothetical EBA tariff). The tables confirm the above perception that growth of trade is largely
explained by growth in existing products. It seems that preferential margins do not correlate to
changes in exports of new products. There appears to be no discernable link between higher
preferences and higher extensive margins. There is, nonetheless some evidence that at lower
levels of preference margins (0<Pref.<10), there has been some growth in new exports. The
tables also suggest that growth at the extensive margins where the tariff preference is zero
seems to be the same than where the tariff preference is above 20. Whilst an in depth analysis
of the relationship between preferential margins and export of new products is beyond the
23
This means that the preference margin in this table will include any other type of preferences awarded
to a given country such as bilateral preferences or preferential quotas. To control for changes in
preferential margins in time we use the preferential margins as they stand in 2008 for the classification.
64
scope of this section, evidence points to there being very little by way of increases in new
exports as preferential margins rise.
Table 2.15: Differences between the Hypothetical MNF and Applied Tariffs
INTENSIVE EXTENSIVE
Share of
total trade
Change
2002-
2008 Existing Old New
EB
A
Preference =0 58.11% 145.95% 102.67% -7.60% 4.93%
0<Preference<5 7.29% 16.77% 71.36% -39.21% 67.86%
5<Preference<10 2.52% 3.94% 140.12%
-
252.08% 211.96%
10<Preference<15 8.72% 110.38% 100.54% -1.67% 1.14%
15<Preference<20 1.45% 20.57% 99.59% -11.75% 12.16%
Preference>20 21.91% 41.11% 97.05% -1.97% 4.92%
TOTAL 100.00% 96.66% 101.37% -7.40% 6.03%
GS
P+
Preference =0 63.72% 109.92% 102.59% -8.34% 5.75%
0<Preference<5 13.11% 66.38% 82.74% -11.83% 29.10%
5<Preference<10 6.23% 83.01% 98.84% -2.20% 3.35%
10<Preference<15 3.14% 97.16% 99.60% -1.02% 1.42%
15<Preference<20 1.62% 60.94% 96.80% -3.42% 6.61%
Preference>20 12.18% 64.36% 97.45% -2.62% 5.17%
TOTAL 100.00% 72.51% 100.06% -6.06% 6.00%
GS
P
Preference =0 57.08% 66.50% 98.64% -3.20% 4.56%
0<Preference<5 4.63% 57.10% 96.05% -3.59% 7.53%
5<Preference<10 8.13% 162.15% 95.02% -3.61% 8.58%
10<Preference<15 3.75% 130.91% 99.62% -2.06% 2.43%
15<Preference<20 3.45% 13.56% 98.79% -2.10% 3.32%
Preference>20 22.95% 70.19% 100.22% -0.65% 0.43%
TOTAL 100.00% 95.21% 98.32% -3.02% 4.70%
Source: Own calculations using EU database
65
Table 2.16: Difference between the Hypothetical MFN & Hypothetical Applicable Tariffs
INTENSIVE EXTENSIVE
Share of
total trade
Change
2002-
2008 Existing Old New
EB
A
Preference =0 56.26% 160.79% 103.13% -6.71% 3.58%
0<Preference<5 4.99% 43.82% 55.61% -28.73% 73.12%
5<Preference<10 2.06% -2.82% 10.48% 504.54% -415.02%
10<Preference<15 10.27% 75.37% 102.52% -4.27% 1.75%
15<Preference<20 1.45% 8.00% 106.13% -40.17% 34.04%
Preference>20 24.96% 29.64% 98.20% -3.99% 5.79%
TOTAL 100.00% 96.66% 101.37% -7.40% 6.03%
GS
P+
Preference =0 57.10% 113.28% 102.72% -7.58% 4.86%
0<Preference<5 3.78% 44.38% 84.78% -14.74% 29.96%
5<Preference<10 4.03% 61.38% 98.20% -9.30% 11.10%
10<Preference<15 5.01% 74.69% 96.68% -2.15% 5.47%
15<Preference<20 5.80% 119.87% 99.02% -0.31% 1.30%
Preference>20 24.27% 91.85% 97.96% -3.33% 5.37%
TOTAL 100.00% 72.51% 100.06% -6.06% 6.00%
GS
P
Preference =0 61.93% 65.08% 98.62% -3.01% 4.39%
0<Preference<5 12.00% 81.19% 95.21% -9.45% 14.23%
5<Preference<10 7.15% 79.63% 94.29% -2.40% 8.11%
10<Preference<15 3.72% 102.27% 99.03% -1.38% 2.35%
15<Preference<20 3.33% 158.98% 100.30% -0.81% 0.52%
Preference>20 11.86% 65.58% 99.57% -0.10% 0.53%
TOTAL 100.00% 95.29% 98.32% -3.02% 4.70%
Source: Own calculations using EU database
2.4.3 Section Summary
This section assessed whether there is any evidence that either preference margins or
utilisation rates, etc are related to measures of development or development needs. It uses five
different methods of measuring the extent to which countries take up the EU‘s preferences and
correlates them to four different measures of development.
The result of this analysis is mixed. The research identified some evidence of the expected
correlations for the GSP+ group of countries, but very little for the EBA or GSP countries. The
results suggest that higher rates of utilisation are associated with higher rates of growth but little
relationship between growth rates and preference margins. However, there was some evidence
that preference margins tend to be higher for countries with lower levels of GDP per capita.
We also considered the extent to which there is any evidence that the GSP regimes lead to
export diversification. The analysis suggests that the growth of trade is largely explained by
growth in existing products. It seems that preferential margins do not correlate to changes in
exports of new products. There appears to be no discernable link between higher preferences
66
and higher extensive margins. There is, nonetheless some evidence that at lower levels of
preference margins, there may have been some growth in new exports. The evidence points to
there being very little by way of increases in new exports as preferential margins rise.
67
2.5 Section 2: Conclusions
Section 2 utilises detailed (10-digit) trade and tariff data to quantify the preferential access
granted to developing countries and the different preferences granted across the three GSP
schemes preference regimes. It focused on four main areas:
1. The structure of the EU‘s GSP system or the degree of coverage and preferential access
granted under the EU‘s existing regimes.
2. The suitability of the GSP regimes in terms of the fit between the preferences offered by
the EU and the structure of developing country exports
3. The competitiveness between developing countries and whether these countries
become more competitive vis-à-vis other countries due to improved preferences.
4. Assessing whether the EU‘s GSP regimes are well directed to countries‘ development
needs
The section notes that the importance of "preferences" in total EU imports is typically low. GSP,
GSP+ and EBA account for 4.18 percent, 0.46 percent and 0.46 percent of total EU imports
respectively. Nevertheless there are four sectors where trade entering under all the GSP
regimes constitutes more than 20 percent of total EU imports - footwear, animal or veg. fats, live
animals and raw hides.
In the GSP and GSP+ systems there are several products (particularly in agriculture) protected
by the EU. Under GSP there are 4781 additional duty free tariff lines, under GSP+ there are
9717, and under EBA 11053.
Of the differences between the GSP and GSP+ many of these differences occur in textiles and
clothing products. This suggests that a country that is highly concentrated in the textiles and
clothing industries is likely to benefit considerably more from GSP+ preferences than from GSP
preferences.
On average the preference regimes do not account for a substantial amount of the relevant
countries' trade with the EU, suggesting that, on average, the structure of the GSP regimes may
not be well directed towards the export needs of developing countries. This is because either
the preferences on offer are not being utilised, or that the countries‘ export structures is such
that they already get access to the EU with low MFN tariffs.
For the majority of GSP countries, given their export structure, being offered GSP tariffs as
opposed to MFN tariffs only gives these countries a small preference margin. Even if all GSP
countries were to be offered improved GSP+ tariffs, this would again not make a big difference
on average for the majority of such countries.
The structure of the EU‘s preference regimes is such that the scope for offering significant
preferential access for developing countries is largely limited to a few sectors. Overall, it is only
in agriculture and processed foods, and textiles and clothing that there is much scope for
68
improved preferential access, and by and large this really only applies to the GSP countries, as
these preferences are already being offered to the GSP+ and EBA countries. This may have
important implications for the incentives in given developing countries with regard to the
orientation of the structure of their exports.
The assessment of the extent to which the utilisation of the EU‘s preferences on offer is
correlated to measurements of development offered highly mixed results. There is some
evidence of this relationship for the GSP+ group of countries, but very little for the EBA or GSP
countries.
Finally, any growth of trade in these countries is largely explained by growth in existing
products. Preferential margins do not correlate to changes in exports of new products. There
appears to be no discernable link between higher preferences and higher extensive margins.
There is, nonetheless some evidence that at lower levels of preference margins there has been
some growth in new exports. Nevertheless, evidence points to there being very little by way of
increases in new exports as preferential margins rise.
69
3 Utilisation Rates The preceding section focused on looking at margins and their differences across preference
regime and on the (relative) amounts of trade covered by this. It is now important to consider the
extent to which countries actually utilise these preferences, what determines the degree of
utilisation and if there is any evidence on the extent of any impact of that utilisation. The detailed
information on utilisation rates by country is given in Appendix 4, Table A9, while in this section
we provide more summary information both by type of regime and also by sector.
3.1 Descriptive stats on utilisation rates Figures 3.1 and 3.2 below provide summary information on utilisation rates for the EBA and
GSP countries. As the number of GSP+ countries is much smaller, there is little value in
providing this information in a figure. However, for these countries utilisation rates are typically
high. For seven countries they are greater than 90 percent, for four countries they are greater
than 80 percent, for two countries they are greater than 70 percent, and for one country the rate
is just below 70 percent. For the EBA countries, if we look at utilisation rates in Figure 3.1 we
see that 28 of the countries have rates greater than 75 percent, while 13 countries have rates
less than 50 percent. If we look at utilisation rates for the GSP countries, in Figure 3.2 we see
that 82 countries have rates in excess of 75 percent, while there are 26 countries with less than
10 percent. Overall then the data suggests that utilisation rates are pretty high.
Figure 3.1: EBA Utilisation Rates
9
13
8
28
0
5
10
15
20
25
30
<10% <25% <50% <75% <100%
No
of
Co
un
trie
s
Source: Own calculations using EU database
70
Figure 3.2: GSP Utilisation Rates
26
2 3
21
61
0
10
20
30
40
50
60
70
<10% <25% <50% <75% <100%
No
of
Co
un
trie
s
Source: Own calculations using EU database
The above analysis does not allow us to grasp the degree to which preferences are suited to the
different countries and the extent to which these are being utilised. By cross-tabulating
countries‘ preference utilisation against share of trade eligible for preferences we can capture
the extent to which the degree of utilisation actually ‗matters‘ to a given country. In Table 3.1 we
carry out this exercise for EBA countries.
Here, it is interesting to look at the four extremes of the table. Countries in the top left corner of
the table are those which show both the lowest rates of preference utilisation and the lowest
shares of trade eligible for preferences. The structure of these countries exports to the EU seem
to be to be particularly badly suited for the EBA regime and even when they can utilise some
preferences, they may be finding it hard to do so.24 If we then take Tuvalu in the top right hand
corner we see how it has a very low rate of utilisation but a high rate of trade that is eligible for
preferences. This suggests that the coverage of the EBA regime seem to be well suited to
Tuvalu export structures to the EU, but that it may be finding it hard to take advantage of these
preferences. Countries located in the bottom left hand corner of the table have very high
utilisation rates but very low shares of trade eligible for preferences. Finally, countries in the
bottom right hand corner are those where both utilisation and eligibility is high. These countries
24
The countries identified in this top left section of the table appear to be ones which have suffered or are
suffering internal conflicts which may explain the low rates of utilisation.
71
are those whose export structures to the EU are best suited to the EBA regime, and which are
also able to take advantage of the preferences granted. Overall, EBA countries seem to cluster
around the left and bottom edges of the table which suggests that utilisation is relatively high,
but for many, the share of trade eligible for preferences is low.
Table 3.1: EBA Suitability
share of trade with EU eligible for preferences
<10 <25% <50% <75% <100%
Uti
lisa
tio
n
<10%
Chad, East
Timor, Liberia,
Somalia,
Sudan, Niger,
Sierra Leon Kiribati
Tuvalu
<25%
Samoa
<50%
Guinea,
Afghanistan
Bhutan
<75%
Benin, Angola,
Burundi,
Congo Dem
Rep,
Guinea
Bissau, Mali Haiti
Central African
Republic
<100% Rwanda,
Lesotho, Equ.
Guinea, Sao
Tome, Burkina
Faso
Togo,
Mauritania,
Zambia
Comoros,
Ethiopia,
Uganda
Eritrea,
Gambia,
Djibouti,
Tanzania,
Senegal,
Malawi,
Solomon
Islands
Madagascar,
Yemen,
Vanuatu,
Nepal,
Mozambique,
Cape Verde,
Laos,
Cambodia,
Bangladesh,
Maldives
Source: Own calculations using EU database
Table 3.2 then mimics the above table for GSP countries. Here we see a more even distribution
of countries around the table where most countries lie in the bottom right quadrant implying both
good utilisation rates and well suited preferences to the structure of trade with the EU. There are
however important exceptions where these two incidences are low which are found in the top
left hand corner of the table.
72
Table 3.2: GSP Suitability
share of trade with EU eligible for preferences
<10% <25% <50% <75% <100%
Uti
lisa
tio
n
<10%
Norfolk Island,
Iraq, Belarus,
Marshall
Islands,
American
Samoa,
Bermuda,
Wallis and
Futuna, Nauru,
Cocos Island,
Cayman
Islands, Brit.
Virgin Is.,
Antarctica,
Niue,
South Georgia,
Min. Out Terr.,
Brunei,
Anguilla,
Bouvet Island
Pitcairn, Heard
Island and
McDonald,
Macao, Palau,
Tonga, Guam,
Northern
Mariana
Islands
<25%
Tokelau
Kyrgyzstan
<50%
Antigua and
Barbuda, Cook
Islands
St Helena and
dependencies
<75%
Azerbaijan,
China, Russia,
Libya, New
Caledonia,
Virgin Islands,
Congo (rep)
Trinidad and
Tobago
Malaysia,
Philippines,
UAE, Jordan,
French
Southern
Territories,
Thailand,
Indonesia
Oman,
Mayotte,
Vietnam
Micronesia,
Turks and
Caicos
Islands, St
Kitts and Nevis
<100%
Nigeria, St
Vincent and
the
Grenadines,
Syria,
Kazakhstan,
Iran,
Paraguay,
Algeria,
Gabon,
Botswana
Qatar,
Bahamas,
Saudi Arabia,
Uruguay,
Brazil,
Cameroon,
Turkmenistan,
Ukraine,
Argentina,
Tajikistan
Ghana,
Grenada,
Chile, Cote
d'Ivoire, British
Indian Ocean
Terr., Surinam,
South Africa,
Kuwait, Dom.
Republic,
Uzbekistan,
Netherlands
Antilles,
Mexico, Aruba,
Egypt,
Lebanon
French
Polynesia,
Armenia, India,
Montserrat,
Papua New
Guinea,
Barbados,
Bahrain,
Zimbabwe,
Cuba,
Dominica,
Tunisia,
Guyana,
Kenya
Moldova,
Namibia,
Belize,
Pakistan,
Morocco,
Falkland
Islands,
Mauritius,
Greenland, St.
Lucia,
Swaziland, Fiji,
Jamaica,
Seychelles
Table A.9 in Appendix 4 breaks down this information into considerably more detail. This table
provides information on the amount of trade which actually entered the EU under the different
73
preference regimes in 2008, but where we also distinguish between eligibility across regimes.
Hence, if you consider the first row of the table we see that 92.45 percent of Afghanistan‘s
exports to the EU were eligible for MFN=0 and entered the EU with MFN zero tariff. We also
see that 4.92 percent of their exports could have entered under GSP>0 rates, but in fact entered
under MFN>0 rates. The total amount of trade entering via the MFN route was therefore 97.46.
The category GSP covers GSP, GSP+ and EBA. Given that Afghanistan is an EBA country we
see that 2.49 percent of its exports were eligible for EBA preferences and entered the EU
utilising those preferences. Only 34 percent of total imports that were eligible for EBA access
actually utilised those preferences.
3.1.2 Section Summary
This section considered the extent to which countries actually utilise these preferences, what
determines the degree of utilisation and if there is any evidence on the extent of any impact of
that utilisation. The evidence suggests that utilisation rates are typically high, but with of course
some variation across countries. We identify those countries whose structure of exports is such
that the GSP regimes may not be well suited on the basis of both low rates of utilisation and low
shares of trade eligible for preferences; as well as those where the converse is true.
74
3.2 The determinants of preference utilisation A mismatch between preferences which have been granted and the degree to which they were
taken up is likely to arise either because exporters may not be aware of the preferences being
granted, or because the benefits of the preferences may not exceed the costs of adhering to
them. In turn, this is likely to be a function of the alternative tariff which the beneficiary country is
likely to face, it could arise from onerous administrative procedures or from rules of origin
restrictions. In this section, we examine possible determining factors which help to shed light on
the degree of preference utilisation rate.
We start by considering the relationship between the average preference margins and two
different ways of considering the extent to which preferences are utilised for the entire sample of
GSP countries for 2008. Figures 3.2 and 3.3 both indicate that the preference margin is
calculated on the basis of each country‘s 10 digit trade, and where we calculate the difference
between the hypothetical MFN tariff that would have applied on that trade and the hypothetical
preferential tariff that applies to that trade.
In Figure 3.2 we correlate this with the utilisation rate, which is the share of MFN non-zero trade
which entered using preferential rates. In Figure 3.3 we correlate this with the difference
between the actual share of trade which entered the EU utilising GSP preferences and the
amount of trade which could in principle have entered the EU either under MFN=0 tariff or that
was eligible for GSP preferences. Take Afghanistan for example. The amount of trade that was
eligible to enter duty free, given that it is an EBA country, was just under 100 percent, which
includes lines where the MFN tariff is equal to zero. In reality, 2.49 percent entered under
EBA=0, with 92.45 entering under MFN=0, and the remainder (5.01 percent) under MFN>0.
Hence, in this case, the high level of non-GSP preferential trade, occurs because a large
proportion of trade can enter under MFN=0 anyway.
Figure 3.3: Correlation between Average Preference Margin and Utilisation Rates
0
20
40
60
80
100
120
0 5 10 15 20
Uti
lisat
ion
Rat
e
Preference Margin (1)
Source: Own calculations using EU database
75
Figure 3.4: All GSP Regimes: Correlation between Preference Margin and Non-Utilisation
of Preferences
0
2
4
6
8
10
12
14
16
18
20
-20 0 20 40 60 80 100 120
Pre
fere
nce
Mar
gin
(1
)
Unused Preference Trade
Source: Own calculations using EU database
In Figure 3.3 there is a positive correlation with a correlation coefficient of 0.21 suggesting that
across the entire sample there is some evidence that utilisation tends to be higher the larger are
the preference margins. Figure 3.4 is based on the possible non-utilisation of preferences so
one might expect the converse relationship which is indeed what we find. We see a negative
correlation (with a correlation coefficient of -0.72), again suggesting that preference utilisation is
associated with the height of the preference margins. Both of these figures suggest that the
degree of use of preferences is correlated with the height of the preference margin. This issue is
explored more formally below.
3.2.1 Preference utilisation: econometric analysis
Trade preferences are not always utilised. Therefore any evaluation of preferential regimes
needs to explain the reasons for non-utilisation. The existing literature explains that non-
utilisation is mainly due to the costs of compliance associated to preferential regimes. A first
element to consider is compliance with product specific rules of origin. In order to be eligible for
preferential treatment, exporters need to comply with rules that establish a minimum threshold
of domestic transformation in the production process from inputs imported abroad. While trying
to avoid export deflection of finished products from non-preferential countries, RoOs de facto
discourage some forms of outward processing and outsourcing originated in non-preferential
partners that may constitute a substantial share of trade flows.
Other costs associated to the use of trade preferential schemes are administrative. While
exports under MFN regimes only need standard documentation such as a ―made in‖ certificate
usually issued by the chamber of commerce, preferential schemes require specific certificates of
76
origin that can only be issued by certain government institutions such as customs or specific
ministries. This usually implies additional documentation that in some cases maybe
cumbersome and costly25.
Several authors have estimated these costs of compliance at between 3 percent and 6 percent.
Manchin (2006) estimates a required preference margin to cover compliance costs above
4.5 percent. Carrère and de Melo (2004) estimates for compliance costs of NAFTA rules of
origin are 6.16 percent. These estimates are based on an estimated threshold margin below
which non-utilisation occurs with more frequency. However, there are two main problems with
this approach. First, an implicit assumption is the link between administrative costs, preference
margin and also export prices. If larger preference margins need to compensate for these
administrative costs, this can only be done by paying higher prices or exporting higher volumes;
as compared to the situation where MFN tariffs are paid. However, Section 3.3 suggests that
larger preference margins are not necessarily translated into higher prices for some preferential
regimes. Second and more important, we observe preference utilisation at very low preferential
margins. For example, 50 percent of flows and 53 percent of the value share of preferential
imports eligible for preferences use these preferences when margins are below 6 percent, and
25 percent (24.92 percent of value share) when margins are below 2.7 percent.26
Non-utilisation of EU preferences
The extent of non-utilisation -of preferences varies across countries (see Table A9 in Appendix
4). Overall it is not very significant for most countries and products. Figures 3.5 and 3.6 indicate
estimates of the probability distribution functions of the share of non-utilisation exports on total
exports27 for countries and 10 digits products in 2007. The figures indicate the frequency of the
different utilisation shares across countries and products. Clearly, for most of the countries and
products the shares of non-utilisation flows as a share of total exports is below 20 percent.
25
Alfieri and Cirera (2007) document anecdotal evidence for Mozambique of non-utilisation cases where
the signature of the relevant certificate of origin could not be produced on time for the date of the
shipment. 26
The figures are highly comparable when broken down between GSP, GSP+, EBA and other
preferences. 27
Excluding exports with unknown entry regime in the EU
77
Figure 3.5: Probability Distribution Function of Preference Non-utilisation Exports as a
Share of Total Exports in 2007 – by Country 0
24
68
Den
sity
0 .2 .4 .6 .8 1nonuti
kernel = epanechnikov, bandwidth = 0.0277
Kernel density estimate
Source: Own calculations using EU database
Figure 3.6: Probability Distribution Function of Preference Non-utilisation Exports as a
Share of Total Exports in 2007 – by Product
05
10
De
nsi
ty
0 .2 .4 .6 .8 1nonuti_p2
kernel = epanechnikov, bandwidth = 0.0110
Kernel density estimate
Source: Own calculations using EU database
78
However, this picture changes slightly when we consider non-utilisation of preferences of
eligible exports for preferences for countries and products in Figures 3.7 and 3.8. Now, two
groups of countries and products emerge. The first cluster on the left of the distribution
represents most countries/products, and in this cluster non-utilisation shares are very low and,
therefore, not important. On the other hand, a smaller cluster of countries/products appear on
the right of the distribution with very large non-utilisation shares.28 Therefore, non-utilisation of
preferences appears to be a polarised phenomenon; not important for most countries/products
and very important for a small subset of countries/products.
Figure 3.7: Probability Distribution Function of Preference Non-utilisation Exports as a
Share of Eligible Exports in 2007 – by Country
0.5
11.
52
Den
sity
0 .5 1nonuti_eligible
kernel = epanechnikov, bandwidth = 0.1019
Kernel density estimate
Source: Own calculations using EU database
28
Some of these countries with low preferential utilisation in 2007 are: Chad, Iraq, Micronesia, Pitcairn,
Virgin Islands or Afghanistan
79
Figure 3.8: Probability Distribution Function of Preference Non-utilisation Exports as a
Share of Eligible Exports in 2007 – by Product 0
12
3
Den
sity
0 .2 .4 .6 .8 1nonuti_eli
kernel = epanechnikov, bandwidth = 0.0372
Kernel density estimate
Source: Own calculations using EU database
To understand further non-utilisation of preferences we need to move from country and product
averages towards specific data at the product and country level. This analysis is possible since
we observe for each year the different regime of entry in the EU of exports at 10 digits level,
which allow us to establish the determinants of non-utilisation.
We estimate a reduced form equation for analysing the probability of preference utilisation
based on the literature on compliance costs, tariffs and margins in (5). The variable Yijt =1 if the
trade flow of product j from country i for a specific tariff regime is eligible to preferences and use
them, and Yijt =0 in case of non-utilisation. An important element of our data is the fact that
because each flow is defined by tariff regime of entry to the EU, we can observe both utilisation
and non-utilisation, for the same product, origin and period, which adds additional within variety
variation to the sample (see Section 3.3.2 for a more detailed explanation of the data).
The problem that arises when estimating (5), however, is that we need to restrict the sample to
only flows eligible to preferences. This raises issues of sample selection, since some of the
determinants of preferential eligibility may also explain utilisation of the preferences; potentially
biasing the coefficient estimates. In order to correct for potential selection bias, we employ a
Heckman procedure and estimate a selection equation for the determinants of preferential
eligibility in (6), where S=1 if the export flow for that product, country and year is eligible for
trade preference and zero otherwise, and use the inverse Mills ratio as an additional regressor
for (5) as a control for potential unexplained factors from preference eligibility.29
29
The inverse Mills ratio is the ratio of the probability density function over the cumulative distribution
function of a distribution. It is used in regression analysis to take account of a possible selection bias.
80
Y*=βX + ε (5)
Y=Y* if S=1
Y is not observed when S=0
S*=γZ + u (6)
S=0 if S*≤0
S=1 if S*≥0
Table 3.3 shows the results of the estimations. Columns (1) and (2) show the results of the
selection model when the dependent variable is a dummy variable for utilisation We mainly use
―gravity‖ geographical and common language variables to identify the selection equation.
Regarding the utilisation equation, we use GDP per capita as proxy for institutional
development. We estimate other specifications using the World Bank cost of doing business
index to proxy ―red tape‖, with similar results.30 These can be found in the Appendix 5, Table
A.14..
For the selection equation on preference eligibility, the results suggest that smaller, more
populated and poorer countries are more likely to be eligible to preferences in the EU, as well as
distant, former colonies, contiguous countries and countries with common language. Finally,
more stringent rules of origin31 increase the probability of preferential eligibility, although it is
difficult to capture the direction of causality since it is possible that more stringent RoOs are
implemented on products with larger preferential coverage.
Regarding the main specification of interest - utilisation of preferences - the results correspond
to what should be expected, although the low level of the pseudo R2 indicates the importance of
unexplained factors in explaining utilisation. Richer countries are more likely to utilise
preferences. As expected, the size of the preference margin available for exporting increases
the probability of preference utilisation. Although the coefficients in the table are the estimated
coefficients and not the marginal effects, the estimated marginal effect of the preference margin
is 2.02, indicating that a 1 percent increase in the preference margin increases the probability
of utilising preferences by 2 percent. Also, more stringent RoO reduce the probability of utilising
preferences.
Concretely, the marginal effect for the RoO index is -0.04, suggesting a mild decrease of -0.04
in the probability of utilisation from increasing 1 level the degree of RoO rigidity. Finally, the
30
Due to the lack of data on costs of doing business for 2008, the panel for this specification is from 2002
to 2007. 31
As RoO index we use the synthetic index developed in Cadot et al. (2007) at the HS-6 tariff level. This
index ranges from 1, very flexible, to 7 very stringent. The index ranks restrictiveness according to
whether involves a change of tariff, subheading, heading or chapter, or in the case of value content
requirement depending on the percentage required.
81
inverse Mills ratio coefficients are negative and statistically significant, indicating the need for
correcting for sample selection since unexplained factors for preference eligibility may impact
utilisation negatively.
In order to analyse the robustness of the results, we re-estimated the same specifications,
changing the dependent variable. Rather than using a dummy variable that measures whether
an eligible trade flow requests preferences, we use as dependent variable the value share of
imports eligible for preferential treatment that use the preferential regime, columns (3) and (4).
In this case, rather than working with flows defined by the tariff regime of entry to the EU, we
use only one flow for each product, country and year. The coefficients confirm the previous
findings.
In conclusion, once corrected for the determinants of preference eligibility, the use of
preferences is correlated with the size of the preferential margin, the flexibility of rules of origin
and how large are bureaucratic costs in the exporting country. As a result, the most accurate
way of assessing the impact of preferential regimes on exports needs to consider preference
utilisation rather than simply eligibility (as is usually implemented in aggregate gravity
estimations). This implies working with effective tariffs paid rather than nominal tariffs or nominal
preferential membership. Furthermore, the results indicate that a positive impact on preference
utilisation, and as result on exports as indicated in Section 4.3, could be achieved by improving
rules of origin and export procedures in export countries.
82
Table 3.3: Determinants of Non Utilisation
Dummy for utilisation Ratio utilisation share
over eligible references
(1) (2) (3) (4)
Selection
Preference
Utilisation Selection
preference
utilisation
GDP_capita -0.3168*** 0.0701*** -0.2906*** 3.1635***
(0.0009) (0.0013) (0.0012) (0.0541)
Distance -0.4404*** -0.3148***
(0.0025) (0.0036)
Contiguity 0.1680*** 0.3815***
(0.0051) (0.0073)
Common
language
0.0182*** -0.0275***
(0.0028) (0.0040)
Colony 0.3245*** 0.2817***
(0.0028) (0.0039)
Preference
margin
5.1004*** 48.8073***
(0.0342) (1.3507)
RoO 0.0933*** -0.0100*** 0.0947*** -1.2926***
(0.0008) (0.0010) (0.0011) (0.0447)
year_2003 0.0500*** -0.0348*** 0.0723*** -3.4209***
(0.0040) (0.0045) (0.0056) (0.1998)
year_2004 -0.0941*** 0.0270*** -0.1206*** -0.8122***
(0.0040) (0.0047) (0.0057) (0.2137)
year_2005 -0.0800*** -0.0190*** -0.1014*** -2.0806***
(0.0042) (0.0050) (0.0059) (0.2291)
year_2006 -0.1206*** 0.0136** -0.1728*** 0.5273*
(0.0042) (0.0051) (0.0059) (0.2359)
year_2007 -0.0336*** -0.0129* -0.0416*** -4.0767***
(0.0043) (0.0051) (0.0061) (0.2337)
year_2008 0.0026 -0.0439*** -0.0248*** -5.4161***
(0.0043) (0.0051) (0.0061) (0.2347)
Lambda1 -0.5726*** -12.9783***
(0.0067) (0.2322)
Lambda2
Constant 5.9564*** -0.3721*** 3.8913*** 63.9515***
(0.0235) (0.0106) (0.0337) (0.4901)
Observations 1459559 901157 792449 792449
Log-likelihood -817875 -604057
Pseudo-R2 0.157 0.0300
Sigma 30.96 30.96
Rho -0.419 -0.419
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
83
3.2.1 Section Summary
This section examines possible determining factors which help to shed light on the degree of
preference utilisation rate. The existing literature notes that non-utilisation is mainly due to the
costs of compliance associated to preferential regimes, such as compliance with product
specific rules of origin, export administration.
Section 3.2 suggests that larger preference margins are not necessarily translated into higher
prices for some preferential regimes. Further, the analysis observed preference utilisation at
very low preferential margins. The extent of non-utilisation of preferences varies across
countries. For most of the countries and products the shares of non-utilisation flows as a share
of total exports is below 20 percent. Overall, the analysis suggests that non-utilisation of
preferences is a polarised phenomenon, which is not important for most countries/products but
very important for a small subset of countries/products.
The analysis also suggests that richer countries are more likely to utilise preferences.
Predictably, the size of the preference margin available for exporting increases the probability of
preference utilisation. However, improving rules of origin and export procedures in export
countries also has a positive impact on the ability of these countries to utilise preferences.
3.3 Price margins – or who captures the preference rent?
The purpose of this section is to analyse empirically who appropriates the rents created in the
EU market by preferential regimes. The importance of trade preferences for exporters primarily
depends on the coverage of trade flows and the utilisation of such preferences. If exporters
have capacity to export products covered by the preferential scheme and the costs of
compliance with the scheme are small enough, tariff preferences provide a competitive
advantage to exporters vis a vis other MFN exporters. However, in addition to coverage and
utilisation, tariff preferences may impact the prices that exporters receive by introducing a
wedge in the border price of that product. Preferences effectively create a rent.
To illustrate the idea of a price rent associated to preferential trade, we can think of exports for a
simple homogenous product x from exporter i to the EU being sold at world prices p* in a
competitive market. A country too small to influence the world price, is entering the EU paying
an MFN tariff at a c.i.f price pxicif = p*/(1+ τ) determined by the equilibrium at ―e‖ and exporting M-
Md. In the short-run, keeping exporters i‘s export share constant, if the tariff is removed there is
a gap p*>pxicif (distance ab in the Figure below) which corresponds to the rent τ pxi
cif. This rent
can be distributed between the exporter and the importer.32 If pxicif rises to p* then there is full
transmission of the preference rent to exporter prices and the exporter appropriates the full
amount. On the other hand, if pxicif remains the same, then importers absorb all the price rent
32
The main assumption here is that the tariff reduction is not passed to the consumer. For example, in a
monopolistic competition setting with Dixit-Stiglitz preferences, the exporter price would remain
unchanged and the price for consumers would be lower, increasing the demand for that variety.
84
that then may (or may not) be passed on to consumers by lowering prices. As a result, one
important question that arises when assessing preferential schemes is who appropriates the
rent, exporters or importers.
Few studies have looked empirically at this issue, although existing evidence suggests lack of
full transmission of preference margins to exporters. Olarreaga and Ozden (2005) for example,
study the impact of AGOA on export prices of African exporters of apparel to the US. The
authors find that only a small share of the tariff rent remained in the hands of African exporters.
Ozden and Sharma (2004) focus on exports of apparel to the US under the Caribbean basin
Initiative (CBI). They find that preferential exporters appropriate two thirds of the preference
margin, increasing their prices 9 percent. Alfieri and Cirera (2008) find for a group of primary
commodities an incomplete pass-through from tariff margin to price margin ranging between 0.4
and 0.6.
Given the rich and comprehensive nature of the dataset provided for this study, we can shed
some light on the issue of price rents moving beyond the study of specific products. Concretely,
we compute the impact of preference margins under GSP and other preferential schemes on
price rents for a sample formed by thousands of products at 10 digits classification and all
exporters to the EU market.
Figure 3.9: Prices and the Preference Rent
p
Imports M
ROW (τ)
D
X (τ)
Md
ROW
a
b
e P*
X
Pcif
85
3.3.1 Methodology
The main challenge when estimating the degree of pass-through from tariff margin to price rents
is the choice of counterfactual. We can observe the price (proxied by the unit value) of a country
exporting under a preferential regime or MFN, but we do not know what the price would have
been under a different scheme. As a result, we need some proxy for the counterfactual price.
There are several approaches in order to find these proxy prices, all of which have
shortcomings.
Quality Differentials
A proxy for the international price, as an approximation to the price that preferential exporters
should receive when exporting under the MFN regime, is the average MFN unit value. The main
problem of this proxy is the fact that there exist large differences in unit values within HS
product categories, likely the result of quality differentials (Schott, 2004). In this case, different
varieties within the same each HS product may be competing in different quality segments and
the average MFN price will not be a good approximation for the price at each quality segment.
One possibility in order to correct for quality problems is to use the ratio of the same country unit
value under MFN in the same period that exporting also under a preferential scheme. The fact
that countries do not always use preferences implies that we may observe exports from the
same country, product and period under different regimes, and, therefore, under different tariffs.
Therefore, under the assumption that quality differentials within exports of the same country and
product are minimal, this may constitute the best proxy.
There are two main problems to this approach. The first problem is that by using only those
observations where we observe in the same country/product/period both, preference utilisation
and non-utilisation, we effectively carry out two sample selections. First, we exclude those
observations not eligible for preferential treatment. Second, for those eligible, we only use those
cases where both utilisation and non-utilisation of preferences are observed and the price ratio
can be computed; excluding those observations eligible for preference that only use the
preference in the same period or only not use it. Thus, if some of the determinants of both
selections also explain the price margin, such as income per capita of exporter, then OLS
estimates of the price margin equation are biased. In order to correct this we need to use a
Heckman (1979) procedure with a selection equation able to control for the different
alternatives. This can be done by employing a multinomial logit framework for selection, where
we explain discrete outcomes such as non-eligibility, utilisation, non-utilisation and both
utilisation and non-utilisation happening in the same period.
A third problem is the fact that non-utilisation of preferences can sometimes be the result of
specific problems at the border such as getting the certificate of origin on time. If this is the
case, we would expect to get the same price under MFN and preferential scheme, because the
exporter would have to face the burden of the sporadic customs inefficiency. This would imply
86
that with this specification there would be no tariff rent transmission to prices. However, in reality
would still be possible that the price under preference could still be higher than the price under
an MFN contract.
Export Pricing
A more important problem is, as suggested above, the price rent appears as a specific case of
homogenous products in competitive markets. Thus, under alternative competition frameworks,
changes in tariffs may change exporters‘ strategic price decisions. Chang and Winters (2002)
for example, show in a Bertrand monopolistic setting how changes in preferential treatment in
MERCOSUR have impacted on prices of exporters. This implies that in addition to controlling for
aggregate price increases, we need to control for the degree of product competition, which may
impact the degree of pass-through.
Transmission to export prices
An alternative approach is to look at whether changes in tariffs impact exporters‘ prices. This is
the general case under which preferential treatment or changes in MFN tariffs are a subset.
Obviously, any changes in MFN tariffs can be linked to a price rent only if there is no full
transmission to consumer prices, but we can still analyse whether changes in tariffs in general
are transmitted to export prices and whether this transmission is different for MFN and
preferential tariff changes.
3.3.2 The Data
We use import data at the country level and disaggregated at HS-10 supplied by the EC. Such a
fine level of disaggregation allows us to minimise quality differences between product varieties33
in the same product category. Trade flows are aggregated each year per country, product and
tariff regime. The tariff regimes are: MFN; GSP, GSP+ or EBA; other preferential regime; tariff
suspension, and; MFN under quota or preferential under quota. In around 80 percent of the
observations we only observe one tariff regime, but on the remaining cases we may observe
two regimes (more than two in only 1 percent of observations). We match import data
observations with tariff data from TARIC.34
Observations with trade flows below 500 Euros are dropped, since they do not represent any
meaningful trade. We use unit values as proxy for prices. Any errors on the inputted values or
quantities reported are likely to generate noisy unit values. For this reason we apply the Hadi
(1992) filter for outliers, for each product and year. After cleaning the dataset we have around
1.5 million observations.
33 We use the term variety to define a product originated in a specific country 34
There are gaps in the tariffs supplied likely the result of some seasonal tariffs not supplied. Also, some
ad valorem conversions have not been possible when there was the need for reference prices. The total
loss of observations represents around 5per cent of the value of imports.
87
Figure 3.10 plots the probability distribution function of the log price ratio for the periods where
both preference utilisation and non-utilisation are observed. The figure illustrates the frequency
of each of the values of the log ratio in the sample. A value of zero corresponds to the logarithm
of value one, and, therefore, the ratio when both prices are equal. The probability distribution
function (pdf) is slightly skewed to the right, however, the average value for the log ratio of -
0.092 indicating higher probability that prices when preferences are non-utilised are larger than
preferential prices. This is reflected in a longer tail to the right of the distribution. In general,
however, this univariate analysis indicates similar recurrence of cases where prices under
utilisation are both smaller and larger than non-utilisation prices. The problem of the univariate
analysis is the fact that we need to control for other factors that may explain prices and also the
determinants of utilising or non-utilising those preferences. Therefore, in order to determine the
real impact of the tariff margin on prices the following section estimates a reduced form equation
for analysing the impact of tariffs on prices.
Figure 3.10: Kernel Estimate of pdf for Log Ratio of Prices
0.2
.4.6
.81
Dens
ity
-10 -5 0 5 10lratio
Histogram Kernel estimate
Normal density
Source: Own calculations using EU database
3.3.3 Econometric Analysis
We construct an export price equation based on the existing literature. In an imperfect
competition setting, prices depend on rival prices (Chang and Winters, 1992), which we proxy
as the average price for that product on the EU market. Second, prices depend on technology
and unit costs that the exporter has for that product, their market power, whether they have a
tariff margin and any costs of compliance related with using preferential schemes.
),,,,(_
prefccpfp (1)
We parameterize equation (1) in logarithms as:
88
ijttjiij
ijt
mfn
jt
ijtjtijt ecpp
)1(
)1(
*32
_
10 (2)
Where the log of the export price pijt depends on the average log price for the product on that
year p-jt, the market share of the country on the same year and product Øijt, the ratio between
the MFN tariff and the effective tariff paid (preference margin), and a set of fixed and time
effects.
We assume that the specific unit cost c ij does not change over time, and in order to estimate
equation (2) we use country product pair λ fixed effects, variety, that controls for all specific
country and product fixed effects.
jiijij c (3)
ijttij
ijt
mfn
jt
ijtjtijt epp
)1(
)1(
*32
_
10 (4)
We estimate equation (4) using two different dependent variables. The first specification, the
price ratio specification, uses only the ratio between preference utilisation and non-utilisation
unit values when these are observed in the same period. The second specification, the export
price specification, uses the import unit value. Table 3.4 shows the results when using the
restricted sub-sample where price ratios can be computed; this is when both utilisation and non-
utilisation are observed. We report both OLS and variety (product for each country) fixed effects
with year dummies. Increasing the country‘s market share (as proxy of market power) on this
product tends to increase the price margin from preferences. Increasing the average price for all
exporters tends to reduce the price margin. Finally, and most important, increasing the
preference margin is positively transmitted to the price margin, with a pass-through close to
perfect pass-through. This result is confirmed when we use the tariff rates when utilising and
non-utilising preferences as separate regressors. Increases in preferential tariffs reduce the
price ratio by reducing the preference margin, and increases in MFN tariffs increase the price
ratio by increasing the margin.
