UNIVERSITY OF GHANA THE EFFECTS OF AGOA AND EU-ACP TRADE PREFERENCES ON GHANA’S BILATERAL EXPORTS: A COMPARATIVE STUDY BY PRINCE BAAH (10277227) A THESIS SUBMITTED TO THE UNIVERSITY OF GHANA, LEGON IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE AWARD OF MASTER OF PHILOSOPHY (M.PHIL) DEGREE IN ECONOMICS JULY, 2015 University of Ghana http://ugspace.ug.edu.gh ii DECLARATION I, PRINCE BAAH, hereby declare that this thesis is the original research undertaken by me under the guidance of my supervisors. Neither the whole nor a part of this thesis has been presented for another degree elsewhere. ………………………………………… PRINCE BAAH (10277227) ……………………………………… DATE ……………………………….. …………………………………… DR. FESTUS EBO TURKSON DR. FRITZ AUGUSTINE GOCKEL (SUPERVISOR) (SUPERVISOR) ………………………………. ………………………………….. DATE DATE University of Ghana http://ugspace.ug.edu.gh iii ABSTRACT Ghana has been a recipient of several trade preferences fashioned to increase its export earnings and diversify its exports. Among these trade preferences are the notable EU- ACP and AGOA adopted by the EU and US respectively. The study draws motivation from the relative sparcity of literature that seeks to investigate the effect of these preferences on individual developing countries and the inconclusiveness of existing studies. The main objective of the study is to find out if the EU-ACP and AGOA trade preferences have had any effects on the total bilateral exports of Ghana. Estimating a gravity model augmented with measures of trade preference agreements and a multilateral resistance term, the study uses bilateral export flows and key gravity covariates from the CEP II, UN-COMTRADE and WDI on Ghana’s export destinations over the period 1960-2013. After taking into account the unobserved individual heterogeneity, controlling for the potential endogeneity of the trade preference dummies, and correcting for serial correlation and heteroskedasticiy, the study finds that EU-ACP and AGOA trade preferences have a negative and insignificant impact on Ghana’s total bilateral exports. The results therefore indicate the need for Ghana to focus on expanding its export base in order to significantly benefit from the two PTAs. It is also recommended that the EU and US expand the product coverage of their respective preference schemes to cover commodities in which Ghana has a comparative advantage. University of Ghana http://ugspace.ug.edu.gh iv DEDICATION To the Baah and Issahaka families in consolidation of our family ties. University of Ghana http://ugspace.ug.edu.gh v ACKNOWLEDGEMENT I thank the Almighty God for his mercies and abounding gift of knowledge that has seen me through the completion of this thesis. My profound gratitude goes to Dr. Festus Ebo Turkson and Dr. Fritz Gockel for their good supervision. My primary debt is to Mr and Mrs Oduro for their motivation and encouragement throughout this programme. For their affection, constructive suggestions, as well as moral support, I express my heartfelt appreciation to my mum and siblings (Mavis Baah, Louis Asare and Priscilla Martey). I am also indebted to many friends and colleagues from the economics department for their support. I acknowledge the valuable contribution made by Rebecca Addo, Deborah Amartey, Maxwell Adika, Abdallah Issahaka Wumpini, Vera Acheampong, George Acheampong, Iddrisu Rashid, Manan Moro, Salim Mumuni Bawumiah and Ibrahim Issahaka Dokurugu in helping to improve the study. I will never forget the meritorious contribution of my father, Mr. Mohammed Issahaka Salifu. University of Ghana http://ugspace.ug.edu.gh vi TABLE OF CONTENTS DECLARATION .......................................................................................................... ii ABSTRACT ................................................................................................................. iii DEDICATION ............................................................................................................. iv ACKNOWLEDGEMENT .............................................................................................v TABLE OF CONTENTS ............................................................................................ vi LIST OF FIGURES ..................................................................................................... ix LIST OF TABLES .........................................................................................................x LIST OF ABBREVIATIONS ...................................................................................... xi CHAPTER ONE INTRODUCTION 1 1.1 Introduction to International Trade ....................................................................1 1.2 Background of the Study ...................................................................................4 1.2.1 Ghana’s trade in Retrospect ........................................................................9 1.3 Problem Statement .......................................................................................... 12 1.4 Objectives of Study ......................................................................................... 14 1.5 Rationale of the Study ..................................................................................... 14 1.6 Organization of Study ...................................................................................... 16 CHAPTER TWO OVERVIEW OF EU AND US TRADE PREFERENCES 17 2.1 Introduction to Trade Preferences .................................................................... 17 2.1.1 Unilateral and Non-Reciprocal Trade Preferences..................................... 18 2.1.2 Trade Creation, Facilitation and Erosion ................................................... 20 2.2 The Scope and Depth of AGOA ...................................................................... 21 2.2.1 Membership ............................................................................................. 23 2.2.2 Product coverage ...................................................................................... 25 2.2.3 Rules of Origin ......................................................................................... 26 University of Ghana http://ugspace.ug.edu.gh vii 2.2.4 Preference margin .................................................................................... 29 2.2.5 Assessment of the Export Performance of SSA under AGOA ................... 30 2.2.6 Ghana in AGOA ....................................................................................... 32 2.3 The Scope of the EU-ACP Trade Agreement ................................................... 34 2.3.1 Historical Overview of the EU-ACP Trade Relationship........................... 35 CHAPTER THREE LITERATURE REVIEW 42 3.1 Introduction ..................................................................................................... 42 3.2 Theoretical Review .......................................................................................... 42 3.2.1 Models of International trade .................................................................... 42 3.2.2 Theoretical Framework of Trade Preferences ............................................ 46 3.3 Empirical Review ............................................................................................ 48 3.4 Conclusion ...................................................................................................... 63 CHAPTER FOUR METHODOLOGY 65 4.1 Introduction ..................................................................................................... 65 4.2 Theoretical Foundations of the Gravity model ................................................. 65 4.3 Model Specification......................................................................................... 69 4.4 Data and Variables .......................................................................................... 76 4.4.1 Description of Variables ........................................................................... 76 4.5 Estimation Techniques .................................................................................... 79 4.6 Diagnostic tests ............................................................................................... 85 4.6.1 Serial Correlation and Heteroskedasticity ................................................. 85 4.7 Conclusion ...................................................................................................... 87 CHAPTER FIVE DATA ANALYSIS 88 5.1 Introduction ..................................................................................................... 88 5.2 Descriptive Analysis ........................................................................................ 88 University of Ghana http://ugspace.ug.edu.gh viii 5.3 Diagnostic Tests .............................................................................................. 90 5.3.1 Time Effects ............................................................................................. 90 5.3.2 Heterogeneity Bias ................................................................................... 90 5.3.3 Serial Correlation ..................................................................................... 91 5.3.4 Heteroskedasticity .................................................................................... 92 5.4 Empirical Results and Discussion .................................................................... 92 5.4.1 Economic Size .......................................................................................... 95 5.4.2 Distance ................................................................................................... 95 5.4.3 Control variables ...................................................................................... 96 5.4.4 EU-ACP ................................................................................................... 96 5.4.5 AGOA ...................................................................................................... 97 5.5 Conclusion ...................................................................................................... 98 CHAPTER SIX CONCLUSION AND RECOMMENDATIONS 99 6.1 Introduction ..................................................................................................... 99 6.2 Summary and Conclusion ................................................................................ 99 6.3 Recommendations ......................................................................................... 100 6.4 Delimitation of Study .................................................................................... 102 REFERENCES .......................................................................................................... 103 APPENDICES ........................................................................................................... 120 University of Ghana http://ugspace.ug.edu.gh ix LIST OF FIGURES Figure 1-1: Total goods, EU trade flows with Ghana (annual data 2004-2013) ............... 10 Figure 1-2: Ghana’s trade with the United States (2005 to 2012) ................................... 11 Figure 2-1: US imports from AGOA countries, 2000 - 2013 .......................................... 31 Figure 2-2: Ghana’s exports to the United States as a percentage of Ghana’s total exports (1996 to 2013) .....…………………………………………………33 Figure 2-3: EU Trade Preferences: The “Pyramid of Privilege” ..................................... 35 University of Ghana http://ugspace.ug.edu.gh x LIST OF TABLES Table 2-1: Breakdown of RTAs in force ........................................................................ 17 Table 2-2: EU-ACP agreements and their respective membership .................................. 39 Table 5-1: Summary Statistics of Dependent and Independent Variables, 1990 - 2013 ... 89 Table 5-2: Breusch and Pagan LM test for random effects ............................................. 91 Table 5-3: Wooldridge test for Serial Correlation in Panel data ...................................... 91 Table 5-4: Breusch-Pagan test for Heteroskedasticity .................................................... 92 Table 5-5: Estimates of gravity model using FE, RE and HT, 1990-2013. ...................... 94 University of Ghana http://ugspace.ug.edu.gh xi LIST OF ABBREVIATIONS ACP – African, Caribbean, and Pacific AGOA – African Growth and Opportunity Act ATPA – Andean Trade Preference Act CBERA – Caribbean Basin Economic Recovery Act CBTPA – Caribbean Trade Partnership Act DC – Developing Countries DFQF – Duty-free Quota-free DWH – Durbin-Wu-Hausman EBA – Everything But Arms ECOWAS – Economic Community of West African States EDF – European Development Fund EEC – European Economic Community EU – European Union FEM – Fixed Effect Model FTA – Free Trade Agreement GATT – General Agreements on Tariffs and Trade GDP – Gross Domestic Product GSP – Generalized System of Preferences HO – Heckscher-Ohlin HOPE – Haitian Opportunity through Partnership Encouragement HT – Hausman-Taylor LDC – Least Developed Countries LSDV – Least Square Dummy Variable MFA – Multifibre Agreement University of Ghana http://ugspace.ug.edu.gh xii MFN – Most Favoured Nation MRT – Multilateral Resistance Term NAFTA – North American Free Trade Act NRPTA – Non-Reciprocal Preferential Trade Agreement NTB – Non-Trade Barrier OECD – Organisation for Economic Co-operation and Development OLS – Ordinary Least Square PTA – Preferential Trade Agreement REM – Random Effect Model RoO – Rules of Origin RTA – Regional Trade Agreement SDT – Special and Differential Treatment SSA – Sub-Saharan Africa UN – United Nation UNCTAD – United Nations Conference on Trade and Development US – United States USITC – United States International Trade Commission WAMZ – West African Monetary Zone WTO – World Trade Organisation University of Ghana http://ugspace.ug.edu.gh 1 CHAPTER ONE INTRODUCTION 1.1 Introduction to International Trade An important catalyst for the development of countries in this dispensation is trade. Increased trade is widely believed to be one of the keys to alleviating poverty and improving standards of living in developing countries (Persson, 2012). Quite a number of studies have identified a positive association between economic growth and trade growth (Pattillo et al., 2005; Olken and Jones, 2005). Olken and Jones (2005) estimated that the share of trade in income increases on average by 13 percentage points with growth in trade. In order to improve trade, there has been the gradual and conscious effort to remove most, if not all restrictions to trade, especially tariffs. According to Krueger (1999), global trade has been enhanced by the gradual removal of trade restrictions, after the second world war. Krueger (1999) estimated that global trade in the postwar era had grown at more than twice the rate of growth of real world GDP. The significance of trade has led to the proliferation of Regional Trade Agreements (RTAs). Regional trade agreements are largely enforced on grounds of cooperation among nations. By their nature, RTAs promote trade liberalisation by providing member countries with a competitive urge to liberalisation (Turkson, 2012). Empirical analyses have been focused on estimating the ex-post effect of RTAs on exports and economic growth (Krueger, 1999). The performance of RTAs can be said to have influenced the adoption of Preferential Trade Agreements (PTAs) by developed countries. PTA as defined by the World Trade Organization (WTO) refers to all unilateral trade preferences University of Ghana http://ugspace.ug.edu.gh 2 as well as non-reciprocal preferential schemes provided by developed countries for developing countries. Since the early 1970s, developed countries have provided developing countries with preferential market access through unilateral and non- reciprocal preference schemes (Persson, 2012). The first non-reciprocal preference scheme implemented was the General System of Preferences (GSP) which was aimed at providing preferential treatment to all developing countries. Some developed countries, however, have fashioned other preference programmes based on the GSP for specific groups of developing countries. The European Union (EU) and the United States (US), which are the two largest importers of goods from developing countries have several of such agreements in place (Nilsson, 2007). The EU and the US have been and continue to be the most important markets for developing countries (Hoekman et al., 2006). The two dominant preferential trade agreements implemented by the EU and US for African countries are the EU-ACP and the African Growth and Opportunity Act (AGOA) respectively. The EU-ACP is a preference program designed by the EU for African, Caribbean and Pacific (ACP) countries. AGOA also provides preferential access to US markets, to Sub-Saharan African (SSA) countries. Several studies have been conducted to examine the effects of existing preferential trade agreements on beneficiaries. Most of the studies have found positive effects of the PTAs (e.g. Agostino et al., 2007; Gil-Pareja et al, 2014; Mattoo et al., 2002). For instance, Agostino et al. (2007) estimated that all non-reciprocal preferential trade agreements granted by 8 major OECD countries (i.e EU15, USA, Japan, Norway, Switzerland, Australia, Canada and New Zealand) have a positive impact on total exports from developing countries. There are also studies that have found that preferential trade University of Ghana http://ugspace.ug.edu.gh 3 agreements, either exert negative impact, or do not have any significant effect on their beneficiaries (Nilsson, 2002; Özden and Reinhardt, 2004; Gamberoni, 2007; Cline, 2004). Some of the concerns raised by these studies have been that although tariffs are eliminated by preference schemes, there is the subtle implementation of other protectionary measures such as Rules of Origin (RoO), quality control and other bureaucratic trading obstacles.1 Others have also cited the inability of beneficiaries to diversify their exports as another reason that undermines the effectiveness of trade preferences. All preferential trade agreements cover some specific products. The restrictiveness in the coverage of trade preferences could be justified along the view shared by classical economists (mainly Adam Smith and David Ricardo) that, specialization in trade would result in providing trading partners with a comparative advantage in the commodity they specialize, as well as increase world output. Concerns have, however, been raised regarding the insecurity of certain specializing countries in trade, if the trade does not encourage the diversification of the product base of countries involved (see Prebisch, 1950; Hesse, 2006; Hausmann and Klinger, 2007). It is important to determine the extent to which specialisation has to be encouraged. That is, to what extent can developing countries depend solely on just the production of “traditional commodities”. Mold (2005) posits that the criticisms against unilateral trade are fundamentally premised on the argument that preferences might actually reduce the incentives to diversify: by increasing the potential rents from traditional exports. Mold (2005) argued that preferences could contribute to “locking in” developing countries. According to Hesse (2006), the dependence of countries on few export commodities exposes them to adverse external shocks, resulting in deteriorating terms of trade. 1 See Candau and Jean (2005); Brenton and Manchin (2003); Jayasuriya et al. (2006) for detailed analysis. University of Ghana http://ugspace.ug.edu.gh 4 Developing countries often suffer when the demand for their products is throttled. This could be because, they usually have relative advantage in the production and export of primary products (Osakwe, 2007). So, for trade to be beneficial to all parties, it should support the possibility of diversification in the export base of underdeveloped countries and not just concentrate on the specialization of a few volatile commodities. 1.2 Background of the Study The widespread significance of trade among nations precipitated the formation of a general trade agreement which was referred to as the General Agreement on Trade and Tariff (GATT). The GATT was initially signed by 23 countries in 1947 during the Geneva I round. The GATT later progressed through several rounds of trade negotiations before it was eventually replaced by the World Trade Organization (WTO) in 1995.2 These rounds were useful in having a general consensus on any necessary modifications to the existing trade agreements. Fundamental to the GATT was the concept of trade equity, which was clearly defined under the Most Favored Nation (MFN) Principle. The MFN (established under article I3 of the GATT) stipulates that countries must treat all trading partners equally. That is, countries must treat all imports from all trading partners without bias. This goes to establish that the tariff treatment on the imports from all trading partners should be uniform irrespective of the exporting country. The MFN principle was vehemently opposed by Least Developed Countries (LDCs) as well as Developing Countries (DCs) 2 See Appendix I for list of rounds of trade negotiations 3 Refers to the clause that spells out the MFN as forming the fundamental and universal basis for trade. University of Ghana http://ugspace.ug.edu.gh 5 on the premise that equity in trade was not properly presented by the tenets of the principle. That is, the provisions under the MFN defeated its purpose. They argued that equal treatment of economically unequal trading partners did not constitute equity in trade. They rather called for Special and Differential Treatments (SDT). It is on this ground that the premier preferential trade agreements that offer special and differential treatment to LDCs and DCs known as the Generalized System of Preferences (GSP) schemes were initiated. This was made possible after GATT parties agreed to a waiver of article 1 under the “Enabling Clause” (as part of the Tokyo Round of negotiations) in 1979. The underlying caveat of the enabling clause was that systems had to be “generalized, non-discriminatory and non-reciprocal” with respect to beneficiary countries. The amendment of the GATT recognized the special economic needs of developing countries. Most industrialized countries have developed preferential trade agreements under the GSP scheme for developing countries (Persson, 2012). The basic principle behind the GSP is to provide a wide range of goods originating from developing countries with preferential market access (usually in the form of lower tariff rates or duty-free status) to developed country markets in order to spur on economic growth (Gil-Pareja, et al, 2014). Many developing countries have been recipients of several trade preferences fashioned to increase their export earnings and to diversify their exports. According to Hoekman, et al. (2006), the rational for granting preferential market access emerged in the context of the arguments in favour of SDT and was premised on the argument that “developing countries needed to foster industrial capacity both to reduce import dependence and to diversify away from traditional commodities that were subject to long-term declining University of Ghana http://ugspace.ug.edu.gh 6 terms of trade (and often also affected by short-term price volatility)”. Developing countries view trade preferences as providing trade opportunities (Candau and Jean, 2005). Particularly, Sub-Saharan African countries seem to regard trade preferences probably because they believe that trade preferences improve the productive capacity of beneficiaries sufficiently enough to enable them withstand competition from producers from highly industrialized countries. Collier and Venables (2007) concluded that given the right conditions, it is possible for African countries to accelerate their export sector growth through trade preferences. From the perspective of the beneficiaries, trade preferences could be perceived as providing a level playing field for trading activities between the advanced nations and the underdeveloped nations (Supper, 2001). In an attempt to approximate the productivity gap, Chang (2002) employed the GDP per capita at purchasing power parities to estimate that, the difference between the richest European countries and the poorest African countries is likely to be in the order of 40-60 to one. Notwithstanding the acclaimed benefits and apparent popularity of trade preferences among beneficiaries, some studies in trade have found reason to doubt the efficacy of trade preferences (e.g. Mattoo et al, 2002 among others). Also, the non-reciprocal nature of some existing preferential trade agreements have been criticised by mainly non beneficiaries and trade experts whose arguments are grounded on the issue of trade discrimination and the enervation of multilateral liberalization. This is because, these non-reciprocal preferential trade agreements have distorted world trade flows and generate benefits and costs for both preferred (recipient) and excluded countries (Hoekman and Özden, 2005). Mold (2005) argued that under excessive Rules of Origin University of Ghana http://ugspace.ug.edu.gh 7 (RoO), the amount of trade diverted between two bilateral partners involved in preferential agreement is larger than the trade created as a result of the agreement. The criticisms against Special and Differential Treatment (SDT) have long been in existence since the adoption and implementation of the Generalized System of Preference. Patterson (1965) elaborated succinctly many of the criticisms repetitively raised against preferential trade. Johnson (1967) similarly, evaluated the possible impacts of preferential trade on donors and recipients as well as excluded countries. Some of the main criticisms outlined in both works included the possibility of fostering specialization in the production of preference granted commodities irrespective of the inherent comparative urge recipient countries may or may not have (Hoekman and Özden, 2005). This concern was quite widespread because in many instances, certain countries (often least developed) have been caught in the “trade trap” of wanting to exploit trade preferences even though their production capabilities suggest otherwise. The “trade trap” herein used refers to a state where preference beneficiary countries do not have comparative superiority in the production of the product they have been granted preferential access (either duty-free, quota-free or both). In such a situation, there is the temptation of forgoing the production of the commodity they have the production urge and rather concentrate on taking advantage of the offered preferential treatment by specializing in the products eligible for duty-free access (even though they may not necessarily be efficient in that production). Continuing with the criticisms against SDT, Johnson (1967) agreed with Patterson (1965) on the conflict potential of preferential agreements. Johnson (1967) reiterated the point that the existence of preferences could lead to the insurgence of political friction between beneficiaries and excluded countries; University of Ghana http://ugspace.ug.edu.gh 8 especially among developing countries who are at different stages of development (Hoekman and Özden, 2005). Moving forward, these criticisms are still valid in recent times. One important question confronting the WTO members is whether the basic principle of the GATT/WTO regime of non discrimination specified under the Most Favoured Nation (MFN) principle should continue to be hallowed in the face of policies that advocate active discrimination against some members (including many developing countries) in order to provide better market access to developing and least developed countries, thus “aid through trade” (Panagariya, 2003).4 Ghana, as many Sub-Saharan African countries, is a signatory to trade agreements; some reciprocal, others non reciprocal. The engagement of developing economies in trade preferences, according to (Crawford and Fiorentino, 2005) has stemmed from the desire to implement domestic trade reforms which are aimed at opening up their economies to competitive liberalization as well as facilitating the integration of their economies into the world trading system. Ghana belongs to regional trade agreements such as the Economic Community of West African States (ECOWAS), and the West African Monetary Zone (WAMZ). Among the trade agreements Ghana subscribes to, two main non-reciprocal ones have gained the attention of trade analysts since their inception. These non reciprocal preferential trade agreements are the African Growth and Opportunity Act (AGOA) and the EU-ACP trade agreements. AGOA is a non reciprocal trade agreement between the United States of America and some selected Sub-Saharan African countries. The EU-ACP agreement on the other hand, is an agreement between the European Union 4 Cited by Hoekman B. and Özden C. (2005) University of Ghana http://ugspace.ug.edu.gh 9 and the African, Caribbean and Pacific countries. Ghana has been a beneficiary under these two unilateral, non-reciprocal preferential trade agreements. 