The effects of the everything but arms initiative on the agricultural exports of African, Caribbean and pacific least developed countries

(1)

University of Amsterdam

The effects of the Everything But Arms initiative on the

agricultural exports of African, Caribbean and Pacific

Least Developed Countries

Master’s thesis International Economics and Globalisation

Author: Daniël Heijtel Supervisor: drs. N. J. Leefmans

Student number: 10105050 Second reader: dr. K. Mavromatis

daniel.heijtel@student.uva.nl Date: 29 February 2016

This thesis focuses on the effects of the Everything But Arms initiative on the agricultural exports of African, Caribbean and Pacific Least Developed Countries to the European Union. The Everything But Arms initiative grants all Least Developed Countries duty-free and quota-free access to the internal market of the European Union. A log linear augmented gravity model is constructed for all agricultural exports of 70 African, Caribbean and Pacific countries to the EU-15 between 1995 and 2013, which is analysed using a fixed effects panel data regression. This thesis finds that the Everything But Arms initiative failed to stimulate the agricultural exports of African, Caribbean and Pacific Least Developed Countries to the European Union.

(2)

2

1. Introduction

In an effort to help Least Developed Countries (LDCs) integrate further into the global economy, the European Union (EU) introduced the Everything But Arms (EBA) initiative in 2001. Under this arrangement the EU has been granting all LDCs duty-free access to the EU market for all their exports, except for arms and ammunition. When the EBA entered into force, transitional provisions were installed for the import of bananas, rice and sugar. From October 2009, all these remaining restrictions have been fully liberalised. Since then, the EBA has granted full duty-free and quota-free access for all products from all LDCs, except for the import of arms and ammunition (European Commission, 2009: 1). This means that LDCs could have a competitive advantage over non-LDCs for products over which tariffs or quotas are applied by the EU for non-LDCs. This potential competitive advantage could stimulate exports and economic growth in LDCs. According to the European Commission, the EBA initiative is the most generous form of preferential treatment to LDCs globally (European Commission, 2014a: 1).

Despite the large theoretical advantages that the EBA offers to LDCs, the actual impact of the EBA initiative remains unclear. In a report on the effects of the EBA proposal, Stevens and Kennan state that although the direct effects on the least developed states will be positive, the impact will be modest given the limited supply capacity of these countries (Stevens and Kennan, 2001: 1). Furthermore, many of the LDCs already received duty-free access for most of their products destined for the EU market prior to the EBA initiative; for African, Caribbean and Pacific (ACP) LDCs, preferential access was already guaranteed under the Cotonou agreement which governs the trade relationship between the EU and ACP countries. The remaining nine LDCs were all eligible for a special tranche of the Generalised Scheme of Preferences (GSP) which provided them with additional preferences over those available to most developing countries (Stevens and Kennan, 2001: 2). Hence, the trade regulations for LDCs were already more favourable than those for non-LDCs prior to the implementation of the EBA. On top of that, the EBA entails strict administrative requirements and rules of origin which can be hard to meet, especially for LDCs.

Although the EBA initiative is very interesting from an international economics point of view, there is little empirical research on the effectiveness of the EBA, especially after the full liberalisation of the transitional provisions for bananas, rice and sugar. Gradeva and Martínez-Zarzoso attempted to estimate the effects of the EBA regime with an augmented gravity model for exports from the 79 ACP countries to the EU-15 for the time period of 1995 to 2005 (Gradeva and Martínez-Zarzoso, 2011: 1). In this study, which is discussed extensively in the literature review, the authors find that the EBA

(4)

4

agreement seems to decrease the value of exports in the absence of additional aid. This outcome appears to be very surprising. The authors mention that this result could be driven by the fact that their study has been performed prior to the full liberalisation of the trade in rice, sugar and bananas. Another explanation for this result is the more flexible rules of origin under the Cotonou agreement in comparison to the complex regulations under the EBA. The authors state that after the liberalisation of the transitional measures, the EBA agreement might be more effective (Gradeva and Martínez-Zarzoso, 2011: 20).

The literature on the EBA suggests that for ACP LDCs, the effects of the EBA are probably limited to agricultural products. Under the Lomé Convention, which was the predecessor of the Cotonou agreement, products originating in the ACP states that were not covered by the Common Agricultural Policy (CAP) could be imported into the EU free of custom duties and charges having equivalent effects. However, ACP countries received a product-specific low preference margin on most products regulated by the CAP, while the remaining products regulated by the CAP were subject to special protocols. This principle has been applied to successive agreements with ACP countries (European Commission, 2014b: 10). Hence, if the EBA initiative indeed stimulates ACP LDCs’ exports, this effect is most likely to be observed in the agricultural sector. It is also important to note that the effect of EBA is likely to increase over time. The transitional measures on bananas, rice and sugar make it harder for LDCs to reap the fruits of EBA, especially since these products are of much importance to the LDCs. It is likely that with the gradual abolition of the transitional measures, the EBA becomes more effective.

It would therefore be very interesting to know the actual impact of the EBA after the liberalisation of the transitional measures specifically on the agricultural sector. If the EBA does indeed foster ACP LDCs agricultural exports, this would confirm that generous preferential trade agreements like the EBA are beneficial for LDCs’ economic development. This thesis will therefore investigate the effects of the EBA on ACP LDCs’ agricultural exports. The research question reads: did the Everything But Arms initiative increase agricultural exports of African, Caribbean and Pacific Least Developed Countries to the European Union between 2001 and 2013?

In order to estimate the effects of the EBA initiative, an augmented gravity model will be constructed in order to analyse the total value of agricultural exports from ACP countries to the EU. The choice for ACP countries has been made for various reasons. According to Stevens and Kennan (2001), the EBA initiative will improve access to the EU market for ACP LDCs, and to a much smaller extent for non-ACP LDCs. This difference is due to the fact that, prior to the implementation of the

(5)

5

EBA, the ACP LDCs were far more important suppliers of the affected items than non-ACP LDCs (Stevens and Kennan, 2001: 9). This means that the effects of the EBA are most likely to be observed in ACP countries. Another advantage of the choice for ACP countries follows from the observation that ACP countries have a very similar export structure and are direct competitors in some industries (Gradeva and Martínez-Zarzoso, 2011: 11). ACP countries are therefore highly comparable. Roughly half of the ACP countries are eligible for the EBA, which means that the other half provides an excellent counterfactual sample. Furthermore, ACP countries were subject to very similar import tariffs prior to the introduction of the EBA, which makes it more likely that any observed changes after the introduction of the EBA are indeed caused by the EBA arrangement.

This thesis follows the method employed by Gradeva and Martínez-Zarzoso, which is described in more detail in the literature review. However, contrary to Gradeva and Martínez-Zarzoso, this thesis will treat the value of agricultural exports rather than total exports as the dependent variable. The reason for this choice is that the literature suggests that an improvement in exports is most likely to occur in the agricultural sector. This thesis also extends their period of analysis. Where Gradeva and Martínez-Zarzoso analyse the time period of 1995 to 2005, this thesis uses data from 1995 to 2013. The reason for choosing 1995 as a starting point is simply because the EU-15 did not exist before 1995. Also, in that year, the ACP countries in the analysis were already facing the same tariffs and regulation under the Lomé convention. The reason for not including 2014 is the major GSP reform that has taken place since 1 January 2014. A more elaborate explanation of these choices is given later in this thesis. The data thus represents the total value of EU-15 imports of all agricultural products from individual ACP states in millions of euros between 1995 and 2013.