As suggested above, the results for this specification are only indicative since the sample is
reduced to around 340,000 observations, which is the number of observations when we can
observe both utilisation and non-utilisation of preferences for the same country, period and
product. Therefore the estimates are likely to experience sample selection bias. Furthermore,
the results show a very low R2, indicating lack of explanatory power for price variations. This is
likely the result of not having information on variety costs, which is likely to be the main
determinant of prices and their variation.
89
Table 3.4: Export Price Ratio Specification
(1) (2) (3) (4)
OLS1 FE1 OLS2 FE2
Average price -0.0334*** -0.1967*** -0.0330*** -0.1969***
(0.0025) (0.0060) (0.0025) (0.0060)
Market share 0.0052** 0.0060*** 0.0049** 0.0059***
(0.0019) (0.0018) (0.0019) (0.0018)
Preference margin 1.0245*** 0.8008***
(0.0624) (0.1180)
Tariff pref -0.7624*** -1.0275***
(0.1004) (0.1547)
Tariff MFN 1.0692*** 0.4960**
(0.0644) (0.1790)
year_2003 0.0254* 0.0120** 0.0254* 0.0117**
(0.0101) (0.0040) (0.0101) (0.0041)
year_2004 0.0221* 0.0105* 0.0213* 0.0100*
(0.0108) (0.0044) (0.0108) (0.0044)
year_2005 -0.0448*** -0.4321*** -0.0453*** -0.4331***
(0.0123) (0.0148) (0.0123) (0.0148)
year_2006 -0.0431*** -0.4341*** -0.0446*** -0.4351***
(0.0127) (0.0146) (0.0127) (0.0146)
year_2007 0.0366** 0.0406*** 0.0348** 0.0400***
(0.0116) (0.0051) (0.0116) (0.0051)
year_2008 0.0538*** 0.0551*** 0.0517*** 0.0544***
(0.0121) (0.0052) (0.0122) (0.0052)
Constant -0.0783*** 0.2763*** -0.0856*** 0.2996***
(0.0120) (0.0159) (0.0123) (0.0190)
Observations 333945 333945 333945 333945
R-squared 0.0069 0.0054 0.0071 0.0054
Number of variety 99985 99985
R2 within 0.0054 0.0054
R2 between 0.00436 0.00398
R2 overall 0.00408 0.00367
log-likelihood -220720 -220716
Robust standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
In order to correct for potential sample selection bias from reducing our sample to those periods
where both utilisation and non-utilisation are observed, we need to implement a selection
procedure. We follow Bourguignon et al. (2004) and estimate a multinomial logit model for the
different utilisation alternatives. Concretely, we estimate the following equation, where Yi is a
discrete variable with value 0 to 3 according to whether a trade flow is only MFN eligible,
preferences fully utilised, preferences non-utilised, or both, as compared to using an MFN
regime. The price ratio is only observed for Yi=3
90
Y*=βX + ε, for Y=0,1,2, 3
P*=βX + ε
P=P* if Y=3
The interpretation of the estimated coefficients in the selection equation is complex, and needs
to be understood as the impact of each variable with respect to the baseline category, MFN
eligibility. The objective of the selection equation is to control for the sample selection bias,
rather than eligibility and utilisation (See Section 3 on the determinants of utilisation). In order to
explain the different utilisation regimes, we use an index that measures RoO rigidity, a dummy
variable with value one is the good is a homogenous good according to Rauch‘s classification
and GDP per capita as the identifying variables for the selection equation.
Table 3.5 shows the results for the selection equation. Since the selection model is a
multinomial Logit, the coefficients need to be interpreted in relation to the baseline category, the
MFN regime. Larger MFN tariffs and smaller applied tariffs increase the probabilities of both,
preference utilisation and non-utilisation, vis-a-vis MFN eligibility; via increasing the preference
margin. That is, larger margins increase the probability that a trade flow is eligible for
preferences and these preferences are used or not used, compared to the trade flow being
eligible to MFN treatment. Income per capita reduces both, preference eligibility and utilisation,
since richer countries are less likely to receive preferences. Stringent RoOs reduce utilisation
and homogenous goods according to Rauch‘s classification are less likely to being eligible for
preferences.
Table 3.5: Multinomial Logit for Selection Utilisation
(1) (2) (3)
utilisation non-utilisation utilisation & non-utilisation
Applied tariff
-10.6679***
-12.7839***
-12.5863***
(3.1430) (3.1430) (3.1430)
MFN tariff 19.4376*** 24.7515*** 25.6474***
(3.1435) (3.1449) (3.1440)
GDP_capita -0.4485*** -0.3876*** -0.3371***
(0.0020) (0.0045) (0.0042)
RoO index 0.0309*** -0.0384*** -0.0048
(0.0020) (0.0039) (0.0036)
Homogenous -0.9826*** -1.7230*** -1.6781***
(0.0144) (0.0271) (0.0249)
year_2003 0.0197* 0.0975*** 0.1599***
(0.0089) (0.0182) (0.0173)
year_2004 -0.1562*** -0.1784*** -0.2722***
(0.0089) (0.0186) (0.0177)
year_2005 -0.5712*** -0.4328*** -0.6964***
(0.0098) (0.0197) (0.0189)
91
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Once we have estimated the selection equation, we can use the estimated selectivity terms in
the price ratio equation and corrected for selection. Table 3.6 reports Bourguignon et als' (2004)
preferred method. When this method is used the preference margin pass-through is halved to
0.51. (Table A.19 in Appendix 5 compares the results using different methods).
Summing up, the estimations suggest that preference margins are transmitted to exporters,
although the degree of pass-through is reduced to around 0.5 when we control for potential
sample selection.
year_2006 -0.5319*** -0.4541*** -0.6973***
(0.0096) (0.0197) (0.0189)
year_2007 -0.4923*** -0.5156*** -0.7141***
(0.0101) (0.0202) (0.0193)
year_2008
Constant 2.6030*** -2.3160*** -1.8320***
(0.0196) (0.0472) (0.0431)
Observations 1245924 1245924 1245924
Pseudo R2 0.469 0.469 0.469
log-likelihood -866249 -866249 -866249
92
Table 3.6: Export Price Ratio Specification with Multinomial Selection (pref. margin)
VARIABLES Bourguignon
Average price -0.0084
0.0017
Market share -0.0033
0.0008
Preference
margin
0.5154
0.0440
year_2003 0.0049
(0.0062)
year_2004 0.0569***
(0.0067)
year_2005 0.1037***
(0.0097)
year_2006 0.0934***
(0.0096)
year_2007 0.1057***
(0.0082)
year_2008
m1 -3.2770***
(0.3162)
m2 -0.6589***
(0.1135)
m3 0.4366***
(0.1608)
Constant -1.0711
0.0770
Observations 283332
R-squared 0.0076
Robust standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
To check the robustness of the results, we also estimate equation (4) using the export price as
explanatory variable. This allows us to use the entire dataset, without the need to control for
sample selection. Table 3.7 shows the results when analysing the degree of pass-through to
export prices. Since we do not compare prices from the same country as in the previous
specification, we need to control for quality differentials. Any variety specific quality issues will
be absorbed by the fixed effects, and we also control for country quality differentials between
countries with GDP per capita. In addition, we add a dummy for those export flows
corresponding to non-utilisation episodes to check whether in these cases export prices are
lower or higher.
93
The average product price has a positive impact on the export price, indicating similar sign of
rival response or positive price trends on average for each specific market. The country‘s
market share, the proxy for market power, increases the export price as expected. Somewhat
puzzling are the coefficients on GDP per capita, which is consistently negative although
marginally significant, and on non-utilisation of preferences, which is positive. If variety fixed
effects can perfectly control for quality differentials, then the negative sign on income per capita
could be explained by higher cost competitiveness in richer countries. In addition, non-utilisation
episodes have higher export prices, which may indicate that part of the additional tariffs paid by
exporters, are transmitted to their price.
Regarding the two main variables of interest, the tariff rate applied and the preference margin,
the results are similar to the previous specification. There is a positive pass-through elasticity of
0.64 from preference margins to export prices. When the preference margin effect is
decomposed using interactive dummies with the effective regime of entry, an interesting result is
the fact the positive pass-through disappears for exports under EBA and GSP, although the
coefficient on the former is not statistically significant. This result suggests that preference
margins are positively transmitted to export prices mainly for Cotonou and other FTA regimes.
Again, one problem of the estimates is the very low R2, which indicates very low explanatory
power of the estimated specifications on explaining overall export prices. The most likely reason
for this is the lack of any data on costs for each product and country, which is the most
important determinant of prices.
These results are confirmed when using the effective and MFN tariffs separately as regressors
rather than as a ratio. Larger effective tariffs reduce prices by reducing the margin, and larger
MFN tariffs increase export prices by increasing the margin. We also include the decomposition
of the tariff effect on export prices by preferential regime. Unfortunately, this decomposition is
not very meaningful since most preferential tariffs are zero and, therefore, not possible to
identify over non-preferential tariffs. As a result, the coefficients are not statistically significant.
Summing up, preferential margins are positively transmitted to price margins and export prices.
However, it is less clear that there is positive transmission of margins when the preferential
regime used is GSP or EBA.
94
Table 3.7: Export Price Specification
Robust standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
(1) (3) (4) (5) (7) (8)
OLS1 FE1b FE1c OLS2 FE2b FE2c
Average Price 0.9377*** 0.4722*** 0.4721*** 0.9392*** 0.4718*** 0.4718***
(0.0008) (0.0018) (0.0018) (0.0008) (0.0018) (0.0018)
Market Share 0.0158*** 0.0408*** 0.0409*** 0.0154*** 0.0408*** 0.0408***
(0.0004) (0.0006) (0.0005) (0.0004) (0.0006) (0.0006)
Preference margin 0.2930*** 0.6415*** 0.1146**
(0.0194) (0.0234) (0.0399)
non_utilisation 0.1176*** 0.1176*** 0.1188*** 0.1194***
(0.0018) (0.0020) (0.0018) (0.0019)
GDP_capita -0.0285* -0.0303* -0.0281* -0.0276*
(0.0125) (0.0125) (0.0125) (0.0125)
Margin*cotonou 0.7194***
(0.0716)
Margin*pref 0.9667***
(0.0499)
Margin*eba -0.0021
(0.0915)
Margin*gsp -0.3839***
(0.0693)
Tariff paid 0.0395 -0.6646*** -0.6669***
(0.0211) (0.0235) (0.0240)
MFN tariff 0.4972*** 0.1992*** 0.2011***
(0.0223) (0.0523) (0.0525)
Tariff*cotonou 0.0698
(0.9403)
Tariff*pref -0.2263*
(0.0917)
Tariff*eba 0.5256
(0.7732)
Tariff*gsp 0.0615
(0.0421)
Constant -0.0506*** 1.2483*** 1.2635*** -0.0774*** 1.2693*** 1.2646***
(0.0029) (0.1031) (0.1031) (0.0031) (0.1031) (0.1031)
Observations 1568723 1481623 1481623 1568723 1481623 1481623
R-squared 0.8159 0.7731 0.7733 0.8161 0.7731 0.7731
Number of variety2 436652 436652 436652 436652
R2 within 0.7731 0.7733 0.7731 0.7731
R2 between 0.711 0.711 0.712 0.712
R2 overall 0.714 0.714 0.715 0.715
log-likelihood -1.025e+06 -1.025e+06 -1.025e+06 -
1.025e+06
95
3.3.4 Section Summary
This section examined the extent to which the exporters in the beneficiary countries appropriate
the rents created in the EU market by the preference regimes. A number of estimations are
undertaken in this section, based on various methods, including Bourguignon et. al. preferred
method, We find that both market share and the size of the preference margin is positively
transmitted to the price margin, which suggests that preference margins are transmitted to
exporters. The extent of this pass-through is reduced to around 50% when we control for
sample selection. The analysis also suggests that preference margins are positively transmitted
to export prices mainly for Cotonou and other FTA regimes, it is less clear that there is a
position transmission of margins under the GSP or EBA preferential schemes.
96
3.4 Section 3: Conclusions This section considered the extent to which countries actually utilise these preferences, what
determines the degree of utilisation and if there is any evidence on the extent of any impact of
that utilisation. The evidence suggests that utilisation rates are typically high, but with of course
some variation across countries. We identify those countries whose structure of exports is such
that the GSP regimes may not be well suited on the basis of both low rates of utilisation and low
shares of trade eligible for preferences; as well as those where the converse is true.
The discussion also considered the possible determining factors which help to shed light on the
degree of preference utilisation rate. The existing literature notes that non-utilisation is mainly
due to the costs of compliance associated to preferential regimes, such as compliance with
product specific rules of origin, export administration.
The analysis suggests that larger preference margins are not necessarily translated into higher
prices for some preferential regimes. Further, the analysis observed preference utilisation at
very low preferential margins. The extent of non-utilisation of preferences varies across
countries. For most of the countries and products the shares of non-utilisation flows as a share
of total exports is below 20 percent. Overall, the analysis suggests that non-utilisation of
preferences is a polarised phenomenon, which is not important for most countries/products but
very important for a small subset of countries/products.
The analysis also suggests that richer countries are more likely to utilise preferences.
Predictably, the size of the preference margin available for exporting increases the probability of
preference utilisation. However, improving rules of origin and export procedures in export
countries also has a positive impact on the ability of these countries to utilise preferences.
The extent to which the exporters in the beneficiary countries appropriate the rents created in
the EU market by the preference regimes was also examined. A number of estimations are
undertaken in this section, based on various methods, including Bourguignon et. al. preferred
method, We find that both market share and the size of the preference margin is positively
transmitted to the price margin, which suggests that preference margins are transmitted to
exporters. The extent of this pass-through is reduced to around 50% when we control for
sample selection. The analysis also suggests that preference margins are positively transmitted
to export prices mainly for Cotonou and other FTA regimes, it is less clear that there is a
position transmission of margins under the GSP or EBA preferential schemes.
97
4 Gravity Modelling Gravity models are an extremely important component of the applied economist‘s toolkit, and
have been widely and successfully used in a very wide range of context. The basic gravity
modelling framework assumes that trade between countries will depend on their respective
sizes and income levels, the distance between them, any common cultural/linguistic factors, and
then on key policy variables (such as being a member of a regional trade agreement, having a
common currency….). A gravity model is thus typically used in order to assess the impact of
either differences in policy or changes in policy on flows of goods, services, and investment
between countries. Gravity models can thus assess the aggregate and (if correctly specified)
sectoral impact on trade flows on a given country or country groupings, shed light on the
possible welfare consequences, as well as on the impact on trade creation and trade.
For the purposes of this report we undertake three complementary sets of gravity modelling
exercises. These are, firstly aggregate modelling of trade and investment; secondly more
disaggregated modelling of trade at the sectoral level based on particular sectors identified in
the preceding sections of the report; and thirdly highly disaggregated analysis of trade between
the EU and developing countries in order to ascertain with a degree of accuracy that has not
previously been possible the extent to which the preference margins implied by the different
regimes impact on trade flows.
4.1 Aggregate modelling of trade and investment
In the following applied analysis we study the effect of unilateral trade preference given to
developing countries by the EU since the 70s both on exports from these countries and on FDI
flows to these countries. We base our econometric analysis on three different versions of the
gravity equation, where we aim to identify various possible explanatory factors for the
determinants of first, bilateral exports and secondly FDI outflows.
4.1.1 The target model
The classical gravity equation constitutes a device commonly used to estimate the effects of
many different phenomena in international trade. Timbergen (1962) and Poyhonen (1963)
developed its first applications in international economics. In its most elementary version, the
equation establishes that the volume of trade flows between two countries depends positively
on their economic dimensions, measured by the level of their GDP and population, and
negatively by the transport costs captured by the absolute distance between their biggest
economic centres.
The name of the model derives from the analogy with Isaac Newton‘s theory on the gravitational
attraction of two masses according to which the bigger the sizes of the masses and the smaller
the distance, the greater will be the attraction. Linnemann (1966) added population as a further
element of the dimension of the country, specifying his model as the following:
ijijjijiij DistPPYYX lnlnlnlnlnln 54321
98
where Xij is the value in dollars of the aggregate trade flows from country i to country j; Yi and Yj
are the GDP of these countries; Pi and Pj are their populations and Distij is the absolute distance
in kilometres between them; εij is the error with distribution N(0,2) according to the parametric
assumption.
In the subsequent versions of this model some dummy variables have been introduced with the
aim to capture the effects of either geographical or institutional factors, which increase or shrink
the distance between two countries. The result is an augmented gravity equation that includes
three types of determinants of bilateral trade flows: i) characteristics of supply in the exporter
country, ii) characteristics of demand in the importer country and iii) elements which favour or
obstruct the specific trade flows (common border, common language, past colonial links and
geographical characteristics).
The most common way to consider the effects of regional integration into the extended gravity
equation is to include dummy variables for the RTAs in force during the sample period. Each
dummy takes the value of one if the bilateral trade in the dependent variable is between two
countries which are in the same RTA, zero otherwise. With this specification of the model, the
effects of preferential trade policies are defined as deviations from the volume of trade predicted
by the baseline extended gravity equation. Similarly, we can include dummies to capture the
effect of unilateral preferences such as GSP, GSP +, EBA and Cotonou.
Multiple theoretical approximations exist in order to justify the gravity equation. Anderson (1979)
proposed a theoretical foundation based on the Armington hypothesis: that is, consumers
differentiate the products on the basis of their origin. In its more complete versions this model
includes non-tradable goods, tariffs and transportation costs. The preferences about tradable
goods are represented by a CES utility function with constant elasticity of substitution and non-
homotheticity of preferences between tradable and non-tradable goods. Bergstrand (1985)
developed a general equilibrium model of world trade with products differentiated on the basis of
their country origin. The preferences of the consumers are described with a CES utility function,
with the possibility that the elasticity of substitution between imported goods differs from that
existing between imported and domestic goods. In the supply side there is only one production
factor, which is not internationally mobile and whose allocation between different markets
depends on the production function with constant transformation elasticity, such that the
transformation elasticity between the domestic and foreign production is different from that
defined between distinct foreign productions.
The traditional gravity equation is achieved after assuming that each market in question is small
compared to the rest of the world and technologies and preferences are the same in the rest of
the world. These conditions give the ―generalized gravity equation‖ which is a gravity equation
without restrictions on the parameters. Bergstrand (1985) developed another general
equilibrium model which he called ―H-O-Chamberlin-Linder‖ and used it to derive a new version
of the generalized gravity equation. The economies have two sectors in a context of
monopolistic competition and there are two production factors, labour and capital, whose
99
relative endowments differ across countries. This theoretical structure is a conjunction point
between the Hecksher-Olin (H-O) model, which implies a context of perfect competition, and the
models with one sector based on monopolistic competition.
The aim of this work was showing that the gravity equation could be compatible with either the
inter-industry trade described by the H-O model or the intra-industry trade described instead in
the Helpman-Krugman model. Bergstrand provided an alternative interpretation of the
explanatory variables: the GDP of the exporter country can be seen as an approximation of the
product in terms of unities of capital whereas its per-capita GDP is an approximation of the
capital/labour ratio; on the other hand, GDP and GDP per capita of the importer country can be
seen as its expenditure capability and non-homothetic preferences. Anderson and Wincoop
(2003) show theoretically that the gravity equation should include an implicit price index, given
by the level of prices in both countries and the trade costs, that they call it ―multilateral trade
resistance‖.35 This term reflects the openness of the importer country to all goods and the
openness of the world to the exporter country‘s goods. Indeed, Deardorff (1997) argues that the
gravity equation does not prove the validity of one theory or another but just confirms a ‗fact of
life‘.
We generalize the gravity equation in order to explain with this economic device bilateral trade
and FDI outward stocks between countries. In particular FDI outward stocks are treated as trade
in the original version of the model and explained by geographical and institutional distance and
economic size of the country partners: bilateral FDI flows are positively determined by the
economic size of the partners and negatively by the distance (geographical, cultural and
institutional) which separates them. This generalization of the gravity equation, already used in
some recent application (see Brenton et al.1999, Eaton and Tamura 1996, Bevan and Estrin,
2000) derives from a small part of the theoretical literature on international trade based on
general equilibrium models (see Markusen and Venables, 2000). In these models multinational
activity is endogenous and driven by the trade off between costs of establishing a new plant
abroad in order to supply the domestic market and costs of exporting in that market. Trade and
FDI are therefore substitute in this theoretical context and the choice made by the firms is a
function of the specific characteristics of the two countries that can be captured by their
economic size and relative costs of transaction (such as transport costs or costs determined by
institutional differences).
Our main aim is to evaluate the impact of the unilateral preferences given by the EU on trade
and FDI.
The specification of the gravity model is the following:
35
Other authors call the same term ―remoteness”.
100
ln X ijt 1 ln popit 2 ln pop jt 3 lnYit 4 lnY jt 5 ln dist ij 6bord ij
7comlang ij 8landlocked j 9landlocked i 10landarea j 11landarea i
12colony ij nRTAijtn
n
mGSPijn
m
ryeart
t
i j ij ijt
where i is the exporter and j the importer in the model for trade (and i is the country that
receives the foreign investment, j is the investing country in the model for FDI), t denotes time
and the variables are defined as:
- Xijt is the value of deflated exports to j from i (deflated FDI outwards from i to j) ;
- Y is the deflated GDP at time t;
- pop is population at time t;
- landarea is the area of the country;
- dist is the distance between i and j;
- lang is a dummy taking the value one when i and j have a common language, zero
otherwise;
- bord is a dummy taking the value of one if i and j share a border;
- comlang if the countries have the same main official language;
- colony is a dummy taking the value 1 if the countries have a past colonial relationship;
- landlocked is a dummy taking the value one if the country is enclosed by land;
- m
m
ijtGSP is a set of dummies for all the generalized systems of preferences since the
1970s. Each of those dummies, takes the value of one when I exports to j and i received
trade preferences from j.
-
RTAijtn
n
is the set of 3 dummy variables for EU, RTAs between the EU and other
countries and non EU RTAs. Each of those dummies, takes the value of one if i and j are
both in the kind of regional trading bloc that the dummy characterize, zero otherwise,
-
year t
t
is a set of 11 dummy variables which captures time specific effects,
- i is the exporter (host) effect;
- j is the importer(investing country) effect;
- ij is the country pair effects;
- ijt is the error, normally distributed and with zero mean.
4.1.2 Data
Trade data come from the COMTRADE dataset and cover all the available bilateral imports
between 183 countries (with gaps) over the period 1996-2008. Trade values are reported in
millions of dollars. The outward stocks of FDI data comes from the OECD International
Investment statistics on-line dataset. The dataset covers (with many gaps) the period 1996-
2007 and contains data on investment stocks of OECD countries in their international partners.
The annually reported values are in US dollars and this should approximate a correction for the
differential in exchange rates across countries. We have therefore deflated both trade and FDI
101
values with the US GDP deflator provided by World Bank's Global Development Finance
database whose base year (GDP deflator=100) is 2000. Population and real GDP data (with
some gaps) have been obtained from a standard source: the World Bank‘s World Development
Indicators.
We used CEPII data for the country-specific variables: landlocked status, distance, physically
contiguous neighbours, language and common colonial past. Absolute distance data refers to
the distance between the capitals of each pair of countries. The list of RTAs entered into force
before December 2008 has been derived from the WTO web site (www.wto.org). Information on
the GSP schemes so far comes from the UNCTAD website
(http://www.unctad.org/Templates/Page.asp?intItemID=1418&lang=1).
4.1.3 Estimation and Results
Econometric issues: modelling bilateral FDI flows by correcting for sample selection or
unobserved heterogeneity.
The specifications of the gravity model follow the empirical strategy of controlling for as many
natural and institutional causes as possible, and then looking for the effects of GSP schemes in
the residuals. However, after specifying the model and the empirical strategy we faced a
problem: how to estimate the model? As Cheng and Wall (2004) pointed out, the perceived
success of the gravity model has always been stated on the basis of goodness of fit, the R2,
which is usually high, without any further analysis regarding its econometric properties.
Most studies using the gravity equation implement the estimation with ordinary least squares
(OLS) either on cross-sectional or pooled cross-sectional data. However, some analysts have
recently recognized a bias in this estimation technique of the gravity equation by basing their
assessment on its theoretical derivation (Cheng and Wall, 2004). The theory behind the model,
in fact, seems to have shed light on specific factors (also called unobserved heterogeneity) that
characterize bilateral relationships and are omitted in the standard augmented gravity equation.
These factors can be captured by fixed effects estimators, even though there is small
agreement about the specification of these fixed effects.36 Therefore this suggests that the right
way to carry out our analysis would be a panel estimation of the gravity equation. In particular
we will consider two types of unobserved heterogeneity: country pair heterogeneity and importer
and exporter (investing and host country for FDI) heterogeneity.
Studying the determinants of international bilateral trade (and outward stock of investments) by
using a gravity model would imply focus only on positive values without taking into account the
selection mechanism of the outcomes. A problem of sample selection bias arises if some
component of the trade (investment) decision is relevant to determine the level of trade
(investment) but it is not taken into consideration in the regression analysis. Controlling for the
36
See Anderson and Wincoop (2003), Glick and Rose (2001) and Mátyás (1997).
102
observable characteristics when explaining trade (investments) with the gravity equation is
insufficient as some additional process is influencing the level of trade (investment), namely, the
process determining whether a country trades with the partner (invests in the partner), or does
not trade at all (does not invest at all or disinvests). If these unobservable characteristics are
correlated with the observables then the failure to include an estimate of the unobservables will
lead to incorrect inference regarding the impact of the observables on investment (Vella, 1998).
The probable non-randomness of the sub-sample chosen (that is, in this case, only positive
values of bilateral trade flows outward FDI stocks) if ignored in the context of an econometric
estimation, might create problems of misspecification and therefore inconsistent estimates37.
The underlying idea is that important information, derived from the selection process, would be
excluded by running a regression only on positive values (that is proper investments). Following
Heckman‘s approach (1979) in fact, it is possible to demonstrate that the conditional mean we
are interested in obtaining with our application is actually:
E[lnX|Z, X>0] = Z’β + σλ(X’β) instead of E[lnX|Z, X>0] = Z’β,
That is what we would get by carrying out a least square regression analysis (where lnX is the
natural logarithm of bilateral trade flows or FDI outward stocks, Z is the set of our exogenous
regressors and λ(X‘β/σ) is the term which corrects for the presence of unobservables, which
affect both the investment decision and the level of investments and is called the inverse Mills
ratio in the theoretical literature). As explained by Heckman, estimating the model without the
correction term would mean to omit a regressor and commit a misspecification error which could
generate inconsistent estimates of our parameters of interest.
It is then necessary to test for the presence of selection using Heckman‘s two step procedure.
This procedure augments the regression of the variable of interest on its exogenous regressors
by an estimate of the omitted term. This estimate is obtained by a first step probit regression
which represents the selection equation and gives the probability of positive outcome (that is
participation) happening. In order to identify the parameter of interest in the second stage, the
first step probit equation should contain at least one instrument, that is, one variable that does
not appear in the model of interest.38
37
Silva and Tenreyro (2006) solve the problems of the zero in the dependent variable by using a Poisson
pseudo-maximum likelihood estimator. However this estimator imposes strong assumptions on the
conditional mean and variance of the dependent variable. Econometric models are the smallest (and
often arbitrary) sets of assumptions that are required to identify the parameter of interest. Here we
decided to impose a different kind of assumptions.
38However, Heckman described a model in which the possible outcomes in the selection are only two
(participation or non-participation). Therefore we need to extend his procedure to a more general case in
which the outcomes can be more than two. In our application the outcomes are three, as already noted:
investment, non-investment and dis-investment. This leads to a first stage that implies the estimation of a
multinomial logit (instead of a probit) to model the decision mechanism which produces three outcomes
(investing, not investing or disinvesting) and the second stage models the investments with a gravity
103
In our analysis we present the estimates obtained with the Heckman two-step procedure as
well, however we argue that the results obtained with the panel framework are less biased and
take into account more heterogeneity, therefore cleaning the residual in a more meaningful way.
4.1.4 Trade
With a dataset of aggregate bilateral import flows from 1996 to 2008 we analyze the impact of
the Generalized System of Preferences on imports from beneficiary countries to those countries
that have implemented these schemes of preferences.
We assume that bilateral imports are explained by an extended version of the gravity equation
that includes a set of dummies for the existing GSP schemes. Each dummy is equal to one if
the importer country is the one that implemented the scheme and the exporter country is a
beneficiary within that scheme. In particular, for all the agreements and schemes that involve
the EU, we consider the membership of the partners as mutually exclusive according to a
specific hierarchical relationship across these agreements and schemes (Cotonou > GSP EBA
> GSP PLUS > GSP).
We consider three different econometric models based on the gravity equation that may capture
the data generating process behind bilateral trade flows.
In the first model, defined as Tobit type 2 (and estimated with the Heckman 2 step procedure),
we consider the possibility of endogenous selection in positive trade. We explicitly model the
propensity (or probability) to trade of a country pair and allow the unobservable characteristics
of this pair, which contribute to the decision of trading, to be correlated with the unobservable
characteristics of the pair that determine the actual level of trade. We do that by using an
exclusion restriction in the equation that explains the propensity to export to a country. The
exclusion restriction is a time variant score for the economic freedom in the exporting country.
This score is provided by the Heritage foundation. The economic theory that lies behind is that
countries with higher fixed costs to exports will be less likely to export. And fixed costs are likely
to be explained by frictions in the economy such as lack of economic freedom.
In the second model we take into account the country pair heterogeneity with pair trade-
direction specific fixed effect (this means that the pair in which x import from y it is considered
different from the pair in which y imports from x). This model allows for control of all the
unobservable characteristics of the country pairs that are time invariant and are likely to be
correlated with observed explanatory variables.
Finally we consider a model in which we take into account importer and exporter heterogeneity
through a double set of fixed effects. This model allows us to control for all the unobservable
characteristics of importers and exporters that are time invariant and are likely to be correlated
equation which includes a correction term for the selection. This two- stage estimation is complicated by
the longitudinal dimension of the available data.
104
with observed explanatory variables. In this specific context the coefficients identified by this
model will be our preferred. In the last two models we will consider the selection as time
invariant and fixed effect specific and therefore swiped away by the fixed effects estimation or
equivalently exogenous. The full set of results is given in Table A.14 in Appendix 5.
The coefficients for the original gravity model variables across the three models are similar to
the ones previously found in the literature (see Baldwin and Taglioni 2006). The signs of the
coefficients for the standard variables are coherent with several other studies that employed the
same specification of the model. They represent the elasticities of exports to exporter and
importer GDPs and populations, and to the distance between them. The exports are expected to
be positively related to national incomes, whose coefficients should be positive and not
significantly different from one.
The expected signs on the population variables are not unambiguous in the literature: in the
aggregate trade there can be positive coefficients for both exporter and importer that, according
to Bergstrand (1989), can indicate respectively, labour-intensive exports or exports of necessary
goods. The sign for the absolute distances between trade partners is negative and slightly
greater than one, as usual. The coefficients for the dummies capturing the institutional and
geographical factors which obstruct or favour trade have the expected signs: they have to be
interpreted as the deviations from the prediction of the baseline gravity model.
However the coefficients for the dummies of interest – the ones that represent the GSP
schemes – are new to this literature. Looking at the Tobit type 2 estimates, obtained with a
Heckman two step procedure, we can see that all of the GSP schemes seem to positively affect
the probability of trading (―selection‖ column). However, when we look at the level of trade
conditional on the positive selection into trade we can see that the beneficiaries countries
appear to trade systematically less with their partners that have the preferences than the
remaining non-preferential countries. However, it is important to note, this result can be the
effect of a strong omitted variable bias due to the fact that the traditional variables of the gravity
equation are not able to capture all the elements that affect a bilateral trade relationship, even
after controlling for endogenous selection.
A very complete model that takes into account all the time invariant unobservables that affect
bilateral trade is one that includes country pair (trade direction specific) fixed effect. The
drawback of this model is that it drops from the estimation all the variables that are time
invariant for the country pair over the period considered: we do not identify coefficients for
distance, common language, common colony and common border. Equally most of the GSP
scheme dummies in this model are not identified: the reason is because these schemes existed
since the 70s and therefore they were time invariant for the period covered by the data (1996-
2008). The EU_GSP, EU_GSP_PLUS, EU_GSP_EBA and Cotonou coefficients are identified
since over time (from 1995) there is a change in the membership of both the EU and the
scheme participants (with the introduction of GSP PLUS in 2006 and EBA in 2001 and the
cessation of Cotonou in 2007). But mainly these coefficients pick up the effect of trade between
the preferential partners and those countries that entered the EU after 1996, and this might
105
explain the negative sign associated with them - the dummies only pick up the trade between
the new members and the beneficiary countries.
For these reasons we prefer the importer and exporter fixed effects model where, although
some of the richness of time invariant controls is lost, the coefficients for GSP schemes are
identified by all available data points over the all sample of data. All the coefficients for schemes
involving the EU countries as importers now have positive and significant coefficients.
In this model the implied percentage variation which they represent is given, on average and
ceteris paribus, by the following formula: [exp(coefficient)-1]*100. The regression therefore
suggest that the increase in exports to the EU as result of the respective preferential schemes is
as summarized in Table 4.1. Here we see a positive impact of trade which is strongest for the
GSP+ and EBA countries, and somewhat weaker for the GSP and Cotonou countries. All of
these results are statistically significant. These results must be interpreted as increases in EU
imports from countries that benefited from a preferential scheme where the baseline is given by
all of the other countries that export to the EU and that weren't included in any preferential
scheme.
Table 4.1: Percentage Change in Aggregate Trade
Scheme % increase in exports to
the EU countries
GSP 15.48%
EBA 25.86%
GSP PLUS 29.04%
COTONOU 10.62%
Source: Own calculations
4.1.5 FDI
For FDI we use data on outward stocks of OECD countries in the rest of the world, from 1997
to 2007 (with many missing value). We model FDI flows with a gravity equation extended with
two sets of dummies. The first set of dummies includes one for each system of preferences
agreed and each dummy equals one for all beneficiary country, independently on the investing
OECD country partner. These dummies are meant to identify the effect, if any, on aggregate
FDI in host countries that are included in the preference scheme. The idea behind this is to see
if trade preferences from more developed countries made them more suitable as hosts for
foreign direct investments irrespective of the source of that investment.
The second set of dummies aims to characterize the investment relationship between EU
countries (that are part of the OECD) and the beneficiary countries of the schemes GSP, GSP
Plus, EBA and Cotonou. In particular, each of these dummies equals one if the investing
country is an EU/OECD country and the host country is a beneficiary of the scheme.
106
We consider again 3 different econometric models based on this extended version of the gravity
equation: Tobit type 2 for the endogenous selection into (positive) investment, country pair fixed
effects and investor and host country fixed effects, and the full set of results is given in Table
A.15 in Appendix 3. As previously discussed in the context of the trade regressions in principle
the model most suitable for our analysis is the investing and host country fixed effect that allows
us to identify the coefficients of interest with more observations and therefore variation. As we
can see from the table, when we look at these estimates all the coefficients of interest are
identified, they are all positive and very similar in their magnitude. This seems to suggest that
FDI outward stocks of the OECD (non-EU) countries have risen in those countries that have
been given preferences from the EU through the GSP and EBA schemes, respectively by 37%
and 179%. However because of multicollinearity that arises for the presence in the model of
many country and trade policy dummies and the lack of complete time series of FDI outward
stocks for each country-pair, we cannot identify the dummies that equals one if the investing
country is an EU country and the host country is a beneficiary of the scheme.
Table 4.2 summarises the estimated impact on investment into the beneficiary countries
associated with each of the regimes, from the country pair fixed effect model. The impact
appears to be potentially very large, with increases in investment in comparison to non-
beneficiary countries in excess of 200%. These numbers should be treated with a high
appropriate degree of caution, as there are a substantial number of missing observations.
Indeed if we compare the trade with the FDI regressions we see that for the FDI regressions we
have less than 6% of the number of observations that we have for the trade regression.
Moreover the distribution of log FDI stocks that are available, has a very long right tail -that is
there are a relatively few very high FDI stocks – some of which are in countries that were
included in the EU GSP schemes. Their relative higher variation might have affected the
suspiciously high estimates.
Hence, while the results do indicate a positive impact of preferences on investment (all of these
results are statistically significant), we would strongly caution against treating the absolute
numbers reported here too literally.
Table 4.2: Percentage Change in FDI
Scheme % increase in FDI stocks
of non-EU countries
GSP 349.07%
EBA 545.59%
GSP PLUS 213.30%
COTONOU 354.94%
Source: Own calculations
4.1.6 Section Summary
The aim of this section was to evaluate the impact of the unilateral preferences given by the EU
on aggregate trade and FDI. The section employs three different versions of the gravity
107
equation to study the effect of unilateral trade preference given to developing countries by the
EU since the 1970s, both on exports from these countries and on FDI flows to these countries..
This section uses the COMTRADE dataset covering nearly all of the available bilateral imports
between 183 countries over the period 1996-2008, and CEPII data for the country-specific
variables such as distance and contiguity etc.
The results from the preferred model, with importer and exporter fixed effects model such a
positive impact of the GSP regimes on exports, range from 10 to 30 percent. A similar set of
regression on investment flows also indicated a positive impact of the GSP regimes, though we
caution about interpreting the numbers literally.
108
4.2 Sectoral multilateral gravity modelling of trade In this part of the report we complement the aggregate analysis reported on above, but
considering the possible impact of preferences for six TDC sectors, that have been identified on
the grounds that there are larger preference margins associated with these sectors, and/or on
the grounds of their relative importance in the trade of the countries concerned. The sectors that
we consider here are TDC sectors II (Vegetable Products), IV (Prepared Foodstuffs), XII
(Footwear), XIa (Textiles), XIb (Clothing), and XVI (Machinery).
The structure of the gravity model is exactly the same as that reported on earlier, with the
difference that the bilateral flows in the regressions are now sector specific as opposed to
aggregate. In these results we focus on the results produced by a gravity model with importer
and exporter fixed effect. The full set of results is in Table A.16 in Appendix 5, where we see
that the coefficients for the main/standard variables of the gravity model all have the expected
coefficients. The percentage variation in trade implied by the coefficients for the EU preferential
scheme dummies are reported in Table 4.3, below, separately for each sector.
From the table we see that preferential schemes appear to have a differential impact across the
sectors being analysed. For the sector TDC XVI seem clear the reduction in trade across
different preferences agreed over the period in consideration, whereas for the sector TDC II the
change is always (at least) positive. For sector TDC IV there has been an increase in exports to
the EU for countries in all preferential schemes except for the EBA countries. For sectors
TDCXII, TDCXIa and TDCXIb instead there has been a positive increase in imports in the EU
only from countries that benefited from EBA scheme, but not for those countries exporting under
GSP, GSP+ or Cotonou. All of the coefficients were statistically significant, if not stated
otherwise. These results must be interpreted as increases in EU imports from countries that
benefited from a preferential scheme where the baseline is given by all of the other countries
that export to the EU and that weren't included in any preferential scheme.
Table 4.3: Percentage Change in Trade at Sectoral Level
Schemes TDC II
Veg.
Products
TDC IV
Prep.
Food
TDC XII
Footwear
TDC XIa
Textiles
TDC XIb
Clothing
TDC XVI
Mach’y
EU_GSP 43.33% 5.79% -18.78% -33.63% -27.16% -36.55%
EU_EBA 0 -32.69% 71.257% 54.18% 14.68% -64.08%
EU_GSP_PLUS_2006 255.37% 26.36% -51.76% -37.49% -36.49% -50.24%
COTONOU 54.34% 50.23% -61.90% -62.76% -67.43% -57.641
4.2.1 Section Summary
In this section we explored the impact of the GSP regimes on a set of specific sectors. These
were chosen in part because of their importance in many LDC countries trade, and in part
because of the existence of more substantial preference margins. For the sector TDC XVI
(Machinery) there is a clear a reduction in trade across the different preference regimes over the
109
period in question, whereas for the sector TDC II (Vegetable Products) the change is always (at
least) positive. For sector TDC IV (Prepared Foodstuffs) there is an increase in exports to the
EU for countries in all preferential schemes except for the EBA countries. For sectors TDCXII
(Footwear), TDCXIa (Textiles) and TDCXIb (Clothing) instead there is a positive increase in
imports in the EU only from countries that benefited from the EBA scheme, but not for those
countries exporting under GSP, GSP+ or Cotonou.
4.3 The impact of preferences on trade flows at the product level The aggregate gravity model allows for identifying the impact of the EU GSP regime over other
countries‘ GSP schemes and normal MFN trade. However, there are two main caveats of this
type of analysis. First, most GSP countries enjoy preferences only for a subset of products.