1.2.1 Ghana’s trade in Retrospect Ghana has maintained a cordial trade relationship with several developed countries. The major trading partners of Ghana include the Netherlands, UK, France, China, United States, and Nigeria. The major destinations of Ghana’s exports include South Africa, United Arab Emirates (UAE), Switzerland, France and Italy. Undoubtedly, EU members dominate Ghana’s export destinations. The exports of Ghana are mainly agricultural products and natural mineral resources. They include Cocoa, timber, gold, bauxite, and aluminium. Ghana imports mainly capital equipment, petroleum, and foodstuffs (Observatory of Economic Complexity).5 1.2.1.1 Trade with the European Union Historically, Europe has been Ghana’s largest trading partner. The volume of Ghana’s exports to the EU has increased steadily over the years. The annual average growth rate of the EU’s import from Ghana for the period 2009 – 2013 stands at approximately 32 percent. For instance, in 2013, Ghana’s exports to the EU (28 members) constituted about 48 percent of its total exports. Also, about 23 percent of Ghana’s total import value was sourced from the EU.6 5 http://atlas.media.mit.edu/profile/country/gha/ 6 The source of the data is Eurostat IMF but was retrieved from the European Commission, Directorate- General for Trade. University of Ghana http://ugspace.ug.edu.gh 10 Figure 1-1: Total goods, EU trade flows with Ghana (annual data 2004-2013) Source: Eurostat comext statistical regime 4 It is observed that the EU has been experiencing positive trade balance with Ghana throughout the period with the exception of 2011. The genesis of the preferential trade agreement between the EU and Ghana goes back to the Lomé Convention. Hence, Ghana has benefited from non-reciprocal market access to the EU since 1975. Ghana signed onto this agreement as part of the African, Caribbean and Pacific countries. This trade agreement between the EU and ACP countries has undergone several modifications up till its current state where it is being made WTO- compatible in what is referred to as the Economic Partnership Agreement (EPA). 1.2.1.2 Trade with the United States The United States is one of the major trading partners of Ghana. In 2013, two way trade between Ghana and the US totalled about $1.4 billion, with Ghana exporting and importing about $366 million and $1.1 billion worth of commodities respectively.7 The export profile of Ghana’s trade with the US is mainly characterized by primary products. For instance, in 2013, Ghana’s five largest export commodities to the US were cocoa, 7 www.ustr.gov/countries-regions/africa/west-africa/ghana University of Ghana http://ugspace.ug.edu.gh 11 mineral fuel, wood, rubber, and vegetables in that order. Also, Ghana’s imports were mainly vehicles, machinery, meat, and cereals. Ghana tends to be one of the top five African destinations for US products. On the regional level, US total trade with SSA amounted to $63.3 billion in 2013. Exports to the region from the US accounted for about 62 percent of the total trade between them. Whereas, imports to the US amounted to about $24 billion.8 Figure 1-2: Ghana’s trade with the United States (2005 to 2012) Data source: Trade Map Figure 1-2 above shows that, Ghana recorded trade deficits throughout the specified period. The exports of Ghana to the United States quite remarkably increased in 2011. Ghana is a beneficiary of the non-reciprocal preferential trade agreement designed by the United States for eligible countries in Sub-Saharan Africa referred to as the African Growth and Opportunity Act (AGOA). This trade agreement was designed to encourage beneficiaries’ trade with the United States. Before the enactment of AGOA, the trade 8 www.trade.gov/agoa/pdf/2013-sub-saharan-Africa-trade.pdf -1500000 -1000000 -500000 0 500000 1000000 1500000 2000000 2005 2006 2007 2008 2009 2010 2011 2012 exports imports balance University of Ghana http://ugspace.ug.edu.gh 12 preference designed by the US that was in force and predominantly used by SSA is the Generalized System of Preferences (GSP). 1.3 Problem Statement The increase in exports from developing countries to industrialized nations' markets has long been considered an essential element to reduce poverty, promote sustainable development and reap the potential benefits of globalization for the developing world (Gil-Pareja, et al, 2014). A good strategy to achieving this objective hinges on the support of developed countries in the form of Non-Reciprocal Preferential Trade Agreements (NRPTA). According to Agostino et al (2007), “a NRPTA is a concession granted by a developed country to developing countries on a unilateral basis, that is without reciprocal preferences for the donor’s exports”. Existing NRPTA include, but not limited to the GSP, EU-ACP, and AGOA. Developed countries’ trade preferences for developing countries have been one of the largest issues in North-South trade for the last century (Persson, 2012). Several researchers have sort to evaluate the impacts of existing NRPTA on preference receiving regions (Sapir, 1981; Nouve and Staatz, 2003; Lederman and Özden, 2004; Perssson and Wilhelmsson, 2007; Gil-Pareja, et al, 2014). However, not much has been done in the area of determining the effect of these NRPTA on individual beneficiaries. The two dominant NRPTAs in Sub-Saharan Africa are the EU-ACP agreement and the AGOA. Ghana has been a beneficiary of the non-reciprocal trade agreement between the EU and ACP since 1975. Although the EU-ACP agreement began with the Yaoundé agreement in University of Ghana http://ugspace.ug.edu.gh 13 1964, Ghana became a beneficiary from the Lomé conventions (beginning from 1975) and has remained a beneficiary through to the Cotonou agreement. Ghana has continued to initial onto the Economic Partnership Agreement (EPA). With respect to trading with the United States, Ghana continues to be a beneficiary under AGOA since its inception in 2000, with its pending review in 2015. With AGOA review pending, and the transformation of EU-ACP into a WTO compatible preference scheme (EPA), it has become imperative to establish a connection between these two non-reciprocal trade agreements and Ghana’s bilateral export performance over the years. Most of the previous studies have considered the effects these trade preferences have on the exports of regional blocs or group of countries. For instance, Brenton and Ikezuki (2004) found that in 2002, beneficiaries under AGOA benefited hugely considering the total exports of beneficiaries. Further investigation into the composition of exports, however, revealed that only two oil exporting countries (Gabon and Nigeria) contributed more than three-quarters of the total exports. With this, it becomes problematic to conclude that Ghana benefited from AGOA in 2002 on grounds that the effects of AGOA on beneficiaries in that year was positively significant as revealed by Brenton and Ikezuki (2004). Several studies affirm that the effects of trade preferences differ from one country to another, and that the effects on some countries are so strong to bias the regional effects (see Brenton and Ikezuki, 2004; Mattoo et al., 2003; Shapouri and Trueblood, 2003). The study, therefore, focuses on investigating the effects of the EU-ACP and AGOA agreements on Ghana’s bilateral exports. University of Ghana http://ugspace.ug.edu.gh 14 1.4 Objectives of Study The main objective among others is to investigate the effects of AGOA and EU-ACP trade preferences on Ghana’s bilateral exports. To achieve this objective, the study estimates the effects of EU-ACP agreements and compares it with the effects of AGOA on Ghana’s bilateral exports. In an effort to address the principal objective, the study seeks to provide answers to the following questions:  Has the EU-ACP trade agreement had any significant effect on Ghana’s bilateral exports?  Has the AGOA had any significant effect on Ghana’s bilateral exports?  Which of these trade preferences (AGOA or EUACP) has had the most significant effect on Ghana’s bilateral exports?  Do AGOA and EU-ACP have jointly significant effect on Ghana’s bilateral exports? 1.5 Rationale of the Study Against a backdrop of “aid fatigue”, it is widely believed that developing countries can benefit more from opportunities to increase their exports rather than depending heavily on foreign aid. Hence, the massive advocacy for enhanced trade for underdeveloped countries. This advocacy is captured in the popular slogan “trade, not aid” or “aid for trade”9 which was introduced at the first UN Conference on Trade and Development (UNCTAD). From the perspective of aid beneficiaries who now call for comprehensive 9 See Hoekman and Prowse (2005); Prowse (2010) for details on aid for trade. University of Ghana http://ugspace.ug.edu.gh 15 trade preferences, the rationale for their appeal and advocacy for preferential trade in place of aid is premised on the opinion that trade is beneficial both from the point of view of economic growth and poverty reduction. Indeed, because of its direct link to productive activities, it is often argued that trade is better than aid (Mold, 2005). To this end, it is commonly implied that market access agreements like the European Union’s EU-ACP agreement and AGOA by the United States, for developing countries would be of immense benefits compared to any anticipated benefits of aid. Existing literature that have estimated the ex-post effects of AGOA and EU-ACP have yielded mixed results. Also, existing studies generally consider the effects of AGOA and EU-ACP on beneficiaries as a whole without special considerations for the specific effects on individual beneficiaries. Recent Studies that have compared the effects of these preferences have also considered the relative effects of the trade preferences on their respective beneficiary groups (see Nilsson and Davies, 2013; Persson and Wilhelmsson, 2013). Owing to this gap in literature, the study estimates the effects of AGOA and EU- ACP on Ghana’s bilateral exports, compare their independent impacts and eventually form a basis for suggesting the direction of policy concentration when it comes to the prioritization of trade policy. Moving forward, it is worthwhile for Ghana to know the exact effects of the trade preferences on exports, instead of relying on conclusions drawn from the investigating the effects of the trade preferences on regional economic bloc. The conclusion of most literatures on the effect of both EU-ACP and AGOA on beneficiaries differs considerably among studies. Whereas some have identified positive University of Ghana http://ugspace.ug.edu.gh 16 trade creation/diversion effects10 others have observed otherwise. This brings to bear the fact that the effects of these trade agreements on their beneficiaries are not straightforward. As a result, using the results from studies where beneficiaries are aggregated to inform policy direction of Ghana may be misleading. This study adds to the existing literature by seeking to investigate the effects of AGOA and EU-ACP on Ghana’s bilateral exports. The study would then compare the relative effects of these two NRPTAs. Again, the study seeks to determine whether Ghana’s concurrent membership in these two NRPTAs has significantly affected Ghana’s bilateral exports. 1.6 Organization of Study The study is divided into six chapters. The first chapter deals with the introduction to the research. Following the introduction is the overview of the study as chapter two. This chapter discusses the main provisions of the AGOA and the EU-ACP agreements. Chapter three, reviews most of the existing literatures relating to the effect of trade preferences on exports. In chapter four, the study deals with methodological issues. Here, the tenets of the gravity model (which would form the primary basis of our empirical analysis) is perused and justification for its adoption provided accordingly. Other issues regarding data type, their sourcing, handling and modification are addressed in this chapter. The study then proceeds to present and interpret the statistical results of the empirical analysis in chapter five. Finally, in Chapter six, the conclusion of the results obtained are expounded and recommendations prescribed. 10 For example see Frazer and Van Biesebroeck (2010) University of Ghana http://ugspace.ug.edu.gh 17 CHAPTER TWO OVERVIEW OF EU AND US TRADE PREFERENCES 2.1 Introduction to Trade Preferences The first United Nation’s Conference on Trade and Development held in Geneva in 1964 was the breakthrough for the acceptance of trade preferences. The resolution of the conference prescribed that developed countries could grant trade preferences to all developing countries under the enabling clause. This led to the establishment of the General System of Preferences (GSP). Following the GSP, several Regional Trade Agreements (RTAs) have emerged, most of which are Free Trade Agreements (FTAs). As of June 2014, the WTO reported that the GATT/WTO had been notified of 585 RTAs, out of which 379 were in force. Table 2-1: Breakdown of RTAs in force Agreement Number Share Free Trade Agreement (FTA) 220 58% Economic Integration Agreement (EIA) 118 31% Customs Union (CU) 26 7% Partial Scope Agreement 15 4% Total 379 100% Source: WTO website Underdeveloped countries have been beneficiaries of several trade preferences fashioned to improve their export earnings. Trade preferences, refer to the considerate, special and differential treatment developed countries grant to underdeveloped countries University of Ghana http://ugspace.ug.edu.gh 18 (beneficiaries). The differential treatment usually comes in the form of lower or no tariffs on exports from these beneficiary nations. It is worth noting that, trade preferences do not only cover the reduction or elimination of tariff, but generally covers the removal of all sorts of trade barriers. It is in this light that Persson (2012) in defining trade preferences states that if a country grants trade preferences to another country, this simply refers to the fact that the latter country faces less restricted trade barriers than the donor country’s other trade partners. The existing agreements between and among nations, whether bilateral or multilateral could be classified as either Regional Trade Agreement (RTA) or Preferential Trade Agreement (PTA). The generic use of these terms interchangeably does not; in strict adherence to trade terminologies, conform to the operational definition provided by the WTO. According to the WTO, RTAs are defined as trade agreements concluded between countries not only belonging to the same geographical region. On the other hand, PTAs are used often to refer to all unilateral trade preferences as well as non-reciprocal preferential schemes.11 2.1.1 Unilateral and Non-Reciprocal Trade Preferences Trade preferences can be referred to as unilateral when they are designed by one country or group of countries without necessarily consulting the targeted beneficiaries. Generally, unilateral preferences in force are non-reciprocal in nature. That is, they do not require beneficiaries to provide duty-free quota-free access in return. Examples of non-reciprocal unilateral trade preferences include the GSP, AGOA, and EU-ACP agreements. 11 https://www.wto.org/english/tratop_e/region_e/rta_pta_e.htm University of Ghana http://ugspace.ug.edu.gh 19 One of the early concerns raised about unilateral trade preferences has been that they are structured and adopted by most developed countries to provide bias treatment for their previous colonies or political allies who are underdeveloped. Through this, unilateral trade preferences offered by advanced countries are used as a policy tool to discriminate among countries on a quid pro quo basis (Johnson, 1967). Another concern of unilateral trade preferences has to do with the frequency with which amendments are made to the granted unilateral trade preference. Persson (2013) in analyzing the effects of such short lived changes, concentrated on the firm level effects stating that unilateral changes at short notice create a great deal of uncertainty for firms potentially interested in making use of the preferential access. Persson (2013) argued that, unilateral preference schemes must provide stable preferential access for a considerable period of time, say ten years. In contrast to the unilateral agreements is the much prescribed multilateral trade preferences. These refer to trade agreements agreed on a much broader consensus. This involves the “meeting of minds” of all stakeholders involved in the trade agreement. There are some concerns raised regarding the possible weakening of multilateral liberalization in the presence of the proliferation of preferential trade agreements. That is, whether the spread of preferential trade agreements helps global trade liberalization. While some insist that participation in preferential trade agreements decreases the incentive to strive for multilateral liberalization, others are of the view that preferential trade would rather provide participating countries, especially beneficiaries the opportunity to prepare and compete on the global stage, hence, facilitating multilateral liberalization. University of Ghana http://ugspace.ug.edu.gh 20 2.1.2 Trade Creation, Facilitation and Erosion Trade creation occurs when domestic production in the preference-granting country is replaced by more efficient imports (Condon & Stern, 2011). Fundamental to the propagation and the pronouncement of trade preferences is the underscoring trade creation relevance. Trade creation is believed to be phenomenal when trade preferences are granted to countries whose presence in the global trade arena is quite insignificant. Trade preferences afford such countries the opportunity to expand their export capacity as the market for their exports is unconditionally provided given the creation of trade by the preference-giving countries. There are, some observed setbacks to the trade creation advantage of trade preferences. Most beneficiaries of trade preferences do not necessarily experience massive trade creation benefits as presented by the theory. The idealized objectives of most trade preferences with respect to trade creation have been hampered in many ways by certain mitigating factors which seem to empirically define the trade landscape. For instance, most beneficiaries of trade preferences are underdeveloped countries, whose exports are dominantly primary products as well as raw materials. These primary products are often subject to very low MFN tariffs. As a result, beneficiaries are not able to fully exploit the duty-free exemptions given to their exports under preferential trade agreements. This highlights the point that the trade creation argument in support for trade preferences could be greatly overemphasized and grossly exaggerated. The presence of very low MFN rates erodes the preferences. In cases where trade creation benefits may be profound, the preferences granted are highly guarded with strict Rules of Origin (RoO). Generally, the phenomenon that seems to mitigate against the achievement of trade creation in full potential is the preference erosion. University of Ghana http://ugspace.ug.edu.gh 21 2.2 The Scope and Depth of AGOA The United States in an effort to help bolster the export-led growth of underdeveloped countries, have over the years implemented both general and specific trade preferences. The long standing trade preference among all trade initiatives of the U.S. is the Generalised System of Preference (GSP) which has been in operation since 1974. The institution of the GSP program was in response to a general call for special and differential treatment by underdeveloped countries. Following the GSP, several regional trade preferences have been enforced to help improve the export capacity of underdeveloped countries. Hence, an exploitation of the waiver to the MFN principle that makes it possible for developed countries to consider the implementation of unilateral non-reciprocal trade preferences. One of such dominant trade preferences is the African Growth and Opportunity Act. AGOA is a non-reciprocal trade initiative designed and implemented by the United States for the 48 Sub-Saharan African countries intended to provide duty-free quota-free (DFQF) access to beneficiaries. AGOA does not only involve tariff considerations, but also makes it mandatory for a meeting of U.S and African government officials to be held annually to discuss trade and economic issues. This annual meeting is referred to as the AGOA Forum. AGOA was first authorized by the United States congress and signed into law on May 18, 2000 as Title 1 of the Trade and Development Act of 2000. Following the first authorization, several technical changes have been made to the agreement. In all, AGOA has been amended five times with its recent renewal due for expiration on September 30, University of Ghana http://ugspace.ug.edu.gh 22 2015 instead of the initial year of expiration (i.e. 2008). The amendments to AGOA have led to the adoption of various legislation in the implementation of the AGOA program. The AGOA legislations began with the AGOA I. On August 6, 2002, the first amendment to the AGOA was made and the AGOA II surfaced replacing the AGOA I. AGOA II expanded the preferential access for exports of beneficiary countries. Again, AGOA III came into force after President Bush signed into law the AGOA Acceleration Act of 2004 (Public Law 108-274) to replace the AGOA II in 2004. The main additions to the AGOA II included the extension of preferential access and the third country fabric provision to September 30, 2015 and 2007 respectively.12 In December 2006, President Bush signed into law the Africa Investment Incentive Act of 2006 (Public Law 109-432)13 popularly known as AGOA IV to replace AGOA III. Underpinning the establishment of AGOA is the primary objective of facilitating and promoting a stable and sustainable market-led economic growth and development in SSA. Also AGOA is intended to deepen the trade ties between US and the SSA region.14 AGOA is not the only regional trade program the U.S has aside the GSP. There are also the Caribbean Basin Economic Recovery Act (CBERA), the Caribbean Trade Partnership Act (CBTPA), the Andean Trade Preference Act (ATPA), and the Haitian Opportunity through Partnership Encouragement (HOPE) Act. However, among these preferential programs, AGOA has been described as the flagship of US commercial and development policy with Sub-Saharan Africa (Brenton and Hoppe, 2006). The office of the United States Trade Representative through the AGOA implementation guide published in 2010 12 Retrieved from www.trade.gov/index.asp 13 USITC (2014). 14 https://ustr.gov/issue-areas/trade-development/preference-programs/african-growth-and-opportunity-act- agoa University of Ghana http://ugspace.ug.edu.gh 23 identifies three important benefits of AGOA to eligible Sub-Saharan African exporters. To begin with, AGOA extends the duty-free treatment under the GSP program for eligible Sub-Saharan African countries. Penultimately, AGOA eliminates most of the limitations of the GSP program for eligible SSA countries, and finally, it expands the product coverage of the GSP program but only for products of SSA. 2.2.1 Membership AGOA is a trade program designed by the United States for only Sub-Saharan Africa (SSA). The SSA as defined under AGOA consists of 48 countries. The premier AGOA agreement was signed in 2000 with 32 SSA countries as beneficiaries. As of August 2014, there were 41 SSAs eligible for AGOA benefits. Currently, there are 39 SSA beneficiary countries under AGOA (West Africa - 15, East Africa - 14, Central Africa - 6, and Southern Africa – 4).15 2.2.1.1 Eligibility criteria An SSA country qualifies for AGOA considerations, if the country is determined by the president of the US to have met the criteria set out in section 104 of the AGOA act (Trade and Development Act, 2000) and section 502 of the GSP. The eligibility status of countries could be terminated if the president of the US determines or has any reason to suggest that an eligible Sub-Saharan African country is not making continual progress in meeting the eligibility requirements. The membership of AGOA has varied over the years since its inception in 2000. At its inception, 34 Sub- Saharan African countries were designated as eligible for trade benefits; including Ghana. 