Subsequently a fixed effects panel data regression will be used to estimate the effects, using the natural logarithm of the total value of agricultural exports of individual ACP countries to the EU as the dependent variable. The independent variables are the natural logarithm of GDP, the natural logarithm of population, a dummy variable indicating whether a country is an LDC and two dummy variables indicating eligibility for the EBA for the period of 2001 to 2007 and for the period of 2008 to 2013 respectively. The choice of two separate dummies follows from the observation that the EU drastically changed its trade relationship with ACP countries in 2008. This method is explained in more detail in the methodological part of this thesis. The use of fixed effects estimation allows the comparison of a large amount of different countries since it accounts for country specific time-invariant characteristics like distance, language and culture. This greatly simplifies the augmented gravity model of trade. After running this regression, the same model is extended with controls for

(6)

6

time fixed effects and country specific time trends. In the end this model should provide an answer to the research question of this thesis.

To understand the context in which EBA preferences are granted, chapter 2 of this thesis is devoted to a review of the trade policy of the European Union. Subsequently the focus turns to previous studies on or related to the EBA initiative in chapter 3. Chapter 4 discusses the methodology of this study. Chapter 5 presents the analysis and the empirical results of the augmented gravity model. Finally the results of this analysis are used in chapter 6 to answer the research question, after which the limitations of the results are discussed in chapter 7.

(7)

7

2. The trade policy of the European Union

This chapter reviews the trade policy of the EU. When reviewing the trade policy of the EU, it is vital to have a precise definition of the concepts of LDC and ACP countries. Therefore, the first section of this chapter elaborates on these concepts. Subsequently, this chapter briefly reviews the basis of the EU trade policy and the various preferential trade regimes offered by the EU. This knowledge is crucial in order to place the EBA arrangement in the right perspective and in order to understand the design of the quantitative analysis. Then the focus turns on the EBA arrangement and the trade regimes that are applicable to the ACP countries.

2.1 Definitions

The United Nations (UN) recognises Least Developed Countries (LDCs) as a category of states that are deemed highly disadvantaged in their development process. These countries face the risk of remaining in a situation of underdevelopment and deeper poverty. Additionally, these countries are highly vulnerable to external economic shocks, natural and man-made disasters and diseases. As such, LDCs are in need of attention from the international community (UNCTAD, 2015a). There are currently 48 countries that are classified as a LDC. The list of LDCs is reviewed every three years by the United Nations Economic and Social Council. The criteria that are used to determine LDC status are gross national income per capita, human assets such as indicators of nutrition, health, school enrolment and literacy, and economic vulnerability. Benefits of an LDC status include development financing, a preferential position in the trading system and technical assistance (UNCTAD, 2015b). The complete UN list of Least Developed Countries can be found in appendix 1.

ACP countries are countries that belong to the African, Caribbean and Pacific group of States. This group is composed of 48 states from Sub-Saharan Africa, 16 states from the Caribbean and 15 states from the Pacific region. All of these countries, except for Cuba, are signatories of the Cotonou Agreement which binds them to the European Union. The main objectives of the ACP group include, among others, the sustainable development of its member states and their integration into the world economy, and coordination with regards to the implementation of the Partnership Agreements with the European Union (ACP Secretariat, 2015). More information on the trade relationship between the ACP countries and the European Union is given in section 2.5. The complete list of ACP countries can be found in appendix 2.

(8)

8

2.2 The basis of the EU trade policy

The EU manages trade and investment relations with non-EU countries through the EU’s trade and investment policy. Trade policy is an exclusive power of the EU, which means that individual EU member states cannot legislate on trade matters or conclude international trade agreements (European Commission, 2015a). The EU’s trade policy is established in articles 206 and 207 of the Treaty on the Functioning of the European Union. In article 206 of this treaty, it is stated that:

By establishing a customs union in accordance with Articles 28 to 32, the Union shall contribute, in the common interest, to the harmonious development of world trade, the progressive abolition of restrictions on international trade and on foreign direct investment, and the lowering of customs and other barriers (European Union, 2012: 93).

Article 207.1 reads:

The common commercial policy shall be based on uniform principles, particularly with regard to changes in tariff rates, the conclusion of tariff and trade agreements relating to trade in goods and services, and the commercial aspects of intellectual property, foreign direct investment, the achievement of uniformity in measures of liberalisation, export policy and measures to protect trade such as those to be taken in the event of dumping or subsidies. The common commercial policy shall be conducted in the context of the principles and objectives of the Union's external action (European Union, 2012: 94).

It is important to note that the member states of the EU all impose the same tariff rates on imports. However, these tariffs vary for different non-EU countries. As a member of the World Trade Organisation (WTO), the EU is obliged to apply the Most Favoured Nation (MFN) treatment. The requirement for MFN treatment means that countries cannot discriminate between their trading partners. If some country is granted a special favour, the same favour has to be granted to all WTO members. This means that the EU cannot lower its import tariffs for a specific country or group of countries without lowering the same tariff by the same amount for all WTO members. The MFN treatment is included in the General Agreement on Tariffs and Trade, the General Agreement on Trade in Services and the Agreement on Trade-Related Aspects of Intellectual Property Rights (WTO, 2015). However, some exceptions to MFN treatment are allowed. For example, countries can set up a free trade agreement that applies only to goods traded within the group. This allows the EU to apply different tariffs to different countries and allows the EU to operate as a customs union. Countries are also allowed to give developing countries special access to the home market under the

(9)

9

Enabling Clause, which enables developed WTO members to give differential and more favourable treatment to developing countries (WTO, 2015a) (WTO, 2015b).

These exceptions lead to a large variety of different import tariffs for products destined for the EU from different countries. By the end of 2013, there were free trade agreements in place between the EU and Central America1_{, Colombia and Peru, South Korea, Mexico, South Africa and Chile. Economic}

Partnership Agreements were being implemented between the EU and the CARIFORUM states2, Papua New Guinea, Zimbabwe, Mauritius and the Seychelles. Since free trade agreements are a core component of many Association Agreements as well as Customs Unions, the EU also has free trade deals in force with many countries and territories in Europe and the Southern Mediterranean3. Overall, in 2013 the EU had trade agreements in place with some 50 countries. On top of that, there are many negotiations concerning new free trade agreements that are either still on-going or finished but not yet applied (European Commission, 2013: 1-7).

Since the variety in import regulations is so large, it is hardly possibly to compose an exhaustive overview of all of the EU’s import tariffs for all products. However, such an overview is not necessary for this thesis; only the import regulations that apply to ACP’s agricultural exports are needed to answer the research question. The trade regimes applicable to ACP countries are the standard Generalised Scheme of Preferences, the GSP+, the Lomé / Cotonou conventions, the Economic Partnership Agreements, the Market Access Regulation and the Everything But Arms initiative. All of these regimes are outlined in the following sections. These trade regimes are first described in general, and then with a particular focus on the implications of these arrangements on agricultural exports of ACP countries in the relevant time period. The various trade regimes are summarised in a timeline on page 16 of this thesis.