Therefore, measuring preferential access with one dummy can overestimate the impact of
preferential schemes because MFN trade flows are included as preferential. Second, GSP
preferences are not fully utilised due to costs of compliance and rules of origin. So again the
impact of preferences may be overestimated, since as suggested by previous sections of this
report preference utilisation matters for understanding the impact of trade preferences.
In order to overcome this problem, we need to include in the gravity model each flow according
to the trade regime used. That can be done at the aggregate level, by splitting flows according
to preference use and MFN use. Or it can be done at the product level. The advantage of doing
it at the product level is the fact that we can use tariffs rather than an MFN dummy, and,
therefore, control for the fact that a large number of flows have zero MFN rates. Either way is
consistent with non-full utilization and no other option exists for estimating the ―true‖ impact of
the GSP scheme on trade flows.
This section therefore complements the aggregate gravity part of the report taking into
consideration non-utilisation issues. In doing so, we use disaggregated flows at the product
level, which allows us to determine the real tariff paid for each export flow into the EU. The main
disadvantages of this approach is that there are no clear theoretical underpinnings for a gravity
model at such level of disaggregation and the fact that we can only compare flows to the EU
and not to other export markets, since data is not comparable at ten digits and we do not have
data on preference utilisation for other countries. This means that we will not be able to pick up
whether e.g. Bangladesh is exporting more to the EU as a result of preferences in a given
sector than it is exporting to the US. We can, however, pick whether a country which has GSP,
GSP+ or EBA preference is exporting to the EU more than a country which does not have those
preferences. More importantly, we can capture the importance of preferential flows of one
product compared to flows of the same product from the same country when the preference is
not requested and receives MFN treatment. One caveat that applies, however, is the fact that
we can distinguish whether preference was requested in the origin country, but not whether the
shipment obtained preference treatment at the port of entry. Keeping these caveats in mind, the
disaggregated gravity model that considers utilisation is the best approximation to measuring
the real impact of preferences on trade flows.
110
We estimate the model in (7) with the level variables in logarithm form. Exports from origin i in
product j in time t under regime r;39 depend on size, GDP and population, geographic and
distance variables (GEO), the tariff or tariff margin, and variety, country, time and product
specific terms. We assume that the time invariant elements in equation (7) can be absorbed by
variety, product for each origin, fixed effects as in equation (8) and estimate (9). The assumption
is that export flows can be explained by gravity variables, time dummies and variety fixed effects
that will capture any variety specific elements.
ijrttjiijijt
n
n
inititijrt uctarGEOPOPGDPX
1
1
10 (7)
jiij
n
n
inij cGEO 1
(8)
ijrttijtijtititijrt utarPOPGDPX 110 (9)
Table 4.4 shows the main estimates of equation (9). In total for the period from 2002 to 2008 we
have around 1.5 million observations. OLS estimates only include time dummies, while fixed
effects estimates are defined at the variety level and also include time dummies. Most variables
are statistically significant at 0.1 percent level.
The sign of GDP is positive across all specifications, while the sign on population changes to
negative when controlling for variety fixed effects. Two of the geo-economic indicators in the
OLS specifications have the expected sign, distance and contiguity, while common language
and former colony have negative signs. This may be the result of the fact that since we do not
know the country destination in the EU, these values take value one if the origin of the good was
a colony or had common language with any of the EU countries, reducing considerably any
variation and the effectiveness of the dummies as proxies.
39
An export flow can have several entry regimes that correspond to different tariffs in the same period
ranging from MFN to several preferential regimes. Each is associated with a different tariff. This can be
the result of the introduction of a quota or a temporary suspension of a preference, or the case of both
preference utilisation and non-utilisation in the same period.
111
Table 4.4: Gravity Model at Product Level-Tariff Regime
(1) (2) (3) (4) (5)
OLS FE1 FE2 FE3 FE4
GDP 0.4205*** 2.2281*** 2.2310*** 1.8964*** 1.9079***
(0.0035) (0.0307) (0.0307) (0.0293) (0.0294)
Population 0.0084* -1.2468*** -1.2329*** -0.8886*** -0.8967***
(0.0038) (0.0907) (0.0906) (0.0865) (0.0868)
Distance -0.0460***
(0.0098)
Contiguity 0.0968***
(0.0190)
Com language -0.1104***
(0.0112)
Colony -0.1883***
(0.0110)
Tariff -5.8708*** -11.8659*** -8.2541*** -1.3119*** -1.3066***
(0.1165) (0.0422) (0.0810) (0.0807) (0.0807)
RoO -0.0062
(0.0033)
Preference margin 4.0991*** -0.9258*** -0.9238***
(0.0785) (0.0894) (0.0894)
Margin*cotonou 3.1861*** 3.1798***
(0.1421) (0.1420)
Margin*pref 4.8160*** 4.8063***
(0.0913) (0.0913)
Margin*eba -1.7434*** -1.7447***
(0.1872) (0.1871)
Margin*gsp -5.5974*** -5.6372***
(0.1487) (0.1487)
Margin*gspplus 0.8144*** 0.8118***
(0.1953) (0.1952)
Non-utilisation -1.2624*** -1.2632***
(0.0045) (0.0045)
year_2003 -0.0382*** -0.0896*** -0.0873*** -0.0621*** -0.0622***
(0.0045) (0.0046) (0.0045) (0.0043) (0.0043)
year_2004 -0.1565*** -0.3423*** -0.3380*** -0.3260*** -0.3264***
(0.0051) (0.0056) (0.0056) (0.0053) (0.0054)
year_2005 -0.1396*** -0.2864*** -0.2813*** -0.2509*** -0.2530***
(0.0060) (0.0071) (0.0071) (0.0068) (0.0068)
year_2006 -0.0766*** -0.2864*** -0.2780*** -0.2454*** -0.2487***
(0.0063) (0.0088) (0.0088) (0.0084) (0.0084)
year_2007 -0.0442*** -0.3108*** -0.3054*** -0.2375*** -0.2415***
(0.0069) (0.0106) (0.0106) (0.0101) (0.0101)
year_2008 -0.0533*** -0.3691*** -0.3650*** -0.2678*** -0.2744***
(0.0071) (0.0122) (0.0122) (0.0116) (0.0117)
Constant 3.0678*** -1.6628*** -1.9260*** -1.3081*** -1.3453***
(0.0855) (0.3347) (0.3343) (0.3191) (0.3205)
Observations 1459559 1459559 1459559 1459559 1451541
R-squared 0.1084 0.0823 0.0847 0.1664 0.1667
R2 within 0.0823 0.0847 0.166 0.167
R2 between 0.104 0.102 0.116 0.118
R2 overall 0.0774 0.0777 0.0985 0.0997
112
Number of variety 423913 423913 423913 421405
Robust standard errors in parentheses; *** p<0.001, ** p<0.01, * p<0.05
The impact of RoO rigidity on flows is negative but not statistically significant and when
controlling for variety fixed effects, these absorb the variables that are constant over time. More
interesting is the impact of the tariff and the tariff margin. With regards the former, as expected,
higher applied tariffs imply lower export flows. Regarding the latter, preference margins have a
positive impact on exports.
We also decompose the impact of the preference margin according to the preferential regime
used and add a dummy to check the impact on exports of those episodes of non-utilisation of
preferences. Unfortunately for the case of tariffs we cannot use the regime used since most
tariffs for EBA, and to a lesser extent for GSP, are at zero rates, and, therefore, cannot be
identified. The results of the interactive preference margins are interesting, since they indicate
that the positive impact of preference margins occur for FTA regimes and Cotonou, while
margins are associated with lower exports for the EBA and GSP preferences, and to a lesser
extent for GSP+.40 It could be possible that exports of energy products that have low preferential
margins may be driving these results. For this reason, specification (5) estimates the same
specification but excluding all exports from chapter 27 ―Mineral fuels and oils‖. The results are
almost identical to the full sample. Regarding non-utilisation, the estimated coefficient is
consistently negative, indication that non-utilisation reduces the level of exports.
The previous estimates do not take into consideration the fact that we do not observe exports in
most countries for all products. There are in reality a large number of zero exports when we
consider all potential exports. As a result, our sample selection is not random, but obeys to
observed flows more likely in larger countries. The problem when correcting for this sample
selection for disaggregated exports at 10 digits is that the dataset is too large to carry out the
estimations.41 One way to overcome this problem is to aggregate the dataset at HS-4 digits and
estimate the model with selection.
Table 4.5 shows the results of controlling for sample selection. The ―export‖ column, selection
equation, explains the probability of exporting that specific HS-4 category, while the ―export
value‖ columns explain the level of exports. The results are similar to the previous tables.
However, now population and common language have the expected sign and RoO has a
puzzling positive sign explaining the level of exports. Regarding the most important coefficients
on applied tariffs and preference margins, the coefficients have the expected signs. Higher
applied tariffs reduce exports and higher margins increase exports. We also decompose the
impact of the preference margin for GSP and EBA. This time, however, due to aggregation to
HS4 the preferential regime does not reflect utilisation, since both, utilisation and non-utilisation,
40
The impact of the different margins by regime is the sum of the coefficient on the margin plus the
specific regime coefficient. In the case of GSP+ the sum is around -0.11. 41
Including zeroes for all the potential flows for all countries and all products during the period 2002-2008
will increase the sample to more than 20 million observations, out of reach of the computing capacity of a
standard PC using STATA.
113
are lost during the aggregation. We do, however, add a dummy if there has been some non-
utilisation for that HS-4 category, country and period. At this level of aggregation and without
distinguishing flows by utilisation, the margins for different regimes are highly correlated, and
therefore, we only can use GSP and EBA. Interestingly, both interactive coefficients are
positive, which indicates that the coefficients are absorbing the potentially differentiated impact
of the other preferential margins, Cotonou and other FTAs. This highlights the danger of
obtaining biased estimators when the true utilised preferential regime is not considered, and the
need to work with highly disaggregated data.
One potential problem of the previous results is the fact that they are estimated as pooled
regression. If the error term should be modelled using a panel structure then the standard
Heckman selection term is no longer valid for correcting selection bias Vella (1998). 42 To correct
for selection in the panel specification, Table 4.6 presents the results when using Wooldridge‘s
methodology for sample selection and panel data. We first estimate a probit regression for the
probability of exporting each year as a country cross-section and calculate the inverse Mills ratio
adding the selection term at each year. Then, we add this inverse Mills ratio to a pooled level
regression for the level of exports. The results are similar to the standard Heckman estimates
presented above. The significance of the correlation of residuals between the selection and
level equation, as well as the inverse Mills ratio (lambda), indicate the need for correcting for
sample selection problems. The main coefficients remain similar. Tariffs reduce the level of
exports, while preferential margins increase the level of exports.
42
Vella, F. (1998) ―Estimating models with sample selection bias: A survey,‖ Journal of Human
Resources, 1998, vol 33 pp 127-169.
114
Table 4.5: Gravity at HS4 with Selection
Selection (1) (2) (3)
Probability
export
Export value Export
value
Export
value
GDP 0.5607*** 0.5598*** 0.6199***
(0.0182) (0.0163) (0.0187)
Population 0.0923*** 0.0984*** 0.0483**
(0.0183) (0.0164) (0.0187)
GDP_capita 0.2201***
(0.0008)
Distance 0.0214*** -0.1856*** -0.1764*** -0.1360***
(0.0026) (0.0106) (0.0105) (0.0106)
Contig 0.3401*** 0.6761*** 0.6431*** 0.6605***
(0.0053) (0.0338) (0.0315) (0.0344)
Com language -0.2098*** 0.0983*** 0.0931*** 0.0929***
(0.0032) (0.0217) (0.0202) (0.0221)
Colony -0.0726*** -0.1303*** -0.1366*** -0.1901***
(0.0034) (0.0141) (0.0140) (0.0142)
Tariff -4.2572*** -4.7980*** -5.7787***
(0.1014) (0.1024) (0.1045)
RoO 0.0184*** 0.0385*** 0.0150*** 0.0147***
(0.0008) (0.0039) (0.0039) (0.0039)
Margin 2.6141*** 2.3094***
(0.0764) (0.1132)
Margin*gsp 0.7321***
(0.1488)
Margin*eba 1.4627***
(0.2284)
Non-uti 0.5011***
(0.0105)
year_2003 0.0030 -0.0539** -0.0587*** -0.0674***
(0.0044) (0.0177) (0.0177) (0.0176)
year_2004 -0.0269*** -0.1804*** -0.1839*** -0.1743***
(0.0044) (0.0179) (0.0179) (0.0179)
year_2005 -0.1288*** -0.1707*** -0.1611*** -0.1697***
(0.0045) (0.0212) (0.0206) (0.0212)
year_2006 -0.1102*** -0.0963*** -0.0868*** -0.0874***
(0.0045) (0.0204) (0.0200) (0.0204)
year_2007 -0.1406*** -0.0367 -0.0308 -0.0397
(0.0045) (0.0219) (0.0213) (0.0220)
year_2008 -0.1130*** -0.0757*** -0.0692*** -0.0827***
(0.0046) (0.0209) (0.0205) (0.0209)
Athrho -0.1064** -0.1302*** -0.0932*
(0.0383) (0.0342) (0.0394)
lnsigma 1.0454*** 1.0457*** 1.0391***
(0.0032) (0.0035) (0.0030)
Constant -2.4942*** 4.4369*** 4.4826*** 3.7352***
(0.0232) (0.1897) (0.1740) (0.1939)
Observations 1415565 1415565 1415565 1415565
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
115
Table 4.6: Gravity Model at HS4 with Wooldridge Panel Selection
(1) (2) (3)
Model1 Model2 Model1
GDP 0.5127*** 0.5258*** 0.5703***
(0.0073) (0.0073) (0.0074)
Population 0.1407*** 0.1328*** 0.0984***
(0.0076) (0.0076) (0.0076)
Distance -0.1852*** -0.1745*** -0.1363***
(0.0102) (0.0102) (0.0102)
Contig 0.5858*** 0.5743*** 0.5686***
(0.0233) (0.0233) (0.0233)
Com language 0.1553*** 0.1361*** 0.1516***
(0.0145) (0.0145) (0.0144)
Colony -0.1329*** -0.1427*** -0.1918***
(0.0136) (0.0136) (0.0135)
Tariff -4.2400*** -4.7781*** -5.7582***
(0.1146) (0.1148) (0.1228)
RoO 0.0345*** 0.0123** 0.0106**
(0.0038) (0.0039) (0.0038)
Margin 2.5957*** 2.2550***
(0.0938) (0.1330)
Margin*gsp 0.7914***
(0.1819)
Margin*eba 1.4981***
(0.2801)
Non-uti 0.5017***
(0.0105)
Lambda -0.5816*** -0.5635*** -0.5560***
(0.0383) (0.0383) (0.0378)
Constant 4.7481*** 4.6628*** 4.0681***
(0.1058) (0.1056) (0.1057)
Observations 326660 326660 326660
R-squared 0.1852 0.1881 0.1940
Robust standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
The results of the disaggregated gravity under different specifications indicate that as expected
higher applied tariffs reduce the level of exports, while tariff margins increase exports. This
results control for non-utilisation of preferences, so when countries are able to use those
preferences the results are as expected. A surprising result is, however, that when the
preference margin is decomposed by preferential regime, the average preference margin is
driven by Cotonou and FTA preferences, while EBA, GSP and GSP+ are associated with a
negative impact on exports.
116
4.3.1 Section Summary
This section complements the aggregate gravity analysis by considering non-utilisation issues. It
uses disaggregated flows at the product level to allow for consideration of the real tariff paid for
each export flow into the EU.
The impact of RoO rigidity on flows appears to be negative but not statistically significant. The
impact of the tariff and the tariff margin indicated that higher applied tariffs imply lower export
flows, while preference margins do have a positive impact on exports.
The results of the disaggregated gravity under different specifications indicate that the positive
impact of preference margins occurs primarily for the EU‘s free trade regimes and under what
was the Cotonou regime, while margins are associated with lower exports for the EBA and GSP
preferences, and to a lesser extent for GSP+. The analysis also suggests that non-utilisation
reduces the level of exports.
117
4.4 Section 4: Conclusions The aim of this section was to evaluate the impact of the unilateral preferences given by the EU
on aggregate trade and FDI, on trade in specific sectors, as well as on trade in aggregate but
where the regressions are undertaken at the product level.
With regard to aggregate trade we employ three different versions of the gravity equation to
study the effect of unilateral trade preference given to developing countries by the EU since the
1970s, both on exports from these countries and on FDI flows to these countries. We use the
COMTRADE dataset covering nearly all of the available bilateral imports between 183 countries
over the period 1996-2008, and CEPII data for the country-specific variables such as distance
and contiguity etc.
The results from the preferred model, with importer and exporter fixed effects model such a
positive impact of the GSP regimes on exports, range from 10 to 30 percent. A similar set of
regression on investment flows also indicated a positive impact of the GSP regimes, though we
caution about interpreting the numbers literally.
The sectors in the sectoral regressions were chosen in part because of their importance in
many LDC countries trade, and in part because of the existence of more substantial preference
margins. For the sector TDC XVI (Machinery) there is a clear a reduction in trade across the
different preference regimes over the period in question, whereas for the sector TDC II
(Vegetable Products) the change is always (at least) positive. For sector TDC IV (Prepared
Foodstuffs) there is an increase in exports to the EU for countries in all preferential schemes
except for the EBA countries. For sectors TDCXII (Footwear), TDCXIa (Textiles) and TDCXIb
(Clothing) instead there is a positive increase in imports in the EU only from countries that
benefited from the EBA scheme, but not for those countries exporting under GSP, GSP+ or
Cotonou.
Finally the preceding analyses were complemented by regressions which used disaggregated
flows at the product level to allow for consideration of the real tariff paid for each export flow into
the EU. This indicated that the impact of RoO rigidity on flows appears to be negative but not
statistically significant. The impact of the tariff and the tariff margin indicated that higher applied
tariffs imply lower export flows, while preference margins do have a positive impact on exports.
The results of the disaggregated gravity also indicated that the positive impact of preference
margins occurs primarily for the EU‘s free trade regimes and under what was the Cotonou
regime, while margins are associated with lower exports for the EBA and GSP preferences, and
to a lesser extent for GSP+. The analysis also suggests that non-utilisation reduces the level of
exports.
118
5 Computable General Equilibrium Evaluation of GSP
5.1 Introduction To isolate the effects of tariff changes under the GSP 2006-2008 scheme from other exogenous
influences, a global computable general equilibrium (CGE) modelling approach is adopted in
this section. The analytic framework is the GLOBE model, a multi-regional and multi-sectoral
CGE model of global production and trade developed by McDonald, Robinson and Thierfelder
(2007). The model is calibrated to the new GTAP7 database that reflects the global input-output
structure of production and trade by origin and destination in 2004. The database distinguishes
113 geographical regions and 57 commodity groups.
For the present study, we construct a 32-region aggregation of the GTAP database which
identifies a range of individual and composite GSP, GSP+ and EBA ―countries/regions‖ with an
appropriate level of geographical detail for the partner countries as identified in the GTAP
dataset. As shown in Table 5.1, the regional aggregation agreed with the Commission includes
eight GSP+ countries / regions, four EBA countries / region blocs, and a range of other
developing GSP region blocs. The agreed sectoral aggregation distinguishes 19 commodity
groups and activities and aims to identify separately the product groups most affected by the
GSP scheme at the deepest possible disaggregation level (Table 5.2). The model includes five
primary production factors: skilled labour, unskilled labour, capital, land and natural resources.
A distinct advantage of using a global CGE model for the purpose at hand is that it allows a
comprehensive integrated internally consistent assessment of the trade creation, trade
diversion, sectoral employment and structural transformation effects triggered by the GSP
scheme; as well as an evaluation of the aggregate welfare effects by country that takes full
account of indirect open-economy general equilibrium feedback linkages.
The model framework allows analysing the incremental impact of the switch from the pre-2006
GSP (i.e. 2004) to the 2006-09 GSP regime as well as an evaluation of the total impact of the
GSP in the form of a comparison of the 2006-09 equilibrium with a ―no-GSP‖ anti-monde. In this
latter scenario, all EU import tariffs faced by the GSP, GSP+ and EBA beneficiaries will be
raised to MFN level. Furthermore, we consider a switch from the observed levels of utilization of
GSP preferences to a full utilization of preferential GSP tariffs. Finally, we simulate a complete
elimination of all EU import tariffs for GSP countries.
The following section provides a brief non-technical outline of the GLOBE model. Section 5.3
highlights a number of key features of the benchmark data set and Section 5.4 presents the
results of the simulation analysis.
119
5.1.1 Section Summary
This section describes the GLOBE model as an analytical framework to isolate the effects of
tariff changes using a multi-regional and multi-sectoral CGE model of global production and
trade. The use of this model allows for a comprehensive assessment of the trade creation, trade
diversion, sectoral employment and structural transformation effects triggered by the GSP
scheme. It also provides for an evaluation of the aggregate welfare effects by country.
This section describes how this model enables an examination of the incremental impact of the
switch from the pre-2006 GSP to the 2006-09 GSP regime as well as an evaluation of the total
impact of the GSP. It indicates the impact of raising all EU import tariffs faced by the GSP,
GSP+ and EBA beneficiaries to MFN level. It can also examine levels of utilization of GSP
preferences to a full utilization of preferential GSP tariffs and simulate a complete elimination of
all EU import tariffs for GSP countries.
5.2 The GLOBE Model GLOBE is a theory-grounded, comparative-static, multi-region, multi-sectoral CGE model of
global production and trade developed by McDonald, Robinson and Thierfelder (2007).43 The
model version used here is calibrated to the new GTAP7 database that reflects the global input-
output structure of production and trade by origin and destination in 2004.
International Trade
Domestically produced commodities are assumed to be imperfect substitutes for traded goods.
Import demand is modelled via a series of nested constant elasticity of substitution (CES)
functions; imported commodities from different source regions to a destination region are
assumed to be imperfect substitutes for each other and are aggregated to form composite
import commodities that are assumed to be imperfect substitutes for their counterpart domestic
commodities The composite imported commodities and their counterpart domestic commodities
are then combined to produce composite consumption commodities, which are the commodities
demanded by domestic agents as intermediate inputs and final demand (private consumption,
government, and investment). Export supply is modelled via a series of nested constant
elasticity of transformation (CET) functions. The composite export commodities are assumed to
be imperfect substitutes for domestically consumed commodities, while the exported
commodities from a source region to different destination regions are assumed to be imperfect
substitutes for each other. The composite exported commodities and their counterpart domestic
commodities are then combined as composite production commodities.
The use of nested CET functions for export supply implies that domestic producers adjust their
export supply decisions in response to changes in the relative prices of exports and domestic
commodities. This specification is desirable in a global model with a mix of developing and
developed countries that produce different kinds of traded goods with the same aggregate
43
For recent applications of this model to the analysis of preferential trading arrangements see e.g. Polaski et al. (2009), World Bank (2009), CARIS (2008), McDonald, Thierfelder and Robinson (2008) and McDonald and Willenbockel (2008).
120
commodity classification, and yields more realistic behaviour of international prices than models
assuming perfect substitution on the export side.
Production, Input Demand and Factor Markets
Production relationships by activities are characterized by nested Constant Elasticity of
Substitution (CES) production functions. Activity output is a CES composite of aggregate
intermediate inputs and aggregate value added, while aggregate intermediate inputs are a
Leontief aggregate of the individual intermediate commodity inputs and aggregate value added
is a CES composite of primary factors demanded by each activity. The determination of product
supply and input demand is based on the assumption of profit maximizing behaviour.
Two alternative factor market regimes are considered in this study – a standard neoclassical
long-run full employment closure and a closure that allows for unemployed unskilled labour in
the developing regions of the model. Under the latter closure, factor markets in developed
countries are characterized by inelastic factor supplies and the model solves for market-clearing
factor prices like under the neoclassical closure. In developing regions, however, the real wage
of skilled and unskilled labour is fixed in terms of the domestic consumer price index and the
supply of skilled and unskilled labour is infinitely elastic at that wage. In this specification, any
shock that would otherwise reduce the equilibrium wage will instead lead to increased
unemployment. In both factor market regimes, the primary factors except activity-specific natural
resource endowments are mobile across production activities, but immobile across borders.
Final Domestic Demand by Commodity
The commodity composition of government consumption demand and investment demand is
fixed, with demand patterns from the benchmark data set. Households are utility maximizers
who respond to changes in relative prices and incomes. In this version of the model, the utility
functions for private households take the Stone-Geary form and hence consumer demand by
commodity is described by a Linear Expenditure System (LES) specification.
Macro Closure
For this exercise a ―neutral‖ or ―balanced‖ set of macro closure rules is specified. Current
account balances for all regions are assumed to be fixed at initial benchmark levels in terms of a
global numeraire and real exchange rates adjust to maintain external equilibrium. The global
numeraire is the basket of goods underlying the EU consumer price index. Any change in, say,
the nominal value of export earnings at world market prices The assumption of fixed current
account balances ensures that there are no changes in future ―claims‖ on exports across the
regions in the model. That is, net asset positions are fixed. In addition, we assume a ―balanced‖
macro adjustment to the trade policy shocks within countries. Changes in aggregate absorption
are assumed to be shared equally (to maintain the shares from the base data) among private
consumption, government, and investment demands. Household and government saving rates
adjust residually to establish the macroeconomic saving-investment balance in each region.
Benchmark Data and Calibration
121
The model is calibrated to a social accounting matrix representation of the GTAP 7.0 database
(Narayanan and Walmsley (eds.), 2008) that combines detailed bilateral trade, and protection
data reflecting economic linkages among regions with individual country input-output data.
These account for intersectoral linkages within regions for the benchmark year 2004.
Production, trade and income elasticities are drawn from the GTAP behavioural data base.
122
Table 5.1: Regional Aggregation of the Model
Code Description Status Notes
EU European Union ex post-2004 entrants Bulgaria, Romania
RoOECD Rest of OECD+
SriLanka Sri Lanka GSP+
Peru Peru GSP+
Ecuador Ecuador GSP+
Colombia Colombia GSP+
CostaRica Costa Rica GSP+
GSP+ LA GSP+ Other Latin America GSP+ Bolivia, Paraguay, Guatemala, Panama,
Nicaragua, El Salvador, Rest of Ctrl America
GSP+ EE GSP+ Eastern Europe GSP+ Armenia, Azerbaijan
Georgia Georgia GSP+
Cambodia Cambodia EBA
Bangladesh Bangladesh EBA
EBA RoAs EBA: Rest of Asia EBA Afghanistan, Bhutan,Laos,Maldives,
Myanmar,Nepal
EBA SSA EBA: Sub-Saharan Africa EBA Angola, DR Congo, Ethiopia, Madagascar,
Malawi, Mozambique, Senegal, Tanzania,
Uganda
China China GSP
Philippines Philippines GSP
India India GSP
Pakistan Pakistan GSP
Thailand Thailand GSP
RoAsia Rest of Asia GSP
Argentina Argentina GSP
Brazil Brazil GSP
Caribbean Caribbean GSP
Russia Russia GSP
Ukraine Ukraine GSP
RoSEE Rest of Southern and Eastern Europe GSP
CtrlAsia Central Asia GSP
NAfrica North Africa GSP
RoSSA Rest of Sub-Saharan Africa GSP/EBA Non-EBA and composite mixed EBA/GSP
SSA regions in GTAP7
SAfrica South Africa GSP
Emerged Emerged DCs (GSP) Hong Kong, Taiwan, Singapore, Korea,
Chile, Mexico *
RoWorld Rest of World (GSP) Middle, East, Iran, Turkey, Uruguay, Guyana,
Falklands *
* Non-beneficiary countries in italics
123
Table 5.2: Commodity Aggregation of the Model
Code Description
Rice Paddy rice, processed rice
Vegetables, fruits Vegetables, fruit, nuts
Other crops Wheat, other cereal grains, plant-based fibres, crops nec
Oils, fats Oil seeds, vegetable oils and fats
Sugar prd Sugar cane, sugar beet, processed sugar products
Livestock prd Livestock except fish, raw milk, animal products except meat
Fishing prd Fishing products
Fossile fuels Fossile fuels: Coal, oil, gas, petroleum, coal products
Mineral prd Minerals nec, mineral products
Other food prd Meat, dairy products, food products nec, beverages, tobacco
Textiles Textiles
Apparel Apparel
Leather prd Leather products
Other light mnf Light manufacturing: Forestry and wood products, paper
products, publishing, other manufacturing
Chemicals Chemical, rubber, plastic products
Metal prd Metals and metal products
Transport equip Motor vehicles and parts, other transport equipment
Machinery, elec equip Electronic equipment, machinery and equipment nec
Services Construction, utilities, services
5.3 Patterns of Trade and Production in the Benchmark Equilibrium This section presents information on selected key characteristics of the 2004 benchmark
equilibrium data required for a deeper understanding of the simulation results reported in
Section 5.4.
Table 5.3 shows the ratio of total exports of goods of services to GDP for the 32 model regions
as well as each region‘s EU share in total 2004 export revenue. For the GSP beneficiaries,
these two ratios co-determine the order of magnitude of the economy-wide impacts of changes
in the GSP regime. The export-GDP ratio ranges from less than 0.2 for India and Pakistan to 1.0
for Cambodia. North Africa, Bangladesh and the Rest of Southern and Eastern Europe region
sell more than 50 percent of their total exports of goods and services to the European Union,
while for the GSP+ other Latin America region, the EU market accounts only for 13 percent of
total exports. The final column of Table 5.3 shows the origin composition of total EU imports
excluding intra-EU imports.
Table 5.4 reports the regional composition of EU imports from non-EU sources by commodity
group in 2004, highlighting all market shares in excess of 5 percent. Table 5.5 exhibits the EU
share in a region‘s exports by commodity group, e.g. 76 percent of Bangladesh‘s textile exports
124
go to the European Union and this flow accounts for 6 percent of the European Union‘s extra-
EU textile imports.
Finally, Table 5.6 shows the sectoral contributions to GDP generated for each region, e.g. the
Bangladeshi textile industry generates 5.9 percent of the country‘s labour and real capital
income in 2004. The figures in Table 5.6 are crucial for the explanation of the simulation results
reported in the following section.
125
Table 5.3: Selected Benchmark Macro Indicators by Country
GDP (fc)
EX/GDP
%
EU
EX/
EX %
Share in
EU IM %
EU 92517.2 44.8 60.7 -
Sri Lanka 191.9 39.1 35.5 0.17
Peru 653.4 20.2 22.3 0.19
Ecuador 271.9 37.2 18.1 0.12
Colombia 868.5 21.6 19.6 0.24
Costa Rica 170.8 68.6 37.7 0.29
GSP+ Other Latin America 1805.5 34.5 13.5 0.54
GSP+ Eastern Europe 112.3 45.2 45.9 0.15
Georgia 39.3 38.7 23.4 0.02
Cambodia 42.4 99.9 31.3 0.09
Bangladesh 519.9 20.7 54.0 0.38
EBA: Rest of Asia 214.4 30.0 30.7 0.13
EBA: Sub-Saharan Africa 637.7 41.1 26.1 0.44
China 16381.6 41.7 24.3 10.79
Philippines 798.5 63.7 19.2 0.63
India 5896.5 17.6 29.4 1.97
Pakistan 861.9 19.1 31.7 0.34
Thailand 1488.8 81.3 20.5 1.61
Rest of Asia 4158.0 70.4 18.5 3.50
Argentina 1247.5 31.2 23.1 0.58
Brazil 4935.2 23.2 26.7 1.98
Caribbean 1534.2 30.4 32.5 0.98
Russia 4592.0 36.0 44.8 4.80
Ukraine 495.3 78.2 28.3 0.71
Rest of South and East
Europe 1764.6 49.2 54.0 3.04
Central Asia 620.2 56.0 33.4 0.75
North Africa 2504.6 40.4 57.0 3.74
Rest of Sub-Saharan Africa 2168.0 46.3 32.1 2.09
South Africa 1915.5 32.2 35.4 1.42
Emerged DCs 18618.4 56.9 15.3 10.49
Rest of OECD 163023.2 15.2 24.8 39.97
RoWorld 10854.4 46.6 23.9 7.84 Notes: GDP at factor price 2004 in 100 billion US$;
EX/GDP: Exports-GDP Ratio;
EU EX/ EX: Share of exports to the EU in total exports of the region;
Share in EU Im: Share of a region‘s exports to the EU in EU‘s total extra-EU imports.
126
Table 5.4: Regional Origin Shares in Extra-EU Imports by Commodity (%)
Ric
e
Ve
ge
tab
les,
fru
its
Oth
er
cro
ps
Oils
, fa
ts
Su
ga
r p
rd
Liv
es
tock
prd
Fis
hin
g p
rd
Fo
ss
ile f
ue
ls
Min
era
l p
rd
Oth
er
foo
d p
rd
Te
xti
les
Ap
pa
rel
Le
ath
er
prd
Oth
er
lig
ht
mn
f
Ch
em
icals
Me
tal
prd
Tra
ns
po
rt e
qu
ip
Ma
ch
ine
ry, e
lec
eq
uip
Se
rvic
es
SriLanka 0.3 0.1 0.9 0.0 0.0 0.0 0.3 0.0 0.2 0.2 0.7 1.8 0.1 0.3 0.1 0.0 0.1 0.0 0.2
Peru 0.1 0.9 1.4 0.0 0.1 0.2 0.1 0.0 1.1 1.4 0.2 0.1 0.1 0.1 0.0 0.7 0.0 0.0 0.2
Ecuador 0.1 5.1 1.0 0.0 0.1 0.0 0.1 0.0 0.0 1.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.1
Colombia 0.1 2.9 2.7 0.5 0.1 0.1 0.1 0.6 0.0 0.5 0.1 0.1 0.3 0.0 0.0 0.4 0.0 0.0 0.2
CostaRica 0.1 3.8 0.8 0.0 0.1 0.1 0.0 0.0 0.0 0.4 0.0 0.0 0.0 0.2 0.0 0.0 0.0 0.7 0.2
GSP+ LA 0.7 3.9 3.3 2.4 2.5 0.7 0.4 0.8 0.7 1.2 0.2 0.1 0.5 0.2 0.1 0.8 0.0 0.0 0.9
GSP+ EE 0.0 0.0 0.1 0.0 0.0 0.0 0.1 1.0 0.1 0.0 0.0 0.0 0.0 0.1 0.0 0.1 0.0 0.0 0.1
Georgia 0.0 0.1 0.0 0.0 0.0 0.1 0.0 0.0 0.1 0.1 0.0 0.0 0.0 0.0 0.0 0.1 0.0 0.0 0.0
Cambodia 0.9 0.0 0.0 0.0 0.0 0.1 0.1 0.0 0.0 0.0 1.2 0.8 0.4 0.0 0.0 0.0 0.0 0.0 0.1
Bangladesh 0.2 0.1 0.1 0.0 0.3 0.0 0.1 0.0 0.1 0.5 6.5 5.2 0.8 0.0 0.0 0.0 0.0 0.0 0.1
EBA RoAs 2.1 0.0 0.1 0.0 0.7 0.3 0.2 0.2 0.0 0.2 0.7 1.0 0.1 0.1 0.0 0.0 0.0 0.0 0.2
EBA SSA 0.6 1.1 4.7 0.4 6.0 1.4 2.6 0.7 0.6 2.1 0.3 0.3 0.2 0.4 0.0 1.3 0.0 0.0 0.4
China 10.5 3.4 1.8 1.5 0.1 14.8 2.7 0.8 8.1 6.0 14.5 21.7 30.5 20.1 7.0 10.7 5.2 21.0 6.3
Philippines 0.9 0.4 0.1 0.9 0.1 0.1 0.6 0.0 0.1 0.6 0.5 0.5 0.3 0.5 0.1 0.1 0.2 1.9 0.4
India 20.5 1.9 2.5 1.9 1.4 1.6 0.6 0.2 2.3 1.9 7.9 6.3 6.7 3.9 1.8 2.0 0.9 0.5 2.7
Pakistan 8.1 0.2 0.1 0.0 4.5 0.7 0.0 0.0 0.0 0.3 5.1 2.6 0.7 0.3 0.1 0.1 0.0 0.0 0.2
Thailand 24.3 1.6 0.5 0.1 0.6 0.2 0.7 0.0 0.8 3.7 2.2 1.9 2.3 2.7 1.0 0.6 1.3 2.2 2.0
RoAsia 7.4 1.1 5.2 9.4 0.8 3.7 3.6 0.3 2.3 3.4 5.3 5.9 21.9 6.7 2.5 1.0 0.8 4.5 3.5
Argentina 0.0 4.5 2.9 17.9 0.1 4.1 0.3 0.0 1.1 3.6 0.2 0.1 0.6 0.2 0.3 0.2 0.2 0.0 0.6
Brazil 0.2 3.6 13.4 34.8 3.7 4.1 0.8 0.2 7.6 7.5 0.5 0.2 4.3 3.6 1.0 2.4 2.5 0.5 1.4
Caribbean 1.1 1.6 0.8 0.3 13.2 1.2 0.5 0.3 0.3 2.5 0.3 0.3 0.2 0.4 0.7 1.2 1.8 0.3 2.0
Russia 0.0 0.4 0.9 0.9 0.2 1.9 0.2 23.1 4.8 1.6 0.6 0.3 0.7 4.3 3.4 12.7 0.6 0.2 2.6
Ukraine 0.2 0.4 1.2 2.6 0.5 2.2 0.0 0.6 2.2 0.6 0.4 1.3 1.0 0.7 0.6 3.0 0.2 0.2 0.6
127
Ric
e
Ve
ge
tab
les,
fru
its
Oth
er
cro
ps
Oils
, fa
ts
Su
ga
r p
rd
Liv
es
tock
prd
Fis
hin
g p
rd
Fo
ss
ile f
ue
ls
Min
era
l p
rd
Oth
er
foo
d p
rd
Te
xti
les
Ap
pa
rel
Le
ath
er
prd
Oth
er
lig
ht
mn
f
Ch
em
icals
Me
tal
prd
Tra
ns
po
rt e
qu
ip
Ma
ch
ine
ry, e
lec
eq
uip
Se
rvic
es
RoSEE 0.2 1.4 1.9 1.8 13.1 11.4 6.9 1.5 2.2 3.9 6.1 14.8 13.8 5.4 1.9 5.8 2.2 1.6 2.3
CtrlAsia 0.1 0.1 2.4 0.0 0.0 0.5 0.0 4.5 0.0 0.1 0.4 0.0 0.0 0.0 0.1 1.7 0.0 0.0 0.4
NAfrica 2.9 7.4 1.1 4.8 2.0 4.0 8.4 15.3 2.2 3.0 5.1 13.0 4.3 0.8 1.6 1.6 0.4 0.9 2.7
RoSSA 0.6 6.7 22.6 0.8 33.2 1.6 3.2 4.2 13.4 6.0 1.5 0.6 1.0 2.3 0.6 0.7 1.4 0.2 1.5
SAfrica 0.0 8.5 1.1 0.1 0.6 1.9 3.1 1.4 8.9 2.6 0.6 0.3 0.4 1.7 0.7 4.4 1.2 0.5 1.0
Emerged 0.4 6.2 1.5 0.2 0.5 2.7 3.0 1.3 5.5 4.5 8.3 4.7 2.2 4.7 9.1 10.3 12.2 16.7 13.8
RoOECD 13.6 16.9 19.0 14.3 3.5 24.8 55.5 25.8 23.9 34.1 9.9 2.8 4.6 34.8 61.5 32.4 62.4 44.8 45.2
RoWorld 3.8 15.7 5.8 4.2 12.2 15.7 5.8 17.2 11.4 6.5 20.9 13.3 1.9 5.3 5.6 5.7 6.5 3.1 8.0
Sum 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100
Shares > 5% highlighted.