15 See Appendix II University of Ghana http://ugspace.ug.edu.gh 24 In 2001 and 2002, Swaziland and Cote D’Ivoire were designated as the 35th and 36th beneficiaries respectively. Shortly afterwards, on January 1, 2003, the Gambia and the Democratic Republic of Congo were designated as the 37th and 38th AGOA beneficiaries respectively, followed by Angola in 2004. Just as some SSA countries are gradually being considered for trade benefits under AGOA, others are also losing their eligibility status. For instance, on 31st December, 2003, Eritrea and the Central African Republic were removed as AGOA beneficiaries. Following in this fate were Mali, Madagascar, the Gambia, Guinea Bissau, Mauritania and Madagascar. Some of the countries have also regained beneficiary status after it is determined that they satisfy the requirements as stipulated in section 104 of the AGOA and section 502 of the 1974 Act. The Ivory Coast regained beneficiary status. Also, Guinea Bissau was reinstated on the 23rd December, 2014 in a proclamation by Barak Obama, the president of the United States. Mali, in a similar presidential proclamation also regained beneficiary status on January 2014. Presently, some SSA countries are considered ineligible for AGOA benefits. These include Zimbabwe and Sudan. Those who have their beneficiary status terminated include South Sudan, the Gambia, Swaziland, and the Democratic Republic of Congo. Appendix II shows a list of the countries eligible as at February, 2015. There are also special provisions for the consideration of a lesser developed beneficiary country. To be considered as a lesser developed beneficiary country, the gross national product per capita of the beneficiary country in 1998 should have been less than $1500 as computed by the World Bank (Williams, 2014). The exceptions so far to this LDC University of Ghana http://ugspace.ug.edu.gh 25 criteria are Namibia and Botswana, who although have a measured GNP per capita above the threshold are considered for LDC status. Lesser developed countries stand to benefit from additional benefits such as an extension of the product coverage of AGOA. 2.2.2 Product coverage In many cases, the number of products covered under preferential schemes is often used in the assessment and determination of the value of the preferential scheme involved. More products on offer will increase the value of a preferential scheme for the recipient country (Persson, 2013). This opinion, however, is not shared by most recent experts. AGOA provides duty-free quota-free access for the exports of eligible SSA countries to the United States. This preferential treatment covers all products covered under the US GSP scheme and proceeds to extend the product coverage onto other products which are not covered under the GSP. Essentially, AGOA extends preference on more than 4,600 products which are eligible under the US GSP program. In addition, the AGOA legislation adds a further 1,800 tariff lines. As at 2014, 1740 products are covered under AGOA.16 Tariff treatments of products are determined based on the import sensitivity of the products. A product that is highly import sensitive would undoubtedly be ineligible for preferential access. For instance, some key agricultural product lines such as cotton and sugar are not accorded duty-free access under AGOA (Leo and Ramachandran, 2014). AGOA covers some agricultural, petroleum and non agricultural commodities. The agricultural commodities include vegetables, fresh cut roses, and citrus products (either 16 See Brenton and Ikezuki (2004) and Dean and Wainio (2006) for detailed examination of the product coverage of AGOA. University of Ghana http://ugspace.ug.edu.gh 26 fresh or juice). Goods such as footwear, handbags, gloves, and watches fall under the non agricultural commodities category. One major product covered by the AGOA that informs high preference utilisation by some beneficiaries is the textiles and apparel. The coverage of this product is significant because most of the Sub-Saharan economies are major cotton producers (Shapouri and Trueblood, 2003). Apart from the products strictly covered under AGOA, least developed beneficiary countries qualify for duty-free treatment on additional 1,400 products. AGOA extends the product coverage under the GSP from about 4,600 items to more than 6,400 items (International Trade Administration). 2.2.3 Rules of Origin Rules of Origin (RoO) are stipulated procedures set by preference-giving countries used to determine whether or not a product qualifies to be considered under a trade agreement as originating from a beneficiary country. The RoO basically places restriction on which kind of product is considered as economically originating from a particular country. RoO are profound in unilateral non-reciprocal trade preferences. Preference granting countries use RoO as a means of differentiating between products transiting through beneficiary countries and products produced by beneficiaries. The objective behind the use of Rules of Origin as additional provisions in trade preference is to avoid trade deflection. Trade deflection refers to the situation where goods made in preference-excluded countries are routed through a preference-receiving country with no significant value added to the good. Without RoO, some preference- receiving countries will be used merely as transit zones by other countries in order to University of Ghana http://ugspace.ug.edu.gh 27 benefit from the trade preference. RoO are of immense benefits, especially when some of the inputs used in the production of products by preference-receiving countries are outsourced. This is an attempt to ensure that the envisaged benefits of the granted preference accrue solely to the beneficiaries involved. One of the main reasons used to justify strict Rules of Origin is that “they encourage the development of integrated production structures within developing countries to maximize the impact on employment and to ensure that it is not just too low value-added activities which are undertaken in the LDCs” (Brenton, 2003). It has however been argued to the contrary, that, the enforcement of strict RoO undermines the utilisation of preferences by beneficiaries (Mold, 2005). This is because, products from the preference-receiving nations may not qualify for duty-free access given the strict RoO. For instance, it is argued that increasing the value added requirement by beneficiaries (mainly LDCs) under trade preference schemes, is likely to create an underutilisation of the preference scheme rather than an exploitation. Almost all preference-receiving countries have inadequate capital requirement for production. As a result, they tend to rely on other countries for capital and other inputs for manufacturing. So implementing strict RoO which would prevent them from importing inputs from such sources would amount to reduced production. AGOA, has its unique provisions concerning the Rules of Origin. The RoO under AGOA can be broadly categorised as general rules and sector-specific rules. The general rules apply to all AGOA eligible products irrespective of sectors, and the specific ones are additions to the general rules that apply only to specific sectors.17 One profound 17 Refer to Brenton and Özden (2005) for details on US Rules of Origin University of Ghana http://ugspace.ug.edu.gh 28 advantage of AGOA over the GSP has been the non-restrictive rules of origin for apparel. Also, whereas apparel and textile articles are exempted from the GSP program, AGOA provides certain eligible Sub-Saharan beneficiary countries duty-free market access for these products. 2.2.3.1 Apparel RoO Apparel is one of the few commodities with which there exists much potential for significant expansion in exports. According to Brenton and Özden (2006), apparel and textile provide a unique opportunity since both the US and EU in the recent decade, have under the Multifibre Agreement (MFA) imposed strict protectionist measures on their apparel imports. They found that, the MFN tariffs that apply to apparel imports into the US and the EU are 20% and 12% respectively even though MFA quotas have completely been removed. This presents the fact that apparel could be viewed as one of the “cash cows” under the AGOA agreement. There is also a special provision made for sourcing fabrics and yarns from other sources (most efficient). This global sourcing of fabrics is captured in the AGOA as “the third country fabric rule”. The third country fabric rule is a special provision that allows the imports of apparel into the U.S from lesser developed SSA countries to qualify for duty- free treatment even if the fabrics and yarns used in their production are sourced from non-AGOA countries. For instance, under this provision, a least developed SSA country can import yarns and fabrics from China, use them in the production of apparel and still qualify to export to the U.S under AGOA. This third country fabric rule was set to be removed entirely in 2007 but was re-authorized in August 2012. Nonetheless, there are University of Ghana http://ugspace.ug.edu.gh 29 quantitative restrictions to this requirement. Currently, there are 24 Sub-Saharan African countries18 that qualify for the 3rd Country Fabric Rule.19 2.2.4 Preference margin This is the difference between the MFN tariff rates and the tariffs applicable under trade preferences. Preference margins are used to determine the value of trade preferences. According to Persson (2013), preference margin is one of the most important factors in determining the potential value of trade preferences. The higher the preference margin, the larger the value of the trade preference and hence, high preference utilisation. The preference margin could be expressed mathematically as; 20 Where refers to the preference margin on import commodity from preference beneficiary to preference-giving country at time . The calculation of preference margins for trade preferences appears to be challenging. This is because a comparison of the preferential rates with the MFN rates, although convenient, may yield overestimated results (Cipollina et al, 2013). Fugazza and Nicita (2011), Hoekman and Nicita (2011), and Carrère et al. (2010) in recent works, adopted a competition-adjusted margin proposed by Low et al. (2005), as their measure of preferential margin. This measure is calculated as the percentage point difference between the weighted average tariff rate applied to the rest of the world and the preferential rate applied to the beneficiary, where weights are represented by trade shares 18 They are Benin, Botswana, Burkina Faso, Cameroon, Cape Verde, Chad, Ethiopia, Ghana, Kenya, Lesotho, Liberia, Madagascar, Malawi, Mauritius, Mozambique, Namibia, Republic of Congo, Nigeria, Rwanda, Senegal, Sierra Leone, Tanzania, Uganda, and Zambia. 19 See http://agoa.info/about-agoa/apparel-rules-of-origin.html for details on apparel RoO 20 Adopted notations from Persson (2013) University of Ghana http://ugspace.ug.edu.gh 30 in the preference granting market (Low et al, 2005). Another approach used by Cipollina and Salvatici (2011) defined preference margin as the ratio between a reference tariff and the applied rates faced by each exporter. Brenton and Hoppe (2006) estimated that the average preferential margin under AGOA is 7.7 percent. Studies have shown that with the exception of apparel, the margin of preference of products covered by AGOA are low (see Van Grasstek, 2003; Dean and Wainio, 2006). 2.2.5 Assessment of the Export Performance of SSA under AGOA The zenith of the objectives underscoring the enactment of AGOA is to bolster the exports of beneficiary nations in SSA, and inherently promote the proliferation of democracy and strict adherence to the rule of law in SSA countries. The exports of SSA countries under AGOA represents a small share of U.S imports. There has been an observed steady increase in the value of exports from SSA beneficiaries to the United States. In 2000, the United States was the largest single country export market for total SSA exports, accounting for about 23 percent of total SSA exports. AGOA’s share of total SSA exports to the U.S increased from 29 percent in 2001 to 36 percent in 2002 (Mattoo et al, 2002). The United States International Trade Commission (USITC, 2014) reported that generally, the US imports from SSA countries are dominated by imports entering under AGOA. According to Bonarriva and Guth (2014), there has been an observed high preference utilisation of the AGOA over the period 2008 - 2013; with AGOA exports of beneficiaries contributing about a significant 70 percent of all imports from AGOA countries. For instance, the 2014 report by USITC shows that between 2001 University of Ghana http://ugspace.ug.edu.gh 31 and 2013, U.S imports under AGOA increased by about 10 percent per year, from $7.6 billion to $24.8 billion. The commodity composition of US imports from AGOA countries is predominantly crude petroleum. Petroleum trade largely accounts for the observed increase in the exports of AGOA beneficiaries into the U.S. The bulk of SSA’s exports to the US mainly comprises of oil and related products. Figure 2-1 below shows the relative contribution of crude petroleum and non-crude petroleum products to total US imports from AGOA countries. Figure 2-1: US imports from AGOA countries, 2000 - 2013 Source: AGOA: Trade and Investment Performance Overview, USITC (2014) The figure above shows that the imports of Crude petroleum after the year 2000 has been above the non-crude imports of the US. In 2002, it was estimated that Gabon and Nigeria accounted for more than three-quarters of trade covered by AGOA (Brenton and Ikezuki, 2004). This means that beneficiaries who do not export petroleum are less likely to benefit immensely from the increased US imports under AGOA. According to Leo and University of Ghana http://ugspace.ug.edu.gh 32 Ramachandran (2014), oil exporting African nations have consistently accounted for more than 90 percent of African exports to the US market. This raises concern for countries whose exports are dominated by non-petroleum products. 2.2.6 Ghana in AGOA Ghana has been a beneficiary of AGOA since its inception in 2000. Under AGOA, the US has imported quite a substantial amount of commodities from Ghana. The import bill of the US for consumption (excluding crude petroleum) from Ghana under AGOA, increased from $2,052,800 in 2010 to $72,731,100 in 2011. This import figure suffered a decline through 2012 to 2013 recording 16,988,100 and 2,811,000 respectively (USITC, 2014 p327). A decomposition of US imports under AGOA reveal that, US imports from Ghana are mainly energy related products, followed by agricultural products (AGOA official website: http://agoa.info/profiles/ghana.html). Ghana’s bilateral trade balance with the US has been negative throughout the period 2000–2012. However, Angola, a major oil exporting country in SSA has consistently experienced bilateral trade surpluses with the US. Also Botswana, a major exporter of diamond and apparel in SSA has observed significant trade surpluses with the US (see Appendix VII). Almost all of US imports from Botswana consists of textile and apparel.21 Again, comparing Ghana’s exports to the US before and after AGOA revealed that, the percentage of Ghana’s exports to the US to Ghana’s total exports began to take a decline right after it peaked after the enactment and implementation of AGOA. Prior to the AGOA, the proportion of Ghana’s total exports to the US was increasing. After the enactment of AGOA in the year 2000, the share of exports to the US plummeted to a 21 http://agoa.info/profiles/botswana.html University of Ghana http://ugspace.ug.edu.gh 33 record high in 2001. However, after recording this high share in 2001, the proportion of Ghana’s total exports to the us dropped suddenly in 2002 and since has been averagely declining steadily. This is shown in the Figure 2-2 below: Figure 2-2: Ghana’s exports to the United States as a percentage of Ghana’s total exports (1996 to 2013) Source: UN Comtrade Database (2015) From figure 2-2 above, between 1997 to 2000, the contribution of Ghana’s exports to the US to Ghana’s total exports was increasing. In 2001, this share recorded its peak at about 7%. This indicates that the initial effect of the AGOA on ghana’s exports to the united States was profound. However, right after this peak, the share fell drastically to about 3% in 2003. From 2003 till 2013, US share of Ghana’s total exports has been fairly declining with a period average of about 2.5 percent. 0 1 2 3 4 5 6 7 8 96 97 98 99 2000 1 3 4 5 6 7 8 9 10 11 12 13 P E R C E N T A G E YEAR University of Ghana http://ugspace.ug.edu.gh 34 2.3 The Scope of the EU-ACP Trade Agreement The European Union over the years has adopted the use of preferential trade agreements as a key strategy in assisting the developing world catch up with global trade trends and to reduce poverty (Turkson, 2012). The EU is by far the largest contributor to the proliferation of trade agreements worldwide with more than 50 RTAs notified to the WTO (Candau and Jean, 2005). The implementation of trade preferences by the EU dates back to about 30 decades ago through the inception of the GSP. Borne out of the traditional GSP are several non- reciprocal unilateral trade preferences for targeted regional blocs of developing and least developed nations. Most notable among these preferences are the Everything But Arms (EBA) initiative, the EU-ACP agreement, the GSP for LDCs, and the EU preferences to the Mediterranean. All EU preference schemes are fundamentally designed to improve trade with their beneficiaries. However, the potential usefulness of each trade preference differs in terms of quality (Persson and Wilhelmsson, 2007). Persson and Wilhelmsson (2007) in classifying selected EU preferences according to preference quality, used the preference margin, product coverage, rules of origin and safeguard clauses as the criteria for developing what they termed the “Pyramid of Privilege”. This is shown in Figure 2-3 below. University of Ghana http://ugspace.ug.edu.gh 35 The arrow in the diagram shows that the quality of preferences increases moving from the GSP through the Mediterranean preference to the ACP agreement. It can be inferred that firms in ACP countries would undoubtedly request for preferential treatment under the EU-ACP agreement, although they may be eligible for the GSP as well. Similarly, Mediterranean firms would opt for the Mediterranean preference ahead of the GSP.22 2.3.1 Historical Overview of the EU-ACP Trade Relationship The trade relationship between the group of African, Caribbean and Pacific (ACP) countries23 and the European Union finds its ancestry from the Treaty of Rome signed in 1957 to establish the European Economic Community (EEC).24 One of the rationales for the Rome Treaty was to enshrine the solidarity and commitment of the members of the EEC to their colonies and their overseas territories (Bartels, 2007). As part of the treaty of Rome, technical and financial aid was made available to ACP countries through the creation of the European Development Funds (EDFs) (Nolte, 2002). The first EDF lasted 22 A country will not be eligible for both the EU-ACP and the Mediterranean preference. Hence there is no overlapping of their membership. 23 The ACP group was founded in 1975 through the Georgetown Convention. 24 Laaksonen et al (2007) Source: Persson and Wilhelmsson (2007) Figure 2-3: EU Trade Preferences: The “Pyramid of Privilege” ACP Generalised System of Preferences Mediterranean Preferences University of Ghana http://ugspace.ug.edu.gh 36 for five years (1958 to 1963). Following this EDF, a much comprehensive trade arrangement that considered the migration of political colonies to independent states was necessary. This led to the establishment of the first Yaoundé convention in 1963. This convention, which lasted for six years, established financial, technical and trade cooperation between the EEC and 18 associated African States and Malgache (Madagascar) (Laaksonen et al., 2007). The Yaoundé I was replaced with the Yaoundé II convention, which also lasted for six years (1969 – 1975). In 1975, once the United Kingdom joined the European Community, other countries (mainly Commonwealth nations) were included in the arrangement (Haveman and Schatz, 2003). In effect, nine countries of the EEC concluded a trade agreement with 46 ACP countries in Togo. This arrangement was referred to as the Lomé I convention. The lomé convention has been considered as a highly innovative model of international cooperation (Laaksonen et al., 2007). According to Bartels (2007, p42), “The main flashpoint of the Lomé convention was the principle of reciprocity, which was initially defended by the EEC and a hard core of Yaoundé associates (especially Senegal) but strongly opposed by those joining the system. The opposition was successful: the EEC soon abandoned the principle and the 1975 Lomé convention enshrined the principle of non-reciprocal trade preferences”. The Lomé I introduced a system that sought to compensate ACP countries for shortfalls in their export earnings emanating from price fluctuations or fluctuations in the supply of commodities such as cocoa, coffee, tea and groundnuts. Successive amendments followed the 1975 convention. That is, the subsequent entry into force of Lomé II, III and IV conventions in 1980, 1985 and 1990 respectively. All these variants of the Lomé University of Ghana http://ugspace.ug.edu.gh 37 convention had their main policy fulcrum. Whereas the Lomé II introduced a system that was intended to assist ACP countries who are heavily dependent on mining for export, the Lomé III changed the policy focus from industrial development to self-sufficiency and food security with the final amendment laying much emphasis on the human right issues, democracy and governance. Within the Lomé IV convention, the scope of the EU- ACP agreement was widened to include human rights as an integral prerequisite for cooperation. It also made budgetary allocations for environmental consideration through the EDF budget. This was intended for the preservation of the vast rainforest in the ACP member states. The end of the Lomé conventions saw the inception of the Cotonou agreement which was signed in 2000 between the European Union (then 15 members) and 77 ACP countries.25 The objectives of the Cotonou Agreement as specified in the general provisions was to promote and expedite the economic, cultural and social development of the ACP states, with a view of contributing to peace and security and to promoting a stable and democratic political environment (European Commission, 2014). The establishment of the Cotonou Agreement was rooted on five main thematic areas;  A comprehensive political dimension  Participatory approaches (through institutional provisions)  A strengthened focus on development based poverty reduction  A new framework for economic and trade cooperation, and  A reform of financial cooperation 25 Heavily drawn from Laaksonen, et al. (2007) University of Ghana http://ugspace.ug.edu.gh 38 Other relevant facets of the Cotonou agreement included the respect for human rights and democratic principles, a special paradigm of anti-corruption, comprehensive stakeholder engagement in economic, social and institutional policy reforms which ought to be endorsed by the European Commission. Issues pertaining to development driven poverty alleviation agenda, intellectual property, and labour were extensively covered by the institutional requirements, for the consideration of beneficiary status in the Cotonou agreement. Aside the extension of the conventional period from 5 to 20 years, other key innovations marked the period of the Cotonou convention. These included the inclusion of the civil society and the private sector as new actors. With all these reforms enshrined in the convention, the fulcrum remained fostering cooperation in trade related areas such as competition policy, trade and environment through trade liberalisation between the parties involved.26 2.3.1.1 Membership The membership of the EU-ACP agreements have been progressively increasing with successive amendments to the agreements. For instance, migrating from Lomé I through to the Cotonou convention, 5, 2, 4, and 7 countries joined the successive conventions respectively. The initial Cotonou Convention comprised of 77 ACP countries (47 African countries, 15 Caribbean countries, 14 Pacific countries, and 1 CARIFORUM country) and the 15 member European Union. From 2000 up until 2013, about 13 other European countries have joined the EU. Also, Cuba and South Africa have joined the ACP 26 This section is mainly referenced from Panagariya (2002). University of Ghana http://ugspace.ug.edu.gh 39 countries, making them number 79. Table 2-2 below shows the ACP membership of the respective agreements and their year of entry into force. Table 2-2: EU-ACP agreements and their respective membership Agreement Signed Entry into Force Membership Treaty of Rome 1957 1958 Belgian, French and Italian colonies Yaoundé I 1963 1964 Benin, Burkina Faso, Burundi, Cameroon, Central African Republic, Chad, Congo (Rep), Congo DR, Ivory Coast, Gabon, Madagascar, Mali, Mauritania, Niger, Rwanda, Senegal, Somalia, Togo. Yaoundé II 1969 1971 Yaondé I plus Mauritius (1943) Arusha 1969 1971 Kenya, Tanzania, Uganda Lomé I 1975 1976 Original Parties: Yaoundé II and Arusha plus Bahamas, Barbados, Botswana, Equatorial Guinea, Ethiopia, Fiji, Gambia, Ghana, Grenada, Guinea, Guinea Bissau, Guyana, Jamaica, Lesotho, Liberia, Malawi, Nigeria, Samoa, Sierra Leone, Sudan, Swaziland, Tonga, Trinidad and Tobago, Zambia Acceding Parties: Cape Verde (1977), Comoros (1976), Djibouti (1978), Dominica (1979), Kiribati (1979), Papau New Guinea (1977), Sáo Tomé and Principe (1977), Seychelles (1976), Solomon Islands (1978), St. Lucia (1979), St Vincent and the Grenadines (1980), Surinam (1976), Tuvalu (1979) Lomé II 1979 1981 Lomé I plus Antigua and Barbuda (1982), Belize (1982), St. Kitts and Nevis (1984), Vanuatu (1981), Zimbabwe (1981) Lomé III 1984 1986 Lomé II plus Angola (1986), Mozambique (1986) Lomé IV 1989 1991 Lomé III plus Dominican Republic (1991), Eritrea (1993), Haiti (1991), Namibia (1990) Cotonou 2000 2000 Lome IV plus Cook Islands (2000), Micronesia (2000), Marshall Islands (2000), Nauru (2000), Niue (2000), Palau (2000), Timor Leste (2005) Source: Persson and Wilhelmsson (2013). University of Ghana http://ugspace.ug.edu.gh 40 2.3.1.2 Product Coverage The migration of commodities onto the EU preferences for ACP countries increased under the Lomé IV. Under Lomé IV, nearly all the products from ACP countries qualified for entry into the EEC without any form of restrictions nor reciprocal obligations. Trade preferences in the Lomé convention covered about 99 percent of the industrial products of ACP countries without quantitative limits (Panagariya, 2002). One of the landmark adjustment the introduction of the Lomé conventions made to the Yaoundé II convention was the founding of the Sugar, beef and veal, banana, and rum Protocols. These Protocols refer to arrangements that require that the European Community imports a specific quota of the affected commodities each year from the ACP producers at guaranteed prices, which have to be higher than the world market prices (Laaksonen et al., 2007). The ACP producers are however, selected by the EEC. In 2006, the European Commission reported that under the Cotonou Convention (which succeeded the Lomé conventions), 97 percent of ACP exports to the EU are exempted from custom duties. 