2.3 The Generalised Scheme of Preferences

The EU’s Generalised Scheme of Preferences (GSP) was introduced in 1971. The aim of the GSP is to help developing countries worldwide by making it easier to export their goods to the EU by reducing

1

This group consists of Costa Rica, El Salvador, Guatemala, Honduras, Nicaragua and Panama.

2

These are Antigua and Barbuda, Bahamas, Barbados, Belize, Dominica, Grenada, Guyana, Haiti, Jamaica, Montserrat, Saint Kitts and Nevis, Saint Lucia, Saint Vincent and the Grenadines, Suriname and Trinidad and Tobago.

3

These are Andorra, San Marino, Turkey, Faroe Islands, Norway, Iceland, Switzerland, the former Yugoslav Republic of Macedonia, Albania, Montenegro, Bosnia and Herzegovina, Serbia, Algeria, Egypt, Israel, Jordan, Lebanon, Morocco, Palestinian Authority, Syria and Tunisia.

(10)

10

tariffs on these goods. Through the additional export revenue which is generated, the GSP fosters growth in their income and supports economic growth and job creation (European Commission, 2014c: 2). The GSP is subject to WTO law; the so-called Enabling Clause allows for an exception to the WTO’s MFN principle. The GSP provides a sliding scale of preferences according to different needs, namely the standard arrangement, the GSP+ and the EBA initiative (European Commission, 2014c: 2-3).

2.3.1 The standard Generalised Scheme of Preferences and GSP+

Until December 2013, the standard GSP reduced duties on approximately 66% of all EU tariff lines to 111 countries and territories. Another 25% of all EU tariff lines were already at 0% normal MFN duty in 2013. In total, only 9% of all tariff lines are outside the GSP (European Commission, 2014c: 11). However, the tariff lines outside the GSP were mostly agricultural products. The agricultural products outside GSP face MFN tariffs, provided that the exporting country does not have a more favourable trade agreement with the EU than standard GSP. In 2012, eligible countries exported goods worth €40.7 billion under this scheme, which is 70.2% of all EU imports benefiting from GSP preferences (European Parliament, 2015). In theory, all ACP countries were eligible for standard GSP in the time period this thesis focuses on. However, the rest of this chapter will show that with Lomé / Cotonou and EBA, all ACP countries had more beneficial trade agreements at their disposal until the end of 2007. We can therefore assume that standard GSP preferences were not utilised by ACP countries until the end of that year. From 2008 onwards, several ACP countries had to revert to standard GSP since they were no longer eligible for more beneficial trade agreements, while other ACP countries could profit from favourable agreements like EPAs and the MAR. This division is described later in this chapter.

Introduced in 2005, the GSP+ is a special incentive arrangement for sustainable development and good governance. GSP+ has provided for zero duties on the approximately 66% of all tariff lines for which reduced duties apply under the standard GSP. GSP+ is only available for developing countries considered to be vulnerable. In order to be eligible for this scheme, the country must have ratified 27 core international conventions in the fields of human and labour right, the environment and good governance. In 2012, the 15 countries that then qualified exported goods worth €4.9 billion to the EU, which represented 8.5% of all EU imports under GSP preferences (European Parliament, 2015) (European Commission, 2014c: 3-5). None of the ACP countries were beneficiaries of the GSP+ scheme in the analysed time span.

(11)

11

Effective from 1 January 2014, the GSP has seen major reform. The standard GSP and the GSP+ have remained in place but have become more focused. The number of GSP beneficiaries has been reduced from 177 to 88. The countries that no longer benefit from GSP have either another preferential market access agreement that is equivalent to or better than GSP, or are high or upper-middle income economies that no longer need unilateral preferences such as the GSP to successfully trade with the rest of the world. Also, there have been some changes regarding the graduation of competitive sectors. These sectors are considered to be competitive at the highest international level and thus do not need preferences. Third, some tariff lines have been added to the new standard GSP or GSP+ scheme (European Commission, 2014c: 7-11). Although the EBA has not been changed, the changes to the standard GSP and GSP+ have strengthened the EBA arrangement. By reducing the GSP to fewer beneficiaries and increasing graduation of competitive sectors, the competitive pressures from other developing countries on LDCs’ exports have been reduced, which provides LDCs significantly more opportunity to export (European Commission, 2014c: 13). It is important to note that these changes are not relevant for the present analysis; the GSP reform affects the tariffs and regulation for certain ACP states only and thus compromises the comparability of ACP states. The easiest way to solve this problem is simply to exclude this year from the analysis. Therefore, this thesis does not look at the years after 2013. However, these changes should be taken into account to make any reasonable prediction about the effectiveness of the EBA in the future.

2.3.2 Everything But Arms

The EU’s Everything But Arms arrangement was implemented in 2001 to give all LDCs full duty-free and quota-free access to the European market for their export with the exception of arms and armaments. This arrangement is the most generous form of preferential treatment to LDCs globally. Entry into the EBA is automatic and, unlike other GSP arrangements, has no time limit (European Commission, 2014a: 1). When the EBA arrangement entered into force in 2001, transitional provisions were installed for bananas, rice and sugar, which allowed for a duty-free import of these products under a gradually expanding quota regime. For bananas, the quota regime ended already in 2006. The restrictions for sugar and rice had lapsed completely at the end of August and September 2009 respectively. From October 2009 onwards, the EBA thus grants full duty-free and quota-free market access for all products from all LDCs, except for arms and ammunitions (European Commission, 2009: 1).

The EBA regulation was adopted on February 26th 2001 and is outlined in EC Regulation 416/2001 (UNCTAD, 2011). This regulation builds on Regulation (EC) No 2820/98, in which a multiannual scheme of generalised tariff preferences was applied for the period of 1 July 1999 to 31 December

(12)

12

2001 providing for more favourable tariff treatment for LDCs. Article 6 of this regulation was replaced by the following:

1. Without prejudice to paragraphs 2 to 4, Common Customs Tariff duties on all products of Chapters 1 to 974, except those of Chapter 935, originating in the least developed developing countries listed in Annex IV, shall be entirely suspended.

2. Common Customs Tariff duties on the products of CN code 0803 00 196 shall be reduced by 20 % annually starting on 1 January 2002. They shall be entirely suspended as from 1 January 2006.

3. Common Customs Tariff duties on the products of tariff heading 10067 shall be reduced by 20 % on 1 September 2006, by 50 % on 1 September 2007 and by 80 % on 1 September 2008. They shall be entirely suspended as from 1 September 2009.

4. Common Customs Tariff duties on the products of tariff heading 17018 shall be reduced by 20 % on 1 July 2006, by 50 % on 1 July 2007 and by 80 % on 1 July 2008. They shall be entirely suspended as from 1 July 2009.

(…)

This regulation applied from 5 March 2001 onwards (European Commission, 2001: 1-3). The transitional measures on rice and sugar both ended one month later than determined in EC Regulation 416/2001.