128
Table 5.5: EU Share in Countries’ Total Exports by Commodity (%)
Ric
e
Ve
ge
tab
les,
fru
its
Oth
er
cro
ps
Oils
, fa
ts
Su
ga
r p
rd
Liv
es
tock
prd
Fis
hin
g p
rd
Fo
ss
ile f
ue
ls
Min
era
l p
rd
Oth
er
foo
d p
rd
Te
xti
les
Ap
pa
rel
Le
ath
er
prd
Oth
er
lig
ht
mn
f
Ch
em
icals
Me
tal
prd
Tra
ns
po
rt e
qu
ip
Ma
ch
ine
ry, e
lec
eq
uip
Se
rvic
es
To
tal
Ex
po
rts
EU 86.1 88.2 74.7 73.8 63.3 72.6 89.9 69.1 57.0 73.0 65.9 67.8 61.0 66.8 65.4 71.0 68.6 57.1 47.1 60.7
SriLanka 39.8 26.8 22.8 1.0 38.5 23.4 12.6 34.3 58.0 18.7 43.7 34.2 56.4 49.2 37.5 5.7 68.8 30.4 42.9 35.5
Peru 46.8 35.3 58.0 31.3 7.5 31.5 20.4 5.1 17.1 30.0 12.0 10.8 36.9 11.2 10.5 15.7 9.9 11.5 42.0 22.3
Ecuador 40.0 45.6 30.5 1.7 4.1 33.2 3.7 0.2 5.3 39.4 8.4 31.6 16.5 14.8 7.6 5.0 12.2 10.8 25.0 18.1
Colombia 33.1 61.4 25.3 44.4 0.6 1.9 16.0 17.4 3.6 22.6 7.7 4.5 29.8 4.3 3.8 22.3 2.4 4.5 38.2 19.6
CostaRica 38.0 49.1 35.2 1.7 5.0 12.9 1.8 38.8 5.1 19.7 7.5 3.8 9.5 27.9 6.1 4.9 20.3 44.9 48.0 37.7
GSP+ LA 19.6 39.9 36.1 14.4 8.1 20.9 5.9 4.9 19.8 18.3 2.4 1.9 42.7 14.0 5.4 17.2 5.9 11.0 38.2 13.5
GSP+ EE 14.7 13.6 27.0 0.1 0.8 4.4 74.1 63.6 33.3 8.3 11.7 55.9 27.9 40.5 14.0 27.1 0.9 18.2 29.7 45.9
Georgia 0.3 51.8 10.6 25.6 0.0 40.8 67.8 57.7 36.8 27.2 45.4 42.3 64.5 25.5 41.7 11.2 5.6 28.1 28.3 23.4
Cambodia 57.6 16.8 8.1 12.5 47.4 28.5 29.8 23.4 35.5 11.8 60.6 17.0 44.6 7.9 3.5 5.3 43.2 21.2 45.7 31.3
Bangladesh 42.8 44.8 10.2 1.6 94.2 19.4 6.5 1.7 71.5 46.3 75.8 50.5 47.6 30.2 15.6 10.0 41.6 5.1 25.3 54.0
EBA RoAs 36.3 1.5 20.2 1.1 84.5 29.0 5.2 28.4 2.6 18.9 58.3 56.7 24.8 12.5 3.7 2.6 16.6 18.0 38.1 30.7
EBA SSA 18.4 45.0 45.0 20.5 57.8 39.7 76.5 9.8 39.3 60.7 33.3 31.0 54.9 43.2 5.6 68.0 39.7 35.9 34.2 26.1
China 17.9 18.0 13.2 19.2 5.5 27.2 6.4 10.7 22.8 14.8 13.5 20.7 21.0 24.7 20.9 20.5 25.4 26.2 40.0 24.3
Philippines 46.8 6.0 38.5 24.8 1.6 25.7 13.2 2.0 4.6 18.2 29.2 11.4 36.8 27.5 12.5 3.2 13.1 18.7 39.2 19.2
India 11.1 26.6 19.2 15.6 35.4 37.1 26.8 7.8 14.7 23.2 34.5 46.5 59.6 18.4 21.6 19.4 34.1 26.5 45.9 29.4
Pakistan 8.6 15.9 10.6 1.6 55.0 46.8 4.5 1.1 13.3 30.6 29.9 49.0 38.9 46.9 32.6 25.1 8.6 30.4 30.8 31.7
Thailand 6.2 22.5 17.3 2.4 1.2 7.7 10.3 2.4 15.6 17.2 20.1 26.9 28.2 25.8 9.3 11.9 19.8 17.9 42.7 20.5
RoAsia 4.8 17.0 31.3 11.3 18.3 17.8 9.3 1.9 15.9 15.7 23.0 23.2 58.0 20.5 13.7 7.8 22.5 15.1 42.4 18.5
Argentina 0.1 63.1 14.5 28.3 1.6 68.2 40.1 0.8 45.3 32.7 19.9 33.7 15.7 16.3 10.5 13.1 8.9 12.0 43.8 23.1
Brazil 11.2 65.4 42.1 45.6 2.1 42.1 32.2 9.2 28.5 28.3 15.0 29.6 26.7 31.7 18.2 17.6 19.2 17.2 40.1 26.7
Caribbean 38.9 59.8 28.8 28.5 39.0 53.3 25.4 8.0 19.4 42.5 13.5 6.0 18.8 23.1 22.4 32.5 61.4 23.8 42.7 32.5
Russia 2.3 52.1 17.1 27.4 6.6 37.2 8.2 60.9 53.3 15.5 28.7 31.3 55.1 37.9 33.9 33.5 17.0 14.9 41.0 44.8
129
Ric
e
Ve
ge
tab
les,
fru
its
Oth
er
cro
ps
Oils
, fa
ts
Su
ga
r p
rd
Liv
es
tock
prd
Fis
hin
g p
rd
Fo
ss
ile f
ue
ls
Min
era
l p
rd
Oth
er
foo
d p
rd
Te
xti
les
Ap
pa
rel
Le
ath
er
prd
Oth
er
lig
ht
mn
f
Ch
em
icals
Me
tal
prd
Tra
ns
po
rt e
qu
ip
Ma
ch
ine
ry, e
lec
eq
uip
Se
rvic
es
To
tal
Ex
po
rts
Ukraine 45.3 54.2 21.5 39.1 14.7 44.3 37.7 35.2 59.9 13.1 57.2 86.5 79.9 43.0 26.9 23.0 7.9 26.2 27.1 28.3
RoSEE 35.9 49.3 38.2 41.1 52.1 58.9 82.8 56.3 43.9 37.0 72.2 89.9 90.1 61.3 40.1 53.9 56.5 58.5 37.4 54.0
CtrlAsia 2.2 2.7 19.5 9.4 0.0 22.7 4.8 41.6 0.3 18.2 33.8 21.7 11.6 15.1 29.7 27.4 1.6 13.8 27.6 33.4
NAfrica 7.2 70.9 37.8 77.9 32.2 71.0 87.6 64.6 38.0 51.7 77.8 85.8 87.5 55.1 45.6 46.2 66.6 71.6 36.4 57.0
RoSSA 22.6 69.8 48.6 18.3 64.6 10.1 71.1 15.9 73.0 52.5 42.4 19.1 58.8 49.2 25.7 13.5 62.7 47.7 40.4 32.1
SAfrica 0.0 65.6 27.2 7.8 3.7 38.9 63.6 56.3 57.6 34.5 34.1 24.8 32.6 32.3 20.3 22.6 28.5 35.2 45.1 35.4
Emerged 5.0 14.9 17.5 5.9 5.1 9.1 8.4 5.3 14.9 10.2 9.4 10.8 12.2 11.6 12.5 13.5 16.4 14.4 22.2 15.3
RoOECD 6.7 23.6 9.4 13.0 6.3 13.1 38.2 39.3 23.8 17.4 13.8 17.9 23.8 20.5 28.6 17.7 19.6 23.1 31.7 24.8
RoWorld 8.8 47.2 35.7 31.1 34.3 40.6 28.2 12.4 34.2 26.6 54.2 55.3 31.2 18.4 23.0 19.0 50.3 35.5 38.6 23.9
130
Table 5.6: Sectoral Shares in GDP by Country (%)
Ric
e
Ve
ge
tab
les,
fru
its
Oth
er
cro
ps
Oils
an
d f
ats
Su
ga
r p
rd
Liv
es
tock
prd
Fis
hin
g p
rd
Fo
ss
ile f
ue
ls
Min
era
l p
rd
Oth
er
foo
d p
rd
Te
xti
les
Ap
pa
rel
Le
ath
er
prd
Oth
er
lig
ht
mn
f
Ch
em
icals
Me
tal
prd
Tra
ns
po
rt e
qu
ip
Ma
ch
ine
ry, e
lec
eq
uip
Se
rvic
es
Su
m
EU 0.0 0.5 0.6 0.1 0.1 0.6 0.1 0.5 1.1 2.7 0.5 0.5 0.2 3.2 2.9 2.4 2.0 4.4 77.8 100
SriLanka 3.0 6.3 7.9 0.3 0.1 1.4 2.7 0.6 2.1 7.5 2.1 4.9 0.4 2.1 3.0 0.3 0.3 0.6 54.3 100
Peru 0.6 2.2 5.0 0.8 1.0 1.7 1.0 1.5 4.3 8.0 3.0 1.6 0.9 7.7 3.1 4.2 2.4 4.5 46.3 100
Ecuador 0.8 3.4 2.4 0.4 0.7 1.4 1.5 21.1 1.1 5.3 0.9 0.9 0.2 2.6 0.9 0.2 0.1 0.2 55.9 100
Colombia 0.4 2.8 2.8 0.5 0.6 3.8 0.5 8.0 1.7 4.1 0.4 0.7 0.2 2.0 2.4 1.9 0.4 0.7 66.2 100
CostaRica 0.2 3.8 2.5 0.5 0.6 1.9 0.3 0.0 1.0 4.8 0.4 0.7 0.1 4.3 2.4 0.6 0.3 10.1 65.5 100
GSP+ LA 0.4 2.4 2.0 1.1 0.8 2.6 0.6 14.2 1.6 4.3 0.9 1.1 0.3 1.9 2.4 1.6 0.9 0.5 60.3 100
GSP+ EE 0.0 4.8 4.3 0.4 0.1 2.4 0.2 29.1 0.5 5.1 0.0 0.0 0.0 1.7 0.0 0.7 0.9 0.1 49.7 100
Georgia 0.0 6.7 3.1 0.0 0.6 10.6 0.3 0.2 1.7 5.1 0.0 0.1 0.1 1.3 0.6 2.5 1.3 0.5 65.3 100
Cambodia 4.3 3.0 2.9 0.7 0.3 4.0 7.0 2.0 0.4 1.5 4.8 17.3 0.8 3.2 0.6 0.1 0.3 0.7 46.0 100
Bangladesh 8.2 1.9 2.6 0.3 0.7 1.6 2.1 1.4 0.7 1.1 5.9 1.6 0.1 3.5 0.4 0.4 0.4 0.3 66.6 100
EBA RoAs 12.5 7.1 2.3 1.6 0.6 2.4 2.7 4.9 0.7 1.9 2.4 0.9 0.1 4.7 0.4 0.6 0.6 0.9 52.7 100
EBA SSA 1.1 6.4 11.4 1.0 1.0 3.5 1.6 15.7 1.5 4.3 0.5 0.5 0.1 4.7 0.9 1.1 0.1 0.3 44.1 100
China 1.3 5.8 1.1 0.4 0.1 3.3 1.4 4.1 3.0 2.0 2.1 1.3 0.7 6.1 4.1 4.6 1.9 8.9 47.9 100
Philippines 3.2 2.3 1.4 0.9 0.5 2.8 2.5 0.3 1.2 3.9 0.6 1.6 0.2 2.5 2.2 2.3 1.3 17.9 52.4 100
India 3.9 4.5 6.8 3.3 1.6 2.1 0.9 2.5 1.3 4.8 2.1 0.4 0.3 3.6 2.2 2.1 1.4 1.9 54.1 100
Pakistan 1.3 3.5 5.2 0.5 2.4 10.3 0.4 1.1 1.9 2.7 5.1 1.0 0.1 1.0 0.5 0.1 0.4 0.5 62.2 100
Thailand 3.1 2.6 1.7 0.5 0.7 0.8 1.2 2.6 1.9 3.1 3.0 1.9 1.0 4.2 3.7 1.3 3.4 8.4 55.0 100
RoAsia 2.3 2.3 1.9 2.6 0.3 1.3 1.5 11.1 2.8 3.3 1.7 0.9 0.8 4.9 4.9 1.9 2.0 8.3 45.3 100
Argentina 0.0 1.1 2.4 3.4 0.1 1.8 0.0 5.9 1.5 3.4 0.3 0.6 0.5 2.6 2.4 1.4 1.2 0.9 70.5 100
Brazil 0.3 0.6 2.9 1.4 0.5 1.7 0.0 2.6 1.4 2.7 0.5 0.5 0.4 2.9 2.8 2.0 1.9 3.5 71.3 100
Caribbean 0.2 1.4 2.2 0.4 0.9 1.6 0.2 1.8 1.3 3.5 1.4 0.5 0.1 3.4 4.3 2.7 2.0 6.0 66.0 100
131
Ric
e
Ve
ge
tab
les,
fru
its
Oth
er
cro
ps
Oils
an
d f
ats
Su
ga
r p
rd
Liv
es
tock
prd
Fis
hin
g p
rd
Fo
ss
ile f
ue
ls
Min
era
l p
rd
Oth
er
foo
d p
rd
Te
xti
les
Ap
pa
rel
Le
ath
er
prd
Oth
er
lig
ht
mn
f
Ch
em
icals
Me
tal
prd
Tra
ns
po
rt e
qu
ip
Ma
ch
ine
ry, e
lec
eq
uip
Se
rvic
es
Su
m
Russia 0.0 2.2 1.6 0.2 0.1 1.4 0.1 16.5 1.7 2.3 0.1 0.2 0.1 2.3 1.2 3.9 0.6 2.5 62.9 100
Ukraine 0.1 1.5 2.3 0.7 0.2 2.2 0.0 2.7 1.1 1.8 0.2 0.3 0.2 0.7 0.1 3.1 1.4 1.7 79.5 100
RoSEE 0.0 4.1 4.1 0.5 0.3 2.6 0.1 2.3 1.7 4.8 1.4 1.6 0.8 4.6 2.2 2.6 2.6 4.0 59.9 100
CtrlAsia 0.1 2.6 2.8 0.1 0.1 3.8 0.3 18.6 2.9 3.2 0.3 0.1 0.1 0.9 0.9 5.1 0.3 1.5 56.2 100
NAfrica 0.6 5.3 2.8 0.4 0.4 1.5 0.7 17.6 2.8 3.7 1.4 2.1 0.8 1.7 1.5 1.4 0.5 0.9 54.0 100
RoSSA 0.6 7.8 8.2 0.7 0.4 2.3 1.0 20.1 3.1 3.4 0.6 0.5 0.1 3.8 1.3 1.1 0.6 0.7 43.5 100
SAfrica 0.0 1.2 0.5 0.1 0.2 0.8 0.1 1.8 2.3 2.9 0.5 0.7 0.2 3.6 2.9 6.0 1.4 1.8 72.9 100
Emerged 0.4 1.2 0.8 0.1 0.2 0.7 0.3 0.7 1.7 2.5 1.6 0.6 0.1 3.2 4.1 3.4 2.9 10.5 64.7 100
RoOECD 0.1 0.4 0.4 0.1 0.1 0.3 0.1 1.1 0.8 2.0 0.3 0.2 0.0 2.9 2.5 2.1 1.9 4.4 80.4 100
RoWorld 0.1 2.7 1.1 0.2 0.2 1.5 0.2 21.8 1.3 2.0 0.8 0.3 0.1 3.2 1.6 1.3 1.0 2.2 58.1 100
132
5.3.1 Section Summary
This section provides for a deeper understanding of the simulation results, by providing detailed
details of key features of the underlying data such as the ratio of total exports of goods of
services to GDP for the 32 model regions; each region‘s EU share in total 2004 export revenue;
and the regional composition of EU imports from non-EU sources by commodity group in 2004.
5.4 Simulation Results Five different computable general equilibrium simulation scenarios are considered in this
section, and these are summarised in Table 5.7..
The first simulation evaluates the switch from the pre-2006 to the 2006-2009 EU GSP regime
while taking account of the actual observed degree of utilization of preferential GSP tariffs
including GSP+ and EBA preferences. The next two simulation runs aim to provide an overall
assessment of the EU GSP through a comparison of the observed 2004 benchmark equilibrium
with an anti-monde in which the EU GSP does not exist at all. In this case, the observed applied
benchmark tariff rates for the EU GSP beneficiary regions including EBA and GSP+ countries
switch to 2004 or 2006 MFN rates in the counterfactual equilibrium. The FULLGSP scenario
contemplates a switch in EU import tariff rates from the observed 2006 effective levels to the
effective levels that would prevail under a 100% utilization of GSP, GSP+ and EBA preferences.
The final scenario explores, to what extent developing countries could benefit from a further
extension of preferential treatment by simulating a complete removal of all EU duties on imports
from existing GSP beneficiaries.
Table 5.7: Simulation Scenarios
Code Scenario Description
GSP06 Change from applied 2004 EU GSP to applied EU 2006 GSP tariffs
MFN04 Abolition of EU GSP: Change from EU GSP tariffs to 2004 MFN tariffs
MFN06 Abolition of EU GSP: Change from EU GSP tariffs to 2006 MFN tariffs
FULLGSP Switch from observed 2006 utilization to 100% Utilization of EU GSP tariffs
ZEROTM Complete elimination of all EU import tariffs for GSP countries
Table 5.8 provides summary information on the size orders of the simulated percentage
changes of the power of EU imports by commodity group for each of the five scenarios, that is,
the percentage change of (1 + tariff rate), which provides a measure of the change in the price
faced by EU purchasers on impact, before secondary general equilibrium feedback effects that
affect the ex-tariff supply price of imports have played out.44 The table reports the simple
average across all GSP regions for each commodity group as well as the largest reduction – or
44
The percentage change in the power of the tariff is a far more meaningful measure of the impact of a tariff variation than the percentage change or the percentage-point change of the tariff rate. For instance, a 50 percent tariff cut applied to an initial EU tariff of 5 percent changes the price faced by EU consumers by about 2.4 percent, while a 50 percent tariff cut applied to an initial EU tariff of 50 percent, changes the EU consumer price by 16.7 percent (in the absence of general equilibrium feedbacks).
133
largest increase in the case of the MFN scenarios – among the region-specific changes in the
power of tariffs.
5.4.1 Change from 2004 to 2006 EU GSP – GSP06
As indicated by Table 5.8, the changes in applied GSP EU import duties between the 2004 and
the 2006 GSP regime at the commodity group level – which reflect both changes in the GSP
duty rates and changes in the actual utilization of preferences – are generally moderate to small
except for a sub-set of agricultural commodities including fruits, vegetables and rice.
Correspondingly, with some exceptions, the additional aggregate economy-wide welfare gains
for GSP beneficiaries due to the switch to the 2006-08 GSP regime, as measured by the
percentage change in real absorption45 in Table 5.9a and 9b,46 remain generally small. The
notable exceptions are the GSP+ countries Costa Rica and Ecuador and to a lesser extent
Colombia. All three countries benefit from a large boost to their EU vegetable and fruit exports
(Table 5.13) associated with a significant terms-of-trade improvement (Table 5.11). Ecuador
also benefits from a strong increase in its rice exports to the European Union. As the additional
EU demand for these commodities entails a noticeable real exchange rate appreciation for the
three countries, their EU exports of other commodities not subject to noteworthy tariff reductions
drop to some extent under the neoclassical full employment closure.
Table 5.17 reports the resulting impacts on the sectoral structure of production and factor
employment. To maintain a proper perspective on the percentage changes in gross real output
by sector in this table, the sectoral shares in GDP for each country shown in Table 5.6 need to
be borne in mind. The vegetable and fruit sector expands by 12 percent in Costa Rica and
Ecuador and by 6 percent in Colombia. The sector contributes between 3 and 4 percent of total
GDP in these economies in the 2004 benchmark period. In order to be able to expand, the
sector must drag labour, land and capital from other uses, and hence other domestic sectors
need to contract to some extent in this comparative-static simulation exercise with a fixed total
factor endowment.
Other strong sectoral expansion effects triggered by the incremental switch to the 2006-08 EU
GSP regime highlighted in Table 5.17 occur, e.g. in Sri Lanka‘s textile and transport equipment
sectors, Georgia‘s chemical, rubber and plastics industry and North Africa‘s oilseeds sector.
45
Real absorption is the sum of economy-wide private consumption, government consumption and
investment expenditure evaluated at constant benchmark period prices 46
A comparison of the aggregate welfare effects for the two labour market closures in Tables 9a and 9b
shows that the directions of the effects as well as the broad order of magnitudes is very similar between
the two labour market specification. Therefore the following discussion and subsequent tables will focus
exclusively on the standard neoclassical full employment closure in order to keep the exposition concise
and to avoid unnecessary repetition.
134
5.4.2 A World without the EU GSP – MFN04/06
We now turn to the overall assessment of the EU GSP through a comparison of the observed
2004 benchmark equilibrium with a counterfactual equilibrium in which the EU GSP does not
exist at all.
In this simulation experiment, the observed applied benchmark tariff rates for the EU GSP
beneficiary regions including EBA and GSP+ countries switch to 2004 or 2006 MFN rates in the
counterfactual equilibrium. In other words, all EU GSP preference margins are eliminated in this
scenario. Since the results for the MFN04 and MFN06 scenario are very similar as shown in the
aggregate Tables 5.8 to 5.11, the following discussion and the detailed tables focus on the
MFN04 scenario.47
The figures in the second and third column of Table 5.9 show the aggregate comparative-static
welfare effects associated with an abolition of the EU GSP. Accordingly, entries with a negative
sign indicate positive welfare gains attributable to the existence of the GSP. Among the EBA
regions in the model, Cambodia and Bangladesh benefit most from the scheme, while the EBA
Sub-Saharan Africa composite region as a whole appears to gain very little.48 Among the GSP+
countries the biggest gainers are again Ecuador and Costa Rica. Not surprisingly, welfare gains
are on the whole considerably smaller for the ordinary GSP countries, for which the preference
margins vis-à-vis MFN tariffs are moderate. Exceptions are North Africa and the Rest of
Southern and Eastern Europe region.
For the countries and regions enjoying the largest real absorption gains due to the GSP, these
gains are mirrored by a significant terms of trade appreciation compared to the counterfactual
no-GSP equilibrium (Table 5.11), as a result of the additional export demand from the EU. From
a macro perspective, it is precisely this terms-of-trade gain that allows countries to raise their
real absorption of final goods and services at a given factor endowment. This is since the terms-
of-trade gain means that more import goods can be obtained for each real unit of exports
shipped abroad. Regions which are more open to international trade as measured by the
export-GDP ratio in Table 5.3 gain more from a terms-of-trade gain of a given magnitude. For
instance, Pakistan and Cambodia both enjoy a terms-of trade gain on the order of 0.8 percent
due to the presence of the EU GSP. For the open economy of Cambodia with its export-GDP
ratio of 100 percent, this terms of trade gain translates into an aggregate welfare gain of 1.3
percent, whereas for the relatively closed economy of Pakistan with its export-GDP ratio of 19
percent, the aggregate welfare gain is barely noticeable.
Small aggregate welfare gains may go along with significant GSP impacts on exports to the EU
and domestic production at the sectoral level. In the case of Pakistan, for example, the
simulation suggests that without preferential access to EU markets, Pakistan‘s apparel exports
to the EU would shrink by 9 percent and as a result its domestic apparel sector would contract
47
This is arguably the ―neater‖ of the two MFN scenarios, since in both scenarios the MFN tariffs faced by
non-GSP beneficiaries are held fixed at their 2004 benchmark equilibrium levels. 48
However, it needs to be borne in mind here, that the RoSSA composite region also includes a number
of EBA countries along with non-LDC countries in sub-Saharan Africa.
135
by more than 3 percent (Tables 5.13 and 5.17). Pakistan‘s exports of processed food products
to the EU would drop by more than 20 percent, but since the export-output ratio and the EU
share in total exports of processed food products is relatively small, Pakistan‘s food processing
sector would contract by only 0.3 percent.
Tables 5.13 and 5.17 highlight the cases of strong sectoral EU export and domestic production
impacts of the EU GSP for each country.49 Apart from the reported significant trade and output
effects for a sub-set of agricultural commodities and regions, substantial expansionary50 impacts
of the EU GSP occur in particular in the textile, apparel and leather goods industries within a
number of GSP beneficiary regions.
5.4.3 Full Utilization of EU GSP Preferences – FULLGSP
Since concern about the underutilization of preferences due to administrative costs, restrictive
rules of origin and other obstacles, is a recurrent theme in the literature surrounding the GSP
(e.g. DeMaria, Drogue and Matthews, 2008), this simulation scenario considers a switch from
observed 2006 utilization to full utilization of EU GSP preferences.51
Table 5.8 shows that the average reduction in the power of the EU GSP import tariffs by sector
associated with a hypothetical move to full utilization is moderate across the board, ranging from
-0.1 to -2.4 percent, although these simple averages hide some double-digit percentage
reductions for individual countries for processed food products, textiles, apparel and rice.
Correspondingly, the aggregate economy-wide welfare gains resulting under this scenario
exceed 0.2 percent of benchmark absorption in only two cases, namely for GSP+ beneficiary Sri
Lanka and EBA beneficiary Cambodia. Tables 5.14 and 5.18 reveal the sources of these gains.
Sri Lanka significantly underutilizes the EU preferential treatment of processed food products,
textiles and apparel, while for Cambodia underutilization is most pronounced for its EU exports
of oilseeds and vegetable fats, rice and apparel. Sri Lanka‘s textile and apparel industries
expand by 3 to 4 percent and Cambodia‘s textile sector grows by more than 1 percent under full
utilization. The two tables exhibit a number of other very strong sectoral effects for individual
regions, e.g. Georgia and the GSP+ EE‘s apparel industries expand by 9.4 and 13.6 percent
respectively, yet because in both cases the benchmark contribution of the apparel sector to
GDP is miniscule, these strong sectoral effects do not translate into substantial economy-wide
welfare effects.
49
To increase the readability of the sectoral tables, sectors with very moderate impacts across all GSP
regions have been suppressed here and subsequently. 50
Recall that expansionary impacts of the EU GSP are indicated by a negative sign in these tables, since
the figures show the simulated impact of a counterfactual abolition of the system. 51
This scenario is implemented in the model by first determining the hypothetical percentage changes in
the power of tariffs due to a switch from actual applied 2006 EU tariffs towards tariffs under full
utilization.The observed benchmark equilibrium tariff powers are then reduced by these percentages.
136
5.4.4 Further Reform of the EU GSP: The Extreme Case - ZEROTM
Finally, in order to explore to which extent developing countries could potentially benefit from a
further extension of preferential EU treatment, we briefly consider the extreme borderline case
of a complete removal of all EU duties, including duties for graduated sectors on imports from
existing GSP beneficiaries.
As shown in Table 5.8, this simulation scenario involves, not surprisingly, substantial tariff
reductions for sugar, rice, fruits and processed food products, moderate reductions for textiles
and apparel and small reductions for other manufacturing sectors where remaining average
applied EU tariffs are already very low in the benchmark equilibrium.
The aggregate welfare effects reported in Table 5.9 indicate large gains for a subset of the Latin
American GSP+ countries, including Costa Rica, Ecuador and Columbia, as well as the standard
GSP countries Thailand, Argentina and Brazil. It is noteworthy that all EBA regions in the model
lose out – a clear-cut case of preference erosion. Preference erosion is likewise the main
explanation for the negative, or very small positive, economy-wide real absorption effects in a
range of other GSP regions. Like in all other scenarios under consideration, the welfare impact
on the EU is of a negligible order.
Tables 5.15 and 5.19 display the sectoral trade and production effects by country. As expected,
the largest effects occur in the agro-food sectors directly affected by substantial tariff reductions
identified above as well as in the textile and apparel sectors, where average reductions in the
EU border price tariff wedge are moderate on average but larger for individual countries.
5.4.5 Section Summary
This section uses five simulation scenarios to i) Evaluate the switch from earlier to later GSP
regimes; ii) and iii) to provide an assessment of the EU GSP scheme by comparing it with a
scenario where it does not exist; iv) a switch in EU import tariff rates to a scenario where there
the GSP schemes are fully utilised; and v) the extent that developing countries could benefit
from greater preferences if the EU removed all its duties from GSP beneficiary imports.
The results from this analysis indicate that the impact of changes over different GSP time-
frames are generally modest except for a sub-set of agricultural commodities. Among the EBA
regions Cambodia and Bangladesh benefit most from the scheme, while the EBA Sub-Saharan
Africa region gained very little. The biggest gainers are Ecuador and Costa Rica. The welfare
gains are generally a lot less for the ordinary GSP countries. The exceptions are North Africa
and the Rest of Southern and Eastern Europe region.
137
In addition to significant trade and output effects for a sub-set of agricultural commodities and
regions, the other most substantial expansionary52 effect of the EU GSP occur in the textile,
apparel and leather goods industries within a number of GSP beneficiary regions.
The average reduction in the power of the EU GSP import tariffs by sector associated with a
hypothetical move to full utilization is moderate. There are two cases of interest however. Sri
Lanka‘s textile and apparel industries expand by 3 to 4 percent and Cambodia‘s textile sector
grows by more than 1 percent under full utilization. Sri Lanka significantly underutilizes the EU
preferential treatment of processed food products, textiles and apparel, while for Cambodia
underutilization is most pronounced for its EU exports of oilseeds and vegetable fats, rice and
apparel.
When we consider the complete removal of all EU duties the welfare effects indicate larger
gains for a subset of the Latin American GSP+ countries, including Costa Rica, Ecuador and
Columbia, as well as the standard GSP countries Thailand, Argentina and Brazil. All EBA
regions in the model lose out from this simulation. Preference erosion is the main explanation
for the negative, or very small positive, economy-wide real absorption effects in a range of other
GSP regions.
52
Recall that expansionary impacts of the EU GSP are indicated by a negative sign in these tables, since
the figures show the simulated impact of a counterfactual abolition of the system.
138
Table 5.8: Percentage Changes in the Power of EU Import Tariffs by Scenario & Commodity Group
Average Max or Min
GSP06 MFN04 MFN06 FULLGSP ZEROTM GSP06 MFN04 MFN06 FULLGSP ZEROTM
Rice -18.1 5.4 2.1 -1.8 -29.7 -63.5 53.3 24.2 -22.9 -57.4
Vegetables, fruits -5.2 10.1 7.2 -0.3 -9.8 -31.2 132.2 88.0 -8.8 -45.2
Other crops 0.4 1.8 1.9 -0.2 -5.0 -0.9 22.5 23.1 -1.8 -31.5
Oils, fats -0.2 1.8 1.7 -0.6 -3.3 -3.6 21.5 15.7 -8.8 -36.9
Sugar prd -1.3 14.7 11.8 -0.1 -40.6 -6.6 71.8 60.8 -2.2 -73.0
Livestock prd 0.0 1.0 0.9 -0.1 -1.7 -0.6 16.1 14.0 -1.4 -12.2
Fishing prd 0.0 3.8 3.7 -1.2 -2.9 -0.4 14.4 14.4 -7.4 -9.8
Fossile fuels 0.0 0.2 0.2 -0.4 0.0 0.0 2.1 2.1 -3.6 -0.6
Mineral prd 0.0 1.3 1.4 -0.4 -0.6 -1.3 11.2 11.2 -3.4 -3.7
Other food prd -0.4 8.7 8.2 -0.8 -8.3 -5.5 39.6 24.8 -11.8 -37.4
Textiles -0.1 4.7 4.8 -1.3 -3.4 -3.5 11.2 11.2 -10.1 -9.8
Apparel -0.1 4.0 4.0 -2.4 -4.1 -1.8 11.2 11.2 -10.5 -9.9
Leather prd -0.1 2.2 2.3 -0.7 -2.4 -2.0 11.2 11.2 -3.7 -9.1
Other light mnf 0.0 0.6 0.6 -0.3 -0.3 0.0 2.1 2.1 -1.6 -1.5
Chemicals -0.1 2.4 2.5 -0.5 -0.7 -2.3 6.0 6.0 -3.0 -3.4
Metal prd 0.0 0.9 0.9 -0.2 -0.5 0.0 3.2 3.2 -2.2 -2.7
Transport equip -0.2 1.6 1.9 -0.7 -1.0 -6.0 13.4 13.4 -3.3 -6.2 Machinery, elec
equip 0.0 0.7 0.7 -0.6 -0.3 0.0 3.0 3.0 -2.2 -2.0
139
Table 5.9: Change in Real Absorption by Country and Scenario –
Full Employment Closure
GSP06 MFN04 MFN06 FULLGSP ZEROTM
EU 0.00 0.03 0.03 -0.02 -0.06
SriLanka 0.07 0.06 -0.02 0.21 0.16
Peru -0.01 -0.07 -0.07 0.01 -0.03
Ecuador 0.77 -0.45 -0.35 0.01 1.52
Colombia 0.20 -0.06 -0.05 0.00 0.35
CostaRica 1.03 -0.42 -0.28 0.08 1.68
GSP+ LA 0.08 -0.11 -0.09 0.01 0.20
GSP+ EE -0.01 0.09 0.09 -0.03 -0.19
Georgia 0.03 0.11 0.07 -0.01 -0.14
Cambodia 0.02 -1.27 -1.23 0.81 -0.23
Bangladesh -0.01 -0.31 -0.31 0.14 -0.18
EBA RoAs -0.01 0.09 0.08 -0.04 -0.12
EBA SSA 0.00 -0.15 -0.14 0.00 -0.01
China 0.00 -0.01 -0.01 0.10 0.24
Philippines -0.01 0.07 0.06 -0.02 0.04
India 0.00 0.00 0.00 0.00 0.01
Pakistan 0.00 -0.05 -0.04 -0.02 -0.08
Thailand 0.05 -0.15 -0.14 0.13 0.76
RoAsia 0.01 -0.16 -0.16 0.08 0.37
Argentina 0.05 -0.22 -0.20 0.02 0.45
Brazil 0.02 -0.08 -0.07 0.02 0.44
Caribbean 0.00 -0.03 -0.02 -0.01 -0.01
Russia 0.01 -0.12 -0.12 0.03 0.24
Ukraine 0.00 -0.09 -0.09 0.01 0.27
RoSEE 0.00 -0.26 -0.25 -0.01 0.06
CtrlAsia 0.01 -0.10 -0.09 0.01 0.11
NAfrica 0.00 -0.37 -0.36 0.01 0.13
RoSSA 0.03 -0.30 -0.26 0.00 0.20
SAfrica -0.01 -0.09 -0.09 0.01 0.08
Emerged 0.00 -0.04 -0.04 0.05 0.15
RoOECD 0.00 0.01 0.01 -0.01 -0.03
RoWorld 0.01 -0.18 -0.18 0.02 0.15
Changes > 0.25% highlighted.
140
Table 5.10: Change in Real Absorption by Country and Scenario –
Unlimited Supply of Unskilled Labour in Developing Countries
GSP06 MFN04 MFN06 FULLGSP ZEROTM
EU 0.00 0.02 0.02 -0.01 -0.05
SriLanka 0.11 0.07 -0.05 0.32 0.28
Peru -0.03 -0.09 -0.09 0.01 -0.09
Ecuador 1.25 -0.75 -0.59 0.01 2.39
Colombia 0.31 -0.12 -0.10 0.01 0.54
CostaRica 1.11 -0.47 -0.32 0.08 1.77
GSP+ LA 0.13 -0.15 -0.12 0.01 0.31
GSP+ EE -0.01 0.08 0.07 -0.02 -0.20
Georgia 0.06 0.08 0.01 -0.01 -0.10
Cambodia 0.02 -1.30 -1.26 0.88 -0.27
Bangladesh -0.01 -0.65 -0.65 0.31 -0.31
EBA RoAs -0.02 0.13 0.12 -0.05 -0.03
EBA SSA 0.00 -0.33 -0.33 0.01 0.01
China 0.00 0.01 0.01 0.13 0.29
Philippines -0.01 0.08 0.08 -0.02 0.05
India 0.00 0.00 0.00 0.00 0.06
Pakistan 0.00 -0.14 -0.13 -0.03 -0.07
Thailand 0.05 -0.15 -0.14 0.14 0.84
RoAsia 0.01 -0.16 -0.16 0.09 0.41
Argentina 0.08 -0.32 -0.28 0.02 0.67
Brazil 0.03 -0.10 -0.09 0.02 0.65
Caribbean 0.01 -0.08 -0.08 -0.01 0.12
Russia 0.01 -0.16 -0.15 0.04 0.28
Ukraine 0.00 -0.14 -0.14 0.01 0.36
RoSEE 0.00 -0.55 -0.55 -0.02 0.07
CtrlAsia 0.01 -0.12 -0.12 0.01 0.16
NAfrica 0.00 -0.69 -0.68 0.00 0.22
RoSSA 0.06 -0.53 -0.46 0.00 0.34
SAfrica -0.01 -0.12 -0.12 0.02 0.11
Emerged 0.00 -0.04 -0.04 0.06 0.19
RoOECD 0.00 0.01 0.01 -0.01 -0.03
RoWorld 0.01 -0.26 -0.26 0.02 0.19
Changes > 0.25% highlighted.
