University of Southern California Dissertations and Theses

The gravity model at work: An empirical study on the effects of governance on bilateral trade flows with special emphasis on intra-Arab trade

(USC Thesis Other)

The gravity model at work: An empirical study on the effects of governance on bilateral trade flows with special emphasis on intra-Arab trade

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content THE GRAVITY MODEL AT WORK: AN EMPIRICAL STUDY ON THE EFFECTS OF GOVERNANCE ON BILATERAL TRADE FLOWS WITH SPECIAL EMPHASIS ON INTRA-ARAB TRADE Copyright 2004 by Rania Samir Miniesy A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (POLITICAL ECONOMY AND PUBLIC POLICY) August 2004 Rania Samir Miniesy Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UMI Number: 3145247 INFORMATION TO USERS The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleed-through, substandard margins, and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion. ® UMI UMI Microform 3145247 Copyright 2004 by ProQuest Information and Learning Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. ProQuest Information and Learning Company 300 North Zeeb Road P.O. Box 1346 Ann Arbor, Ml 48106-1346 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. DEDICATION To my father, who taught me how to love God, people, life and knowledge. To the beloved members of my family, who are truly the backbone of my life. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ACKNOWLEDGEMENTS First of all, I would like to thank God for never ever letting me down, for always being with me, especially in the hardest times, and for granting me more than I wished for. My wholehearted thanks go to my dissertation advisor and committee chair, Dr. Jeffrey Nugent, for all his support, encouragement, guidance, and insights not just with my research but also throughout my years at USC. The idea and a part of this research is a result of a research paper for which professor Nugent, thankfully, took me as a co-author. My advisor also took me as a co-author in a few other research papers and for this great trust and faith in my work, I thank him a lot. Moreover, professor Nugent was always helpful, kind, and understanding, especially during my last episode of back problems as well as my pregnancy, where he was more like a father than a professor. I would also like to thank Dr. Laurie Brand and Dr. Rod McKenzie for kindly finding time to serve on my qualification exam guidance committee as well as on my dissertation committee and also for their constructive feedback on earlier drafts of the dissertation. Extended thanks go to professor Brand for being a friend and for the many corrections that she provided in my almost final draft. I would also like to thank Dr. Nora Hamilton, Dr. Howard Sherman, and Dr. Laurence Shute for serving on my qualification exam guidance committee and providing invaluable comments on my dissertation proposal. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. iv Many thanks go to Dr. Farideh Motamedi for helping me as well as many other PEPP students in fulfilling the program requirements - without any obstacles - and thus making our lives much easier. Special thanks go to the late Dr. John Elliott for directing the PEPP program. I would also like to thank Ms. Young Miller and Ms. Sheila Williams for their help with many administrative issues. Many thanks go to all the professors who instructed me in the Economics, International Relations, and the Political Science departments and from whom I learned a lot, especially Dr. Timur Kuran, Dr. Hyward Alker, Dr. Todd Sandler, and Dr. Sunhyuk Kim. I would also like to thank all my classmates who were always kind and helpful. I also thank many of my friends who gave me constant support and encouragement, especially Ola Darwish, Bahira Metwally, Mariam Saada, and Hania Abou alShamat I would also like to thank USAID/Egypt for sponsoring me for four years and the Graduate School for granting me the Final Year Dissertation Fellowship. I would have never been able to start, continue, and finish this Ph.D. without the love, help, support, and encouragement of my mother, Nadia Mostafa, who left Egypt so many times to come and stay with me to grant me all her unconditional love, support, and prayers. Besides carrying out all my house chores, and taking full care of my now toddler son, she frequently typed my term papers, and always always uplifted my spirits whenever I was down. Moreover, she did all that with a big smile. She has always been my best friend and she will always remain to be. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. V My deepest gratitude and heartfelt thanks go to my husband and my best friend, Islam Kandil, for first leaving his job to support me as I was realizing my goal - acquiring a Ph.D.; for the many hours he took out of his leisure to help me with his computer expertise in the collection and manipulation of the data; and for again taking time off from his job to accompany me in this last trip to Los Angeles to defend my dissertation. Besides all that, he supported me, emotionally, during all the ups and downs of my Ph.D. I owe the rest of my family all my thanks for their encouragement, support, prayers, and for having faith in me, especially my brother, Dr. Ehab Miniesy; his wife, Gihan Raafat; their two wonderful daughters, Nada and Shams; my uncle, Tag Eldin Bartlett; and my grandmother, Samira Ghaly. I also thank my in-laws, especially my mother-in-law, Nagwa ElMaghrabi, for being there for me, especially during my last episode of back problems and also to my two sisters-in-law, Amira and Heba Kandil, for helping me with my son, thus giving me more time to study. Last but not least I would like to thank my son, Ali Kandil, for just being there. His sweet innocent smile gave a new meaning to my life, put everything in perspective, and gave many things in life their proper weights. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. TABLE OF CONTENTS Dedication.................................................................................................................................ii Acknowledgements............................................................................................................... iii List of Tables........................................................................................................................ viii Abstract...................................... xi 1. Introduction.......................... 1 2. The Gravity M odel..............................................................................................................6 2.1 Theoretical Foundations and Empirical Success...................................................... 6 2.2 Literature Review........................................................................................................10 2.2.1 On Economic Factors..........................................................................................11 2.2.1.1 Regional Trade Agreements....................................................................... 11 2.2.1.2 Exchange Rate Volatility............................................................................ 15 2.2.1.3 Currency Union............................................................................................18 2.2.1.4 Other Economic Variables.........................................................................20 2.2.2 On Political Factors.................................. 21 2.2.3 On Other Factors............................................................................. 25 2.2.4 Specifically on Intra-Arab Trade.......................................................................31 2.3 Purpose of this Study and Contribution to the Literature......................................33 3. Governance and World Trade................................................................................ 36 3.1 The M odel................................................................................................................... 36 3.2 The Data.......................................................................................................................40 3.3 Expected Signs........................................................................... 49 3.4 World Trade Results....................................................................... 72 3.4.1 Table by Table Results and Interpretations............................ 72 3.4.2 Summary of Results................................... 108 4. Governance and Arab Trade .......................... I l l 4.1 Arab Regionalism / Arab Economic Integration............. I l l 4.1.1 History of Failures - When, What, and W hy?........................ 112 4.1.2 Any Prospects for Intra- Arab Trade / Arab FTA?..................... 117 4.1.3 Renewed Interest in Arab Economic Integration.........................................119 4.1.4 Intra-Arab Trade....................... 122 4.1.4.1 Recent Trends in Intra-Arab Trade ..................................... 122 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.1.4.2 Explanations for the Low Intensity of Regionalism / Low Trade 126 4.1.4.3 Is Intra-Arab Trade Low? Empirical Results from Old and New Data ...................................................................................................................................130 4.2 Gravity Model and Arab Trade............................................................................ 131 4.2.1 Intra-Arab Trade Results................................................................. 131 4.2.2 Arab-World Trade Results.................................................. 156 4.2.3 Intra-Arab and Arab-World Results Comparison................. 175 5. Conclusion............................................................... 178 5.1 Summary................................................................................................................... 178 5.2 Policy Implications................................................................................................... 190 5.3 Further Research ........................................................................................204 Bibliography ....... 206 Appendices...........................................................................................................................219 Appendix A: List of Countries and Summary Statistics........................................... 219 Appendix B: Data Manipulation...................................................................................223 Appendix C: Experimentation with the Four Governance Variables.......................233 Appendix D: Some Important Aspects of GAFT A .................................................... 247 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. v iii LIST OF TABLES Table 3.1. Regression results from estimating the gravity model for World total bilateral trade using OLS, where the null values in the dependent variable are assumed genuinely missing..........................................................................................73 Table 3.1b. Predictions based on column 6 Table 3.1...................................................... 80 Table 3.2. Regression results from estimating the gravity model for World total bilateral trade using OLS, where the null values in the dependent variable are assumed zeros ................. 82 Table 3.2b. Predictions based on column 6 Table 3.2...................................................... 88 Table 3.3. Regression results from estimating the gravity model for World total bilateral trade using Tobit, where the null values in the dependent variable are assumed zeros.................................................................................................................89 Table 3.4. Regression results from estimating the gravity model for World total bilateral trade using OLS, where a lagged dependent variable is included as one of the explanatory variables......................................................................................... 92 Table 3.5. Regression results from estimating the gravity model for World total bilateral trade for pooled data (1985-2000) using OLS, where each of the six governance components is used separately and where the null values in the dependent variable are assumed genuinely missing..................................................95 Table 3.6. Regression results for estimating the gravity model for World energy trade using OLS......................................................................................................................100 Table 3.7. Regression results for estimating the gravity model for World other trade using OLS.......................................................................................... ........................ 103 Table 3.8. A summary for World trade regression results......................... 109 Table 4.1. Direction of Arab Trade, 1998 (In billions of U.S. dollars)........................ 123 Table 4.2. Trends in Intra-Regional Trade, 1970-98 (As a share of total exports in the region) ...... 124 Table 4.3. Indicators of Intra-Arab Trade, 1998 ............ 125 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ix Table 4.4. Regression results from estimating the gravity model for intra-Arab total bilateral trade using OLS, where the null values in the dependent variable are assumed genuinely missing........................................................................................132 Table 4.4b. Predictions based on Table 4.4, column 6 ..... 138 Table 4.5. Regression results from estimating the gravity model for intra-Arab total bilateral trade using OLS, where the null values in the dependent variable are assumed zeros............................................................. 140 Table 4.5b. Predictions based on Table 4.5 column 6 .....................................................144 Table 4.6. Regression results from estimating the gravity model for intra-Arab total bilateral trade for pooled data (1985-2000) using OLS, where each of the six governance components is used separately and where the null values in the dependent variable are assumed genuinely missing........................... 146 Table 4.7. Regression results from estimating the gravity model for intra-Arab energy trade flows using OLS for pooled data (1985-1997).............................................. 151 Table 4.8. Regression results from estimating the gravity model for intra-Arab other trade flows using OLS for pooled data (1985-1997).............................................. 153 Table 4.9. Regression results from estimating the gravity model for Arab-World total bilateral trade using OLS, where the null values in the dependent variable are assumed genuinely missing...................................................................................... 157 Table 4.10. Regression results from estimating the gravity model for Arab-World total bilateral trade using OLS, where the null values in the dependent variable are assumed zeros...............................................................................................................159 Table 4.9b. Predictions based on Table 4.9 column 6 .....................................................164 Table 4.10b. Predictions based on Table 4.10 column 6 ................................................ 165 Table 4.11. Regression results from estimating the gravity model for Arab-World total bilateral trade for pooled data (1985-2000) using OLS, where each of the six governance components is used separately and where the null values in the dependent variable are assumed genuinely missing ...... 167 Table 4.12. Regression results from estimating the gravity model for Arab-World energy trade flows using OLS for pooled data (1985-1997).................... 171 Table 4.13. Regression results from estimating the gravity model for Arab-World other trade flows using OLS for pooled data (1985-1997)....................... 173 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. X Table 4.14. A summary for Intra-Arab trade regression results....................................176 Table 4.15. A summary for Arab-World trade regression results.................................177 Table 5.1. Summary & comparison for World, Intra-Arab and Arab-World trade regression results.............................. 183 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. xi ABSTRACT Does governance affect international trade? In other words, does the existence/lack of voice and accountability, political stability, government effectiveness, regulatory quality, rule of law, and control of corruption within two countries enhance/restrict their bilateral trade? This dissertation aims to answer this question, with special emphasis on intra-Arab trade. Good governance is essential for Arab countries’ economic and social development, it is one of four factors cited as the most important changes needed to improve living standards throughout the Middle East and North Africa. Whether governance affects/is affected by economic development has been one of the hottest research subjects in both comparative politics and political economy in the past twenty years. Nevertheless, its effect on international trade has been largely ignored. The main purpose of this study is to utilize a gravity-type model to investigate just that. In addition to the gravity model variables commonly used, the model includes a governance variable, which combines the six individual governance components mentioned above, and some new and rarely used variables such as variables measuring income inequality (Gini), financial development (M2GDP), whether economies are and were ever centrally planned (CPY and CPE), and whether countries have account restrictions (current and capital). This study includes data for 186 countries for 1985, 1990, 1995, 1997 & 2000, a year which no gravity studies have included. Moreover Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. this study places special emphasis on intra-Arab trade, an area largely ignored. As such, these are my modest contributions to the literature. The parameters of the model are estimated for World, Intra-Arab, and Arab- World trade flows using OLS. Sensitivity analyses were carried out (different assumptions, estimation techniques, and specifications) to explore the robustness of the findings. Of the seven variables that represent this study’s contribution, governance is the variable that is the most significant and the most robust and that has a positive effect on all trade flows, with rule of law and control of corruption being the most important components. CPE significantly impedes Intra-Arab and Arab-World trade. Current significantly reduces only Intra-Arab trade. Capital impedes World and Arab-World trade. Concerning M2GDP, it significantly fosters only Arab-World trade. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 “Without good governance, without the rule of law, predictable administration, legitimate power, and responsive regulation - no amount of funding, no short-term economic miracle will set the developing world on the path to prosperity” (Kofi Annan 1997 in Ridgway and Talib 2003). 1. INTRODUCTION Do aspects of domestic governance affect international trade? In other words, does the existence or lack of voice and accountability, political stability, government effectiveness, regulatory quality, rule of law, and control of corruption within a pair of countries enhance or restrict trade flows between them? The objective of my dissertation is to provide an answer to this question, with special emphasis on intra- Arab trade. The concept of good governance is becoming increasingly important worldwide. Indeed, a number of international organizations use this concept to refer to the high standards they insist upon from states participating in the global trading regime and other international activities. For instance, trading blocs such as the North American Free Trade Agreement (NAFTA) and the European Union (EU) have repeatedly asserted that a commitment to good governance is a prerequisite for increased trade relations (Najem 2003). Good governance is one of four factors that have been cited as the most important changes needed to improve living standards throughout the MENA (Middle Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. East and North Africa) region over the coming two decades.1 It is not a separate challenge, to be worked on at its own pace, but rather, it complements and reinforces other reform efforts in private investment, trade, and economic diversification. Good governance not only improves capacity and incentives within governments, but also fosters a larger role for civil society in governance. Although better governance cannot ensure optimal economic policies, it is essential to guard against persistently poor policies and to guarantee that the good policies needed to meet MENA’s growth potential enjoy legitimacy and are faithfully and speedily implemented (World Bank 2003b: 8). Good governance is essential for Arab countries’ economic and social development. In the field of economic development, better governance means improving the environment for productive investment through: reducing the scope for arbitrary government policies, reducing uncertainty and costs of doing business, and ensuring effective public services for businesses. For social development, better governance improves delivery of public services, especially education, health, telecommunications, and water and sanitation. Contrary to c o m m on perceptions, the weaknesses in delivery in these services do not stem from technical weaknesses in capacity, but rather from problems in governance, where weak accountability mechanisms permit weak performance, especially as the public good or service 1 The other three factors concern: unlocking the employment potential, implementing gender equality especially in economic activities, and promoting trade liberalization and creating investment opportunities. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. becomes more complex. Effective accountability for each of the major players in service delivery will address these deficiencies and thus promote social development.2 Given the importance of good governance, highlighted above, its effect on international trade has been largely ignored and most studies concentrated on the effects on trade of mainly economic factors. There is no doubt that economic considerations influence international trade, but economics does not narrate the whole story, politics also plays a role and so do other factors. Some trade relations cannot just be simply explained “by the economic invisible hand of comparative advantage. Political and cultural hands are all too visible” (Bliss and Russett 1998:1127-28). In spite of the growing attention in the social sciences to the interdependence between economic and political spheres, empirical work in this area especially concerning domestic governance and international trade is limited. This dissertation is a contribution to the political economy of trade that integrates knowledge and insights from both the economics and the political science disciplines. Many variables, especially economic ones, have been used to help understand bilateral trade flows. Economic variables include but are not limited to: Gross Domestic Product (GDP), population, per capita GDP (GDPPC), distance as a proxy for transportation costs, Free Trade Agreements (FTAs), exchange rate regime, exchange rate volatility, currency union, current and capital account restrictions, 2 The three major players in service delivery are: the policymakers, the service providers, and the citizens. Effective accountability will improve accountability between policymakers and service providers, service providers and clients, and clients and policymakers (World Bank 2003a: 12-14). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 income growth and convergence, tariff rates, FDI and many others. Some political variables have also been utilized such as: common colonizer, colonial relationships, democracies or political similarity, foreign policy orientation, political disputes, conflict and cooperation, alliances, defense pacts, —etc. Still other factors that can affect bilateral trade flows have been included such as relative location, immigration, infrastructure, common border, language and intellectual property rights. Studies that used the above mentioned variables utilized a gravity-type model to examine the effect of such variables on bilateral trade flows. To the best of my knowledge, no study has attempted to explore the effects on the volume of international trade of governance variables such as government effectiveness, regulatory quality, rule of law, and control of corruption. Very few studies investigated the effects of democracy and political stability on international trade. The objective of this dissertation is to employ a gravity equation using some of the frequently used factors and the governance variables to examine their effect on trade flows, especially intra-Arab trade. In addition to the governance variables, this study will employ some variables that have either never been used before such as a measure for income inequality, and a measure for financial development, or have been used in a couple of studies, such as two measures for whether economies concerned are centrally planned and were ever centrally planned, and two measures for whether countries have current and/or capital account restrictions. This study will include data for 100+ countries from 1985 up until and including 2000, a year no previous gravity studies have included. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5 This dissertation is composed of five chapters, this being Chapter 1. Chapter 2 will provide evidence for the theoretical foundations of the gravity model and underline its empirical success, present a comprehensive literature review of some of the important studies that have employed the gravity model as applied to trade flows and the most important variables they investigated, and highlight the purpose of this study as well as the contribution to the literature. Chapter 3 will present the augmented gravity model and the sources of the data, discuss the expected signs of the coefficients of the variables used in the model based on theoretical foundations and previous empirical findings, and display the results and the interpretations of estimating the model on World bilateral trade flows under a number of assumptions, using different estimation techniques and employing different specifications. Chapter 4 will discuss in great depths Arab regionalism, present separately the results of the estimations carried for intra-Arab and Arab-World trade flows, and compare these results. Chapter 5, the concluding chapter, will provide a short summary of the main results, discuss some policy implications, and highlight areas for further research. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6 2. THE GRAVITY MODEL This chapter is composed of three sections: the first section is concerned with briefly providing evidence for the theoretical foundations of the gravity model and underlining its empirical success. The second section will present a literature review of some of the important studies that have employed the gravity model as applied to trade flows, and that represent a good summary for most of the work done and variables investigated in this area. As for the third section, it will highlight the purpose of my study as well as my contribution to the literature. 2.1 Theoretical Foundations and Empirical Success The gravity model has been long recognized for its consistent empirical success in explaining various types of flows, such as migration, commuting, tourism, commodity shipping, and bilateral trade flows. Typically, the log-linear equation specifies that a flow from origin x to destination y can be explained by economic forces at the flow’s origin and destination points, as well as economic forces assisting or obstructing the flow’s movement from origin to destination (Bergstrand 1985). Regarding bilateral trade flows, the gravity model is an empirical model that in its simplest form models the flow of trade between a pair of countries as being proportional to their mass, i.e. national income, and inversely proportional to the distance between them. The gravity model acquired its name from a similar function Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7 in Newtonian physics, which describes the force of gravity (Rose 2000). Distance belongs to the vector of factors that either enhance or resist trade. Physical distance is indeed the most common resistance factor and it is included in virtually all studies that use the gravity model to explain bilateral trade flows. It is included mainly as a proxy for transportation costs. Other resistance/enhancement factors have been used as mentioned in the introduction in chapter one and as will be explained in more detail below. What is most appealing, in my opinion, about the gravity model is that its origin as an means of analyzing bilateral trade flows is not a hypothetical deductive development but rather an intuitive approximation, as Sanso, Cuairan, and Sanz (1993) claimed. The fact that the gravity model, as applied to bilateral trade flows, has a remarkably consistent history of success as an empirical tool, which is not usual for economics, is also appealing. According to Deardorff (1984), the empirical success of the gravity equation is due to the fact that it can explain some real phenomena, such as the trade between industrialized countries, the intra-industry trade and the lack of dramatic reallocation of resources when trade liberalization processes have taken place, that the conventional factor endowment theory of international trade cannot. Note, however, that the utilization of the gravity model for predictive purposes has been inhibited owing to an absence of a strong theoretical foundation despite the model’s consistently high statistical explanatory power. However, Helliwell (1998) observed that in recent years the model has undergone a transformation from being “a Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 8 theoretical orphan” to one that can be derived from standard trade theories. The fact of the matter is that due to the gravity model’s empirical success, there have been numerous attempts to shed some light on its economic underpinnings. For example, Linnemann (1966) showed how the standard gravity model could be derived from a quasi-Walrasian general equilibrium model of export supply and import demand. Learner and Stem (1970) showed how a probability model of trade patterns could be used to derive the gravity model. Anderson (1979) suggested a theoretical foundation in terms of an expenditure system with goods differentiated by countries of origin. Bergstrand (1985) used a general equilibrium world trade model from which he derived a gravity equation by making certain assumptions, including perfect international product substitutability. With nationally differentiated products, Bergstrand was also able to present empirical evidence supporting the notion that the gravity model is a reduced form of a partial equilibrium subsystem of a general equilibrium model. Helpman and Krugman (1985) offered a theoretical foundation by embedding the gravity equation in a model of monopolistic competition with increasing returns to scale, which also has the advantage of yielding predictions regarding the sectoral pattern of trade. As a result, Deardorff (1995) argued that the claim that the gravity model of bilateral trade flows has no theoretical foundation is not true. In fact, he showed that the gravity model is consistent with the Heckscher-Ohlin theory of trade - both with frictionless and impeded trade. Anderson and Van Wincoop (2001) refined Bergstrand’s (1985) analysis by incorporating the ‘ relative distance effect’, which Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 9 underlines the likelihood that trade between two peripheral countries - geographically speaking - would be greater than between two core countries, after controlling for bilateral distance and country size. Other theoretical foundations were also provided in Bergstrand (1989&1990) and Eaton and Kortum (2001). Rose (2000), summarizing the attempts of many scholars to find theoretical foundations for the gravity model, stated that theoretical aspirants have claimed the singular empirical success of the gravity model. These include: the ‘Armington’ model of nationally differentiated goods; models with increasing returns and monopolistic competition; models with national technological differences; models with reciprocal dumping; models of homogenous goods; and models with internationally varying factor endowments. However, he asserted that the question of which theoretical model best describes the empirical findings of the gravity model is a matter of dispute. Baier and Bergstrand (2001) observed that studies that addressed the gravity model’s theoretical foundations are usually offered as theoretical substitutes where one chooses his preferred set of assumptions and model. They, however, argued and showed that these theoretical foundations are complementary to one another, each being a special case of a more general model. The important point here is that it could be argued that the gravity model stands firmly on theoretical grounds and that “it can no longer be said that the gravity model exists in a theoretical vacuum” (Greenaway and Milner 2000: 579). What impresses me and many others about the model is that, as Learner and Levinsohn (1995, p. 1384) put it, the gravity model provides “some of the clearest and Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 10 most robust empirical findings in economics” regardless of whether or not it has theoretical foundations. It is, as Bayoumi and Eichengreen (1997: 142) termed it, the “workhorse for empirical studies of the pattern of trade” and as Rauch (1999: 10) noted the “ standard empirical framework used to predict how countries match up in international trade.” 