Mid-term Evaluation of the EU s Generalised System of Preferences:

Transcription

1 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.

2 Contents: List of Tables... 4 List of Figures... 5 List of Abbreviations... 6 Executive Summary Introduction Motivation and objectives Section Summary Overview of the GSP Section Summary Preferential Access, Trade and Competitiveness The structure of the EU s GSP regimes Section Summary GSP and developing country exports EBA countries GSP+ countries: GSP countries: Section Summary Impact of preference regimes on other LDCs Similarities in export structures Relative Export Competitive Pressure Index Section Summary GSP and LDC development needs Preferences and development Analysis of changes in intensive and extensive margin Section Summary Section 2: Conclusions Utilisation Rates Descriptive stats on utilisation rates Section Summary The determinants of preference utilisation Preference utilisation: econometric analysis Section Summary Price margins or who captures the preference rent? Methodology The Data Econometric Analysis Section Summary Section 3: Conclusions Gravity Modelling Aggregate modelling of trade and investment The target model Data Estimation and Results

3 4.1.4 Trade FDI Section Summary Sectoral multilateral gravity modelling of trade Section Summary The impact of preferences on trade flows at the product level Section Summary Section 4: Conclusions Computable General Equilibrium Evaluation of GSP Introduction Section Summary The GLOBE Model Patterns of Trade and Production in the Benchmark Equilibrium Section Summary Simulation Results Change from 2004 to 2006 EU GSP GSP A World without the EU GSP MFN04/ Full Utilization of EU GSP Preferences FULLGSP Further Reform of the EU GSP: The Extreme Case - ZEROTM Section Summary Section 5: Conclusions Qualitative Assessment of the GSP Implementation & effects of international conventions Lessons from the literature From ratification to implementation: legal analysis Challenges of implementation: lessons from case studies Quantification of implementation effects Section Summary Costs and benefits of fostering sustainable development and good governance GSP+ beneficiaries perspective Section Summary Selection criteria for GSP Vulnerability criteria International conventions Section Summary Section 6: Conclusions Conclusions and policy recommendations What do we learn from the analysis undertaken? Policy options Amending/improving existing schemes Alternative policies References:

4 List of Tables Table 2.1: EU Imports by Preference Regime Table 2.2: Coverage of EU Preferential Regimes Table 2.3: Coverage of EU Preferential Regimes (share of tariff lines) Table 2.4: Share of Tariff Lines by Regime and Size of Tariff (2008) Table 2.5: Average Tariff by Regime and TDC Sector (2002 and 2008) Table 2.6: Preference Margins by TDC Sector Compared to MFN (2002 & 2008) Table 2.7: Preference Usage by Regime Type Table 2.8: Summary Results on Export Similarity Table 2.9: Summary RECPI by Regime Type Table 2.10: Competitive Pressure by Country upon each Regime Type all trade Table 2.11: Competitive Pressure by Country upon each Regime Type MFN > Table 2.12: Preferences and Development Table 2.13: Annual Growth of Exports by Category Table 2.14: Export Growth Decomposition by Varying Identification Procedures Table 2.15: Differences between the Hypothetical MNF and Applied Tariffs Table 2.16: Difference between the Hypothetical MFN & Hypothetical Applicable Tariffs Table 3.1: EBA Suitability Table 3.2: GSP Suitability Table 3.3: Determinants of Non Utilisation Table 3.4: Export Price Ratio Specification Table 3.5: Multinomial Logit for Selection Utilisation Table 3.6: Export Price Ratio Specification with Multinomial Selection (pref. margin) Table 3.7: Export Price Specification Table 4.1: Percentage Change in Aggregate Trade Table 4.2: Percentage Change in FDI Table 4.3: Percentage Change in Trade at Sectoral Level Table 4.4: Gravity Model at Product Level-Tariff Regime Table 4.5: Gravity at HS4 with Selection Table 4.6: Gravity Model at HS4 with Wooldridge Panel Selection Table 5.1: Regional Aggregation of the Model Table 5.2: Commodity Aggregation of the Model Table 5.3: Selected Benchmark Macro Indicators by Country Table 5.4: Regional Origin Shares in Extra-EU Imports by Commodity (%) Table 5.5: EU Share in Countries Total Exports by Commodity (%) Table 5.6: Sectoral Shares in GDP by Country (%) Table 5.7: Simulation Scenarios Table 5.8: Percentage Changes in the Power of EU Import Tariffs by Scenario & Commodity Group Table 5.9: Change in Real Absorption by Country and Scenario Table 5.10: Change in Real Absorption by Country and Scenario Table 5.11: Terms of Trade Change by Region and Scenario Table 5.12: Change in Aggregate Export Volume by Country and Scenario Table 5.13: Change in Export Volume to the EU by Commodity GSP Table 5.14: Change in Export Volume to the EU by Commodity MFN Table 5.15: Change in Export Volume to the EU by Commodity - FULLGSP Table 5.16: Change in Export Volume to the EU by Commodity ZEROTM Table 5.17: Change in Real Output by Sector and Region GSP Table 5.18: Change in Real Output by Sector and Region MFN