141
Table 5.11: Terms of Trade Change by Region and Scenario
GSP06 MFN04 MFN06 FULLGSP ZEROTM
EU 0.0 0.1 0.1 0.0 -0.2
SriLanka 0.2 0.0 -0.3 0.7 0.8
Peru -0.1 -0.2 -0.2 0.0 -0.3
Ecuador 1.2 -0.9 -0.7 0.0 1.9
Colombia 0.9 -0.3 -0.3 0.0 1.5
CostaRica 1.2 -0.4 -0.3 0.1 1.7
GSP+ LA 0.2 -0.2 -0.2 0.0 0.3
GSP+ EE 0.0 -0.1 -0.1 0.0 0.1
Georgia 0.2 -0.1 -0.2 0.0 0.2
Cambodia 0.0 -0.8 -0.7 0.5 -0.4
Bangladesh 0.0 -1.9 -1.9 0.7 -0.6
EBA RoAs 0.0 0.1 0.1 0.0 0.2
EBA SSA 0.0 -0.6 -0.6 0.1 0.1
China 0.0 0.0 0.0 0.2 0.3
Philippines 0.0 0.1 0.1 0.0 0.0
India 0.0 -0.1 -0.1 0.0 0.3
Pakistan 0.0 -0.8 -0.8 0.0 0.2
Thailand 0.0 -0.1 -0.1 0.1 0.6
RoAsia 0.0 -0.1 -0.1 0.0 0.1
Argentina 0.1 -0.5 -0.4 0.0 0.7
Brazil 0.1 -0.2 -0.1 0.0 1.3
Caribbean 0.0 -0.3 -0.3 0.0 0.5
Russia 0.0 -0.1 -0.1 0.0 0.2
Ukraine 0.0 -0.2 -0.2 0.0 0.5
RoSEE 0.0 -1.2 -1.2 0.0 0.3
CtrlAsia 0.0 -0.1 -0.1 0.0 0.0
NAfrica 0.0 -0.8 -0.7 0.0 0.1
RoSSA 0.1 -0.6 -0.5 0.0 0.2
SAfrica 0.0 -0.2 -0.2 0.0 0.1
Emerged 0.0 0.0 0.0 0.0 0.1
RoOECD 0.0 0.0 0.0 0.0 -0.1
RoWorld 0.0 -0.3 -0.3 0.0 0.2
142
Table 5.12: Change in Aggregate Export Volume by Country and Scenario
GSP06 MFN04 MFN06 FULLGSP ZEROTM
EU 0.02 -0.14 -0.14 0.04 0.31
SriLanka 0.20 -0.02 -0.20 -0.05 0.89
Peru -0.03 0.15 0.15 0.07 -0.10
Ecuador 0.12 0.07 0.10 0.06 0.03
Colombia -0.48 0.09 0.07 0.03 -0.92
CostaRica -1.11 0.55 0.41 0.06 -1.98
GSP+ LA -0.11 0.11 0.09 0.01 -0.32
GSP+ EE 0.01 -0.19 -0.19 0.04 0.28
Georgia 0.01 -0.32 -0.30 0.03 0.34
Cambodia -0.03 0.23 0.20 0.08 -0.18
Bangladesh 0.00 -0.68 -0.68 -0.55 -0.26
EBA RoAs 0.01 0.05 0.05 0.01 1.05
EBA SSA 0.00 -0.05 -0.05 -0.05 0.04
China -0.01 0.06 0.06 -0.04 -0.18
Philippines 0.00 -0.04 -0.04 0.01 -0.16
India 0.00 -0.08 -0.08 0.04 0.32
Pakistan 0.02 -0.84 -0.86 0.04 0.48
Thailand -0.06 0.15 0.14 -0.05 -0.70
RoAsia -0.02 0.17 0.17 -0.04 -0.26
Argentina -0.09 0.42 0.37 -0.03 -0.93
Brazil -0.04 0.17 0.16 -0.02 -0.95
Caribbean -0.01 -0.10 -0.10 0.02 0.03
Russia -0.02 0.18 0.17 -0.03 -0.29
Ukraine 0.01 0.02 0.01 0.01 -0.21
RoSEE 0.01 -0.93 -0.93 -0.01 -0.15
CtrlAsia -0.01 0.07 0.07 -0.01 -0.10
NAfrica 0.00 -0.09 -0.09 -0.01 -0.14
RoSSA -0.03 0.16 0.13 0.00 -0.08
SAfrica 0.01 0.07 0.07 -0.01 -0.08
Emerged -0.01 0.07 0.06 -0.02 -0.14
RoOECD 0.00 -0.02 -0.02 0.01 0.07
RoWorld -0.01 -0.01 -0.01 -0.02 -0.20
143
Table 5.13: Change in Export Volume to the EU by Commodity – GSP06
Ori
gin
Ric
e
Ve
ge
tab
les,
fru
its
Oth
er
cro
ps
Oils
, fa
ts
Su
ga
r p
rd
Min
era
l p
rd
Oth
er
foo
d p
rd
Te
xti
les
Ap
pa
rel
Le
ath
er
prd
Tra
ns
po
rt e
qu
ip
Ma
ch
ine
ry, e
lec
eq
uip
wEU -7.6 -1.4 0.2 0.0 -0.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0
wSriLanka 21.2 -1.0 -0.3 0.4 4.4 0.1 -0.1 5.6 2.0 2.0 6.9 -0.9
wPeru 24.7 2.3 -0.3 -0.3 0.5 -0.1 -0.1 -0.2 -0.1 -0.2 -0.1 -0.2
wEcuador 32.9 27.2 -6.2 -3.1 -1.7 -1.8 -2.5 -3.5 -2.7 -4.0 -3.2 -6.9
wColombia -6.0 24.5 -3.0 -1.5 0.6 -0.8 -1.2 -2.0 -1.3 -2.3 -1.2 -2.3
wCostaRica -9.8 24.2 -6.6 -3.4 -0.7 -1.8 -2.0 -3.5 -3.4 -5.1 -0.9 -3.7
wGSP+ LA -6.9 17.0 -0.9 -0.7 0.5 -0.3 -0.4 -0.7 -0.5 -0.5 -0.5 -0.8
wGSP+ EE -5.5 -1.0 0.1 -0.1 -0.2 0.0 -0.1 0.0 -0.1 0.0 0.0 0.0
wGeorgia 6.9 -0.9 -0.1 -0.3 0.4 -0.3 4.8 -0.3 -0.1 -0.3 -0.3 -0.4
wCambodia 6.9 1.8 -0.1 -0.2 -0.3 0.0 -0.1 -0.2 -0.1 -0.1 0.0 -0.1
wBangladesh 6.1 -1.2 0.1 -0.1 0.1 0.0 -0.1 0.0 0.0 0.0 0.0 0.0
wEBA RoAs -4.3 -1.1 0.1 -0.1 -0.2 0.0 -0.1 0.0 0.0 0.0 0.0 0.0
wEBA SSA 29.1 -1.1 0.0 -0.1 -0.2 0.0 -0.1 0.0 -0.1 0.0 0.0 -0.1
wChina 11.2 0.5 0.1 -0.1 2.1 0.0 -0.1 -0.1 -0.1 -0.1 0.0 0.0
wPhilippines -4.2 -1.2 0.0 -0.1 -0.3 0.0 -0.1 0.0 0.0 0.0 0.0 0.0
wIndia 9.3 -1.0 0.0 -0.1 5.6 0.0 -0.1 -0.1 -0.1 -0.1 -0.1 -0.1
wPakistan 6.2 -1.2 0.0 -0.1 1.4 -0.1 -0.2 -0.1 -0.1 -0.1 -0.1 -0.1
wThailand 27.0 0.4 -0.8 -0.3 3.7 -0.1 -0.1 -0.2 -0.1 -0.2 -0.1 -0.2
wRoAsia 28.3 -1.1 0.0 -0.1 -0.3 0.0 0.0 0.0 -0.1 -0.1 -0.1 -0.1
wArgentina 31.3 4.3 -6.5 -0.5 0.9 -0.1 1.6 -0.5 -0.2 -0.3 -0.3 -0.3
wBrazil 37.7 1.6 -1.7 -0.3 5.1 -0.1 2.1 -0.2 -0.1 -0.3 -0.2 -0.2
wCaribbean 44.7 6.5 -0.2 -0.2 1.7 -0.1 -0.1 -0.2 -0.1 -0.2 -0.1 -0.1
wRussia -1.9 -1.0 1.0 -0.1 -0.2 0.0 -0.7 -0.1 0.0 -0.1 -0.1 -0.1
wUkraine -4.0 -0.9 -2.1 -0.1 -0.3 0.0 1.0 -0.1 -0.1 0.0 0.0 0.0
wRoSEE -0.2 -0.6 -1.3 0.0 -0.2 0.0 -0.1 0.0 0.0 0.1 0.0 0.0
wCtrlAsia -6.0 -1.1 0.0 -0.1 -0.2 0.0 -0.1 -0.1 -0.1 -0.1 -0.1 -0.1
wNAfrica 9.1 -1.1 0.0 4.6 0.3 0.0 -0.1 -0.1 -0.1 -0.1 -0.1 -0.1
wRoSSA 130.2 2.7 -0.1 -0.2 0.3 -0.1 -0.1 -0.2 -0.2 -0.2 -0.2 -0.3
wSAfrica 20.4 -1.3 0.1 0.4 0.5 0.0 0.0 0.0 0.0 0.0 0.0 0.0
wEmerged 28.8 -1.2 0.2 0.0 -0.3 0.0 0.1 -0.1 -0.1 0.0 -0.1 -0.1
wRoOECD -5.4 -1.1 0.1 -0.1 -0.3 0.0 -0.1 0.0 -0.1 0.0 0.0 0.0
wRoWorld 29.1 -1.1 0.1 0.0 0.7 0.0 0.3 -0.1 -0.1 -0.1 -0.1 -0.1
Export expansions > 5% highlighted
144
Table 5.14: Change in Export Volume to the EU by Commodity – MFN04
Ric
e
Ve
ge
tab
les,
fru
its
Oth
er
cro
ps
Oils
, fa
ts
Su
ga
r p
rd
Liv
es
tock
prd
Fis
hin
g p
rd
Fo
ss
ile f
ue
ls
Oth
er
foo
d p
rd
Te
xti
les
Ap
pa
rel
Le
ath
er
prd
Ch
em
icals
Tra
ns
po
rt e
qu
ip
Ma
ch
ine
ry, e
lec
eq
uip
EU 0.5 0.8 0.2 0.3 8.6 0.1 0.3 -0.3 0.3 1.2 1.3 0.7 0.1 0.0 -0.1
SriLanka -0.5 0.9 0.1 -0.8 7.3 0.4 -0.6 -0.1 -0.3 0.3 1.1 -3.3 -2.2 -3.0 -1.3
Peru 0.9 -1.9 0.4 -0.9 7.5 0.4 -0.5 0.3 -5.3 -6.8 -10.6 0.4 -2.2 0.3 0.9
Ecuador 2.2 -5.9 -2.3 2.1 8.8 0.6 -0.3 1.7 -8.2 -9.3 -2.0 2.7 2.1 2.1 3.9
Colombia 1.1 -2.0 1.1 -3.2 8.3 0.4 0.1 0.3 -9.6 -10.1 -8.6 0.4 -5.5 0.9 0.6
CostaRica 1.8 -5.4 2.1 1.3 8.4 0.5 0.2 0.0 -9.8 -10.5 1.1 -2.9 -1.6 -0.7 1.9
GSP+ LA 1.4 -6.1 0.3 1.2 -26.6 -2.9 -5.7 -0.4 -8.0 -4.8 -1.1 0.5 -3.5 0.8 0.6
GSP+ EE 0.7 -0.8 0.8 0.7 7.3 0.4 0.5 0.2 0.3 1.2 0.4 1.1 -2.9 0.4 0.6
Georgia -0.3 -0.7 0.8 0.9 -12.8 0.4 0.5 -0.2 -1.8 2.1 2.5 -1.0 -3.8 0.9 0.4
Cambodia -26.1 -38.6 3.7 4.1 7.9 0.8 0.5 4.2 -5.9 -8.8 -1.3 -13.5 6.2 1.4 3.0
Bangladesh -8.7 -5.5 0.8 3.0 -23.4 1.4 0.8 5.8 -6.1 -7.0 -0.4 -5.1 0.7 -8.9 5.6
EBA RoAs 0.6 0.8 0.8 0.5 5.5 0.3 0.5 -0.4 0.5 1.6 2.4 1.0 0.3 0.3 0.1
EBA SSA 1.7 0.1 -1.9 -2.1 -31.5 0.4 -4.9 0.7 -7.2 -10.8 -9.6 -1.1 0.7 0.8 2.2
China -1.6 -2.3 0.2 0.4 7.8 0.3 0.3 -0.1 -2.6 1.0 1.6 1.2 -0.7 -1.4 0.5
Philippines 0.5 0.8 0.5 0.4 7.6 0.3 0.3 0.1 0.6 1.3 1.5 1.0 0.4 0.4 0.3
India 0.4 0.8 0.8 -0.8 7.0 0.4 -0.7 -3.3 -2.3 0.2 -0.1 -1.4 -3.2 -2.2 -1.0
Pakistan 1.3 0.7 0.6 1.8 4.3 1.1 -4.6 1.2 -22.1 -5.5 -8.6 -0.6 -4.1 -0.5 2.5
Thailand -2.6 -18.9 -0.1 -1.1 6.6 0.5 0.5 0.2 -0.2 0.5 0.7 1.4 0.4 -2.3 0.1
RoAsia 0.8 0.9 0.6 0.2 7.2 -0.3 -0.7 0.1 -2.6 0.6 1.8 -3.9 -1.2 -1.2 0.6
Argentina 1.4 -5.8 0.8 2.3 8.3 0.2 -0.3 0.5 -6.6 2.2 -0.2 1.9 -1.7 -0.5 -0.1
Brazil 1.2 -0.9 0.9 1.3 5.9 0.3 -1.1 0.0 -3.6 -0.1 -0.3 2.2 -2.3 -1.9 -0.6
Caribbean 0.1 -3.7 0.2 0.9 -26.2 0.4 -2.7 -3.0 -3.9 -10.3 -3.4 -2.6 -3.7 0.9 0.1
Russia 1.3 0.5 -3.6 0.4 7.6 0.4 -0.7 0.3 -5.3 0.2 1.2 0.5 0.1 0.2 0.3
Ukraine 0.7 0.2 -2.3 -0.7 7.7 0.5 0.4 -0.7 -2.8 1.5 0.9 0.7 -1.4 -0.7 -1.3
RoSEE -35.7 -0.6 -16.0 2.2 -22.9 -6.7 -2.8 2.2 -8.9 -10.0 -16.0 -8.0 -1.7 1.3 1.7
CtrlAsia 1.2 1.1 0.9 1.0 7.3 0.4 0.4 0.3 0.7 1.8 1.9 1.4 0.8 0.8 0.9
NAfrica 3.8 -2.7 0.7 -18.3 5.0 0.9 -4.7 1.5 -6.6 -9.3 -10.1 -4.3 -2.5 -1.1 0.7
RoSSA 2.1 -7.5 0.4 0.1 -35.4 0.5 -5.1 0.7 -9.9 -9.4 -9.5 -0.5 1.0 2.7 2.8
SAfrica -14.6 -0.2 -0.2 -1.2 5.9 0.5 0.3 0.3 -1.1 -2.8 -7.8 -0.9 -2.5 -1.4 -0.8
Emerged 1.2 -1.7 -0.4 -1.0 2.1 -1.5 0.0 0.0 -2.5 1.3 1.6 1.1 -3.3 0.0 0.6
RoOECD 0.8 0.9 0.5 0.6 8.1 0.3 0.3 -0.1 0.5 1.2 1.4 1.0 0.4 0.4 0.4
RoWorld 0.3 -0.7 -1.9 -1.4 -29.9 0.4 -4.4 0.5 -6.9 -6.3 -7.4 -3.4 -2.0 -4.0 -2.4
Export contractions > 5% highlighted.
145
Table 5.15: Change in Export Volume to the EU by Commodity - FULLGSP
Ric
e
Ve
ge
tab
les,
fru
its
Oils
, fa
ts
Fo
ss
ile f
ue
ls
Oth
er
foo
d p
rd
Te
xti
les
Ap
pa
rel
Le
ath
er
prd
wEU 0.0 0.0 0.0 0.1 0.0 -0.2 -0.4 -0.7
wSriLanka -1.0 -0.2 -1.8 1.9 5.3 8.2 10.5 -1.7
wPeru -0.2 -0.1 -0.2 -0.2 0.0 0.0 -0.1 -0.7
wEcuador -0.2 -0.1 -0.3 -0.1 0.0 0.0 -0.2 -0.3
wColombia -0.2 -0.1 -0.2 -0.1 0.0 0.1 1.6 -0.9
wCostaRica -0.3 -0.2 -0.2 3.4 0.3 0.3 10.5 -0.9
wGSP+ LA -0.3 -0.1 -0.3 -0.1 0.0 5.6 9.8 -0.8
wGSP+ EE -0.2 0.0 -0.2 -0.1 10.7 6.9 21.7 0.3
wGeorgia 0.0 0.0 -0.2 -0.2 0.0 15.8 20.0 -0.7
wCambodia 7.1 4.6 7.8 -2.7 -1.0 2.9 6.7 -3.2
wBangladesh 8.0 -0.8 -1.6 0.6 -1.2 -0.3 9.2 -5.7
wEBA RoAs -0.2 -0.1 -0.2 0.0 -0.1 -0.4 -0.8 -0.8
wEBA SSA 26.1 -0.1 -0.2 -0.1 0.0 -0.1 0.1 -0.7
wChina -0.4 -0.2 -0.5 -0.5 -0.3 1.3 1.9 4.7
wPhilippines -0.1 -0.1 -0.2 -0.2 -0.2 -0.4 -0.6 -0.9
wIndia -0.3 -0.1 -0.3 0.0 -0.1 0.2 0.5 1.4
wPakistan -0.1 -0.1 -0.2 -0.1 -0.1 -0.3 -0.6 -0.1
wThailand -0.5 -0.2 0.4 -0.5 1.1 0.5 0.6 4.3
wRoAsia -0.4 -0.2 2.9 -0.3 0.4 0.7 1.6 0.8
wArgentina -0.2 -0.1 -0.3 -0.2 -0.2 -0.4 -0.1 1.4
wBrazil -0.3 -0.1 -0.3 -0.2 -0.2 -0.4 -0.2 2.4
wCaribbean -0.1 -0.1 -0.2 1.2 -0.1 -0.3 -0.5 -0.5
wRussia -0.1 0.2 -0.1 -0.1 -0.2 -0.1 1.6 -0.6
wUkraine -0.1 0.0 -0.1 0.0 -0.1 0.9 1.7 -0.9
wRoSEE 0.0 0.0 -0.1 0.3 0.0 -0.3 -0.7 -1.0
wCtrlAsia -0.2 -0.1 -0.3 -0.1 -0.1 -0.4 -0.6 -0.9
wNAfrica 3.9 -0.1 -0.2 0.1 -0.1 -0.4 -0.5 -0.8
wRoSSA -0.1 -0.1 -0.2 -0.1 -0.1 -0.3 -0.6 -0.9
wSAfrica 0.0 -0.1 -0.2 -0.2 -0.1 -0.3 -0.5 -0.8
wEmerged -0.4 -0.2 0.5 0.1 -0.2 1.7 2.1 4.2
wRoOECD -0.2 -0.1 -0.2 -0.1 -0.1 -0.3 -0.5 -0.8
wRoWorld -0.1 -0.1 -0.2 -0.1 -0.1 -0.5 -0.7 -0.9 Export expansions > 2.5% highlighted.
146
Table 5.16: Change in Export Volume to the EU by Commodity – ZEROTM
Ric
e
Ve
ge
tab
les,
fru
its
Oth
er
cro
ps
Oils
, fa
ts
Su
ga
r p
rd
Liv
es
tock
prd
Fo
ss
ile f
ue
ls
Oth
er
foo
d p
rd
Te
xti
les
Ap
pa
rel
Le
ath
er
prd
Ch
em
icals
Me
tal
prd
Tra
ns
po
rt e
qu
ip
Ma
ch
ine
ry, e
lec
eq
uip
wEU -24.0 -2.4 -1.2 -1.5 -20.6 -0.2 0.4 -0.6 -1.2 -1.4 -2.3 0.0 0.3 0.0 0.3
wSriLanka 58.7 -2.3 3.0 -2.4 132.3 -0.9 -2.3 4.8 14.0 11.1 3.7 -3.5 -3.6 -1.0 -3.7
wPeru -5.6 2.6 -1.5 -2.3 53.4 -1.0 -0.9 -1.0 -1.9 -2.4 -2.7 -1.3 -0.9 -1.2 -1.6
wEcuador 21.2 49.5 -13.7 -5.1 85.9 -5.6 -6.5 -5.0 -8.6 -7.0 -10.0 -7.3 -7.6 -7.0 -14.2
wColombia -12.2 41.4 -6.0 -4.5 147.5 -2.8 -2.3 -1.5 -5.6 -4.5 -7.1 -3.7 -4.1 -3.1 -5.3
wCostaRica -24.8 34.6 -10.0 -6.5 30.5 -4.6 -0.3 0.0 -6.6 -7.2 -9.6 -3.2 -6.0 -2.0 -6.9
wGSP+ LA -17.3 28.8 -2.6 -3.6 122.4 -1.6 -1.1 0.9 -3.4 -3.0 -3.5 -2.0 -1.8 -1.9 -3.1
wGSP+ EE -18.2 -2.3 -2.0 -1.7 -16.4 -0.5 -0.3 5.8 7.2 7.5 -1.1 1.6 1.2 0.1 -0.3
wGeorgia 24.7 -1.9 -2.0 7.0 44.7 -0.7 -0.5 7.9 5.7 12.8 1.8 2.5 1.7 0.6 -1.6
wCambodia 30.8 -2.5 -2.2 -2.0 -15.1 -0.8 0.6 -0.7 -2.4 -1.7 -3.6 0.6 0.7 -1.0 -0.6
wBangladesh 32.4 -2.5 -1.4 -1.7 41.0 -0.4 1.9 -0.8 -0.8 -1.7 -1.8 0.0 0.2 0.1 1.2
wEBA RoAs -10.0 3.5 -2.3 -3.3 49.0 -1.3 -1.1 1.5 7.8 11.8 -1.1 -1.5 -2.1 -2.1 -2.6
wEBA SSA -3.3 -2.4 -2.0 -2.0 33.8 -0.7 -0.3 -1.1 -2.0 -2.3 -2.8 -0.9 -1.4 -1.1 -1.4
wChina 77.1 20.3 2.3 -2.1 67.6 -0.5 -1.3 9.4 8.1 10.9 10.8 2.0 1.3 -0.9 0.3
wPhilippines -8.0 1.2 2.0 5.5 97.5 -0.7 -0.7 10.3 9.9 9.3 6.5 -0.7 -0.9 0.6 -1.5
wIndia 42.2 -2.2 -0.7 -1.8 2.1 1.3 -1.2 7.1 6.8 11.7 1.0 -0.8 -1.5 2.1 -1.6
wPakistan 51.1 -2.6 -2.2 -2.4 -13.4 -0.9 -0.7 -0.3 2.7 -1.2 -0.1 -1.0 -1.1 -1.3 -1.5
wThailand 102.3 -0.9 2.0 -0.5 16.3 0.1 -2.2 14.2 6.0 7.5 10.2 -1.4 -2.5 5.6 -1.5
wRoAsia 10.2 -1.9 -1.7 4.3 -8.0 -0.2 -1.0 5.3 8.6 10.5 11.7 -0.5 -1.6 1.6 -1.2
wArgentina 3.4 5.2 15.1 -6.3 42.3 6.3 -2.9 16.6 -0.8 0.6 0.1 -3.2 -2.4 -3.0 -4.2
wBrazil -14.0 0.0 21.9 -5.9 151.2 2.3 -3.0 33.1 3.7 5.0 -1.2 -2.7 -2.0 -3.9 -4.8
wCaribbean 56.0 15.5 -2.5 -2.4 128.3 2.7 -0.4 1.2 -3.3 -3.2 -4.1 -0.5 -2.1 -1.6 -2.4
wRussia 69.9 -0.9 34.2 -1.8 -10.4 -0.8 -0.4 7.5 5.5 8.0 -1.9 0.0 -0.3 -0.3 -1.7
wUkraine -13.7 -1.9 46.2 -2.7 89.3 -0.3 -1.4 10.9 9.8 15.6 0.8 -0.7 -1.6 -1.6 -2.3
wRoSEE -8.4 -1.3 12.4 -0.5 75.7 1.9 -0.2 5.0 -1.5 -2.6 -3.8 -0.5 -0.6 -0.7 -0.9
wCtrlAsia 44.2 -2.0 -0.2 -2.6 -17.5 -0.6 -0.8 11.4 4.3 9.4 -1.9 -0.9 -0.2 -0.6 -0.8
wNAfrica 55.4 -1.2 -2.5 86.5 -13.3 -0.7 -0.6 0.5 -3.3 -3.2 -4.2 -1.3 -1.7 -1.2 -2.6
wRoSSA -16.3 2.8 -3.1 -2.7 90.2 -1.1 -0.8 1.3 -3.4 -3.3 -4.8 -2.0 -2.2 -2.9 -3.3
wSAfrica 32.6 -2.3 -0.9 -0.8 27.7 -0.5 -0.6 7.5 -1.4 -1.3 -2.8 -1.0 -1.0 2.4 -1.5
wEmerged 5.8 -1.2 4.5 2.3 14.3 4.6 -0.7 5.6 9.1 11.0 8.7 0.7 -0.4 5.2 -0.6
wRoOECD -18.1 -2.5 -1.9 -2.1 -19.6 -0.7 -0.4 -1.2 -1.8 -2.1 -2.7 -0.8 -0.7 -0.9 -0.9
wRoWorld 56.7 -1.4 0.0 9.7 121.4 -0.3 -0.5 6.3 -2.3 -2.9 -2.4 -0.8 -1.1 -1.4 -1.9
Export expansions > 5% highlighted.
147
Table 5.17: Change in Real Output by Sector and Region – GSP06
Ric
e
Ve
ge
tab
les,
fru
its
Oth
er
cro
ps
Oils
an
d f
ats
Su
ga
r p
rd
Oth
er
foo
d p
rd
Te
xti
les
Ap
pa
rel
Le
ath
er
prd
Ch
em
icals
Me
tal
prd
Tra
ns
po
rt e
qu
ip
Ma
ch
ine
ry, e
lec
eq
uip
EU -5.3 -0.8 0.1 0.0 -0.1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
SriLanka 0.1 0.0 -0.3 -0.7 -0.3 -0.1 2.7 0.4 0.4 -0.6 -0.8 2.5 -0.7
Peru 0.0 0.2 0.0 0.1 0.0 0.0 0.0 0.0 0.0 0.1 -0.1 0.0 0.0
Ecuador -0.3 11.7 -3.4 -0.7 -0.1 -0.8 -1.1 -0.3 -0.8 -1.3 -2.2 -1.7 -5.1
Colombia -0.2 6.1 -2.2 -0.7 -0.3 -0.2 -1.1 -0.5 -1.3 -0.7 -1.2 -0.4 -1.5
CostaRica -1.9 12.6 -4.0 -1.9 -0.8 -0.6 -2.6 -2.6 -4.3 -0.3 -2.7 0.0 -3.7
GSP+ LA 0.0 2.1 -0.4 -0.2 0.1 0.0 -0.5 -0.3 -0.1 0.0 -0.3 -0.1 -0.6
GSP+ EE 0.0 0.0 0.0 0.0 0.1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
Georgia -0.1 0.0 0.0 -0.2 -0.3 0.4 -0.2 -0.2 -0.1 2.0 -0.5 -0.2 -0.3
Cambodia 0.2 0.0 0.0 0.0 0.0 0.0 -0.1 0.0 -0.1 -0.1 0.0 0.0 0.0
Bangladesh 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
EBA RoAs 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.1 0.0 0.0 0.0
EBA SSA 0.2 -0.1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
China 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
Philippines 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
India 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
Pakistan 0.2 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
Thailand 1.1 -0.2 -0.3 -0.1 -0.1 -0.1 -0.1 0.0 -0.1 -0.2 -0.1 0.0 -0.1
RoAsia 0.1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 -0.1 0.0 0.0 0.0 0.0
Argentina 0.1 1.9 -0.7 -0.3 0.1 0.2 -0.3 0.0 -0.1 -0.1 -0.1 -0.1 -0.2
Brazil 0.1 0.2 -0.1 -0.1 0.0 0.2 0.0 0.0 -0.1 0.0 -0.1 -0.1 -0.1
Caribbean 1.5 1.0 -0.1 0.0 0.3 0.0 -0.1 -0.1 -0.1 0.0 -0.1 -0.1 -0.1
Russia 0.1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
Ukraine 0.0 0.0 -0.1 0.0 0.0 0.1 0.0 0.0 0.0 0.0 0.0 0.0 0.0
RoSEE 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.1 0.0 0.0 0.0 0.0
CtrlAsia 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
NAfrica 0.1 -0.1 0.0 2.5 0.0 0.0 -0.1 0.0 0.0 0.0 0.0 0.0 -0.1
RoSSA 0.6 0.2 -0.1 -0.1 0.1 0.0 -0.1 -0.1 -0.1 -0.1 -0.1 -0.2 -0.2
SAfrica 0.0 -0.4 0.1 0.1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
Emerged 0.0 -0.1 0.0 0.1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
RoOECD 0.0 -0.1 0.1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
RoWorld 0.8 -0.1 0.0 0.0 0.1 0.0 -0.1 0.0 0.0 0.0 0.0 0.0 -0.1
Output expansions > 1% highlighted.
148
Table 5.18: Change in Real Output by Sector and Region – MFN04
Ric
e
Ve
ge
tab
les,
fru
its
Oils
an
d f
ats
Su
ga
r p
rd
Fis
hin
g p
rd
Oth
er
foo
d p
rd
Te
xti
les
Ap
pa
rel
Le
ath
er
prd
Ch
em
icals
Tra
ns
po
rt e
qu
ip
EU 0.3 0.5 0.1 3.8 0.2 0.1 0.5 0.6 0.3 0.0 -0.1
SriLanka 0.0 0.1 -0.4 -0.2 0.0 0.0 0.2 0.4 -1.0 -0.7 -1.1
Peru -0.1 -0.2 -0.2 -0.1 -0.1 -0.3 -0.2 -0.2 0.0 0.0 0.1
Ecuador -0.3 -2.2 0.0 -0.3 -1.0 -1.2 0.4 0.1 0.4 1.0 0.9
Colombia 0.0 -0.4 -0.4 0.0 -0.1 -0.2 -0.2 -0.1 0.2 0.1 0.1
CostaRica 0.2 -2.7 0.1 0.2 -0.4 -0.6 0.2 0.8 0.7 0.1 -0.1
GSP+ LA -0.1 -0.6 0.0 -0.4 -0.2 -0.2 0.6 0.3 0.1 0.1 0.1
GSP+ EE -0.4 0.0 -0.2 -0.5 0.3 0.0 0.2 0.2 0.2 -0.5 -0.2
Georgia -0.1 0.0 0.1 -0.6 0.2 -0.1 0.9 1.5 0.1 -1.8 0.3
Cambodia -0.7 -0.5 1.0 0.0 -0.6 -0.1 -4.5 2.4 -5.4 4.5 -0.2
Bangladesh 0.0 0.1 0.7 -0.3 -0.4 -0.3 -1.8 1.1 -0.1 1.2 1.8
EBA RoAs 0.0 -0.1 -0.2 0.4 0.2 0.0 0.8 1.5 0.1 -0.3 -0.1
EBA SSA 0.2 0.1 0.1 -2.3 -1.0 -0.9 -1.8 -1.0 0.1 0.3 0.9
China 0.0 0.0 -0.2 -0.1 0.0 0.0 0.0 0.2 0.2 -0.1 -0.1
Philippines 0.0 0.0 -0.2 0.0 0.0 0.0 0.2 0.1 0.2 -0.1 0.0
India 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.1 -0.6 -0.1 -0.1
Pakistan 0.3 0.1 0.4 0.2 -0.1 -0.3 -0.9 -3.3 0.4 0.3 0.7
Thailand -0.1 -0.9 0.0 0.0 0.0 0.0 0.2 0.1 0.3 0.3 -0.4
RoAsia -0.1 0.0 -0.3 0.0 -0.1 -0.1 0.2 0.8 -2.6 0.0 0.0
Argentina -0.1 -2.5 1.5 -0.2 -0.6 -0.7 0.7 0.0 0.7 0.6 0.6
Brazil -0.2 -0.1 0.4 -0.1 -0.2 -0.3 0.1 0.0 1.0 0.1 0.0
Caribbean 0.2 -0.5 -0.1 -3.7 -0.2 -0.3 -0.3 -0.1 0.0 -0.1 0.4
Russia -0.1 -0.1 -0.1 -0.3 0.0 -0.2 -0.3 0.2 0.1 0.1 0.1
Ukraine -0.1 -0.1 -0.3 0.0 -0.1 -0.2 0.8 0.8 0.5 -0.4 0.2
RoSEE -0.2 0.1 0.5 -1.3 -1.4 -0.3 -5.9 -13.2 -5.5 0.5 1.4
CtrlAsia -0.1 -0.1 -0.1 -0.8 -0.1 -0.1 0.4 0.4 0.2 0.2 0.1
NAfrica 0.2 -0.3 -8.9 0.1 -1.5 -0.4 -4.3 -5.3 -1.7 0.1 0.4
RoSSA 0.4 -0.5 0.1 -13.0 -1.2 -1.3 -1.5 -0.5 0.3 0.8 2.1
SAfrica -1.3 -0.1 -0.4 0.1 0.0 -0.1 -0.2 -0.2 0.0 -0.1 0.0
Emerged 0.0 -0.1 -0.2 -0.1 0.0 -0.1 0.1 0.1 0.1 -0.2 0.0
RoOECD 0.0 0.0 -0.1 0.0 0.0 0.0 -0.1 -0.1 0.0 0.0 0.0
RoWorld 0.1 -0.1 -0.1 -1.3 -0.4 -0.3 -3.4 -3.5 -0.7 0.1 -1.2
Output contractions > 1% highlighted.
149
Table 5.19: Change in Real Output by Sector and Region – FULLGSP
Ric
e
Ve
ge
tab
les,
fru
its
Oth
er
cro
ps
Oils
an
d f
ats
Su
ga
r p
rd
Oth
er
foo
d p
rd
Te
xti
les
Ap
pa
rel
Le
ath
er
prd
Oth
er
lig
ht
mn
f
Ch
em
icals
Me
tal
prd
Tra
ns
po
rt e
qu
ip
Ma
ch
ine
ry, e
lec
eq
uip
EU 0.0 0.0 0.0 0.0 0.0 0.0 -0.1 -0.2 -0.4 0.0 0.0 0.0 0.0 0.1
SriLanka 0.0 0.2 -0.7 -1.7 -0.9 0.0 3.7 3.1 -0.9 -1.9 -1.6 -2.4 -0.6 -1.7
Peru 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
Ecuador 0.0 -0.1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.1 0.1 0.1
Colombia 0.0 -0.1 0.2 0.0 0.0 0.0 0.1 0.1 -0.2 0.0 0.0 0.0 0.0 0.0
CostaRica 0.0 -0.2 0.3 0.0 0.0 0.0 0.1 0.4 -0.2 0.1 -0.1 -0.1 0.0 -0.2
GSP+ LA 0.0 0.0 0.0 0.0 0.0 0.0 0.2 0.2 -0.1 0.0 0.0 0.0 0.0 0.0
GSP+ EE 0.1 0.0 0.0 0.0 0.0 0.1 1.7 13.6 0.8 -0.1 0.3 0.1 0.0 0.5
Georgia 0.2 0.0 0.0 -0.1 0.0 0.0 5.2 9.4 -0.1 0.0 0.4 -0.1 0.0 0.1
Cambodia 0.3 0.0 -0.5 -0.3 0.1 -0.2 1.2 0.1 -3.1 -1.0 -3.1 -1.6 0.1 -0.6
Bangladesh 0.0 0.0 -0.3 -0.6 0.0 -0.3 0.5 4.5 -4.9 -0.1 -0.9 -0.9 -1.5 -3.3
EBA RoAs 0.0 0.0 0.0 0.1 0.0 0.0 -0.2 -0.5 -0.2 0.1 0.1 0.2 0.0 0.2
EBA SSA 0.2 0.0 0.0 0.0 0.0 0.0 0.1 0.0 -0.1 0.0 0.0 -0.3 0.3 0.1
China 0.0 0.0 -0.1 -0.1 0.0 0.0 0.0 0.1 0.8 0.0 0.0 -0.1 -0.1 -0.2
Philippines 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 -0.2 0.0 0.1 0.1 0.0 0.0
India 0.0 0.0 0.0 0.0 0.0 0.0 0.1 0.3 0.7 0.0 0.0 0.0 0.0 0.0
Pakistan 0.0 0.0 0.1 0.0 0.0 0.0 0.1 -0.2 0.0 0.1 0.1 0.1 0.0 0.1
Thailand -0.1 0.0 -0.1 -0.1 0.0 0.1 0.1 0.1 0.8 0.0 0.0 -0.3 0.0 -0.2
RoAsia 0.0 0.0 -0.1 0.3 0.0 0.0 0.3 0.4 0.5 0.0 0.0 -0.2 0.0 -0.2
Argentina 0.0 -0.1 0.0 -0.1 0.0 0.0 0.0 0.0 0.2 0.0 -0.1 -0.1 0.0 0.0
Brazil 0.0 0.0 0.0 -0.1 0.0 0.0 0.0 0.0 0.7 0.0 0.0 -0.1 0.0 0.0
Caribbean 0.0 0.0 0.0 0.0 0.0 0.0 0.1 0.0 0.0 0.0 0.0 0.0 0.0 0.1
Russia 0.1 0.0 0.0 0.0 0.0 0.0 0.0 0.2 -0.1 0.0 0.1 -0.1 0.0 0.0
Ukraine 0.0 0.0 0.0 0.0 0.0 0.0 0.6 1.4 -0.5 -0.1 0.1 -0.1 0.0 0.1
RoSEE 0.0 0.0 0.0 0.0 0.0 0.0 -0.2 -0.5 -0.7 0.0 0.0 0.1 0.1 0.1
CtrlAsia 0.0 0.0 0.2 0.0 0.0 0.0 0.0 0.0 -0.1 0.0 0.0 0.0 0.0 0.0
NAfrica 0.1 0.0 0.0 0.0 0.0 0.0 -0.2 -0.3 -0.3 0.0 0.0 0.0 0.0 0.0
RoSSA 0.1 0.0 0.0 0.0 -0.1 0.0 0.0 0.0 -0.4 0.0 0.0 0.0 -0.2 0.4
SAfrica -0.1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.1 0.0
Emerged 0.0 0.0 0.0 -0.1 0.0 0.0 0.3 0.2 0.4 0.0 0.0 -0.1 0.1 -0.1
RoOECD 0.0 0.0 0.0 0.0 0.0 0.0 0.1 0.0 0.0 0.0 0.0 0.0 0.0 0.0
RoWorld 0.1 0.0 0.0 0.0 0.0 0.0 -0.3 -0.3 -0.2 0.0 -0.1 -0.1 -0.1 -0.1
Output expansions > 1% highlighted.
150
5.5 Section 5: Conclusions
This section follows the GLOBE model as its framework of analysis. This is a multi-regional
and multi-sectoral CGE model of global production and trade to assess trade creation, trade
diversion, sectoral employment and structural transformation effects triggered by the GSP
scheme. It also allows for an evaluation of the aggregate welfare effects by country that
takes full account of indirect open-economy general equilibrium feedback linkages.
The main conclusions from the CGE analysis can be summarized as follows:
The incremental change in applied EU GSP tariff rates from the pre-2006 to the
2006-08 system generates only small aggregate welfare gains for GSP beneficiaries
except for a sub-set of Latin American GSP+ countries.
Among the EBA regions in the model, Cambodia and Bangladesh benefit most from
the existence of the EU scheme, while the EBA Sub-Saharan Africa composite
region as a whole gains very little. However, due to data constraints not all actual
EBA countries in sub-Saharan Africa are included in this model region. Among the
GSP+ countries the biggest gainers are again Ecuador and Costa Rica. Not
surprisingly, welfare gains are on the whole considerably smaller for the ordinary
GSP countries, for which the preference margins vis-à-vis MFN tariffs are moderate.
Exceptions are North Africa and the Rest of Southern and Eastern Europe region.
Apart from significant trade and output effects for a sub-set of agricultural
commodities and regions, substantial expansionary impacts of the EU GSP occur in
particular in the textile, apparel and leather goods industries within a number of GSP
beneficiary regions.
With few exceptions and contrary to received wisdom, underutilization of existing EU
GSP preferences does not appear to be a major factor that would reduce the actual
realized gains from the existing GSP scheme in relation to the potential gains under
full utilization of existing preferences.
A hypothetical complete removal of all EU duties on imports from existing GSP
beneficiaries would lead to large gains for a subset of the Latin American GSP+
countries as well as the for the standard GSP countries Thailand, Argentina and
Brazil. In contrast, all EBA regions in the model lose out in this speculative borderline
scenario – a clear-cut case of preference erosion.
In all scenarios under consideration, the aggregate welfare impacts on the EU are of
a negligible order.
151
6 Qualitative Assessment of the GSP+
An important component of the EU‘s GSP strategy towards developing countries is to
provide additional preferences under the GSP+ scheme to vulnerable non-LDCs that have
ratified and effectively implemented 27 international conventions. The conventions are
related to the core political, human and labour rights as well as sustainable development and
good governance.
They include:
the elimination of discrimination against women;
the prohibition of torture;
the right to strike;
the banning of child labour;
sustainable management of the environment,
good governance and the fight against drug production and trafficking;
the Kyoto Protocol;
the Convention on International Trade in Endangered Species; and
the UN Convention against Corruption.
The GSP+ regime has been created especially for ―vulnerable countries with special
development needs‖. As of early 2010, GSP+ covered Armenia, Azerbaijan, Bolivia,
Colombia, Costa Rica, Ecuador, El Salvador, Georgia, Guatemala, Honduras, Mongolia,
Nicaragua, Paraguay, Peru and Sri Lanka.
Before 2005 different ―GSP‖ regimes were in place. The special incentive arrangement for
the protection of labour rights was conditioned on the same list of eight core ILO conventions
that have been retained in the current GSP+ scheme. However, this arrangement had only
two beneficiaries, Sri Lanka and Moldova, with the latter no longer under the GSP
preference regime. Georgia applied for inclusion in the scheme in 2000/2001,53 but no
positive decision has yet been taken. The GSP Drugs regime was not strictly linked to the
ratification and implementation of conventions, although the Regulation provided for the
Commission's assessment of beneficiary countries' ―social development, in particular the
respect and promotion of the standards laid down in the ILO Conventions referred to in the
ILO Declaration on Fundamental Principles and Rights at Work‖ and ―environmental policy,
in particular the sustainable management of tropical forests‖.54
This part of the report provides a qualitative assessment of the GSP+ scheme, focusing on
its sustainable development dimension. The first section deals with progress in ratification
and de jure and de facto implementation of international conventions. The following section
53
See the Commission notice OJ, 27.4.2001, C 127/13, http://eur-
lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:C:2001:127:0013:0013:EN:PDF 54
See COUNCIL REGULATION (EC) No 2501/2001 of 10 December 2001 applying a scheme of
generalised tariff preferences for the period from 1 January 2002 to 31 December 2004 OJ
31.12.2001, L 346/1.
152
discusses the costs and benefits of implementation of conventions. The third section
analyses the selection criteria for GSP+.
6.1.1 Implementation & effects of international conventions
This section addresses the following questions:
What are the trends in de jure compliance and ratification of the core conventions on
sustainable development?
What is the de facto efficacy of the GSP+ provisions on labour, the environment,
human rights and good governance?
6.1.2 Lessons from the literature
Conditionality in policies towards the developing world has been the subject of particularly
hot debates and controversies. Examples of the recent reviews of conditionality related to
aid and other policies include Mold (2009) and Paczynski (2009). There is also a large body
of literature focusing specifically on trade-related conditionality, e.g. IMF (2005), World Bank
(2006), Saner and Guilherme (2007) and Hafner-Burton (2005 & 2009).
We do not attempt to conduct a full review of this literature but rather focus on selected
lessons that may be of interest in the analysis of the de jure and de facto efficacy of GSP+
provisions on labour, the environment, human rights and good governance.
One question is whether ratification of international conventions on human rights, labour
rights, etc. is related to actual progress in the relevant fields. As regards human rights and
human rights conventions, the analyses of Hathaway (2002, 2003) and Neumayer (2005)
suggest the lack of a link between the conventions‘ ratifications (or time passed since
ratifications) and progress in human rights. Clearly, all these studies face a problem of
finding reasonable proxies for respect of human rights and data deficiencies may also play
some role in determining these results.
Hafner-Burton (2005) presents a more upbeat view on trade-related human rights
conditionality. She identifies preferential trade arrangements (PTAs) that supply hard
standards tying the material benefits of integration to compliance with human rights
principles. Her empirical analysis suggests that such PTAs appear to mitigate repression
levels in the countries involved. In contrast, the commitments of the countries themselves to
human rights conventions or PTAs supplying soft human rights standards (not tied to market
benefits) are found not to produce a systematic improvement in respect for human rights.
Analysis of the effects of international labour conventions and the effects of labour clauses in
trade agreements faces similar challenges to those encountered in the human rights
literature. The problems start with measurement. Compa (2003) surveys various ways of
measuring progress in compliance with core labour standards on workers‘ freedom of
association, which is at the core of ILO Conventions 87 and 98 (both of which enter GSP+
conditionality). He observes that ratification of the relevant ILO convention cannot in any way
be taken as a proxy for a measure of respect of workers‘ freedom of association. Neumayer
and de Soysa (2005) agree with this assessment, extending it to all core ILO conventions.
153
This may be due to the fact that ratification followed by noncompliance does not lead to any
negative consequences for a state following such a policy. Compa (2003) makes it clear that
an objective, wide-ranging system of monitoring progress in labour rights and standards is
yet to be built.
Labour related provisions are found in multiple bilateral and unilateral trade arrangements,
especially those where the EU or the US is a partner. Despite their popularity, robust
analysis of the actual effects of these provisions appears scant (Bourgeois et al. 2007; Horn,
Mavroidis and Sapir, 2009). Most articles and reports take a narrative approach or are based
on case studies rather than attempting international comparisons. The measurement
problems indicated above may be one explanation for this. Another explanation might be
related to the heterogeneity of labour provisions in existing trade agreements.