2.3.1.3 Rules of Origin (RoO) The provisional requirements of EU’s RoO have been modified through the several preferential schemes implemented by the EU. Brenton (2008) compares the RoO regarding vessels under the EBA and Cotonou Agreement. He succinctly observes that, the EBA required vessels to be registered and sailed under the flag of either the beneficiary country or EU. However, under the Cotonou agreement, the vessel could be registered and sailed under the flag of the EU or any ACP state. Again, the Cotonou agreement made amendments to the ownership and crew composition of vessels by University of Ghana http://ugspace.ug.edu.gh 41 allowing 50 percent of the crew, and the master and officers to be nationals of any ACP state or the EU, which under the EBA is allowed at 75 percent. University of Ghana http://ugspace.ug.edu.gh 42 CHAPTER THREE LITERATURE REVIEW 3.1 Introduction This chapter focuses on providing a theoretical review, which concentrates on identifying the theoretical basis of trade as well as presenting the theoretical framework that provides rational theoretical justification for trade preferences in international trade. Also, several existing literatures on preferential trade are perused and the key and trending arguments concerning the empirical veracity of trade preferences are uncovered and expounded. 3.2 Theoretical Review Any theoretical discussion of preferential trade cannot be done without considering the theoretical foundations of international trade itself. A review of international trade theories is necessary to understanding the rigorous theoretical dimensions of preferential trade. 3.2.1 Models of International trade The premier discussion of international trade is often attributed to Adam Smith owing to the publication of his book “The Wealth of Nations (1776)”. However, much of international trade was discussed during the pre 1776 period. Whereas the discussion of international trade under the pre 1776 era and the Adams Smith regime focused on inter- industry trade, emerging trends in trade have highlighted the eminent significance of intra-industry trade. This has resulted in the theoretical development of several trade University of Ghana http://ugspace.ug.edu.gh 43 models that have been propounded in an attempt to provide explanations for intra- industry trade. Therefore, the discussion of international trade models could be analysed under the blocs of inter-industry trade and intra-industry trade. Among the inter-industry trade theories are the Ricardian Model and the Heckscher-Ohlin Model. Models that have been developed to explain intra-industry trade include the monopolistic competition theory. 3.2.1.1 The Ricardian Model Based on Adam Smith’s concept of absolute advantage, David Ricardo developed a comprehensive trade theory (often referred to as the Ricardian Model) based on sectoral comparative advantage. The Ricardian model postulates that a country should concentrate or specialize in the production of a commodity if the opportunity cost of producing that commodity is lower as compared to other countries. This stands as one of the very useful models in explaining inter-industry trade. Essentially, the Ricardian model implies that difference in the relative productivity of sectors is enough to explain why countries would trade among themselves. David Ricardo, in propounding the comparative advantage theory, maintained the assumptions of Adams Smith. Thus, two countries, two goods, and one factor of production (labour). Also, perfect competition and constant returns to scale were assumed. The assumption of constant returns to scale explains the notion of constant opportunity cost as used exhaustively in the Ricardian Model. Other key explicit assumptions include, constant technology and full employment. The concept of complete specialization advanced by the Ricardian model suggests an increase in global output as a result of free trade. Also, the consumption possibilities of both trading partners increase. University of Ghana http://ugspace.ug.edu.gh 44 Following from the Ricardian Model, other theories have made adjustments to some assumptions underpinning the Ricardian model. 3.2.1.2 The Heckscher-Ohlin (HO) Theory. The traditional Heckscher-Ohlin model increases the factor assumption made under the Ricardian model to two. By so doing, holding onto the constant opportunity cost assumption is flawed. Hence, the HO model assumes increasing opportunity cost. Again, the HO model assumes incomplete specialization. The assumptions of the Ricardian model which are maintained in the HO model include constant returns to scale, perfect competition, no transportation cost, no barriers to trade and similarity of taste. Fundamental to the conclusion of the HO model is the notion of relative factor difference. With regards to the factor endowments, the HO model proceeds to assume that there exist different factor endowment ratios, perfect internal mobility of factors but complete immobility internationally, and no factor intensity reversals. In essence, the conclusion of the traditional two-factor, two-good, two-country version of the HO model states that each country exports the good that intensively uses its relatively abundant factor. It follows then that in this model, each country exports (embodied in goods) its relatively abundant factor and imports its relatively scarce factor (Baldwin, 2008). 3.2.1.3 The Specific Factors Model This model differs slightly from the HO model. Unlike the HO model, the specific factors model assumes that each country is endowed with three distinct factors. One factor is assumed to be general (labour) whereas the other two are assumed to be sector-specific; that is, they are only used in particular sectors. This model leads to an important proposition which states that the structure of trade is determined by the relative University of Ghana http://ugspace.ug.edu.gh 45 abundance of the specific factors. This proposition is not significantly different from the foundations of the HO model. As a result, the specific factors model is often viewed by some trade economists as the short run scenario of the HO model. Notwithstanding the similarities, this model presents more insight into the income distribution effect of free trade. According to this theory, the effect of free trade on the owners of the general factor (i.e. workers) is ambiguous. The effect depends on the consumption bias of the workers. For workers whose consumption is biased towards the export commodity, they may become worse off while workers whose consumption preference is for the importing good become better off. However, the effect of free trade on the owners of the specific factors can be unambiguously determined. Owners of the factor specific to the production of the export good become better off, whereas owners of the factor specific to the import sector lose out. The above mentioned classical trade theories are challenged by some empirical observations pertaining to the pattern of global trade. These empirical challenges are dominated by the observation of intra-industry trade. These trade theories do not necessarily explain explicitly the proliferation of intra-industry trade. Owing to this, other trade theories emerged to provide theoretical justification for the observed intra-industry trade trend. One of such models is the monopolistic competition model developed by Edward H. Chamberlin in 1933. According to this model, product differentiation combined with increasing returns to scale can easily explain intra-industry trade (Chacholiades, 1990). So, relaxing the constant returns to scale assumption in the classical theories forms the first step to explaining trade in similar commodities. The gains from trade under the monopolistic competition model is higher than those advanced University of Ghana http://ugspace.ug.edu.gh 46 by the classical theories. This is because of the presence of economies of scale in the monopolistic competition model. In addition to the monopolistic competition model, other models such as the Linder’s thesis (also known as the spillover theory), the technological gap and product cycle theory have attempted to provide theoretical justification for intra-industry trade. The international trade theories discussed in the preceding paragraphs advance a clear case for free trade. Core to these theories is the general conclusion that free trade is usually mutually beneficial. Hence, the need to encourage free trade. However, the realism of free trade is challenged by the empirical existence of restrictions to trade. These trade restrictions as observed in world trade impose huge costs to trade. So, to promote free trade, it is necessary to eliminate these trade costs or absorb them (through trade preferences). 3.2.2 Theoretical Framework of Trade Preferences What forms the basis of trade preferences is the apparent need for the elimination of tariff and non-tariff barriers to trade. Trade preferences, be it reciprocal or non-reciprocal present the opportunity to trade in the face of minimum tariff and non-tariff costs. Aside the apparent reduction in tariff and non-tariff costs, Bora et al (2002) and Mold (2005) present a partial equilibrium analysis of the welfare effects of Preferential Trade Agreements. Their analyses present a three-country, one homogenous good static partial equilibrium framework. The three countries are basically A, B and C. Countries A and B are assumed to be the trade partners with country C being the rest of the world. A and B are assumed to sign a non-reciprocal preferential trade agreement in which A is the donor University of Ghana http://ugspace.ug.edu.gh 47 and B, the beneficiary. So A is the importer and B is the exporter. The supply from C is assumed to be too large to be influenced by changes in the import volumes in A. So, the exporter supply curve to A is perfectly horizontal. Given the horizontal nature of the supply curve, prices would not respond to any changes in import volumes after the preferential agreement, hence, import quantities would not change. Other assumptions made to support this framework include perfectly substitutable homogenous good, and common world. Both studies also assume perfect competition and non-increasing returns to scale in both countries. With these assumptions, it can be established that preferential agreements reduce the cost of exporting commodities from B to A compared to from B to C. Consequently, there will be a shift in country A’s demand from C to B. Mold (2005) advanced two arguments in favour of the purported increase in the supply of B. The first is what he terms the ‘terms of trade effect’ and secondly the ‘displacement effect’. With the terms of trade effect, he argued that B will have better prices in the market of A because of the eliminated tariff cost compared to C. The displacement effect, he used to refer to the fact that all production of B will be sold in A in order to exploit the preferential advantage. Following the displacement effect, the export supply curve of B will be the same as its domestic supply curve. So in effect, preferential agreements increase the exports of the preference-receiving country. Also, Collier and Venables (2007) point that, the benefits of trade preferences accrue through two mechanisms. They cited the transfer of rent to recipient countries as the first mechanism. They argued that the removal of tariffs in trade preferences transfers the preference margin, which would have received by the developed country importer (either as tariff revenue or quota rent) to producers in the preference receiving country. The University of Ghana http://ugspace.ug.edu.gh 48 second mechanism they identified is an export supply response to the trade preferences. They argued that trade preferences present beneficiaries with the opportunity to increase the supply of their exports to also meet the foreign demand. The brief theoretical model and arguments discussed above highlights the possible deterministic mechanisms through which trade preferences can affect bilateral trade flows under the specified assumptions. The empirical veracity of the theoretical expositions on trade preferences has been examined by several researchers. A snapshot of this wide array of empirical literature seems to confirm the inconclusiveness of the empirically observed impact of trade preferences on bilateral trade flow. 3.3 Empirical Review The effects of trade agreements on trade flows, empirically, is not straightforward as the theoretical framework of trade preferences may seem to suggest. Many studies have been conducted in the SSA context and the results do not differ from the observed international trend. In reviewing the existing literature, the study considers primarily, studies that adopt the gravity approach in estimating the effects of preferential trade on the exports of developing or less developed countries (preferably, Sub-Saharan African Countries). In addition, some selected studies that employ descriptive measures would be reviewed as well. A brief review of the effects of Regional Trade Agreements precedes the perusal of studies that have considered the impact of either/both US or/and EU trade preferences on total exports as well as the exports of individual commodities. University of Ghana http://ugspace.ug.edu.gh 49 Considering SSA countries, several researchers have interrogated the possible effects Regional Trade Agreements (RTA) could have on bilateral trade. Turkson (2012) assessed the ex post bilateral trade effect of the EU-ACP PTA and sub-regional RTAs on bilateral trade involving SSA countries. Estimating a gravity equation through including the sub-regional RTAs and EU-ACP PTA as dummies, controlling for their endogeneity, as well as accounting for zero-valued trade flows and multilateral price resistance, the findings of the study indicated that EU-ACP PTA and RTAs within ECOWAS and SADC have a positive and significant impact on bilateral trade involving SSA countries. In another study, regarding the impact of RTAs on bilateral trade flows, Afesorgbor (2013) in an attempt to possibly explain the potential heterogeneity in the existing studies as well as compare the different results emanating from their distinct respective estimation methods, conducts a meta-analysis of previous empirical studies. Afesorgbor (2013) found that, the impacts of African RTAs are generally positive and highly sensitive to different estimation methods. With regards to NRPTA in general, Agostino, Aiello and Cardamone (2007) concluded after considering all NRPTAs granted by 8 major OECD countries that NRPTAs have a positive impact on total exports from developing countries. A similar and much recent study conducted by Gil-Pareja, et al (2014) investigated whether and to what extent nonreciprocal preferential trade agreements in general have increased the exports of beneficiary developing countries. Also, in the same study, they investigated the extent to which specific NRPTA have affected the exports of developing countries. They considered 177 countries over the period 1960 to 2008 at four year intervals. Adopting a panel estimation technique, and accounting for zero trade flows, individual NRPTA trade University of Ghana http://ugspace.ug.edu.gh 50 effects, and firm heterogeneity, their study revealed robust evidence that, non-reciprocal preference agreements in general have had an economically significant positive effect on exports. Moving to the individual preference effects, the study revealed, after controlling for unobserved bilateral heterogeneity, that whereas some of the selected non-reciprocal preferential trade agreements have had a positive effect on developing countries' exports (although at varying confidence intervals), others did not exhibit any evidence of a positive effect on the exports of beneficiaries. Of the NRPTAs that revealed positive effects were AGOA and the EU-ACP. Studies that have interrogated the impacts of unilateral preferential trade agreements on trade flows in SSA, have mainly focused on the preferential access offered by the United States and the European Union. In assessing the impacts the EU and US preferential trade agreements have had on bilateral trade flows in developing countries (often SSA), the analyses have mainly been targeted at either explaining if such preferences have been trade creating or trade diverting. With respect to US non-reciprocal preferences, the results of existing literature have been quite contradictory. Whereas most studies find significant positive effects of US trade preferences; mainly the GSP and AGOA (e.g Mattoo et al., 2002; Shapouri and Trueblood, 2003; Keck and Lendle, 2012) others have concluded otherwise (e.g Özden & Reinhardt, 2004; Lederman and Özden, 2004; Mueller, 2008; Cooke, 2011). In what follows, the study briefly presents the results of each of these studies, beginning with those that do not find positively significant effect of AGOA on total exports. To begin with, Lederman and Özden (2004) found that the impact of AGOA on beneficiaries was significantly negative. Their study employed the gravity equation to conduct a Tobit University of Ghana http://ugspace.ug.edu.gh 51 estimation considering data for 173 countries and 98 product categories for 2001. In specifying their gravity equation, they included bilateral covariates such as distance to capture transport cost, sharing a common language or a border, and being landlocked to also represent elements of natural trade cost to trade. The outcome by Lederman and Ozden (2004) could possibly be due to the fact that they considered data at the very initial stages of AGOA. So, beneficiary countries may have now been putting in place infrastructural development to fully exploit the AGOA trade preference. Hence, concluding based on these results may not be necessarily a fair estimate of the impact of AGOA. Moving forward, Mueller (2008) used time series data from 2000 to 2004 to assess the impact of AGOA on the total non-oil imports of the US. Employing the gravity model, he found negative, yet insignificant results, implying that AGOA had no significant impact on the non-oil exports of beneficiaries. Seyuom (2007) on the other hand, interestingly, found positive, albeit non-significant effect of AGOA on beneficiaries. His analysis also employed the gravity model on SSA export data to the US up to 2004. In a much recent study, Cooke (2011) used a matching approach to study the impact of AGOA on SSA beneficiaries. The results from his study showed that AGOA beneficiaries tend to export a lesser share of their exports to the US. The results thus point to a negative significant impact of AGOA on SSA beneficiaries. Contrary to these results, Shapouri and Trueblood (2003) estimated that countries’ responses to the AGOA at its initial phase had increased in 2002 compared to 2001. This was in line with their estimation of 15.7 percent preference margins for all AGOA products. Again, they observed that the estimated elasticities for all the commodity groups almost doubled from 2001 to 2002. Shapouri and Trueblood (2003) further moved University of Ghana http://ugspace.ug.edu.gh 52 to qualify their conclusions by adding that the high response does not necessarily reflect a high total export growth for all countries. In that, the observed and envisaged impacts of AGOA will not be the same for all countries. These conclusions of Shapouri and Trueblood (2003) derived after a partial equilibrium analysis conform to the earlier conclusions by Mattoo et al. (2002). However, Mattoo et al. (2002) conceded that although their analysis estimated an improvement of SSA’s exports owing to AGOA, their study could not estimate the resulting trade diversion or trade creation that underpin the observed export increase. In a study by Brenton and Ikezuki (2004), they found that the benefits of AGOA mainly accrue to countries who are beneficiaries of the clothing scheme under AGOA. As part of their findings, they assert that “the ability to export clothing products under preferences with liberal rules of origin is the key factor currently determining whether the African Growth and Opportunity Act has a significant impact on non-oil exporting African countries”. This falls in line with the conclusion by Shapouri and Trueblood (2003) that the benefits of AGOA differed from country to country. Brenton and Hoppe (2006) and Dean and Wainio (2006) affirmed the results of Brenton and Ikezuki (2004) by concluding that, the value of AGOA preferences excluding apparel is minimal. Even with respect to the apparel benefits under AGOA, Nouve (2005) argued otherwise that the increase in apparel exports under AGOA could rather have dire repercussions on the economies involved. He premised his argument on the fact that, countries benefiting from increased apparel export under AGOA will focus more on the apparel sector resulting in the apparent reallocation of resources from other exports sector to the apparel sector. This would dwindle the performance of other sectors of the University of Ghana http://ugspace.ug.edu.gh 53 economy, and may not be entirely beneficial to the country experiencing the increased apparel exports. Following up from the literary works at the initial phases of AGOA, Frazer and Van Biesebroeck (2007) used a triple difference in differences gravity model to analyse the effect of AGOA on the exports of AGOA beneficiaries from 2000 to 2006. Their study concluded that AGOA has had marginally positive and significant impact on beneficiaries. Keck and Lendle (2012) estimated that the ratio of imports claiming AGOA preference to imports of products eligible under AGOA in 2008 stood at 92 percent. This is representative of a high utilisation of the AGOA scheme by the beneficiary countries. This notwithstanding, previous works have highlighted the potential country and product differences in the utilisation rates in the face of a high aggregated or average utilisation rate. Brenton and Ikezuki (2004) noted that whereas the average utilization rate of US preferences (GSP plus AGOA) in 2002 was over 80 percent, 24 out of the 39 beneficiaries (about 62 percent) had utilization rates below the average. Quite outstanding was the fact that 13 of the beneficiaries (representing one- third) had utilisation rates below 20 percent. A similar pattern was observed by Brenton and Hoppe (2005) when they estimated the value of AGOA preferences on a country by country basis using data from 2005. They observed that for 26 out of the 37 AGOA beneficiaries, the value of preferences was less than 2 percent of the value of exports to the US in 2005. The perused literature above makes it very clear that the effects of AGOA on the exports of beneficiaries is not straightforward as may be suggested by the theoretical framework of trade preferences. Even in the face of positively significant effects, the majority of University of Ghana http://ugspace.ug.edu.gh 54 such studies holds the possibility of differing independent country effects of AGOA. That is, the effects of AGOA on independent beneficiaries may be entirely different from the effect obtained from aggregating the exports of all beneficiaries. Hence, the need to delineate the effect of AGOA on Ghana from the aggregate effect. The study, therefore adds to these literatures by trying to estimate the impact of AGOA specifically on Ghana’s bilateral exports. Some studies have also examined the viability of EU trade preferences. When it comes to preference utilisation, EU preferences are generally believed to be highly utilised (Keck and Lendle, 2012). On the contrary, other studies suggest that EU preferences are systematically under-utilised by exporters (Candau and Jean, 2005). Brenton and Manchin (2003) found that EU preference utilisation is much less in practice than on paper. In that, empirically, the degree of improved market access measured in terms of the exports of beneficiaries to the EU is less. They argued that, inasmuch as this finding is not a new finding, very little or nothing has been done to try and change this situation. Brenton (2003), in a related study, showed that in spite of the generally low utilization of EU trade preferences (especially the EBA) among LDCs, there are considerable variations across beneficiary countries with regards to the impact of the EU preferences on bilateral exports. This, according to Brenton (2003) emanates from the differences in the export structures of countries. The study further identified that there is a group of LDCs for whom EU trade preferences do not have significant impacts on their exports. The reason advanced in defence of this observation is that the exports of these countries are mainly products with which zero MFN duty apply. In addressing the low utilisation problem, Brenton (2003) attributed the observed underutilisation to the restrictiveness of University of Ghana http://ugspace.ug.edu.gh 55 the rules of origin. As shown by Brenton and Manchin (2003), the precise impact of EU preferential trade agreements relies on the detailed rules that the EU imposes to govern the eligibility for preferential treatment; especially for small countries with narrow industrial base. This brings to bear the restrictions imposed by strict rules of origin on preference utilisation, and its astounding relevance in assessing the impact of trade preferences. Strict Rules of Origin have been cited by recent researchers as one of the fundamental impediments to the potential utilisation of trade preferences. Brenton (2003) premised his scepticism about the actual benefit of the EPA mainly on the stringency of the Rules of Origin. Mattoo et al. (2003) similarly singles out Rules of Origin as the particular factor that contributes heavily to what they referred to as the “generosity” of AGOA. Mattoo et al. (2003) confirmed their assertion by modelling a general equilibrium framework and arguing that the potential benefit of AGOA in the absence of RoOs would have been five times greater. Similarly, Brenton and Ikezuki (2004) hinted that the benefits of AGOA under the clothing scheme appeared fragile under the considerations of tightening the apparel Rules of Origin in 2004. The complexity and rigorous detailing of RoO make it virtually impossible to downplay the repressive effects of their strict enforcement on trade preferences. Candau and Jean (2005) do not mince words when they posit that “given the cost and sometimes the complexity of these constraints, the benefit of preferential agreements cannot be considered as automatic, costless nor unconditional”. This claim informed their research agenda of assessing specifically, the extent to which EU trade preferences are used by exporters, and their value therein. Candau and Jean (2005) found that on