All countries identified by the UN as LDCs benefit from the EBA. If a country moves up the development ladder, it is no longer considered ‘least developed’ by the UN and EBA preferences are no longer required. However, to mitigate possible trade flow shocks, a transition period of three

4

Chapters 1 to 97 consist all products.

5

Chapter 93 includes arms and ammunition.

6_{CN code 0803 00 19 refers to bananas.} 7

Products of tariff heading 1006 refers to rice.

8

(13)

13

years is guaranteed after losing LDC status, during which EBA preferences still apply (European Commission, 2014a: 1).

2.4 The trade regimes for ACP countries

Until 2000, the economic cooperation between ACP countries and the EU was determined by the Lomé Convention. The main characteristics of the Lomé Convention were the partnership principle, the contractual nature of the relationship and the combination of aid, trade and political aspects. The first Lomé Convention, which was signed in 1975, included non-reciprocal preferences for most exports from ACP countries to the European Economic Community. Lomé I, II and III had a duration of five years, and the introduction of Lomé II and III did not bring any major changes to the Lomé Convention. In 1990, Lomé IV was introduced, which covered a period of 10 years, although it was revised after 5 years (European Commission, 2005).

The Cotonou agreement was introduced as a follow-up agreement of the Lomé Convention. This agreement, signed in June 2000, concluded a partnership between the EU and ACP countries for a 20-year period from 2000 to 2020 and has been the framework for the EU’s relationship with the 79 ACP countries since. The Cotonou agreement was designed to establish a comprehensive partnership based on development cooperation, political cooperation and economic and trade cooperation (European Commission, 2015b). However, the unilateral trade preferences granted under the Lomé and Cotonou agreements were not compatible with the WTO’s MFN treatment rules; the Enabling Clause requires that tariff preferences granted by developed countries must not discriminate between developing countries. Since the Lomé and Cotonou agreements foresaw in specific preferences for limited groups of developing countries, these agreements are not covered by the Enabling Clause (FAO, 2016). The EU’s preferential trade regime therefore used to operate under waivers to the MFN rules. The last waiver for the preferential EU – ACP trade relationship was extended to 31 December 2007. Given this deadline and the obligation to replace the waiver with WTO-consistent regional trade agreements, the EU and ACP countries started negotiations on Economic Partnership Agreements (EPAs) in 2002. From 1 January 2008 onwards, the economic relationship between the EU and ACP countries has been governed by these EPAs, which do adhere to the WTO’s MFN rules (European Parliament, 2014).

The basic principle of the Lomé Convention and the Cotonou agreement was that all products originating in the ACP states should be imported into the Community free of custom duties and charges having equivalent effects. However, ACP countries were granted only restricted access to EU markets regulated by the Common Agricultural Policy (CAP). This resulted in a broad classification of

(14)

14

agricultural products into three categories. The first category contained products that originated in ACP countries and were not produced within the EU, which could be imported free of duty. In the second category were products imported from ACP countries and covered by the CAP, which received a product-specific low preference margin. Finally there were four products which were covered by specific regulations, namely bananas, beef, sugar and rum. In the sugar protocol, the EU undertook to purchase and import at guaranteed prices 1.3 million tons of cane sugar originating from specific ACP states9 and India per year for an indefinite period. The banana protocol provided that, in respect of banana exports to the EU market, the preferential access of ACP states would be maintained until the expiry of the Lomé convention. Under the beef protocol, certain ACP states were allowed to export specified quantities of boneless meat to the EU at import duties reduced by 90%. Rum was subject to annual quotas within which rum could be imported duty-free, but this protocol was eliminated in the Cotonou agreement. The specific regulations on bananas, sugar and beef were terminated with the expiry of the Cotonou trade preferences in 2007 (European Commission, 2014b: 10).

Prior to the implementation of the EPAs, it was decided that negotiations would take place with seven regional groupings of ACP states (European Commission, 2014b: 10). Although the EPAs have to be compatible with WTO rules, the trade liberalisations are asymmetric. The EU provides duty-free and quota-free access for all exports from ACP countries, while the ACP countries are allowed to exempt up to 20% of EU exports to their domestic market from tariff reductions. At the same time, ACP countries are allowed to phase in their tariff reductions over periods of up to 25 years.

By the end of 2007, no EPA had yet been agreed on. The European Commission therefore decided to split the negotiations in two stages. The first stage would be the conclusion of interim EPAs involving a free trade area on goods alone between the EU and ACP countries in order to prevent a loss of market access after 2007, followed by further negotiations towards comprehensive EPAs to be concluded at the regional level (European Commission, 2014b: 13). Alongside the EPAs, the EU introduced the Market Access Regulation (MAR), which ensures that ACP countries continue to benefit from free access to the EU market while they are signing and ratifying the EPAs. For countries whose trade regime under the Cotonou agreement had expired, the MAR was crucial to avoid trade disruptions (European Parliament, 2014). The MAR is described in Council Regulation (EC) No. 1528/2007. This regulation states:

9

The countries benefiting from the sugar protocol were Congo-Brazzaville, Cote d’Ivoire, Kenya, Madagascar, Malawi, Mauritius, Swaziland, Tanzania, Uganda, Zambia, Zimbabwe, Barbados, Belize, Guyana, Jamaica, St. Kitts & Nevis, Suriname, Trinidad & Tobago and Fiji (Matthews, 2008: 32).

(15)

15

1. Subject to Articles 6, 7 and 8, import duties shall be eliminated on all products of Chapters 1 to 97 but not 9310 of the Harmonised System originating in a region or state listed in Annex I. Such elimination shall be subject to the transitional safeguard and surveillance mechanisms set out in Articles 9 and 10 and the general safeguards mechanism provided for in Articles 11 to 22 (European Commission, 2007a: 3).

The MAR thus provides duty-free and quota-free access to the EU market for all products, except arms and ammunition. However, this regulation also arranges transitional measures for rice and sugar, which are the same as those under the EBA. Since the transitional measures for bananas had already lapsed under EBA, these were also excluded from the MAR. This means that as of 2008, all countries covered by the MAR could export their agricultural products under the same favourable conditions as offered under EBA. However, this regulation was only applicable to those 35 ACP countries which had initiated EPAs in 200711. The remaining ACP states could only profit from GSP or the EBA from 2008 onwards. The MAR was eventually extended until 1 October 2014. Only countries that at least ratified or applied their interim EPAs by that date retained duty-free and quota-free access to the EU market, unless they are LDCs and thus benefit from the EBA. Otherwise, the GSP applies (European Commission, 2014b: 14-15) (European Commission, 2015c).

2.5 Summary of the EU trade policy for ACP countries

We can draw some interesting conclusions from the various trade agreements that have been outlined above. First, we can conclude that the group of ACP countries is indeed an appropriate group to study the effects of the EBA. Not only are about half of all ACP countries LDCs and are the ACP countries direct competitors of each other, but they are also very comparable in terms of participation in various trade agreements. All ACP countries except Cuba are signatories of the Cotonou agreement (European Commission, 2014b: 11). Also, none of the ACP countries are beneficiaries of the GSP+ scheme. Disregarding sugar, bananas and beef, which were regulated under special protocols, this means that all ACP countries were facing the same tariffs and regulation when exporting to the EU until the introduction of the EBA under the Lomé and Cotonou agreements.