2.2 Literature Review The very basic gravity model, as mentioned earlier, requires the operationalization of the trading partners’ mass, or in other words economic size, and the distance between them. For the first purpose almost all studies employ Gross Domestic/National Product (GDP/GNP). Many of these studies in addition use either population or GDP/GNP per capita as a second indicator of economic size. In short, GDP/GNP and distance are the common factors in virtually all studies that explain bilateral trade flows through the gravity equation. As the objective of one study differs from another, so does the choice of the other variables to be included in the model. Some scholars use only economic factors to investigate their effect on bilateral trade flows, others use political factors only, others mix and match between economic and political factors, and still others use other factors. In what follows, I will present a literature review on, in my opinion, the most important studies and variables used, and some in more detail than others. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 11 2.2.1 On Economic Factors 2.2.1.1 Regional Trade Agreements Gravity models have been extensively used to evaluate the effects of regional trading arrangements, (RTAs), on trade, especially over the last ten - fifteen years. Most gravity analyses of RTAs usually address one or more of the following issues: whether there is a regional bias to trade, meaning whether regional blocks are a natural feature of international trade with near neighbors; whether there is an identifiable RTA effect, meaning whether intra-regional trade is stimulated by the formation of an RTA; if an RTA is concluded how much additional intra-regional trade might be expected, usually referred to as ‘trade potential’; and whether the formation of an RTA creates a domino effect, where non-members feel threatened by such RTA and thus either petition for membership or form their own RTA (Greenaway and Milner 2002). Aitken (1973) addressing the first two issues found that membership in the European Economic Community (EEC), the European Free Trade Area (EFTA) and a neighboring country dummy variable significantly influence European trade. Geraci and Prewo (1977) found trade between the Organization for Economic Cooperation and Development (OECD) countries to be significantly influenced by average tariff rates, preferential trading group membership and c o m m on language. A neighboring country dummy variable however was not found to be significant. On the first two issues also, Frankel, Stein and Wei (1995), after controlling for distance, common borders and common language, investigated whether Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 12 geographical clusters of countries trade more with each other even in the absence of a formal RTA. They found that in some regions as in Latin America and Western Europe that was true but in other areas, such as East Asia and North America, that wasn’t the case. The authors also explored whether commitments to an RTA exaggerated this regional bias and they found that it did in all cases and is of growing significance through time even in the European Union (EU). Other studies that addressed the regional bias and the trade effects of RTAs and that are listed in Greenaway and Milner (2002) include Frankel and Wei (1993); Frankel (1993); Bayoumi and Eichengreen (1998); Dhar and Panagariya (1999); Sharma and Chua (2000) and Hassan (2001). On the trade effect of RTAs, one of the variables I use in my model, Aitken and Obutelewicz (1976) found preferential agreements to positively influence trade between European and African countries. Brada and Mendez (1985) examined six integration schemes: namely, two developed country schemes (EEC and EFT A); three developing country schemes (the Central American Common Market (CACM), the Latin American Free Trade Agreement (LAFTA) and the Andean pact); and one centrally planned country scheme (the Council for Mutual Economic Assistance (CMEA)). They investigated and decomposed their ability to increase inter-member trade into environmental, policy, and system effects. They found that environmental factors caused the greatest variation in trade creation, inter-member distance being the most important environmental factor. They also, interestingly, found that although the CMEA was not Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 13 able to achieve its full potential for increasing inter-member trade because of a combination of policy and system effects associated with these centrally planned economies, the CMEA’s effectiveness still did not differ from that of unions among market economies. Pollins (1989) criticized the way in which the effects of economic integration on trade are usually measured for their heavy reliance on the use of dummy variables to distinguish members from non-members. He argued that this allows no means for distinguishing between the highly cooperative nation pairs and the weakly cooperative ones. He thus introduced his weighted cooperation sent variable that is used in a gravity equation that contains dummy variables for a number of trade agreements, namely the General Agreement on Tariffs and Trade (GATT), LAFTA, CMEA and EC. Such dummy variables are used as control variables in the exploration of the effect of conflict and cooperation. Still, he found that the coefficients for these control variables were all positive except for LAFTA, which was negative, but only in its early years. The conclusion of his article being that trade flows were influenced significantly by broad political relations of amity and enmity between countries. Nations adjusted trade ties to satisfy security and economic welfare goals as well. Mansfield and Bronson (1997a) analyzed the effects of not only preferential trading arrangements but also alliances on bilateral trade flows because they believed that their interaction is central to explaining patterns of commerce. The authors employed a gravity model using data on a large number of countries for the period 1960 - 1990 (five years interval). They found that both factors positively affected Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 14 trade and were significant, and alliances that included a major power conducted significantly more trade than their non-major power counterparts. What is more interesting, however, is their finding that countries that were both parties to a common preferential trading arrangement and a common alliance engaged in substantially greater trade than did members of either type of institution alone. There are other studies that have investigated the trade effects of RTAs and found them to be absent and sometimes even negative. For instance, Finger, Ng and Soloaga (1998) and Soloaga and Winters (1999) found no effects for the Southern Common Market (MERCOSUR), and neither did Sharma and Chua (2000) for the Association of South East Asian Nations (ASEAN) nor Flassan (2001) for either ASEAN or (South Asian Association for Regional Cooperation) SAARC. On the issue of trade potential, this has been most widely applied to preferential trade arrangements between the EU and the Central and Eastern European Countries (CEECs), with the works of Hamilton and Winters (1992) and Nilsson (2000) being good examples. Their main finding was that the RTA arrangements that have been put in place to prepare for the accession of transition economies, in particular the Europe Agreements have caused substantial growth in EU-CEEC trade. The conclusion being that the expected effects of further EU enlargement to the east will be modest since most adjustment has already occurred. As for the issue of the domino effect, Greenaway (2000) through focusing on all RTA’s on the period 1965-1993 and Sapir (2001) through focusing on EU and EFTA on the period 1960-1992 found evidence that supported the idea that domino Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 15 effects may have been important in stimulating EU enlargements and the creation of new RTAs around the world. 2.2.1.2 Exchange Rate Volatility The breakdown of the Bretton-Woods system of fixed exchange rates and the beginning of the generalized floating rates in 1973 have led policy makers and researchers to investigate the nature and extent of the impact of exchange rate volatility on the volume of international trade. Since that time, a debate began especially within economic circles and that is still going on because there is no consensus among economists regarding how exchange rate volatility affects the volume of trade from either a theoretical or an empirical perspective. In this part, we focus exclusively on empirical studies making use of the gravity model. The paper will shed light on some of the theoretical work as well as some of the other empirical studies that investigated the expected sign of exchange rate volatility without making use of the gravity model in the next chapter. Brada and Mendez (1988) used a gravity model to investigate the effect of exchange rate regime and exchange rate volatility on bilateral trade flows. They argued that the unanticipated exchange rate volatility that is associated with a freely floating exchange rate regime could result in governments erecting trade barriers to offset the destabilizing effects in exchange rates. These trade barriers would result in a lower volume of trade flows. Even though exchange rate volatility reduced trade, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 16 Brada and Mendez (1988) asserted that its effect is smaller than that of restrictive commercial policies. D ell’Ariccia (1999) analyzed the effects of exchange rate volatility on bilateral trade flows through the use of a gravity model and panel data from Western Europe for the period 1975-1994. He found that exchange rate uncertainty depressed international trade. However, the negative effect of such volatility on trade was very small. What is more interesting in this article is that Dell’Ariccia tested his model using different volatility measures, different temporal windows and both real and nominal exchange rates, and still found his results to be robust, all the coefficients of exchange rate volatility were negative and significant at 1% level. Nilsson and Nilsson (2000) investigated the net effects of developing countries’ choice of exchange rate regime on their exports. They classified the exchange rate regimes into six categories, ranging from single-currency pegging to independently floating and placed each of them as a dummy variable in a gravity model that is estimated in cross-sections of bilateral export flows for more than hundred developing countries. They found that the more flexible the exchange rate regime, the greater the exports of developing countries. Rose (2000) used a gravity model to estimate the separate effects of exchange rate volatility and currency unions on international trade. He used a panel data for 186 countries for five years spanning 1970 through 1990. Rose argued that a common currency is not equivalent to reducing exchange rate volatility to zero as some scholars argued; rather, his empirical findings showed that two countries that have a currency Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 17 union trade substantially more than countries with their own currencies. His point estimate is that trade between currency union countries is over three times higher. Exchange rate volatility reduced trade, but to a much smaller degree. Rose then went to lengths to show that his results were robust and insensitive to exact econometric methodology. Aristotelous (2001) investigated the effect of exchange rate volatility and exchange rate regime on the British exports to the United States utilizing a gravity model and using data for the period 1889-1999. His paper concluded that neither exchange rate volatility nor the different exchange rate regimes that spanned his period of study, namely fixed, managed-float, and freely floating exchange rate regimes had an effect on export volume. Cho, Sheldon, and McCorriston (2002) used a sample of bilateral trade flows across ten developed countries between 1974 and 1995 to explore the impact of exchange rate volatility on the growth of agricultural trade as compared to other sectors, namely machinery, chemicals, and other manufacturing. The authors argued that using aggregate data ignores the fact that the impact of exchange rates may vary across sectors. This may occur because different sectors have different degrees of openness to international trade, and/or because different sectors have different industry concentration levels and make different use of long-term contracts. Cho, Sheldon, and McCorriston used alternative measures of volatility and the results were generally robust. The conclusion being that agricultural trade has been more adversely affected by medium to long-run volatility in real exchange rates as compared to other sectors, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 18 especially when the sample excluded bilateral trade between European Monetary System (EMS) countries. 2.2.1.3 Currency Union Related to exchange rate volatility is currency union or common currency. Rose’s (2000) study discussed earlier, was carried out to resolve the argument between ‘Sceptics’ (and most economists) and ‘Europhiles’. ‘The former argued that the EURO might only increase the intra-EU trade a little because exchange rate volatility was low before the European Monetary Union (EMU), and whatever volatility remained could, through the utilization of forward contracts and other derivatives, be inexpensively hedged. On the other hand, ‘Europhiles’ believed that sharing a common currency would cause an increase in the depth of trading relations, while preventing the ‘beggar thy neighbor’ competitive devaluations that can destroy a common market. As noted above, Rose (2000) findings showed that entering a currency union delivers an effect that is over an order of magnitude lager than the effect of decreasing exchange rate volatility from one standard deviation to zero. Following Rose (2000) a number of studies were carried to test the impact of currency union/common currency on bilateral trade flows. For instance, Nitsch (2000) using Rose’s (2000) same set of data and after making some modifications and extensions showed empirically that the estimated currency union effect on trade was reduced by about one-half (than that found by Rose) after simple manipulations of the data set and the regression specification. In other words Nitsch demonstrated that the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 19 effect o f a common currency fell from a factor of 3.3 (in Rose) to about 2.5. The factor was even reduced to 2 in three out of the five years, suggesting that a common currency only doubled rather than tripled trade. Moreover, Nitsch found that it was possible to find a specification, where the effect of currency unions on trade was zero. He also argued that the trade-multiplying effect varied across different currencies - from a completely unaffected trade patterns for countries that adopted the US Dollar to trade flows that exceeded average trade by 30,000% for countries that adopted the Australian dollar. This in Nitsch’s opinion suggested that projections about the potential trade-multiplying effect of adopting a common currency are not reliable. Frankel and Rose (2002) estimated the effects of currency unions/boards on both trade and income in a two-stage approach. They showed that, in the first stage, belonging to a currency union/board tripled trade with other union members. In the second stage, they showed that every 1% rise in a country’s overall trade relative to GDP increased income per capita by at least one-third of a percent. Glick and Rose (2002) tried to answer the question of whether leaving a currency union tended to reduce international trade. They carried out this study on 217 countries in the period 1948-1997, where a large number of countries left currency unions. Glick’s and Rose’s fixed effects estimates indicated that entry into or exit from a currency union caused bilateral trade flows to approximately double or be halved respectively, holding other things constant. They argued that these results were not only economically and statistically significant but relatively robust as well. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 20 Yeyati (2003) examined the impact of a currency union on bilateral trade flows. His new approach was to differentiate between multilateral currency union (MCU) and unilaterally dollarized countries ‘siblings’, which are countries that unilaterally adopt the currency of a larger country ‘parent’ as a legal tender. Yeyati found that the magnitude of the impact of a common currency on trade flows was significantly larger for ‘parent-sibling’ or ‘sibling-sibling’ pairs than it was for members of a multilateral currency union. 2.2.1.4 Other Economic Variables Economic variables other than the ones mentioned above have also been used to investigate their effects on trade flows. Gaile and Grant (1989) employing data from 122 countries and for four decades since World War II examined how power and location - measured as distance between the centroids of the trading countries - affected trade flows through using a gravity model. The variables used to measure power were GNP (economic power), defense expenditure (military power), and percent of population involved in primary, secondary, and third level education (power of knowledge). The authors found that power-distance relationships statistically accounted for the majority of the trade variance among nations. Tamirisa (1999) used the gravity model on a sample of 40 industrial, developing, and transition countries for 1996 data to examine the effect of current and capital account restrictions on bilateral exports. She found that controls on current payments and transfers formed a negligible barrier to trade. Capital account Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 21 restrictions, on the other hand, reduced exports into developing and transition economies but not into industrial economies. These results, the author argued, might reflect variation in the extent of liberalization across countries and types of control, where controls on current payments and transfers have been largely abolished worldwide, while controls on capital flows still prevail in many developing and transition economies, but not in industrial countries. Tamirisa’s conclusion was that further capital account liberalization might significantly foster trade. Baier and Bergstrand (2001) desired to disentangle from one another the relative empirical contributions of transport-cost declines, tariff reductions, income convergence, and income growth to the increase of world trade after World War II. To do so, the authors used a gravity model based on data from 16 OECD countries between late 1950s and late 1980s. Their empirical results revealed that of the increased trade flows, income growth (real GDP growth) explained about 67%, tariff rate reductions and preferential trade agreements (PTAs) about 25%, transport cost declines about 8%, and virtually none by real GDP convergence. 2.2.2 On Political Factors That countries have certain international strategic and diplomatic interests that affect their international economic policy is not doubtful. The use of sanctions, for instance, is an illustration of the effect of international politics upon international trade policy. Aside from such obvious examples of linkages between international politics and Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 22 international trade policy, Summary (1989) raised an important question. Are international political considerations important enough to affect trade flows? A number of studies carried mostly by political scientists, especially in the last ten years have used quantitative methodologies to answer this question. Srivastava and Green (1986) found cultural similarity, economic union, political instability and former colonial status to be significant determinants of trade between nations. Summary (1989) using data for sixty-six U.S. trading partners in 1978 and 1982 developed a gravity model to identify and quantify the factors affecting bilateral trade flows between the United States and other countries. She found trade flows between the U.S. and its trading partners to be affected not only by economic factors but by international political factors as well, such as the amount of arms trade, and the number of foreign agents registered in the United States, which is a measure of the degree of political alliance or friendliness between the U.S. and that nation. The number of U.S. government employees located in the trading country was found to affect U.S. exports only. Political rights were found to be insignificant. Pollins (1989) argued that the impact of international conflict and cooperation on international trade is not limited to boycotts, embargoes, or direct ways to influence the behavior of others. Trade ties may change as a result of foreign policy alignments and conflicts. For instance, U.S.-Iranian economic ties were tom by the reorientation of Iran’s foreign policy following the 1979 revolution. At a more subtle level, Pollins argued, expansion and contraction of bilateral trade flows may be influenced by the warming or cooling of diplomatic ties between countries and he gave the example of Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 23 the growth in the U.S.-Egyptian trade in the wake of Camp David. To test the hypothesis that trade flows are influenced by the general diplomatic cooperativeness or hostility, amity or enmity between trading partners, Pollins utilized a gravity model and used data on 25 countries for the period 1960-1975. He found that, as mentioned earlier, countries adjusted their commercial relations (through changing tariffs, quotas, and similar trade measures) to satisfy security objectives even as they cared for economic concerns such as production capacity, prices, and transportation costs. This might imply that the economic impacts are usually overestimated as they exclude security as well as other political objectives. Dixon and Moon (1993) theorized that similarity in political system and similarity in foreign policy orientation should enhance trade levels. They tested this hypothesis for U.S. exports to 76 importing nations over the period 1966-1983. To account for similarity in political system, the authors employed a measure of the openness of the importer’s institutional structures. This concept is operationalized with the eleven-point democracy scale contained in the Policy II data collection (Gurr, Jaggers, and Moore 1989). In other words, the authors used the institutionalization of democracy as a measure for political similarity. As for foreign policy orientation the authors’ measure is based on United Nations (UN) voting agreements. Dixon and Moon found that institutionalized democracy - political similarity - positively affected U.S. exports, meaning that democratic states traded more than other pairs of states, but fell short of conventional standards of significance. On the other hand, foreign policy Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 24 orientation - UN voting agreement - positively affected U.S. exports and was statistically significant at well below the 5 percent level. Bliss and Russett (1998) utilized a body of data that included between 882 and 1,042 pairs of countries in the period 1962 - 1989 to examine the extent to which shared democratic political institutions have an impact on trade between countries. Their results are fully consistent with those of Dixon and Moon (1993). Even after controlling for distance, size of the economy, and alliance bonds, democratic countries traded with each other more than they did with countries that have other types of political systems. Bliss and Russett also found no support for the hypothesized effect of alliances on trade (that it increases trade), in contrast to the findings of Gowa and Mansfield (1993) and Morrow, Siverson, and Tabares (1998) concerning multipolar alliances. Regarding the influence of militarized disputes on trade flows, Bliss and Russett found it only moderate and inconsistent. Morrow, Siverson, and Tabares (1998) through a gravity model examined trade flows between all directed dyads of the major powers from 1907 to 1990, except during the two world wars and the year immediately following the end of each. Their aim was to simultaneously test three hypotheses: whether trade flows are greater between nations with similar interests than those with dissimilar interests; whether trade flows are greater in democratic dyads than non-democratic dyads; and whether trade flows are greater between allies. The authors’ findings were that: a militarized dispute did not significantly reduce trade between the disputants, while similarity of political relations increased trade flows; mutual democratic institutions also increased Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 25 trade flows between nations; and the impact of alliances on trade flows was not certain: they increased trade flows between allies in a multipolar system while they decreased trade flows between allies in a bipolar system. This last finding is exactly the opposite of Gowa and Mansfield’s (1993) study conclusion. Long (2003) using two different alliance data sets, distinguished between alliances that include commitments of military assistance in the event of an attack, i.e., defense pacts, from those that do not, i.e., non-defense pacts, to test the hypothesis that countries that have defense pacts trade more than counties that do not. Long used a gravity model to examine such proposition on trade between major powers and employed trade data from the period 1885 to 1990. Long’s results demonstrated that countries in a defense pact alliance traded more than countries that were in a non defense pact alliance, whose trade was statistically indistinguishable from trade between non-allies. Long believed that his study could explain the inconclusive findings in previous studies concerning the effects of alliances on trade levels since previous studies did not distinguish between defense and non-defense pacts. 2.2.3 On Other Factors A number of scholars used a number of other factors to investigate their impact on trade flows. Gould (1994) employed a gravity model to study U.S. trade with 47 nations to find that immigration positively influenced trade flows, with exports being more strongly affected than imports. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 26 Head and Ries (1998) hypothesizing that immigrants may increase trade with their country of origin due to superior knowledge of, or preferential access to, market opportunities, used Canadian trade data with 136 partners from 1980 to 1992 to explore such proposition. To do so they employed an augmented gravity model. Head and Ries ascertained that a 10% increase in immigrants led to a 1% increase in exports and a 3% increase in imports. This finding is in line with their hypothesis that immigrant connections to the home country and knowledge lower transactions costs, which thus increase trade flows. As for the reason behind the larger increase in imports over exports, the authors argued that preferences for home country played a role. Head and Ries also found that independent immigrants - those who are selected according to a point system based on education, occupational demand, and other criteria and thus tend to be more skilled than other immigrants - had the largest influence on trade, while refugees the least and family immigrants in between. The entrepreneur class, created in 1976, however, did not appear to have energized Canadian trade. It is also interesting to note that the authors discovered that East Asian immigrants expanded trade more than immigrants from other regions. Bougheas, Demetriades, and Morgenroth (1999) argued that distance is not the only variable that models transport costs as many studies show. They demonstrated theoretically that transport costs are not only a function of distance but also of the availability of public infrastructure, such as roads, ports, and telecommunication networks. Accordingly better infrastructure reduces transport costs, which is thus expected to increase the volume of trade. The authors empirically tested their Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. hypothesis using the gravity model on data from European countries from 1970 to 1990. Their results suggested that the infrastructure variables’ coefficients were positive and significant, a finding that supported their hypothesis. Bougheas, Demetriades and Morgenroth also showed that the inclusion of the infrastructure indicators improved the fit of the gravity model in most cases as reflected by a higher adjusted R2. The authors’ argument is intuitive especially when you know that because of infrastructure problems, shipping costs from Africa to Europe are 30% higher for plywood and 70% for tuna than those from Asia to Europe (World Bank, 1994). Bougheas, Demetriades, and Morgenroth cautioned however that increases in the volume of infrastructure are not always welfare improving when taking into account the resource/opportunity cost of infrastructure because sometimes for some relatively high levels of infrastructure the benefits of extra investments in terms of higher volume of trade, are outweighed by loss in the final output. Limao and Venables (2001), in a study similar to the one just mentioned above, argued that infrastructure is an important determinant of transport costs. This is especially true for landlocked countries. Their analysis of bilateral trade data showed that a deterioration of infrastructure from the median to the 75t h percentile raised transport costs by 12% and reduced trade volume by 23%. Fink and Braga (1998) estimating the effects of Intellectual Property Rights (IPRs) on bilateral trade flows also using a gravity model, found that, on average, higher levels of protection had a positive and significant impact on non-fuel trade. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 28 This result was not confirmed however when confining the estimation to high technology goods. In this case, IPRs had no statistically significant impact. A common border, a variable that is commonly used as an indicator of geographical proximity and cultural similarity between a pair of countries, has been included in many studies that employed the gravity model that aimed to examine the effect of another variable on trade flows and not primarily that of borders. A number of studies, however, explored the effect of national borders - as the main variable - on trade flows. A fewer studies did so by utilizing a gravity model. McCallum (1995) using data from the 1988 input-output tables for Canada, a data set which Nitsch referred to as “apparently unique in allowing the analysis of trade patterns between regions of different countries” (Nitsch 2000:1091), found that Canadian provinces traded twenty times as much with other provinces as they did with equally distant states in the USA. This suggested a surprisingly large ‘home bias’ in international trade, since the national border between Canada and the US is usually assumed to be one of the most easily passable lines in the world. Helliwell (1996&1998) extended McCallum’s basic sample (now 1988-1994) and carried out a number of robustness checks, finding that variations in the estimated border effect to be only minor. Wei (1996) performed a similar investigation on OECD countries in the period 1982-1994 and, using exactly the same specifications as in McCallum and Helliwell, found that an OECD country traded with itself about ten times more as it did with another OECD country. However, when Wei included some potentially important regressors that were not included in the previous two works, such as Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 29 remoteness, common language and common border, the home bias fell dramatically to 2.6. Wei also found that exchange rate volatility had a negligible effect on the home bias and that the home bias experienced slow and steady decline during the period of study (due to high substitutability in the goods market among OECD countries) especially among EC members, where trade barriers are minimized. Along the same lines, Nitsch (2000) utilizing a gravity model and data from EU countries from 1979-1990 investigated the impact of national borders on international trade flows within the EU countries. He discovered that after adjusting for size, distance, common language, common border, and remoteness, an average EU country still exported seven to ten times more to itself than to an EU partner, suggesting that national borders still matter. It is however important to point out that Rose (2000), Rose and Van Wincoop (2001), and Frankel and Rose (2002) among others argued that a big part of this ‘home bias’, which is usually attributed to just the presence of national borders, stems from the fact that a single currency is used inside a country. Rose and Van Wincoop (2001) estimated that half the typical border variable was due to different sovereign monies. Just like the border variable, the common language variable, which is used as a proxy of cultural similarity and which is expected to lower information and transaction costs and consequently to increase trade flows, is used in many gravity studies as a dummy variable and commonly not the basic variable of interest. A few studies have focused on language as the main variable of interest. For instance, Hutchinson (2001) used an index to analyze the importance for trade flows Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 30 of the difficulty for a native English speaker in learning a foreign language. After controlling for the standard gravity model variables as well as the stock of migrants from a particular country, Hutchinson’s results indicated that the U.S. bilateral trade flows between 1970 and 1986 was lower with a country that had a dominant language which was relatively more difficult to leam. The effect was especially crucial for U.S. imports of consumer manufactures. Hutchinson (2002) using U.S. trade data with 33 countries in 1995 employed a gravity model to investigate the effect on trade flows of the proportion of population that speaks English as a first language or as a second language in the other country. After controlling for GDP, population, and distance, the proportion of the population who speak English as a second language was shown to be more important than the proportion of those who speak English as a first language for increasing trade with the United States. Moreover, both English as a first and as a second language were found to be more important for imports than for exports. So far, the paper reviewed the effect on trade of a number of variables that studies were originally set out to investigate. Other variables that are also used in gravity models, but may be not as the primary variables of interest include but are not limited to: common colonizer, colonial relationship, tariff rates, prices, sum of areas of trading partners, landlocked, and common nation. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 31 2.2.4 Specifically on Intra-Arab Trade Gravity models, as noted above, have been used extensively to measure the effects of numerous variables on bilateral trade flows among countries. However, their use in the literature on intra-Arab trade is, to the best of my knowledge, limited to handful: Ekholm, Torstensson, and Torstensson (1995&1996), Al-Atrash and Yousef (2000), and Miniesy, Nugent, and Yousef (2004). The reasons for this shortcoming, according to Al-Atrash and Yousef, are attributed to the demanding data requirements for estimating a gravity model, of which many might be missing for Arab countries, and the dominant role of oil in trade that could bias the relative importance of intra-Arab trade as well as Arab trade with the rest of the world. The main question that these studies, more or less, have attempted to answer was whether intra-Arab trade was too low. Ekholm, Torstensson, and Torstensson (1995) utilized regression estimates of gravity models to define a “natural” level of trade between two partners given determinants such as size, level of development, distance, and historical or cultural affinities. If a predicted value is higher than the actual value then this indicates a potential for more trade. They concluded that intra-regional trade in the Middle East was not “too low”. Ekhholm, Torstensson and Torstensson (1996) questioned whether the peace process (the one that is now completely shattered) between the Israelis and the Palestinians and thus normalization of relations will lead to more trade among the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 32 countries of the Middle East and North Africa (MENA) and between these countries and Europe. To answer this question, the authors used a gravity model to estimate what the MENA countries’ trade would have been if normal political relations had prevailed. They concluded that a successful peace process in itself might not have large effects on either trade or growth, but might have indirect effects in so far as a successful peace process promotes trade liberalization and economic reforms. Al-Atrash and Yousef (2000) also used the gravity model to investigate whether intra-Arab trade is too low. They employed a standard gravity model to examine their hypothesis. They augmented the simple model with variables that captured the effect of trade policies, cultural factors, and regional trading arrangements. Their data set consisted of 18 Arab countries and 43 other countries that represented above 90% of the exports and imports of the Arab world. The time period that they have chosen is 1995-97, for which they calculated the average level for each variable. As for the distance, it is measured as the direct distance between two capitals. Al-Atrash and Yousef found, according to their estimates, that intra- Arab trade and Arab trade with the rest of the world were indeed lower than what would be predicted by the gravity equation. The model actually suggested that overall intra-Arab trade should be about 10-15% higher than what was observed. The results strengthen the case for further trade liberalization in the Arab world and possibly in the context of greater regional integration. Miniesy, Nugent, and Yousef (2004) also utilized the gravity model to test, among other things, whether intra-Arab trade is really low relative to its theoretical Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. potential, and to simulate the effect of a full-fledged free trade agreement between Arab countries on their trade level. In addition to the usual variables used in the gravity model - GDP, population, and distance - Miniesy, Nugent, and Yousef used other variables such as exchange rate variability, common language, common colonial or other historical experience, common currency, free trade agreements and common border. Their results showed that in 1992, every Arab country has achieved less in the way of intra-regional trade than predicted by the model, however the differences have not been large for Egypt, Jordan, and Morocco. For the region as a whole, trade flows were only 41.2% of the predicted level. The creation of an FTA between Arab countries would more than double such trade relative to what would be expected on the basis of gravity model considerations. 2.3 Purpose of this Study and Contribution to the Literature Very few studies have investigated the effects of different aspects of domestic governance on bilateral trade flows. Even fewer used a gravity model to do so such as the scholars whose works I discussed earlier, who examined the effects of democracy and political instability and freedoms on trade flows. To the best of my knowledge, no study has attempted to explore the effects on the volume of international trade of government effectiveness, regulatory quality, rule of law, and control of corruption, variables that will be explained in detail in the very next chapter. The impact of these variables on trade is the primary focus of this study. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 34 But why aspects of governance? My choice is based on three reasons. First, the different aspects of governance and whether they affect or are affected by economic performance/development have been one of the hottest research subjects in both political science and economics in the past twenty years, especially in comparative politics and political economy. Nevertheless, their effect on international trade has been largely ignored. Second, the different aspects of governance, especially the lack of government effectiveness, regulatory quality, rule of law, and control of corruption may very well underlie the apparent reasons regarding low intra-Arab trade. Third, knowing the effects of such governance factors on bilateral trade flows may have important policy implications that would require first addressing such factors before promoting FT As between Arab countries. In addition to the basic gravity model variables, my study will employ many other variables, some of which have either never been used before such as a measure for income inequality, and a measure for financial development, or have been used in a couple of studies, such as two measures for whether economies concerned are centrally planned and were ever centrally planned, and two measures for whether countries have current and/or capital account restrictions. This study will include data for 100+ countries from 1985 up until and including 2000, a year which no gravity studies have included. Moreover this study, in addition to investigating the effects of all these variables on trade for all 100+ countries, will place special emphasis on intra- Arab trade, an area largely ignored as mentioned earlier. As such, these will be my modest contributions to the literature. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Now, lets move to chapter three where the model, sources of the data, expected signs of the model’s coefficients, and results for World trade flows regressions are discussed. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 36 3. GOVERNANCE AND WORLD TRADE The central objective of this dissertation is to investigate the impact of governance on intra-Arab trade flows. However, to have some idea of how the results compare to world trade flows, the paper will draw on the experience of the world for a reasonable period of time. Such analysis is the purpose of this chapter, which is composed of four sections. The first will introduce the model to be estimated. The second section will present the sources of the data. The third section will discuss the expected signs of the coefficients of the model. And the last section will display the estimation results and offer interpretations. 3.1 The Model The parameter values of the gravity model are estimated on the basis of a panel of well over 100 countries for the years 1985, 1990, 1995,1997, and 2000. The model3 is specified as follows: Ln (Bilatijt) =f}0 + Pi Ln (GDP{ GDPj) t +/32Ln (GDPPCi GDPPCj) , + f$ 3 Ln Disty + / ? 4 Ln (Areat + Areaj) + fis LLg+ fieBorderij + 07 Langy + fa Region# + Nation^ + f}1 0 Colonizer# + Pn Colonial^ + finERV# + CU# + aj (Gov_dst + Gov_dsj)t 3 Prices and tariffs are not used as explanatory variables because of the very limited and the low quality of their data. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 37 + o .2 CPYijt + 03 CPEijt + 04 (Ginii + Ginij) t + Os(M2GDPi M2GDPj) t + 0$ Currentyt + 07 Capitalyt + syt Where i and j denotes countries, t denotes time, and the variables are defined as follows: . Bilatjjt is the nominal value of bilateral trade between i and j at time t, . GDPi, GDPj are nominal GDPs of i and j, . GDPPCj, GDPPCj are Per Capita GDPs of i andj, . Distjj is the Great Circle Distance between i andj in miles, . Area;, Areaj are the areas in Square Kilometers of i andj, • LLij is a dummy variable, which is 0 if no countries are landlocked, 1 if only one country is landlocked, and 2 if both countries are landlocked, . Borderjj is a binary variable, which is 1 if i andj share a border and 0 otherwise, . Langjj is a binary variable, which is 1 if i andj share an official language and 0 otherwise, . Regionjjt is a binary variable, which is 1 if i andj belong to an RTA in year t, . Nationjj is a binary variable, which is 1 if i andj are part of the same nation, . Colonizerjj is a binary variable, which is 1 if i andj share the same colonizer in or after 1945, . Colonialjj is a binary variable, which is 1 if i colonized j or vice versa. . ERVjjt is the volatility of the bilateral nominal exchange rate between i andj in period t, . CUjjt is a binary variable, which is 1 if i and j use the same currency at time t, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. . Gov_dsi, Gov_dSj are governance indices of i and j, . CPYjjt is a dummy variable, which is 0 if no country is centrally planned at year t, 1 if only one country is centrally planned, and 2 if both countries are centrally planned, . CPEy is a dummy variable, which is 0 if no country was ever centrally planned, 1 if only one country was ever centrally planned, and 2 if both countries were ever centrally planned. . Ginii, Ginij are the Gini coefficients of i and j, . M2GDP,, M2GDPj are the money supply as a percentage of GDP of i and j, . Currentjjt is a dummy variable, which is 0 if no country has current account restrictions at time t, 1 if only one country has current account restrictions, and 2 if both countries have current account restrictions. . Capitalijt is a dummy variable, which is 0 if no country has capital account restrictions at time t, 1 if only one country has capital account restrictions, and 2 if both countries have capital account restrictions. . syt is the error term. As indicated by the form of the above equation, the model is log-linear in some of the continuous variables. This means that for these variables, the estimated coefficients represent elasticities, that is, the percentage changes in the bilateral trade values due to given percentage changes in the continuous explanatory variables. For the other continuous variables, the estimated coefficients represent the unit changes in the dependent variable due to a unit increase in the continuous independent variables. As for the coefficients of the dummy variables, they indicate the percentage change in Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 39 the dependent variable due to a unit increase in the binary variable from 0 to 1, and from 0 to 1 and 1 to 2 in the other dummy variables. The parameters will be estimated using Ordinary Least Squares (OLS). A common problem regarding the estimation of bilateral trade flows is that sometimes such flows are zero simply because the two countries do not trade. In this case, a standard log-linear model with a log-normally distributed error term cannot, by definition, explain these zero trade flows and thus OLS cannot be used. Exclusion of zero trade flows on the other hand may also be invalid because it would lead to a potential sample selection bias. To get over this problem, the parameters will be estimated using a Tobit limited dependent variable model. Both techniques will then be compared to determine if the estimated parameters are significantly affected. According to Al-Atrash and Yousef (2000), oil may bias the results through exaggerating the level of Arab trade with the rest of the world and also through underestimating the potential for intra-Arab trade given the similar economic structure of the oil economies. For this reason, the paper will follow Miniesy, Nugent, and Yousef (2004) in disaggregating the trade matrices of each year into energy (specifically oil and gas flows) and non-energy/other subtotals. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 40 3.2 The Data4 The trade data for all years except 2000 were taken from Feenstra et al (2000). Missing data in this source in these years as well as 2000 trade data were taken from the International Monetary Fund’s (IMF) Direction of Trade Statistics (DOTS) CD- Rom (2003). Bilateral trade flows are divided into energy and other (non-energy) in all years except 2000 since IMF’s DOTS CD-Rom has only aggregate data. For data on GDP, GDPPC and M2GDP, the World Bank’s 2001 World Development Indicators’ (WDI’s) CD-Rom and on-line WDI 2002 for 2000 data were used.5 Some missing values for GDP and population were taken from the UN’s National Accounts Statistics: Main Aggregates and Detailed Tables (85-2000). Concerning M2GDP, M2 refers to quasi money or broad money.6 It includes M l (transactions money which consists of currency and checking accounts) as well as saving accounts and similar assets, which are very close substitutes for transactions money. For data on Areas, LL, border, language, nation, colonizer, and colonial, the Central Intelligence Agency’s (CIA’s) website was used7. For distance, the coordinates (from center of country) were taken from the CIA’s website. These coordinates through a certain formula available on line 4 In this part, the paper will only discuss the sources of the data. For a list o f the countries used in this study, please see Appendix A, Table A.I. For Summary Statistics, please see Table A.2. For more details and information concerning data manipulation please consult Appendix B. 5 Was available at http://publications. worldbank.org/wdi 6 It is sometimes also called “asset-money” or “near-money”. 7 Available at http://www.odci.gov/cia/publications/factbook/index.html (World Factbook 2002) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 41 calculated the Great Circle Distance.8 The distance data between Czechoslovakia, East Germany, USSR, and Yugoslavia and their trading partners were taken from Rose (2000).9 Using the information at the World Trade Organization’s (WTOs) website1 0 , countries that have the following RTAs were included in the variable Regional11: EC, EFTA, CACM (Central American Common Market), CARICOM (Caribbean Community and Common Market), PATCRA/SPARTECA (South Pacific Regional Trade and Economic Cooperation Agreement), ASEAN (Association of South East Asian Nations), CER (Closer Trade Relations Trade Agreement), NAFTA (North American Free Trade Agreement), BAFTA (Baltic Free Trade Area), US-Israel, and Canada-US. To measure the exchange rate volatility between countries i and j at time t, the paper employed Rose’s (2000) methodology. Rose calculated the exchange rate volatility through estimating the standard deviation of the first-difference of the monthly natural logarithm of the bilateral nominal exchange rate (using the IMF’s International Financial Statistics (IPS) Line ae) in the five years preceding period t. That is, for the 1985 Egypt/US observation, the standard deviation of the first- difference of the log Egyptian/American exchange rate is estimated through using 8 Available at www.wcrl.ars.usda.gov/cec/java/lat-long.htm 9 There is a slight difference between the formula this paper used in calculating distance and that of Rose. However, using Rose for the above mentioned countries was the best the paper could do since the CIA’s website did not provide their coordinates. 1 0 Available at www.wto.org/english/tratop_e/region_e/eif_e.xls 1 1 Agreements included are agreements that are relatively more recognized. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 42 monthly data from 1980 through 1984. For 1997 and 2000 observations, the monthly data are taken for 5 years from 1992-1996 and from 1995-1999 respectively.1 2 For currency union, data is taken from Rose (2000) and corrections thereof in Glick and Rose (2002), but only for common currency not pegged currencies or currency boards. For CPY and CPE, the dummy variables were constructed on the basis of knowledge about the use of central planning in the past. In some cases, that would involve membership in CMEA. For the Gini values a combination of two sources was used: Deininger and Squire (1997) and the United Nations University (UNU)AVorld Institute for Development Economic Research (WIDER) - United Nations Development Program (UNDP) World Income Inequality Database (2000). The Gini coefficient is a measure of the degree of inequality of income distribution. It takes the value between 0 and 1; 0 referring to absolute income equality while 1 referring to extreme income inequality. Thus the closer the Gini coefficient is to zero, the more equal is the income distribution. For data on current and capital account restrictions, the IMF’s Annual Report on Exchange Arrangements and Exchange Restrictions (1986-2001) was used. Current/Capital account restrictions are restrictions on payments for current/capital 1 2 Nominal rather than real measures o f exchange rate volatility were used for two reasons: First, as Rose (2000) reasoned, the absence of monthly or quarterly price data needed to calculate real exchange rate volatility would substantially reduce the sample size. Second, as Chowdhury (1993) observed, the models estimated using nominal or real measures of exchange rate volatility had their results qualitatively similar. Still, no consensus is reached on the best measure of exchange rate volatility. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 43 international transactions. In other words, they are restrictions on payments to IMF member countries other than those imposed for security reasons. They are official actions that directly affect the availability or cost of foreign exchange or impose undue delays (IMF 1986). Since governance is the main variable of interest, we treat it in some detail. According to Kaufmann, Kraay, and Zoido-Lobaton (1999b: 1), governance is defined as “[T]he traditions and institutions by which authority in a country is exercised. This includes (1) the process by which governments are selected, monitored and replaced, (2) the capacity of the government to effectively formulate and implement sound policies, and (3) the respect of citizens and the state for the institutions that govern economic and social interactions among them.” As Kaufmann, Kraay, and Zoido-Lobaton (2002) argued, “Voice and Accountability” and “Political Stability” capture the first part of the definition of governance. “Government Effectiveness” and “Regulatory Quality” capture the second part, while “Rule of Law” and “Control of Corruption” capture the third part. Data on these governance components was taken from the International Country Risk Guide (ICRG) produced by the Political Risk Services (PRS) group, where the components of the political risk index were used, which report subjective assessments of the factors influencing the business environment in the countries studied. The paper employs not only Kaufmann, Kraay, and Zoido-Lobaton’s definition of governance, but also the variables they thought appropriate to be included in each of the six governance components mentioned above. More precisely, “Voice and Accountability” has two subcomponents from ICRG data: Military in Politics and Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 44 Democratic Accountability. “Political Stability” has one subcomponent: Internal Conflict. “Government Effectiveness” has two subcomponents from ICRG data: Government Stability and Bureaucratic Quality. “Regulatory Quality” has one subcomponent: Investment Profile. “Rule of Law” has one subcomponent: Law and Order. And “Control of Corruption” has one subcomponent: Corruption.1 3 In what follows, the paper will briefly discuss what each subcomponent aims to measure as they are represented in ICRG (2003), leaving the calculations to the part on governance data in Appendix B. The ICRG data has two very desirable features: (1) its large sample of developed and developing countries (130+) and (2) its length of coverage over time (1982-current). The ICRG data depends on polls of experts. The central advantage of polls of experts is that they are explicitly designed for cross country comparability, and great effort is put in the benchmarking process, which i < 14 ensures this. Military in Politics: the military is not elected by anyone, thus its involvement in politics even at a marginal level is a diminution of democratic accountability and therefore represents a risk. The threat of military intervention might force elected governments to make inefficient changes in policy or even the government itself. Military in politics also represents a high risk since it is an indication that the government is unable to function effectively and that the country has an uneasy 1 3 Only the variables to be included in each governance component are what has been borrowed from Kaufmann, Kraay, and Zoido-Lobaton (2002) and not their calculations or calculation techniques. 1 4 For more on measures based on polls of experts, their advantages and disadvantages, consult Kaufmann, Kraay, and Zoido-Lobaton (1999). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 45 environment for foreign businesses. Lower military in politics points indicate greater degree of military intervention in politics and thus a higher level of political risk. Democratic Accountability: this measure quantifies how responsive a government is to its people. Since the less responsive a government is the more likely it is that it will fall, peacefully or violently, this variable not only indicates whether or not there are free and fair elections, but also how likely is the government to remain in power or remain popular. The lower the democratic accountability score, the higher the political risk. Internal Conflict: this measure assesses the political violence in a country and its impact on governance. Highest scores are assigned to countries that have no armed opposition to the government and to those whose governments do not indulge in arbitrary violence, direct or indirect. The lowest scores go to civil war-torn countries. Generally, the risk rating for internal conflict is the sum of three components: civil war, terrorism/political violence, and civil disorder. Government Stability: it is an assessment of the government’s ability to perform its declared programs, as well as its ability to stay in office. The risk rating assigned to government stability is the summation of three components: government unity, legislative strength, and popular support. Bureaucratic Quality: it is an assessment of the institutional strength and quality of the bureaucracy, which is of high quality if it minimizes revisions of policy when governments change. Therefore, high scores are assigned to countries where the bureaucracy has the strength and expertise to govern without substantial changes in Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 46 policy or interruptions in government services, while low scores are given to countries where a change in government tends to be traumatic not only in terms of policy formulation but also in terms of administrative functions. Investment Profile: it is an assessment of factors affecting the risk to investment. It covers issues like contract viability/expropriation, profits repatriation, and payment delays. Just like all the other governance components, a higher score indicates a better investment profile and thus a lower risk. Law and Order: Law assesses the strength and impartiality of the legal system, and order assesses the popular observance of the law. Corruption: it measures corruption within the political system, which distorts the financial and economic environment, impedes the efficiency of government and business by enabling people to assume positions of power through patronage rather than ability, and introduces an inherent stability in the political system. It also measures financial corruption in the form of demands for bribes connected with import and export licenses, tax assessments, exchange controls, police protection, or loans. Such financial corruption makes it difficult to conduct business effectively. This measure is also concerned with actual or potential corruption in terms of patronage, nepotism, job reservations, favor for favor, secret party funding, and suspiciously close ties between politics and business. While the aforementioned aspects of governance are admittedly subjective, there are many reasons for believing their use to be beneficial. First, objective data, e.g., on corruption, is almost by definition very difficult to obtain. Thus if the effect of Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 47 governance on trade flows or on any other dependent variable is to be investigated, subjective data will have to be used. Second, while a country may enjoy a set of sound institutions according to some standards, the confidence of residents of this country in these institutions is required if those residents are to participate in and contribute to good governance. Thus perceptions of the quality of governance may be as important as objective differences in institutions across countries (Kaufmann, Kray, and Zoido-Lobaton 1999b). Third, subjective perceptions might have greater explanatory power for future economic outcomes than past objective data. For example, Kaufmann, Mehrez and Schmukler (1999), in the context of the East Asian financial crisis, found that investor perceptions of future financial instability had significant explanatory power for future actual volatility. Fourth, the data used do not really aim to “measure” such aspects of governance, but rather to give them “indices”. The numbers associated with “indices” unlike “measures” do not mean something themselves. Their aim is primarily to sort countries into broad groupings according to levels of governance and to indicate changes over time. Fifth, although different sources of government data use different units to measure governance and thus using any one of them could lead to different outcomes and different policy implications, still Kaufmann, Kray, and Zoido-Lobatdn (1999a & 2002) found substantially large pairwise correlations among different governance Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 48 sources within each of the governance components, reflecting the common component of governance in these sources.1 5 Thus using ICRG data is quite comparable to other governance data found in other sources with the advantage of the high coverage of countries and over time. But if governance has six components, why does it appear as a single variable in the gravity equation presented earlier? The reason is that when trade flows were regressed on all the variables including the six governance components two problems arose: First, all the six components were highly correlated to each other suggesting that it is preferable not to include all of them in the same regression.1 6 To get around this problem, the six governance components were added together (using Principal Components Analysis) to form a single standardized weighted sum variable. Second, this single variable was highly correlated to GDPPC. To get around this problem, a separate auxiliary regression of the standardized weighted governance variable of the individual countries was run on GDPPC and then the deviation between the predicted governance and the actual governance variables was used to come up with an index 15 Still, Kaufmann, Kray, and Zoido-Lobaton (1999) warned that in general governance measures might not be very precise and thus caution must be exercised when using them and when interpreting their results. 16 Multicollinearity might cause some of these variables to show opposite signs to what they should actually have thus causing confusions. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 49 for each country. Putting these together for trading partners the governance variable 17 (a deviated standardized weighted sum) was finally reached. 3.3 Expected Signs GDP & GDPPC It is widely recognized that a country’s capacity to trade is directly related to its GDP because greater productive capacity leads to greater supply of tradables and higher income (mass) raises the demand for importables. In all the studies that used the gravity model, GDP/GNP turned out to have a positive and significant influence on trade flows. Thus the sign of the coefficient of GDP, Pi, is expected to be positive. GDPPC is GDP divided by population. The effect of the GDP component in GDPPC on bilateral trade flows is expected to be positive as just stated. Moreover, as Linder (1961) indicated, countries with similar levels of per capita income can be expected to demand somewhat similar products. Entrepreneurs tend to turn domestic demand opportunities into production activities and hence into exports as well. However, the population component of GDPPC, by itself, may either reduce or promote trade. A large population in the importing country may indicate a large domestic market, a higher degree of self-sufficiency and less need to trade, or a large 17 Four different governance calculations were experimented with. As explained further in Appendix B, the choice was made to use Gov_ds because it is more theoretically appealing than Gov_dw. Gov_s and Gov_w were not used because of their high correlation with GDPPC. Generally, regression results didn’t differ much with using the four governance calculations. For sample regression tables using these four calculations consult Appendix C. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 50 domestic market that eats up most of the produce and thus implies low export supply potential for the exporting country. A large population on the other hand may promote the division of labor, and because of economies of scale in production allow both 1 8 greater imports and exports. It also may increase the variety of goods. Not all studies recognized that population or GDPPC have inconclusive effect on trade. For instance, Linnemann (1966), Learner and Stem (1970), Aitken (1973), and Brada and Mendez (1985), argued that the population coefficient of the exporting country should be negative. Brada and Mendez (1985) also argued that the population coefficient of the importing country should be positive. Mansfield and Bronson (1997a) on the other hand argued that the population coefficient should be negative whether the country is the importing or the exporting one. Others, like Pollins (1989)1 9 , Nilsson and Nilsson (2000), and Al-Atrash and Yousef (2000), recognized that the population coefficient is ambiguous. Empirically, mixed results for GDPPC or population were found. Anderson and Marcouiller (2002) empirically argued that the omission of indices of institutional quality, like tariffs, and security variables (enforceability & transparency) not only biased the estimates of the gravity models, but also obscured a negative relationship between GDPPC and the share of total expenditure devoted to traded-goods. When the tariff and security variables were dropped from the 18 A big part of this argument is from Nilsson and Nilsson (2000). 19 Pollins thus decided to forgo including population as a second indicator of economic size for this ambiguity reason. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 51 regression, part of the positive effect of security on trade was misattributed to GDPPC. When these theoretically appropriate regressors were included, it was revealed that GDPPC actually had a negative effect on the traded goods expenditure share. From all that is stated above, the expected sign of the coefficient of GDPPC, p2, is ambiguous. Distance This study is interested only in the effect on trade flows of a variable measuring 9 0 absolute distance as opposed to relative distance , and regardless of its best measure21, distance is one of the important ‘resistance’ factors that is usually included 99 in the gravity model as a proxy for transportation and transaction/information costs. 20 Recently, a number of studies argued that relative distance, which is commonly referred to as ‘remoteness’ has an impact on the volume o f trade. Deardorff (1995) for instance, debated that remote countries such as Australia and New Zealand can be expected to trade more with each other than two countries which are separated by the same absolute distance but are geographically well positioned near other markets as Germany and Portugal. Nitsch (2000) used the remoteness variable as well as the distance variable in his model, which aimed to examine the impact of national borders on international trade within the European Union. He however found that the inclusion of remoteness measures did not affect his basic results. 11 In the gravity literature, there is no consensus on the best measure of distance. For instance, Gaile and Grant (1989) argued that the straight line distance between the centroid o f each pair of countries is the best way to measure distance. It is however to be outweighed by relative distance concepts, which take economic and cultural distance into account. They used the ‘Arcdis’ program, which takes into account differences in latitude when calculating distance over large areas. Al-Atrash and Yousef (2000) measured distance as the direct distance between two capitals, which they believed would tend to exaggerate the distance between countries the larger their size. They argued however that the usage of a border variable would lessen this bias against large countries. Rose (2000) used the Great Circle distance to calculate distance. He also used the Hirschberg Centroid measure, and the Fitzpatrick- Modlin Great Circle distance between the most populous cities within the pair of countries to examine the sensitivity o f his results to alternate measures o f distance. Interestingly, none of his key results were sensitive to this choice. 22 Some studies have used distance however as a surrogate for tariffs; and others for cultural and linguistic similarity as Pollins (1989) mentioned in his article. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 52 Commonly, countries close to each other are likely to have more precise information and more direct acquaintance with each other’s culture and economy than a third country, which is distant from both of them.2 3 This in turn suggests that information costs increase with distance. Transportation costs are also expected to rise as the distance between any pair of countries increases. In short, transportation as well as transaction and information costs would be expected to rise as the distance between trade partners increases, thereby lowering bilateral trade flows. Thus, the sign of the coefficient of distance, P3 , is expected to be negative, which will then be consistent with all other published studies that employed the distance variable in a gravity model. Areas & Landlocked (LL) A number of studies such as Rose (2000) and Miniesy, Nugent and Yousef (2004) used a complementary measure of distance, which is the sum of the geographic areas of the trading partners. This variable, with the same logic for the distance variable, was found to be negatively and significantly related to bilateral trade flows. Thus the coefficient of the areas variable, P4, is expected to be negative. Rose (2000), Frankel and Rose (2002), and Glick and Rose (2002) also used a landlocked variable, which is also a proxy for transportation costs and which they found to be negative and significant. It is expected that landlocked countries endure higher transport costs and thus trade less than non-landlocked countries. 23 A similar but a bit different argument is made in Linder (1961) and Brada and Mendez (1985). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 53 Consequently, the sign of the coefficient of the landlocked variable, p5 , is expected to be negative. Border Another factor representing lower transport and possibly information and transaction costs is the possession of a common border. Consequently, the effect of border on bilateral trade flows and thus its coefficient, (3 6 , is expected to be positive. Indeed, in most if not all of the studies that used the border variable, its coefficient turned out to be positive and significant as in Al-Atrash and Yousef (2000), Rose (2000), and Glick and Rose (2002) to name just a few. While this variable could also have an indirect negative effect on trade flows between one of the partners and other countries not sharing the common border or on trade within the country (McCallum (1995) and Nitsch (2000)), this study limits itself to the direct effect. Language A common language facilitates trade through the ease of communication. It is used as a proxy of cultural similarity and is expected to lower information and transaction costs and consequently to have a positive impact on bilateral trade flows. Bliss and Russett (1998), Rose (2000), and Glick and Rose (2002) found the common language variable in their samples to be positive and significant. Thus the coefficient of the language variable, (3 7, is expected to be positive. While the various different common Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 54 languages could in principle have different effects on trade, we restrict our attention to the effect of a common language, regardless of which language that is.2 4 Regional As Greenaway and Milner (2002) argued, it is a fairly common finding in the gravity literature that RTAs affect trade positively, even if distance is controlled for. That is, countries that are members of the same RTA trade more with each other than would otherwise be expected. After all, such types of organizations are designed to do just that through their discriminatory use of tariffs between trade agreement partners and non-partners. Empirical evidence for that was given in chapter two. There are undoubtedly RTAs where no effective trade liberalization has occurred and where finding a positive trade effect would be quite surprising. In those instances, trade effects may be absent and sometimes even negative as also mentioned earlier. Since this chapter is not particularly interested in the effects of certain RTAs on trade flows, all regional agreements were considered equal in that no separate dummies for different RTAs were used. A number of studies where the main objective is not to study the effect on trade flows of RTAs such as Dell’Ariccia (1999), Rose (2000), and Nitsch (2002), included RTA as an independent variable in their gravity equations and found that 24 It is interesting to note that Al-Atrash and Yousef (2000) observed that English-speaking countries traded more with each other, while the French Language was positive but insignificant. Nitsch (2002) on the other hand found English and French to be positive and equally significant, Spanish to have no effect on trade flows between two countries while two Arabic speaking countries tend to trade less with each other than countries with different languages i.e. negative coefficient for the Arabic language. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 55 countries that belonged to one RTA traded more than countries that did not, other things being equal. Based on all the above, the coefficient of the regional variable, Ps, is expected to be positive, especially since the RTAs included in the Regional variable are ones that were actually adopted and implemented practically, not simply on paper. Nation, Colonizer, and Colonial Overseas territories/departments of another country are expected to trade more with the home country than independent countries. Not only may the different territories of a single country share the same nation, cultural similarity, and language but also they may share a common currency. Thus the coefficient, p9, is expected to be positive. Rose (2000) and Miniesy, Nugent, and Yousef (2004) found the same nation coefficient to be positive and in many cases significant. Countries that share a common colonizer, which indicates historical similarity, are likely to trade more with each other than countries that do not share a common colonizer, other things being equal. Thus the coefficient pio is expected to be positive. It is also expected that countries that had a colonial relationship, where one colonized the other, would subsequently trade more than countries that did not share such relationship, other things equal. The coefficient Pn is thus expected to be positive. However, the effect of the colonial relationship on trade might get smaller as time passes by. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 56 Svedberg (1981) found European trade with developing countries to be significantly affected by former colonial ties. Srivastava and Green (1986) found former colonial status, among other things, to significantly determine trade between nations. Mansfield and Bronson (1997a) found a colonial relationship coefficient to be positive and significant. Rose (2000), and Glick and Rose (2002), also found a common colonizer as well as previous colonial relationship variables to be positive and mostly significant at 1% level. Exchange Rate Volatility (ERV) There is no consensus in either the theoretical or the empirical literature on the impact of nominal or real exchange rate volatility on international trade, no matter how it is measured. Several theoretical studies, for instance Clark (1973), Baron (1976), Peree and Steinherr (1989), and many others have shown that an increase in exchange rate volatility reduced the volume of international trade. A typical argument in this literature is that if exchange rate movements were not fully anticipated, an increase in exchange rate volatility would increase risk that in turn would lead risk-averse market participants to reduce their trading activity and to reallocate production towards domestic markets. This argument perceives traders as bearing undiversified exchange risk and thus if hedging was impossible or costly and those traders were indeed risk- averse, an increase in exchange risk would reduce the risk-adjusted expected revenue from trade, and consequently the incentive to trade falls. A number of empirical studies provided evidence in support of this view like Akhtar and Hilton (1984); Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 57 Cushman (1983&1986); Thursby and Thursby (1987); Brada and Mendez (1988), Koray and Lastrapes (1989); Pozo (1992); Chowdhury (1993); D ell’Ariccia (1999); Rose (2000); Cho, Sheldon, and McCorriston (2002) and many others. Some other theoretical studies like those of Franke (1991) and Seem and Van Hulle (1992) have shown that trade benefits from exchange rate volatility. Their models focused on the firm’s option to adjust its production in response to the exchange rate. Higher exchange risk promotes trade because when firms are allowed to optimally respond to exchange rate changes (by increasing resources in the export sector to offset the drop in expected utility of export revenue caused by the increase in exchange rate volatility), the entire cash flow from exporting become convex functions of the exchange rate and consequently the expected cash flow increases when the volatility increases. A number of empirical studies supported this view. Asseery and Peel (1991), and Kroner and Lastrapes (1993) model found that an increase in exchange rate volatility might be associated with an increase in international trade, while McKenzie and Brooks (1997) found an obvious positive association. Still other theoretical studies like Mann (1989) and Feenstra and Kendall (1999) argued that exchange rate volatility might not have an impact on trade and may rather have an effect in some other fashion such as on prices and foreign direct investment. Empirically, Gotur (1985), Bailey, Tavlas and Ulan (1986), and Aristotelous (2001) found that exchange rate volatility did not have any significant effect on trade flows. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Two interesting studies tried to theoretically provide a rationale for the contradictory empirical evidence in the literature. Both studies argued that the effect of volatility on trade would depend upon the source of volatility. Barkoulas, Baum, and Caglayan (2002) employing a signal extraction framework, asserted that an increase in the variance of the general microstructure shock reduced the volume of trade, while the variances of exchange rate fundamentals and the noise of the signal of future policy innovations had an ambiguous effect on the volume of trade flows. Secru and Uppal (2003) developed a general-equilibrium stochastic-endowment economy with imperfect international commodity markets, and treating both variables as endogenous argued that an increase in the volatility of the relative endowment process would lead to an increase in both exchange rate volatility and the expected trade volume. On the other hand, a drop in the shipment cost could imply a decrease in exchange rate volatility and an increase in the expected trade volume. From all of the above, the paper does not expect a certain sign for the coefficient of the exchange rate volatility variable, i.e. the expected sign of pi2 is ambiguous. 25 A general microstructure shock comes from innovations to the exchange rate process from the effects of portfolio shifts among international investors, excess speculation, bandwagon effects, bubbles and rumors, or the effects o f technical trading by chartists or “noise traders”. Fundamental factors are factors driving the exchange rate process and are related to monetary policy. Noisy signal refer to information through policy announcements, and central bank intervention or statements. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 59 Currency Union (CU) Theoretically speaking, a currency union or a common currency should never reduce trade flows. It should either encourage trade or at the very least leave trade patterns as is. The reason being that a common currency reduces transaction costs. The intuition is simple “trade between areas that use a single money is cheaper and easier than trade between areas with their own monies” (Mundell (1961), quoted in Frankel and Rose 2002). A common currency usually represents a serious government commitment to long-term integration. In turn, this commitment could motivate the private sector to engage in greater international trade. A common currency could alternatively induce greater financial integration that could lead to higher trade in goods and services. Empirically, a number of studies were carried to investigate the effect of a common currency on trade flows. The conclusion being that a common currency increased trade. As discussed earlier, Rose (2000), Nitsch (2002), Frankel and Rose (2002), Glick and Rose (2002), and Yeyati (2003), found with no exception that a common currency has a positive effect on trade. The debate between the studies is not about the sign of the coefficient of the common currency, but rather on its magnitude. Accordingly, P1 3 , the coefficient of the common currency variable is expected to be positive. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 60 Governance (Gov_ds) Voice and Accountability (VA):26 The question that needs to be investigated in this part is whether theoretically and empirically countries that are democratic/ have higher VA scores trade more with each other than countries that are not/do not have high VA scores. In other words, should we expect VA to positively affect trade? Private actors in democratic countries might prefer trading with those in other democratic states for a number of reasons: First, trade is enhanced by familiarity and trust that in part are derived from political and social similarity among countries. In general, private actors are more knowledgeable about relevant business trends in politically and socially similar nations. Moreover, social connections are easier to build and to maintain (Dixon and Moon 1993). Second, trade, which usually has a long-term component,2 7 might be less disrupted due to military conflict (Dixon and Moon 1993; and Bliss and Russett 1998). Although not an unmitigated truth, many studies showed that countries that are similarly committed to democratic procedures seldom engage with each other in military confrontation (Doyle 1986, Maoz and Abdolali 1989, and Bremer 1992). Third, the rule of law and shared norms among democracies might provide insurance for entrepreneurs against expropriations, which they might encounter in an autocracy (Bliss and Russett 1998). More generally, 26 It will be for convenience frequently referred to as democratic accountability or shortly democracy. 27 For exporters, the long-term component of trade is represented in marketing arrangements, advertising campaigns, and product design, etc. All these factors represent sunk costs and might only be justified by expectations of a stream of sales over time. Importers encounter similar pressures: the necessity of replacement parts, system upgrades, training, etc, which urges them to initiate only trade relationships that can be maintained. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6 1 limited governments based on the rule of law, which have historically gone hand in hand with democracies, gives foreign firms greater confidence that their interests will not be handled arbitrarily (Morrow, Siverson, and Tabares 1998). Democratic states might prefer and encourage their private agents, through lower tariffs, higher quotas and similar trade encouraging measures, to trade with agents in other democratic states. This is because a democratic trading state will feel its security less threatened by another democratic country than by an autocracy since the latter might use the gains from trade, if any, in enhancing its political stability or increasing its military capability that might one day pose a threat to the former (Bliss and Russett 1998). A number of empirical studies support the hypothesis that democratic countries trade more with each other than non-democratic countries as discussed in chapter 2. In the mid-1970s, more than two-thirds of all countries could reasonably be called authoritarian. This percentage has been dramatically reduced; less than a third of all countries are now authoritarian (Potter 1997). Could this imply that more trade will take place between democratic pairs? Verdier (1998) provided an argument that suggested that democratic convergence might in certain cases choke trade. He argued that if trade is propelled by factor endowments, then similarity in political regimes (democratic convergence) like similarity in factor endowments chokes trade. Since similar regimes tend to empower the same classes of producers, Verdier argued, specialization based on factor endowments would empower as many free traders as protectionists, with negative consequence on trade. On the other hand, if trade is Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 62 fueled by scale economies and countries specialize along product lines, then democratic convergence will not hurt trade. His conclusion being that democracy promotes trade only between industrialized countries. A number of empirical studies found the relationship between democracy and trade to be weak such as Summary (1989), Gowa and Mansfield (1993), Mansfield and Bronson (1997b), and Penubarti and Ward (2000). To sum up, the effects of democracy/democratic accountability/VA on trade flows is ambiguous. Political Stability (PS): The objective here is to explore whether, other things being equal, politically stable countries trade more with each other than non-politically stable countries. Intuitively, this should be the case. Since as explained earlier, trade has a long term component, agents in politically stable countries will prefer trading with other agents in other politically stable states because political stability reduces uncertainty and permit reliable economic predictions and thus lower the risk involved in long term contracts/investments. Empirically, I am aware of only the work of Srivastava and Green (1986) who found among other things that political instability negatively and significantly affected exports between nations while it had a little effect on imports. Accordingly, if political stability is a variable by itself, the sign of its coefficient is expected to be positive. Government Effectiveness (GE): Theoretically and empirically, the question to be answered is, should countries with high scores of GE trade more with each other than countries with low scores? Stable governments and high bureaucratic quality Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 63 imply that policies are not frequently changing due to government’s inability to carry out its programs, stay in office, or due to frequent revisions needed when new governments are placed. This accordingly implies more transparent policies and less confusion about which policies have been eliminated and which are still in place, trade policies being no exception. Empirically, to the best of my knowledge, there are no studies that investigated the impact of GE or any of its subcomponents on trade flows. However, it is expected that it will positively affect trade based on the theoretical arguments stated above. Regulatory Quality (RQ): Theoretically, Bliss and Russett (1998) argued that entrepreneurs and traders are likely to be more confident in the continuity of their business practices in democratic countries, where the risk of expropriation is not high. Leaving democracy aside for now, the same argument may still apply. The UAE and Singapore are not democratic but have high RQ scores especially for the late 1990s. Entrepreneurs/traders in countries with high RQ scores are more likely to trade with other countries with high RQ scores, where expropriation, payment delays and obstacles to profit repatriation are minimal. Empirically, to the best of my knowledge, the effect of RQ, in that context, on trade flows has not been explored before. If RQ is a variable by itself, its effect on trade flows is expected to be positive. Rule o f Law (RL): The rule of law plays a crucial role in trade because efficient trade requires reliable enforcement of the agreements governing the exchange. Anderson and Marcouiller (2002) argued that trade expands substantially when it is supported by strong institutions - especially by a legal system that is Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 64 capable of enforcing commercial contracts. In their study, Anderson and Marcouiller asserted that if the seven Latin American countries in their sample (Argentina, Brazil, Chile, Colombia, Mexico, Peru, and Venezuela) were to have the same enforceability and transparency scores as the mean scores of the EU members, Latin American import volumes would rise by 30%. Accordingly, RL’s impact on trade is expected to be positive. Control of Corruption (CC): Although control of corruption may be expected to have positive effect on trade flows, in the literature this is not an outright truth. There is a strand in the literature on corruption, contributed by both economists and non-economists (such as the works of Leff 1964; Huntington 1968; and Lui 1985) which suggested that in the context of pervasive and cumbersome regulations in many developing countries, corruption may actually improve efficiency and thus promote growth in that context. It is viewed as “efficient grease”. As Samuel P. Huntington (1968:386) put it: “In terms of economic growth, the only thing worse than a society with a rigid, over-centralized, dishonest bureaucracy is one with a rigid, over-centralized, honest bureaucracy.” Many other scholars criticized that ‘efficient grease’ hypothesis. For instance, Bardhan (1997) argued that distortions that are supposed to be mitigated by the effects of corruption are themselves and also corruption itself at least preserved or aggravated by the same common factors. This is because these distortions are often part of the built-in corrupt practices of a patron client political system. Bardhan also criticized the “speed money” argument. He asserted that many times corrupt officials instead of Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 65 speeding up, actually cause administrative delays to attract more bribes. Krueger (1974), Shleifer and Vishny (1993 and 1994), Bliss and Di Telia (1997), and Kaufmann and Wei (2000), among others, have modeled the problems of corruption. For instance, using a general equilibrium model and utilizing data from three world wide firm surveys, Kaufmann and Wei (2000) rejected the ‘grease’ argument, as they did not find evidence that the business sector reduced time waste or capital costs by paying bribes. In fact they found the opposite to be the case, meaning that firms that paid more in bribes were more likely to need more management time with bureaucrats to negotiate reductions. For our purpose, since we have both theoretical arguments concerning the effect of corruption on trade/development, the expected sign of the control of corruption variable is ambiguous. As discussed earlier, governance is the standardized weighted sum of all the six components. Four of the components are expected to affect trade positively, namely political stability, government effectiveness, regulatory quality and rule of law. The other two components, voice and accountability and control of corruption however have theoretically ambiguous effects on trade flows.2 8 But since in the weighted sum, the six components almost have equal weights, it is expected that the 28 Please note that the effect on trade of some of the governance individual components comes through GDPPC and that’s why an auxiliary regression, which takes away the influence of GDPPC, was run to get around this problem, as discussed in chapter 2 . Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 66 four positive components will outweigh the two ambiguous ones resulting in an overall positive sign for the coefficient of the governance variable, ai. Centrally planned by year and ever (CPY & CPE) CPY investigates whether in a specific year, two centrally planned economies trade more with each other than non-centrally planned economies, all other things being equal. Brada and Mendez (1985) observed that the literature on the trade patterns of command economies suggested that these economies would trade less with each other than comparable market economies. This could be due to the fact that many of the centrally planned economies were keener on import substitution, self-sufficiency and limited global orientation than they were on promoting trade. According to this argument CPY should be expected to have a negative effect on trade flows. However, another argument might reflect the keenness of centrally planned economies to trade with each other and to heavily do so if possible so as to reinforce their ideological stance. This would mean a positive effect on trade flows. Empirically, Brada and Mendez (1985) as mentioned in chapter two, found that the combined effect in CMEA of central planning and of integration policy caused that integration to under fulfill its potential. However, this under fulfillment was by the same amount as that of the EEC, which suggested that the system of central planning did not seem to place significantly greater barriers to encouraging inter-member trade than did policies adopted among integrating market economies. Mansfield and Bronson (1997a) in their study also discussed in chapter two included a variable Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 67 reflecting whether both trading partners have centrally planned economies. They discovered that the coefficient of this variable was not only positive but statistically significant at 1% level as well. From all of the above, the paper is not sure what sign to expect for the coefficient of the CPY variable. The expected sign of ( X 2 is thus ambiguous. CPE investigates whether two trading partners traded more if they were ever centrally planned. This could be treated as some sort of history sharing and could result in a positive effect on trade. Alternatively, this could mean that structurally these two countries -owing to the previously centrally planned nature of their economies- might face the same problems that might impede trade. The fact that many command economies turned into market economies, may support the structural problems hypothesis. This reasoning leads to the expectation that CPE will have a negative impact on bilateral trade flows, i.e., 0 1 3 is expected to be negative. Gini How should income inequality within two nations affect their trade with each other? Theoretically, Keynes (1936) argued that as the level of income increases both the average and marginal propensity to consume could be expected to fall. Accordingly, the Keynesian consumption function would lead us to expect that a redistribution of Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 68 income away from equality could have the effect of lowering aggregate consumption 29 and therefore imports m an open economy context. Linder (1961) presented a more explicit model in which income inequality may affect trade - also through demand. He argued that a country’s exports of manufactures are likely to be dependent on that country’s own demand for such products. Actual production and export are hypothesized to follow from prior demand. Linder explained this by saying that there would be no production or export of a product unless an entrepreneur would have seen an investment opportunity. The Linder model gives rise to two hypotheses: First, manufactures trade is likely to be intensive between countries at the same level of per capita income. Second, even for countries at similar levels of per capita income, the demand patterns will be most concentrated on overlapping commodities allowing for sizable reciprocal demand for manufactures imports the lower the income inequality within each country. Thus, income inequality both within and across countries is likely to reduce trade. Empirically, the closest studies to my variable are those of Hunter and Markusen (1988) and Hunter (1991) which have showed aggregate demand functions to be non-homothetic and suggested that the non-homotheticity of these demand functions has the effect of lowering trade volumes by as much as 25 percent. Although the non-homotheticity in demand functions is a necessary condition for income 29 Although this could be offset by the effect of greater savings on investment (coming from the greater inequality), in Keynesian theory, this would not be expected. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 69 distributional effects on international trade, these studies did not explicitly explore the impact of within-country income inequality on trade flows. From all the above, the sign of the coefficient of Gini variable, a 4, is expected to be negative. M2GDP M2 is a useful indicator of trends in the growth of money supply (Samuelson and Nordhaus 1995). M2GDP, or M2 as a percentage of GDP is a measure of financial development. Since some sort of financing is usually required for carrying out international trade, its impact on trade flows is expected to be positive, where countries that are relatively more financially developed are more likely to trade with each other, other things being equal than countries that are less financially developed. That is, the coefficient of the M2GDP variable, as, is expected to be positive. Empirically, to the best of my knowledge, no study has investigated the effect of M2GDP on trade flows. Current and Capital Account Restrictions Current account restrictions are considered trade impeding. As a matter of fact, one of the main purposes of the IMF was to help in the elimination of foreign exchange restrictions, which stood in the way of world trade growth. Capital account restrictions on the other hand were not perceived as inconsistent with this objective. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 70 Actually capital account restrictions were considered crucial for supporting the system of fixed exchange rates, which was/is thought to foster trade (Tamirisa 1999). Tamirisa (1999) argued that theoretically, the effect of current and capital account restrictions on trade is ambiguous. She asserted that the overall impact of current and capital restrictions on trade depends on the structure and effectiveness of these restrictions and their interaction with other distortions in the economy. The very basic economic effect of current restrictions is that they act as a tax on the foreign currency that is required for purchasing foreign goods and services. This in turn cut the quantity imported and/or increased the domestic relative price of imports leading to trade reduction. Capital account restrictions might impede trade through limiting opportunities for hedging foreign exchange risks and financing trade. Furthermore, current and capital restrictions usually raise transaction as well as other trade-related - 3 -i costs thus reducing trade. Moreover, these restrictions often foster evasion and rent- seeking that are considered unproductive costs. Not only that, but current and capital restrictions might reduce trade by limiting transfer of technology, managerial expertise, and skills through foreign direct investment, which are likely to be discouraged by controls on repatriation of profits and dividends, surrender 30 The debate is still going on in the literature concerning which exchange rate regime fosters trade. 31 This is because current restrictions tend to choke the development of liquid and efficient foreign exchange markets and modem payment instruments. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 71 requirements, etc. The empirical evidence indicated that in the absence of tariff barriers, foreign direct investment tends to increase the host countries’ trade.3 2 On the other hand, capital account restrictions can encourage trade through a number of macroeconomic channels such as limiting short-term speculative capital flows and retaining domestic savings. Still, the specific effect on trade through these channels depends dramatically on the interaction of these restrictions with other distortions in the economy.3 3 However, as Tamirisa argued, these effects are not that consequential in practice and could be offset by capital flight and the reduction in capital inflows and thus it is not surprising to find no empirical evidence supporting this argument. In short, theoretically, the effects of current and capital account restrictions on trade flows are more likely to be negative. Empirically, studies concerning their impact on trade flows have been very scarce. I am aware of only two studies. Tamirisa (1999), concluded that current and capital restrictions impeded trade, where the former had minor effects because they have been largely abolished worldwide, while the latter had significant negative effects for developing and transition economies but not for industrial countries, where they have been largely liberalized. Rose (2000) used current and capital accounts restrictions in one of his specification 32 In the presence of tariff barriers, restrictions on EDI may encourage trade. Because in this case, EDI and exports are alternative strategies and if EDI is allowed, a multinational company might choose to avoid paying tariffs by supplying the host country’s market through a subsidiary company. 33 Tamirisa argued that limiting short term speculative capital flows might increase trade through limiting exchange rate volatility. This is however not that clear cut and straightforward as discussed previously. Capital account restrictions can encourage trade through retaining domestic savings if these savings led to higher investment in export sectors. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 72 sensitivity analyses and found them to have had significant negative effects on bilateral trade flows. Thus, this paper expects the signs of the coefficients of the current and capital control variables and 07 to be negative. 3.4 W orld Trade Results 3.4.1 Table by Table Results and Interpretations Table 3.1 presents the regression results from estimating the gravity model for world total bilateral trade for: each individual year, the whole pooled sample (1985-2000), and the pooled sample not including year 2000 with and without interaction terms. These were carried out using OLS, where the null values in the dependent variable were assumed to be genuinely missing.3 4 The paper includes the pooled sample not including year 2000 with and without interaction terms because these results will be compared to the regression results for estimating the model for ‘energy’ and ‘other’ bilateral trade flows, which do not contain 2000 observations as mentioned earlier. On the effect of economic size on bilateral trade flows, consistent with the expectations of the gravity model, the impact of the product of GDPs on total bilateral trade flows is positive and highly significant, while the coefficients of the product of GDPPCs, are mostly positive and mostly significant. Also consistent with 34 Since heteroscedasticity might be a problem because pairs of countries are likely to be highly dependent across years, standard errors were calculated on the basis o f the white heteroscedasticity- consistent matrix, which produce robust standard errors. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 3.1. Regression results from estimating the gravity model for World total bilateral trade using OLS, where the null values in the dependent variable are assumed genuinely missing. 1985 Coef. (R.Std.Er) 1990 Coef. (R.Std.Er) 1995 Coef. (R.Std.Er) 1997 Coef. (R.Std.Er) 2 0 0 0 Coef. (R.Std.Er) Pooled 1985-2000 Coef. (R.Std.Er) Pooled 1985-1997 Coef. (R.Std.Er) Pooled w. erv interaction 1985-1997 Coef. (R.Std.Er) Pooled w. colonial interaction 1985-1997 Coef. (R.Std.Er) Pooled w. border interaction 1985-1997 Coef. (R.Std.Er) gdp 1.04 *** I II *** I 12 *** I 12 *** I n *** 1.09 *** 1.07 *** 1.08 *** 1.07 *** ^ 07 *** (0.03) (0.03) (0 .0 2 ) (0.03) (0 .0 2 ) (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) gdppc -0.03 -0.06 0.05 * 0.05 * 0 .0 1 0 .0 2 * 0.03 ** 0.03 ** 0.03 ** 0.03 ** (0.04) (0.04) (0.03) (0.03) (0.03) (0 .0 1 ) (0 .0 2 ) (0 .0 2 ) (0 .0 2 ) (0 .0 2 ) distance -1.60 *** -1.44 *** -1.41 *** -1 37 *** -1 32 *** -1.38 *** -1 41 *** -1.42 *** -1.41 *** -1 41 *** (0.07) (0.07) (0.05) (0.05) (0.05) (0.03) (0.03) (0.03) (0.03) (0.03) areas -0.27 *** -0.28 *** “0 18 *** -0.13 *** -0.09 *** -0.14 *** -0.15 *** -0.15 *** -0.15 *** -0.15 *** (0.05) (0.05) (0.03) (0.03) (0.03) (0 .0 2 ) (0 .0 2 ) (0 .0 2 ) (0 .0 2 ) (0 .0 2 ) 11 0 .2 0 ~0 45 -0 37 *** -0 29 *** -0 .2 1 *** -0.31 *** -0.34 *** -0 33 *** -0.34 *** “ 0 34 *** (0 .1 2) (0 . 1 1) (0.07) (0.07) (0.07) (0.04) (0.04) (0.04) (0.04) (0.04) border 0 .0 1 0.80 *** 0.51 ** 0.56 *** 1.26 *** 0.76 *** Q 5 7 *** 0.55 *** 0.57 *** O g7 *** (0.33) (0.25) (0 .2 2 ) (0.19) (0 .2 1 ) (0 .1 1) (0 .1 2 ) (0 .1 2) (0 .1 2) (0 .2 1 ) border_'85 -0 .2 2 (0.37) border_'90 (D) border_'95 -0.38 (0.29) border_'97 -0.47 * (0.26) language 0.49 *** 0.58 *** 0 .6 6 *** 0.60 *** 0 .6 8 *** 0.61 *** 0.58 *** 0.55 *** 0.58 *** 0.58 *** (0.15) (0.13) (0 .1 0 ) (0 .1 0) (0 .1 0) (0.05) (0.06) (0.06) (0.06) (0.06) regional 1.25 *** 1.16 *** I 0 9 *** 1.31 *** I ig *** 1.34 *** 1 33 *** 1 30 *** 1 3 3 *** 1 33 *** (0.32) (0.30) (0.26) (0.25) (0.29) (0.13) (0.14) (0.14) (0.14) (0.14) Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 3.1. Regression results from estimating the gravity model for World total bilateral trade using OLS, where the null values in the dependent variable are assumed genuinely missing (continued) 1985 Coef. (R.Std.Er) 1990 Coef. (R.Std.Er) 1995 Coef. (R.Std.Er) 1997 Coef. (R.Std.Er) 2 0 0 0 Coef. (R.Std.Er) Pooled 1985-2000 Coef. (R.Std.Er) Pooled 1985-1997 Coef. (R.Std.Er) Pooled w. erv interaction 1985-1997 Coef. (R.Std.Er) Pooled w. colonial interaction 1985-1997 Coef. (R.Std.Er) Pooled w. border interaction 1985-1997 Coef. (R.Std.Er) nation (D) C D ) (D) (D) (D) (D) (D) (D) (D) (D) colonizer 0.36 ** 0.36 ** Q 0.80 *** 0.61 *** 0.59 *** 0.57 *** 0.58 *** 0.57 *** 0.57 *** (0.19) (0.18) (0.15) (0.15) (0.14) (0.07) (0.08) (0.08) (0.08) (0.08) colonial 0.44 0.67 0.56 0.16 0.53 0.63 * 0.63 * 0.60 * 1 .1 2 ** 0.63 * colonialj (0.70) (0.81) (0.75) (0.76) (0.92) (0.34) (0.36) (0.36) (0.52) -0.33 (0.30) (0.36) erv erv_'85 erv_'90 ervJ95 erv_'97 -1.56 * (0.84) 0 .0 2 (0.45) -0.18 (0.39) -1.33 ** (0.56) 6.73 *** (0.76) -0.06 (0.23) -0.63 *** (0.24) 2 .0 1 *** (0.63) 0.09 (0.35) -1.05 *** (0.30) -2.28 *** (0.45) -0.63 *** (0.24) -0.64 *** (0.24) cu 1.06 * 1 7 9 *** 1 84 *** 2.05 *** 1.06 1 54 *** 1.82 *** 1 .8 6 *** 1 82 *** 1.81 *** (0.60) (0.42) (0.47) (0.34) (0.69) (0.24) (0 .2 2 ) (0 .2 2 ) (0 .2 2 ) (0 .2 2 ) gov_ds 0 19 *** 0 1 4 *** 0.15 *** 0 14 *** 0 .1 1 *** 0 0 9 *** 0 1 0 *** 0.13 *** 0 .1 0 *** 0 *** (0.04) (0.03) (0 .0 2 ) (0 .0 2 ) (0 .0 2 ) (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) cpy 0.27 (0.25) (D) CD) (D) (D) 0.14 (0.18) 0 .2 1 (0.18) 0 .1 0 (0-19) 0 .2 1 (0.18) 0 .2 2 (0.18) Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 3.1. Regression results from estimating the gravity model for World total bilateral trade using OLS, where the null values in the dependent variable are assumed genuinely missing (continued) 1985 Coef. (R.Std.Er) 1990 Coef. (R.Std.Er) 1995 Coef. (R.Std.Er) 1997 Coef. (R.Std.Er) 2 0 0 0 Coef. (R.Std.Er) Pooled 1985-2000 Coef. (R.Std.Er) Pooled 1985-1997 Coef. (RStd.Er) Pooled w. erv interaction 1985-1997 Coef. (R-Std.Er) Pooled w. colonial interaction 1985-1997 Coef. (R.Std.Er) Pooled w. border interaction 1985-1997 Coef. (R.Std.Er) epe -0 .6 8 *** -0.36 *** 0.31 *** 0.04 -0.09 -0.06 -0 .1 2 ** -0.07 -0 .1 2 ** -0 .1 2 ** (0.18) (0 .1 2) (0.09) (0.08) (0.07) (0.04) (0.05) (0.05) (0.05) (0.05) gini -0 .0 1 0 .0 0 2 0 .0 1 *** 0.0009 -0 .0 1 ** -0.003 * -0 .0 0 1 0 .0 0 1 -0 .0 0 1 -0 .0 0 1 (0 .0 1 ) (0 .0 0 ) (0 .0 0 ) (0 .0 0 ) (0 .0 0 ) (0 .0 0 ) (0 .0 0 ) (0 .0 0 ) (0 .0 0 ) (0 .0 0 ) m2 gdp 0 .0 0 0 2 *** 0 .0 0 0 2 *** 0 .0 0 0 2 *** 0 .0 0 0 2 *** 0 .0 0 0 1 *** 0 .0 0 0 1 *** 0 .0 0 0 2 *** 0 .0 0 0 2 *** 0 .0 0 0 2 *** 0 .0 0 0 2 *** (0 .0 0 0 ) (0 .0 0 0 ) (0 .0 0 0 ) (0 .0 0 0 ) (0 .0 0 0 ) (0 .0 0 0 ) (0 .0 0 0 ) (0 .0 0 0 ) (0 .0 0 0 ) (0 .0 0 0 ) current -0.23 ** -0 .0 2 0 .0 1 0.03 -0.07 -0 .0 1 0.06 * 0 .0 2 0.06 * 0.06 * (0 . 1 0) (0.09) (0.07) (0.06) (0.06) (0.03) (0.03) (0.03) (0.03) (0.03) capital -0.55 *** -0.56 *** -0 29 *** -0.18 * -0.06 -0.30 *** -0.40 *** -0.37 *** -0 40 *** _ q 40 (0 . 1 2) (0 .1 0) (0.07) (0.09) (0 .1 0) (0.03) (0.04) (0.04) (0.04) (0.04) _cons -15.00 *** -20.61 *** -26.13 *** -26.06 *** -25.82 *** -23.36 *** -22.26 *** -2 2 .8 6 *** -22.27 *** -22.28 *** (1.70) (1.43) (0.91) (0.94) (0.92) (0.45) (0.52) (0.53) (0.52) (0.52) No.of Obs 1652 2134 2998 2992 3645 13421 9776 9776 9776 9776 R-squared 0.65 0.67 0.75 0.75 0.73 0.71 0.71 0.71 0.71 0.71 RMSE 2 .0 2 1.99 1.79 1.78 1 .8 6 1.90 1.90 1.89 1.90 1.90 Notes: 1) D stands for dropped. 2) Standard errors are beneath the coefficients. 3) ***, **, * denotes significance at 1, 5, and 10% respectively. 76 expectations, the effects on total bilateral trade of distance, area?,, and LL are negative and highly significant. Regarding the transaction-cost reducing variables: border, language, regional, colonizer, colonial and CU, they all have positive effects on total trade flows and are mostly highly significant, with the exception of colonial, which is significant only in the pooled results. The transaction-cost increasing variable, ERV, has its coefficients mostly negative and mostly significant. Interaction terms have been used for a number of variables, namely border, colonial, and ERV?5 This is done for those cases in which from the separate year regression, the results seemed to reflect systematic variation in the parameters over time. Their role is to capture the change in the effect of the variable/s for the period or the year/s indicated. For border, we have the original variable, border, together with the yearly interaction terms in the same regression (column 10). As explained earlier, the effect on total bilateral trade flows of border is positive and highly significant, on the other hand, the yearly interaction variables have negative signs that are decreasing. This 35 For interaction terms that capture changes over the entire period, such as the one for colonial, a new variable, call it ‘Time’ was created, which took the value 0 for the first year, which is 1985, 1 for the second year - 1990, 2 for the third year - 1995, 3 for the fourth year - 1997, and 4 for the fifth year - 2000. The variable concerned was then multiplied by this time variable to create colonial_i. As for interaction terms that capture changes over the years indicated, such as the ones for border and ERV, a number of variables, call them yl till y5 were created, for each year, the value of the variable is equal 1 for that specific year and 0 otherwise. For instance y l= l in 1985 and 0 otherwise, y2=l in 1990 and 0 otherwise, and so on. The concerned variable was then multiplied by these new variables to create the interaction variables. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 77 means that the effect of the border variable is declining over time. This could possibly be due to the fact that transaction and information costs have fallen. A colonial interaction term is used to see if the effect of a colonial relationship on trade flows varies over time, as mentioned in the expected signs section. Indeed this is the case; the original colonial variable itself has a positive sign but the interaction term has a negative sign. This means that the effect is decreasing each year by 0.33 points, indicating that perhaps after 3 periods the effect will change from positive to negative. This is not hard to grasp since the effects of a colonial relationship may have been strong right after the end of the colonial relation but could be declining over time. Compared to the other two variables, ERV has the most dramatic changes over time. The original ERV variable itself is not included in the same regression with the interaction terms because its effect on bilateral trade flows does not seem to have a time trend, it just seems to vary over time. As you can see, the ERV interaction terms show that the effect of ERV is not only declining over time but also changing from positive to negative. This could be due to changes in the world economic conditions such as the strong and fast economic development of the Asian tigers before the 1997 financial crises and the expected introduction of the Euro by the end of 2000. The former, through significant increases in trade, could have suggested that trade flows are not really affected by ERV, while the expected introduction of the Euro could have rendered ERV between a number of countries (EU) zero. Hence with no effect on trade flows. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 78 Concerning the impact on total trade flows of the new variables that represent my contribution to the literature, consistent with expectations, Gov_ds and M2GDP have positive signs and are highly significant throughout. CPY is not significant at any conventional level indicating that perhaps its effect on total trade flows is negligible. Consistent with expectations, CPE has mostly a negative impact on trade flows and is mostly significant.3 6 The Gini coefficients are mostly insignificant at any conventional level indicating Gini’s negligible effect on trade flows.3 7 At first glance, the coefficients of current account restrictions, contrary to expectations, are mostly positive and mostly insignificant. However, further investigations proved that the correlation between current and capital account restrictions although not high (~ 0.45) is the cause of this reversed sign. When capital account restrictions were dropped, current account restrictions became negatively signed (except in 1997) and significant. To make this story short, the current account restrictions variable seems to adversely affect trade after all. The last variable, capital account restrictions, has negative coefficients throughout, and is mostly highly significant, which is consistent with the gravity model expectations.3 8 As for the intercept, it is negative and highly significant throughout, which indicates that the 36 Further investigations showed that the correlation between the CPE and the Gini variables in 1995 & 1997 (-0.51) might have caused the positive sign of the CPE variable in both years. When the same specification is run again after dropping the Gini variable, CPE became negative (-0.05) but insignificant in 1995, while it became negative (-0.18) and significant at the 1% level in 1997. 37 Similar investigations to the one carried on CPE showed no changes in the sign and little changes in the significance of the Gini variable when the CPE variable was dropped, indicating that the correlation between the CPE and the Gini variables is not the cause for the latter’s sometimes positive sign. 3 8 Because of its strong effect, the correlation with current account restrictions did not affect capital account restrictions’ signs or significance. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 79 unmeasured trade distortions tend to reduce trade rather than promote it. Overall, the fact that the values of R2 based on this model are never lower than 0.65, while the values of RMSE are never higher than 2.02, provides strong support for the goodness of fit of that model.3 9 Based on the regressions of column 6 (pooled 1985-2000)40 another table was prepared, Table 3.1b, that shows the effect on total trade flows of increases in variables either by one standard deviation around the mean for the continuous variables or from 0 to 1 in binary dummy variables or from 0 tol/1 to 2, or 0 to 2 in other dummy variables.4 1 For instance, a one standard deviation increase in GDP increases total trade flows by 1673% or more than 16 times, while a one standard deviation increase in distance decreases trade flows by 66%. Following GDP, variables that have large effects on trade flows are: currency union, which indicates that countries sharing a common currency trade three times more than countries that do not share a common currency (consistent with Rose’s (2000) findings); forming a regional agreement; and having a common border. 39 The main reason I did not exclude one of the variables that are correlated with each other and that caused signs’ reversals is that the correlations usually were not high - not reaching 0.55 in all cases while other variables were highly correlated and showed no such collinearity problems as GDP and GDPPC. In other words, I didn’t suspect that such relatively low collinearity would cause such problems. 0 I chose pooled from 1985-2000 because it has the largest number of observations, and thus would provide better predictions as opposed to any other sample. Note: 1) Because the regression used here is based on the pooled data, the standard deviations of some of the variables are abnormally large like that of M2GDP, Gini, and GDP, which means that a one standard deviation increase in either of them is not just a small increase or a one percent increase. 2 ) The results of the predictions should not be taken too seriously because they depend on the specifications and the samples used. Predictions using other specification and sample to the one used above will show different results. This is especially true to the sensitive variables explained later. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 80 Table 3.1b. Predictions based on column 6 Table 3.1 A=Coef. B=Std. Dev. A*B exp(A*B) - 1 exp(A) -1 gdp 1.0861 2.6471 2.8751 1673% 1673% gdppc 0.0247 2.0447 0.0505 5% 5% distance -1.3755 0.7799 -1.0728 -6 6 % -6 6 % areas -0.1381 1.3641 -0.1884 -17% -17% 11 -0.3096 0.5228 -27% * -46% ** -27% * border 0.7605 0.1698 114% 114% language 0.6092 0.3530 84% 84% regional 1.3403 0.1128 282% 282% nation C D ) colonizer 0.5917 0.2875 81% 81% colonial 0.6300 0.0431 8 8 % 8 8 % erv -0.0596 0.0814 -0.0048 0 % 0 % cu 1.5442 0.0789 368% 368% gov_ds 0.0905 1.8384 0.1664 18% 18% cpy 0.1422 0.0821 15% * 33% ** 15% * cpe -0.0558 0.5747 -5% * - 1 1 % ** -5% * gini -0.0030 13.3970 -0.0395 -4% -4% m2 gdp 0 .0 0 0 1 1971.9340 0.2749 32% 32% current -0.0072 0.7214 -1% * -1% ** - 1 % * capital -0.2991 0.5267 -26% * -45% ** -26% * * change due to change in dummy varaible from either 0 to 1 or 1 to 2 ** change due to change in dummy variable from 0 to 2 Other variables that have non-minor positive effects on total trade flows include a colonial relationship, common language, and common colonizer. M2GDP and Gov_ds suggest that a one standard deviation increase in M2GDP and Govjds will increase trade flows by 32% and 18% respectively, which is less than a one half. The variable with the largest negative effect on trade is of course distance. LL, capital account restrictions and areas follow it, which is very consistent with expectations. Minor effects on trade are associated with variables such as CPE, GDPPC, Gini, and current account restrictions. As for ERV, it has a negligible effect on trade flows. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. As mentioned in the beginning of chapter 3, a standard log-linear model with a log-normally distributed error term cannot, by definition, explain zero trade flows and thus using OLS is not advisable. However the paper will use OLS in estimating the coefficients of Table 3.2 to check variables’ sensitivities to the different assumptions and then compare them to Table 3.3, which uses Tobit. Just like Table 3.1, Table 3.2 presents the regression results from estimating the gravity model for world total bilateral trade for: each individual year, the whole pooled sample (1985-2000), and the pooled sample not including year 2000 with and without interaction terms but differs from Table 3.1 in that the null values in the dependent variable were assumed to be zeros instead of genuinely missing. To be able to take logs the zero values were presented as ‘0.0001’ instead of a ‘0’ whose log is undefined. As in Table 3.1, all the coefficients of GDP, language, regional, colonizer, CU, and Gov_ds are positive and highly significant (with only a few exceptions). Similarly as in Table 3.1, the coefficients of distance, areas, LL, ERV, and capital account restrictions are negative and highly significant (with also a few exceptions). Concerning ERV, the negative interaction terms enforce the hypothesis that exchange rate volatility adversely affects trade. The increasing and decreasing magnitudes however suggest that the adverse effect of ERV on trade itself varies over time increasing at times and decreasing at others. Two comments should be made at this point. First, with very few exceptions, all coefficients in Table 3.2 have higher magnitudes than the corresponding Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 3.2. Regression results from estimating the gravity model for World total bilateral trade using OLS, where the null values in the dependent variable are assumed zeros. 1985 Coef. (R.Std.Er) 1990 Coef. (R.Std.Er) 1995 Coef. (R.Std.Er) 1997 Coef. (R.Std.Er) 2000 Coef. (R.Std.Er) Pooled 1985-2000 Coef. (R.Std.Er) Pooled 1985-1997 Coef. (R.Std.Er) Pooled w. erv interaction 1985-1997 Coef. (R.Std.Er) Pooled w. colonial interaction 1985-1997 Coef. (R.Std.Er) Pooled w. border interaction 1985-1997 Coef. (R.Std.Er) gdp 2 7 0 *** 2 7 | *** 2.58 *** 2.25 *** 2 27 *** 2.46 *** 2.48 *** 2 .4 9 *** 2.48 *** 2 ,4 9 *** (0.13) (0 .1 1) (0.09) (0.08) (0.09) (0.04) (0.05) (0.05) (0.05) (0.05) gdppc 0 .2 1 0.03 -0.19 * 0.03 0.17 * -0.04 -0.05 -0.06 -0.05 -0.05 (0.16) (0.14) (0 .1 0) (0.09) (0 .1 0) (0.04) (0.05) (0.05) (0.05) (0.05) distance -3.16 *** -2.90 *** -3 17 *** -2.59 *** -2 31 *** -2.73 *** -2 89 *** - 2 91 *** -2.89 *** -2.89 *** (0.28) (0.24) (0.18) (0.17) (0.17) (0.09) (0 .1 0) (0 . 1 0) (0 . 1 0) (0 . 1 0) areas -1.19 *** -0.99 *** -1 2 0 *** -0.57 *** -0.84 *** -0 92 *** - 0 92 *** -0 92 *** -0 92 *** (0.18) (0.16) (0 .1 2) (0 . 1 1) (0 .1 1 ) (0.06) (0.07) (0.07) (0.07) (0.07) 11 -0.77 * -1.33 *** -0.05 -0.69 *** -1.28 *** -0.93 *** -0 72 *** -0.69 *** .0.72 *** -0.72 *** (0.44) (0.38) (0.25) (0.24) (0.23) (0.13) (0.15) (0.15) (0.15) (0.15) border -0.23 0.93 -0.89 -0.37 0.73 0 .1 1 -0.13 -0.16 -0.13 0.95 (1.35) ( 1.0 0 ) (0.56) (0.52) (0.59) (0.32) (0.38) (0.38) (0.38) (0.85) borderJ85 border_'90 borderJ95 border_'97 language 1 9g *** 1.84 *** 1.84 *** 2 *** 2.28 *** 2.08 *** 2.03 *** 2 .0 0 *** 2 03 *** -0.26 (1.43) (D) -1.35 (0.97) -2 .0 2 ** (0.95) 2 03 *** (0.54) (0.49) (0.32) (0.31) (0.38) (0.18) (0 .2 0 ) (0 .2 0 ) (0 .2 0 ) (0 .2 0 ) regional 1.75 * 1.85 * -0.94 0.06 1 .0 1 0.85 ** 0.75 * 0.71 0.75 * 0.74 * (1.04) (1.09) (0.73) (0.64) (0.65) (0.37) (0.44) (0.44) (0.44) (0.44) oo to Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 3.2. Regression results from estimating the gravity model for World total bilateral trade using OLS, where the null values in the dependent variable are assumed zeros (continued). 1985 Coef. (R.Std.Er) 1990 Coef. (R.Std.Er) 1995 Coef. (R.Std.Er) 1997 Coef. (RStd.Er) 2 0 0 0 Coef. (R.Std.Er) Pooled 1985-2000 Coef. (R.Std.Er) Pooled 1985-1997 Coef. (R.Std.Er) Pooled w. erv interaction 1985-1997 Coef. (R.Std.Er) Pooled w. colonial interaction 1985-1997 Coef. (R.Std.Er) Pooled w. border interaction 1985-1997 Coef. (R.Std.Er) nation (D) (D) (D) (D) (D) (D) (D) (D) (D) C D ) colonizer 2.54 *** 2.84 *** I 9 2 *** j 70 *** I 9 0 *** 2.09 *** 2 .1 2 *** 2 13 *** 2 1 2 *** 2 1 2 *** (0.65) (0.58) (0.47) (0.47) (0.55) (0.24) (0.27) (0.27) (0.27) (0.27) colonial -3.88 -3.53 -2.84 -2.71 -1.93 -2.65 ** -2.95 ** -2.98 ** -1.58 -2.95 ** colonialj (2.57) (2.85) (2.57) (2.32) (2.47) (1.14) (1.30) (1.30) (2.04) -0.92 ( 1.1 0) (1.30) erv erv_'85 erv_'90 erv_'95 erv_'97 -12.54 *** (2 .8 8 ) 0.38 (1.56) -6 .6 8 *** (1.55) -9 93 *** (2 .0 2 ) 11.82 *** (2 .6 8 ) -3.64 *** (0.83) -5.49 *** (0 .8 8 ) -7.12 *** (2.24) -3.02 ** ( 1.2 0 ) -6 .2 2 *** (118) -8 .6 6 *** (1.60) -5.48 *** (0 .8 8 ) -5.50 *** (0.89) cu 5.15 *** 4.28 ** 4.82 *** 4 9 2 *** 5.88 *** 5 2 1 *** 4 g9 *** 4.88 *** 4 39 *** 4.86 *** (1.39) (2 .0 1 ) (0.80) (1.15) (1.28) (0.60) (0.67) (0.67) (0.67) (0.67) gov_ds 0.18 0.42 *** 0.76 *** 0.31 *** 0.33 *** Q 40 *** 0.47 *** 0.52 *** 0 4 7 *** 0.47 *** (0.13) (0.13) (0.08) (0.07) (0.08) (0.04) (0.04) (0.04) (0.04) (0.04) cpy -1.51 (1.03) (D) (D) (D) (D) -2.25 *** (0.84) -2.14 ** (0.85) -2 .0 0 ** (0.85) -2.14 ** (0.85) -2.14 ** (0.85) Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 3.2. Regression results from estimating the gravity model for World total bilateral trade using OLS, where the null values in the dependent variable are assumed zeros (continued). 1985 Coef. (R.Std.Er) 1990 Coef. (R.Std.Er) 1995 Coef. (R.Std.Er) 1997 Coef. (R.Std.Er) 2 0 0 0 Coef. (R.Std.Er) Pooled 1985-2000 Coef. (R.Std.Er) Pooled 1985-1997 Coef. (R.Std.Er) Pooled w. erv interaction 1985-1997 Coef. (R.Std.Er) Pooled w. colonial interaction 1985-1997 Coef. (R.Std.Er) Pooled w. border interaction 1985-1997 Coef. (R.Std.Er) cpe -1.53 ** -0.52 -0 .0 1 1 09 *** 0.67 *** 0.14 -0 .1 0 -0 .1 0 -0 .1 0 -0 .1 0 (0 .6 6 ) (0.44) (0.31) (0.26) (0.25) (0.14) (0.17) (0.18) (0.17) (0.17) gini -0.06 *** -0.03 * 0 .0 1 0.003 .0 04 *** -0.03 *** -0 .0 2 *** -0 .0 2 *** -0 .0 2 *** -0 .0 2 *** (0 .0 2 ) (0 .0 2 ) (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) m2 gdp -0.0003 -0 .0 0 0 2 -0 .0 0 0 1 * -0.00003 0 .0 0 0 1 *** -0.00006 ** -0 .0 0 0 1 *** -0 .0 0 0 1 *** -0 .0 0 0 1 *** -0 .0 0 0 1 *** (0 .0 0 0 ) (0 .0 0 0 ) (0 .0 0 0 ) (0 .0 0 0 ) (0 .0 0 0 ) (0 .0 0 0 ) (0 .0 0 0 ) (0 .0 0 0 ) (0 .0 0 0 ) (0 .0 0 0 ) current -0 .6 8 * 0.63 ** 0 7 0 *** 0.17 0.84 *** 0 27 *** 0.23 ** 0 .2 0 * 0.23 ** 0.23 ** (0.37) (0.32) (0.23) (0.18) (0 .2 1 ) (0 .1 0) (0 . 1 1) (0 -1 2) (0 .1 1) (0 .1 1) capital -0.43 -1.83 *** -1.77 *** -0.13 -1.27 *** -1.27 *** -1.33 *** -1.29 *** -1 33 *** -1.33 *** (0.42) (0.36) (0.24) (0.34) (0.34) (0 .1 2) (0.13) (0.14) (0.13) (0.13) _cons -68.91 *** -75.13 *** -63.94 *** -65.66 *** -68.28 *** -66.04 *** -64.48 *** -64.79 *** -64.50 *** -64.57 *** (6 .2 2 ) (4.90) (3.08) (3.14) (3.31) (1.57) (1.82) (1.85) (1.82) (1.82) No.of Obs 2080 2628 3403 3321 4186 15618 11432 11432 11432 11432 R-squared 0.46 0.47 0.50 0.50 0.45 0.47 0.48 0.48 0.48 0.48 RMSE 8 .2 2 7.91 6.67 6.28 7.28 7.25 7.22 7.21 7.22 7.22 Notes: 1) D stands for dropped. 2) Standard errors are beneath the coefficients. 3) ***, **, * denotes significance at 1, 5, and 10% respectively. oo 4 * . 85 coefficients in Table 3.1. This is because OLS that used to estimate Table 3.1, by excluding zero trade flows (if they were actually zero), exerts -by definition - a downward bias on the coefficients. Second, it could be stated that all the above- mentioned variables with positive or negative effects on total trade flows are so far fairly robust to the different assumptions. Unlike Table 3.1, the impact on trade flows of GDPPC and border are mostly insignificant.4 2 Surprisingly, the effect of colonial on trade flows is negative and mostly significant for the pooled data. This is an outright contrast to the assumption that countries with a colonial relationship trade more with each other than pairs of countries that do not share such characteristic. This could actually suggest that the reverse is what happened: after independence, previously colonized countries tried to limit ties, at least in terms of lowering trade, with their colonizers and as time passed by and as the independent countries felt more secure with their independence trade ties got better. Concerning CPY, unlike in Table 3.1, its effect on trade flows is negative and mostly significant. This probably allows the self-sufficiency argument to outweigh the reinforcement of ideological stance argument. As for CPE, its coefficients are 42 For GDPPC, this might suggest that most of the effect of economic size is captured by the GDP variable and thus GDPPC has little to add making its effect on trade flows negligible. As for border, although insignificant, border’s effect on trade flows might be negative sometimes as shown when zero trade flows are included because neighboring countries might face border disputes, which might negatively impact trade, than non-neighboring countries. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 8 6 mostly insignificant suggesting that it has a negligible effect on trade flows.4 3 Regarding the Gini variable, its effect on trade flows is negative throughout, with a couple of exceptions, and also significant mostly at the 1% level.4 4 This indicates that pairs of countries that enjoy better income distribution trade more with each other than country pairs that suffer from high income inequality, consistent with expectations. Astonishingly, the M2 GDP coefficient is not only negative throughout (except in 2000) but also mostly highly significant, especially in the columns with pooled data. The paper cannot find an appropriate explanation for this surprising result.4 5 As for the current account restrictions variable, it is mostly positive and significant. However, further investigations revealed that correlation with capital account restrictions was the cause of most positive current account restrictions’ coefficients and when capital account restrictions was dropped, current account restrictions became negative, but were largely insignificant indicating that current account restrictions has a negligible effect on trade flows. As in Table 3.1, the intercept has negative and highly significant coefficients throughout. So far it is obvious that variables like GDPPC, border, colonial, CPY, CPE, Gini, M2GDP, and current account restrictions are sensitive to the different assumptions used. As for the 43 When Gini was dropped, which is correlated to CPE, some CPE coefficients became negative but were still insignificant 44 In 1997 excluding the CPE variable in the further investigations caused the Gini variable to become negatively signed and significant at 5% level. 45 Closer investigations showed that changes that occurred in the signs of GDPPC, border, colonial, CPY, and M2GDP as compared to Table 3.1 were not a problem of correlation to other variables. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 87 goodness of fit, it has fallen dramatically where the greatest value for R2 is 0.5 while the smallest value for RMSE is 6.67. Once again, based on the regressions of column 6 (pooled 1985-2000) but this time for Table 3.2, another table was prepared, Table 3.2b, that presents the effect on total trade flows of increases in continuous and dummy variables. The largest effects are associated with changes in the GDP and the CU variables, just as Table 3.1b showed. Following are the effects of a common colonizer, a common language, a regional agreement, and a colonial relationship. This is quite similar to the findings in Table 3.1b with the exception of border, which lagged behind because of the inclusion of zero trade flows in the dependent variable. It is obvious that the magnitudes of the changes are much larger than the magnitudes presented in Table 3.1b because Table 3.2 itself has higher magnitudes for the coefficients due to OLS and its downwardly biased coefficients when excluding zero dependent variables. Gov_ds is not far behind; where a one standard deviation increase cause trade flows to increase by 113%. Consistent with the findings in Table 3.1b, current account restrictions, Gini, ERV, CPE, and GDPPC are at the tail of the list although shuffled this time, GDPPC and CPE were even insignificant reflecting negligible effect on trade flows. Added to them are the border and the M2GDP variables that were very sensitive to the assumption that assigned zero values to the null values in the dependent variables instead of considering them genuinely missing. Once again, distance, capital account restrictions, areas, and LL have some of the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 88 Table 3.2b. Predictions based on column 6 Table 3.2 A=Coef. B=Std. Dev. A*B exp(A*B) - 1 exp(A) - 1 gdp 2.4574 2.8101 6.9055 99677% 99677% gdppc -0.0433 2.0580 -0.0891 -9% -9% distance -2.7315 0.7536 -2.0584 -87% -87% areas -0.8429 1.3314 -1.1223 -67% -67% 11 -0.9265 0.5460 -60% * -84% ** -60% * border 0.1113 0.1587 1 2 % 1 2 % language 2.0796 0.3444 700% 700% regional 0.8538 0.1047 135% 135% nation (D) colonizer 2.0923 0.2867 710% 710% colonial -2.6531 0.0400 -93% -93% erv -3.6398 0.0845 -0.3076 -26% -26% cu 5.2116 0.0744 18239% 18239% gov_ds 0.4043 1.8732 0.7573 113% 113% cpy -2.2484 0.0909 -89% * -99% ** -89% * cpe 0.1391 0.5673 15% * 32% ** 15% * gini -0.0273 13.5047 -0.3680 -31% -31% m2 gdp -0 .0 0 0 1 1883.9120 -0.1106 - 1 0 % - 1 0 % current 0.2702 0.7198 31% * 72% ** 31% * capital -1.2733 0.5143 -72% * -92% ** -72% * * change due to change in dummy varaible from either 0 to 1 or 1 to 2 ** change due to change in dummy variable from 0 to 2 largest negative effects on trade flows to be only preceded by colonial and CPY, which like border and M2GDP were shown to be very sensitive to the assumption made. In Table 3.3, the Tobit limited dependent variable technique is used to estimate the gravity model for world total bilateral trade flows for only the pooled sample including year 2000 for both the base model and with ERV interaction terms. The objective of this exercise is to investigate whether the coefficients are sensitive to the estimation technique used. By definition, the Tobit technique is used when the dependent variable could have limited values - in our case zero values, and thus the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 3.3. Regression results from estimating the gravity model for World total bilateral trade using Tobit, where the null values in the dependent variable are assumed zeros. Pooled 1985-2000 Coef. (Std. Err.) Pooled w. erv interaction 1985-2000 Coef. (Std. Err.) gdp 2 7 2 *** 2 72 *** 0.05 0.05 gdppc -0.06 -0.06 0.06 0.06 distance -2 98 *** -2 99 *** 0.11 0.11 areas -0.99 *** -0.98 *** 0.07 0.07 1 1 -1.08 *** -1.07 *** 0.14 0.14 border -0.06 -0.07 0.48 0.48 language 2.36 *** 2.35 *** 0.22 0.22 regional 0.68 0.67 0.67 0.67 nation (D) (D) colonizer 2.42 *** 2.43 *** 0.27 0.27 colonial -3.33 ** -3.34 ** 1.70 1.70 erv -4.32 *** 0.85 erv_'85 -7.59 *** 2.01 erv_'90 -2.62 ** 1.22 erv_'95 -4.92 *** 1.21 erv_'97 -6.30 *** 1.58 ervJOO 1.27 2.14 cu 5.87 *** 5.87 *** 0.94 0.94 gov_ds 0 .4 7 *** q 4 9 *** 0.04 0.04 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 90 Table 3.3. Regression results from estimating the gravity model for World total bilateral trade using Tobit, where the null values in the dependent variable are assumed zeros (continued). Pooled 1985-2000 Coef. (Std. Err.) Pooled w. erv interaction 1985-2000 Coef. (Std. Err.) cpy -2.69 *** -2 39 *** 0.77 0.77 cpe 0.18 0.08 0.16 0.16 gini -0.03 *** -0.03 *** 0 .0 1 0 .0 1 m2 gdp -0.0001 ** -0.0001 ** (0 .0 0 0 0 ) (0 .0 0 0 0 ) current 0.36 *** 0.38 *** 0 .1 1 0 .1 1 capital -1.47 *** -1.51 *** 0.15 0.15 _cons -74.20 *** -73.78 *** 1.85 1 .8 8 _se 8.33 8.33 0.05 0.05 No. of obs 15618 15618 uncensored 13421 13421 left-censored 2197 2197 Pseudo R2 0.09 0.09 Notes: 1) D stands for dropped. 2) Standard errors are beneath the coefficients. 3) ***, **, * denotes significance at 1, 5, and 10% respectively. observations are left censored. Comparing Table 3.3 to Table 3.2 column 6, one notices that there is no difference in the signs of any of the coefficients (except for border, which is insignificant so its sign doesn’t really matter), while significance varies only slightly in only some variables (except for regional, which became insignificant). The only major difference is that the magnitudes of the coefficients Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 91 estimated by Tobit are larger than their corresponding ones in Table 3.2 estimated by OLS except for the border and the regional variables. This supports the argument that OLS indeed downwardly bias the coefficients, but the variables as a whole, except for regional, were not sensitive to the estimation technique. Our next exercise is to introduce the lagged dependent variable as one of the explanatory variables. The purpose of this exercise is to investigate whether the previous trade values are related to their current values. The effect of the lagged dependent variable on trade flows if positive, which is what is expected, indicates persistence of tastes and preferences. The results for such exercise are presented in Table 3.4. Table 3.4 presents the regression results from estimating the gravity equation, with the lagged dependent variable included in the right hand side for world total bilateral trade using OLS for the whole pooled sample (1985-2000) and for the pooled sample not including year 2000. In the first two columns the null values were assumed to be genuinely missing and in the last two columns the null values were assumed to be zeros. The impact of the lagged dependent variable on trade flows is positive and highly significant throughout, which is consistent with expectations. Comparing column 1 in Table 3.4 to its corresponding match but without the lagged dependent variable - column 6 in Table 3.1, notice that most of the variables as well as the intercept have similar signs and significance (except for the significance of current Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 3.4. Regression results from estimating the gravity model for World total bilateral trade using OLS, where a lagged dependent variable is included as one of the explanatory variables. Null values assumed genuinely Null values assumed zeros missing Pooled Pooled Pooled Pooled (1985-2000) (1985-1997) (1985-2000) (1985-1997) Coef. Coef. Coef. Coef. (R.Std.Err) (R.Std.Err) (R.Std.Err) (R.Std.Err) lagged bilat 0.08 *** 0.08 *** q 4 2 *** q 4 4 (0 .0 0 ) (0 .0 0 ) (0 .0 1 ) (0 .0 2 ) gup 0.96 *** 0 9 7 *** 1.46 *** 2 49 (0 .0 2 ) (0 .0 2 ) (0.06) (0.07) gdppc -0 .0 2 -0.03 -0.16 *** -0.27 *** (0 .0 2 ) (0 .0 2 ) (0.05) (0.07) distance -1.16 *** -1 23 *** -1.45 *** -1.69 *** (0.03) (0.04) (0 . 1 0) (0.13) areas -0 13 *** -0 .2 0 *** -0.53 *** -0 70 *** (0 .0 2 ) (0.03) (0.06) (0.08) 11 -0 31 *** -0 37 *** -0 .6 8 *** -0.24 (0.04) (0.06) (0.14) (0.19) border 0.84 *** 0.55 *** 0.08 -0.42 (0 . 1 2) (0.15) (0.29) (0.37) language 0.50 *** 0.49 *** 1.15 *** q 9 2 *** (0.06) (0.08) (0.19) (0.24) regional p ig *** 1.06 *** 0.48 * -0.06 (0.14) (0.17) (0.28) (0.36) nation C D ) C D ) (D) (D) colonizer 0 3 9 *** 0.34 *** 1.06 *** 0 9 9 *** (0.09) (0 .1 1) (0.27) (0.33) colonial 0.80 ** 0.82 * -1.27 -1.57 (0.38) (0.45) (0.83) (1.05) erv q 9 7 *** 0.47 1.34 -0.29 (0.26) (0.29) (0 .8 8 ) (0.98) cu 1 11 *** 1.58 *** 2.84 *** 1.70 * (0.34) (0.30) (0.75) (1 .0 2 ) gov_ds 0.08 *** 0 .1 0 *** 0 .2 2 *** 0.42 *** (0 -0 1 ) (0 .0 2 ) (0.04) (0.05) cpy (D) (D) (D) (D) cpe 0 .1 0 ** -0.04 0.56 *** 0.004 (0.05) (0.07) (0.16) (0.23) gini 0 .0 0 2 0 .0 1 *** -0 .0 1 * 0.0005 (0 .0 0 ) (0.00) (0.01) (0.01) m2 gdp 0.00014 *** 0.00018 *** -0.00005 * -0 .0 0 0 1 1 ** (0.00001) (0.00002) (0.00002) (0.00005) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 93 Table 3.4. Regression results from estimating the gravity model for World total bilateral trade using OLS, where a lagged dependent variable is included as one of the explanatory variables (continued). Null values assumed genuinely Null values assumed zeros missing Pooled Pooled Pooled Pooled (1985-2000) (1985-1997) (1985-2000) (1985-1997) Coef. Coef. Coef. Coef. (RStd.Err) (R.Std.Err) (RStd.Err) (R,Std.Err) current -0.08 ** 0.03 0.05 0.08 (0.03) (0.05) (0 .1 2) (0.16) capital -0 17 *** -0.32 *** _0 9 9 *** -1.04 *** (0.04) (0.05) (0.13) (0.17) _cons -2 0 .2 1 *** -19.21 *** -38.12 *** ~34 14 (0.57) (0.76) (1.99) (2.50) No.of Obs 8555 4910 9974 5788 R-squared 0.74 0.74 0.57 0.61 RMSE 1.81 1.81 6.55 6.34 Notes: 1) D stands for dropped. 2) Standard errors are beneath the coefficients. 3) ***, **, * denotes significance at 1, 5, and 10% respectively. account restrictions) with the magnitudes of the coefficients slightly different. The exceptions being GDPPC, ERV, CPE, Gini and CPY variables. The first four encountered changes of signs and significance while the latter was dropped. Similar conclusions follow from comparing column 2 in Table 3.4 with its corresponding match in Table 3.1 - column 7. The CPE variable however did not undergo a sign change but it became insignificant. These outcomes perhaps indicate that GDPPC, ERV, CPY, CPE and the Gini variables are sensitive to the specifications employed. Concerning the columns where the null values were assumed to be zeros, compare column 3 in Table 3.4 to its corresponding one in Table 3.2 - column 6. Almost all the variables have similar signs and significance, with the exceptions of Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 94 GDPPC, CPE, colonial, ERV and current account restrictions, where the first two became significant, while the last three became insignificant. Notice that R2 has increased and RMSE has decreased when the lagged dependent variable is included, indicating a better goodness of fit. Similarly, comparing column 4 in Table 3.4 to column 7 in Table 3.2, the significance of a number of variables changed such as GDPPC, LL, regional, colonial, ERV, Gini, and current account restrictions, where only GDPPC became significant, while the rest became insignificant. Also as in column 3, R2 increased and RMSE decreased when the lagged dependent variable was included. So far, the governance variable has proved robust to all different assumptions, estimation techniques, and specifications. In yet another stimulating exercise, the individual governance components: VA, PS, GE, RQ, RL, and CC are introduced as explanatory variables each at a time. The goal behind such exercise is to explore the sensitivities of the variables to these specifications, and to try to determine which governance component is the most important. Table 3.5 presents the regression results from estimating the gravity model, with the individual governance components each at a time, for world total bilateral trade using OLS for the whole pooled sample (1985-2000), where the null values were assumed to be genuinely missing.4 6 This Table can be compared to column 6 in Table 3.1. Almost none of the signs of the variables or their significance changed (with the 46 Like governance, for each of the six components an auxiliary regression was run on GDPPC (to get over the high correlation problem) and the deviation was used to come up with the indices used. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 3.5. Regression results from estimating the gravity model for World total bilateral trade for pooled data (1985- 2000) using OLS, where each of the six governance components is used separately and where the null values in the dependent variable are assumed genuinely missing. Coef. (R.Std.Err) Coef. CR.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R,Std.Err) gdp P09 *** 1.08 *** 1.08 *** 1.08 *** 1 08 *** 1 0 9 *** (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) gdppc 0 .0 2 0.03 ** 0.03 ** 0.03 ** 0 0 4 *** 0.03 * (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) distance -1.37 *** -1.37 *** -1.37 *** -1 37 *** -1 37 *** -1.36 *** (0.03) (0.03) (0.03) (0.03) (0.03) (0.03) areas -0.14 *** -0.13 *** -0.14 *** -0 14 *** -0.13 *** -0 14 *** (0 .0 2 ) (0 .0 2 ) (0 .0 2 ) (0 .0 2 ) (0 .0 2 ) (0 .0 2 ) 11 -0.31 *** -0.32 *** -0.30 *** -0.31 *** -0.33 *** -0.30 *** (0.04) (0.04) (0.04) (0.04) (0.04) (0.04) border 0.76 *** 0.77 *** 0.76 *** 0.76 *** q 7 3 *** 0 7 7 *** (0 .1 1 ) (0 .1 1) (0 .1 1) (0 .1 1) (0 .1 1) (0 .1 1) language 0.62 *** 0.61 *** 0.62 *** 0.62 *** 0.60 *** 0.62 *** (0.05) (0.05) (0.05) (0.05) (0.05) (0.05) regional 1.35 *** 1.35 *** 1.36 *** 1.35 *** 1.35 *** I 3 7 *** (0.13) (0.13) (0.13) (0.13) (0.13) (0.13) nation (D) (D) (D) (D) (D) (D) colonizer 0.59 *** 0.59 *** 0.58 *** 0.59 *** 0.59 *** 0.59 *** (0.07) (0.07) (0.07) (0.07) (0.07) (0.07) colonial 0.61 * 0.60 * 0.63 * 0.62 * 0.65 * 0.64 * (0.34) (0.35) (0.34) (0.35) (0.33) (0.34) erv -0.26 -0.34 -0.004 -0 .1 2 -0.19 -0.32 (0 .2 2 ) (0 .2 2 ) (0.23) (0.23) (0.23) (0 .2 2 ) cu 1.52 *** 1.49 *** 1.52 *** 1.52 *** 1.55 *** j 4 9 (0.24) (0.24) (0.25) (0.24) (0.24) (0.24) Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 3.5. Regression results from estimating the gravity model for World total bilateral trade for pooled data (1985- 2000) using OLS, where each of the six governance components is used separately and where the null values in the dependent variable are assumed genuinely missing (continued). Coef. Coef. Coef. Coef. Coef. Coef. (R.Std.Err) (R.Std.Err) (RStd.Err) (RStd.Err) (R.Std.Err) (RStd.Err) va_d 0.03 *** (0 .0 1 ) ps_d Q Q4 (0 .0 1 ) ge_d 0.03 *** (0 .0 0 ) rq_d 0 .0 2 *** (0 .0 1 ) rl_d 0.09 *** (0 .0 1 ) cc_d 0.04 *** (0 .0 1 ) cpy -0 .0 2 0 .0 0 1 0.05 0.03 0 .1 1 -0.05 (0.18) (0.18) (0.18) (0.18) (0.18) (0.18) cpe 0 .0 1 -0.03 -0 .0 0 1 0.03 -0.05 0.04 (0.04) (0.04) (0.04) (0.04) (0.04) (0.04) gini -0.003 * -0.003 ** -0.004 ** -0.003 ** -0 .0 0 2 -0.003 * (0 .0 0 ) (0 .0 0 ) (0 .0 0 ) (0 .0 0 ) (0 .0 0 ) (0 .0 0 ) m2 gdp 0.00015 *** 0.00014 *** 0.00013 *** 0.00013 *** 0.00014 *** 0.00014 *** (0 .0 0 0 0 1 ) (0 .0 0 0 0 1 ) (0 .0 0 0 0 1 ) (0 .0 0 0 0 1 ) (0 .0 0 0 0 1 ) (0 .0 0 0 0 1 ) current -0.03 -0 .0 1 -0.03 -0.03 0 .0 0 2 -0.05 (0.03) (0.03) (0.03) (0.03) (0.03) (0.03) capital -0.28 *** -0 27 *** -0.34 *** -0 29 *** -0 27 *** -0.25 *** (0.03) (0.03) (0.04) (0.03) (0.03) (0.03) _cons -23.46 *** -23.34 *** -23.00 *** -23.32 *** -23.48 *** -23.54 *** (0.45) (0.45) (0.45) (0.45) (0.45) (0.46) Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 3.5. Regression results from estimating the gravity model for World total bilateral trade for pooled data (1985- 2000) using OLS, where each of the six governance components is used separately and where the null values in the dependent variable are assumed genuinely missing (continued). Coef. (RStd.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) No. of Obs 13421 13421 13421 13421 13421 13421 R-squared 0.71 0.71 0.71 0.71 0.71 0.71 RMSE 1.90 1.90 1.90 1.90 1.90 1.90 Notes: 1) D stands for dropped. 2) Standard errors are beneath the coefficients. 3) ***, **, * denotes significance at 1, 5, and 10% respectively. V O 98 exception of the significance of GDPPC and Gini which differ only slightly). CPY, CPE, and current account restrictions have mixed signs and significance levels that are dramatically different but that still show insignificance so the signs do not really matter. The coefficients of the governance components all have positive signs and are highly significant. Rule of law seems to have the largest effect; possibly suggesting that rule of law may be the governance characteristic with the greatest effect on trade flows. Control of corruption and political stability follow rule of law in importance, and are closely followed by the other governance components. These results, which rank the importance of the governance components, although beneficial, shouldn’t be taken too seriously because these ranks might change with the change of assumptions, estimation techniques, and model specifications. However, the finding that rule of law consistently has the largest impact is by no means counter-intuitive because it is expected that agents/countries would be less inclined to trade with other agents/countries in countries lacking both reliable courts to which they can take their disputes and guarantees against arbitrary treatment. Tables 3.6 and 3.7 present the results of estimating the gravity model separately for energy and other trade. One would expect the gravity considerations to be quite a bit weaker for energy trade than other trade. To some extent this is the case. One can see that the R is considerably lower for the energy trade in Table 3.6 than that of the other trade in Table 3.7. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 99 Table 3.6 and 3.7 present the regression results from estimating the gravity equation for energy and other trade for the whole pooled sample (1985-2000) and the pooled sample not including year 2000 with and without the interaction terms. This is done once where the null values were assumed to be genuinely missing (1st four columns), and once where some of the missing values were assumed to be genuinely missing while the others were assumed to be zeros when they were found zeros in all the other years (last four columns). In both cases the zero values were excluded. One can clearly see that the coefficients of GDP, regional, colonizer and CU are all positively signed and highly significant in both tables, consistent with expectations. Similarly those of Distance, LL, and capital account restrictions as well as the intercept are all negatively signed and highly significant throughout both tables. Also consistent with expectations, the effect on trade flows of language, colonial, and Gov_ds are positive and significant in both Tables, but the significance differs only slightly between energy and other trade. With respect to colonial, in both tables the interaction term is negative meaning that the colonial effect on trade is decreasing as explained previously in Table 3.1. In Table 3.7, consistent with expectations, coefficients of areas, ERV and CPE are negative and mostly highly significant, while those of border and M2GDP are positively and mostly highly significant. These same variables showed some different (yet sometimes intuitive) results as far as energy trade flows are concerned. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 3.6. Regression results for estimating the gravity model for World energy Null values assumed genuinely missing trade using OLS. Some null values assumed genuinely missing , while others are Pooled (1985-1997) Coef. (R.Std.Err) Pooled w. erv interaction (1985-1997) Coef. (R.Std.Err) Pooled w. colonial interaction (1985-1997) Coef. (R.Std.Err) Pooled w. border interaction (1985-1997) Coef. (R.Std.Err) Pooled (1985-1997) Coef. (R.Std.Err) Pooled w. erv interaction (1985-1997) Coef. (R.Std.Err) Pooled w. colonial interaction (1985-1997) Coef. (R.Std.Err) Pooled w. border interaction (1985-1997) Coef. (R.Std.Err) gdp 0.59 *** 0.60 *** 0.59 *** 0.59 *** 0.59 *** 0.61 *** 0.59 *** 0.59 *** (0.04) (0.04) (0.04) (0.04) (0.04) (0.04) (0.04) (0.04) gdppc 0 .1 2 *** q 2 2 *** 0 1 2 *** 0 22 *** 0 .1 2 *** 0 .1 2 *** 0 .1 2 *** 0 .1 2 *** (0.05) (0.05) (0.05) (0.05) (0.05) (0.05) (0.05) (0.05) distance - 1 .6 8 *** -1.69 *** - 1 .6 8 *** - 1 .6 8 *** -1.69 *** -1.69 *** -1.69 *** -1.69 *** (0.08) (0.08) (0.08) (0.08) (0.08) (0.08) (0.08) (0.08) areas 0 3 2 *** 0.30 *** 0 32 *** 0.31 *** 0 32 *** 0.29 *** 0.31 *** Q *** (0.06) (0.06) (0.06) (0.06) (0.06) (0.06) (0.06) (0.06) 11 -1.70 *** “1 72 *** -1 70 *** -1 70 *** -1.70 *** *1 72 *** -1.70 *** -1 71 *** (0.13) (0.13) (0.13) (0.13) (0.13) (0.13) (0.13) (0.13) border -0.16 -0.17 -0.16 0.07 -0.16 -0.17 -0.16 0.09 (0.23) (0.23) (0.23) (0.53) (0.23) (0.23) (0.23) (0.53) border_'85 (D) (D) border_'90 -0.35 -0.34 (0.64) (0.65) border_'95 -0.35 -0.36 (0.60) (0.60) border_'97 -0.17 -0.18 (0.62) (0.62) language 0.34 ** 0.34 ** 0.34 ** 0.34 ** 0.33 ** 0.33 ** 0.33 ** 0.33 ** (0.15) (0.15) (0.15) (0.15) (0.15) (0.15) (0.15) (0.15) o o Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 3.6. Regression results for estimating the gravity model for World energy trade using OLS (continued). Null values assumed genuinely missing Pooled (1985-1997) Coef. (R.Std.Err) Pooled w. erv interaction (1985-1997) Coef. (R.Std.Err) Pooled w. colonial interaction (1985-1997) Coef. (R.Std.Err) Pooled w. border interaction (1985-1997) Coef. CR.Std.Err) Pooled (1985-1997) Coef. (R.Std.Err) Pooled w. erv interaction (1985-1997) Coef. (R.Std.Err) Pooled w. colonial interaction (1985-1997) Coef. (R.Std.Err) Pooled w. border interaction (1985-1997) Coef. (R.Std.Err) regional 0.83 *** 0.83 *** 0 83 *** 0.83 *** 0.83 *** 0.83 *** 0.83 *** 0.83 *** (0.26) (0.26) (0.26) (0.26) (0.26) (0.26) (0.26) (0.26) nation (D) (D) (D) (D) (D) (D) (P) (D) colonizer I i9 *** 1 .2 2 *** I ig *** I Y2 *** 1.16 *** I 12 *** 1 .1 2 *** (0.27) (0.26) (0.27) (0.27) (0.27) (0.27) (0.27) (0.27) colonial 0.85 * 0.84 * 0.95 0.85 * 0.87 * 0 .8 6 * 0.97 0.87 * (0.48) (0.48) (0.75) (0.48) (0.48) (0.48) (0.74) (0.48) colonial_i -0.06 -0.06 (0.44) (0.44) erv 1.84 ** 1.84 ** 1 .8 8 *** 1.83 ** 1.83 ** I gy *** (0.72) (0.72) (0.72) (0.72) (0.72) (0.72) erv_'85 11.76 *** 11.85 *** (2.24) (2.24) erv_’90 0.70 0.61 (1.11) (1.11) erv_’ 95 1.30 1.33 (0.90) (0.90) ervJ97 3 9 9 *** 3 9 9 *** (1.28) (1.28) cu 2.16 *** 2.24 *** 2.16 *** 2 *** 2 23 *** 2.31 *** 2 23 *** 2 .2 1 *** (0.67) (0 .6 6 ) (0.67) (0 .6 6 ) (0.67) (0 .6 6 ) (0.67) (0 .6 6 ) Some null values assumed genuinely missing , while others are o Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 3.6. Regression results for estimating the gravity model for World energy Null values assumed genuinely missing trade using OLS (continued). Some null values assumed genuinely missing , while others are Pooled (1985-1997) Coef. (R.Std.Err) Pooled w. erv interaction (1985-1997) Coef. (R.Std.Err) Pooled w. colonial interaction (1985-1997) Coef. (R.Std.Err) Pooled w. border interaction (1985-1997) Coef. (R.Std.Err) Pooled (1985-1997) Coef. (R.Std.Err) Pooled w. erv interaction (1985-1997) Coef. (R.Std.Err) Pooled w. colonial interaction (1985-1997) Coef. (R.Std.Err) Pooled w. border interaction (1985-1997) Coef. (R.Std.Err) gov_ds 0 .1 0 *** 0.08 ** 0 .1 0 *** 0 .1 0 *** 0 .1 0 *** 0.08 ** 0 .1 0 *** 0 1 0 *** (0.03) (0.04) (0.03) (0.03) (0.03) (0.04) (0.03) (0.03) cpy -0.08 -0.56 -0.08 -0 .1 0 -0.08 -0.56 -0.08 -0 .1 0 (0.69) (0.69) (0.69) (0.69) (0.69) (0.69) (0.69) (0.69) cpe 0.33 ** 0.45 *** 0.33 ** 0.33 ** 0.33 ** 0.45 *** 0.33 ** 0.33 ** (0.15) (0.15) (0.15) (0.15) (0.15) (0.15) (0.15) (0.15) gini 0 .0 2 *** 0.03 *** 0 .0 2 *** 0 .0 2 *** 0 .0 2 *** 0.03 *** 0 .0 2 *** 0 .0 2 *** (0 .0 0 ) (0 .0 0 ) (0 .0 0 ) (0 .0 0 ) (0 .0 0 ) (0 .0 0 ) (0 .0 0 ) (0 .0 0 ) m2 gdp 0.000004 0.0000004 0.000004 0.000004 0.000005 0 .0 0 0 0 0 1 0.000005 0.000005 (0.00003) (0.00003) (0.00003) (0.00003) (0.00003) (0.00003) (0.00003) (0.00003) current 0.28 *** 0 .2 2 ** 0 2 9 *** 0.28 *** 0.28 *** 0 .2 1 ** 0.28 *** 0.28 *** (0 . 10) (0 . 10) (0 .1 0) (0 .1 0) (0 .1 0) (0 .1 0) (0 . 10) (0 . 10) capital -0.47 *** -0.50 *** -0.47 *** -0.48 *** -0.48 *** -0.51 *** -0.48 *** -0.48 *** (0 .1 0 ) (0 .1 1) (0 .1 0) (0 .1 0) (0 . 1 0) (0 .1 1) (0 .1 0 ) (0 . 1 0) _cons -9.39 *** _9 9 9 *** -9.39 *** -9.40 *** -9 37 *** -9 97 *** -9.38 *** -9 39 *** (1.42) (1.45) (1-42) (1.43) (1-42) (1.45) (1.42) (1.43) No. of Obs 2783 2783 2783 2783 2788 2788 2788 2788 R-squared 0.33 0.33 0.33 0.33 0.33 0.33 0.33 0.33 RMSE 2.87 2 .8 6 2.87 2 .8 8 2 .8 8 2.87 2 .8 8 2 .8 8 Notes: 1) D stands for dropped. 2) Standard errors are beneath the coefficients. 3) ***, **, * denotes significance at 1, 5, and 10% respectively. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 3.7. Regression results for estimating the gravity model for World other trade using OLS. Null values assumed genuinely missing Pooled w. Pooled w. Pooled w. Pooled w. Pooled w. colonial border Pooled w. colonial border Pooled erv interaction interaction interaction Pooled erv interaction interaction interaction (1985-1997) (1985-1997) (1985-1997) (1985-1997) (1985-1997) (1985-1997) (1985-1997) (1985-1997) Coef. Coef. Coef. Coef. Coef. Coef. Coef. Coef. CR.Std.Err) (R.Std.Err) (RStd.Err) (RStd.Err) (R.Std.Err) (R.Std.Err) (R.Std.Err) (RStd.Err) gdp 1.03 *** 1 0 4 *** 1 03 *** 1 03 *** 1.07 *** 1.08 *** 2 0 7 *** j Q7 *** (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) gdppc -0.003 -0 .0 1 -0 .0 0 2 -0.003 -0 .0 1 -0 .0 1 -0 .0 1 -0 .0 1 (0 .0 2 ) (0 .0 2 ) (0 .0 2 ) (0 .0 2 ) (0 .0 2 ) (0 .0 2 ) (0 .0 2 ) (0 .0 2 ) distance - 1 .2 1 *** -1 23 *** - 1 .2 1 *** - 1 .2 1 *** -1.29 *** -1.30 *** -1.29 *** -1.29 *** (0.03) (0.03) (0.03) (0.03) (0.03) (0.03) (0.03) (0.03) areas -0.18 *** -0.18 *** -0.18 *** -0.18 *** -0 .2 0 *** - 0 2 1 *** -0 .2 0 *** -0 .2 0 *** (0 .0 2 ) (0 .0 2 ) (0 .0 2 ) (0 .0 2 ) (0 .0 2 ) (0 .0 2 ) (0 .0 2 ) (0 .0 2 ) 11 -0 19 *** -0.17 *** -0.18 *** -0.18 *** -0.19 *** -0 18 *** -0.19 *** -0.19 *** (0.04) (0.04) (0.04) (0.04) (0.05) (0.05) (0.05) (0.05) border Q *** 0 4Q *** 0 4 2 *** 0.38 0 41 *** 0 40 0 41 *** 0.76 *** (0.13) (0.13) (0.13) (0.31) (0.14) (0.14) (0.14) (0.26) border_'85 (D) -0.33 (0.41) border_'90 0.33 (0.39) (D) border_'95 -0 .0 2 (0.38) -0.41 (0.35) border_'97 -0.16 (0.37) -0.61 * (0.33) language 0.56 *** 0.52 *** 0.56 *** 0.56 *** 0.54 *** 0.51 *** 0.54 *** 0.54 *** (0.05) (0.05) (0.05) (0.05) (0.06) (0.06) (0.06) (0.06) Some null values assumed genuinely missing , while others are o U ) Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 3.7. Regression results for estimating the gravity model for World other trade using OLS (continued). Null values assumed genuinely missing Pooled (1985-1997) Coef. (R.Std.Err) Pooled w. erv interaction (1985-1997) Coef. (R.Std.Err) Pooled w. colonial interaction (1985-1997) Coef. (R.Std.Err) Pooled w. border interaction (1985-1997) Coef. (R.Std.Err) Pooled (1985-1997) Coef. (R.Std.Err) Pooled w. erv interaction (1985-1997) Coef. (RStd.Err) Pooled w. colonial interaction (1985-1997) Coef. (R.Std.Err) Pooled w. border interaction (1985-1997) Coef. (R.Std.Err) regional 1.48 *** 1.44 *** 1 48 *** 1.48 *** 1.43 *** 1.39 *** 1.43 *** 1.42 *** (0.13) (0.13) (0.13) (0.13) (0.13) (0.13) (0.13) (0.13) nation (D) (D) (D) (D) (D) (D) (D) (D) colonizer 0.62 *** 0.64 *** 0.62 *** 0.62 *** 0.59 *** 0.60 *** 0.59 *** 0.59 *** (0.08) (0.08) (0.08) (0.08) (0.08) (0.08) (0.08) (0.08) colonial 0.78 *** 0.74 ** 1 2 2 *** q 7 g *** 0.70 ** 0.67 ** 1.17 ** 0.70 ** (0.30) (0.31) (0.44) (0.30) (0.33) (0.34) (0.49) (0.33) colonial_i -0.29 -0.31 (0.25) (0.27) erv -0.35 -0.35 -0.35 -0 .6 8 *** -0 .6 8 *** -0.69 *** (0.24) (0.24) (0.24) (0.25) (0.25) (0.25) erv_'85 1.67 *** 1.07 * (0.62) (0.64) erv_'90 1 0 9 *** 0.37 (0.31) (0.35) erv_'95 -1 31 *** -1.42 *** (0.31) (0.31) erv_'97 -2.56 *** -2.60 *** (0.49) (0.47) cu 1.62 *** 1.64 *** 1.63 *** 1 2 0 *** 1 .2 0 *** 1 2 0 *** (0.40) (0.40) (0.40) (0.40) (0.35) (0.35) (0.35) (0.35) Some null values assumed genuinely missing , while others are assumed zeros when they were zeros in all the other years o Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 3.7. Regression results for estimating the gravity model for World other trade using OLS (continued). Null values assumed genuinely missing Pooled (1985-1997) Coef. (RStd.Err) Pooled w. erv interaction (1985-1997) Coef. (R.Std.Err) Pooled w. colonial interaction (1985-1997) Coef. (R.Std.Err) Pooled w. border interaction (1985-1997) Coef. (R.Std.Err) Pooled (1985-1997) Coef. (RStd.Err) Pooled w. erv interaction (1985-1997) Coef. (R.Std.Err) Pooled w. colonial interaction (1985-1997) Coef. (R.Std.Err) Pooled w. border interaction (1985-1997) Coef. (RStd.Err) gov_ds 0.09 *** 0 .1 2 *** 0.09 *** q Q9 *** o n *** 0.13 *** O.ii *** q i i *** (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) cpy 0.26 0.19 0.26 0.27 0.31 * 0.24 0.31 * 0.32 * (0.19) (0 .2 0 ) (0.19) (0.19) (0.19) (0.19) (0.19) (0.19) cpe -0.25 *** -0.19 *** -0.25 *** -0.25 *** -0 29 *** -0.24 *** -0 29 *** -0.29 *** (0.05) (0.05) (0.05) (0.05) (0.05) (0.05) (0.05) (0.05) gini -0 .0 0 1 0 .0 0 2 -0 .0 0 1 -0 .0 0 1 -0 .0 0 2 0 .0 0 1 -0 .0 0 2 -0 .0 0 2 (0 .0 0 ) (0 .0 0 ) (0 .0 0 ) (0 .0 0 ) (0 .0 0 ) (0 .0 0 ) (0 .0 0 ) (0 .0 0 ) m2 gdp 0 .0 0 0 2 *** 0 .0 0 0 2 *** 0 .0 0 0 2 *** 0 .0 0 0 2 *** 0 .0 0 0 2 *** 0 .0 0 0 2 *** 0 .0 0 0 2 *** 0 .0 0 0 2 *** (0 .0 0 0 0 1 ) (0 .0 0 0 0 1 ) (0 .0 0 0 0 1 ) (0 .0 0 0 0 1 ) (0 .0 0 0 0 1 ) (0 .0 0 0 0 1 ) (0 .0 0 0 0 1 ) (0 .0 0 0 0 1 ) current 0.04 -0 .0 2 0.04 0.04 0.03 -0 .0 1 0.03 0.03 (0.03) (0.03) (0.03) (0.03) (0.03) (0.03) (0.03) (0.03) capital -0.41 *** -0 37 *** -0.41 *** -0 41 *** -0 41 *** -0 38 *** -0 41 *** -0 41 *** (0.04) (0.04) (0.04) (0.04) (0.04) (0.04) (0.04) (0.04) _cons -20.93 *** -2 1 .6 8 *** -20.94 *** -20.95 *** -21.84 *** -22.43 *** -21.85 *** -21.87 *** (0.50) (0.51) (0.50) (0.51) (0.53) (0.54) (0.53) (0.53) No. of Obs 7959 7959 7959 7959 8611 8611 8611 8611 R-squared 0.73 0.72 0.72 0.72 0.71 0.71 0.71 0.71 RMSE 1 .6 6 1.67 1.67 1.67 1.79 1.79 1.79 1.79 Some null values assumed genuinely missing , while others are Notes: 1) D stands for dropped. 2) Standard errors are beneath the coefficients. 3) ***, **, * denotes significance at 1,5, and 10% respectively. o U \ 106 The effect of areas on energy trade is positive and highly significant. Further investigations showed that correlation of areas with other variables had nothing to do with this finding. It might be the case that areas captured the effect of how big a country is and not the effect of transportation costs on the basis that different and may be more energy resources might be found in a bigger country as opposed to a country with a small area. Yet, of course, countries small in size like for instance Kuwait have more petroleum than Sudan, which is a bigger country. However, energy here does not include petroleum only, it includes natural gas, coal, other oils, mineral fuels, etc, a number of which might be found in one country the bigger its size. With respect to border and M2GDP, their coefficients are insignificant in Table 3.6. This is intuitive because neighborhood/financial development is an irrelevant variable as far as energy trade goes. It all really depends on where the energy resources are and thus border’s/M2GDP’s negligible effect on energy trade flows is quite expected. The effect of ERV on energy trade flows is positive and significant. The interaction terms reinforces this positive effect although they themselves vary across the years. This might support the argument that trade can be considered as an option held by firms.4 7 This might also indicate some reversed causality. 47 This might be specifically true for energy trade, where for instance oil exporting countries knowing how vital their resources are might postpone making export deals when increases in exchange rate volatility, that are in their favor, are expected or rapidly make their export deals when decreases in exchange rate volatility, that are not in their favor, are expected. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 107 Surprisingly the CPE variable has positive and significant effect on energy trade flows. This could be due to the fact that Russia orchestrated a lot of this with pipelines of natural gas, etc. However, further investigations revealed that CPE’s correlation with the Gini variable is the cause of such positive values. When the Gini variable was dropped, all CPE coefficients became negative but were insignificant suggesting that CPE has a negligible effect on energy trade flows. In Table 3.6, GDPPC has positive coefficients throughout and are all highly significant, indicating that countries with higher GDPPC trade more with each other in energy. In Table 3.7, GDPPC is insignificant.4 8 The impact of CPY on other trade flows in Table 3.7 is shown to be positive but not always significant. A similar finding was discussed in Table 3.1 before. Its effect on energy trade flows is however insignificant indicating that CPY has a negligible effect on energy trade flows. This is not so hard to grasp because, as explained earlier, energy trade depends to a great extent on which countries have the energy resources and not on how countries with common centrally planned economies will behave towards each other. The Gini variable has insignificant coefficients in Table 3.7 indicating negligible effect of income inequality on other bilateral trade flows. On the other hand, the effect of the Gini variable on energy trade flows is positive and highly significant. Further investigations reinforced this finding. This perhaps indicates that 48 This supports the argument discussed earlier that GDP probably captured all the increase in trade due to economic size thus making the effect of GDPPC on other trade negligible. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 108 countries with energy resources might inherently have high income inequalities and that they trade more with each other - intra-industry trade - that is. Current account restrictions show mostly positive, but insignificant, effects on other trade flows. However, further investigations showed that after dropping capital account restrictions, current account restrictions became negative and highly significant, indicating that its correlation with capital account restrictions is the cause for the surprising positive values. This consequently means that the effect on other trade flows of current account restrictions is negative and significant. The impact of current account restrictions on energy trade flows is positive and mostly highly significant throughout Table 3.6. Further investigations showed that correlations with capital account restrictions did not cause any changes in the signs of the coefficients of current account restrictions, but showed changes in their significance, where on average current account restrictions had coefficients that were not significant, indicating that current account restrictions have negligible effects on energy trade. 3.4.2 Summary of Results This part is intended to summarize the findings in this chapter. As clear from Table 3.8, some variables have been persistently robust to the different assumptions employed (missing and zero null values), to the different estimation techniques used (OLS and Tobit), and to the different specifications modeled (total trade flows, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 109 Table 3.8. A summary for World trade regression results Variable/ Table no. 3.1 3.2 3.3 3.5 3.7 Result gdp +ve, sig +ve, sig +ve, sig R (+ve, sig) gdppc +ve, sig insig insig S distance -ve, sig -ve, sig -ve, sig R (-ve, sig) areas -ve, sig -ve, sig -ve, sig R (-ve, sig) 11 -ve, sig -ve, sig -ve, sig R (-ve, sig) border +ve, sig insig +ve, sig RR (+ve, sig) language +ve, sig +ve, sig +ve, sig R (+ve, sig) regional +ve, sig +ve, sig insig +ve, sig RR (+ve, sig) colonizer +ve, sig +ve, sig +ve, sig R (+ve, sig) colonial +ve, sig -ve, sig +ve, sig RR (+ve, sig) erv -ve, sig -ve, sig -ve, sig R (-ve, sig) cu -t-ve, sig +ve, sig +ve, sig R (+ve, sig) gov_ds +ve, sig +ve, sig +ve, sig R (+ve, sig) cpy insig -ve, sig insig RR (insig) cpe -ve, sig insig -ve, sig RR (-ve, sig) gini insig -ve, sig insig RR (insig) m2 gdp +ve, sig -ve, sig +ve, sig RR (+ve, sig) current -ve, sig insig -ve, sig RR (-ve, sig) capital -ve, sig -ve, sig -ve, sig R (-ve, sig) Notes: 1) Blank cells in columns 3.3 & 3.5 indicate no significant changes due to using different estimation technique or including gov. components respectively. 2) VA, PS, GE, RQ, RL &CC all +ve and highly significant, with RL highest magnitude followed by CC& PS. individual governance components, and other trade flows49). These variables are: GDP, language, colonizer, CU, and Gov_ds, all of which showed mostly positive effects on trade flows and were mostly significant at the 1% level. Other parameters that have also been persistently robust include distance, areas, LL, ERV, and capital account restrictions, where these showed mostly negative effects on trade flows and were mostly significant at the 1% level. 49 Lagged dependent variable and Energy trade flows are not included because they are very special cases. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 110 There are eight variables that in general have been relatively robust. These are border, regional, colonial, M2GDP (showing mostly positive and significant effects), CPY, Gini (showing mostly insignificant effects), CPE, and current account restrictions (showing mostly negative and significant effects), which showed changes due to only one sensitivity analysis (assumption, technique, or specification). The only variable that displayed the highest sensitivities in terms of showing different results in two or more of the three sensitivity analyses - different assumptions, techniques, and specifications employed - was GDPPC. Having investigated the effect of the model discussed earlier on World trade flows, let’s now explore its effect on intra-Arab and Arab-World trade flows in the following chapter. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. I l l 4. GOVERNANCE AND ARAB TRADE The main purpose of this chapter is to explore the effect of domestic governance on intra-Arab trade flows. To be able to investigate whether intra-Arab trade has specific features that characterize trade only between Arab countries, the results from intra- Arab regressions are compared to results from Arab-World regressions also presented here. This chapter is composed of two main sections. Section one will present a comprehensive background on Arab economic integration, for which intra-Arab trade is a measure, while the second section will display the results from intra-Arab and Arab-World regressions and offer interpretations. 4.1 Arab Regionalism / Arab Economic Integration Regionalism is broadly defined as preferential trade agreements among a number of nations that geographically lie within the same region. It is discriminatory because under its umbrella preferences are extended only to partners. At the same time, regionalism represents a step towards freer trade among partners. Regional economic integration has five main stages. The first stage is the Preferential Trading Arrangement (PTA). The second stage is the Free Trade Area (FTA). The third stage is Customs Union. The fourth stage is a common market, and the final stage is Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 112 economic unity50. Whatever the form or stage of the regional arrangement/integration, increased intra-regional trade ranks high among the priorities and is the yardstick against which the success or failure of regionalism is measured (Abouyoub 1992; and De Melo and Panagariya 1993). 4.1.1 History of Failures - When, What, and Why? The Arab countries share a great deal in the way of common histories and cultural characteristics51. However, they also have several differences, and their diversity in the economic sphere is well pronounced. They are diverse in terms of size, natural resource endowments, standard of living, and development trajectories. Some countries are primarily agricultural such as Mauritania and Sudan; others are primarily energy producers such as GCC members, and others have an emerging industrial base such as Egypt and Morocco (Al-Atrash and Yousef 2000). Attempts at Arab economic cooperation date from as early as 1950 when the Arab Joint Defense and Economic Cooperation Agreement was signed under the auspices of the Arab League. The promotion of trade has long been considered the basis for cooperation and economic integration, and thus joint Arab efforts have 50 In PTA, customs barriers such as tariffs and other barriers that might impede the flow of trade are reduced for the partners. An FT A, which is also sometimes referred to as Free Trade Agreement/Arrangement involves zero tariffs among partners and positive, but not necessarily identical, tariffs with the rest of the world. An FTA also involves the elimination of nontariff barriers. In a Customs Union, a common external tariff is imposed by partner countries vis-a-vis the rest of the world. A common market is a unified market that allows the free movement of labor, firms, services and capital. Economic unity is the highest degree of regional integration, and in which partners place their economic sovereignty in the hand of a unified economic authority that supervises the adoption of unified policies regarding monetary, tariff, tax and other economic aspects. 51 Arab Countries are defined based on membership in the Arab League. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 113 attempted to liberalize intra-Arab trade through various channels. Such channels include among other things multilateral and bilateral trade agreements. A multilateral agreement was reached on the regulation of transit trade in 1953, which was seen as a preparatory phase to establishing a common market later on (Zarrouk 1992). In practice this agreement failed due to three main factors. First, overland links from Egypt - the most populous Arab state - were disrupted by the trade boycott of Israel. Second, trade ties remained strong with Britain and France - the former colonial powers - which was clear in Egypt’s and Jordan’s currencies being tied to Sterling, while those of Syria, Lebanon, and the Maghreb countries were part of the French Franc Zone. And third, all the economies involved depended on primary agricultural produce and thus had little to offer each other in trade (Wilson 1994). Egyptian president Gamal Abd al-Nasser stimulated the most ambitious attempt at Arab economic cooperation in 1958 with the founding of the United Arab Republic (UAR) between Egypt and Syria. However, it did not succeed owing to the fact that the two countries were geographically separated by Israel and they had different colonial experiences and economic outlooks. Egypt was the stronger partner in the Union, and Syria resented having its economic plans dictated by Nasser (Wilson 1994). In 1964, the Arab Common Market (ACM) agreement was signed by Egypt, Iraq, Jordan, and Syria, and joined later by Libya (1977), Mauritania (1980), and the People’s Democratic Republic of Yemen (1981). The aim of the agreement was to create a common market in stages to provide an advanced framework for intra-Arab Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 114 trade promotion. The first phase was to establish an FTA, and the principal measure was the elimination of tariffs and non-tariff barriers on agricultural products as well as manufactures according to a specific timetable. Unfortunately, the decision to permit exceptions allowed member countries to take advantage of this loophole, which undermined the agreement (Zarrouk 1992). Other factors that weakened the ACM include: the suspension of the membership of Egypt - the Arab world’s largest economy - when it broke ranks to sign a peace treaty with Israel in 1979, and the closing of borders between other member countries from time to time due to political frictions (Miniesy, Nugent, and Yousef 2004). The failure of economic integration at the Pan-Arab level motivated a search for alternative strategies. One approach was to seek integration through the establishment of joint projects. This approach was anointed by many regional organizations as a more feasible way of achieving integration from the “bottom up”, where firms that could and would trade at least within the region rather than directly through trade agreements are created. Thus the 1970s witnessed a plethora of such projects in different sectors ranging from agriculture and cattle farming to shipping and mining. However, it soon became obvious that whatever the benefits of such projects, they could not by themselves constitute a valid approach to economic integration, especially since many of the joint ventures investments have been made by the public sector (El-Naggar 1992). In 1981 the Agreement for the Facilitation and Promotion of Trade was signed by member states of the Arab League, and entered into force in 1982. The aim of this Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 115 agreement was to fully eliminate tariffs, nontariff barriers, and taxes of similar effect for manufactured and semi manufactured goods. The problem with this agreement is that it allowed negotiators to liberalize some manufactured products but not others. Moreover, the negotiations did not lay out a time schedule for the stages to complete liberalization. As a result, only a very few Arab manufactured products remained candidates for tariff exemptions. Furthermore, because tariff exemptions meant tariff revenue losses, and because the agreement did not set the grounds for a compensation scheme to balance such losses, many signatories were discouraged from speeding up the trade liberalization process (Zarrouk 1992). In addition to the specific reasons for the failure of many attempts at regional integration, regionalism was a by-product of Arab nationalism. It was driven by political, not economic motives, and thus the numerous attempts at Arab regionalism produced very little true integration (Shafik 1998). Galal (2000) made the same argument as Shafik did, where he asserted that regional integration becomes a reality when the parties involved have sufficient economic and political incentives. In the absence of such incentives regional integration does not occur even if politicians declare their intentions to the contrary. Thus the lack of economic incentives limited the previous integration attempts in the region despite the talks about Arab nationalism and the repeated attempts at an Arab common market. The failure of many attempts at Arab regionalism led some Arab states to devote their efforts to pursuing smaller-scale regional groupings. At the sub-regional level, the most important development of the 1980s was the creation of the Gulf Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 116 Cooperation Council (GCC) in 1981 among Bahrain, Kuwait, Oman, Qatar, Saudi Arabia, and United Arab Emirates. These countries agreed to eliminate tariffs on products produced by firms that were majority-owned by nationals of the country and for which domestic value added constitutes at least 40% of the cost of the finished product. The creation of the Arab Maghreb Union (AMU) in 1989 among Algeria, Libya, Mauritania, Morocco, and Tunisia is another important development at the sub regional level (Miniesy, Nugent, and Yousef 2004). To concentrate only on trade in goods is to overlook most of the economic flows within the region. Other flows include trade in labor services as well as capital flows. Labor flows constituted a more important form of trade within the Arab countries than goods’ flows. This trade however has been extremely reduced by the Gulf war and its attendant political disruptions. Concerning capital flows; they were in terms of FDI and aid. Private capital flows took place mainly among the GCC countries and in the shape of FDI to Egypt. Aid has flowed also from the Gulf States to other Arab countries, namely Egypt, Jordan, Lebanon, Syria, and Yemen. The aid flows were extremely large in 1980 and declined with the decline of the price of oil and the political disruptions in the region. After the Gulf war, the aid declined even further. However, the oil-rich countries are more aware than they were before of the need for political stability in the region. Trade with and financial investment in other Arab countries are more likely than aid to lead to constructive long-term relations (Fisher 1993). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 117 4.1.2 Any Prospects for Intra- Arab Trade / Arab FTA? El-Naggar (1992) argued that the prospects for intra-Arab trade may be significantly better than had previously been the case for three main reasons: First, most of the highly protected Arab countries are engaged in trade liberalization in the context of adjustment programs signed with the IMF and the World Bank. Second, oil-rich countries are pursing a thorough going program of diversification, which is likely to open up significant opportunities for intra-trade. And third, the establishment of the Arab Trade Financing Program (ATFP) that came into effect in 1989 under the direction of the Arab Monetary Fund. The ATFP was established following the realization that financing intra-Arab trade is a crucial factor in encouraging Arab exporters and importers to trade with each other on a continuous and regular basis. It thus aims at enhancing the competitiveness of Arab exporters through the provision of badly needed finance and credit guarantees. The ATFP has also taken the initiative in preparing for the establishment of a trade information network for Arab users. This system will help Arab exporters to better reach Arab markets and to gain new comparative advantage, through collecting data and information on trade laws and regulations, preferential arrangements, and available export opportunities in the Arab countries (Zarrouk 1992). Zarrouk (1992) made an attempt to measure potential static and dynamic gains from across-the-board liberalization of tariff and nontariff barriers between Arab countries. For static gains, he found out that an agreement that leads to full (100%) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 118 concessional tariff cuts on intra-Arab imports on manufactures could expand intra- Arab trade by $400-$500 million in the first year after the agreement. Such increases represent 1% of non-oil total Arab exports or 9% of non-oil intra-Arab exports. Coming to dynamic gains, although very difficult to measure, Zarrouk attempted to provide ‘hypothetical’ dynamic gains based on highly simplified assumptions and he found out that under such assumptions, intra-Arab exports of manufactures would grow at an annual average rate of 3.9% of non-oil total intra-Arab exports during the first five years following the liberalization, and at a rate of 7.5% during the second five years. Zarrouk concluded by strongly encouraging Arab countries to take effective measures for a multilateral trade liberalization scheme covering all products of Arab origin and allowing for trade barriers to be phased out in stages. Fischer (1993) argued that there is no ‘realistic’ prospect of Middle East-wide regional integration on either the NAFTA or EC models in the near future. He added that although FTA among a subset of countries - Arab countries - is possible, the resulting benefits would at most be small, and thus negotiating PTAs would not make sense. Fischer attributed the reasons for such poor prospects to the regional political conflicts and hostilities, emphasizing that the share of military spending in GNP in the region is almost 10 percentage points above the world average, which in turn eats up resources that could be used for investment in human and physical capital. Fischer also argued that the economic attributes of the countries in the region (high debt, and other structural problems) limit the scope for trade-creating effects. He suggested instead cooperation in functional areas, such as water management, and agreements on Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 119 the creation of regional infrastructure, which are more promising though perhaps difficult to achieve. However, Fischer believed that there are large ‘potential’ gains from increased integration and cooperation but that they are unlikely to be reaped until the region becomes more peaceful and less tom by conflicts, and countries pursue vigorous stabilization and adjustment goals. In discussing Fischer’s (1993) work, Baldwin agreed with Fischer’s argument concerning the absence of realistic prospect of Middle East - wide regional integration. Baldwin believed that it is a stylized fact that South-South PTAs do not work, because they get signed “with a flourish of brotherhood and unity, but almost as soon as the ink of the headlines dries, the deals begin to be ignored”. Baldwin attributed this to two reasons: First, policy-makers perceive their country’s own liberalization as a cost that must be incurred in order to obtain greater access for domestic firms to export markets. And second, Southern exporters are not very interested in selling to other Southern markets; the reasons for this are embedded in the mercantilist prism. 