average the University of Ghana http://ugspace.ug.edu.gh 56 utilisation of EU tariff preferences turns to be strong with the exception of the of the EBA for LDCs in South Asia. They cited the strict RoO on textiles and clothing as the main reason for the observed underutilisation. Krueger and Krishna (1995) in their analysis of trade preferences demonstrated how rules of origin can act as “hidden protectionism” and induce a switch in demand in free trade partners from low-cost external inputs to higher- cost partner inputs to ensure that final products actually receive duty free access.27 The analyses of RoO have taken a whole new dimension. Current research analyses of RoO have adopted a cost-benefit approach. The envisaged benefits of implementing a comprehensive RoO in any preferential trade agreement have been echoed through the efforts of preference granting countries in justifying the enforcement of the respective RoOs. The primal relevance of RoOs is to avoid trade deflection. By so doing, ensuring that the full benefits of trade preferences accrue solely to targeted beneficiaries. Another complementary advantage as hinted by existing literature in this respect28 is the branding of preference-receiving countries as investment hubs through an uncompromising enforcement of the RoO. Candau and Jean (2005) described the positive side of Rules of Origin as creating an indirect, yet significant incentive to invest in preference-receiving countries in order to benefit from preferential market access granted to those beneficiaries. Along the lines of restricting the utilisation of preferential trade agreements, a contributory factor is the usage of quality standards. In a study focused on India, Jayasuriya et al. (2006) employed the gravity model in investigating the impact of 27 As cited in Brenton and Manchin (2003) 28 See Candau and Jean (2005); Brenton (2003); Kipe (2003); Inama (2003) University of Ghana http://ugspace.ug.edu.gh 57 increasingly stringent and differing standards set by developed countries on exports by India’s food processing industries, and found that high quality standards undermined India’s exports of processed food. Their study also showed that food exported from India to EU countries and the US faced extremely restrictive standards. Concentrating on food safety standards, Baboola, et al. (2008) analysed data for processed food exports from 15 developing countries over 17 years using a gravity equation, and estimated that a percent increase in food safety standards decreased exports by approximately 0.5 percent. In spite of the identified underutilisation of trade preferences by beneficiaries, many researches have continually emphasized the relevance of trade preferences to DCs and LDCs. According to Persson and Wilhelmsson (2006), low trade flows which account for the underutilisation of trade preferences would have been exacerbated in the absence of trade preferences. Dwelling on the effects of EU trade preferences on the exports of beneficiaries, Verdeja (2006) estimated cross section regressions using both Fixed Effects (FE) and Random Effect (RE) models over the period 1973 to 2000. He found that ACP countries gained substantially from non-reciprocal preferential trade preferences granted by the EU. Thus, the EU-ACP agreement greatly influences the exports of beneficiaries positively. Specifically, the estimated ACP variable was significant and positive in 8 out of the 10 periods covered, with coefficients ranging from 0.525 to 1.271. This outcome followed from the study by Nilsson in 2002. Nilsson (2002) employed the gravity equation in estimating a cross-section regression from 1973 to 1992. He found that the effects of the GSP and Lomé conventions on EU imports are significant for most of the years. But, the GSP effects are consistently smaller than the effects of the Lomé conventions. Some University of Ghana http://ugspace.ug.edu.gh 58 studies have also identified negative effects of EU preferential schemes on exports of beneficiaries. For instance, Langhammer (1983) focused on estimating the effect of the EU GSP on the bilateral export of manufactured products of beneficiaries for the 1978- 1980 period. His analysis revealed that the binary variable representing GSP status was negative and significant in the period considered. He alluded to the use of quantitative restrictions against beneficiaries as one factor that explained the yielded results. Similarly, Golhar (1996) employed a gravity model and estimated a negative effect of the EU GSP. Özden and Reinhardt (2004) in a related study found that beneficiaries of the GSP perform better once they become ineligible for the program. The findings of Özden and Reinhardt (2004) support the results obtained by Golhar (1996). Specifically, Özden and Reinhardt (2004) estimated that the export-to-GDP ratio of countries removed from the EU GSP was 2.6% higher, compared to beneficiaries. Using the OLS estimation technique, Özden and Reinhardt (2004) estimated the coefficient of the GSP dummy to be - 0.252 and statistically significant. Thus, removal of the GSP is associated with a 22.3% ( difference in exports. Nilsson (2007) compared the effects of EU and US trade policies on developing country exports using the gravity equation. Estimating yearly cross-sections for 2001-2003, he found quite significant and positive effects of EU trade policies on exports of beneficiaries. Nilsson estimates the effects to be about 30 percent for all developing countries and 50 percent for low income countries. Again, the results indicated that the Cotonou Agreement had increased exports to the EU relatively more compared to exports to the US under AGOA. This outcome is supported by Haveman and Schatz (2003) who estimated that EU preferences have increased exports from LDCs by about 45 percent as University of Ghana http://ugspace.ug.edu.gh 59 compared to 10 percent in the case of the US. Davies and Nilsson (2013) conducted a similar comparative analysis of EU and US trade preferences for their respective beneficiaries. Their analysis showed that a larger share of EU imports benefited from duty-free MFN tariffs compared to the US. All these studies that compare the EU and US trade preferences adopted the gravity equation in estimating the effect of the trade preferences. The estimation of the trade preference effect has been one of the econometric issues grappling the minds of trade analyst. The issue has been the determination of a most appropriate model that captures and accurately estimates the effects of trade preferences. A key issue of concern in this regard has been the measurement of trade preferences. Many studies in analysing and estimating the effects of trade preferences have adopted the gravity model which was first used in the bilateral trade analysis by Tinbergen (1962) and Pöyhönen (1963). Other studies, however, adopt a partial equilibrium modelling framework (for instance Shapouri and Trueblood, 2003). The explanatory variables used by Tinbergen in explaining bilateral trade flows were the income levels of the importing and exporting countries as well as the bilateral distance between the trading partners. Income was used as an indication of the economic size (mass) of the trading countries whereas distance was used as a proxy of transport costs. Since the pioneering studies by Tinbergen and Pöyhönen, different variants of the gravity model have been adopted in estimating the trade effects of numerous preferential trade schemes. These variants of the gravity model basically depend on the a broader definition of what constitutes trade cost. For instance, Agostino et al (2007) in using the gravity model introduced population to augment the income measure of the economic size of the University of Ghana http://ugspace.ug.edu.gh 60 trading countries. Also, country-specific bilateral variables such as common border, language, colonial ties and landlocked were included to measure the effect of natural cost (cost associated with how countries are spread globally and the natural nature of countries). In the estimation of trade preferences, a common approach adopted in capturing preferences in the gravity model is to include them as independent dummy variables. Among studies that capture trade preferences as dummies are Turkson (2012), Persson and Wilhelmsson (2013), Davies and Nilsson (2013), Agostino, et al. (2007), Aiello and Demaria (2009), Laaksonen et al. (2007), Nilsson (2002), Verdeja (2005). For example, Turkson (2012) in estimating the effects of non reciprocal preferential trade agreement between SSA and the EU introduced two dummies (EUSSA to capture the non-reciprocal nature of the PTA and SSAEU PTA to capture the preferential treatment offered). Persson and Wilhelmsson (2013) used a large set of preference dummies which covered the MED, GSP, LDC GSP, and the ACP conventions by the EU. Similar to this, Agostino et al (2007) used three non-reciprocal preference dummies, that is, GSP, LDC GSP and other preferences. Agostino, et al. (2007) identified two inconsistencies in using the existing gravity equation to assess the impact of non-reciprocal preferential trade policies (NRPTPs). The bias they identified are the use of of the aggregate export flows at country level and also the econometric approach adopted in estimating the NRPTPs. In addressing these two inconsistencies, their study adopted bilateral export data at several aggregate levels. They considered total exports, total agricultural exports and export data, at the 2-digit level. In this respect, they concluded after considering all NRPTPs granted by 8 major OECD University of Ghana http://ugspace.ug.edu.gh 61 countries that NRPTPs have a positive impact on total exports from developing countries. However, using the export data at 2-digit level, it emerged that the effect of ordinary GSP, GSP for LDCs and other preferences is not always positive and statistically significant (Agostino, et al., 2007- page 30). Concerning the econometric estimation, they took into account the unobservable country heterogeneity as well as the endogeneity of trade preferences. Addressing the problem of unobserved country heterogeneity, Agostino, et al. (2007) used the Least Squares Dummy Variables (LSDV) approach. This approach, as used by the study decomposed the error term of the gravity specification into comprising time-invariant import and export-country fixed effects. This study would, instead, adopt the fixed and random effect models to account for the heterogeneity bias. Furthermore, their study tested for the endogeneity of trade preferences by using the Durbin-Wu-Hausman (DWH) endogeneity test, that compares the Ordinary Least Square and Instrumental Variable estimations. The results from the DWH test allowed for the rejection of the hypothesis that the preferential variables are endogenous in all estimations. The estimation also addressed the selection bias posed by zero-trade values. Turkson (2012) addressed the problem of possible endogeneity of preference dummies by adopting the Hausman-Taylor panel method. The Hausman-Taylor estimator assumes that some explanatory variables are correlated with the individual level random effects, but none is correlated with the idiosyncratic error. This estimator thus uses the average values of the time-variant exogenous variables and the deviations from these averages as instruments for the time invariant endogenous variables (Turkson, 2012). Egger and Pfaffernayr (2003) also noted that the Hausman-Taylor panel method incorporates time invariant variables that are correlated with bilateral specific effects in the estimation. University of Ghana http://ugspace.ug.edu.gh 62 Laaksonen et al. (2007) placed a spotlight on the Lomé conventions and thoroughly described the components and requirements of these agreements between the European Union and ACP countries. This study, however, focuses on the Cotonou Agreement. Laaksonen, et al further employed the gravity model to estimate the impact of the lomé convention on the disaggregated exports (at the product level) of 20 Lomé and 12 non- Lomé countries; focusing on cocoa, coffee and sugar. In modelling their gravity equation, they considered the real GDPs and populations of both the importing and exporting countries. Following Agostino, et al (2007), the study also addressed the problem of endogeneity by using the method of 2 Stage Least Square (2SLS) Instrumental Variable to estimate the gravity equation. They employed the 2SLS method after conceding that the potential two-way feedback between a country’s exports and its GDP could account for the presence of endogeneity. Again, to capture the effects of the Lomé Conventions, dummy variables were introduced into the models to represent each convention. The results of the study suggested that broadly, the Lomé preferences had a positive impact on the exports of beneficiaries. Notwithstanding, the product level aggregation analysis showed that cocoa and sugar exports seemed to have benefited from the Lomé preferences, whereas the effects on coffee exports was ambiguous. In a quite similar paper, however, expanding the trade preferences under scrutiny, Cardamone (2011) uses monthly data at HS8-digit level in a gravity equation to evaluate the impact of trade preferences on fruits and vegetables (focusing on five products; fresh grapes, apples, pears, oranges and mandarins). The study concludes that the impact of EU trade preferences is commodity-specific. In that, the impact differs according to the commodity concerned. In this regard, she particularly identified that the GSP has a University of Ghana http://ugspace.ug.edu.gh 63 positive impact on the export of fresh grapes only to the EU whereas, the EU-ACP preference (specifically, the Cotonou agreement) is much more effective in increasing the exports of oranges. Aiello and Demaria (2009) complemented this analysis by focusing on the EU GSP. They evaluated the impact of trade preference in enhancing the exports of developing countries’ exports to the EU. Considering 769 products from 169 developing countries, their study estimated a gravity model and concluded that the EU GSP has a positive and significant impact on the agricultural exports of beneficiary countries. The treatment of the GSP in this study follows the dummy representation as in the reviewed literature above. 3.4 Conclusion The economic effects of trade preferences on preference receiving nations are captured by the free trade analysis of international trade. The theoretical literature seems to justify a positive effect of trade preferences on beneficiaries. However, the perused empirical studies show that the effects of EU and US trade preferences, as well as Non-Reciprocal Preferential Trade Agreements in general on beneficiaries (mainly SSA), could be either positive or negative. Most of the studies have investigated the effects of these trade preferences on regional blocs, ignoring the differing impacts on individual countries. In estimating the effects of trade preferences on the regional blocs, most of these studies adopted a gravity equation which included bilateral covariates such as distance, common currency, language and colonial history. Few of these studies have included a multilateral University of Ghana http://ugspace.ug.edu.gh 64 resistance term to capture the influence of the relative positioning of trading partners. This study adds to existing literatures by bridging these identified gaps. University of Ghana http://ugspace.ug.edu.gh 65 CHAPTER FOUR METHODOLOGY 4.1 Introduction This chapter begins by giving a detailed overview of the theoretical foundations of the gravity model. The chapter continues to present a discussion of the exact methodology, highlighting the specific estimation techniques employed and describes the data used for the estimation. Major econometric issues pertaining to panel regression are discussed. At the same time, evidence of how the study addresses these identified problems are discussed and justification provided. 4.2 Theoretical Foundations of the Gravity model The gravity model was introduced by Tinbergen (1962) and developed by Linneman (1966) to establish and explain the relationship between the trade flows of two countries. From international trade theory, the volume of trade between two different countries basically hinges on two broad factors. That is, the size of the countries involved and the costs associated with the trade. The size of the countries can be segmented into economic size, geographical size and population size. The trade cost is known to consist of tariff and non-tariff barriers to trade. The theoretical analysis of the tariff cost is somewhat straightforward compared to the non-tariff cost. Notwithstanding the outcome of several studies, which suggest that free trade agreements increases trade flows through reducing University of Ghana http://ugspace.ug.edu.gh 66 trade costs, it is still unclear as to which element of these trade agreements plays a significant role. The concept of the gravity model emerged from Newton’s law of universal gravity, which states that the attractive force between two objects is dependent on the gravitational constant, the masses of the two objects and the distance between the objects. Newton’s result was that, there exists a positive relation between the attractive force between two objects and the masses of the objects. However, the distance between the two objects is inversely related to the attractive force. Similar to Newton’s gravity, the gravity model as used in economics seeks to provide explanations to some of the factors that explain the “trade force” between countries. That is, what factors are responsible for influencing international trade? Just as Newton’s law of gravity predicts, the gravity model predicts an inverse relationship and positive relationship between the volume of trade and the distance between countries and the sizes of the countries respectively. The theoretical justifications of these hinted relationships border greatly on trade cost. All variables that reduce trade cost would tend to foster trade. Distance, according to the model, imposes trade costs through transportation cost. That is, the more countries are separated by way of geographical distance, the lesser the trade flows between those countries. The larger the distance between countries, the higher the transport cost involved in trading between the countries. An increase in transportation cost increases trade cost. As a result, it becomes very costly for such nations to trade. Hence, a fall in the trade between such countries. University of Ghana http://ugspace.ug.edu.gh 67 Several researchers have tried to provide theoretical justification for the conclusions of the gravity model. However, the much referenced justification was by Anderson (1979). Anderson (1979) developed a gravity equation based on the micro foundations of monopolistic competition. In analysing the gravity model, Anderson (1979) assumed the heterogeneity of goods. He differentiated goods based on their respective countries of origin.29 Anderson (1979) proceeded to identify that individuals reveal their preferences over all of the differentiated goods. This suggests that countries, to say the least, would consume at least some of every good from every country. So, Anderson (1979) supported the rational behind trade between countries on the basis of the differentiation of goods. Hence, one key conclusion that can be inferred from Anderson (1979) is that, all countries trade someway somehow. He eventually concluded that larger countries export and import more. This conclusion followed the premises that all goods are traded, and that all countries trade. This work formed the groundbreaking piece to pave way for the development of other gravity-related theories. Bergstrand (1985 and 1989) adopted a monopolistic competition approach in justifying the gravity equation.30 In his work, Bergstrand further developed the work by Anderson (1979) in understanding the microeconomic underpinnings of the gravity model. He went about his analysis in a general equilibrium framework with optimizing behaviour. Bergstrand’s approach suggested that the gravity equation depended on the size of the respective countries, transport cost, preferences for differentiated goods, and endowment. The unique characteristic of the Bergstrand approach is the incorporation of what seems to be Hecksher-Ohlin (HO) conclusion in the gravity model. Deardorff (1997) 29 This assumption is popularly known as the “Armington assumption” (see Armington, 1969). 30 The monopolistic competition framework was developed by Paul Krugman. See Krugman (1980). University of Ghana http://ugspace.ug.edu.gh 68 emphasizes the incorporation of the HO model with the gravity equation. From his work, Deardorff (1997) shows that the gravity model is consistent with the HO model where each country specializes in a distinct good. Anderson and Wincoop (2003) in a follow up to the work of Anderson (1979), derived a gravity model which equated the bilateral trade between two countries to the product of their respective national incomes, and the bilateral trade barrier between these two countries. The work of Anderson and Wincoop (2003) is widely seen as the standard reference for the theoretical foundation of the gravity model (Nilsson & Matsson, 2009). The derivation of the gravity model was based on the assumptions of identical homothetic consumer preferences, market clearance and symmetric trade costs, among others. Anderson and Wincoop (2003) identified that the gravity model is grounded on certain main assumptions. They referred to these as the building blocks of the gravity model. According to them, the first building block is that all goods are differentiated by place of origin. They proceed to assume that each region is specialized in the production of only one good. Hence, supply of each good is fixed. The second building block they emphasize is identical, homothetic preferences, approximated by a Constant Elasticity of Substitution (CES) utility function. The research community has expressed keen interest in defining and including several variables that prove to be empirically useful to trade determination in the specification of the gravity equation. Aside the traditional independent variables identified by the gravity theory, there are other new variables that have been identified and introduced as contributing greatly to trade flows. They include cultural and political affinity (such as colonial ties, currency similarities), trade preferences, and population. Over the past University of Ghana http://ugspace.ug.edu.gh 69 decade, the gravity equation has emerged as the empirical workhorse in international trade to study the ex-post effects of trade preferences on bilateral trade flows (Cipollina, et al., 2013). 4.3 Model Specification For the purpose of estimating the trade effects of AGOA and EU-ACP trade agreements on Ghana’s bilateral exports, the study adopts the gravity model. The gravity model has demonstrated an excellent empirical robustness in describing trade flows (Kepaptsoglou et al., 2010). Also, the gravity model is the most popular among simulation and econometric approaches mainly because of its robust performance and limited need for parameter assumptions (Philipin and Molini, 2003).31 The study adopts the traditional variables of the gravity model as introduced by Tinbergen (1962). Other variables are, however, introduced into the model to satisfy new developments in international trade theory as well as to help explain developments in Ghana’s bilateral exports. This section begins by specifying the gravity model in its simplest original form and eventually, augmenting it to suit the principal objective of the study. The gravity equation relates bilateral trade flows to Gross Domestic Product (GDP), distance and other factors that affect trade barriers (Anderson and Wincoop, 2003). Agostino, et al. (2007) also noted that a gravity model represents trade flows as being determined by variables that describe the demand of importing countries, the supply conditions of exporting countries, and the bilateral trade barriers. In the gravity model, 31 As cited in Kepaptsoglou, et al. (2010) University of Ghana http://ugspace.ug.edu.gh 70 the independent variables enter the equation multiplicatively. The traditional gravity equation following from Tinbergen (1962) and Anderson (1979) is of the form; (4.3.1) is trade flow from country i to j; and measure the economic size (GDP) of country i and j respectively. For the purpose of this study, the land area of the trading countries and their respective populations are included to further gauge the economic masses of the trade partners (e.g Afesorgbor, 2013; Turkson, 2012). is a set of observable bilateral variables that impose trade cost and is the error term. The simple specification of the gravity model uses distance as a proxy of trade cost. The distance measure is expected to be negatively correlated to trade. This is because distance increases the cost involved in trading, especially transport cost. Also, it is observed that countries tend to trade less with partners who are geographically distant. Chaney (2011) confirmed this by succinctly stating that “as long as firms engage in direct communication with their client and suppliers, and information permeates through these direct interactions, one ought to expect that aggregate trade is inversely proportional to distance”. For econometric purposes, a log-linearized form of equation (4.3.1) is often estimated. However, several studies have pointed out the possible flaws with this procedure. Often cited is the study by Silva and Tenreyro (2006). In their study, they found overwhelming evidence that suggested that the variance of the error terms in the normal log-linear representation of the gravity equation are not zero. That is, using the log-linear University of Ghana http://ugspace.ug.edu.gh 71 specification, least square estimates would suffer from heteroskedasticity. In spite of this, the study adopts the log-linear specification and corrects for heteroskedasticity in the model. The log-linear specification of (4.3.1) is presented below as; (4.3.2) Decomposing the set of bilateral variables ( in equation (4.3.2) forms the basis for the observed difference in the specification of the gravity model used by existing literature. Fundamentally, several studies separate distance from this set of bilateral variables in the model specification. This is simply because whereas the other bilateral variables are introduced as dummies, distance is not a dummy variable. So, in the log specification, the natural log of the distance is taken whereas the dummy variables are not logged. The inclusion of these variables depends essentially on how broad trade cost is defined. Trade costs have been classified as primarily emanating from two sources; natural sources and artificial sources. The natural costs encapsulate all costs incurred as a result of the geographical location of countries. These include transportation cost, costs associated with the landlocked nature and contiguity of countries. The artificial trade cost, on the other hand, refers to “man-made” costs. They are mainly tariff and non-tariff costs. With regards to the natural costs, several literatures have included different measures of these costs in their gravity specification. McCallum (1995) estimated a traditional gravity equation with two measures of natural cost (i.e. bilateral distance and a ‘same-country’ dummy). The ‘same-country’ dummy was included to determine the extent to which internal trade outweighed cross-border trade in terms of trade volumes. McCallum (1995) after estimating his augmented gravity equation for US states and University of Ghana http://ugspace.ug.edu.gh 72 provinces in Canada, and controlling for distance and size, found that trade between provinces were twenty-two times more than trade between states and provinces. This highlighted the presence of substantial cross-border trade cost. Following this study, Anderson and Wincoop (2003) refined the theoretical basis of the gravity model by including “Multilateral Trade Resistance (MRT)” variable. The inclusion of this variable was motivated by the proven belief that the overstated impact of national borders, as found by McCallum (1995) could be alluded to an artificial trade cost. This resistance was not captured by McCallum’s specification of the gravity model which assumed free trade (Verdeja, 2006). Studies such as Wei (1996), Evans (2003), and Anderson and Wincoop (2003) have also contributed to the studies that seek to estimate the effect of border-related trade costs in the gravity model. Anderson and Wincoop (2003) specified their augmented gravity equation as;32 33 (4.3.3) Where, and are the exporter and importer MRT, and is the bilateral trade cost between the two countries. Van Bergeijk and Brakman (2010) argued that the MRT is relevant in determining bilateral trade, as the relative position of two countries influences the trade between the two countries. The method proposed by Anderson and Wincoop (2003) in estimating the MRT was computationally demanding (Afesorgbor, 2013). As a result, some studies proposed alternative estimation methods. Wei (1996) developed proxy variables (ie. exporter and importer remoteness index) as measures of the MRT. 32 This section draws heavily from chapter 2 of Turkson (2012). 