10_{Chapters 1 to 97 include all products. Chapter 93 comprises arms and ammunition.} 11

In January 1st 2008, the countries to which the MAR applied were Antigua and Barbuda, the Bahamas, Barbados, Belize, Botswana, Burundi, Cameroon, The Comoros, Cote d’Ivoire, Dominica, the Dominican Republic, Fiji, Ghana, Grenada, Guyana, Haiti, Jamaica, Kenya, Lesotho, Madagascar, Mauritius, Mozambique, Namibia, Papua New Guinea, Rwanda, St. Kitts and Nevis, St. Lucia, St. Vincent and the Grenadines, the Seychelles, Suriname, Swaziland, Tanzania, Trinidad and Tobago, Uganda and Zimbabwe (European Commission, 2007a: 11). Zambia was added to this list in 2008.

(16)

16

Countries that were not eligible for the EBA were all facing the same tariffs and regulations until the expiry of the Cotonou agreement at the end of 2007, except for sugar, bananas and beef. After the expiry of the Cotonou agreement, the special protocols for sugar, bananas and beef were abandoned. The countries that did not yet sign an interim EPA and were not covered under the MAR lost their ACP preferences after 2007. This means that from 2008 onwards, there were two groups of non-LDC ACP countries. The first group could export duty-free and quota-free to the EU under the MAR or an EPA, and the second group was subject to normal GSP preferences. These normal GSP preferences have been available to all ACP countries in the entire period of study, but prior to 2008 all ACP countries had more beneficial agreements at their disposal.

The equality in tariffs and trade regulation means that it is possible to make use of a highly accurate counterfactual sample when analysing the effects of the EBA. Before 2008, the counterfactual sample simply consists of all ACP non-LDCs. After 2008, the counterfactual sample itself is divided in two groups, namely those ACP non-LDCs that did sign either an interim EPA or were covered in the MAR and those ACP non-LDCs that could only benefit from the ordinary GSP. Focusing on agricultural products specifically, this means that before 2008, the counterfactual sample only had restricted access to the European market. After 2008, one group of ACP non-LDCs could export their agricultural products duty-free and quota-free or under the same transitional measures as under EBA, while other group faced more restricted access to the European market than before. The various trade agreements that were available for ACP countries in the analysed time period are summarised in the following time line:

Finally, it is crucial to note that after the expiry of the Cotonou trade agreement, the effectiveness of the EBA for ACP LDCs could be completely different. Under the EPAs and MAR, some non-LDC countries can also enjoy duty-free and quota-free access to the EU while the rules of origin and administrative requirements are less strict than under the EBA. This deteriorates the competitive position of ACP LDCs. On the other hand, there is a group of ACP countries that no longer enjoys ACP preferences, which increases the competitive position of ACP LDCs. Although it is hard to reach

1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013

Lomé / Cotonou

Everything But Arms Standard Generalised Scheme of Preferences

(17)

17

conclusions about the effects of the EBA after 2008 up front, we can conclude that the effects of the EBA for ACP LDCs are likely to be different before and after 2008. Although the EBA itself was not changed in 2008, the expiry of the Cotonou agreement and the introduction of the EPAs and the MAR should be taken into account. Therefore, in the analysis, the EBA until 2008 and the EBA after 2008 should be treated as if they were different arrangements. This approach avoids the possibility that the results are obscured by the fact that the EBA has different effects before and after 2008.

(18)

18

3. Literature review

The empirical research on the effects of the EBA so far can broadly be divided into two categories. The first category uses computable general equilibrium models to quantify the effect of implementing the EBA initiative. Usually such models are employed to forecast the future impact of given policies on the exports and welfare of developing countries and the EU. Evenett (2008) reviewed different computable general equilibrium based studies on EU preferences and their welfare effects, most of which were related to the EBA initiative. He found that:

Most of the estimates of the gains to LDCs from this initiative ranged between USD 300-400 million. This initiative was found to have cost the EU between USD 200-300 million. These studies consistently showed that the welfare of the non-LDC developing countries, including those who receive EU GSP preferences but are not eligible for the EBA initiative, decreased (although the amount varied considerably across countries). The EBA initiative redistributes welfare as well as creating it in LDCs. Many of the estimates reported that the net effect on world welfare of the EBA initiative is almost nil. Evenett (2008) concluded from this that this initiative is largely about redistributing income from EU citizens and non-beneficiary developing country exports to the LDCs (European Commission, 2014b: 44).

However, the major limitation of computable general equilibrium models is that they only estimate the likely effects of a policy, without using actually observed data. This method is therefore not appropriate for evaluating the actual effects of the EBA. In the second category there are studies that use the gravity model of trade to estimate the effect of the EBA initiative on bilateral trade flows. This method uses ex post data and applies econometric techniques to assess the impact of the EBA initiative (Gradeva and Martínez-Zarzoso, 2011: 8-9) (European Commission, 2014b: 46). Since ex post econometric studies are much more useful to answering the research question of this thesis, the literature review focuses on these studies. This thesis builds on the work of Gradeva and Martínez-Zarzoso. Therefore, their work is discussed extensively in the next section. The focus then turns to other empirical studies on the impact of the EBA on the exports of developing countries.

3.1 Graveda and Martínez-Zarzoso

In the article ‘Are Trade Preferences more Effective than Aid in Supporting Exports? Evidence from the Everything But Arms Preference Scheme’, Gradeva and Martínez-Zarzoso (2011) examine the effects of the EBA trade preference regime and official development assistance on exports from ACP countries to the EU. In addition to the separate effect of each of these tools with respect to an

(19)

19

improvement of the recipient country’s exports, the combined impact of the EBA and aid flows is also examined. With this aim, an augmented gravity equation is estimated for bilateral trade flows from the 79 ACP countries to the EU-15 for the time period between 1995 and 2005 using panel data techniques (Gradeva and Martínez-Zarzoso, 2011: 1-2).

Gradeva and Martínez-Zarzoso first discuss the possible transmission mechanisms through which aid and trade preferences could affect export growth in the recipient countries. The authors state that the idea of trade preferences as a development tool is based on the theory of export-led growth. Opening the domestic market of the recipient country to the world markets should support an effective reallocation of the factors of production towards the sectors and industries exploiting the comparative advantages of a particular country. Stronger competition also boosts the innovation process. For these reasons, it can be expected that countries which are more engaged in international trade experience bigger incentives for faster growth and higher welfare. The gains from trade in comparison to aid come especially from spillovers and dynamic benefits which are induced through higher integration into world markets. It should be mentioned that non-reciprocal trade preferences exclude one of the major sources of benefits, namely liberalisation within the country itself through the opening of its own market to the rest of the world (Gradeva and Martínez-Zarzoso, 2011: 5).