5 Table 5.19: Change in Real Output by Sector and Region FULLGSP Table 6.1: Ratifications of 27 Conventions in Present and Past GSP+ Beneficiaries Table 6.2: Kendall Rank Correlation Coefficients between Export Concentration Ratios and Selected Economic and Geographical Characteristics Table 7.1: Impact of changing the graduation threshold: List of Figures Figure 2.1: Incidence of Preference Margins at 10-digit level Figure 2.2: Distribution of countries by the share of the EU in total exports Figure 2.3: EBA Preference Margins Figure 2.4: EBA Countries: Share of MFN>0 Trade Figure 2.5: GSP+: Tariffs and MFN Preference Margins Figure 2.6: GSP+: GSP and EBA Margins Figure 2.7: GSP+: Change in Share of Trade Eligible for Duty Free Access Figure 2.8: GSP+ Countries: Share of MFN>0 Trade Figure 2.9: GSP Countries: Share of MFN>0 Trade Figure 2.10: GSP Countries: Frequency Chart with Difference between MFN and GSP Figure 2.11: GSP Average Preference Margins Figure 2.12: Change in Share of Duty Free Eligible Exports under MFN & GSP Figure 2.13: Changes in Exports by Type and Grouping ( ) Figure 2.14: Correlation between Intensive and Extensive Margins and Applied Tariffs Figure 2.15: Correlation between Preference and Export Margins Figure 3.1: EBA Utilisation Rates Figure 3.2: GSP Utilisation Rates Figure 3.3: Correlation between Average Preference Margin and Utilisation Rates Figure 3.4: All GSP Regimes: Correlation between Preference Margin and Non-Utilisation of Preferences Figure 3.5: Probability Distribution Function of Preference Non-utilisation Exports as a Share of Total Exports in 2007 by Country Figure 3.6: Probability Distribution Function of Preference Non-utilisation Exports as a Share of Total Exports in 2007 by Product Figure 3.7: Probability Distribution Function of Preference Non-utilisation Exports as a Share of Eligible Exports in 2007 by Country Figure 3.8: Probability Distribution Function of Preference Non-utilisation Exports as a Share of Eligible Exports in 2007 by Product Figure 3.9: Prices and the Preference Rent Figure 3.10: Kernel Estimate of pdf for Log Ratio of Prices

6 List of Abbreviations AGOA: CAFTA: CAP: CES: CET: CGE: CITES: CPIA: EBA: EU: FDI: FK: GEI: GNI: GSP: HDI: HPI: ILO: LDCs: MFN: NAFTA: NGOs: NTMs: OECD: OLS: PTAs: PPP: RCA: RECPI: RoO: RTAs: TOT: WTO: UNCTAD: UNICEF: African Growth and Opportunity Act Central American Free Trade Agreement Common Agriculture Policy Constant Elasticity of Substitution Constant Elasticity of Transformation Computable General Equilibrium Convention on International Trade in Endangered Species World Bank s Country Policy and Institutional Assessment Everything But Arms European Union Foreign Direct Investment Finger Kreinin index Gender Equality Index Gross National Income General System of Preferences Human Development Index Human Poverty Index International Labour Organisation Least Developed Countries Most Favoured Nations North American Free Trade Agreement Non-Governmental Organisations Non-Trade Measures Organisation of Economic Cooperation and Development Ordinary Least Squares Preferential Trade Arrangements Purchasing Power Parity Revealed Comparative Advantage Relative Export Competitive Pressure Index Rules of Origin Regional Trade Agreements Terms of Trade World Trade Organisation United Nations Conference on Trade and Development United Nations Children s Funds 6

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. 7

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 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 percent of GSP countries exports to the EU entered the EU with a zero MFN tariff, percent of GSP+ countries' exports, and 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 percent, 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. 8

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 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 9

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 10

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. 11

12 1 Introduction The EU s Generalised System of Preferences (GSP 1 ) 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 The most recent scheme is applicable from 1 st January 2009 to 31 st December 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+. 12