Wells (2006) provides a case study of the US-Cambodia Textile Agreement that operated
between 1999 and 2005. Its unique feature was linking increased market access to
systematically and publicly monitored increased compliance with labour standards, based on
ILO assessments. The Agreement is considered very successful both in fostering a
significant improvement of compliance with core ILO labour standards, as well as in
providing a major boost to the garment industry, with a four-fold increase in exports over the
lifetime of the Agreement. Other studies provide a less optimistic assessment of other PTAs
signed by the US. Greven (2005) claims that, apart from the agreements with Cambodia and
Jordan, other US PTAs appear not to have resulted in any progress on labour rights.
The distinction between positive incentives (as in the US-Cambodia Textile Agreement) and
negative conditionality (sanctions in response to violations of labour rights) seems very
important. The latter is often found not to be effective, or even possibly prone to lead to a
worsening of the situation. With regard to child labour, Doepke and Zilibotti (2009) claim that
international pressure in the form of boycotts and trade sanctions is likely to be
counterproductive. Their explanation focuses on the specific constellation of interests of
involved parties (e.g. working children vs. adult workers in the formal labour market),
highlighting an important lesson that the effectiveness of particular measures can be
context-specific, and hence differ between countries, periods or economic sectors.
Frundt (1998) evaluates the effects of labour-related conditionality on the US GSP scheme
in Central American countries. He concludes that GSP conditions had partial success in El
Salvador. Compliance fared little better in Guatemala, although attitudes improved. In
Honduras, Costa Rica, Nicaragua, and Panama, GSP also appeared to have achieved
temporary successes. The Dominican Republic is identified as a country where the trade
requirements displayed their greatest effectiveness, resulting in substantive labour reform.
Frundt (1998) also discusses a more general issue, that of the usefulness of labour-rights
trade conditionality and in particular challenges the notion held by some experts that
conditionality may encourage an increase in informal labour and hence inhibit trade and
workers‘ benefits.
Compa and Vogt (2003) review the 20 years experience of labour rights clause in the US
GSP. The authors conclude that, on balance, the GSP clause has played an important and
positive role in stimulating actions by other international actors, as well as in directly
improving workers‘ situations, at least in some countries.
154
One study looking directly at the GSP+ experience with labour provisions is Orbie and Tortell
(2009). The authors start from analysing the patterns of ratification of core ILO conventions.
They conclude that the timing of convention ratifications suggest that at least in some cases
there appears to be a direct link between the conditionality of the GSP+ and countries'
decisions to ratify. This likely applies to Bolivia, Colombia, Venezuela, Mongolia, and El
Salvador, all of which ratified one or more of the core labour conventions during the period
2005-06. El Salvador is identified as the most obvious example of the working of such a
mechanism. The country was granted GSP+ in late 2005 with the condition of completing
ratification of the two then unratified core ILO conventions (87 & 89). El Salvador finally
ratified both conventions only in September 2006, just short of the EU deadline, when the
risk of losing GSP+ status became real.
Orbie and Tortell (2009) further analyse the GSP+ effects on implementation of labour
standards. In measuring progress in labour standards they utilise ILO committees‘ reports,
although attempting to decode their diplomatic language and at the same time constructing a
scale according to which the countries could be assessed. They do so by constructing a
‗hierarchy of condemnation‘. The hierarchy has five levels and the position of a country on
this scale is determined by the discussion of country cases by specific types of ILO
committees and the use of certain words and phrases in the committees‘ reports. The
general conclusion from this work that focuses entirely on GSP+ beneficiaries (i.e. no control
group is analysed) is that during 2005-2008 the GSP+ did not lead to an overall
improvement in implementation of labour standards in the analysed countries (as measured
by the ―levels of condemnation‖ by ILO committees).
The literature on environmental provisions in trade agreements mostly focuses on the
experience of NAFTA (OECD, 2007; Bourgeois et al., 2007). It is predominantly descriptive
and qualitative in nature. Also, it typically does not attempt to measure progress in
environmental matters – possibly because of the lack of appropriate data.
Overall, it is probably fair to say that the literature is far from offering any consensus view on
the effectiveness of human rights, labour standards, governance and environmental
provisions in trade agreements and unilateral preferences. This is largely due to the lack of
tangible indicators of progress in these areas and major differences in the depths of
commitments. Also, in the case of many recent PTAs, it is simply too early to judge the
effectiveness of such commitments. However, the existing empirical literature points to the
importance of effective monitoring of the actual implementation of these provisions, as
ratification is often not followed by implementation and to the use of positive incentives as
opposed to negative conditionality.
6.1.3 From ratification to implementation: legal analysis
The GSP Regulation states that GSP+ status can be granted to a vulnerable country that
meets the following three requirements55:
It has ratified and effectively implemented all the 27 listed conventions
55
Council Regulation (EC) No 732/2008 of 22 July 2008, Article 8.
155
It undertakes to maintain the ratification of the conventions and their implementing
legislation and measures
It accepts regular monitoring and review of its implementation record in accordance with
the implementation provisions of the conventions it has ratified
It is further required that the European Commission shall keep under review the status of
ratification and effective implementation of the 27 listed conventions by examining available
information from relevant monitoring bodies. It must also produce a summary report on the
status of ratification and available recommendations by relevant monitoring bodies. The
Commission must additionally inform the Council if any of the conventions have not been
effectively implemented.
At the level of de jure implementation, the above requirements amount to monitoring three
levels of de jure implementation:
i) the ratification of the convention itself
ii) the post-ratification transposition of the convention requirements into national
legislation
iii) the post-ratification compliance by the GSP+ beneficiary with the convention‘s
internal reporting procedures
The following analysis focuses on changes in de jure implementation on these three levels.
Table 6.1 gives the details on recent ratifications of the 27 conventions in 18 countries that
were beneficiaries of the GSP+ regime at some point since its introduction. Of these
countries, 15 enjoy GSP+ status at the time of writing this report (early 2010), one country
(Moldova) was granted autonomous trade preferences in 2008 and was removed from the
list of GSP beneficiaries, one country (Paraguay) was late with its submission for
prolongation of the GSP+ status and hence lost the status at least until mid-2010, while
Venezuela lost GSP+ preferences in mid-2009 following information that it failed to ratify one
of the 27 conventions.
In February 2010, a decision was taken to temporarily withdraw Sri Lanka from the list of
GSP+ beneficiaries following the results of an investigation that identified significant
shortcomings in implementation of three UN human rights conventions – the International
Covenant on Civil and Political Rights, the Convention against Torture and the Convention
on the Rights of the Child. Following a six-month transitory period, the decision to withdraw
preferences will enter into force in summer 2010.
156
Table 6.1: Ratifications of 27 Conventions in Present and Past GSP+ Beneficiaries
Country GSP status No. of
ratifications
till end-2004
Recent ratifications
Armenia GSP+ since
2009
23 Freedom of Association and Protection of the
Right to Organise (N° 87), 2/01/2006
Minimum Age for Admission to Employment
(N° 138), 27/01/2006
Prohibition and Immediate Action for the
Elimination of the Worst Forms of Child Labour
(N° 182), 2/01/2006
Convention on International Trade in
Endangered Species, 23/10/2008
UN Convention Against Corruption, 8/03/2007
Azerbaijan GSP+ since
2009
25 Convention on Biological Diversity, 1/04/2005
Cartagena Protocol on Biosafety, 1/04/2005
UN Convention Against Corruption, 1/11/2005
Bolivia GSP+ since
2006
24 Forced or Compulsory Labour (N° 29),
31/05/2005
Convention on the Prevention and Punishment
of the Crime of Genocide, 14/06/2005
UN Convention Against Corruption, 5/12/2005
Colombia GSP+ since
2006
24 Prohibition and Immediate Action for the
Elimination of the Worst Forms of Child Labour
(N° 182), 28/01/2005
Stockholm Convention on Persistent Organic
Pollutants, 22/10/2008
UN Convention Against Corruption, 27/10/2006
Costa Rica GSP+ since
2006
24 Stockholm Convention on Persistent Organic
Pollutants, 06/02/2007
Cartagena Protocol on Biosafety, 07/05/2007
UN Convention Against Corruption, 21/03/2007
Ecuador GSP+ since
2006
26 UN Convention Against Corruption, 15/09/2005
Georgia GSP+ since
2006
24 Stockholm Convention on Persistent Organic
Pollutants, 04/10/2006
International Convention on the Crime of
Apartheid, 21/03/2005
Cartagena Protocol on Biosafety, 4/11/2008
UN Convention Against Corruption, 4/11/2008
Guatemala GSP+ since
2006
24 UN Convention Against Corruption, 3/11/2006
International Convention on the Crime of
Apartheid, 15/05/2005
Stockholm Convention on Persistent Organic
Pollutants, 30/07/2008
Honduras GSP+ since
2006
23 International Convention on the Crime of
Apartheid, 29/04/2005
Stockholm Convention on Persistent Organic
Pollutants, 23/05/2005
Convention on Psychotropic Substances,
23/05/2005
157
UN Convention Against Corruption, 23/05/2005
Cartagena Protocol on Biosafety, 18/11/2008
Sri Lanka GSP+ since
2006
(temporarily
withdrawn in
2010)
26 Stockholm Convention on Persistent Organic
Pollutants, 22/12/2005
Moldova GSP+ till
March 2008
25 International Convention on the Crime of
Apartheid, 28/10/2005
UN Convention Against Corruption, 01/10/2007
Mongolia GSP+ since
2006
24 Forced or Compulsory Labour (N° 29),
15/03/2005
Abolition of Forced Labour (N° 105), 15/03/2005
UN Convention Against Corruption, 11/01/2006
Nicaragua GSP+ since
2006
23 Stockholm Convention on Persistent Organic
Pollutants, 1/12/2005
Convention against Torture and Other Cruel,
5/07/2005
United Nations Single Convention on Narcotic
Drugs, 5/02/2005
UN Convention Against Corruption, 15/02/2006
Panama GSP+ till
end-2008
26 UN Convention Against Corruption, 23/09/2005
Paraguay GSP+ since
2009
25 International Convention on the Crime of
Apartheid, 2/12/2005
UN Convention Against Corruption, 1/06/2005
Peru GSP+ since
2006
26 Stockholm Convention on Persistent Organic
Pollutants, 14/09/2005
El Salvador GSP+ since
2006
24 Freedom of Association and Protection of the
Right to Organise (N° 87), 6/09/2006
Right to Organise and to Collective Bargaining
(N° 98), 6/09/2006
Stockholm Convention on Persistent Organic
Pollutants, 27/05/2008
Venezuela GSP+ till
mid-2009
23 Prohibition and Immediate Action for the
Elimination of the Worst Forms of Child Labour,
(N° 182), 26/10/2005
Kyoto Protocol, 18/02/2005
Stockholm Convention on Persistent Organic
Pollutants, 19/04/2005
UN Convention Against Corruption - not ratified*
Note: * Based on information provided in the Commission Decision of 11 June 2009, OJ L 149/78. In
contrast, the website of the UN convention against corruption lists ratification by Venezuela on
02/02/2009. We were unable to explain the divergence between the sources.
Sources: COM(2008) 656 final and websites of Conventions.
At the level of ratification, the following observations can be made:
Most countries have ratified 1-4 conventions around the dates required by GSP+
conditionality. As discussed elsewhere in this section, in some instances there is no
doubt that the GSP+ has acted as the sole motivation for ratification.
158
The countries that received GSP+ benefits only in 2009 were not less advanced with
regards ratification of conventions in 2005 when the decision on GSP+ for the period
2006-2008 was taken. Paraguay and Azerbaijan ratified all 27 conventions before the
end of 2005. This poses the question as to why these countries did not apply for
GSP+ status in 2005 (or if they applied why they were not accepted). In any case this
proves that fulfilling GSP+ eligibility conditions does not automatically lead to a
successful application for the GSP+ status.
The case of Venezuela shows that non-ratification can indeed lead to the withdrawal
of preferences.
The second level of de jure implementation consists of (ii) the transposition of the protections
set out in the conventions into national legislation. This is a more complicated analysis to
undertake. Such monitoring needs to be based directly on the reports of the relevant
committees of the 27 conventions.
The third level of de jure implementation concerns assessment of whether the beneficiaries
of the GSP+ system are complying with the mandatory procedural requirements of the
conventions, including submitting reports and paying contribution fees. Any non-compliance
is indicative of the level of commitment a beneficiary makes towards implementing a
convention. However, in this report such analysis only includes the mandatory national
reports in level (iii) and not the committees‘ on-going requests for specific information from
the governments.
The most recent full review of the status of ratification and implementation of relevant
conventions is available in the EC report, Reports on the status of ratification and
recommendations by monitoring bodies concerning conventions of annex III of the Council
Regulation (EC) No 980/2005 of 27 June 2005, applying a scheme of generalised tariff
preferences (the GSP regulation) in the countries that were granted the Special Incentive
Arrangement for sustainable development and good governance (GSP+) by a Commission
Decision of 21 December 2005 COM(2008) 656 final. This report was published on 21
October 2008 and describes the state of play as of April 2008.
Below we repeat a similar exercise, but limited to a narrower group of countries. Specifically,
we focus on three GSP+ beneficiaries for whom we also carry out country case studies (see
below): Georgia, Nicaragua and Peru. A November 2009 update of the comments of the
relevant conventions‘ committee reports for the three case study countries is available in
Appendix 7.
For the purposes of this study, a convention is described as ineffectively implemented if the
relevant committee reports indicate concern that domestic legislation has omissions or
requires clarification or amendment. That is, if it has only partially transposed protections of
a given convention into national legislation. Again, for the purposes of this study a
convention is described as implemented if a committee does not have any concerns or has
not made comments to the contrary in its reports.
The main observations of the effective de jure implementation for the three analysed
countries can be summarised as follows. Conventions‘ committee comments suggest that of
159
the 27 conventions which require effective level (ii) legislative transposition into national law
in order to make them operational domestically,
Georgia has implemented 15 conventions effectively, eight not effectively and there
is no information on four conventions.
Nicaragua has implemented 16 conventions effectively, seven not effectively and
there is no information on three conventions.
For Peru, the numbers are 15, nine and three, respectively.
In all three countries, effective implementation was strongest for the conventions covering
the environment and governance and weakest for the core labour conventions.
However, this number crunching can at best only be indicative. For example, although partial
implementation cannot be considered to be effective implementation, beyond this
assumption it is not clear how partial or non-compliant the legislation is with the
requirements of the Convention or how much to censure the partially non-implementing
beneficiary. As the committee reports indicate, a full examination of the national legislation
with respect to all the protections set out in a particular convention is an undertaking that
requires an on-going dialogue and cooperation with the signatory government.
Nonetheless, it is possible to conclude that all 3 countries have not been able to implement
the following conventions effectively in terms of level (ii) transposition into national
legislation:
Freedom of Association and Protection of the Right to Organise Convention, 1948
(No. 87)
Right to Organise and Collective Bargaining Convention, 1949 (No. 98)
The Equal Remuneration Convention, 1951 (No. 100)
The International Covenant on Civil and Political Rights.
Different conclusions can also be drawn from the lack of information on a particular country‘s
implementation record. It is therefore important to include an examination of the third level of
de jure implementation, which concerns (iii) the mandatory national reporting procedures.
The reporting mechanisms of the conventions require regular reporting by the beneficiary
governments regarding the legislative and operational implementation of the convention. If
the beneficiary does not submit the required national implementation reports or financial
contributions, the committee notes this omission. Reporting requirements differ from
convention to convention and may include submitting annual or biennial implementation
reports and self-check lists, conducting pilot tests and keeping up with any financial pledges,
etc. This further complicates cross-convention committee reporting analysis.
Notwithstanding this, an overview of the three countries‘ level (iii) de jure implementation
indicates that:
Georgia: implementation is assumed for 21 conventions, partial for five and unknown for
one
Nicaragua: implementation is assumed for 19 conventions, partial for seven and
unknown for one
Peru: implementation is assumed for 22 conventions, partial for three and unknown for
two
160
Further, all three countries had difficulties complying with the monitoring requirements for the
following two conventions: the Convention on International Trade in Endangered Species of
Wild Fauna and Flora (CITES) and the Convention on Biological Biodiversity.
In summary, this analysis has contended that effective de jure implementation requires the
effective transposition of the protections set out in the conventions into national legislation.
The analysis of level (ii) and level (iii) implementation indicated that this has been most
effective for the environment and governance conventions and least effective for the core
labour requirements in all three examined countries.
Our analysis of progress over time is necessarily somewhat limited not only due to difficulties
in distilling hard evidence from existing reporting systems but also due to the short period of
examination (1.5 years). This is because the only point of reference is provided by the EC
2008 summary report. It is then perhaps not surprising that during such a short period in the
majority of cases our best assessment suggests ―no change‖ on all three levels of de jure
implementation.
Only in the case of the Convention on the Elimination of All Forms of Discrimination against
Women for all three countries do we identify progress in implementation (based on the lack
of reports to indicate otherwise). It is also clear that trends differ between countries. In our
small sample, Nicaragua is found to experience more regresses in implementation than
either Peru or Georgia, while Georgia saw the strongest progress, for five conventions.56
Clearly, such comparisons should be treated with caution.
Going beyond de jure implementation in trying to assess de facto effects proves yet more
challenging. To research the de facto operation of the conventions requires a methodology
more robust than a survey of the information available in the committee reports or
information produced even by prominent international NGOs because of the lack of
comparative data available to assess the situation relative to other countries. It is also clear
from the analysis of the convention committee reports that the beneficiary governments do
not always accept the information provided to the committees by external organisations.57
The committees must draw on only the most reliable and verified information for comment.
However, a survey of reports for the 27 conventions suggests that the information supplied
in the committee reports is not systematic or comprehensive enough to draw any
conclusions at the level of the de facto efficacy of conventions. This is primarily because the
reports rely on the governments to provide or verify information and respond to specific
queries, which necessarily focus on some areas at the expense of others. This does not
provide a strong basis for producing any firm conclusions on levels of de facto efficacy.
56
Our indicative assessment suggests for Georgia - no change in 18 conventions, progress in five,
regression in two, no information on two; for Nicaragua: no change in 20 conventions, progress in two
and regression in four, for Peru: no change in 19 conventions, progress in three, regression in two
and no information on three. See Appendix 6 for details. 57
The committee comments for Georgia on the Minimum Age Convention, 1973 (No. 138), indicate
that the government questioned the ITUC‘s use of UNICEF statistics and subsequently the ITUC was
unable to verify their validity. See Appendix 7.
161
6.1.4 Challenges of implementation: lessons from case studies
The legal analysis above illustrates the challenges of highlighting the effects of entering the
GSP+ and the ratifications of the conventions on progress in their effective de jure and de
facto implementation. It is therefore natural to try to gather evidence ―on the ground‖, that is,
in selected countries that benefit from the GSP+ scheme. This section provides findings from
three country case studies that were carried as part of the GSP evaluation. The selection of
countries followed the preferences of the EC and studies were carried for Georgia,
Nicaragua and Peru by CASE-Transcaucasus, Funides and Metis Gaia, respectively. The
key questions here are:
What is the ('de facto') efficacy of the GSP+ provisions on labour, the environment,
human rights and good governance as perceived by various local stakeholders (NGOs,
labour unions, government, researchers, etc.)?
Has the GSP+ changed anything with regard to the question above, i.e. has the fact that
effective implementation became part of GSP+ conditionality affected the relative
success of implementation and changes in the actual situation as regards labour issues,
human rights, the environment and good governance?
All three case study countries have enjoyed GSP+ status since 2006. Peru and Nicaragua
were previously part of the GSP Drugs arrangement and its predecessors. Georgia has an
Action Plan under the European Neighbourhood Policy designed to help, inter alia, its closer
trade and economic integration with the EU, in particular through gradual regulatory
alignment. This should in particular enable Georgia to progressively become ready to
negotiate, implement and sustain a deep and comprehensive PTA with the EU. Peru is one
of the three Andean Community countries that currently participate in the 'multi-party' trade
negotiations with the EU on an ambitious and comprehensive trade agreement. Nicaragua
participates in the ongoing EU-Central America negotiations on an Association Agreement,
including forming a free trade area.
The importance of trade with the EU differs significantly between Georgia (sending more
than 40 percent of its exports to the EU as of 2008, double the share from 2004), Peru
(around an 18 percent share in total exports as of 2008, down from 25 percent in 2004) and
Nicaragua (9 percent of total exports in 2008, down from 12 percent in 2004).
Some general findings related to all three countries can be summarised as follows:
GSP+ conditionality is believed to have had a very limited impact in encouraging
increased implementation and compliance with convention mandates. In the case of
Peru, the FTA with the US is believed to be more important in this respect. In
Georgia, the ENP Action Plan plays a more important role.
Knowledge of the details on the GSP+ programme appears very limited among the
general public. In particular none of the interviewed exporters using the preference
regime was aware of conditionalities attached to the scheme.
Domestic political dynamics prove to be very important in determining relative
progress in the various fields covered by the conventions. The priorities of a
government in office, the intensity of domestic political struggles and similar factors
162
appear to be decisive in affecting which areas see progress and which see no
change or regress.
Enforcement of existing legislation is much more difficult than passing legislation.
This is related to limited resources and administrative capacity, among other factors.
Availability of financial and other resources matters for the monitoring of the current
situation, quality of institutions, inter-agency cooperation, education and promotion
activities and hence the whole process of effective implementation of conventions. In
some spheres, especially with respect to environmental conventions, access to
international aid is very important.
Some of the interesting country-specific observations include the following:
Peru
Peru ratified all but one convention before 2005, i.e. before the conditionalities of
GSP+ were announced.
In the human rights sphere, pressure from international actors (NGOs, other
countries) has influenced the gradually improving compliance of conventions for the
last two decades. At the same time, the existence of an authoritarian regime and
armed conflicts has exacerbated the violations of human rights.
With regard to environmental conventions, the most significant recent improvement
was the creation of the Ministry of Environment in 2008. Its formal mandate is
stronger and it has more resources than institutions previously dealing with
analogous issues. Decentralisation of certain environmental policies with the creation
of relevant offices in the most vulnerable regions was another positive factor that may
bring benefits in the future. One factor slowing progress is related to weak or
nonexistent sanctions for offences against environmental regulations.
In the sphere of good governance and the fight with drugs, the general weakness of
the state and widespread poverty have contrasted with the huge potential economic
gains from illegal activities and constitute a major impediment to progress.
Nicaragua
Economic constraints account for many of the problems in effective implementation
of conventions. Some obstacles are also associated with other factors, such as
complacency on the side of the government or even disregard for some of the
mandates. This affects progress in civil and political rights and freedoms, labour and
environmental protection and the protection of indigenous and ethnic minorities´
rights.
Conventions tend to be implemented more successfully where ratification was
accompanied by strong support among civil society, policy makers showed stronger
commitment and interest among the donor community was higher. Such conditions
have enabled the allocation of adequate human and economic resources to
implementation, increased the willingness of policy makers to develop
complementary laws and regulations, as well as the institutional framework for
implementation, facilitated inter-agency coordination and coordination with local
authorities and communities and increased public awareness. Thus, some
163
conventions regarding the rights of children, access to education, women and
children‘s health and women‘s rights have been more successful.
A few conventions have been ratified just to fulfil the GSP+ conditionality. In these
cases there was no general interest or even awareness on the side of the general
public and hence effective implementation was negatively affected. This partly
explains the relatively limited progress in implementation of mandates provided in
environmental conventions.
Some conventions have served as important catalysts in guiding public policy,
mobilising civil society and elevating some issues´ priority within the national policy
agenda. This has probably been the case with respect to the rights of children, rights
of women, anti-discrimination policy and some aspects related to labour rights.
Furthermore, some of the conventions related to environmental and governance
issues, particularly those of a more regulatory and control nature such as CITES and
the United Nations Convention against Illicit Traffic in Narcotic Drugs and
Psychotropic Substances, have been more effective in achieving their objectives.
Their mandates were more easily integrated within existing regulatory control and
enforcement mechanisms and agencies, such as police and customs authorities.
The direct impacts of some other conventions have been absent or more difficult to
detect for several reasons. For instance, many aspects of the International
Convention on Economic, Social, and Cultural Rights have historically always been
at the centre of the policy agenda, but progress in this sphere depends on structural
factors, e.g. the prevalence of poverty.
Georgia
At the political level, European integration is a major priority for the Georgian
government. The ENP Action Plan is a relevant reference point for several policy
spheres. To a certain degree all the conditions that are in some way linked to the EU
(including the conventions related to GSP+) are perceived as part of EU
requirements on the road to the European integration of Georgia. It appears that
several stakeholders tend to identify the objective of fulfilling the GSP+ conditionality
with ―meeting EU standards‖.
Interesting developments have taken place with respect to ILO conventions. The
labour code adopted in 2007 has severely limited workers‘ rights and this has led to
strong EU criticism, although not always directly linked to GSP+ preferences. The
following excerpts from the EC Staff document can be illustrative: ―As regards labour
law and rights at work, no progress can be reported as regards unrestricted strike
rights. The 2006 labour code, which was prepared without prior consultation with
trade unions, is not in line with the ILO standards. In particular, it falls short in
addressing the obligations of the ILO Conventions on freedom of association, and on
the right to organize and collective bargaining. Furthermore, the labour code
contradicts both EU standards and the European Social Charter that the country
ratified in July 2005, on a number of fundamental issues such as the duration of
overtime work and termination of employment‖.58
58
The European Commission Staff Working Document (SEC(2008) 393).
164
However, the GSP+ context of the above issue became important at the time when a
decision was to be made on prolongation of GSP+ for the period after 2009.
According to the interviewed representative of a labour union association, ―it
practically depended on the Trade Unions‘ consent whether the country would retain
the GSP+ scheme or not‖. According to various interviewees, the consent of the
trade unions allowed the government to improve its relations with the ILO and hence
avoid a strengthening of the ILO criticism of the labour rights situation and the
possibility of retaining GSP+ status. Assuming this to be a good description of what
has actually happened, it could either be interpreted as confirming the effectiveness
of GSP+ conditionality or (more likely) as an indication of how conditionality can be
manipulated by reaching a temporary agreement that does not solve underlying
issues.
Progress has been limited in environmental policy. Some regulatory improvements
have taken place, but with no major breakthrough. Corruption negatively affects
protection of the environment.
6.1.5 Quantification of implementation effects
The question of the actual effects of the implementation of conventions could in principle be
tackled by looking at measures of progress in the spheres covered by the conventions. The
last decade or so has witnessed the emergence of and increased interest in indicators trying
to capture various aspects of quality of governance, institutions, human rights, etc.
Kaufmann, Kraay and Mastruzzi (2009) provide references to many relevant sources. Such
an attempt appears particularly worthwhile given the scant existing literature based on
econometric evidence. At the same time the lack of serious quantitative studies suggests
that there might be problems measuring the variables of interest, identifying and classifying
the incentives, etc.
In the analysis below we try to reveal the effects of GSP+ conditionality on progress in areas
covered by some existing indicators. We apply a ‗difference in differences‘ approach, where
GSP+ is taken as a treatment applied to the group of countries benefiting from this
preference regime. The change in a relevant indicator (governance, gender, human rights,
etc.) over the period in which GSP+ was in place is contrasted with a change over the same
period in relevant control groups, which in our exercise comprise EBA countries and
countries benefiting from the basic GSP. The underlying logic is that GSP+ may be
mobilising the countries, first to ratify the conventions, but then – possibly more importantly –
to become more serious about de facto implementation of the conventions‘ spirit, compared
to a situation where there are no such forces at play. In other words, in this exercise we do
not consider the fact of a convention‘s ratification as decisive for whether it is likely to be
effective in changing the actual situation in labour market conditions, human rights, etc., but
rather a combination of ratification and potentially more effective implementation due to
GSP+ conditionality. This remark is important in that many non-GSP+ countries have
typically also ratified most of the 27 conventions, in many instances quite a long time ago. It
is also consistent with findings from the literature as discussed above.
There are two obvious limitations to our work. The first is the short history of GSP+, while
significant changes in the fields related to GSP+ conventions typically take longer to
materialise. Secondly, as mentioned above, various social and environmental clauses are
165
also part of other trade arrangements. In other words (part of) the control group is also
exposed to potentially similar sets of incentives.
First we look at gender equity issues. We note that two of the 27 conventions explicitly focus
on the issues of gender equity:
Convention on the Elimination of All Forms of Discrimination Against Women
Convention concerning Equal Remuneration for Men and Women Workers for Work
of Equal Value (No 100).
Some other conventions also explicitly mention the equal rights of men and women in their
texts, e.g.:
The International Covenant on Economic, Social and Cultural Rights
Convention concerning Discrimination in Respect of Employment and Occupation
(No 111).
Therefore we are trying to find a good measure of gender equality that would be available for
a number of countries and at least for the period of the existence of the GSP+ regime. The
candidate of our choice is the Gender Equality Index (GEI) published by the Social Watch
(Social Watch, 2009). It is primarily constructed based on underlying indicators for which
measurement error can be assumed to be relatively low, such as employment data by
gender ( percentage of women in technical positions, percentage of women in management
and government positions, percentage of women in parliaments, percentage of women in
ministerial posts), economic activity data (income gap, activity rate gap) and educational
data (literacy rate gap, primary school enrolment rate gap, secondary school enrolment rate
gap, tertiary education enrolment rate gap). The indicator has been available since 2004.
We run a cross country OLS regression (89 observations), where the dependent variable is
the change of GEI between 2004 and 2008. Independent variables include dummies for
GSP+ and EBA trade preference regimes and several controls, such as the initial (i.e. 2004)
level of GEI, the 2004 level of GDP per capita in PPP terms and the improvement in the
Government Effectiveness index of Kaufmann, Kraay and Mastruzzi (2009) over the period
2004-2009 (to capture the possible effects of other concurrent reforms).59
The results (see Appendix 8 for details) suggest a significant positive effect of the GSP+
regime on improvement in GEI relative to both EBA and GSP groups. GSP+ countries
improved their GEI score by around six percentage points more than the GSP group, while
the difference vis-à-vis EBA countries is close to 12 percentage points (the scale of the GEI
is from 0 to 100 percent).60 Beyond this, an initial GEI level turns out to be highly significant,
suggesting a convergence in levels.61
59
Several other formulations of the econometric model were also tested without affecting the results
in a significant way. 60
To illustrate the scale of differences, six percentage points is the difference in the GEI 2008 score
of the Netherlands or Iceland on the one side and Portugal, Romania or Argentina on the other. 12
percentage points is the difference between the Netherlands or Iceland on the one side and Greece,
Bolivia or Belarus on the other. 61
The details of estimations are available upon request.
166
An analogous exercise (with relevant modifications of controls and estimation techniques) is
repeated for other indicators:
Voice and Accountability (Kaufmann, Kraay and Mastruzzi (2009)) - trying to
measure the extent to which a country‘s citizens are able to participate in selecting
their government, as well as freedom of expression, freedom of association, and a
free media
The Civil Liberties Index (Freedom House (2009)) - covering issues such as freedom
of speech, assembly, demonstration, religion, equal opportunity, excessive
governmental intervention
Control of Corruption (Kaufmann, Kraay and Mastruzzi (2009)) - trying to measure
the extent to which public power is exercised for private gain, including petty and
grand forms of corruption, as well as ―capture‖ of the state by elites and private
interests.
In all these cases no evidence was found of any significant differences in performance
between the trade preference regimes (among GSP+, EBA, GSP).
We also identified other indicators that should capture aspects relevant for the assessment
of conventions‘ effectiveness, such as data on child labour (Statistical Information and
Monitoring Programme on Child Labour of the ILO, the World Bank's Surveys on the
economic activity of children and UNICEF's Multiple Indicator Cluster Surveys) and the
World Bank‘s Country Policy and Institutional Assessment (CPIA), and specifically the CPIA
policy and institutions for environmental sustainability rating. However, data coverage was
insufficient to form the basis of any meaningful econometric analysis trying to capture the
GSP+ effects.
We conclude that available data are largely consistent with the hypothesis that the GSP+
scheme and its conditionality has not yet resulted in significant changes in the situation ―on
the ground‖ in beneficiary countries. One potential exception could be in the sphere of
gender equality, where our results are not inconsistent with the hypothesis of positive GSP+
effects. Incidentally, it is also this sphere which our independently run legal analysis (for the
three case study countries) indicated as the only one where some progress in de jure
implementation can be observed, judging solely from the conventions‘ monitoring reports.62
We note, however, that lack of a coherent theoretical model and deficiencies of the available
indicators suggest a cautious interpretation of the econometric exercise and specifically do
not allow for making any inferences as to causality.
6.1.6 Section Summary
The above analysis can be summarised by the following points:
62
The reports of the Committee for the Convention on the Elimination of All Forms of Discrimination
against Women had no comments for the three case study countries, while previously some concerns
were indicated.
167
It is probably too early to assess whether GSP+ has a chance of becoming an effective
mechanism in promoting sustainable development and good governance. The processes
in these spheres take longer than the scheme‘s timeframe to date.
Given the uniqueness of the GSP+ scheme, lessons from the analysis of other relevant
trade arrangements can only be indicative. One general conclusion from the literature is
that the design of the GSP+ is relatively robust in that there are chances of effectiveness
at least in some countries or in some spheres, while the risk of negative effects is very
limited.
GSP+ appears to be effective in promoting the ratifications of the 27 conventions. De
jure implementation beyond ratification already faces several constraints. We do not find
evidence of any significant positive effects of GSP+ here. There is no indication of
potential adverse GSP+ effects on de jure implementation.
De facto effects are yet more difficult to identify, measure and compare across countries
and time. We find some evidence suggesting positive effects in the sphere of gender
equality. In other spheres, such as corruption, civil liberties, etc., we find no effects. We
do not identify any negative effects on de facto implementation.
6.2 Costs and benefits of fostering sustainable development and good
governance – GSP+ beneficiaries‘ perspective
The underlying idea of the GSP+ scheme is that additional preferences are granted to
countries which have taken on board commitments related to the respect of basic human
and labour rights, environment protection and the principles of good governance. One
aspect of this is that effective implementation of conventions in these spheres may be costly
for developing countries and GSP+ benefits can then be expected to compensate for and
outweigh such costs. This section discusses the character of relevant costs and benefits and
then provides insights into their balance in selected spheres.
The first important note concerns differences between the costs and benefits of effective
implementation of conventions and the costs and benefits of achieving the ultimate
objectives that a given convention is meant to promote. The distinction matters primarily due
to the costs of monitoring, reporting and other actions related to the de jure implementation.
This may be of particular importance in some environmental conventions, where certain
requirements on national reporting or action plan development are very costly to follow for
developing countries. Theoretically there may be a trade-off between fulfilling the costly de
jure requirements of conventions and channelling the resources to policies with a direct
effect on an underlying problem.
Secondly, the analysis is not possible without making a clear distinction between the short-
and long-term. The underlying assumption of the GSP+ scheme (which can of course be
challenged) is that in the long term, pursuing policies in line with sustainable development
and good governance standards will benefit developing countries. The short run matters
because in these spheres progress and tangible benefits typically take a long time to
materialise, while there may be non-negligible costs of policies to be borne immediately.
Given limited resources, countries need to well prioritise their development needs in order to
focus attention on spheres with the most beneficial benefit/cost ratio. This is of course easy
168
to say but extremely difficult to implement in practice given uncertainty on policy outcomes,
discount rates, heterogeneity of stakeholders, etc. In any case, it cannot be taken for
granted that effective implementation of 27 conventions currently forming the GSP+
conditionality is entirely consistent with the development priorities of beneficiary countries.
Self-selection of countries into the scheme cannot be taken as a reassurance given the
above mentioned uncertainties.
Thirdly, a very important distinction needs to be made between stakeholders benefiting from,
and those bearing the costs of, policies aimed at conventions‘ implementations. This is also
related to beneficiaries and victims of practices that conventions try to change. The
illustrative example is provided by the child labour issue. The major costs here are borne by
children who are not given a chance of completing an education that could help them avoid
poverty in the future. The benefits accrue to businesses relying on child labour and the poor
families of working children whose current incomes are increased. As noted by Udry (2003),
this creates a self-reinforcing relation between poverty and child labour: ―because their
parents are poor, children must work and not attend school, and then grow up poor.‖
One illustrative case suggesting that GSP+ may potentially constitute an administrative
burden on countries is related to the failure of Panama (GSP+ beneficiary during 2006-2008)
to re-apply on time to be covered in the 2009-2011 GSP round. There may have been
different reasons for this, and in particular it will be interesting to see if the country applies for
the preferences from July 2010. Still, one possible explanation may be an administrative
failure, where domestic institutions have failed to prepare required documentation on time.
The discussion below is organised according to areas such as human rights, labour rights,
environmental issues and good governance.
Hathaway (2003) notes that from a strictly rationalist point of view, ratifications of human
rights conventions are a puzzle. This is because ratification allows the international
community to intervene (to a degree) in the relationship between the state and its citizens. In
return, a country receives only a set of promises from other countries that they will refrain
from harming their own citizens. The author then discusses various notions of the costs of
commitments to human rights conventions and proposes her own theory. She postulates
that the expected compliance costs are a function of two factors: (1) the difficulty of attaining
conventions‘ standards measured by the degree to which a country‘s practice diverges from
the requirements of a given treaty and (2) the likelihood of realisation of the costs described
under (1), i.e. the likelihood that the state will actually get serious about implementing a
given convention. To the extent that monitoring aspects of GSP+ are believed to be effective
(even if only over a longer horizon) a country‘s decision to apply for GSP+ status may be
considered as an indication of a stronger commitment to human rights promotion.
All GSP+ beneficiaries ratified almost all of the conventions related to human rights a long
time ago and only a few ratifications have taken place since 2004. The costs of effective
implementation of the conventions are mainly related to the social and economic rights
dimension, where adequate provision of education and health services is in practice very
difficult in a number of developing countries. This is typically due to lack of resources (e.g.
due to the existence of a large unofficial economy or inefficient tax collection) and/or misuse
of available resources due to poor management or outright corruption or theft (Hillman and
169
Jenker, 2004). While these costs are high, the benefits are widely believed to outweigh the
costs by a large margin. The findings on the costs and benefits of eradication of child labour
(see below) are applicable here.
The benefits of civil, political and other freedoms that are promoted by the human rights
conventions are hardly measurable in economic terms. The general economic costs of
ensuring respect for human rights should not be particularly large if the process is driven
from within the country. The economic and political costs may be high for particular groups
of the ruling elite – e.g. in the case of oppressive corrupt regimes. We are unaware of
studies that seek to provide a quantitative description of the costs and benefits, especially in
an international comparison. We believe the overall balance to be very positive and hence
maintaining the human rights dimension of the GSP+ should not be particularly controversial
from the perspective of a cost / benefits analysis.
The costs of implementation appear to be a generally important factor in countries' decisions
to adopt international labour conventions. With regard to ILO conventions, Boockmann
(2001) finds strong evidence that the economic costs of ratification are a major factor in
ratification decisions by developing countries. In fact, variables used as proxies of economic
costs (the size of a country and whether it has ratified a predecessor convention, if there was
any) are the only significant determinants. Importantly, this work takes into account a large
number of ILO conventions (around 180) and not just the eight core conventions that are
part of the GSP+ conditionality. The intuitive view that ILO conventions are quite
heterogeneous, in particular with respect to costs of their implementation and that this
should also matter for ratification decisions, is confirmed by Boockmann et al (2009).
The eight core conventions address issues of freedom of association and collective
bargaining, discrimination, forced labour, and child labour. In one of the countries covered by
case studies (Georgia) the recently adopted labour code has been widely perceived as not
guaranteeing the minimum standards in these spheres. At the same time it has allowed
Georgia to see its ranking improve considerably in the World Bank‘s Doing Business report.
In the 2010 report Georgia ranked 9th in the world by the ease of employing workers, a
combined index accounting for factors such as difficulty of hiring and firing, rigidity of
employment, costs of firing, etc. Obviously, the trade-off between preserving basic workers‘
rights and a good Doing Business ranking is only partial, but for a relatively poor country with
weak administration, neglecting some workers‘ rights may be much easier to do than
devising a much more complex scheme with sufficient protection of workers‘ rights and yet
very friendly to employers. To the extent that the Doing Business ranking may be among the
factors considered in investment decisions, better compliance with core ILO conventions
could be associated with costs of foregone investment, jobs etc. Quantification of such
effects would be very difficult to carry out.
Nicaragua, another case study country, has a labour code that is much more worker friendly.
There, the costs of complying with ILO conventions in practice can be identified with the
costs of effective implementation of the labour code. The budgetary resources devoted to
this would probably need to increase substantially to improve compliance. To illustrate the
problem we note that each of the nine departments in the country typically has just one or
two labour inspectors, who may also have insufficient allocation of fuel for their cars used in
inspections. Limited progress in compliance is not surprising in such an environment. Also in
170
Peru, effective implementation of labour rights would require a substantial increase in
administrative resources devoted to this task. To date there is a perception that GSP+ has
not added any new costly obligations to the businesses operating in the formal sector as the
conventions were ratified long ago and external pressure for effective implementation came
earlier and was related to Peru‘s trade agreement with the US.