The authors continue with explaining the EBA initiative. In this part, they make some very interesting remarks. First, the authors claim that the major advantage of the EBA in comparison with other preferential schemes of the EU is the unlimited time period of its implementation. Due to the fact that the arrangement is not subject to periodical renewal, it offers higher certainty for exporters from LDCs and facilitates investment decisions (Gradeva and Martínez-Zarzoso, 2011: 6). Second, the authors note that the EBA initiative may lead to potential losses for non-LDCs developing countries because of the relative preference for LDCs. Especially for non-LDC ACP countries, this negative effect is assumed to be significant since all ACP countries compete in the same industry (Gradeva and Martínez-Zarzoso, 2011: 6-7). Nevertheless, the export share of the 919 products that were liberalised with the introduction of EBA was only 0.03% of total LDCs’ exports to the EU. Bananas, rice and sugar, the three products with delayed liberalisation, had an export share of 0.47% of total LDCs’ exports to the EU. Therefore the newly liberalised products were not of much relevance to the LDCs, at least in the first year of implementation of the EBA. In 2008, total imports entering duty-free under the EBA regime reached €5.8 billion, which represents only 23% of total imports into the EU from LDCs. This indicates that the utilisation rate of the EBA was quite low (Gradeva and Martínez-Zarzoso, 2011: 7).

(20)

20

The authors also mention that until 2008, it could have been more profitable for ACP LDCs to export under the Cotonou agreement than under the EBA scheme due to different rules of origin and administrative requirements, which are much stricter under the EBA scheme. For the ACP LDCs, exporting under the EBA leads to additional documentation, new rules of origin and other regulation about cumulation. The authors also briefly mention the supply capacity and the potential to adjust the production structure of recipient countries as a reason for potential underutilisation of the EBA (Gradeva and Martínez-Zarzoso, 2011: 8).

The authors then turn to their empirical estimation of the effect of the EBA initiative. Of the 79 ACP countries considered in the gravity equation, 41 were considered LDCs during this time period. The authors focus on ACP countries because of the similarity of their export structure and the fact that they are direct competitors, and also because they already had greater preferential access to the EU than non-ACP developing countries before the introduction of the EBA scheme (Gradeva and Martínez-Zarzoso, 2011: 11). The authors then introduce the gravity equation in its log-linear form:

The exact specification of this model is given in appendix 3.

Using fixed effects estimation, the authors find that the GDP of the exporting country has a significant impact on trade flows. However, the GDP of the importing country and the population in both the exporting and the importing country have an insignificant effect. The fixed effects regression does not provide any estimation for the impact of the distance between the trading countries, the effect of a colonial link, or the impact of being either landlocked or an island as exporting country; these variables are country specific time-invariant dummies and cannot be estimated with a fixed effects regression. More surprising is the highly significant and negative impact of the EBA dummy on exports in all regressions, irrespective of whether official development assistance (ODA) is included or not. The interaction term between the EBA and ODA turns out to be a positive and significant determinant of exports of ACP LDCs to the EU-15. Development assistance by itself turns out to be insignificant (Gradeva and Martínez-Zarzoso, 2011: 14).

The authors continue by estimating a Hausman-Taylor model, which allows for some explanatory variables to be correlated with the random term. The main advantage of this method with respect to

(21)

21

the fixed effects estimation is that it also estimates the coefficients of the time-invariant variables. The same model is also estimated with fixed country-pair effects and without the time-invariant dummies. However, the results concerning the variables of interest are unchanged (Gradeva and Martínez-Zarzoso, 2011: 14-15). Since a log-linear form of the gravity equation drops out all zero bilateral flows, the authors also use a Heckman model to avoid a possible sample selection bias. In this model, official development assistance is significant. However, the EBA dummy still shows a strong negative effect on exports (Gradeva and Martínez-Zarzoso, 2011: 15-17).

The EBA agreement thus seems to decrease the value of exports in the absence of additional aid. This outcome appears to be very surprising. However, it should be noted that this thesis is only concerned with agricultural exports, for which the results can differ to that of total exports due to the higher initial tariffs. Additionally, this study was performed prior to the full liberalisation of the trade in rice, sugar and bananas. The authors mention that after this liberalisation, the EBA agreement might be more effective (Gradeva and Martínez-Zarzoso, 2011: 20). The authors also point at the more flexible rules of origin under the Cotonou agreement. Using the EBA preferences requires new rules of origin and regulations, which are complex and demand some time to be introduced. It is therefore possible that the number of ACP LDCs utilising the EBA scheme will increase in the long run (Gradeva and Martínez-Zarzoso, 2011: 20).

3.2 Other empirical studies on the impact of EBA on developing countries

Prior to the implementation of the EBA, Stevens and Kennan (2001) assessed the likely impact of the EBA on the exports of different groups of countries. Based on an analysis of the market structure of the affected goods and countries, the authors find that the direct effects on LDCs will be positive, although the absolute impact will be small given these countries’ limited supply capacity. The EBA should increase market access for ACP LDCs, and to a much smaller extent for non-ACP LDCs. However, industrialised countries, other ACP states and other GSP beneficiaries will face increased competition (Stevens and Kennan, 2001: 1).

The authors note that the EBA will only affect LDCs’ trade for products on which they paid an import tax in the EU prior to the EBA, and for which they have a supply capacity. These products are mostly beef, cheese, maize, bananas, rice and sugar. Although this is only a small number of products, the potential competitive advantage for LDCs is substantial since the import regime to other suppliers is highly protectionist. The non-ACP LDCs will see the greatest improvement in market access, since the status quo before the EBA was more favourable for ACP countries (Stevens and Kennan, 2001: 5-6).

(22)

22

However, in 2001 the ACP LDCs were far more important suppliers of the liberalised products than the non-ACP LDCs (Stevens and Kennan, 2001: 9).

The effects of the EBA for ACP countries is more complicated, since some but not all ACP countries already had preferential access for the products that have been liberalised under EBA prior to the introduction of EBA. In the case of sugar, only those countries that were either party to the Cotonou Sugar Protocol or the Special Preferential Sugar Agreement, in which the EU imported a variable quantity of cane sugar from specific ACP states at a reduced tariff and a price related to the EU domestic price, could benefit. For the beneficiaries of the sugar protocol, the benefits of the EBA were unclear. The sugar protocol guaranteed a high price, which would probably have been much lower under the EBA. However, contrary to the sugar protocol, the EBA provided quota-free access (Stevens and Kennan, 2001: 6-7, 14). With respect to the effects of the EBA on non-LDCs, the authors find that the only case in which the EBA could have had significant adverse implications for other developing countries is over sugar. This is because the EU market was over-supplied, so that any increase in LDC exports will tend to displace exports from another source, either directly or indirectly, if it results in more EU sugar being dumped on the world market (Stevens and Kennan, 2001: 17).

Another interesting study on the effects of the EBA is performed by Gamberoni (2007). In this study the impact of the EU unilateral trade preferences on both the intensive and extensive margin of trade is analysed using a tobit and probit estimation (Gamberoni, 2007: 1). Gamberoni finds that the ACP and EBA preferences decrease trade by 11% and 19% respectively, conditional on trade being present. These numbers reflect the impact of preferences on the average value of exports from beneficiary countries (Gamberoni, 2007: 16). However, after trying another specification for the preference dummies, the author finds that the negative impact of the EBA seems to be driven by ACP LDCs suggesting that LDCs that enjoy ACP preferences tend to use the ACP program rather than the EBA. This finding seems to confirm the hypothesis that the EBA is underutilised due to strict rules of origin, high administrative costs and non-tariff barriers (Gamberoni, 2007: 17-18). Gamberoni also finds that trade preferences for ACP countries seem to have led to an anti-diversification effect when all possible trade products are considered. ACP preferences appear to lead to a concentration of exports in the agricultural sector, where ACP preferences have the highest impact (Gamberoni, 2007: 22-23).