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 o o o o o o Greater impact on those developing countries most in need the most vulnerable, those with the lowest income levels, small islands, landlocked, etc. Higher economic growth, as a result of higher exports and greater integration in the world economy. More regional trade, which may in turn be influenced by possibilities of regional cumulation in the underlying rules of origin. A positive impact on sustainable development, in the context principally of areas such as labour standards, environment etc. Reduction in poverty. Diversification. 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 13

14 issues of the geographical concentration of economic activity (economic geography) and longrun 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). 14

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 nontariff 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 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. 15

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 tenyear 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 Imports from EBA eligible countries also increased by 35 percent in The GSP+ beneficiary countries exports to the EU increased 15 percent in 2006 and a further 10 percent in 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 16

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 products 4 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. 17

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. 18

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 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 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. 19

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. 20

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. 8 This 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. 21

22 MFN=0 MFN>0 GSP=0 GSP>0 GSP+0 GSP+>0 EBA=0 EBA>0 Other pref=0 Other pref>0 Unknown Trade under zero tariffs 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 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 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 22

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, 23

24 under GSP+ there are 9717, and under EBA 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 = percent 22.1 percent 22.1 percent MFN > percent 7.6 percent 0.3 percent Duty Free percent 68.1 percent 77.5 percent Positive pref. tariff percent 2.1 percent 0.0 percent Total percent percent 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 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,

25 Table 2.3: Coverage of EU Preferential Regimes (share of tariff lines) GSP GSP+** EBA GSP GSP+ EBA GSP GSP+ EBA MFN = MFN > Pref. Duty Free Positive pref. tariff 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 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) Change MFN GSP GSP+ EBA MFN GSP GSP+ EBA MFN GSP GSP+ EBA Tariff = Tariff 0<t Tariff 5<t Tariff 10<t Tariff 15<t specific non ad valorised 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 little. 25

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 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) TDC Description MFN GSP GSP+ EBA MFN GSP GSP+ EBA I Live animals; animal products II Vegetable products III Animal or vegetable fats and oils IV Prepared foodstuffs; V Mineral products VI Products of the chem.& allied inds VII Plastics and Articles thereof VIII Raw hides and skins, leather, furskins IX Wood and articles of wood X Pulp of wood or other fibrous Xia Textiles XIb Textile articles (clothing) XII Footwear, headgear, umbrellas XIII Articles of stone, plaster, cement XIV Pearls, precious, semi-precious stones XV Base metals and articles of base metal XVI Machinery and mechanical appliances XVII Vehicles, aircraft, vessels, transport XVIII Optical, photographic... Instruments XIX Arms and ammunition; XX Miscellaneous manufactured articles XXI Works of Art, collectors' piece 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 26

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) change in pref margin TDC Description GSP GSP+ EBA GSP GSP+ EBA GSP GSP+ EBA I Live animals; animal products II Vegetable products III Animal or vegetable fats and oils IV Prepared foodstuffs; V Mineral products VI Products of the chem... & allied inds VII Plastics and Articles thereof VIII Raw hides and skins, leather, furskins IX Wood and articles of wood X Pulp of wood or other fibrous Xia Textiles XIb Textile articles (clothing) XII Footwear, headgear, umbrellas XIII Articles of stone, plaster, cement, XIV Pearls, precious, semi-precious stones XV Base metals and articles of base metal XVI Machinery and mechanical appliances XVII Vehicles, aircraft, vessels, transport XVIII Optical, photographic,... Instruments XIX Arms and ammunition; XX Miscellaneous manufactured articles XXI Works of Art, collectors' piece 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 regime 10, 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. 27

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, percent, 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 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. 28

29 Figure 2.1: Incidence of Preference Margins at 10-digit level 29

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 30

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 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. 31

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. 32

33 Figure 2.2: Distribution of Countries by the Share of the EU in Total Exports EBA <10 <25 <50 <75 <100 EBA GSP <10 <25 <50 <75 <100 GSP 10 9 GSP <10 <25 <50 <75 <100 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 percent of GSP countries exports to the EU entered the EU with a zero MFN tariff, percent of GSP+ countries' exports, and percent of EBA countries' exports. It is interesting that we see a 33

34 big rise in the EBA share of MFN zero trade between 2007 and 2008 from percent to percent and a corresponding decline in the other preferences share from 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 percent, 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. 34