33 As cited by Afesorgbor (2013) University of Ghana http://ugspace.ug.edu.gh 73 Baier and Bergstrand (2007) also suggested the use of time varying bilateral fixed effects to control for the MRT. Rose and Wincoop (2001) also used country specific binary variables to account for the MRT. In line with the difficulty in measuring the MRT proposed by Anderson and Wincoop (2003), the study adopts the approach by Baier and Bergstrand (2007) and Wei (2008) by adopting a remoteness index to proxy the MRT. Other variables that measure natural costs and country differences such as being landlocked, sharing a common currency, border or language, and past colonial ties have been considered by some studies (see among others Agostino et al., 2007; Feenstra et al., 2001, Lederman and Ozden, 2007). The study includes country-specific and bilateral variables such as Contiguity, landlocked, common language, common colonial history and common currency. The contiguity, Landlocked and common language bilateral variables are included to account for major time-invariant factors that affect exports. The dummies for common colonial history and language are used to capture information costs. Countries with a common language and colonial history are likely to know and understand each other well compared to countries without a common history. For this reason, countries are more likely to trade with partner countries who share a similar culture (with respect to language and colonial history). It can be shown that without incorporating these effects into the gravity specification, the parameter estimates of the model can lead to incorrect inference as their values may artificially be inflated or deflated by this misspecification (Mátyás, 1997). These variables have zero within variations and remain constant over time. Hummels (1999), Hutchinson (2002) and Dunlevy (2006) identified common language as an important determinant of trade flows in the gravity model. Also, Radelet and Sachs University of Ghana http://ugspace.ug.edu.gh 74 (1998) found that landlocked countries suffered transport cost which was twice as high as that for coastal countries. Still on trade cost, studies that seek to capture artificial cost have mainly focused on the cost associated with tariff barriers. Notwithstanding, studies such as Head and Mayer (2000), Chen (2004) paved the way for further interrogations into the effects of Non- Tariff Barriers (NTBs) on trade as their results sparked controversies. The results of these studies showed that NTBs have little effect on trade. Tariff barriers are generally represented in the gravity specification by including dummies for the existence of regional trade agreements, and preferential trade agreements. These trade preferences, either reduce or eliminate the tariff barriers to trade. Hence, trade is expected to increase in the presence of these preferences. Now, most preferential trade agreements are being made to cover both tariff and non-tariff costs. So, recent literatures have sort to include these trade preferences to cater for artificial cost and at the same time estimating the significance of these preferences. Quite a number of studies that consider trade preferences have been reviewed in chapter three. In the case of this study, two main preferential trade agreements (that is AGOA and EU-ACP) which entail unilateral reductions in tariff and non-tariff barriers granted by the US and EU respectively are considered. Incorporating all these adjustments into the traditional gravity equation, the model to be estimated is specified in a log-linear form as opposed to the non-linear specification by some studies (e.g Coe et al, 2002; Ewing and Battersby, 2005) as; University of Ghana http://ugspace.ug.edu.gh 75 Where; j = destination of exports from Ghana; t = time; and is a log-normally distributed error term with E(ln ) = 0. From equation (4.3.4), bilateral exports at time t, Xijt is expressed as a function of GDP (Yit and Yjt), the distance between Ghana and export destinations (Dij), a set of control variables that are used as proxies for some bilateral and country characteristics, and the idiosyncratic error term ( . The country-specific variables include Population (POPit and POPjt) which is time-variant. Area (AREAi and AREAj) and the landlocked nature of the importing countries (LLCKj) are time-invariant. The bilateral characteristics included in the model are remoteness of both countries (REMOTEijt), sharing of a common border (CONTij), usage of a common language (LANGij), common currency (COMCURijt), and common colonial ties (COLij). Dummy variables for the EU-ACP agreement and AGOA are included as EUACPjt and AGOAjt respectively. The subscripts i and j refer to Ghana and the importing countries respectively. University of Ghana http://ugspace.ug.edu.gh 76 4.4 Data and Variables The dependent variable is the bilateral exports of Ghana to 192 destination countries34 (Africa-52, America-40, Asia-46, Europe-43, Pacific-11). The explanatory variables consist of the traditional gravity variables (that is GDPs and Distance), and population. Remoteness is also included to proxy the Multilateral Resistance Term. Other covariates of the gravity model such as the common colonial ties, common currency, contiguity, common language, and the landlocked nature of export destinations are added. 4.4.1 Description of Variables The explained variables often used in trade gravity models are bilateral exports and bilateral trade flows. Quite a significant number of studies have treated the average of two-way bilateral trade as the dependent variable (e.g Tomz, et al., 2007; Glick and Rose, 2002; and Rose, 2004). However, almost all theories that underlie a gravity-like specification yield predictions on unidirectional bilateral trade rather than two-way bilateral trade (Gil-Pareja et al., 2014). This informs the choice of bilateral exports as the explained variable used in the analysis. Most importantly, the study chooses bilateral exports instead of bilateral trade flows basically because the focus of the study is to investigate how the trade preferences have impacted Ghana’s export capacity. The explanatory variables employed can be broadly categorised as demand and supply variables, and trade obstructing variables. The mass variables that connote the supply and demand conditions of the trading countries are primarily measures of a country’s economic and market size; proxied by GDP, population and area size. Some studies have 34 See Appendix VII for the list of importing countries considered. University of Ghana http://ugspace.ug.edu.gh 77 explicitly included GDP per capita to measure the purchasing power of both importing and exporting countries (e.g Nilsson, 2002; Verdeja, 2006). This study, however, does not include the per capita measure of GDP. This is to avoid the presence of possible multicollinearity among the variables. The specified model in equation two is likely to suffer multicollinearity should GDP per capita be included as a complementary measure of economic size in addition to the GDP. Verdeja (2006) dropped the GDP per capita variable on grounds that it added no useful information to the equation since size is already being measured by GDP. In addition, a remoteness variable is included to capture the MRT identified by Anderson and Wincoop (2003). The other explanatory variables considered are distance, landlocked, common border, common currency, common language, and common colonial ties. Bilateral Exports (Xijt): This refers to the total value of exports from Ghana to respective export destinations. Owing to data availability, the total bilateral imports of the importing countries from Ghana, denominated in US dollars, are used in place of the bilateral exports. Since a domestic country’s exports are the same as the foreign countries’ imports from the domestic country; accounting for data measurement differences, there is no problem with the data used. According to Head, Mayer and Ries (2010), import reports are more reliable than export reports possibly because governments track imports closely because they are subject to customs duties. The bilateral import statistics were sourced from the UN COMTRADE. Data retrieved from the UN COMTRADE are reported according to the Harmonised System (HS). Gross Domestic Product (Yit and Yjt): Yit and Yjt are used to represent the economic size of Ghana and the export recipients. The variant of GDP used is the GDP measured at University of Ghana http://ugspace.ug.edu.gh 78 current US dollar (millions), sourced from the World Bank’s World Development Indicators (WDI). Distance (Dij): Dij is a time-invariant measure of the geographical distance between Ghana (i) and the respective importing countries (j). Data for the distance measure is sourced from the CEP II gravity database on distance measurement. Dij as computed by CEPII follows the great circle formula which uses the longitudes and latitudes of centers, and captures the weighted distance measure. Remoteness (REMOTEijt): This is included to capture the multilateral resistance to trade. Following Baier and Bergstrand (2002) and Turkson (2012), the study calculates remoteness by taking an average of the weighted mean of the distance from Ghana to all trading partners except country j. The resulting average is log-transformed. The weights refer to the proportions of world GDP held by the trading partners. The share of world GDP is computed as the ratio of partner countries’ GDP to the world GDP. Population (POPit and POPjt): POPit and POPjt are used to represent the total population of Ghana and the respective bilateral import countries over time (measured in millions). Area (AREAi and AREAj): AREAit and AREAjt measure the geographical land size measured in square kilometers of Ghana and the importing countries respectively. AGOAjt: This is a dummy variable, assuming the value of 1 if both Ghana and the importing country are members of the African Growth and Opportunity Act at time t. It assumes a value of 0 if otherwise. University of Ghana http://ugspace.ug.edu.gh 79 EUACPjt: This is a dummy representation of the trade preference agreement between the European Union, and the African Caribbean and Pacific countries. EU-ACP is simply denoted by 1if both Ghana and a member of the EU 28 belong to the EU-ACP agreement at time t. The value of the dummy variable is 0 if otherwise. Other bilateral dummy variables included in the model include CONTij assuming a value of 1 if the two countries involved share a common land border and 0 otherwise. Also, COMCURijt, COLij, and LANGij assume a unitary value if both importing and exporting countries use a common currency, share a colonial history, and share a common official language respectively. Otherwise, the variables assume a value of 0. LLKj measures the landlocked nature of the importing country. It assumes a value of 1 if the importing country is landlocked, and 0 if the importing country is not landlocked. 4.5 Estimation Techniques The study estimates the gravity model specified in equation (4.3.4) using an unbalanced bilateral export panel of about 7449 data observations collected for 192 countries, for the period 1960 to 2013. After eliminating zero export flows from the data, the importing countries and observations used for the analysis reduced to 181 and 4671 respectively. The study adopted a panel estimation in order to capture both between and within variations. Also, panel data give more informative data, more variability, less collinearity among variables, more degrees of freedom and more efficiency (Gujarati, 2004 4ed). Again, panel data allow us to collate data on Ghana’s bilateral exports and at the same time make it possible to account for certain unobserved individual variables that affect University of Ghana http://ugspace.ug.edu.gh 80 Ghana’s bilateral exports but not addressed in the traditional gravity model. For instance, Rault et al (2007) noted that the impacts of historical, cultural and linguistic links in trade flows, if not considered, can lead to biased estimates. Also, the presence of common bilateral relationships, such as past memberships in common trade area could yield similar results. For this matter, the study chooses panel data over cross-section or time series data, because, panel data can enrich empirical analysis. This view is supported by Mátyás (1997), Egger (2000), Ghosh and Yamarik (2004), and Kepaptsoglou, et al. (2010). Matyas (1997) argued that the appropriate specification of the gravity model is a three-way model. Egger (2000) added that, there is the temptation of interpreting coefficient in a cross-section analysis in a conceptually inappropriate manner (ie. interpreting coefficients in the same way) as opposed to the ideal way of interpreting as a composite of within and between effects. Panel analysis, interprets the parameters as elasticities of the influence of the regressors on the regressand. Panel analysis is employed to appreciate the within-country and between-country dynamics of the trade preferences and to control for certain unobserved characteristics of the countries (Wooldridge, 2012). The study begins by running a simple OLS regression. The estimated simple regression equation suffers from omitted variable problems. One way of resolving this problem is by introducing more factors that may affect the dependent variable. This approach may not be reliable as it may be essentially difficult to identify all factors that could affect the dependent variable. As a result, the study introduces time dummies and an additional variable to cater for the time-invariant unobserved variables. Introducing these control variables, we re-specify equation (4.3.4) as; University of Ghana http://ugspace.ug.edu.gh 81 From (4.3.5), the unobserved country heterogeneity ( captures all unobserved factors that affect but are constant over time. This is why the variable has no ‘t’ subscript, to show for its constancy over time. The coefficient of the unobserved effect is set to one. This is because is unobserved and virtually has no natural unit of measurement. So it would be meaningless to try to estimate its partial effect (Wooldridge, 2002). In addition, the time dummies ( are constant across countries, but vary over time. The pooled OLS estimation technique is used to estimate the parameters in equation 4.3.5. One of the advantages of using pooled regression is its ability to explain the past while simultaneously predicting future behaviour of exogeneous variables in relation to endogeneous variables (Moon, et al, 2004). Under the pooled OLS regression of (4.3.5), the parameter estimates would be consistent and unbiased when the covariance of the unobserved heterogeneity and the explanatory variables is assumed to be zero. Hence, the unobserved country heterogeneity is uncorrelated with the explanatory variables. Also, for the pooled OLS estimation to be valid the mean of the error term has to be zero, the variance of the error terms across observations has to be constant, and there should be no correlation between the error terms of different time periods (Moon et al, 2004). According to Moon, et al (2004), the accuracy of the regression model is inevitably University of Ghana http://ugspace.ug.edu.gh 82 measured by the error term. There are a number of methodological disadvantages associated with using the pooled regression. According to Stimson (1985), the basic assumptions underlying traditional OLS regression are violated in a pooled model, and such departures may exhibit severe consequences for the reliability of the estimators. With regards to heterogeneity bias, when the unobserved effect is combined with the idiosyncratic error term to form the composite error term, assuming that the idiosyncratic error term is uncorrelated with the regressors alone does not guarantee an unbiased and inconsistent parameter estimate of the pooled OLS. This is true because any possible heterogeneity bias (ie. correlation between the country-specific unobserved effect and the explanatory variables) would render the parameter estimates biased and inconsistent. Therefore, allowing for a possible correlation between the unobserved effect and the explanatory variables does not make the pooled OLS estimator ideal for the estimation. Egger and Pfaffermayr (2003) argued that the omission of specific country pair effects can bias the estimated coefficients. To overcome this problem of serial correlation, the method of “First Differencing” is often recommended to remove the time-invariant effects. With this, data in adjacent periods are differenced after which pooled OLS is used to estimate the parameters of the differenced equation, given that the error terms of the differenced equation are homoskedastic and serially uncorrelated. Another way to address the unobserved effect which is adopted in the study, is to employ the fixed effects transformation. This approach assumes that there is correlation between the unobserved heterogeneity and the explanatory variables and hence transforms the data into deviations from individual means. By doing so, the fixed effect estimator, also University of Ghana http://ugspace.ug.edu.gh 83 referred to as the within estimator eliminates the time-invariant effects. Basically, fixed effects transforms data into their time-demeaned forms35 and eventually pooled OLS is used to estimate the time-demeaned variables. One disadvantage of the Fixed Effect Model (FEM) is that it trades consistency for efficiency. This is so because, using only within variation ignores the effect of the between-unit variations and hence results in large standard errors. Another disadvantage of using this fixed effect estimation is that, all time invariant variables in the specified model are omitted. As a result, if the variables of interest are constant over time, another estimation technique referred to as the Random Effect Model (REM) is prescribed. The REM assumes that there exist zero correlation between the unobserved effects and the explanatory variables so as to estimate the parameters of any time-invariant and time-variant variable in the model. The selection between random and fixed effects relies on the interests of the analysis, the country sample, the data properties and the underlying theoretical model used. Fixed effects are better for short term prediction purposes. Random effects should be considered if they are adequately consistent and there is an interest in estimating time- invariant effects. For this analysis, the study employs the Hausman (Chi2) test to decide between which of these two estimation techniques to use for final interpretation. The Hausman test tests for the hypothesis that the unobserved country heterogeneity is uncorrelated with the explanatory variables. Thus, a rejection of the null hypothesis of the Hausman test suggests that the fixed effect model is the appropriate estimator for equation (4.3.5). Essentially, the Hausman statistic tests for the orthogonality of the random effects and the regressors, this is thus a test for misspecification (Egger, 2000). 35 That is the difference between the variables and their individual means. See Wooldridge (2002) University of Ghana http://ugspace.ug.edu.gh 84 One problem confronted in estimating the gravity equation is the exogenous specification of the AGOA and EU-ACP preferences (Baier and Bergstrand, 2002). In the specified model AGOA and EU-ACP are represented as right hand exogenous dummy variables. However, Baier and Bergstrand (2005) showed that trade agreement dummies are not strictly exogenous owing to omitted variables, simultaneity bias, and measurement error. Magee (2005) found that two countries are most likely to sign an agreement the closer their geographical locations, the more similar their economic sizes are and the more they trade. GDP, which is a function of exports and imports is potentially endogenous to bilateral trade flows (Baier and Bergstrand, 2002). However, Frankel (1997) reported that the endogeneity of income makes little difference. This study, therefore, focuses on the endogeneity of the AGOA and EU-ACP dummies. In resolving this problem, the study adopts the Hausman-Taylor (HT) estimation technique proposed by Hausman and Taylor (1981). The HT estimator controls for endogeneity by using the average values (and their deviations) of the time varying exogeneous variables as instruments for the time invariant endogeneous variables (Turkson, 2012). Egger and Pfaffermayr (2003) also indicated that a fixed effect model can be used to control for the endogeneity of trade agreements. This study, therefore, compares the results of the Fixed effects with the results from the Hausman-Taylor estimation. The FEM assumes that the regressors are correlated with the unobserved specific effects, whereas the REM assumes that there exists no correlation between the regressors and the individual heterogeneity. However, the Hausman-Taylor model assumes that some regressors are correlated with the heterogeneity bias. The usefulness of the Hausman-Taylor over the REM and FEM is that, the HT allows for University of Ghana http://ugspace.ug.edu.gh 85 controlling the variables that are correlated with bilateral effects in the estimation (Turkson, 2012). The expected signs of the estimated parameters are based on traditional arguments. Typically, we expect a positive sign for the parameter estimates of the economic size variables, common language, and common colonial history. The distance variable together with the landlocked dummy are expected to have negative parameter estimates. 4.6 Diagnostic tests The study tests mainly for the presence of serial correlation and heteroskedasticity in the model. Also, the significance of time effects is tested for. The study, does not focus much on the test for multicollinearity since one advantage of panel data is that they contain less multicollinearity (Hsiao, 2007). 4.6.1 Serial Correlation and Heteroskedasticity The standard error component panel data model assumes that constant serial correlation through the random individual effects and homoskedastic variance of the disturbance (Baltagi, et al., 2008). These assumptions place econometric restrictions on the empirical applications of panel data models. Serial correlation is often an estimation problem associated with time series data. Since Panel data comprise of time elements, panel data is by extension not insulated from the problem of serial correlation. The presence of serial correlation also referred to as autocorrelation in a model simply suggests a mis-specification of the model. This could University of Ghana http://ugspace.ug.edu.gh 86 arise from either the omission of relevant variables, or the inclusion of irrelevant variables, or a combination of both possibilities (Granger and Newbold, 1974). Three well-known major consequences of autocorrelated errors in regression analysis identified by Granger and Newbold (1974) are that; (i) Estimates of the regression coefficients are inefficient. (ii) Forecasts based on the regression equations are sub-optimal. (iii) The usual significance tests on the coefficients are invalid. Owing to these statistical problems identified with serial correlation, the study adopts the test for serial correlation in linear panel models derived by Wooldridge (2002). We employ this test ahead of other tests such as the Baltagi-Wu test. This is because, as proven by Drukker (2003), the Wooldridge test is much robust and has good size and power properties. The Wooldridge test is ideal for this study in testing for serial correlation also because Drukker (2003) found that its tremendous size and power properties is much profound with samples of moderate size, of which the sample for this study is no exception. Another major methodological problem with panel data analysis is the correlation between error terms of different periods. That is, a violation of the constant variance of the error term assumption [i.e var ( ) . The constant variance assumption (homoskedasticity) is very important for the pooled OLS regression to yield parameter estimates with minimum variance. Hence, the study conducts a post estimation test of this assumption so as to determine the efficiency of the pooled regression parameter estimates. Given that, parameters are rendered inefficient in the presence of heteroskedasticity, it follows that inferences from the F and t tests are no University of Ghana http://ugspace.ug.edu.gh 87 longer reliable (Wooldridge, 2003). In as much as the presence of heteroskedasticity affects the efficiency of the estimated parameters, it does not render the OLS estimator biased and inconsistent if all the other assumptions of the classical OLS holds. The pooled regression would yield usable estimates only if the assumption of homoskedasticity holds along side the other assumptions. To test for the presence of heteroskedasticity, the study adopts the Breuch-Pagan (BP) test. Aside the BP test, the study conducts a modified Wald test to test for heteroskedasticity in the fixed effects regression model. To correct for the presence of serial correlation and heteroskedasticity, the study runs a robust command as part of the panel estimation techniques to be compared (that is pooled regression and random and fixed effect; depending on the Hausman test). Another panel estimation technique that corrects for heteroskedasticity and serial correlation compared in the analysis is the Hausman-Taylor panel estimation technique. 