Aiello and Cardamone (2011) assessed the effectiveness of the EBA initiative on the LDCs exports of cloves, vanilla beans, coffee, crustaceans and molluscs over the period of 1995-2006. The results of

(23)

23

this study contrasted with previous studies which found that the EBA was not effective in increasing EU imports from LDCs. The paper showed that the EBA enhanced LDCs’ exports to the EU of crustaceans and vanilla. No conclusion could be drawn when considering cloves, coffee and molluscs. It should be noted that Aiello and Cardamone considered imports at commodity level and not total trade. The fact that the data in this study is highly disaggregated could be the reason that the results of this study differ from the findings in most other studies on the EBA (Aiello and Cardamone, 2011: 148-149).

Pishbahar and Huchet-Bourdon (2008) measured the impact of eleven regional trade agreements on European agricultural imports with an expanded gravity model. The results indicate that most regional trade agreements support the agricultural exports of developing countries to the EU, but for the EBA initiative the results are different. The authors find that the market access of LDCs in the EU agricultural market decreased from 3% to 2.7% during the period of estimation, which is 2000 to 2004 (Pishbahar and Huchet-Bourdon, 2008: 118). The estimations also show that the EBA initiative has had negative effects on the agricultural exports of eligible countries to the EU. Rigorous rules of origin, rigid cumulation rules, substantial processing and transport regulations, sanitary and phytosanitary regulations and the uncertain stability of preferences could be causing these negative effects. Nevertheless, the main explanation for these negative effects is that the Cotonou agreement and the EBA were overlapping during the period of this study. Since the Cotonou agreement imposes less administrative constraints, eligible countries exported under this agreement rather than under the EBA arrangement (Pishbahar and Huchet-Bourdon, 2008: 120).

In 2014, the European Commission published the report ‘Evaluation of Preferential Agricultural Trade Regimes, in particular the Economic Partnership Agreements (EPAs)’. This evaluation assesses the impact of EU preferential trade agreements and arrangements on the development of agricultural trade in the ACP region countries for the period of 1990 to 2012. Since the report is very comprehensive, only the parts that are directly relevant for studying the EBA arrangement are discussed here. When reviewing the agricultural trade between the EU and ACP LDCs, it is found that the group of ACP LDCs exhibits varying but deteriorating performance in the period of 1990 to 2004 and considerably improved performance after 2004, which is the period of the application of the EBA and MAR. It thus appears that the period when more generous EU preferences were applied has coincided with an increase of the share of agricultural exports of ACP LDCs in total agrifood imports of the EU from all ACP countries. However, this does not necessarily mean that there is a causal link (European Commission, 2014b: 24).

(24)

24

To get more insight in the effects of EU preferences, the authors use a Constant Market Share Analysis (CMSA) and a gravity model econometric estimation. Since there is no causal mechanism in the CMSA technique, CMSA is only useful as a descriptive tool (European Commission, 2014b: 53). The gravity model employed is based on dummy variables for different preferential trade agreements. The model includes amongst others dummies as to whether the country was eligible for Lomé 1 to 3, Lomé 4 to Cotonou, EPAs, Standard GSP, GSP+, LDC between 1997 and 2000 and EBA. The data includes 125 agro-food product lines for the time period of 1962 to 2012 (European Commission, 2014b: 64-65). The importing countries consist of all EU-27 countries, included from the year of their accession. In the first approach, the sample of exporting countries consists of all available world exporters for which there is consistent data. The counterfactual sample is thus represented by all developed and developing countries that are not in preference relationships with the EU, but which export to EU countries and, therefore, are trade competitors. In the second approach, only middle and low income countries are included, as well as countries that have had access to an EU trade preference system (European Commission, 2014b: 64-65).

The authors find that, on average, being a member of an ACP/GSP/LDC agreement with the EU has implied a growth in agri-food trade flows of about 45% in comparison with developed and developing countries without PTAs (European Commission, 2014b: 91). However, when looking specifically at the EBA, the effect turns out to be positive but not significant. It appears that, on average, the EBA initiative has not created any relevant positive agri-food trade effect for the countries involved (European Commission, 2014b: 80). The more complex rules of origin of the EBA and the fact that many LDCs already benefitted from zero duty access to the EU within ACP preferences are mentioned as possible drivers of this result. The authors also note that the impact of the EBA on the export diversification in their agri-good trade with the EU has been negative (European Commission, 2014b: 91).

3.3 Summary of the literature review

Some interesting assumptions can be made about the effectiveness of the EBA based on these studies. Strict rules of origin, sanitary regulations, administrative requirements and the supply capacity of the LDCs were often mentioned as barriers that make it harder for the LDCs to reap the fruits of the EBA. Also, the transitional measures that were installed were on exactly those goods that could potentially bring the most benefits to the LDCs. Given the fact that these measures were gradually abolished and the observation that implementing new regulations demands some time, it can be expected that the effectiveness of the EBA increased over time. Especially the expectations of the duty-free and quota-free export of sugar are quite high; since the tariffs on sugar were reduced

(25)

25

with 20%, 50% and 80% in 2006, 2007 and 2008 respectively and fully abolished in 2009, an increase in the effectiveness of the EBA should be observed during these years.

Also, focusing this thesis on examining agricultural exports rather than total exports will likely lead to more accurate results. This is the case because the Lomé and Cotonou agreements already arranged duty-free and quota-free access to the EU market for all ACP exports that were not regulated under the CAP. On top of that, some recent literature suggests that the effects of the EBA should be analysed at a disaggregated level. Aiello and Cardamone (2011) find that this approach could alter the results of studies analysing the effectiveness of preferential trade agreements. In looking at agricultural exports specifically, this thesis makes use of such disaggregated data. Of course, it would be possible to disaggregate the data even further by analysing specific commodities rather than total agricultural exports. However, this would make it harder to generalise potential findings to the overall effectiveness of the EBA.

Most studies conclude that the effects of the EBA on exports of LDCs are negative (Gradeva and Martínez-Zarzoso, 2011) (Gamberoni, 2007) (Pishbahar and Huchet-Bourdon, 2008). Aiello and Cardamone (2011) find that the effects of the EBA on exports of certain commodities were positive. However, all these studies were performed over time periods in which the transitional measures were not yet fully abandoned. The only study reviewed in this thesis that also looked at the more recent effects of the EBA was the evaluation published by the European Commission. This study finds that the effect of the EBA on LDCs’ agricultural exports was insignificant over the time period of 2001 to 2012. However, this study also analyses the effects of the EBA on non-ACP LDCs and uses all other exporting countries as a counterfactual sample. The variety in export structures, the economic infrastructure and regulation among these countries is much larger. This means that exogenous changes are more likely to affect such a sample than a sample with relatively similar countries. Therefore, analysing only ACP countries could lead to different findings. Altogether, this thesis can contribute to the literature by analysing the effects of the EBA on the agricultural exports of ACP LDCs while taking the abolition of transitional measures and recent policy changes into account.