35 MFN=0 MFN>0 GSP=0 GSP>0 GSP+=0 GSP+>0 EBA=0 EBA>0 Other pref=0 Other pref>0 Unknown Table 2.7: Preference Usage by Regime Type Year type Total imports (in millions of Euros) EBA , GSP , GSP , OTHER , EBA , GSP , GSP , OTHER , EBA , GSP , GSP , OTHER , EBA , GSP , GSP , OTHER , EBA , GSP , GSP , OTHER , EBA , GSP , GSP , OTHER 35

36 MFN=0 MFN>0 GSP=0 GSP>0 GSP+=0 GSP+>0 EBA=0 EBA>0 Other pref=0 Other pref>0 Unknown Year type Total imports (in millions of Euros) , EBA , GSP , GSP , OTHER , 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. 36

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 EBA countries For 16 of the 49 EBA countries over 90 percent of their exports to the EU enter under MFN=0 12 ; and for a further seven 13 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 7 14 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 them 15 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. 37

38 No of Countries 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 <2.5% <5% <7.5% <10% <15% >15% 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. 38

39 Figure 2.4: EBA Countries: Share of MFN>0 Trade 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. 39

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 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. 12 Figure 2.5: GSP+: Tariffs and MFN Preference Margins 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. 40

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. 12 Figure 2.6: GSP+: GSP and EBA Margins Where the preceding figure compared the GSP-margin tariffs across EBA-margin 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. 41

42 Figure 2.7: GSP+: Change in Share of Trade Eligible for Duty Free Access 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 percent of their trade, and only one country (Ecuador) paying positive MFN tariffs on more than 25 percent of its trade. 42

43 Figure 2.8: GSP+ Countries: Share of MFN>0 Trade 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: 43

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 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: 44

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 45

46 No of Countries Figure 2.11: GSP Average Preference Margins <2.5% <5% <7.5% <10% <15% >15% 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. 46

47 No of Countries Figure 2.12: Change in Share of Duty Free Eligible Exports under MFN & GSP <10 <25 <50 <75 <100 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 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 47

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 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 The Finger-Kreinin index ranges between 0 and 1. If the structure 17 The mathematical formula for FK index is as follows: X X ih ij (7) FK ihj min, i X ih X ij i i Where i refer to a specific sector (or product), h to the home country and j to the partner country or to the RoW. X ih / Σ X ih is the share of product s i export in country s h total exports, X ij / Σ X ij 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. 48

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 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 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). 49

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 EBA GSP GSP Source: Authors calculations based on TARIC data supplied by the European Commission 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 sx pi RECPI sx i ji pi pi Where s pi represents the share of Uruguayan exports to the EU in total exports to the EU for a given product i and x j,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. 50

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 51

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 GSP GSP EBA GSP GSP 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. RECPI i i s s EBAi EBAi x x pi EBAi 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) 52

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 Equatorial Guinea EBA Bangladesh EBA Chad EBA Mozambique EBA Russian Federation GSP Libya GSP Saudi Arabia GSP Iran GSP Nigeria GSP Kazakhstan GSP Iraq GSP Azerbaijan GSP Algeria GSP Mexico GSP 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. 53

54 Table 2.11: Competitive Pressure by Country upon each Regime Type MFN > 0 EBA GSP GSP+ MFN Bangladesh EBA Mozambique EBA Cambodia EBA Madagascar EBA China GSP India GSP Tunisia GSP Morocco GSP Mauritius GSP Indonesia GSP Egypt GSP Russian Federation GSP Vietnam GSP Pakistan GSP Source: own calculations based on TARIC data supplied by the European Commission 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. 54

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 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 55

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 , the annualised growth rate over the same time period, and GDP per capita in 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. 56

57 Table 2.12: Preferences and Development Regime PM1 PM2 ETS DEA UR HDI HPI GDP growth GDP / capita ALL HDI Correlation HPI Real GDP Growth rate Annualised Growth Rate GDP per capita Source: own calculations EBA GSP GSP ALL EBA GSP GSP ALL EBA GSP GSP ALL EBA GSP GSP ALL EBA GSP GSP 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. 57

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 The aggregates are clustered countries according to the regime that was applicable to them in The first entry of the table tells us that during the period , 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 saw a doubling of annual trade growth rates with respect to 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. 58

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 Annual growth to EU 25 EBA % 8.6% GSP % 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: X i,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. 59

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 ( ) 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. 60

61 Method 3 Method 2 Method 1 Table 2.14: Export Growth Decomposition by Varying Identification Procedures INTENSIVE EXTENSIVE Existing Old New EBA 98.62% -2.12% 3.50% GSP % -1.10% 1.77% GSP 98.46% -0.29% 1.83 EBA GSP GSP EBA GSP GSP 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 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. 61

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. 62

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. 63