4.7 Conclusion This chapter discussed the theoretical foundations of the gravity model and further specified an augmented gravity equation that would be estimated by the study. The method of analysis as well as preliminary diagnostic checks employed were clearly outlined with estimation variables properly defined. Also, the estimation techniques adopted to control for some identified econometric problems are explained in detail in this chapter. University of Ghana http://ugspace.ug.edu.gh 88 CHAPTER FIVE DATA ANALYSIS 5.1 Introduction This chapter presents and discusses the issues relating to the econometric estimation of the panel model specified in the previous chapter. The study uses STATA ahead of SPSS to carry out the regression analysis. This is partly because many scholars have argue that using SPSS to compute time-series related regression analysis is limited in dealing with the problems of autocorrelation or serial correlation, heteroskedasticity, and multicollinearity (Moon, et al, 2004). The analysis begins by conducting a descriptive analysis of the non-binary variables used in the study. We proceed to perform the diagnostic checks discussed in the previous chapter. Finally, the results of the actual estimations are presented and discussed. 5.2 Descriptive Analysis This section briefly discusses the basic statistical properties of the non-binary variables included in the model for the period 1960 to 2013. The summary statistics perused include the mean, standard deviation, minimum and maximum values. This is shown in Table 5-1 below: University of Ghana http://ugspace.ug.edu.gh 89 Table 5-1: Summary Statistics of Dependent and Independent Variables, 1960 - 2013 Source: Author’s Computation using STATA From Table 5-1, total bilateral exports from Ghana averaged about $9.2 million. The maximum value of bilateral exports was approximately $4.6trillion with zero minimum value. The average of Ghana’s GDP for the period averaged $8.9 billion, whereas the GDPs of all the importers for the period recorded an average of $193billion. Comparing the real GDP of Ghana to the importing countries, it suffices to conclude that on the average, an importing country is economically larger than Ghana. Exports of Ghana travelled on average, a weighted-distance of about 6,324 km to their destinations. The standard deviation column measures the dispersion of the variables from their means. Large standard errors reflect the presence of outliers that significantly influence the data. Another means of determining the spread is to consider the difference between the maximum and minimum values of the variables. The larger the range of a variable, the greater the standard deviation of the said variable. The variable with the largest standard deviation reported in Table 5-1 is the GDP of the importing countries. Variables Mean Standard Deviation Minimum Maximum Exports 9197339 1.22e+08 0 4.59e+09 GDPi 8904.279 11000.37 1217.153 47928.72 GDPj 173008.9 815783.5 18.43202 1.68e+07 Distance 6324.417 3656.332 257.2679 19110.09 POPi 15.53727 5.614252 7.130545 25.9046 POPj 34.53568 120.1783 0.030861 1357.38 AREAi 238538 0 238538 238538 AREAj 889743.3 2220108 7 1.71e+07 University of Ghana http://ugspace.ug.edu.gh 90 5.3 Diagnostic Tests In order to determine the appropriate estimation technique for the analysis, the study tests for the preference of multicollinearity, time effects, heterogeneity bias, heteroskedasticiy and serial correlation. To begin with, the Variance Inflation Factor (VIF) test for multicollinearity shown in Appendix XI revealed a generally minimal correlation among the regressors. In that, the VIF of most of the individual variables is less than 10. This outcome is relevant to the study because it purges the estimation results of any suspicion of severe multicollinearity among the independent variables. 5.3.1 Time Effects To test for time effects, the study introduces year dummies (with 1960 as the base year) into the pooled regression. After estimating the regression using OLS, the study tests for the joint significance of the year dummies. The result of the test shows that the year dummies are jointly significant.36 Therefore, the model with time effects is more appropriate. 5.3.2 Heterogeneity Bias One of the major problems of using the pooled regression for Panel analysis as discussed in chapter four is the problem of heterogeneity bias. Thus, there is the need to test for the presence of individual of heterogeneity. Table 5-2 reports the test for individual heterogeneity using the Breusch and Pagan Lagrangian Multiplier test for random effects. This test is also a test for random effects because it forms the basis for choosing between the REM and the pooled regression model. 36 F(51, 4606) = 67.29 Prob > F = 0.0000 University of Ghana http://ugspace.ug.edu.gh 91 Table 5-2: Breusch and Pagan LM test for random effects Var Sd = sqrt (Var) lnX 41.79352 6.464791 e 6.663591 2.581393 u 3.378487 1.838066 Test : Var (u) = 0 Chibar2 (01) = 2914.94 Prob > chibar2 = 0.0000 Source: Author’s Computation using STATA From the test statistic above, the null hypothesis is rejected and the Random Effect Model is chosen over the Pooled Regression. 5.3.3 Serial Correlation The assumption of no correlation between the regressors and the error term is one of the important assumptions that make parameter estimates consistent. To test for serial correlation, the study uses the Wooldridge test. The results from the Wooldridge test shown in Table 5-3 below, confirms the presence of serial correlation in the panel model. Table 5-3: Wooldridge test for Serial Correlation in Panel data H0: No first-order autocorrelation F(1, 152) = 22.084 Prob > F = 0.0000 Source: Author’s computation using STATA University of Ghana http://ugspace.ug.edu.gh 92 5.3.4 Heteroskedasticity For regression estimates to be efficient, the error terms must have equal and constant variance. The study tests the null hypothesis of a constant variance using the Breusch- Pagan test. The test results displayed in Table 5-4 reject the null hypothesis of constant variance and clearly confirms the presence of heteroskedasticity. Also, a test of heteroskedasticity in the fixed model using the Modified Wald test (Appendix IX) confirmed the presence of heteroskedasticity in the fixed effect model. Table 5-4: Breusch-Pagan test for Heteroskedasticity Ho: Constant variance Variables: fitted values of ln X chi2 (1) = 13.26 Prob > chi2 = 0.0003 Source: Author’s computation using STATA To correct for the observed autocorrelation and heteroskedasticity, a robust estimation of the pooled regression and fixed effects model is presented. Also, the study employs the Hausman-Taylor estimation technique to correct for the heteroskedasticity and serial correlation. 5.4 Empirical Results and Discussion The results as shown in Table 5-5 and Appendix X are obtained from estimating the gravity equation (4.3.5) using the pooled regression, random effect, fixed effect and University of Ghana http://ugspace.ug.edu.gh 93 Hausman-Taylor estimators. The result of the Hausman test in Appendix X rejects the null hypothesis of no correlation between the regressors and the individual heterogeneity. This renders the random effect model inappropriate. The results of the Hausman test confirm similar outcome of Turkson (2012). Therefore, the study concentrates on the parameter estimates obtained under the fixed effects and the Hausman-Taylor estimators. The performance of the models employed was fairly satisfactory as most of the core gravity covariates had significant coefficients with expected signs. The coefficient of determination (0.86) indicates that the regressors satisfactorily explains variations in Ghana’s bilateral exports. The importer and exporter GDP, and bilateral distance are highly significant and have their expected sign across all regressions. Importer population, remoteness of importer and exporter, common language, border and currency, as expected, exert positive effects on Ghana’s bilateral exports. Land size of importer, and the landlocked nature of importer countries as expected, exert a negative impact on bilateral exports of Ghana. Area of Ghana is omitted due to collinearity. University of Ghana http://ugspace.ug.edu.gh 94 Table 5-5: Estimates of gravity model using FE, RE and HT, 1960-2013. Independent Variables Dependent Variables: Log (Exportsijt) (1) (2) (3) (4) OLS RE FE Hausman-Taylor Log of GDPit 1.852*** -2.427* 2.257*** 2.839*** (0.143) (1.242) (0.212) (0.145) Log of GDPjt 1.177*** 0.721*** 0.501*** 0.499*** (0.100) (0.0588) (0.157) (0.0707) Log of Populationit 16.62*** (3.603) Log of Populationjt 0.173 1.072*** 2.576*** 2.172*** (0.122) (0.0992) (0.346) (0.160) Log of Distanceij -1.422*** -1.483*** -1.720*** (0.241) (0.192) (0.597) Log of Areaj -0.386*** -0.703*** -1.322*** (0.0852) (0.0784) (0.200) Remotenessijt 7.614*** 0.424** 8.031*** 4.568*** (0.300) (0.215) (0.321) (0.203) Contiguityij 1.258 0.871 0.0538 (0.982) (0.994) (3.208) Common Languageij 0.815* 1.170*** 1.289 (0.420) (0.324) (1.056) Common Currencyijt 0.536 0.162 0.316 0.289 (0.487) (0.408) (0.353) (0.395) Colonial Historyij -1.134** -1.474*** -1.880* (0.536) (0.351) (1.125) Landlockedj -0.462 -0.836*** -1.003 (0.353) (0.315) (1.010) EUACPjt 0.227 -0.290** 0.108 -0.0168 (0.351) (0.141) (0.349) (0.145) AGOAjt -0.217 -0.931 -0.769*** -0.840 (0.417) (0.574) (0.154) (0.557) Constant Included Yes Yes Yes Yes Time Effect Yes Yes Yes Yes No. of Observations 4,671 4,671 4,671 4,671 R-squared 0.867 Number of importing countries 181 181 181 181 Standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1; i and j refer to Exporter (Ghana) and importer country respectively. The coefficients for the year dummies are not reported for brevity. University of Ghana http://ugspace.ug.edu.gh 95 5.4.1 Economic Size From column 4, a percentage increase in Ghana’s GDP increased Ghana’s bilateral exports by about 2.8 percent. Also, Ghana’s bilateral exports increased by about 0.5 percent with a percent increase in the GDP of importers. This result suggests that the import demand for Ghana’s exports is fairly income inelastic. Again, the positive coefficient of importer GDP suggests that exports from Ghana can be considered in general to be normal goods. This outcome conforms to the results of Afesorgbor (2013), Turkson (2012) who considered the effects of trade preferences on bilateral trade of SSA countries. Afesorgbor (2013) adopted a Pseudo Poison Maximum Likelihood (PPML) estimation technique and found that importers GDP positively affects the bilateral trade of SSA countries. The population of importers exerts a positive and significant effect on Ghana’s bilateral exports. Hence, the larger the population of the importing countries, the higher the value of Ghana’s bilateral exports. It is observed that the fixed effects and HT estimations dropped Ghana’s population due to probable collinearity. This is evident in the VIF test for multicollinearity. 5.4.2 Distance Distance exerts a significantly negative impact on Ghana’s bilateral exports. Such that a percentage increase in the distance between Ghana and its trade partners reduces Ghana’s exports by approximately 1.7 percent. This supports the literature that concludes that transport cost negatively affect trade. A survey by Leamer and Levinsohn (1994) identified bilateral distance as one of the clearest and most robust empirical findings in economics. University of Ghana http://ugspace.ug.edu.gh 96 The parameter estimate of the remoteness variable supports the conclusion by Anderson and Wincoop (2003) that a multilateral resistance terms significantly explains bilateral trade flows. The results indicate that countries that are remote with Ghana from the rest of the world trade more. The coefficient is positive as expected and statistically significant. Turkson (2012) found similar effects of the remoteness of trading partners from all trading partners. 5.4.3 Control variables The variables, contiguity, common language and common currency have expected coefficient. However, the results are not significant. The landlocked nature of importing countries exerts a negative, albeit insignificant effect on exports from Ghana. The results from column 4 indicate that Ghana exports less to countries with which it has a common colonial history compared to the countries it has no colonial history with. The estimated results show that Ghana’s bilateral export to countries it has a common colonial history significantly reduces by 85 percent [100(1 - exp-1.880)] compared to countries it shares no common colonial history with. 5.4.4 EU-ACP In line with the main aim of the study, it is estimated that the EU-ACP preference dummy has a negative and insignificant impact on Ghana’s bilateral exports. Column 3 however, estimates a positive, yet insignificant effect of the EU-ACP preference dummy. The estimate obtained under column 3 suggests that bilateral exports from Ghana to the EU, in the presence of the EU-ACP, increased on average by 11 percent , holding the other factors constant. Going by the Hausman-Taylor estimates, it shows that Ghana’s membership in the EU-ACP agreement has a negative impact on export University of Ghana http://ugspace.ug.edu.gh 97 performance towards the EU. That is, Ghana’s bilateral exports under EU-ACP decreased by 1.7 percent [100( compared to bilateral exports to the EU in the absence of the EU-ACP. The results, however, reveal that these estimates are not statistically significant. Therefore, Ghana’s exports to the EU have not been significantly influenced by the EU-ACP trade preference. The outcome in column 3 seems to support the results obtained by Nilsson (2007). Nilsson (2007) estimated the coefficient of the impact of EU trade policy on lower middle income groups to be positive and insignificant. 5.4.5 AGOA The AGOA dummy was estimated to be negative in both columns 3 and 4. That is, AGOA reduced Ghana’s bilateral exports by 54 percent and 57 percent respectively.37 The coefficient was significant in the former and insignificant in the latter. This shows that irrespective of the estimation technique used in this study, the result remains that AGOA has had no positive effect on Ghana’s exports to the United States. Comparing the effects of EU-ACP and AGOA, the results show that under the fixed effect estimation, although the EU-ACP preference yield insignificant result, Ghana’s bilateral exports to the EU have increased under the EU-ACP whereas exports to the US suffered significant decline under AGOA. Going by the Hausman-Taylor estimation, the results indicate that both AGOA and EU-ACP trade preferences have had negative, howbeit insignificant effect on Ghana’s bilateral exports. In spite of the common insignificant result, the negative effect of AGOA (57 percent) is greater than that of the EU-ACP (2 percent). 37 100(1 – exp-0.769) and 100(1 – exp-0.840) respectively University of Ghana http://ugspace.ug.edu.gh 98 Testing for the joint significance of the EU-ACP and AGOA dummies revealed that both preferences do not jointly influence the bilateral exports of Ghana.38 This suggests that Ghana’s concurrent involvement in both preferences has not had any significant effect on its bilateral exports. 5.5 Conclusion The study contributes to the empirical literature which uses gravity models to assess the impact of NRPTPs on bilateral exports. The results of the study show that the exports from Ghana to the EU 28 and US have not been significantly affected by the EU-ACP and AGOA preferences. In the case of a significant effect for AGOA, the results indicated that exports to the US under AGOA reduced compared to exports to the rest of the world in the presence of AGOA. 38 Prob > chi2 = 0.3210 University of Ghana http://ugspace.ug.edu.gh 99 CHAPTER SIX CONCLUSION AND RECOMMENDATIONS 6.1 Introduction This chapter presents the summary and conclusion of the study and further provides some policy measures based on the findings of the study. 6.2 Summary and Conclusion The proliferation of preferential trade agreements has attracted the attention of several researchers. The focus of most of these researchers has been to investigate the ex-post effect of existing trade preferences on the trade performance of developing countries. Among such preferences that have been largely researched are AGOA and EU-ACP agreement. Though several studies have been undertaken to identify the effects of AGOA and EU- ACP on their beneficiaries. The results of these studies seem to be inconclusive. Whereas some estimate positive effects of AGOA and EU-ACP, others have found either a negative or insignificant effect. In addition, only a few of these studies have been identified by the study to have considered the effects of these preferences on beneficiaries individually. Therefore, the principal objective of the study was to investigate the effect of the AGOA and EU-ACP trade preferences on Ghana’s bilateral exports. This study, therefore delineates the effect on Ghana from the effect on SSA as a whole. University of Ghana http://ugspace.ug.edu.gh 100 Using an augmented gravity equation, the study estimated an unbalanced panel model of bilateral export data of Ghana to 181 countries. AGOA and EU-ACP trade preferences were introduced into the gravity model as dummy variables. The study estimated the effects of these preference dummies using the pooled OLS, FE and RE estimation techniques. However, the Hausman test rejected the null hypothesis of no correlation between the unobserved heterogeneity and the regressors. This result suggested the appropriateness of the Fixed Effect Model over the Random Effect Model. Subsequently, a robust estimation of the FE estimation was conducted and reported to control for serial correlation and heteroskedasticity. Controlling for any possible endogeneity in the model using the Hausman-Taylor estimation technique, the study found that both AGOA and EU-ACP do not independently exert any significant effect on Ghana’s bilateral exports. Also a test of the joint significance of both preferences revealed that both trade preferences do not have any significant joint effect on Ghana’s bilateral exports. 6.3 Recommendations Following the results obtained from the study the following recommendations are suggested. The recommendations suggested follow no particular order. To begin with, the study recommends that the EU and US extend their product coverage under the AGOA and EU-ACP respectively. Some of the perused literature in the study showed that with the exception of textile and clothing, most of the products covered under AGOA suffered strict Rules of Origin. For the EU-ACP, rules of origin were noticed to be relatively stricter on other commodities compared to bananas, sugar and University of Ghana http://ugspace.ug.edu.gh 101 beef. It is also noticed that Ghana does not have a comparative advantage in the production of all these commodities compared with the main SSA exporters of the commodities. So, for the AGOA and EU-ACP trade preferences to have significant effects on Ghana’s total bilateral exports, Ghana must negotiate for the extension of the product coverage and a relaxation of the RoO on commodities Ghana has a comparative advantage. Again, with the provision by the EPA to support the development of export capacities of beneficiary countries, Ghana has enough reasons, with the results of this study, to request for funds for the development of its export sector so that it can gainfully benefit from the EPA. For instance, Ghana has enough arable land that can support the cultivation of banana and cotton. So, Ghana can present proposals to the European Commission seeking for funding to help the nation develop the banana and cotton sub- sectors. Moving forward, it is recommended that neighbouring countries who share similar characteristics with Ghana should undertake similar studies to determine the extent to which their exports have benefited from the AGOA and EU-ACP preferences. The results (which I believe would be similar) could provide a good reason for such countries to come together as a group and collectively negotiate with the preference donors, rather than negotiating in regional economic blocs, which may not be universally beneficial. This can help influence the terms of the agreements to their advantage. University of Ghana http://ugspace.ug.edu.gh 102 6.4 Delimitation of Study The study is limited in dealing with the zero export flows in the estimation due to the sophisticated nature of the estimation technique, which is beyond the scope of this study. Also, owing to the unavailability of data on price terms, the study used a remoteness index to proxy for the multilateral resistance term. The study, therefore, recommends that future studies in this area can look at estimation techniques that capture zero trade flows. Also, in order to determine the product with which Ghana’s benefit under the preferential schemes is profound, future studies can consider a similar analysis using disaggregated export data of Ghana. Notwithstanding these limitations, the results of this study are valid. University of Ghana http://ugspace.ug.edu.gh 103 REFERENCES Afesorgbor, S. (2013). Revisiting the Effectiveness of African Economic Integration: A Meta-Analytic Review and Comparative Estimation Methods. Available at SSRN 2316160. African Growth and Opportunity Act Implementation Guide (October 2000). Agostino, M. R., Aiello, F., & Cardamone, P. (2007). Analyzing the Impact of Trade Preferences in Gravity models. Does Aggregation Matter? No. 7294, working papers, TRADEAG – Agricultural Trade Agreements. Aiello, F., & Demaria, F. (2009). Do Trade Preferential Agreements enhance the Exports of Developing Countries? Evidence from the EU GSP. Anderson, E. A., & Wincoop, E. V. (2003). Gravity with Gravitas: A Solution to the Border Puzzle. The American Economic Review, 93(1), 170, American Economic Association. Anderson, J. E. (1979). A Theoretical Foundation for the Gravity Equation. The American Economic Review, 106-116. Baboola, A., Reed, M., Saghaian, S., & Subramaniam, V. (2008). Food Safety Standards and Export Competitiveness in the Processed Food Industries of Asia- pacific countries. Journal of International Agricultural Trade and Development, 1556, 1. Baier, S. L., & Bergstrand, J. H. (2002). On the endogeneity of international trade flows and free trade agreements. University of Ghana http://ugspace.ug.edu.gh 104 Baier, S. L., and Bergstrand, J. H. (2007). Do Free Trade Agreements Actually increase members’ International trade?. Journal of International Economics, 71(1), 72-95. Baldwin, R. E. (2008). The Development and Testing of Heckscher-Ohlin Trade Models: A Review. The MIT Press, Cambridge, Massachusetts; London, England. Baltagi, B. H., Jung, B. C., & Song, S. H. (2008). Testing for Heteroskedasticity and Serial Correlation in a Random Effects Panel Data Model. Centre for Policy Research, paper 56. Retrieved from http://surface.syr.edu/cpr/56 Bartels, L. (2007). The trade and development policy of the European Union. European Journal of International Law, 18(4), 715-756. Battersby, B., & Ewing, R. (2005). International Trade Performance: the gravity of Australia's remoteness. Treasury. Bensassi, S., Jose de Sousa, & Jarreau, J. (2013). Preferential Trade Agreements Proliferation: Sorting out the effects. CEP II, WP No 04. Bergeijk, P. V., & Brakman, S. (2010). The Gravity Model in International Trade: advances and applications. Cambridge: Cambridge University Press. Bergstrand, J. H. (1985). The Gravity Equation in International Trade: some microeconomic foundations and empirical evidence. The Review of Economics and Statistics, 474-481. University of Ghana http://ugspace.ug.edu.gh 105 Bergstrand, J. H. (1989). The generalized gravity equation, monopolistic competition, and the factor-proportions theory in international trade. The Review of Economics and Statistics, 143-153. Bonarriva and Guth (2014). AGOA: Trade and Investment Performance Overview, USITC. Accessed from http://www.usitc.gov/publications/332/pub4461.pdf Bora, B., Kuwahara, A., & Laird, S. (2002). Quantification of non-tariff measures (No. 18). United Nations Publications. Bouët, A., Debucquet, D. L., Dienesch, E., & Elliott, K. (2010). The Cost and Benefits of Duty-Free, Quota-Free Market Access for Poor Countries: Who and What Matters. Working Paper 206: Center for Global Development, Washington DC. Brenton, P., & Manchin, M. (2003). Making EU Trade Agreements work: The Role of Rules of Origin. The world Economy: 755 - 69. Blackwell Publishing Ltd 2003, 9600 Garsington Road, Oxford, OX4 2DQ, UK. Brenton, P., & Özden, Ç. (2005). Trade Preferences for Apparel and the Role of Rules of Origin: The case of Africa. World Bank, Washington, DC. Brenton, P., and Hoppe, M. (2006). The African Growth and Opportunity Act, Exports, and Development in sub-Saharan Africa. World Bank Policy Research Working Paper 3996; International Trade Department; World Bank, Washington. Brenton, P., and Ikezuki, T. (2004). The Initial and Potential Impact of Preferenctial Access to U.S Market under the African Growth and Opportunity Act. World Bank Policy Research Working Paper 3262. University of Ghana http://ugspace.ug.edu.gh 106 Brenton, Paul (2003). Integrating the Least Developed Countries into the World Trading System: The Current Impact of EU Preferences under the Everything but Arms. Policy Research working paper No. 3018, Washington, DC; World Bank. Cadot, O., Carrère, C., De Melo, J., & Tumurchudur, B. (2006). Product-specific Rules of Origin in EU and US Preferential Trading Arrangements: an assessment. World Trade Review, 5(02), 199-224. Candau, F., & Jean, S. (2005). What are EU preferences worth for Sub-Saharan Africa and other developing countries? Working Paper No 2005-19, Centre D’études Prospectives et D’informations Internationales (CEP II). Cardamone, P. (2011). The Effect of Preferential trade agreements on monthly fruit exports to the European Union. European Review of Agricultural Economics, Vol 38 (4) pp. 553-586. Retrieved on