(26)

26

4. Methodology

The econometric estimations in this thesis are based on an augmented gravity model of trade. This thesis builds upon the model used by Gradeva and Martínez-Zarzoso, although the model is somewhat modified. The gravity model is based on the idea that trade is positively affected by the economic mass of the trading countries, which is gauged by their GDP and population, and negatively inﬂuenced by the geographical distance between them. These variables are considered to be the normal gravitational variables that determine trade in the absence of any other disturbances. The gravity model offers the opportunity to model deviations from normal patterns of trade. These deviations are captured by augmenting the model with all the factors that may hinder or promote bilateral trade flows (Aiello and Cardamone, 2011: 129). The existence of preferential trade agreements (PTAs) is such a factor, which means that it can be included in the gravity model to assess its impact on bilateral trade flows. Including PTAs in the gravity model requires a dummy variable, which is either 1 if a certain PTA is in place and 0 otherwise. The use of PTA dummies is simple and transparent and in line with the recent literature (European Commission, 2014b: 64). For this reason, this thesis makes use of dummy variables to assess the effects of the EBA.

This thesis analyses bilateral trade flows between ACP countries and the EU over multiple years. An appropriate tool for such an analysis is fixed effects estimation. The major advantage of fixed effects estimation is that it accounts for country specific time-invariant characteristics, which greatly simplifies the augmented gravity model of trade. Including variables like distance, language, culture, common borders and past colonial ties is no longer necessary, since these are examples of country specific time-invariant characteristics that fixed effects estimation already accounts for.

The data used in the econometric analysis is downloaded from the Eurostat database. The data represents the total value of EU-15 imports of all agricultural products from individual ACP states in millions of euros between 1995 and 2013. The reason for choosing 1995 as a starting point is simply because the EU-15 did not exist before 1995. Also, in that year, the ACP countries in the analysis were already facing the same tariffs and regulation under the Lomé convention. The reason for not including 2014 is the major GSP reform that has taken place since 1 January 2014. This GSP reform affects the tariffs and regulation for certain ACP states only and thus compromises the comparability of ACP states. The easiest way to solve this problem is simply to exclude this year from the analysis. The reason for choosing the EU-15 is to avoid problems due to the accession of new member states to the EU. Agricultural products are defined as all products in sections 0, 1, 2 and 4 minus divisions 27 and 28 of the Standard International Trade Classification, Revision 3 (WTO, 2010).

(27)

27

There are currently 79 ACP countries. Not all these countries are included in the analysis. Cuba is excluded because it did not sign the Cotonou agreement. The trade relationship between the EU and South Africa has been regulated under the Trade and Development Cooperation agreement (European Commission, 2014b: 18). South Africa has therefore been excluded as well. Since Timor-Leste joined the group of ACP countries in 2004, it has been excluded as well. Additionally, the Cook Islands, the Marshall Islands, the Federated States of Micronesia, Nauru, Niue and Palau have been excluded since they were not yet part of the Lomé Convention in 1995 (European Commission, 2014b: 12).

In total, there are 70 countries12 included in the analysis. Of these countries, 3913 were LDCs during the whole period of 1995 to 2013 and could therefore make use of the EBA. Additionally, Cape Verde was an LDC until 1 January 2008. It was allowed to benefit from the EBA arrangement until the end of 2010 (European Commission, 2014a: 2) (European Commission, 2007b).

The gravity model makes use of a dummy variable indicating whether the country is eligible for the EBA in a specific year or not. As mentioned, there is one dummy variable indicating eligibility for the EBA between 2001 and 2007 and one dummy variable indicating eligibility for the EBA from 2008 onwards. This method is used to take account of the policy changes in 2008.

12

These countries are Benin, Burkina Faso, Burundi, Cameroon, Central African Republic, Chad, Congo (Brazzaville), Congo (Kinshasa), Cote d’Ivoire, Gabon, Madagascar, Mali, Mauritania, Niger, Rwanda, Senegal, Somalia, Togo, Kenya, Tanzania, Uganda, the Bahamas, Barbados, Botswana, Ethiopia, Fiji, Gambia, Ghana, Grenada, Guinea, Guinea-Bissau, Guyana, Jamaica, Lesotho, Liberia, Malawi, Mauritius, Nigeria, Samoa, Sierra Leone, Sudan, Swaziland, Tonga, Trinidad and Tobago, Zambia, Cape Verde, Comoros, Djibouti, Dominica, Kiribati, Papua New Guinea, Saint Lucia, Sao Tome and Principe, Seychelles, Solomon Islands, Suriname, Tuvalu, Angola, Antigua and Barbuda, Belize, Dominican Republic, Mozambique, Saint Kitts and Nevis, Saint Vincent and the Grenadines, Vanuatu, Zimbabwe, Equatorial Guinea, Haiti, Eritrea and Namibia.

13

These countries are Angola, Benin, Burkina Faso, Burundi, Central African Republic, Chad, Comoros, Congo (Kinshasa), Djibouti, Equatorial Guinea, Eritrea, Ethiopia, Gambia, Guinea, Guinea-Bissau, Lesotho, Liberia, Madagascar, Malawi, Mali, Mauritania, Mozambique, Niger, Rwanda, Sao Tome and Principe, Senegal, Sierra Leone, Somalia, Sudan, Tanzania, Togo, Uganda, Zambia, Kiribati, Samoa, Solomon Islands, Tuvalu, Vanuatu and Haiti.

(28)

28

The augmented gravity model has a log-linear form and is given by (Model 1):

Where:

 lnYit stands for the natural logarithm of the total value of all agricultural exports from ACP country i to the EU-15 in year t

 αi is the country fixed effect, which captures country specific time-invariant factors like distance, language and culture

 LDCit is a dummy variable indicating whether country i was a LDC in year t

 lnGDPit is the natural logarithm of the gross domestic product of country i in year t

 lnPOPit represents the natural logarithm of the population size of country i in year t

 EBA1it is a dummy variable indicating eligibility for the EBA scheme from 2001 to 2007

 EBA2it is a dummy variable indicating eligibility for the EBA scheme from 2008 onwards

 εit is the error term, which is assumed to be independent and identically distributed

After analysing this model, the same regression is performed with an additional dummy variable to indicate if the country was eligible for the EBA in 2007 in model 2. In that year, the transitional measures were already partially lifted while the Cotonou agreement was still in place. If the effectiveness of the EBA indeed increased over time, the effect of the EBA in 2007 should be higher than the average effect of the EBA between 2001 and 2007. If this is the case, this variable should be positive and significant. Model 2 takes the following form:

Where:

 lnYit, αi, LDCit, lnGDPit, lnPOPit, EBA1it, EBA2it and εit have their usual interpretation

 EBA3it is an additional dummy variable indicating eligibility for the EBA scheme specifically in 2007

The fixed effects regression does control for omitted variables that vary between countries but are constant over time. However, there could be unobserved variables that vary over time and are

The effects of the everything but arms initiative on the agricultural exports of African, Caribbean and pacific least developed countries