In both Peru and Nicaragua, the informal sector is characterised by a particularly difficult
situation with respect to labour rights. Reduction in the size of the informal sector should in
principle allow for both improvement in respect for labour rights and provide additional tax
revenues. Yet, it is apparently very difficult to design not excessively costly policies
effectively reducing the size of the informal sector while avoiding negative side-effects.
Child labour is among the areas where recent years have seen increased interest in the cost
benefit analysis. Such research has mostly been carried out under Understanding Children's
Work, a joint ILO-UNICEF-World Bank project running since 2000. ILO-IPEC (2003), as well
as country case studies, has generally found the benefits of achieving the ultimate objective,
i.e. eradication of child labour, to outweigh the costs by a very high margin. Kassouf et al
(2005) is one of a series of studies carried out under the ILO project. The paper analyses the
case of Brazil, which is likely to be to some extent comparable to some of the Latin American
GSP+ beneficiaries. The analysis covers three types of costs: the cost of providing
education to all children in lieu of work, the cost of programme interventions to alter attitudes
and practices and the opportunity cost of eliminating child labour, i.e. the value of their
labour. Among the benefits, the study considers the economic gains from a more educated
population and the economic advantages resulting from a healthier population, since both
more widespread education and the elimination of hazardous or unsuitable work have
prospective health benefits. Potentially large non-economic (cultural or social) benefits are
not included due to problems in their quantification. The study directly addresses the issue of
different time profiles of the costs and benefits, with benefits generally materialising much
later than costs. The results suggest a very significant net gain from the elimination of child
labour with benefits outweighing costs by a multiple of between five and six.
Other analyses focus on the specific benefits from limiting child labour. Kucera and Sarna
(2004) try to identify the effects of child labour and education on exports in a gravity model
setup. They conclude that child labour is bad and education good for exports, including for
unskilled labour-intensive manufacturing exports.
A cost benefit analysis of environmental conventions is quite difficult for several reasons.
One is related to the key importance of external effects. Environmental processes are in no
way related to political borders and hence environmental benefits for a given country or
region to a large extent depend on the policies implemented by other countries. Conventions
can be thought of as coordination mechanisms trying to ensure that, globally, welfare
improving cooperative equilibrium is selected. This makes valuation of individual benefits
nigh on impossible, especially when uncertainties as to the short- and long-term effects of
environmental policies are taken into account. Focusing on the GSP+ countries, it is also
evident that they have ratified several of the environmental conventions only fairly recently.
Progress with implementation is hence somewhat limited, giving little information on actual
costs. Nevertheless, it is clear that effective implementation of several environmental
conventions would be quite costly for the GSP+ countries.
171
The role of foreign aid is very important in financing the implementation efforts. This also
changes the cost-benefit balance as seen from the perspective of a given country. It could
be argued that the GSP+ conventions have motivated donor resources that would otherwise
not have entered the countries. Given that many of the projects required under the
conventions (reporting, data collection, action plans, etc.) are costly, they would not have
been implemented without external support.
For example, the majority of the activities of the Ministry of Environment and Natural
Resources of Nicaragua (MARENA) are financed externally. It is estimated that 80 percent
of staff are paid through donor project financing and it is through these projects that most of
the actual implementation of the conventions is carried out. The projects have financed
many of the reporting requirements under the conventions, as well as the elaboration of
action plans, such as the National Biodiversity Strategy and Action Plan and the formulation
of the Biodiversity Law. There are likely some benefits from such schemes affecting
governance quality in general.
There are differences in implementation costs between environmental conventions. In the
case of Nicaragua the results of the case study suggest that the Montreal Protocol, CITES
and the Basel Convention on the Transboundary Movement of Hazardous Substances may
have been easier and less expensive to implement. The first reason for this is that their core
regulatory mandates were integrated within existing regulatory and control mechanisms. For
example, Customs and the Ministry of Agriculture already maintained a database of
chemicals requiring authorisation for imports. In the case of the Basil Convention, Customs
already maintained strict control and regulation of the movement of hazardous substances
based on the 1993 law.
The Nicaraguan case study also provides some information on the costs that have been so
far borne by the private sector, although data limitations are quite severe. The general
conclusion is that the impact on the private sector has been limited so far since the
implementation of regulatory controls is in its infancy, particularly in the case of recently
ratified conventions. Furthermore, there is a high degree of non-compliance and
enforcement mechanisms are nonexistent in many cases. In other cases, the regulatory
mechanisms were already in place before ratification of conventions, such as controls on
logging or on the transport of hazardous substances, meaning that these costs were already
borne by the private sector prior to implementation of the conventions. One exception is
perhaps the Montreal Protocol and the requirement to limit the import of ozone depleting
substances. In this case, there apparently was an impact on businesses as evidenced by the
fact that firms operating in the refrigeration and air conditioning sector were against quotas
on imports of ozone depleting substances.
Most of the economic literature suggests potentially significant gains from good governance,
including in particular the reduction of corruption, although this view is not uncontested
(Abed and Gupta, 2002). The economic costs of corruption eradication may strongly depend
on the policy approach. For example, a corruption-reducing higher wage for all government
employees would not be sustainable from a fiscal perspective in most developing countries.
Strengthening control mechanisms may also be costly, although some relatively simple
solutions should also be available. The political costs may be substantial due to the actions
172
of vested interest groups. The information from the case studies suggests that the costs
actually borne have been small, largely owing to very limited implementation. One non-
economic benefit highlighted in the Nicaragua study is an increase in public awareness of
the problem and explicit prohibition of certain activities, even if not yet effectively
implemented.
The cost-benefit analysis of the conventions dealing with drugs is also hindered by
insufficient data. Similarly as in the case of corruption, changing the balance of economic
incentives for people involved in the illegal drugs industry is difficult and potentially costly
given widespread poverty, limited public resources and weak state institutions. Potential
benefits may also be substantial and extend to social, economic, health and other spheres.
The above discussion clearly suffers from lack of data that would enable meaningful cost-
benefit analysis. In large part this is for objective reasons, as both costs and benefits extend
to several different spheres. Valuation of things such as civil liberties, improved
environmental outcomes, lower corruption etc. has been attempted by some studies but
there are limitations on what can be achieved (see e.g. Eyckmans et al, 2004). There is also
a large degree of uncertainty as policy actions typically bear fruit only in the long-term
perspective.
At the same time there appears to be increased interest on the side of certain international
institutions in research into the costs and benefits of conventions. For example, in 2008 the
Secretariat of the Basel Convention initiated work on the Framework for Cost-Benefit
Analysis for use at the national (or regional) level (see also the Basel Convention, 2008).
The job description for an intern who was to work on these issues provide a realistic
description of the current situation: ―To date, little research has been completed that
analyses the socioeconomic cost of the generation, transboundary movement and
environmentally unsound management of hazardous waste, including products at the end of
their useful life, particularly e-wastes. Most of the data available, primarily attained through
the Basel Convention‘s national reporting mechanism, is largely inconsistent. Equally, data
collected through other international organisations and bodies such as the OECD is neither
consistent nor comparable between countries or from the same country.63‖
6.2.1 Section Summary
The costs of effective implementation of the conventions are mainly related to the social
and economic rights dimension. Adequate provision of education and health services is
in practice very difficult in a number of developing countries.
Maintaining the human rights dimension of the GSP+ should not be particularly
controversial from the perspective of a cost / benefits analysis.
The costs of implementation appear to be a generally important factor in countries'
decisions to adopt international labour conventions.
A cost benefit analysis of environmental conventions is quite difficult for several reasons.
One is related to the key importance of external effects. Environmental processes are in
63
We have tried to gather evidence for a cost-benefit analysis of implementing conventions by
contacting the secretariats of the relevant conventions. Unfortunately, we have not received any
information from these sources.
173
no way related to political borders and hence environmental benefits for a given country
or region to a large extent depend on the policies implemented by other countries.
The role of foreign aid is very important in financing the implementation efforts. This also
changes the cost-benefit balance as seen from the perspective of a given country.
The cost-benefit analysis of the conventions dealing with drugs and corruption is
hindered by insufficient data.
6.3 Selection criteria for GSP+
In order to be eligible for the GSP+ program, a country must first be classified as ‗vulnerable‘
by satisfying the following three criteria:
(a) a country cannot be classified by the World Bank as a high-income country during
three consecutive years;
(b) the five largest sections of its GSP-covered imports to the EU must account for over
75 percent in value of its total GSP-covered imports;
(c) its GSP-covered imports to the EU must represent less than 1 percent in value of
total GSP-covered imports to the EU.64
Then to qualify for the additional preferences under the GSP+ program, a ‗vulnerable‘
country must have ratified and effectively implemented 27 international conventions. In
addition to ratification of these conventions, the country is required to provide
comprehensive information concerning the legislation and other measures to implement
them. It must commit itself to accepting regular monitoring and reviewing of its
implementation record. Finally, the country must make a formal request to qualify for the
GSP+.
In this section we analyse and evaluate the adequacy and appropriateness of the above
criteria and whether these remain the most relevant ones in furtherance of the scheme's
objectives.
The starting point for this analysis is an attempt to identify the objectives that the selection
criteria should foster and, if possible, identification of relevant benchmarks.
The EC communication from 2004 on the functioning of the Community's GSP for the ten-
year period from 2006 to 2015 inter alia indicates the following:
The GSP must be stable, predictable, objective and simple
The GSP must be targeted on the countries that are most in need
The GSP should assist (...) countries to attain a level of competitiveness which could
make them self-supporting economically and full partners in international trade
(The GSP) must encourage regional cooperation between developing countries
64
166 middle or low-income countries or territories satisfy these criteria. 49 of them are LDCs and
hence benefit from EBA preferences (apart from Myanmar, which is temporarily withdrawn from the
EU GSP preferences). This leaves 117 countries or territories eligible for GSP+, as of 2008.
174
The goal of promoting sustainable development must be given greater prominence.65
On the targeting of countries, the Communication explains that:
The GSP should focus on those countries most in need, such as the LDCs and the
most vulnerable developing countries (small economies, land-locked countries, small
island states and low income countries) in order to help them play a greater role in
international trade. These are the countries that the GSP must focus on as its first
priority.
(...) In addition, a further obvious group of countries are land-locked and low income
countries, countries that are unable to take advantage of economies of scale or are
beset by logistical problems and those whose economies are not at all diversified.
This goes in particular for the textiles and clothing industry.
In the analysis below we assess whether the current criteria satisfy these objectives.
6.3.1 Vulnerability criteria
The first question one should ask about the vulnerability criteria is whether they are at all
needed. This is relevant because, according to the current definition, few developing
countries are not classified as ‗vulnerable‘. The large majority of countries are highly unlikely
to lose the vulnerable status in the next few years. The condition therefore mostly matters for
a rather narrow group of around 25 countries that are larger and/or close to the EU in
geographic terms and hence have large and/or relatively diversified exports to the EU.
These countries are either currently classified as not vulnerable or not unlikely to lose the
status, according to the current definition (e.g. Argentina, Pakistan, Ukraine, Belarus,
Vietnam, Bangladesh, Morocco, Egypt, etc.). Some of these countries benefit from other
preferences in trade with the EU (or are negotiating such preferences). Bangladesh benefits
from the EBA regime. Yet some other countries (China, Brazil, India, Malaysia, Indonesia,
Thailand, Vietnam) have one or more sections graduated from the GSP. In practice this
implies that the vulnerability condition is a binding impediment for potential access to GSP+
only in a very few cases.
Still, this may matter also for small and undiversified countries in that it potentially affects
whether they will have preferential access to the EU market relative to larger, currently
excluded, countries. An underlying question here is related to the optimal degree of
discrimination in GSP+ preferences. There is also a legal perspective to this as not all
solutions here may end up being WTO-compliant. There is a degree of trade-off between
two broad objectives of the GSP+: supporting countries that are most in need and promoting
sustainable development, good governance, etc. One could think of two scenarios – either
the sustainable development objective of the GSP+ is given prominence and then dropping
the vulnerability criterion altogether might be seriously considered – or alternatively a
65
Excerpts from the Communication from the Commission to the Council, the European Parliament
and the European Economic and Social Committee - Developing countries, international trade and
sustainable development: the function of the Community's generalised system of preferences (GSP)
for the ten-year period from 2006 to 2015, (COM/2004/0461 final).
175
stronger focus on selecting only the most vulnerable countries could be envisaged and then
the criterion could be made more selective.
The current solution can be seen as a compromise between the two options with an
underlying logic that larger countries and countries geographically closer to the EU are more
likely to follow the sustainable development path without the support of GSP+. This is a
reasonable assumption given the easier access to other EU programmes for neighbouring
countries (e.g. Egypt, Morocco, Tunisia, Belarus, Ukraine) and the generally better
administrative and other resources of larger countries. In any case these issues could
perhaps benefit from a dialogue including the developing countries.
The EU vulnerability criteria are economic-based, defined as they are by the country‘s
income and trade. However, this definition may seem somewhat narrow. The UN economic
vulnerability criteria that classify a country as a least-developed country include much more
broad indicators, such as the size of population, remoteness, the share of agriculture,
fisheries and forestry in total GDP, the instability of agriculture production, merchandise
exports concentration, instability of exports of goods and services. Of course, given the
different nature of selectivity in the UN LDCs list and EU extra preferences given only to a
set of developing countries, the definition and perceived level of economic vulnerability do
not have to be the same.
As an example from another GSP system, the 2008 reform of the Norwegian GSP eligibility
conditions extended the preferences originally granted only to the LDCs to also include
―other low income countries‖ based on the OECD DAC List of ODA Recipients.66 This is a
relatively short list of countries (12 in 2009-2010) that includes the two poorest non-LDC
countries that are not vulnerable using the GSP+ definition: Vietnam and Pakistan. One
possible modification of the vulnerability criteria could introduce an alternative of either jointly
meeting the current three criteria or being classified as ―other low income countries‖ based
on the OECD DAC List of ODA Recipients. This would ensure that countries just a bit richer
than LDCs are not excluded from the possibility of applying for GSP+ even if they are large
countries (and hence have high or relatively diversified exports to the EU). One potential
problem with this modification is that the new threshold income per capita level would still be
quite arbitrary. Secondly, all current GSP+ beneficiaries have higher income levels, being
classified as lower or upper middle income countries and territories in the OECD DAC list.
This may suggest that taking up GSP+ commitment is in any case very difficult for poor
countries. If this was indeed the case then the idea of covering ―other low income counties‖
might rather be considered as an option for extending EBA eligibility.
In the remaining part of this section we assume that some definition of vulnerability that
would lead to the selection of a similar number of countries and with similar characteristics is
likely to stay in place in the future. We discuss possible changes to the definitions and
calculation methods and then evaluate the extent to which current criteria are successful in
selecting small, poor and landlocked countries.
Definitions and calculation methods
66
See http://www.regjeringen.no/upload/UD/Vedlegg/Handelspolitikk/gspchanges.pdf
176
The first observation concerns the definition of the criterion itself and specifically the way in
which the figures on covered imports should be calculated.
Article 8 of the Council Regulation (EC) No 732/2008 states with respect to the import share
calculations that, ―The data to be used are: (...) those available on [certain date], as an
annual average over three consecutive years". We note that this formulation is not self-
explanatory and that there are different alternative ways of calculating the averages, leading
to different results. Among other things there are nontrivial questions on handling the years
with no imports from particular countries. Such cases are rather rare but in our database
extending from 2002 to 2008 there are several countries or territories with import data
available only for less than all seven years. Therefore, we believe there is a strong case for
making publicly available a technical explanation on the method actually used in
calculations.
With regard to the predictability and stability of the rule, we note that data on whether a
country is classified as vulnerable or not becomes available very late relative to the period
for which it matters. For example the vulnerability calculations relevant for applications for
GSP+ for the period starting on 1 July 2010 are based on data as of 1 September 2009
(which were published on-line at a yet later date) – around half a year before the deadline for
applications (30 April 2010).
Moreover, it is well known that trade data are subject to revisions and first estimates may
differ from revised data by quite a bit. This may prove decisive for countries being classified
as ‗vulnerable‘ or not. Argentina provides an actual example here. In the EC list of
vulnerable countries calculated based on the statistics available by 1 September 2007,
Argentina was not classified as a vulnerable country given that the share of its five largest
sectors in EU imports stood at 71.8 percent (well below the 75 percent threshold). When we
calculate the same share for the same period (2004-2006) but using the most recent data
available to us, the share of the five largest sectors turns out to be 75.6 percent, that is,
above the threshold. In other words, it is simply errors in preliminary EU import statistics that
have excluded the possibility of Argentina applying for GSP+ preferences for the period from
2008 to mid-2010.
Given the two issues above (the moment of publication and data revisions) we think there is
a strong stability - and a predictability-based case - for the introduction of a transition period
before the country loses its ‗vulnerable‘ status. Such a change would be very easy to
implement and in fact could make the GSP rules more consistent between GSP+ and EBA.
We suggest that for a country that was classified as vulnerable in the past but the most
recent calculations show it is losing its ‗vulnerable‘ status, the effects of this should be
delayed by a certain period, possibly three years, exactly as is the case with the EBA
preference following the removal of a country form the LDCs list (Article 11 (4) of the GSP
regulation67). An additional condition should be that (1) during the three-year transitional
period the country does not become vulnerable again and (2) the original calculations on
67
―When a country is excluded by the UN from the list of the least-developed countries, it shall be
withdrawn from the list of the beneficiaries of this arrangement. The removal of a country from the
arrangement and the establishment of a transitional period of at least three years shall be decided by
the Commission, in accordance with the procedure referred to in Article 27(4).‖
177
which the removal from the list of vulnerable countries is based are not affected by data
revisions during the three-year period.
Using the example of Argentina, the system we propose would result in the following:
As of 1 January 2009 Argentina would enter the three-year transitional period to non-
vulnerable status (implying it would still be eligible to apply for GSP+ for 2009-2011
assuming it met other criteria).
It would be subsequently removed from the transitional status and put back in the
vulnerable category, given the available revised data.
The success of the definition(s) in selecting countries in need
In this section we analyse a set of statistics that may shed light on the question of the
adequacy of the GSP+ vulnerability criteria. We study the characteristics of countries as
provided by the GSP+ definition of ‗vulnerability‘ and other potentially relevant dimensions of
economic development that are usually used to define the vulnerability of countries in
various contexts.
The vulnerability criteria are designed to select small economies, landlocked or small island
states, low income countries with highly concentrated exports and insufficient integration into
the international trading system. We therefore picked a set of indicators to study the
relationship between the guiding principles of the GSP+ system and the vulnerability criteria.
In particular, we look at the level of Gross National Income (in PPP terms) and population to
measure the size of the economy, as well as the Gross National Income per capita (in PPP
terms) to measure the income levels of countries. Further, for each country we include its
size (area in km2) and an indicator as to whether they are landlocked or not. In addition, we
add a couple of measures of the country‘s integration into the international trading system,
that is, the country‘s openness level (share of merchandise trade in GDP) and how long it
has been a member of the WTO. We also add a measure of trade concentration at a lower
level of aggregation, using the Herfindahl index, for exports to the EU calculated at the 10-
digit HS level. This index constitutes a quantitative measure of export concentration (or the
inverse of diversification). The less diversified the composition of exports, the higher the
value of this index. This measure is included to assess how well the concentration indices
calculated at a quite aggregated level of sections of GSP-covered imports to the EU
correspond with the measures of concentration of exports at the product level.
Finally, we study volatility in the terms of trade (ToT). Countries prone to terms of trade
shocks are viewed as more vulnerable, as volatility in terms of trade contributes to the
volatility of macroeconomic outcomes. Macroeconomic volatility due to changes in terms of
trade leads to slower growth and more uneven distribution of income. ToT volatility also
discourages investment in human and physical capital, hampering future economic growth.
In our statistical analysis we consider the export concentration ratio as defined in the GSP
vulnerability criteria, or as the arithmetic average of the share of the five largest sections in
exports to the EU over two periods 2003-2005 and 2006-2008 (columns 1 and 2). In
addition, we also analyse the same share in total world exports of countries over the 2007-
2008 period (column 3). We look at the Kendall rank correlation coefficients between these
average export concentration ratios and selected economic and geographical characteristics
of countries as discussed above. Our sample includes 178 countries.
178
Table 6.2: Kendall Rank Correlation Coefficients between Export Concentration Ratios
and Selected Economic and Geographical Characteristics
Kendall tau Average share of
the 5 highest
sections in exports
to the EU over
2003-2005
Average share of
the 5 highest
sections in
exports to the EU
over 2006-2008
Average share of
the 5 highest
sections in total
(world) exports
over 2007-2008
Column 1 2 3
average
population
2003-2005
('000s)
-0.3424***
average
population
2006-2008
('000s)
-0.3043*** -0.1104
area (km2) -0.2383***
area (km2) -0.2002*** -0.0079
GDP level (PPP)
Average
2003-2005
-0.3927***
GDP level (PPP)
Average
2006-2008
-0.3389*** -0.1495**
GNI per capita
(PPP) Average
2003-2005
-0.1114
GNI per capita
(PPP) average
2006-2008
-0.1053 -0.0859
Herfindahl 2008 0.4256*** 0.4598***
Average
openness 2003-
2005
0.2255***
Average
openness
(2006-2007)
0.1407* 0.4321***
WTO
membership up
to 2008
-0.0247 -0.1315
Relative TOT
Standard
Deviation
2003-2007
0.1407* 0.4321***
Notes: *** denotes significance at the 1% level, ** denotes significance at 5% level.
179
Data sources: Population, Openness (Merchandise Trade as a percentage share of GDP), Net Barter
Terms of Trade – World Development Indicators data base.
Herfindahl index of concentration of exports to the EU based on data provided by the EC at the 10
digit level.
Area – CEPII data base.
The Kendall rank correlation coefficients as presented in Table 6.2 measure the strength of
dependence between the two variables. The Kendall‘s rank correlation coefficient ―tau b‖
ranges between -1 (one ranking is the reverse of the other) to 1 (the two rankings are the
same).
The statistics in Table 6.2 columns 2 and 3 indicate that the vulnerability criterion is
correlated with most of the selected variables and in the majority of cases the directions of
correlations are as expected. For example, EU export concentration is correlated to a
moderate degree with our measures of the size (population, area and GDP level) of the
country. The rank correlation coefficient is negative, which indicates that smaller countries
tend to be associated with higher levels of concentration. However, when we look at the rank
correlation of one measure of development level, gross national income per capita, the
correlation coefficient with export concentration does not turn out not to be statistically
significant (statistically different from zero). Hence, in this sense, poor countries are not
associated with higher export concentration levels on the EU markets. This is a not a
preferred property.
When looking at the levels of openness, the correlation with export concentration is
statistically significant but has a positive sign. This indicates that more open economies in
our sample tend to have less diversified exports to the EU market. This may be due to a
positive link between the size of the country and its trade openness. The rank correlation
between the time span of the WTO membership and the EU export concentration ratio is not
statistically different from zero. Our sample includes 66 countries that are not members of
the WTO and all of them but two (Belarus and Russia) are classified as ‗vulnerable‘ based
on current criteria. Further, we look at the relationship between ToT volatility and EU export
concentration and find it to be positive and statistically significant when studying the 2006-
2008 data.
As expected, the Herfindahl index of trade diversification and the share of the five largest
sections tend to move in the same direction and the correlation coefficient is rather high
(0.42). Hence, it seems that the vulnerability criteria do correspond quite well to what we
believe would be a more accurate measure of trade diversification.
Finally, looking at the number of landlocked countries in our sample, we note that 29 out of
178 countries are landlocked and all of them are classified as ‗vulnerable‘.
Overall, we find that the vulnerability criteria are consistent with the selection of smaller,
landlocked countries, prone to the ToT shocks. In addition the present criteria seem to be
quite well related to the measure of trade diversification at the product level. However, it
seems that the EU export concentration ratios tend to be associated with countries more
integrated into the world trading system as measured by their trade openness levels. And
most importantly (and problematically) our simple analysis does detect any statistically
180
significant association of export concentration with the level of GNI per capita.
This lack of rank correlation between development level and export concentration is not
necessarily particularly troublesome given that almost all poor countries are still classified as
‗vulnerable‘. Still, there are some outliers, those very poor countries that do not meet the
vulnerability criteria. For LDCs this is not an issue as these countries (at present one such
case - Bangladesh) are eligible for a more advantageous set of preferences given by EBA. A
potential problem could affect countries that are just above the LDC level but are still very
poor and not classified as ‗vulnerable‘. By construction this is most likely to affect large
countries.
One modification of the vulnerability criteria avoiding such exclusion problems has been
already mentioned above. It would grant ‗vulnerable‘ status to countries that either jointly
meet the current three criteria or are classified as ―other low income countries‖ based on the
OECD DAC List of ODA Recipients. The income threshold level could of course also be set
differently. This would ensure that countries just a bit richer than LDCs are not excluded from
the possibility of applying for GSP+ even if they are large countries (and hence have high or
relatively diversified exports to the EU). We do not recommend that such a change is
introduced in the near future, but suggest that it may be worth considering and discussing
further.
One might argue that the concentration of exports to the EU does not fully represent the
extent of export concentration and therefore the vulnerability of a given country and that one
should consider the concentration of total exports instead. The reasoning goes that
vulnerable countries in need of support to increase their export volumes and export
diversification should be those that face difficulties in diversifying and expanding their
exports to all markets, not only to the EU market. When we look at the export concentration
of the five largest sections in total exports over 2007-2008 as compared to the EU export
concentration of the five largest sections over 2006-2008, we note as expected that for the
majority of countries (155 out of 178), the levels of world trade concentration are lower. The
two measures are correlated, but not highly (the rank correlation coefficient equals 0.45).
Further, we look at the Kendall rank correlation between total exports over 2007-2008 with
the same economic and geographic characteristics (column 3 of Table 6.2). We note that this
measure is not correlated with the country‘s size as measured by its population or area. We
also observe that the correlation with GDP level is lower than in the case of exports
concentration on the EU market. The total export concentration correlation with ToT shocks
is significantly higher. The remaining correlation coefficients are similar as in the case of
export concentration on the EU market.
Overall, we conclude that on the basis of this simple analysis the total export concentration
does not seem to have any better properties than EU export concentration except for a
better association with TOT shocks, which are not mentioned as one of the objectives of the
vulnerability criteria. Therefore, on this account there are no arguments in favour of using
export concentration on world markets in the vulnerability criteria. We also note that it is not
clear to what extent better access to the EU market through GSP preferences would improve
the diversification of total trade, especially in cases where exports are destination-specific.
181
The time-series properties of vulnerability criteria
In our view, one objective vulnerability criteria should meet is that countries move to non-
vulnerable status when this is the result of a true improvement in the competitive position of
a given economy. Countries should not normally lose their vulnerable status due to
temporary shocks. Hence, it would be ceteris paribus preferred that the vulnerability
definition has a property where it happens only very rarely that a country moves out of the
‗vulnerable‘ status and then quickly back again into vulnerability.
One mechanism that would improve the properties of the vulnerability criteria in this respect
is the introduction of the transition period, as proposed above. Below we put some
alternative definitions of vulnerability to the test, which should shed light on the preferred
properties.
Before proceeding with the discussion we note the limits of this exercise, as stability per se
is not the goal of the GSP system. The goal is to allow those countries which have
significantly improved the diversification or volume of their exports to be excluded from the
preferences to allow more vulnerable countries to enjoy a relatively higher margin of
preference on the EU market to improve their chances of increasing trade volumes and
diversification of exports.
The test compares various possible definitions of vulnerability based on the share of certain
number of covered imports in total covered imports. We are then interested in identifying
cases where such a share first decreases by a certain margin and then quickly goes up
again. It is this type of behaviour that can lead to a country losing ‗vulnerable‘ status when a
share of covered imports in total imports first falls below a certain threshold (e.g. 75 percent)
and then quickly regains it.
We use the dataset obtained from the EC covering the 2002-2008 period as a base for the
simulations. For each of the vulnerability definitions under consideration we calculate the
number of pairs of subsequent years when the import share of n largest sectors (average
calculated over k years) decreased by more than i percentage points in the first year, then
declining by more than j percentage points in the next year. We refer to such cases as
reversals. The definitions exhibiting less reversals are more desirable.
We compare the following six definitions:
1. The currently used definition: simple average over three-year period of percentage
shares of five largest sections of GSP-covered imports in total imports
2. The simple average over a two-year period of the percentage shares of the five
largest sections of GSP-covered imports in total imports
3. The simple average over a four-year period of the percentage shares of the five
largest sections of GSP-covered imports in total imports
4. The simple average over a three-year period of the percentage shares of the three
largest sections of GSP-covered imports in total imports
5. The simple average over a three-year period of the percentage shares of the seven
largest sections of GSP-covered imports in total imports
6. A modification of the calculation of the average share (keeping the currently used
three years and five sectors) in which the average share is calculated as follows:
182
(i) select the five largest sectors in year one & calculate imports in these sectors
(ii) select the five largest sectors in year two & calculate imports in these sectors
(iii) select the five largest sectors in year three & calculate imports in these
sectors
(iv) add three values of imports from steps (i)-(iii) above and divide over total
covered imports in these three years.
We carry out the simulations for different values of i and j ranging from 0.5 percent to
3 percent. The results are mostly intuitive and can be summarised as follows.
Decreasing the number of countries or number of sectors for which the average is calculated
(methods two & four) results in visibly higher number of reversals. In contrast, increasing the
number of years or sectors (methods three & five) decreases the number of reversals,
although the difference compared to method one is typically not large. For example, for i and
j set at 2 percent there is one reversal under methods three & five, and five reversals under
method one. Finally, method six is characterised by significantly more reversals than method
one.
Overall, we conclude that the introduction of the transitory period for phasing out
vulnerability status as discussed above would have much stronger effects on improving the
stability and predictability of (eligibility for) the GSP+ scheme than manipulation of the
methods of calculating import shares.
6.3.2 International conventions
Although revising and implementing legislation is resource and time consuming, another
area where some modifications could be proposed concerns the selection of conventions.
The 2009-2011 GSP+ scheme requires the ratification and effective implementation and
monitoring of all 27 conventions. Some of these seem to duplicate each other, for example:
Convention concerning the Abolition of Forced Labour (No 105)
Convention concerning Forced or Compulsory Labour (No 29)
Or
United Nations Single Convention on Narcotic Drugs (1961)
United Nations Convention on Psychotropic Substances (1971)
Or
Convention concerning Freedom of Association and Protection of the Right to
Organise (No 87)
Convention concerning the Application of the Principles of the Right to Organise and
to Bargain Collectively (No 98)
Or
International Convention on the Suppression and Punishment of the Crime of
Apartheid
International Convention on the Elimination of All Forms of Racial Discrimination
Or
183
Convention on the Rights of the Child
Convention concerning Minimum Age for Admission to Employment (No 138)
Convention concerning the Prohibition and Immediate Action for the Elimination of
the Worst Form of Child Labour (No. 182).
A full textual analysis of the overlapping conventions would indicate whether there were
gaps in legislation that more recent conventions address. For example, Article 1 of the 1930
Convention on Forced Labour (29) states [emphasis added]:
1. Each Member of the International Labour Organisation which ratifies this Convention
undertakes to suppress the use of forced or compulsory labour in all its forms within the
shortest possible period.
2. With a view to this complete suppression, recourse to forced or compulsory labour may be
had, during the transitional period, for public purposes only and as an exceptional
measure, subject to the conditions and guarantees hereinafter provided.
3. At the expiration of a period of five years after the coming into force of this Convention,
and when the Governing Body of the International Labour Office prepares the report
provided for in Article 31 below, the said Governing Body shall consider the possibility of
the suppression of forced or compulsory labour in all its forms without a further transitional
period and the desirability of placing this question on the agenda of the Conference.
The later 1957 Convention 105 concerning the Abolition of Forced or Compulsory Labour is
far more explicit about the protections offered by the convention because it no longer only
considers the possibility of the suppression of forced labour in all its forms:
Article 1
Each Member of the International Labour Organisation which ratifies this Convention
undertakes to suppress and not to make use of any form of forced or compulsory labour--
(a) as a means of political coercion or education or as a punishment for holding or
expressing political views or views ideologically opposed to the established political, social or
economic system;
(b) as a method of mobilising and using labour for purposes of economic development;
(c) as a means of labour discipline;
(d) as a punishment for having participated in strikes;
(e) as a means of racial, social, national or religious discrimination.
The spirit of the two conventions is the same, so it could be argued that the overlap of these
conventions is not a problem if they serve to underpin the same ultimate policy objectives.
However, if, as our analysis suggests, the monitoring systems of the implementation of the
legislation are vital to ensure the effective and non-discriminatory application of the GSP+
system, then the obligation to fulfil the complete and necessary reporting procedures for two
similar but slightly different conventions is onerous and there could be a stronger case for a
discussion on the reduction of the number of conventions alongside the strengthening of
reporting and monitoring mechanisms.
184
A more streamlined GSP+ would, for example, incorporate a shorter list of 22 conventions
as criteria for eligibility, no longer including:
1. Convention concerning Minimum Age for Admission to Employment (No 138);
2. Convention concerning Forced or Compulsory Labour (No. 29);
3. Convention concerning Freedom of Association and Protecting of the Right to
Organize (No. 87);
4. International Convention on the Suppression and Punishment of the Crime of
Apartheid;
5. United Nations Single Convention on Narcotic Drugs (1961).
The reduction would be accompanied by strengthened mandatory reporting and monitoring
of the de facto and de jure implementation of the conventions on a transparent and
procedurally uniform basis. Such a reporting system could enable more efficient and
transparent decision-making with regard to the suspension or re-application of the privileges
set out in the GSP+ system.
On the other hand, and notwithstanding the good governance issue, there are some
conventions that have not been included in the current list of 27 conventions which the UN
considers a ‗core‘ human rights convention and which might be more relevant to promoting
sustainable development. A prime example of this is the International Convention on the
Protection of the Rights of All Migrant Workers and Members of Their Families. The
overview of the committee reports (Appendix 7) for the case study countries has highlighted
that migrant workers are particularly vulnerable to exclusion from the protections set out in
many of the conventions currently included in the GSP+ list. However, the Migration
Convention does not in fact bring in any new human rights for migrant workers but rather
reiterates the rights included in other GSP+ conventions. Therefore, to a degree the above
argument on duplication of commitments applies. The UN, in advocating the ratification of
the Convention to all countries, stresses the importance of drawing the attention of the
international community to the particularly difficult situation of many migrant workers and
members of their families. It also notes that in some countries, legislation implementing other
human rights conventions tends to use terminology covering citizens and/or residents, hence
possibly excluding many migrants. Until now the Convention has been ratified by a minority
of developing countries, mostly from Latin America and Africa. Interestingly, 10 out of 15
current GSP+ beneficiaries have already ratified the Migration Convention.
Conventions combating terrorism may also be considered of significant relevance to
sustainable development. However, among at least 11 UN anti-terrorism conventions there
is not a single comprehensive one that could form a part of the eligibility criteria, which
makes a selection complex.68
68
The list includes: Convention Against the Taking of Hostages (12/79); Convention for the
Suppression of the Financing of Terrorism (12/99); Convention for the Suppression of Terrorist
Bombings (12/97); Convention for the Suppression of Unlawful Acts Against the Safety of Civil
Aviation (Montreal Convention) (9/71); Convention for the Suppression of Unlawful Seizure of Aircraft
(Hague Convention) (12/70); Convention on the Marking of Plastic Explosives for the Purpose of
Identification (3/91); Convention on Offenses and Certain Other Acts Committed on Board Aircraft
(Tokyo Convention) (9/63); Convention on the Physical Protection of Nuclear Material (10/79);
185
Whether a decision is made either to streamline or supplement the list of conventions that
constitute the eligibility criteria for GSP+, it can be expected that such a list will necessarily
evolve over time due to changing circumstances and that legislative revisions will be an on-
going feature of such a mechanism.
Finally, one could also consider a very different approach to sustainable development and
good governance objectives. The eligibility of countries for the benefits of the US GSP
scheme is based on the list of criteria excluding access to GSP and the discretion of the US
President to grant GSP status to all countries not excluded by these criteria (see Grossman
and Sykes, 2005 for details). In particular, GSP status is in this sense conditioned on taking
steps ‗to afford internationally recognised workers rights‘ and cooperation in combating
terrorism (and not supporting terrorism).
The US GSP Subcommittee is responsible for recommending to the President the actions to
take on petitions that seek changes in the programme‘s country coverage. Any interested
party, including government agencies, firms or individuals may petition the Subcommittee to
request modifications in the list of countries eligible for GSP treatment or ask that the
subcommittee remove a country from the GSP. If an inquiry is to be made, the USITC
conducts an investigation parallel to the GSP Subcommittee‘s consideration of petitions.
While such wide interpretive discretion may not be in the interests of legal certainty or
transparency, the ability of interested parties to petition their claims directly to the
Subcommittee offers greater access to decision making processes. It also makes for a more
flexible process of monitoring and investigating de jure and de facto implementation than
relying on the diverse committee reporting systems of the 27 conventions. It could turn out to
be cost-effective in the long term. We recognise different historical processes that have led
to the current shapes of the EU and US GSP systems and do not recommend changing the
logic of GSP+ conditionality by switching to a system more akin to the US one. However, we
stress the importance of analysing the experiences from other systems, and in particular of
their effectiveness in affecting sustainable development, human, labour rights, etc.
Summing up, we see a clear-cut case for neither reducing the number of conventions nor
introducing new ones. There are arguments for both strategies and more experience with
the current scheme might be needed before a decision on modifications is taken. In any
case, the changes should be gradual and involve transition periods so as not to impede the
stability and predictability of the scheme. The discussion on potential modifications should
take into account the experiences of other existing GSP schemes, such as the US one.
6.3.3 Section Summary
The analysis of the adequacy and appropriateness of the GSP+ selection criteria suggested:
Convention on the Prevention and Punishment of Crimes Against Internationally Protected Persons
(12/73); Protocol for the Suppression of Unlawful Acts Against the Safety of Maritime Navigation
(3/88); Protocol for the Suppression of Unlawful Acts of Violence at Airports Serving International Civil
Aviation (2/88).
186
The EU‘s vulnerability criteria are economic-based, defined by a country‘s income
and trade, which may be too narrow.
There is a strong case for making publicly available a technical explanation on the
method actually used in calculations to determine a country‘s vulnerability.
There is a strong case that the introduction of a transition period before the country
loses its ‗vulnerable‘ status would improve stability and predictability of the system.
This reform could make the GSP rules more consistent between GSP+ and EBA.
There is no clear cut case for reforming the list of GSP+ convention although
o There are several overlapping conventions which may cause unnecessary
reporting and monitoring burdens.
o There are some conventions that may be relevant to promoting sustainable
development which have been excluded from the GSP scheme, such as
those protecting migrants workers and combating terrorism.
187
6.4 Section 6: Conclusions
The EU‘s GSP strategy towards developing countries involves offering additional
preferences under the GSP+ scheme to vulnerable non-LDCs that have ratified and
effectively implemented 27 international conventions relating to core political, human and
labour rights, sustainable development and good governance. This section offered a
qualitative assessment of the GSP+ scheme, focusing on its sustainable development
dimension.
The section examined the trends in the compliance and ratification of the core conventions
on sustainable development to understand more about the efficacy of the GSP+ provisions
on labour, the environment, human rights and good governance.
The overview of the existing relevant research indicates that there is still little consensus on
the effectiveness of human rights, labour standards, governance and environmental
provisions in trade agreements and unilateral preferences. In part this is due to the short
time frame for the scheme to make an impact.
The section made the following observations:
Most countries ratified conventions around the dates required by GSP+
conditionality. In some instances GSP+ obligations acted as the sole motivation for
ratification. The case of Venezuela shows that non-ratification can indeed lead to the
withdrawal of preferences.
Of the three case study countries, Georgia, Nicaragua and Peru
o Georgia implemented 15 conventions effectively, eight not effectively and
there is no information on four conventions.
o Nicaragua implemented 16 conventions effectively, seven not effectively and
there is no information on three conventions.
o For Peru, the numbers were 15, nine and three, respectively.
In all three countries, effective implementation was strongest for the conventions
covering the environment and governance and weakest for the core labour
conventions.
All three countries have been unable to transpose the following conventions
effectively into national legislation:
Freedom of Association and Protection of the Right to Organise Convention,
1948 (No. 87)
Right to Organise and Collective Bargaining Convention, 1949 (No. 98)
The Equal Remuneration Convention, 1951 (No. 100)
The International Covenant on Civil and Political Rights
All three countries had difficulties complying with the monitoring requirements for the
following two conventions: the Convention on International Trade in Endangered Species
of Wild Fauna and Flora (CITES) and the Convention on Biological Biodiversity.
188
Some field research findings related to all three countries:
GSP+ conditionality is believed to have had a very limited impact in encouraging
increased implementation and compliance with convention mandates. One potential
exception could be in the sphere of gender equality.
Knowledge of the details on the GSP+ programme appears very limited among the
general public.
Domestic political dynamics prove to be very important in determining relative progress
in the various fields covered by the conventions.
Enforcement of existing legislation is much more difficult than passing legislation. This is
related to limited resources and administrative capacity, among other factors.
Availability of financial and other resources matters for monitoring.
With regard to the analysis of the GSP+ convention selection criteria, Section six found no
unequivocal argument for either reducing the number of conventions or introducing new
ones. Any changes should be gradual and involve transition periods to promote stability and
predictability.