April 28, 2015 from http://erae.oxfordjournals.org/content/38/4/553.short Carrère C., J. De Melo, B. Tumurchudur (2010). Disentangling Market Access Effects of Preferential trading arrangements with an application for ASEAN members under an ASEAN – EU FTA, The world economy 33 (1), 42-59. Chacholiades, M. (1990). International Economics. McGraw-Hill Publication Company. Chaney, T. (2011). The Network Structure of International Trade. National Bureau of Economic Research (No. w16753). Chang, H. J. (2002). Kicking away the Ladder: Development Strategy in Historical Perspective. Anthem Press. University of Ghana http://ugspace.ug.edu.gh 107 Chen, N. (2004). Intra-national versus International Trade in the European Union: why do national borders matter?. Journal of International Economics, 63(1), 93-118. Choi, I. (2001). Unit Root Tests for Panel Data. Journal of International Money and Finance 20,249–272. Cipollina, M., & Salvatici, L. (2011). EU Preferential Margins: measurement and aggregation issues, in Luca De Benedictis e luca Salvatici (eds.). The trade impact of European Union Preferential policies: an analysis through gravity models, Beriln/Heidelberg: Springer. Cipollina, M., Laborde, D., & Salvatici, L. (2013). Do Preferential Trade Policies (actually) increase Exports? An analysis of EU trade policies. Paper prepared for presentation at the Agricultural and Applied Economics Asssociation’s 2013 AAEA &CAES Joint Annual Meeting, Washington DC. Cline, W. R. (2004). Trade policy and global poverty. Peterson Institute. Collier, P., & Venables, A. J. (2007). Rethinking Trade Preferences: how Africa can diversify its exports. The World Economy, 30(8), 1326-1345. Condon, N., and Stern, M. (2011). The Effectiveness of African Growth and Opportunity Act (AGOA) in increasing Trade from Least Developed Countries; A systematic review. London: EPPI-Centre, Social Science Unit, Institute of Education, University of London. Cooke, E. F. A (2011). A Matching Approach to study the Impact of AGOA on Sub- Saharan African Countries. MPRA paper No. 34670. University of Ghana http://ugspace.ug.edu.gh 108 Crawford, J. A., & Fiorentino, R. V. (2005). The Changing Landscape of Regional Trade Agreements. Geneva: World Trade Organization. Dean, J., & Wainio, J. (2006). Quantifying the value of US Tariff preferences for developing countries. Paper presented at the Preference Erosion: Impacts and Policy Responses International Symposium, World Trade Organisation, Geneva, June 2005. Deardoff, A. (1997). Determinants of bilateral trade: does gravity work in a classical word? The Regionalisation of the World Economy, ed. Drukker, D. M. (2003). Testing for Serial Correlation in Linear Panel-data Models. The Stata Journal, 3(2), pp. 168-177. Dunlevy, J. A. (2006). The Impact of Corruption and Language on the Pro-Trade Effect of Immigrants: Evidence from the American States [J]. Rev Econ Stat,88(1). Duru, H. C. (2012). The Effects of the African Growth and Opportunity Act (AGOA): Evidence from Nigeria and South Africa exports (Doctoral dissertation, Texas Tech University). Egger, P. (2000). A note on the Proper Econometric Specification of the Gravity Equation. Economics Letters, 66(1), 25-31. Egger, P., & Pfaffermayr, M. (2003). Structural Funds, EU Enlargement, and the Redistribution of FDI in Europe, WIFO Working Papers, Nr. 195. European Commission (2014). The Cotonou Agreement and multiannual financial framework, 2014-20. University of Ghana http://ugspace.ug.edu.gh 109 Evans, C. L. (2003). The Economic Significance of National Border Effects’. American Economic Review, 93:4. pp. 1291-312 Feenstra, R. C., Markusen, J. R., & Rose, A. K. (2001). Using the Gravity Equation to differentiate among Alternative Theories of Trade. Canadian Journal of Economics, 430-447. Filippini, C., & Molini, V. (2003). The Determinants of East Asian Trade Flows: a gravity equation approach. Journal of Asian Economics, 14(5), 695-711. Frankel, J. (1997). Regional Trading Blocs in the World Economic System. Institute for International Economics, Washington D.C. Frazer, G., & Van Biesebroeck, J. (2010). Trade Growth under the African Growth and Opportunity Act. The Review of Economics and Statistics, 92(1), 128-144. Fugazza, M., and Nicita, A. (2011). Measuring Preferential Market Access. MPRA paper, No. 38565. Gamberoni, E. (2007). Do Unilateral Trade Preferences help Export Diversification? An investigation of the impact of European unilateral trade preferences on the extensive and intensive margin of trade. HEI Working Paper No: 17/2007. Graduate Institute of International Studies, Geneva Ghosh, S., & Yamarik, S. (2004). Are Regional Trading Arrangements Trade Creating?: An application of extreme bounds analysis. Journal of International Economics, 63(2), 369-395. University of Ghana http://ugspace.ug.edu.gh 110 Gil-pareja, S., Llorca-Vivero, R., & Martinez-Serrano, J. A. (2014). Do Nonreciprocal Preferential Trade Agreements increase Beneficiaries’ Exports? Journal of development economics 107 (2014) 291-304. Glick, R., & Rose, A. K. (2002). Does a Currency Union affect Trade? The time-series evidence. European Economic Review, 46(6), 1125-1151. Golhar, A. M. (1996). The Generalized System of Preferences and International Flows for the European Community 1971-1991. California State University. Dept. of Economics. Granger, C. W., & Newbold, P. (1974). Spurious regressions in econometrics. Journal of econometrics, 2(2), 111-120. Gujarati, D. N. (2003). Basic Econometrics Fourth Edition McGraw Hill Gujarati, DN, Basic Econometrics. Hausmann, R., & Klinger, B. (2006). The Evolution of Comparative Advantage: the impact of the structure of the product space. Centre for International Development and Kennedy School of Government Harvard University. Haveman, J. D., & Shatz, H. J. (2003). Developed Country Trade Barriers and the Least Developed Countries. WIDER’s Discussion Paper, (2003/46). Head K., & Mayer, T. (2013). Gravity Equations: workhorse, toolkit, and cookbook. In Handbook of International Economics Vol. 4, eds. Gopinath, Helpman, and Rogoff, North Holland. University of Ghana http://ugspace.ug.edu.gh 111 Head, K., & Mayer, T. (2000). Non-Europe: the magnitude and causes of market fragmentation in the EU. Weltwirtschaftliches Archiv, 136(2), 284-314. Head, K., T. Mayer & Ries, J. (2010). The Erosion of colonial Trade Linkages after Independence, Journal of International Economics, 81 (1), 1-14. Hesse, H. (2006). Export Diversification and Economic Growth. World Bank, Washington, DC. Hoekman, B., & Nicita, A. (2008). Trade Policy, Trade Costs, and Developing Country Trade. World Bank Policy Research Working Paper Series, Vol. Hoekman, B., & Nicita, A. (2011). Trade Policy, Trade Costs, and Developing Country Trade. World Development, 39(12), 2069-2079. Hoekman, B., & Ozden, C. (2005). Trade Preferences and Differential Treatment of Developing Countries: a selective survey. World Bank policy research working paper, (3566). Hoekman, B., & Özden, C. (2005). Trade Preferences and Differential Treatment of Developing Countries: A selective survey. World Bank Policy Research working Paper 3566. Retrieved from https://openknowledge.worldbank.org/bitstream/handle/10986/9264/wps3566.txt? sequence=2 on November 6, 2014. Hoekman, B., & Prowse, S. (2005). Policy Response to Preference Erosion: From Trade as Aid to Aid for Trade. Paper prepared for the international symposium on Preference Erosion: Impacts and Policy Responses, Geneva. University of Ghana http://ugspace.ug.edu.gh 112 Hoekman, B., Martin, W. J., & Braga, C. (2006). Preference Erosion: The Terms of the Debate. World Bank. Hsiao, C. (2007). Panel Data Analysis—Advantages and Challenges. Test,16(1), 1-22. http://agoa.info/about-agoa/country-eligibility.html http://trade.gov/agoa/eligibility/ Hummels, D. L. (1999). Toward a Geography of Trade Costs. Available at SSRN 160533. Hutchinson, W. K. (2002). Does Ease of Communication Increase Trade? Commonality of language and bilateral trade. Scottish Journal of Political Economy, 49(5), 544- 556. Johnson, H. G. (1967). Economic Policies toward Less Developed Countries. Keck & Lendle (2012). New Evidence on Preference Utilization. Kepaptsoglou, K., Karlaftis, M. G., & Tsamboulas, D. (2010). The Gravity Model Specification for Modelling International Trade Flows and Free Trade Agreement Effects: A 10-year Review of Empirical Studies. The Open Economics Journal, 3, 1-13. Krishna, K., & Krueger, A. (1995). Implementing Free Trade Areas: Rules of origin and hidden protection (No. w4983). National Bureau of Economic Research. Krueger, A. O. (1999). Trade Creation and Trade Diversion under NAFTA (No. w7429). National Bureau of Economic Research. University of Ghana http://ugspace.ug.edu.gh 113 Laaksonen, K., Mäki-Fränti, P., & Virolainen, M. (2007). Lomé Convention, Agriculture and Trade Relations between the EU and the ACP countries in 1975 – 2000. Working Paper 06/20, TRADEAG – Agricultural Trade Agreements. Langhammer, R. J. (1983). Ten years of the EEC's generalized system of preferences for developing countries: success or failure? (No. 183). Kiel Working Papers. Leamer, E. E., & Levinsohn, J. (1994). International trade theory: the evidence. National Bureau of Economic Research (No. w4940). Lederman, D., & Özden, Ç. (2004). US Trade Preferences: all are not created equal. Documentos de Trabajo (Banco Central de Chile), (280), 1. Leo, B., & Ramachandran, V. (2014). Getting Serious about Underperformance of the African Growth and Opportunity Act: Policy Options for Supporting Trade Potential in Africa. Rethinking US Development Policy, Center for Global Development. Low, P., Piermartini, R., & Richtering, J. (2005). Multilateral solutions to the erosion of non-reciprocal preferences in NAMA. Maddala, G. S., & Wu, S. (1999). A comparative study of unit root tests with panel data and a new simple test. Oxford Bulletin of Economics and statistics,61(S1), 631- 652. Mattoo, A., Roy, D., & Subramanian, A. (2002). The African Growth and Opportunity Act and its Rules of Origin Generosity Undermined?. University of Ghana http://ugspace.ug.edu.gh 114 Mátyás, L. (1997). Proper Econometric Specification of the Gravity Model. The world economy, 20(3), 363-368. McCallum, J. (1995). National Borders Matter: Canada-US regional trade patterns. The American Economic Review, 615-623. Mold, A. (2005). Trade Preferences and Africa: The State of Play and the Issues at Stake. Moon, S., Sekwat, A. & Stanley, R. E. (2004). Testing the Assumptions of Pooled Time Series Cross-Sectional Regression Analysis through ARIMA and Interrupted Time Series Modelling: Beck and Katz Revisited. Paper prepared for the Western Social Science Conference. Mueller, T. (2008). The Effect Of The African Growth And Opportunity Act (AGOA) On African Exports To The US. In Annual Meeting Of The ISA's 49th Annual Convention, Bridging Multiple Divides, Hilton San Francisco, CA, USA, Mar (Vol. 26, p. 2008). Nilsson, L. (2002). Trading relations: Is the Roadmap from Lomé to Cotonou correct?. Applied Economics, 34(4), 439-452. Nilsson, L. (2007). Comparative Effects of EU and US trade policies on Developing Country Exports. The European Union and Developing Countries: Trade, Aid, and Growth in an Integrating World, 49-70. Nilsson, L., & Davies, E. (2013). A Comparative Analysis of EU and US Trade Preferences for the LDCs and the AGOA beneficiaries. University of Ghana http://ugspace.ug.edu.gh 115 Nilsson, L., & Matsson, N. (2009). Truths and Myths about the openness of EU trade policy and the use of EU trade preferences. DG Trade European Commission Nijkamp Poot Work Pap. Nolte, S. (2002). From Lomé IV to Cotonou and EBA–An Analysis of Trade Preferences and Redistribution of Economic Benefits. Georg-August-Universität Göttningen. Nouve, K. (2005). Estimating the Effects of AGOA on African Exports using a Dynamic Panel Analysis. Available at SSRN 1026204. Nouve, K., & Staatz, J. M. (2003). Has AGOA increased agricultural exports from sub- Saharan Africa to the United States? (No. 11573). Michigan State University, Department of Agricultural, Food, and Resource Economics. Obembe, O. (2011). The Effects of AGOA on Trade and Foreign Direct Investment in Sub-Saharan Africa. Olken, B. A., & Jones, B. F. (2005). The Anatomy of Start-Stop Growth. National Bureau of Economic Research. Osakwe, P. N. (2007). Foreign Aid, Resources and Export Diversification in Africa: A new test of existing theories. Özden, Ç., & Reinhardt, E. (2004). First do no harm: The Effect of Trade Preferences on Developing Country Exports. Emory University Manuscript, April. Panagariya, A. (2002). EU Preferential Trade Arrangements and Developing Countries. The World Economy, 25(10), 1415-1432. University of Ghana http://ugspace.ug.edu.gh 116 Panagariya, A. (2003). Aid through Trade: An Effective Option?. International Trade, 403006. Patterson, G. (1965). Discrimination in International trade, The policy issues: 1945 – 1965. Pattillo, C., Gupta, S., & Carey, K. (2005). Sustaining and Accelerating Pro-Poor Growth in Africa (Washington: International Monetary Fund). Peridy, N., & Abedini, J. (2014). Trade Effects of Regional Integration in Imperfect Competition: Evidence from the Greater Arab Free Trade Area (GAFTA). International Economic Journal, 28(2), 273-292. Persson M. (2012). From trade preferences to trade facilitation: taking stock of the issues. Economics: the open-access, open-assessment E-journal, vol. 6, 2012-7. Persson, M. (2013). Trade Preferences from a Policy Perspective. Working Paper 2013:3, Department of Economics, School of Economics and Management, LUND University. Persson, M., & Wilhelmsson, F. (2007). Assessing the Effects of EU Trade Preferences for Developing Countries. In Y. Bourdet, J. Gullstrand, & K. Olofsdotter (Eds.), The European Union and developing countries: trade, aid and growth in an integrating world (pp. 29-48). Cheltenham, UK: Edward Elgar Publishing. Persson, M., & Wilhelmsson, F. (2013). EU Trade Preferences for and Export Diversification. IFN Working Paper No. 991, 2013; Research Institute of Industrial Economics, Sweden. University of Ghana http://ugspace.ug.edu.gh 117 Pöyhönen, P. (1963). A tentative model for the volume of trade between countries. Weltwirtschaftliches Archiv, 93-100. Prowse, S. (2010). Aid for Trade: Supporting Trade Preference Reform. Center for Global Development, Working Paper, No. 224. Radelet, S., & Sachs, J. D. (1998). Shipping costs, manufactured exports, and economic growth. Regression with Stationarity time series (Chapter two). Retrieved on 18th May, 2015 from academic.reed.edu/economics/parker/312/tschapters/S13_Ch_2.pdf. Riaz, M. (2013). Theoretical Groundwork for Gravity Model. Paper presented at the Proceedings of 3rd International Conference on Business Management, School of Business and Economics, University of Management and Technology, Lahore, Pakistan. Rose, A. K. (2002). Do we really know that the WTO increases trade? (No. w9273). National bureau of economic research. Rose, A. K., & Stanley, T. D. (2005). A Meta‐Analysis of the Effect of Common Currencies on International Trade. Journal of Economic Surveys, 19(3), 347-365. Rose, A. K., and van Wincoop, E. (2001). National Money as a Barrier to Trade: The real case for currency union. American Economic Review: Papers and Proceedings, 2001, 91(2), 386-90. Sapir, A. (1981). Trade benefits under the EEC generalized system of preferences. European Economic Review, 15(3), 339-355. University of Ghana http://ugspace.ug.edu.gh 118 Seyoum, B. (2007). Export performance of developing countries under the Africa Growth and Opportunity Act: Experience from US trade with Sub-Saharan Africa. Journal of Economic Studies, 34(6), 515-533. Shapouri, S., & Trueblood, M. (2003). The African Growth and Opportunity Act (AGOA): Does it really present opportunities. In International Agricultural Trade Research Consortium (IATRC) Conference, Capri, Italy. Silva, J. S., & Tenreyro, S. (2006). The Log of Gravity. The Review of Economics and statistics, 88(4), 641-658. Stimson, James A. (1985). “Regression in Space and Time: A Statistical Essay.” American Journal of Political Science. 4: 915-947. Supper, E. (2001). Is there effectively a level playing field for developing country exports?. UNCTAD, policy issues in International trade and commodities, study series No. 1. Tinbergen, J. (1962). An Analysis of World Trade Flows. Shaping the world economy, 1-117. Tomz, M., Goldstein, J. L., & Rivers, D. (2007). Do we really know that the WTO increases Trade? Comment. The American Economic Review, 2005-2018. Turkson, F. E. (2012). Trade Agreements and Bilateral Trade in Sub-Saharan Africa: Estimating the Trade Effects of the EU-ACP PTA and RTAs. CREDIT Research Paper (No. 12/07). University of Ghana http://ugspace.ug.edu.gh 119 Turkson, F. E. (2012). Trade Costs in Developing Countries (Doctoral dissertation). University of Nottingham. USITC (2014). AGOA: Trade and Investment Performance Overview. A United States International Trade Commission (USITC) report, No 4461, Washington. Van Grasstek (2003). The African Growth and Opportunity Act: a preliminary assessment. New York, US: UNCTAD. Verdeja, L. (2006). EU’s preferential trade agreements with developing countries revisited. Unpublished, University of Nottingham, School of Economics. Wei, S. J. (1996). Intra-national versus international trade: how stubborn are nations in global integration? National Bureau of Economic Research (No. w5531). Williams, B. R. (2014). African Growth and Opportunity Act (AGOA): Background and Reauthorization. Congressional Research Service Report. Wooldridge, J. M. (2003). Introductory Econometrics. South-Western College Publication. Wooldridge, J. M. (2010). Econometric analysis of cross section and panel data. MIT press. Wooldridge, J. M. (2012). Econometric analysis of cross section and panel data, 2nd eds. University of Ghana http://ugspace.ug.edu.gh 120 APPENDICES I : List of GATT/WTO Rounds of Negotiation No. ROUND HOST START YEAR GATT/WTO PERIOD (months) 1 Geneva I Switzerland 1947 GATT 7 2 Annecy France 1949 GATT 5 3 Torquay England 1950 GATT 8 4 Geneva II Switzerland 1956 GATT 5 5 Dillon Switzerland 1960 GATT 11 6 Kennedy Switzerland 1964 GATT 37 7 Tokyo Japan 1973 GATT 74 8 Uruguay Uruguay 1986 GATT 87 9 Doha Qatar 2001 WTO II : List of SSA countries under AGOA Eastern Africa Western Africa Central Africa Southern Africa Burundi* Benin* Angola* Botswana* Comoros* Burkina Faso* Cameroon* Lesotho * Djibouti* Cape Verde* Central African Republic Namibia* Eritrea Cote D’Ivoire* Chad* South Africa* Ethiopia* Gambia Republic of Congo* Swaziland Kenya* Ghana* DR Congo Madagascar* Guinea* Equatorial Guinea Malawi* Guinea-Bissau* Gabon* Mauritius* Liberia* Sao Tome and Principe* Mozambique* Mali* South Sudan** Rwanda* Mauritania* Seychelles* Niger* Somalia Nigeria* Tanzania* Senegal* Uganda* Sierra Leone* Zambia* Togo* Zimbabwe *Current beneficiary **South Sudan technically belongs to the northern Africa, but has been included in the SSA under AGOA. South Sudan was added after the 112th Congress when the amendments of the AGOA were effected.39 39 http://www.whitehouse.gov/the-press-office/2014/06/26/presidential-proclamation-agoa University of Ghana http://ugspace.ug.edu.gh 121 III : List of SSA countries, and their AGOA eligibility status Country Gained eligibility Lost eligibility Regained eligibility Angola 30th Dec. 2003 N/A Benin 2nd Oct. 2000 N/A Botswana 2nd Oct. 2000 N/A Burkina Faso 10th Dec. 2004 N/A Burundi 1st Jan. 2006 N/A Cameroon 2nd Oct. 2000 N/A Cape Verde 2nd Oct. 2000 N/A Central African Republic 2nd Oct. 2000 31st Dec. 2003 N/A Chad 2nd Oct. 2000 N/A Comoros 30th June, 2008 N/A Cote D’Ivoire 16th May, 2002 1st Jan. 2005 25th Oct. 2011 Djibouti 2nd Oct. 2000 N/A DR Congo 1st Jan. 2003 1st January, 2011 N/A Equatorial Guinea N/A Eritrea 2nd Oct. 2000 31st Dec. 2003 N/A Ethiopia 2nd Oct. 2000 N/A Gabon 2nd Oct. 2000 N/A Gambia 28th March 2003 1st Jan. 2015 N/A Ghana 2nd Oct. 2000 N/A Guinea 2nd Oct. 2000 23rd Dec. 2009 25th Oct. 2011 Guinea-Bissau 2nd Oct. 2000 20th Dec. 2012 1st Jan. 2015 Kenya 2nd Oct. 2000 N/A Lesotho 2nd Oct. 2000 N/A Liberia 29th Dec. 2006 N/A Madagascar 2nd Oct. 2000 23rd Dec. 2009 26th June, 2014 Malawi 2nd Oct. 2000 N/A Mali 2nd Oct. 2000 20th Dec. 2012 1st Jan. 2014 University of Ghana http://ugspace.ug.edu.gh 122 Country Gained eligibility Lost eligibility Regained eligibility Mauritania 2nd Oct. 2000 1st Jan. 2006 23rd Dec. 2009 Mauritius 2nd Oct. 2000 N/A Mozambique 2nd Oct. 2000 N/A Namibia 2nd Oct. 2000 N/A Niger 2nd Oct. 2000 23rd Dec. 2009 25th Oct. 2011 Nigeria 2nd Oct. 2000 N/A Republic of Congo 2nd Oct. 2000 N/A Rwanda 2nd Oct. 2000 N/A Sao Tome and Principe 2nd Oct. 2000 N/A Senegal 2nd Oct. 2000 N/A Seychelles 2nd Oct. 2000 N/A Sierra Leone 23rd Oct. 2002 N/A Somalia N/A South Africa 2nd Oct. 2000 N/A South Sudan 20th Dec. 2012 1st Jan. 2015 Swaziland 17th Jan. 2001 1st Jan. 2015 Tanzania 2nd Oct. 2000 N/A Togo 17th April, 2008 N/A Uganda 2nd Oct. 2000 N/A Zambia 2nd Oct. 2000 N/A Zimbabwe N/A N/A: Not applicable Source: Table was prepared by author from a combination of sources.40 40 African Growth and Opportunity Act Implementation Guide (October 2000). Leo, B., and Ramachandran, V. (2014). http://trade.gov/agoa/eligibility/ Proclamation 7350 of October 2, 2000; Federal Reserve. http://agoa.info/about-agoa/country-eligibility.html University of Ghana http://ugspace.ug.edu.gh 123 IV : List of ACP countries in the Cotonou Agreement AFRICA CARIBBEAN PACIFIC Angola Liberia Antigua & Barbuda Cook Islands Benin Madagascar Bahamas Fiji Botswana Malawi Barbados Kiribati Burkina Faso Mali Belize Marshall Islands Burundi Mauritania Dominica Micronesia Cameroon Mauritius Dominican Republic Nauru Cape Verde Mozambique Grenada Niue Central African Republic Namibia Guyana Palau Chad Niger Haiti Papau New Guinea Comoros Nigeria Jamaica Samoa Congo, Rep Rwanda St. Kitts and Nevis Solomon Islands DR Congo Sao Tome and Principe Saint Lucia Timor Leste Ivory Coast Senegal St. Vincent and the Grenadines Tonga Djibouti Seychelles Suriname Tuvalu Equatorial Guinea Sierra Leone Trinidad & Tobago Vanuatu Eritrea Somalia *Cuba Ethiopia South Africa Gabon Sudan Gambia Swaziland Ghana Tanzania Guinea Togo Guinea-Bissau Uganda Kenya Zambia Lesotho Zimbabwe *Not a signatory to Cotonou Agreement Source: ACP website (http://www.acp.int/node/7) University of Ghana http://ugspace.ug.edu.gh 124 V : Members of the EU 28 with the year they joined the EU in parenthesis EU 28 Belgium (1952) Sweden (1995) France (1952) Cyprus (2004) Germany (1952) Czech Republic (2004) Italy (1952) Estonia (2004) Luxembourg (1952) Hungary (2004) Netherlands (1952) Latvia (2004) Denmark (1973) Lithuania (2004) Ireland (1973) Malta (2004) United Kingdom (1973) Poland (2004) Greece (1981) Slovakia (2004) Portugal (1986) Slovenia (2004) Spain (1986) Bulgaria (2007) Austria (1995) Romania (2007) Finland (1995) Croatia (2013) VI : Ghana’s major trading partners and their share of trade Exports Imports Destination Share Source Share South Africa 27% China 20% UAE 9.9% United States 9.6% Switzerland 7.9% Belgium-Luxembourg 5.2% France 7.3% United Kingdom 5% Italy 6.7% Netherlands 5% Source: Observatory of Economic Complexity (OEC) http://atlas.media.mit.edu/profile/country/gha/ University of Ghana http://ugspace.ug.edu.gh 125 Appendix VII: List of Countries considered in the Analysis AFRICA Namibia Costa Rica Bangladesh Syria Luxembourg Angola Niger Cuba Bhutan Taiwan, China Macedonia, FYR Algeria Nigeria Dominica Brunei Tajikistan Malta Benin Reunion Dominican Republic Cambodia Thailand Moldova Botswana Rwanda Ecuador China Turkmenistan Netherlands Burkina Faso Sao Tome and Principe El Salvador Georgia United Arab Emirates Norway Burundi Senegal Greenland Hong Kong, China Uzbekistan Poland Cameroon Seychelles Grenada India Vietnam Portugal Cape Verde Sierra Leone Guadeloupe Indonesia EUROPE Romania Central African Republic Somalia Guatemala Iran, Islamic Rep. Andorra San Marino Chad South Africa Guyana Iraq Albania Slovak Republic Congo, Dem. Rep. Sudan Haiti Israel Austria Slovenia Congo, Rep. Swaziland Honduras Japan Belarus Spain Cote d'Ivoire Tanzania Jamaica Jordan Belgium Sweden Djibouti Togo Mexico Kazakhstan Bosnia Herzegovina Switzerland Egypt, Arab Rep. Tunisia Netherlands Antilles Korea, Democratic Republic Bulgaria Turkey Equatorial Guinea Uganda Nicaragua Korea, Rep. Croatia Yemen Ethiopia(excludes Eritrea) Western Sahara Panama Kuwait Cyprus United Kingdom Gabon Zambia Paraguay Lao PDR Czech Republic Yugoslavia, FR (Serbia/Montene Gambia, The Zimbabwe Peru Lebanon Denmark PACIFIC Guinea AMERICA St. Kitts and Nevis Macao Estonia Australia Guinea-Bissau Antigua and Barbuda St. Lucia Malaysia Faeroe Islands Cook Isds Kenya Argentina St. Vincent and the Grenadines Maldives Finland Fiji Lesotho Bahamas, The Suriname Myanmar France French Polynesia Liberia Barbados Trinidad and Tobago Nepal Germany Nauru Libya Belize United States Oman Gibraltar New Caledonia Madagascar Bermuda Uruguay Pakistan Greece New Zealand Malawi Bolivia Venezuela Philippines Hungary Norfolk Isds Mali Br. Virgin Isds ASIA Qatar Iceland Papua New Guinea Mauritania Brazil Afghanistan Russian Federation Ireland Samoa Mauritius Canada Armenia Saudi Arabia Italy Wallis and Futuna Isds Morocco Chile Azerbaijan Singapore Latvia Mozambique Colombia Bahrain Sri Lanka Lithuania University of Ghana http://ugspace.ug.edu.gh 126 VIII : Bilateral trade of the US with Ghana, Angola and Botswana (2000-2012) University of Ghana http://ugspace.ug.edu.gh 127 IX : Modified Wald test for groupwise heteroskedasticity in fixed effect model Ho: = for all i chi2 (181) = 30779.43 Prob > chi2 = 0.0000 X : Hausman Specification Test Coefficients Difference (b – B) Standard Errors Sqrt(diag(V_b-V_B) Fixed Effects (b) Random Effects (B) Log of GDPit 2.257212 -2.427184 4.684396 Log of GDPjt 0.5013135 0.7211574 -0.2198439 0.453479 Log of Populationjt 2.57592 1.071742 1.504177 0.1547397 Remotenessijt 8.030801 0.423902 7.606899 0.2137648 Common Currencyijt 0.3157595 0.1622375 0.153522 EUACPjt 0.1079525 -0.2899332 0.3978857 0.0501571 AGOAjt -0.7694312 -0.9312034 0.1617722 b = consistent under H0 and Ha; obtained from xtreg B = inconsistent under Ha, efficient under Ho; obtained from xtreg Test: Ho: difference in coefficients not systematic chi2(57) = (b-B)'[(V_b-V_B)^(-1)](b - B) = 575.70 Prob>chi2 = 0.0000 Therefore (V_b-V_B is not positive definite) University of Ghana http://ugspace.ug.edu.gh 128 XI : VIF test for Multicollinearity VARIABLE VIF 1/VIF Log of Populationit 11.19 0.089366 Log of GDPit 9.40 0.106383 Log of GDPjt 4.52 0.221239 Log of Populationjt 4.10 0.243902 Remotenessijt 3.73 0.268097 Log of Areaj 2.83 0.353357 Log of Distanceij 2.11 0.473934 Colonial Historyij 1.96 0.510204 Common Languageij 1.59 0.628931 Contiguityij 1.48 0.675676 EUACPijt 1.46 0.684932 Landlockedj 1.08 0.925926 Common Currencyijt 1.05 0.952381 AGOAijt 1.05 0.952381 Mean VIF 3.40 University of Ghana http://ugspace.ug.edu.gh