189
7 Conclusions and policy recommendations
7.1 What do we learn from the analysis undertaken? Much of the work in this report is based on data which heretofore has not been used for the
analysis of GSP preferences. In particular, we have used detailed 10-digit data on trade and
tariffs where for any given product, country, and year, the data distinguishes between the
regime of entry into the EU. Hence, we do not simply know whether product ―x‖ is eligible for
preferential access to the EU from country ―y‖ together with the appropriate tariff; we also
know how much trade actually entered (or strictly speaking – applied to enter) under that
given regime, and how much trade for the same product, country and year combination may
have entered via a different regime. Hence, we have extremely precise information on
preferential trade between the EU and its partner countries.
On the basis of the analysis undertaken there is clear positive evidence with regard to the
effectiveness of the EU‘s GSP scheme. By this we mean:
1. GSP, GSP+ and EBA respectively offer a markedly greater degree of preferential
access, and hence that the EU is offering improved preferential access to those
countries with a greater developmental need.
2. The econometric evidence also suggests that, in aggregate, preferences do impact
positively on trade as well as on investment, though through different channels. The
impact on aggregate trade is of the order of between 10 percent-30 percent, with
possibly an even bigger impact on investment (though because of underlying data
constraints we caution about a literal interpretation of the numbers here). On a more
disaggregated level, we see sectorally, some evidence of a positive impact on trade –
though by no means does this apply to all sectors, and to each of the preference
schemes; and at the product level also evidence of a positive impact on trade (though
once again not unambiguously so).
3. The CGE modelling also provides support for the positive impact of the preferences on
trade – though once again not for all countries / country groupings; and we also see that
the GSP preferential regimes serve to increase welfare for many developing countries;
4. We also provide evidence that LDC exporters do benefit from the preference margins,
and that the rent is not simply appropriated by the importers;
5. We show that utilisation rates are related to the height of tariffs and to the extent of
preference margins, and that even where preference margins are low, there is utilisation
which suggests that the threshold effects which are often cited in the literature may not
be as strong as previously thought.
6. From the GSP+ analysis we see that there is some evidence that countries do make an
effort to ratify the conventions which are necessary in order for them to be able to obtain
GSP+ status.
190
These are positive and important results. However, there are a number of important caveats
to the preceding which also emerge quite clearly from the work undertaken, which need to
be borne in mind when considering the policy implications arising from this study:
1. The preference margins which the scheme offers are in most sectors low, and there do
not appear to be many significant tariff peaks in these sectors. There are only a few
sectors with significant preference margins – largely TDC Sections I – IV (live animals
and animal products, vegetable products, animal or vegetable fats and oils, and
prepared foodstuffs), XIa (textiles), XIb (clothing) and XII (footwear, headgear,
umbrellas…). The preference margins are typically low because the underlying MFN
tariffs are low. This inevitably means the scope for offering preferential access via tariff
reductions is constrained, and is a structural feature arising from the EU‘s general low
level of MFN tariffs.
2. The structure of many developing countries exports is such that a large number of them
obtain duty free or very low duty access to the EU even without utilising the preferences
offered by the GSP regime. Take Afghanistan for example. Even though it is an EBA
country nearly 93 percent of its exports to the EU are exports in products where the MFN
duty is zero. This again means that the scope for the EU to enable developing countries
to increase their trade by offering them preferential access is inevitably limited.
3. If we add to this the fact that for most developing countries the majority of their exports
do not go to the EU but to third countries, then once again it can be seen that for
structural reasons the extent to which the EU via its preferential scheme can impact on
these countries total exports is constrained.
4. The evidence on the extent to which preference margins are associated with indicators
of development are extremely mixed, and no clear picture emerges which would suggest
that the preferences are particularly well targeted to those countries which are most in
need / vulnerable.
5. There is no evidence that the GSP schemes have led to any export diversification and a
move into new export products on the part of the beneficiary countries.
6. While there is some evidence that the GSP+ scheme may have a positive impact on the
ratification of given conventions, the evidence that there is actual active implementation
of the relevant conventions (especially with regard to labour standards?) is much
weaker. The case studies appear to suggest that countries may ratify in order to meet
the minimum requirement but then do much less to implement those conventions.
It is important to note that a good part of all these caveats is structural in the sense that it is
the inevitable consequence of the mix between the level of the EU‘s MFN tariffs, together
with the structure of LDC trade. Although we have extremely detailed country specific data,
many factors will be at work at the country level. Our analysis is primarily focussed on
drawing aggregate conclusions from looking across a very wide range of countries, where it
is not possible to take into account these individual country issues. Hence it is quite possible
that the GSP regime has been an important factor for given countries in their development.
The point is however, that in aggregate, there is no strong evidence that this is the case.
191
7.2 Policy options Nevertheless the preceding conclusions raise important questions regarding what might be
the EU‘s policy options on the reasonable supposition that it does want to actively use its
trade policy to help LDCs. In terms of the policy options, we can broadly distinguish between
two categories – i) options that try to amend / improve on the existing schemes and ii)
options that consider entirely new and different policies. These are considered separately
below:
7.2.1 Amending/improving existing schemes
a. Improved product coverage: The first issue to consider is the extent to which is may be
possible for the EU to widen the number of products for which it offers preferential
access. Clearly this can only apply to GSP and GSP+ regimes, as the EBA regime
already offers duty free access for almost all goods. For these regimes it is clearly
possible to increase the level of preferential access offered to the relevant countries.
However, two issues arise. If you offer improved access to the GSP and GSP+ countries,
then inevitably this is likely to impact to some degree on some of the EBA countries who
will now see an erosion of their degree of preferential access. This emerges clearly from
the analysis using the Revealed Export Competitiveness Index (RECPI), and also from
the CGE analysis. There is little doubt that improving the level of preferential access for
the GSP countries would have a negative impact on some of the EBA countries. The
second issue, is that for many GSP countries even if they were offered EBA style
preferences given the structure of their trade this would not necessarily make a
substantial difference to them, because much of what they export is already under
MFN=0, or low MFN tariffs.
Note that these two points above are mutually consistent. For a given GSP country a
change in preferences may not have a significant impact on its trade flows, but even a
small change in its trade flows, could have a significant impact on a given EBA country.
Overall then, extending preferences is likely to impact negatively on some EBA
countries, and only help some GSP/GSP+ countries and possibly by not very much.
b. Reduce remaining tariffs on existing preferential products. By and large similar
considerations to those detailed above apply here. Clearly there is potential for the EU to
further reduce the existing preferential tariffs being offered to the GSP and GSP+
countries. However, as discussed earlier, the extent to which this is possible is limited
because of the relatively low level of preferential margins which in turn is driven by the
EU‘s relatively low level of MFN tariffs. Secondly, any improvement in the preferential
access offered to the GSP and GSP+ countries is likely to impact negatively on at least
some of the EBA countries.
c. Increasing utilisation rates: The first point to make here is that overall the evidence in
this report suggests that generally utilisation rates are fairly high – though clearly there
192
are differences across countries and sectors. We also show that utilisation may not be
subject as much to the sort of threshold effects which are often cited in the literature.
Specifically, we show that 50 percent of flows that are eligible for preferences and use
those preferences do so when the margins are below 6 percent, and 25 percent when
the margins are below 2.7 percent. However, while utilisation is typically high the lower is
the margin the lower is the degree of utilisation, and thus trying to maximise utilisation
would enable more LDC trade to take advantage of the preferences on offer. This leads
to a consideration of the underlying reasons driving non-utilisation. Once such reason is
clearly linked to be the administrative costs associated with utilisation. We have no direct
evidence in this report with regard to this, but any action that can be taken to reduce
those costs either in the beneficiary countries or in the EU is likely to increase take up.
d. Rules of Origin: More specifically, an oft-cited reason for non-utilisation is the presence
of rules of origin which may constrain the take up of preferences. There is now a fairly
well established literature on this which shows that rules of origin can be constraining
with regard to developing country exports (both with the EU and with other countries),
and that a relaxation of those rules could be beneficial for exports and development.
Rules of origin were also mentioned as a possible constraint in the case studies
undertaken for this study, and the work on determinants of utilisation also suggested that
rules of origin may be significant. The EU is currently in the process of reviewing its
policy towards rules of origin with regard to developing countries, and we suggest that
there are certain features which would help make those new rules of origin more
development friendly (or at least less subject to the criticism that they may not be
development friendly).
i. First, we suggest that moving more generally to a value-added rule (as has been
suggested by the Commission‘s draft proposal) is a positive step, although of
course the determination of the minimum required value added level then
becomes important. A value-added rule allows for greater flexibility and
negotiability.
ii. It is important to allow for as much cumulation as possible for developing
countries, in order to encourage them to source their inputs from the most
efficient suppliers. In an increasingly globalised world this is important in order for
LDCs to become genuinely competitive as opposed to relying on (declining)
preference margins in a few sectors and products. Specifically this suggest that a
positive step would be to allow for diagonal, if not full, cumulation between all
GSP beneficiaries. Also given that the EU has a number of different rules of
origin regimes in place with different partner countries, we would suggest that the
EU allow either for the application of an MFN principle, or a preferential partner
principle in the application of those rules of origin.69
69
For a fuller discussion of this see Gasiorek et. al, ―Relaxing Rules of Origin or Can those PECS be
flexed?‖, in Baldwin, R and Low, P, ed. Multilateralising Regionalism, Cambridge University Press,
(2009).
193
e. Role of the Graduation Thresholds: Currently graduation is triggered when a country
becomes competitive in one or more product groups, as defined by the TDC sections of
the HS customs code. Preferential access is withdrawn for exports of a section of the
custom code for any country for whom exports of the product group exceed 15 percent of
total EU imports of the same product group under the GSP over the past three
consecutive years.70 The graduation clause is a way of ensuring that if a particular
country becomes competitive in a given product resulting in it becoming a ―significant‖
GSP (greater than 15 percent of EU covered imports) exporter, the preferences for that
particular section are withdrawn. A key aim of the graduation clause is to maintain the
significance of preferences for those countries most in need. It is worth noting that
currently there are relatively few countries and sectors that are subject to graduation.
What is most noticeable is the graduation of China from 14 of the 21 TDC sections.71
Clearly the graduation of a country-sector combination is likely to have a positive impact
on those countries competing in those same products. A detailed analysis would require
a country by country consideration which was outside the scope of this study. However,
one way of considering the importance of this issue for those countries most in need is to
refer back to the RECPI analysis and to consider which countries exert the most
competitive pressure on the EBA. There we saw that the greatest competitive pressure,
on the EBA countries, on average, and in descending order is exerted by China, India,
Tunisia, Morocco, Mauritius, Indonesia, Egypt, Russia, Vietnam and Pakistan. So the
question then arises – how much would changing the graduation criteria impact on the
preferential access of these countries to the EU?
Consider Table 7.1 below. Here we consider which additional TDC country-sector
combination would be excluded as a result of sequentially reducing the graduation
threshold from 15 percent to 12.5 percent, 10 percent and 7.5 percent respectively. The
first column of the table shows that if the threshold were reduced to 12.5 percent, in
addition to the existing graduations, India (Section XIV), and Thailand (Section XVII)
would also become graduated. The table shows that the graduation threshold would
need to be decreased substantially to at least 10 percent or beyond for it to make much
of a difference to the number of sector-country combinations that are graduated.
70
For textiles and clothing, the threshold for withdrawal of basic GSP preferences is 12.5per cent of
the EU‘s total imports of textiles and garments under the GSP 71
Currently Brazil is graduated in TDC Sections IV and IX, China in Sections Vi, VII, VIII, XI – XVIII
and XX, India in Section XIa, Indonesia and Malaysia in Section III, Thailand in Section XIV, and
Vietnam in Section XII.
194
Table 7.1: Impact of changing the graduation threshold:
Graduation Threshold
12.5% 10% 7.5%
Argentina I,III
Brazil III
China I,IV II
Indonesia IX
India XIV VI, VIII XVII
Russia IX VI
Saudi Arabia V,VII
Thailand XVII IV
Ukraine III
Even if the threshold were reduced to 7.5 percent, this would only introduce a further 18
sector-country combinations under graduation. Of course it may well be the case that
these combinations may be particularly important for certain EBA countries and would
help them to increase their exports to the EU significantly. It is also the case that in
reducing the threshold to 7.5 percent four of the countries identified by the RECPI
analysis become graduated in at least one sector.
In summary, then we suggest that changing the graduation thresholds is likely to have
some positive impact on EBA exports (but at the expense of the GSP countries who
graduate), but that in aggregate this would appear to be a fairly crude or blunt way of
helping those countries most in need. However, this argument does need to be
somewhat qualified, as ideally this would require a detailed country by country
examination. It is also worth noting that for any given country, graduation tends to
introduce distortions with respect the relative export prices. Such distortions can lead to
a misallocation of resources.
f. Duration of GSP. The issue here is whether increasing the duration over which the GSP
is set to three years has a positive impact on LDC export flows. Given how recently this
change was introduced, this is not an issue which this study could empirically assess.
However, it seems likely that making sure the system is not subject to yearly change and
fluctuations would improve the potential attractiveness of the preferences on offer
g. Policy with regard to GSP+ In order to improve stability- and predictability of the
vulnerability criteria we recommend the introduction of a three-year transitional period
before a country loses its vulnerable status. This would make the rules consistent with
those applying to EBA eligibility.72
72
It is worth noting that Article 8 of the Council Regulation (EC) No 732/2008 states with respect to the import share calculations that ―The data to be used are: (...) those available on [certain date], as an annual average over three consecutive years". We note that this formulation is not self-explanatory and that there are different alternative ways of calculating the averages leading to different results. Among other things there are nontrivial questions on handling the years with no imports from particular countries. Such cases are rather rare but in our database extending from 2002 till 2008 there are several countries or territories with import data available only for less than all 7 years. Therefore we suggest making publicly available a technical explanation on the method actually used in calculations
195
Our analysis indicates that the current vulnerability criteria are broadly consistent with
selection of smaller, landlocked countries, prone to the terms of trade shocks and with
limited export diversification as measured at the product level. However, the criteria are
not strongly linked to income per capita levels. This is not particularly problematic given
that almost all of the poorest countries are classified as vulnerable. However,
modification of the criteria ensuring that countries below certain income per capita level
are considered vulnerable irrespective of their exports to the EU could be considered.
One such possible modification of the vulnerability criteria could introduce an alternative
of either jointly meeting the current three criteria or being classified as ―other low income
countries‖ based on the OECD DAC List of ODA Recipients. This would ensure that
counties are not excluded from the possibility of applying for GSP+ even if they are large
countries (and hence have high or relatively diversified exports to the EU). One potential
problem with this modification is that the new threshold income per capita level would still
be quite arbitrary. Second, all current GSP+ beneficiaries have higher income levels,
being classified as lower or upper middle income countries and territories in the OECD
DAC list. This may suggest that taking up GSP+ commitment is in any case very difficult
for poor countries. If this was indeed the case then the idea of covering ―other low
income counties‖ might rather be considered as an option for extending the EBA
eligibility.
Another area where some modifications could be proposed concerns the selection of
conventions on which the GSP+ preferences are conditioned. However, we see no clear-
cut case for either reducing the number of the conventions to avoid duplication of their
mandates (e.g. the ILO Convention concerning the Abolition of Forced Labour and the
ILO Convention concerning Forced or Compulsory Labour) or for introducing new ones.
There are arguments in favour of both strategies and more experience with the current
scheme might be needed before a decision on modifications is taken.
Finally, at several points in this report we contend that as well as looking at data for a large
number of countries to try and draw out general patterns and driving forces, it is important to
consider in more detail issues at the country specific level. To do so for each of the GSP
beneficiaries would be a major task. However, if the objective is to increase the effectiveness
of the existing scheme, then we suggest that it would be worth identifying those countries
where preference margins are not insignificant, yet preferences are not being utilized. The
aim would be to examine what the underlying causes are because understanding this is
likely to be an important pointer towards effective policy making.
7.2.2 Alternative policies
The remit of devising policies to help the development process, especially with regard to
those countries most in need is extremely broad, and could result in a very long list of
possible policy options. In the context of this study therefore it is important to keep the focus
on trade and trade related policy measures. However, consideration of alternative (non-tariff)
trade-based policies designed to help those developing countries most in need is much
more difficult to identify. Tariffs are clearly a key trade policy instrument, as well as being an
196
instrument which the enabling clause of the WTO allows to be use preferentially under GSP
style schemes.
In the discussion below we suggest three areas which are worth considering. The first two of
these, loosely entitled ―aid for trade initiatives‖ and ―non-tariff measures‖ are fairly closely
interrelated. The third – import subsidies – is a separate and more speculative suggestion.
a. “Aid for trade initiatives”: As is well known openness to markets (tariff preferences) is
not in and of itself sufficient to lead to higher exports and higher growth rates. The
interrelationship between openness and domestic institutions and infrastructure is
extremely important. There is already a good and growing understanding in the literature
of the importance of some of the transmission mechanisms (in terms of infrastructure,
regulatory environment, financial infrastructure etc), and it is important to put into place
effective mechanisms for identifying on a country basis what the constraints / problems
are, and focussing therefore on the most appropriate aid for trade policies.
We recognise that this is a broad area, and that considerable work is already being done
with regard to this by the European Commission. The point of highlighting it is to
emphasise its importance in the integration of LDCs and especially of those most in need
into the world trading system. This underlines the importance of continued and
strengthened cooperation / coordination between the different branches of the
Commission with regard to such initiatives. We would also suggest that the production of
regular country reports directly focussed on the obstacles to trade expansion in those
countries most in need would help to focus policy both internally and externally.
The preceding is also related to point raised towards the beginning of this report. The
relationship between trade, productivity, and economic growth is not only about improved
access to export markets. Access to export markets may well be significant, although it is
interesting to note that to date much of the empirical literature on this suggests that the
impact of exporting on productivity tends to be somewhat low. Also important is the
degree of domestic openness, and this is often linked to the underlying institutional and
infrastructure environment within a given country. Hence, an important part of helping
developing countries to engage in the world trading system should be to encourage and
assist them in the process of domestic liberalisation where the role of aid for trade is
likely to be extremely important.
b. Non-tariff measures: As mentioned above, to some extent this is related to the
preceding point, but is different in an important regard. Market access is not simply about
the level of the tariffs being imposed by the importing country – there are various other
reasons why access to markets may be impeded ranging from internal domestic
infrastructure issues, to issues of trade facilitation issues, as well as technical measures
and standards etc. There is growing evidence that such ―deep integration‖ issues can
and do indeed impede countries‘ access to some markets. The classic example here
concerns health or technical standards. Where such standards exist, a country not only
has to produce to the required standard but also to prove that it has done so, which in
turn requires accreditation of the appropriate testing and monitoring bodies. For many
industries in developing countries a key issue is the incapacity to meet the required
197
standards. Even where these are met the fixed costs and/or the coordination costs
associated with this may be prohibitive, and hence even where the good is produced to
the required standard it may not be possible to export it.
To the extent that such issues are a constraint for developing countries then this
suggests that trade policy designed to help these countries should focus on identifying
those constraints, and attempting to resolve them. This does not mean offering
developing countries preferential access, for example with regard to lower SPS or
technical standards or requirements). Rather, it means is identifying why and the
circumstances under which these behind the border issues may be a particular problem
for developing countries, and considering the most appropriate policy measures to help
the countries resolve those impediments.
c. Import subsidies/negative tariffs: The final policy option which we suggest worth more
careful consideration is that of import subsidies / negative tariffs which the EU would
offer LDCs. (Olarreaga and Limao (2005)).73
The logic behind this is as follow: As this report noted, the extent to which preferential
access can help developing countries is inevitably constrained by both the structure of
these countries exports and by the height of the possible preference margin which
depends on the EU‘s MFN tariffs (where of course the endogeneity between these two
needs to be recognised). Hence the preference margin has a lower bound.
If the EU were to subsidise developing countries on their exports to the EU, where the
tariff is already zero this would constitute an import subsidy, and where the tariff was
positive and less than the tariff this would be equivalent to a tariff reduction.
There are three key advantages to this proposal. The first, is that the preference margin
no longer has a lower bound, and therefore larger preferences than is currently the case
could be offered to developing countries and in particular to those countries most in
need. The second is that because of the lack of a lower bound a key feature of the policy
could be that all imports irrespective of which product they are in receive the same
subsidy. One of the common criticisms levied against schemes such as the EU‘s current
GSP scheme is that it offers a differential preferential access by product and is thus
inherently distortionary. Countries may be choosing to specialise in those sectors where
the preference margin is highest even if this does not accurately reflect their underlying
comparative advantage. By offering an equal import subsidy to all sectors this
distortionary element would be avoided, and it would encourage countries to specialise
more in those sectors in which they have a comparative advantage.74 Hence the impact
on total exports, as well as exports on the extensive margin (new products) is likely to be
73
There is of course an important issue concerning whether such a ―negative tariff‖ would be WTO compatible and is therefore within the remit of a given developed economy. This requires closer examination. Of course if it turns out that such a policy were currently WTO incompatible, then providing there was support for its introduction we would propose negotiating the appropriate waiver required within the WTO before its introduction. 74
As with any change in policy, the introduction of such a policy is likely to lead to winners and losers across and within countries. We have explored this issue with the use of our CGE model and more detailed results can be supplied on request.
198
greater. The third advantage stressed by Olarreaga and Limao, is that switching to this
sort of policy is likely to make it (politically) easier to achieve MFN tariff cuts and thus
could increase the likelihood of progress in the WTO.
199
8 References: Abed, G.T. and S. Gupta (eds.) (2002), Governance, Corruption and Economic
Performance, International Monetary Fund, Washington DC.
Alfieri, A. and Cirera, X. (2008), ―Unilateral Trade Preferences in the EU: An Empirical
Assessment for the Case of Mozambican Exports‖, DNEAP Discussion Paper 60E
Anderson, J (1979), ―A Theoretical foundation for the gravity equation‖, American Economic
Review, 69(1), pp.106-116.
Anderson J. & van Wincoop E., 2001. "Borders, Trade and Welfare," NBER Working Papers
8515, National Bureau of Economic Research.
Anderson, J & Wincoop, E., (2003) ―Gravity with gravitas: A solution to the border puzzle‖,
American Economic Review, 93, 170-92.
Baldwin, R. (1993) ―A Domino Theory of Regionalism‖, NBER WP 4465, National Bureau of
Economic Research
Baldwin, R. & Taglioni, D., (2006) "Gravity for Dummies and Dummies for Gravity
Equations," NBER Working Papers 12516, National Bureau of Economic Research.
Baltagi, B. H, Egger, P and Pfaffermayr, M, (2008). "Estimating regional trade agreement
effects on FDI in an interdependent world," Journal of Econometrics, Elsevier, vol.
145(1-2), pages 194-208, July.
Bergstrand, J. (1985), "The gravity equation in international trade: some microeconomic
foundations and empirical evidence", The Review of Economics and Statistics, 20,
pp.474-81.
Bevan, A. and S. Estrin (200) ―Determinants of FDI in transition economies‖, Working Paper
342, Centre for New and Emerging Market, London Business School.
Boockmann, B (2001), ―The ratification of ILO conventions: A hazard rate analysis‖
Economics and Politics, vol. 13(3), pp. 281-309, November.
Boockmann, B., D. Diurdjevic, G. Horny and F. Laisney (2009), "Bayesian estimation of Cox
models with non-nested random effects: an application to the ratification of ILO
conventions by developing countries," Documents de Travail 249, Banque de France.
Bourgeois, J, K. Dawar and S. J. Evenett (2007), ― Comparative Analysis of Selected
Provisions in Free Trade Agreements‖, Report for DG TRADE,
http://trade.ec.europa.eu/doclib/docs/2008/march/tradoc_138103.pdf
Bourguignon F., Fournier M. and Gurgand M., (2004) ―Selection Bias Corrections Based on
the Multinomial Logit Model: Monte-Carlo comparisons‖, mimeo Delta.
Brenton, P ect. (1999) ―Export, prices, technology and hysteresis: a disaggregated analysis‖
Discussion paper 99-18, Department of Economics, University of Birmingham.
Brenton, P. and Manchin, M. (2003), ―Making EU Trade Agreements Work: The Role of
Rules of Origin‖, World Economy, May 2003, vol. 26, no. 5
Brenton, P. and Newfarmer, R. (2007) ―Watching more than Discovery Channel: Export
Cycles and Diversification in Development‖, World Bank Policy Research Working
Paper No. 4302, World Bank, Washington DC
200
Cadot, Olivier & Carrere, Celine & De Melo, J. & Tumurchudur, Bolormaa, (2006). "Product-
specific rules of origin in EU and US preferential trading arrangements: an
assessment," World Trade Review, Cambridge University Press, vol. 5(02), pages
199-224, July
CARIS (2008), 'The Impact of Trade Policies on Pakistan's Preferential Access to the
European Union' , Report to the European Commission
Carrere, C. and J. de Melo (2004), ―Are Rules of Origin Equally Costly? Estimates from
NAFTA‖ CEPR Discussion Papers 4437, Centre of Economic Policy Research,
London
Chang, W. and Winters, L. A. (2002), ―How Regional Blocs Affect Excluded Countries: The
Prices Effects of MERCOSUR‖, American Economic Review, Vol. 92, Issue 4, pp.
889-904, September
Cheng, I-Hui and Wall J., (2005). ―Controlling for Heterogeneity in Gravity Models of Trade
and Integration‖ Federal Reserve Bank of St. Louis Review, January/February, 87(1),
pp. 49-63.
Chokheli, Nino et al. (2008), ―Report on implementation of the GSP + related conventions in
Georgia‖, GEPLAC research paper, October.
Compa, L. (2003), ―Assessing assessments: A survey of efforts to measure countries‘
compliance with freedom of association standards‖, Comparative Labour Law &
Policy Journal, Vol. 24, pp. 283-320.
Compa, Lance and Vogt, Jeffrey S. (2003), ‖Labour Rights in the Generalized System of
Preferences: A 20-Year Review‖, Comparative Labour Law & Policy Journal, Vol. 22,
No. 2/3.
Deardorff, A.V. (1997), "Determinants of bilateral trade : Does gravity work in a classical
world ?", in The Regionalization of the World Economy, Jeffrey Frankel ed.,
University of Chicago Press.
DeMaria, F., S. Drogue and A. Matthews (2008) ―Agro-Food Preferences in the EU‘s GSP
Scheme: An Analysis of Changes Between 2004 and 2006‖, Development Policy
Review 26(6), 693-712.
Doepke, Matthias and Fabrizio Zilibotti (2009). ―Do International Labor Standards Contribute
to the Persistence of the Child Labour Problem?‖ NBER Working Paper 15050,
National Bureau of Economic Research.
Dutt, P. & Traca, D. (2008), ―Corruption and bilateral trade flows: Extortion or evasion?‖,
INSEAD working paper.
Eaton, J. & Tamura A., 1996. "Japanese and U.S. Exports and Investment as Conduits of
Growth," NBER Working Papers 5457, National Bureau of Economic Research, Inc.
Egger, P, (2002), ―An econometric view of the estimation of gravity models and the
calculation of trade potentials‖, World Economy, pp.297-312.
Egger, P. & Pfaffermayer, M., (2004), ―Foreign Direct Investment and European Integration
in the 1990s‖, World Economy, pp. 99-110.
201
Evenett, S, & Keller, W, (2002), ―On theories explaining the success of the gravity equation‖
Journal of Political Economy, 110; 281-316.
Evennet, S. (2008) ―The European Union's Generalised System of Preference: An
Assessment of the Empirical Evidence‖, Forthcoming. Centre for Economic Policy
Research, London.
Eyckmans, Johan, Erika Meynaerts and Sara Ochelen (2004), "The Environmental Costing
Model: a tool for more efficient environmental policymaking in Flanders," Energy,
Transport and Environment Working Papers Series ete0405, Katholieke Universiteit
Leuven, Centrum voor Economische Studiën, Energy, Transport and Environment.
Feenstra, R. (2004), Advanced International Trade, Princeton University Press, Princeton,
Ch.5.
Feenstra, R, Markusen, J, & Rose, A, Using the gravity model to differentiate among
alternative theories of international trade‖,
Freedom House (2009), Freedom in the World 2009 http://www.freedomhouse.org.
Freres C. and A. Mord (2004), ―European Union Trade Policy and the Poor. Towards
Improving the Poverty Impact of the GSP in Latin America‖, Madrid, Instituto Complutense
de Estudios Internationales Working Paper 02/04.
Frundt, Henry J. (1998), Trade Conditions and Labour Rights. U.S. Initiatives, Dominican
and Central American Responses, University Press of Florida.
Gamberoni, Elisa (2007), ―Do Unilateral Trade Preferences Help Export Diversification? An
Investigation of the Impact of European Unilateral Trade Preferences on the
Extensive and Intensive Margin of Trade‖, HEI Working Papers 17-2007, Economics
Section, The Graduate Institute of International Studies.
GEPLAC (2007), ―Short Note on the Compliance of Georgian Labour Legislation with
International Labour Standards in Light of EU Generalised System of Preferences‖,
Georgian Law Review, 10/2007-4, pp. 407-417.
Glick, R and A. K. Rose (2001) ―Does a Currency Union Affect Trade? The time series
evidence‖, NBER Working paper 7857, National Bureau of Economic Research.
Gonzalez, J. (2003). Colombia dentro del Sistema Generalizado de Preferencias de la Unión
Europea, Bogotá, ANDI (www.andi.com.co).
Greven T.(2005), ―Social Standards in Bilateral and Regional Trade and Investment
Agreements‖, Dialogue on Globalisation N° 16 / March.
Grossmann, Gene M. and and Anal O. Sykes (2005), ―A preference for development: the law
and economics of GSP‖, World Trade Review, 4 , pp 41-67.
Hadi, A. S. (1992) "A New Measure of Overall Potential Influence in Linear Regression"
Computational Statistics and Data Analysis, 14, 1- 27.
Hafner-Burton, Emilie M. (2005), ―Trading Human Rights: How Preferential Trade
Agreements Influence Government Repression‖, International Organization, 59(3),
pp. 593-629.
Hafner-Burton, Emilie M. (2009), Forced to Be Good: Why Trade Agreements Boost Human
202
Rights, Cornell University Press,
Hamilton, C. & Winters, L.A. (1992), ―‗Trade with Eastern Europe‖, Economic Policy.
Hathaway, O. A. (2002) ―Do Human Rights Treaties make a Differences‖ Yale Law Journal
111: pp. 1935-2042
Hathaway, Oona A. (2003), "The Cost of Commitment", John M. Olin Centre for Studies in
Law, Economics, and Public Policy Working Papers. Paper 273,
http://digitalcommons.law.yale.edu/lepp_papers/273
Heckman, J. (1974) ―Shadow prices, market wages, and labour supply‖, Econometrica
42(4):679–694.
Heckman, J. (1979) ―Sample selection bias as a specification error‖, Econometrica, 47, 153–
61.
Heiko, H,. (2008). "Export Diversification and Economic Growth". Commission on Growth
and Development. Working Paper No. 21
Helpman, E. and Krugman, P. (1985), Market Structure and Foreign Trade: Increasing
Returns, Imperfect Competition and the International Economy. Cambridge, Mass.:
MIT Press
Herin, J. (1986), ―Rules of Origin and Differences Between Tariff Levels in EFTA and in the
EC, European Free Trade Association‖, Economic Affairs Department, Occasional
Paper No. 13
Hillman, Arye L. and Eva Jenkner (2004), ―Educating Children in Poor Countries‖, IMF
Economic Issues, No.33, International Monetary Fund, Washington DC.
Horn, H., P. Mavroidis and A. Sapir (2009), ―Beyond the WTO? An Anatomy of EU and US
preferential Agreements‖, Bruegel Blueprint Series, Volume VII
ICTSD (2003), ―Latin America to Feel Effects Of EU GSP Review', Bridges Weekly Trade
News Digest • Volume 7 • Number 15 • 30th April
ILO-IPEC (2003), Investing in Every Child: An Economic Study of the Costs and Benefits of
Eliminating Child Labour, Geneva.
IMF (2005), Trade Conditionality Under Fund-Supported Programs, 1990-2004 Background
Paper to the Review of Fund Work on Trade, International Monetary Fund. Available
at http://www.imf.org/External/np/pp/eng/2005/021405.htm.
Imbs, J., and Wacziarg,R. (2003). ―Stages of Diversification.‖ American Economic Review,
March,93(1):63-86
Kassouf, Ana Lúcia, Peter Dorman and Alexandre Nunes de Almeida (2005), ―Costs and
benefits of eliminating child labour in Brazil‖, Economia Aplicada, vol.9 no.3, pp. 343-
368, doi: 10.1590/S1413-80502005000300001.
Kaufmann, Daniel, Kraay, Aart and Mastruzzi, Massimo (2009),‖Governance Matters VIII:
Aggregate and Individual Governance Indicators, 1996-2008‖, World Bank Policy
Research Working Paper No. 4978, the World Bank, Washington DC.
Kucera, David and Sarna, Ritash (2004), ―Child Labour, Education and Export Performance‖,
International Labour Office Working Paper No. 52.
203
Kyriazidou, E. (2001), ―Estimation of Dynamic Panel Data Sample Selection Models‖, The
Review of Economic Studies, Vol. 68, No. 3, (July), pp. 543-572
Linnemann, H. (1966). ―An Econometric Study of International Trade Flows‖; Amsterdam,
North Holland
Lofgren, H., Robinson, S. and El-Said, M. (2003), ―Analysis in a General Equilibrium
Framework: The Representative Household Approach‖ in The Impact of Economic
Policies on Poverty and Distribution, Bourguignon, F and Pereira da Silva, L, eds.
World Bank and Oxford University Press, 2003
Maddala, G. S., (1992), Introduction to Econometrics, second edition. New York, Macmillan
Publishing Company
Manchin, M. 2006. "Preference Utilisation and Tariff Reduction in EU Imports from ACP
Countries," The World Economy, Blackwell Publishing, vol. 29(9), pages 1243-1266,
09.
Markusen J & Venables A (1995) ―Mulitnational firms and the new trade theory‖ NBER
working paper 5036, National Bureau of Economic Research.
Markusen J & Venables A, (1997) "Foreign Direct Investment as a Catalyst for Industrial
Development," NBER Working Papers 6241, National Bureau of Economic
Research.
Markusen, J. and Venables, A. (2000), ―The Theory of Endowment, intra-industry and multi-
national trade‖, Journal of International Economics, Vol. 52, Issue 2, pp. 209-234,
December.
Matyàs L. (1997) "Proper Econometric Specification of the Gravity Model" The World
Economy Vol. 20 (3).
McCulloch, N., Winters, L.A, Cirera, X (2002) Trade Liberalization and Poverty: A Handbook,
Centre for Economic Policy Research, London, February, 2002.
McDonald, S., Thierfelder, K., & Robinson, S. (2007). ―Globe: A SAM based global CGE
model using GTAP data‖. USNA Economics Department Working Paper no. 14, US
Naval Academy.
McDonald, S. and Willenbockel, D. (2008) in Bayar, A. (eds), Proceedings International
Conference on Policy Modelling - ECOMOD 2008 Berlin, Florence, MA: EcoMod
Press
McDonald, S., Thierfelder, K., & Robinson, S. (2008). ―Asian Growth and Trade Poles: India,
China, and East and Southeast Asia‖, World Development, Vol. 36, Issue 2, February
2008, pp. 210 – 234
McKenzie, Michael (2008), ―Climate Change and the Generalized System of Preferences‖,
Journal of International Economic Law 11(3):679-695; doi:10.1093/jiel/jgn024
Melitz, M. (2003). "The impact of trade on intra-industry reallocations and aggregate industry
productivity", Econometrica 71, 1695-1725.
Mold, Andrew (2007), ―Pulling Back from the Brink? Evaluations, Options, and Alternatives to
the EPAs‖, Real Instituto Elcano Working Paper 33/2007.
204
Mold, Andrew (2009), Policy Ownership and Aid Conditionality in the Light of the Financial
Crisis, OECD Development Centre, Paris. ,
Narayanan, B. and T. Walmsley (eds) (2008) Global Trade, Assistance and Production: The
GTAP7 Data Base. Centre for Global Trade Analysis, Purdue University.
Neumayer, Eric (2005), ―Do International Human Rights Treaties Improve Respect for
Human Rights?‖, Journal of Conflict Resolution 49(6), pp.925-953.
Neumayer, Eric and Indra De Soysa (2005), ‗Globalization and the Right to Free Association
and Collective Bargaining: An Empirical Analysis‘, World Development Vol. 34, No. 1.
Nieto Solís, José Antonio (2002), ―Exportación Andina Hacia la UE: Indices de
Especialización Comercial y Cuotas de Mercado‖, ICES-Información Comercial
Española, October, Nº 802, pp.173-194.
Nilsson, L & Matsson, N, ‖Truths and myths about the openness of EU trade policy and the
use of EU trade preferences‖, European Commission 2009, mimeograph.
OECD (2005), ―Preferential Trading Arrangements in Agricultural and Food Markets - The
Case of the European Union and the United States‖, Paris.
Olarreaga, M. and Limao, N. (2005), ―Trade preferences to small developing countries and
welfare cost of lost multilateral liberalization‖, World Bank Policy Research Working
Paper No. 3565, World Bank, Washington DC.
Olarreaga, M. and Ozden, C, (2005) ―AGOA and Apparel: Who Captures the Tariff Rent in
the Presence of Preferential Market Access?‖, The World Economy, Vol. 28, No. 1,
pp. 63-77, January
Olivié, Iliana (2009), ―Coherence of Development Policies: Ecuador‘s Economic Ties with
Spain and their Development Impact‖, OECD Development Centre working paper no
290, Paris.
Orbie Jan and Lisa Tortell (2009), ―The New GSP-Plus Beneficiaries: Ticking the Box or
Truly Consistent with ILO Findings?‖, Paper presented at the European Union
Studies Association (EUSA) 11th Biennial International Conference, Marina Del Rey,
California, 23-25 April.
Ozden, C. and Reinhardt, E., (2003), ―The Perversity of Preferences: The Generalized
System of Preferences and Developing Country Trade Policies, 1976-2000‖, World
Bank Policy Research Working Paper No. 2955, January
Ozden, C. and Sharma, G. (2004) ―Price Effects of Preferential Market Access: The
Caribbean Basin Initiative and the Apparel Sector‖, World Bank Policy Research
Working Paper No. 3244
Paczynski, Wojciech (2009), ―European Neighbourhood Policy and Economic Reforms in the
Eastern Neighbourhood‖, CASE Network Studies and Analyses No. 382. Available at
SSRN: http://ssrn.com/abstract=1373951
Persson, Maria, and Fredrik Wilhelmsson. (2006), ―Assessing the Effects of EU Trade
Preferences for Developing Countries‖, Lund University, Department of Economics,
Working Paper 2006:4.
Polaski, S., de Souza Ferreira Filho, J., Berg, J., McDonald, S., Thierfelder, K., Willenbockel,
205
D. and Zepeda, E. (2009) ‖Brazil in the Global Economy. Measuring the Gains from
Trade‖ , Carnegie Report, April
Pöyhönen, P., (1963). ― A tentative model for the volume of trade between countries‖,
Weltwirtschaftliches Archiv 90, 93-100
Saner, Raymond and Ricardo Guilherme (2007), ―The International Monetary Fund‘s
Influence on Trade Policy: A Legal Critique‖. Journal of World Trade, 41(5): pp 931-
981.
Schott, Peter K. (2004) ―Across-Product versus Within-Product Specialization in International
Trade‖, Quarterly Journal of Economics, Vol 119, No. 2. 647-678.
Santos Silva, J.M.C. & Tenreyro, S. (2006), ―Log of Gravity‖, Review of Economics and
Statistics, 88(4), pp. 641-658.
Schott, P. K. (2004). ―Across-Product versus Within-Product Specialization in International
Trade‖ Quarterly Journal of Economics 119 (2): 647–678.
Social Watch (2009), Gender Equity Index (GEI) 2008,http://www.socialwatch.org/indicators.
Soloaga I., and Winters A., (1999), "How has Regionalism in the 1990s affected trade?",
World Bank Policy Research Working Paper 2156, World Bank, Washington DC.
Stevens, C. and J. Kennan (2004), ―The utilization of EU preferences to the ACP‖, paper
presented to technical seminar on Tariff Preferences and their Utilisation, WTO,
Geneva, 31 March, mimeo
Stevens, C. and J. Kennan (2005), ―GSP Reform: a longer-term strategy (with special
reference to the ACP)‖, Report prepared for the Department for International
Development, UK, available at Institute of Development Studies.
Tinbergen, J. (1962). ―Shaping the World Economy‖; in Suggestion for an International
Economic Policy, The Twentieth Century Fund
Vella, F. (1998), ―Estimating Models with Sample Selection Bias: A Survey‖, The Journal of
Human Resources, 33: 127—169.
Wells, Don (2006), "Best Practices in the Regulation of International Labour Standards:
Lessons of the U.S.-Cambodia Textile Agreement", Comparative Labour Law and
Policy Journal v.27 pp. 357-371.
Wooldridge, J. (1995) ―Selection Corrections for Panel Data Models Under Conditional Mean
Independence Assumptions,‖ Journal of Econometrics 68, 115-132, July 1995.
World Bank (2006), Assessing the World Bank Support for Trade, 1987-2004. An IEG
Evaluation, Washington, DC.
Top Related