4.1.3 Renewed Interest in Arab Economic Integration There has been a renewed interest in regional integration between the Arab countries. The Greater Arab Free Trade Agreement (GAFTA), launched by the member states of the Arab league in 1998 is a recent example of such renewed interest. The Executive program for the establishment of the GAFTA aims to revive the 1981 Agreement for the Facilitation and Promotion of Trade among the member states of the Arab League. The objective of the GAFTA is the elimination of import duties and other trade Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 120 52 barriers on goods of Arab origin over a ten-year period. The 18 signatories agreeing to this agreement were required to lower bilateral tariff rates by 10% annually, which means that by 2008 intra-Arab imports, spanning from Morocco in the West to Oman in the East, should enter every country of the region without tariffs or tariff-like barriers (ERF 1998; and Zarrouk 2000).5 3 There are a number of reasons that have been given for the recent interest in creating a region-wide FTA. First, many of the countries in the region have implemented reforms to reduce the economic role of the public sector and shift away from import substitution, which caused external tariffs to be lower now than they were in the past suggesting that there should be less trade diversion in the newly created Free Trade Area (Miniesy, Nugent, and Yousef 2004; and Zarrouk 2000). Second, both the post-Uruguay Round tariff and nontariff reductions and the strengthening of the multilateral trading system have augmented the fears of Arab countries of being marginalized (Zarrouk 2000).5 4 Third, intra-regional trade has been shown to take the form of intra-industry trade, which in recent years has been seen in a more positive 52 All the 22 member states of the Arab League, except Algeria, Djibouti, the Comoros Islands and Mauritania have endorsed the agreement and have committed to the executive program. 53 If interested, Appendix D provides some important aspects of GAFTA. 5 4 Economists and policy-makers have long been divided on the wisdom of regional arrangements. There are those, as Thurow (1992) and Bhagwati (1993), who believe that regionalism would result in the world evolving towards trading blocs, which might turn inward and erect high barriers against non members, and where small countries would be left out of these blocs. On the other hand, there are others, like Summers (1991), who feel that the evolving blocs might speed up the progress towards global free trade since negotiations would then be better carried out among a few blocs than within the large membership of the WTO. Anyhow, the importance of the regionalism question stems from the observed proliferation of preferential regional agreements - over 100 PTAs have notified their existence to the WTO, and many policy-makers are still pursuing more PTAs - notwithstanding the multilateralists’ insistence that national and global welfare would be maximized by opening up to all trading partners rather than to a few (Galal and Hoekman 1997). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 121 light than before. Many analysts believe that greater integration among the Arab countries would bring important benefits, including greater efficiency, deepening of capital markets, fostering FDI, and greater product variety. Greater product variety would also enhance intra-industry trade by weakening monopoly power of local producers (Miniesy, Nugent, and Yousef 2004). Regionalism usually has static and dynamic effects. Concerning the static impact of GAFTA, Zarrouk (2000) argued that GAFTA is not likely to be dominated by trade diversion, because products which are of primary importance to developing countries are usually capital-intensive goods in which GAFTA members do not have a comparative advantage. Zarrouk in a previous study actually found that the majority of dynamic products in intra-Arab trade are not in sectors that are of primary importance in Arab countries’ exports to the EU. There are also potential dynamic effects from establishing a regional FTA. Many industries in the Arab countries cannot be economically viable because domestic sales are insufficient for economies of scale to be realized. Regional economic integration could bring about major economic gains through the realization of such economies of scale, which are otherwise unattainable because of the small size of the Arab countries’ national markets (Wilson 1994). In addition to the economies of scale that integration is expected to provide, Arab regional integration can also promote larger scope for specialization, resulting in more efficient allocation of regional resources and increased market competitiveness. A larger regional market can also stimulate investment within the region by investors Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 122 from inside as well as outside the region in the form of FDI (ERF 1998). Moreover, as Refaat (2000) argued, the Arab FTA and the other Arab bilateral agreements are important in their own right because they can reduce the “hub-and-spokes” effects that are likely to occur as a result of the EU-Med agreements in which the EU would be the “hub” and the individual Med countries the “spokes”. This is problematic because it would mean that FDI that is desired to revitalize the Med countries would paradoxically turn towards the EU itself. 4.1.4 Intra-Arab Trade 4.1.4.1 Recent Trends in Intra-Arab Trade According to Table 4.1, total exports by the Arab countries were about $130 billion in 1998. Over one-half of this went to industrial countries and another one-third went to Asia. Imports totaled about $170 billion, of which about two-thirds, were from industrial countries and another 15% from Asia. In the last three decades, the share of total trade - exports and imports of goods and nonfactor services - in GDP has averaged around 50%, making the region one of the most integrated regions into the global economy. However, the level of integration rose in the 1970s, falling in the early 1980s and recovering in the latter part of the decade where it has remained stable ever since (Nabli and De Kleine 2000). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 123 Table 4.1. Direction of Arab Trade, 1998 (In billions of U.S. dollars) Exports Imports Value Share Value Share Industrial countries 72.3 54.0 113.8 67.5 Developing countries 61.5 46.0 54.9 32.5 Africa 4.0 3.0 2 .8 1.7 Asia 39.0 29.1 26.4 15.7 Central and Eastern Europe 4.5 3.4 9.6 5.7 Arab countries 1 1 .0 8 .2 1 1 .8 7.0 Western Hemisphere 1.7 1.3 3.3 2 .0 Other 1.3 1 .0 1 .0 0 .6 Total 133.8 1 0 0 .0 168.7 1 0 0 .0 Source: IMF Direction of Trade Statistics, 1998 Yearbook. As shown in Table 4.2, intra-regional exports comprise some 8% of total exports in 1998, which compares unfavorably with the level of intra-trade of other regions. For instance, intra-regional trade as a share of total trade is nearly 50% higher in the Andean Pact countries than in the Arab countries, and 7 times higher in the countries belonging to the European Union. Moreover, the share of intra-regional trade has not grown in tandem with trends elsewhere and actually fell in the 1990s. For example, while intra-Arab trade as a share of total trade increased from 5% to 8% between 1970 and 1998, trade among the Andean Pact countries increased from 2% to 11%, trade among the Southern Cone countries increased from 11% to 25%, and trade among members of NAFTA increased from 36% to 51%. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 124 Table 4.2. Trends in Intra-Regional Trade, 1970-98 (As a share of total exports in the region) 1970 1975 1980 1985 1990 1995 1998 All Arab countries 5.2 4.9 4.5 7.8 9.4 6.7 8 .2 Arab countries, Iran and Turkey 6 .0 5.8 6 .1 8.7 10 .8 7.2 8.7 Andean Pact countries 1/ 1.7 3.6 3.5 3.1 4.0 11.3 11.4 Australia & New Zealand 6.1 6 .1 6.4 7.0 7.6 9.9 8 .6 Southern Core Countries 2/ 11.4 11.1 14.3 6.7 1 0 .6 2 1 .6 25.5 East Asian Economies 3 1 19.2 21.3 22.4 20.7 20.7 26.4 2 2 .2 NAFTA 4/ 36.0 34.6 33.6 43.9 41.4 46.2 51.0 European Union 59.5 57.7 60.8 59.2 65.9 62.4 56.8 Sources: IMF Direction of Statistics Yearbook, various editions; and Fund staff calculations. 1/ Colombia, Ecuador, Peru, and Venezuela. 2/ Argentina, Brazil, Chile, Paraguay, and Uruguay. 3/ China, Indonesia, Japan, Korea, Malaysia, the Philippines, and Thailand. Data exclude exports by Taiwan Province of China. 4/ Canada, Mexico, and United States. The Arab countries could be divided into four subgroups based largely on geographical location and production base, namely: the Maghreb countries (Algeria, Libya, Mauritania, Morocco, and Tunisia), the Mashreq countries (Egypt, Jordan, Lebanon, Sudan, and Syria), the Gulf Cooperation Council countries (GCC) (Bahrain, Kuwait, Oman, Qatar, Saudi Arabia, and United Arab Emirates), and other countries (Djibouti, Somalia, and Yemen). As shown in Table 4.3, about 60% of intra-regional Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4.3. Indicators of Intra-Arab Trade, 1998 Exports by: 1/ Selected Arab countries Maghreb countries GCC countries Mashreq Other countries countries (Intra-regional exports, in billions of U.S.. dollars) Exports to: 1/ Arab countries, Of which: 1 2 .0 1 .6 7.5 2 .6 0.3 Maghreb 2 .0 1 .0 0 .6 0.4 0 .0 GCC 6 .8 0 .1 5.3 1 .2 0 .2 Selected Mashreq 2 .6 0.5 1 .2 1 .0 0 .0 Other 0 .6 0 .0 0.4 0 .0 0 .1 (Intra-regional exports, as percent of exports to world) Arab countries, Of which: 8 .2 4.9 7.7 22.7 12.5 Maghreb 1.4 3.1 0 .6 3.3 0 .0 GCC 4.6 0.4 5.5 1 0 .2 7.5 Selected Mashreq 1 .8 1.4 1 .2 8 .6 0 .1 Other 0.4 0 .0 0.4 0 .6 4.9 (Intra-regional exports, as percent of exports to Arab countries) Arab countries, Of which: 100.0 100.0 100.0 100.0 1 0 0 .0 Maghreb 16.7 63.2 7.7 14.7 0 .1 GCC 56.6 7.6 71.4 44.9 59.9 Selected Mashreq 2 1 .8 29.1 15.6 37.7 0 .8 Other 4.9 0 .1 5.2 2.7 39.3 Source: IMF Direction of Trade Statistics, 1998 Yearbook. 1/ Country groupings are: Maghreb: Algeria, Libya, Mauritania, Morocco, Tunisia. GCC: Bahrain, Kuwait, Oman, Qatar, Saudi Arabia, United Arab Emirates. Selected Mashreq countries: Egypt, Jordan, Lebanon, Syria, Sudan. Other countries: Djibouti, Somalia, Yemen. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 126 exports — which amounted to $12 billion in 1998 - were to the GCC countries with another 25% to the Mashreq countries. More important, the proportion of trade within the four subgroups is significantly higher than overall intra-Arab trade: nearly two- third of the Maghreb countries exports to the Arab countries is with other Maghreb countries; three-fourths of the GCC exports to the Arab countries is with other GCC countries; and one-third of the Mashreq exports with the Arab countries is with other Mashreq countries. The fact that most of the intra-Arab trade is within the sub-regions suggests that trade impediments are lower within the subgroups than for the region as a whole and that differences in comparative advantage exist even within the subgroups. 4.1.4.2 Explanations for the Low Intensity of Regionalism / Low Trade Various explanations have been offered for the low level of intra-Arab trade. Below are the most important and commonly cited ones. Most Arab countries, with the exception of the GCC countries, have pursued industrial strategies based on import-substitution and a large public sector. This led, among other things, to high protection as well as nontransparent trade policies. The average tariff for the region as a whole is higher than that of any other region, except Africa55. Tariff escalation is also an important feature of tariff barriers in many Arab 55 The average tariff for Arab countries is estimated at about 17 percent. This compares to an average tariff of 20 percent for African countries, of 13 percent for Western Hemisphere countries, of 12 percent for Asia Pacific, of 10 percent for the Baltic, Russia, and other countries of the former Soviet Union (BRO), and of 9 percent for Europe (excluding BRO). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 127 countries56. Moreover, nontariff barriers are extensive in many countries in the region, which encompass import licensing and quotas, government monopoly over imports, monetary restrictions on imports, and paratariffs - additional fiscal charges on imports (El-Erian and Fisher 1996; El-Naggar 1992; and Zarrouk 1992 & 2000). In addition to trade barriers, the existence of differences in the overall economic strategies and policies between Arab countries tends to discourage intra-regional trade (Allum 1998). In particular, while some countries in the region pursue market-oriented policies and have established a strong record of economic adjustment and reform such as Jordan, Morocco, and Tunisia, other countries maintain a high degree of government involvement as in Libya and Syria. Furthermore, many Arab governments rely on import duties as an important source of revenue instead of introducing broad-based domestic consumption taxes. In addition to all the above, delays and costs involved in clearing customs, obtaining rebates in cases where trade taxes can be exempted when the final products are exported, and bureaucratic red tape to protect rents that accrue to domestic producer interest groups and government officials also played a role in low levels of intra-Arab trade (Zarrouk 2000). Political factors, including economic sanctions and bilateral disputes have also affected intra-regional trade (Fisher 1993; Limam and Abdalla 1999; Wilson 1994; and Zarrouk 2000). Since 1970, there have been numerous bilateral conflicts ranging in intensity from diplomatic crises to border disputes and open wars. This was most 56 Tariff escalation means that tariffs on finished products are far higher than tariff levels on raw materials. It is most pronounced in textiles, clothing, leather, and basic metal industries. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 128 clearly shown in the Maghreb region (Libya and Algeria vs. Morocco) in the 1970s and the Mashreq region in the 1980s (Syria vs. Lebanon) and mainly between Kuwait and Iraq in 1990. Prior to the UN sanctions in 1990, Saudi Arabia’s exports to Iraq exceeded $150 million. Since then, exports have been negligible. On a similar basis, political differences between Algeria and Morocco reduced trade between the two neighbors from $140 million in 1992 to less than $100 million in 1995-6. The lack of product complementarity has often been cited as an important factor impeding intra-Arab trade (Fisher 1993). The relative similarity of resource endowments among many countries in the region argue against intra-regional trade since the region’s comparative advantage is broadly in the same products, which make the countries competitors rather than collaborators in international trade. At the same time, the lack of a diversified export base - particularly in manufactures - limits the opportunities for trade based on product differentiation. Hence, intra-Arab trade fits in neither of the two main models of international trade: the Heckscher-Ohlin model, which bases trade on different factor endowments, and the intra-industry model, which bases trade on product differentiation. However, as noted above, most intra-Arab trade is within sub-regions where, presumably, the lack of product complementarity is greatest.5 7 57 Havrylyshyn (1997) calculated a “complementarity index” that showed that product complementarity in the region is broadly similar to that of other regional groupings like MERCOSUR and APEC, which suggests that there could be much greater potential for regional integration among the Arab countries than is being realized. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 129 Differences in per capita income have also been cited as a factor constraining intra-Arab trade (Fisher 1993; and Zarrouk 1992). Richer Arab countries prefer to import high-quality goods that are more likely to be produced by industrial nations. For instance, Egypt is an important producer of manufactured consumer goods within the region and although Saudi Arabia is an important importer of such goods, Saudi Arabia’s high per capita income induces it to consume higher quality manufactures from European countries instead of manufactures from Egypt. Per capita income per se is not a barrier to trade; otherwise trade between Mexico and the U.S. would be difficult to explain. It is the homogeneity of the export base among many Arab countries coupled with the disparity in income that have generally been cited as hindering intra-Arab trade. Finally, high trade costs, including transport and communications as well as administrative costs have also been cited as a factor constraining intra-Arab trade (Abouyoub 1992; Al-Naggar 1992; Zarrouk 1992; and El-Erian and Fisher 1996). The distance and difficult geographic terrain between some Arab countries make trade links difficult. For instance, the Maghreb countries are geographically closer to Europe than to other Arab countries making trade links easier with Europe. Even where geography is not working against regional trade, the lack of transportation networks, minimal paved roads and rail links across countries, and border closures, all have created disincentives for the movement of goods and services. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 130 4.1.4.3 Is Intra-Arab Trade Low ? Empirical Results from Old and New Data El-Naggar (1992), Zarrouk (1992), Fisher (1993), and El-Erian and Fisher (1996) have all noted that the level of intra-Arab and intra-MENA (Middle East and North Africa) trade is low, accounting for 5-10% of exports to the world - depending on both the definition of the region and on whether or not oil is included. Ekholm, Torstensson, and Torstensson (1995), discussed earlier, and Yeats (1996) who used the trade intensity index5 8 concluded that intra-regional trade in the Middle East was not “too low”. Havrylyshyn (1997) presenting the results of all studies mentioned above argued that the majority of the data support the conclusion that it is in fact too low. He argued that Ekholm, Torstensson, and Torstensson (1995) used a smaller control sample. As for Yeats (1996), Havrylyshyn argued that Yeats’s conclusion that trade intensities are much greater than 1 for the Middle East countries should not be taken alone and has to be set against the values for other regions. For instance, Latin America in 1990, when the unsuccessful regional arrangements were effectively nonfunctional, trade intensities for intra-regional trade of most countries were far higher than in the Middle Eastern group. Moreover the values for the European and East Asian regions were higher than those of the Middle East. Al-Atrash and Yousef 58 The trade intensity index normalizes for a country’s global involvement in trade, and so a propensity greater than 1 means that the proportions of world exports to that country is greater than its proportions of world imports. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 131 (2000) and Miniesy, Nugent, and Yousef (2004), as discussed in chapter 2, also concluded that intra-Arab trade is indeed low. 4.2 Gravity Model and Arab Trade 4.2.1 Intra-Arab Trade Results The objective here is not to investigate whether intra-Arab trade is too low, but rather to explore the effect of many variables, especially governance, on intra-Arab trade. This section presents and interprets the regression results from estimating the gravity equation given in section 3.1 for intra-Arab bilateral trade flows. The only two differences are that the Gini variable was excluded because it dramatically reduced the number of observations and only Arab trading pairs were included. Table 4.4 displays the regression results from estimating the gravity equation for intra-Arab total bilateral trade for: each individual year, the whole pooled sample (1985-2000) once where regional is as given in section 3.2 and once where dummy variables for Arab-regional trade agreements - GCC and AMU - are included, and the pooled sample not including year 2000, again once with regional of section 3.2 and once with GCC and AMU. These were carried out using OLS, where the null values in the dependent variable were assumed to be genuinely missing.5 9 59 Unlike in Chapter 3, Tobit will not be used since it is not expected to change any fundamental conclusions. As was shown in Table 3.3, the Tobit technique did not change the basic results. The only difference was in the magnitudes of the coefficients, which were larger than those estimated using OLS. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 4.4. Regression results from estimating the gravity model for intra-Arab total bilateral trade using OLS, where the null vlaues in the 1985 Coef. R.Std.Er) 1990 Coef. (R.Std.Er) 1995 Coef. (R.Std.Er) 1997 Coef. (R.Std.Er) 2 0 0 0 Coef. (R.Std.Er) Pooled (1985-2000) Coef. (R.Std.Er) Pooled (1985-2000) GCC & AMU Coef. (R.Std.Er) Pooled (1985-1997) Coef. (R.Std.Er) Pooled (1985-1997) GCC & AMU Coef. (R.Std.Er) 0 .8 6 * 1 33 *** 1.61 *** 1 01 *** 0.72 *** 1.12 *** 1 .1 2 *** 1 .2 0 *** 1.2 0 *** (0.43) (0.26) (0 .2 0) (0 .2 0) (0.16) (0 .10) (0 .11) (0 .12) (0.13) -1.41 ** -0.19 -0 .8 8 *** -0.55 -0.47 ** -0.54 *** -0.55 *** -0.60 *** -0.62 *** (0.58) (0.40) (0.33) (0.36) (0 .2 0) (0 .11) (0 .11) (0.13) (0.14) -1.42 ** -1.66 *** -1.62 *** -1 42 *** -1.64 *** -1.73 *** -1.73 *** -1 73 *** -1 72 *** (0.67) (0.41) (0.32) (0.30) (0.31) (0.17) (0 .2 0) (0 .2 0 ) (0.24) -0.04 -0.42 -0.76 *** -0.37 -0.16 -0.46 *** -0.54 *** -0.55 *** (0.41) (0.31) (0 .2 0) (0.26) (0.18) (0 .12) (0 .12) (0.15) (0.15) (D) (D) (D) (D) (D) (D) (D) (D) (D) 0.11 0.38 0.63 0.82 * -0.01 0.30 0.32 0.45 0.49 (1.04) (0.64) (0.52) (0.49) (0.52) (0.31) (0.32) (0.38) (0.40) (D) (D) (D) (D) (D) (D) (D) (D) (D) (D) (D) (D) (D) (D) (D) (D) 0 .0 2 (0.35) -0.16 (0.50) (D) (D) 0.08 (0.43) -0.34 (0.54) (D) (D) (D) (D) (D) (D) (D) (D) (D) g a p gdppc distance areas 1 1 border language regional gcc amu nation Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 4.4. Regression results from estimating the gravity model for intra-Arab total bilateral trade using OLS, where the null vlaues in the dependent variable are assumed genuinely missing (continued). 1985 Coef. (R.Std.Er) 1990 Coef. (R.Std.Er) 1995 Coef. (R.Std.Er) 1997 Coef. (R.Std.Er) 2 0 0 0 Coef. (R.Std.Er) Pooled (1985-2000) Coef. (R.Std.Er) Pooled (1985-2000) GCC & AMU Coef. (R.Std.Er) Pooled (1985-1997) Coef. (R.Std.Er) Pooled (1985-1997) GCC & AMU Coef. (R.Std.Er) colonizer 0.84 0 .6 8 0.37 0.26 -0 .1 2 0.17 0.19 0.23 0.28 (0 .88) (0.51) (0.48) (0.52) (0.39) (0.23) (0.26) (0.27) (0.31) colonial 0 .2 2 -0 .8 6 -1.05 -0.41 0.01 -0 .0 2 0.04 -0.01 (1.46) (0.80) (0.79) (0.87) (0.63) (0.57) (0.59) (0 .6 8) (0.70) erv -47.59 ** 10.13 -5.74 -7.06 * -7.82 * -2.25 -2.27 -2.06 -2 .1 2 (23.78) (6.13) (4.44) (3.75) (4.04) (1.54) (1.55) (1.70) (1.72) cu (D) (D) C D ) (D) C D ) (D) (D) (D) (D) gov_ds -0.61 0.05 0.30 0 .1 2 0.29 *** 0.09 ** 0.09 ** 0.05 0.05 (0.40) (0.18) (0 .2 2) (0.18) (0 .10) (0.04) (0.04) (0.05) (0.05) cpy (D) (D) C D ) (D) (D) (D) (D) (D) (D) cpe -1.79 * -1 32 *** 7 7 *** -0.87 -0.63 * -0.81 *** -0.81 *** -0.80 *** -0.81 *** (0.91) (0.45) (0.51) (0.53) (0.32) (0.19) (0 .2 0) (0.23) (0.24) m2 gdp 0.0003 0.00003 -0 .0 0 0 2 ** -0 .0 0 0 2 ** -0.0001 *** -0 .0 0 0 1 -0 .0 0 0 1 -0.00003 0.0000 (0.0003) (0 .0 0 0 1) (0 .0 0 0 1) (0 .0 0 0 1) (0 .0 0 0 1) (0 .0 0 0 0) (0 .0 0 0 0) (0 .0 0 0 1) (0 .0 0 0 1) current -0.44 -0.35 1.63 *** -1.04 *** -0.69 ** -0.39 ** -0.38 ** -0.46 ** -0.45 ** (0.98) (0.83) (0.51) (0.32) (0.29) (0.15) (0.16) (0.19) (0.19) capital -1.25 C D ) -1.62 *** (D) (D) -0.14 -0.15 -0.13 -0 .1 2 (1.26) (0.57) (0.13) (0.14) (0.16) (0.16) _cons 9.27 -25.32 ** -2 1 .6 8 *** -5.05 5.54 -8.17 ** -8.28 ** -9.90 ** -1 0 .1 2 ** (16.45) (11.18) (5.97) (8.81) (6.69) (3.43) (3.46) (3.98) (4.02) Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 4.4. Regression results from estimating the gravity model for intra-Arab total bilateral trade using OLS, wbere the null vlaues in the dependent variable are assumed genuinely missing (continued). 1985 Coef. (R.Std.Er) 1990 Coef. (R.Std.Er) 1995 Coef. (R.Std.Er) 1997 Coef. (R.Std.Er) 2 0 0 0 Coef. (R.Std.Er) Pooled (1985-2000) Coef. (R.Std.Er) Pooled (1985-2000) GCC & AMU Coef. (R.Std.Er) Pooled (1985-1997) Coef. (R.Std.Er) Pooled (1985-1997) GCC & AMU Coef. (R.Std.Er) No. of Obs 84 76 105 103 104 472 472 368 368 R-squared 0.37 0.59 0.70 0.67 0.63 0.52 0.52 0.51 0.51 RMSE 2.52 1.65 1.45 1.40 1.28 1.73 1.73 1.83 1.83 Notes: 1) D stands for dropped. 2) Standard errors are beneath the coefficients. 3) ***, **, * denotes significance at 1, 5, and 10% respectively. ■fs- 135 On the effect of economic size on intra-Arab total trade flows, consistent with the expectations of the gravity model, the impact of the product of GDPs on total trade flows is positive and highly significant. Concerning the product of GDPPCs, the coefficients are all negative and mostly highly significant. Negative GDPPC coefficients mean that, other things being equal, countries with high GDPPCs trade less with each other than countries with low GDPPCs. In the Arab world, countries with high GDPPCs just happen to be the oil- rich countries (especially GCC countries). Accordingly, this finding might be due to the relative similarity of the oil economies: relative similarity in factor endowments and lack of diversified export base. Another alternative explanation is based on the GCC countries’ relatively small populations with the exception of Saudi Arabia. GDPPC has two components: GDP and Population. The effect on trade of GDP has just been discussed and it proved positive and significant, meaning that the resulting negative effect of GDPPC is that of the population component, where countries with small populations trade less with each other than countries with large populations. Countries with small populations might trade less because they might have low import demand due to relative self sufficiency, or /and low export potential because of may be absence of economies of scale due to inefficient division of labor resulting from low supply of labor, sometimes a by product of small populations. The effect on intra-Arab total bilateral trade flows of distance is negative and mostly highly significant, consistent with expectations. While that of areas is negative and significant only in the pooled samples. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 136 Regarding the transaction-cost reducing variables - border, colonizer, GCC and A M U , and colonial - they are mostly not significant at any conventional level suggesting their negligible effect on intra-Arab trade flows.6 0 The transaction-cost increasing variable - ERV - has mostly negative coefficients and although significant in most of the individual years shows no significance at any conventional level in the pooled samples suggesting a negative but perhaps minor impact on intra-Arab trade flows. The effect on intra-Arab trade flows of Gov_ds is positive (except in 1985). As for the significance, Gov_ds is highly significant in 2000 and the whole pooled samples perhaps indicating that after 1997 governance became an important factor in intra-Arab trade while before that it was insignificant. As for M2GDP, it has coefficients that are mostly insignificant at any conventional level indicating its negligible effect on intra-Arab trade flows. Consistent with expectations, the coefficients of CPE are consistently negative and significant, even highly significant sometimes, indicating that previous centrally planned economies trade less with each other than never centrally planned economies. In the Arab world, only Algeria, Egypt, Iraq, Syria, and Yemen were ever centrally 60 This might have resulted from similar economic structures of neighboring countries as for instance within GCC, or within Maghreb countries. Concerning border alone, border disputes that have aroused from time to time between neighbors, like the Syrian vs. Lebanese 1980s border disputes, the Kuwait vs. Iraq 1990 dispute, war to be more precise, and the Egyptian vs. Sudanese disputes in the 1990s regarding the Hala’ib triangle as well as the other similar triangular areas, might have played a role. Concerning colonial alone, it has only values equal to 1 for Egypt and Sudan. That’s to say that the colonial relationship variable actually investigates more or less the trade relations between Egypt and Sudan, which seems to suggest that it is not a driving force behind Egypt-Sudan trade, which is expected. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 137 planned. That these countries trade less with each other is not puzzling. All of them in one or another point in time pursued industrial strategies based on import- substitution and a large public sector leading to high protection as well as nontransparent trade policies. In addition, these countries more or less have relative similarity of resource endowments limiting trade based on Heckscher-Ohlin, and do not have that diversified export base particularly in manufacturing limiting trade based on product differentiation. Concerning current account restrictions, its effect on intra-Arab trade flows is mostly negative and mostly highly significant. This is consistent with the expectations that countries with current account restrictions trade less with each other than with countries with no restrictions, other things being equal. This is especially true between Arab countries compared to the results in chapter 3, because many Arab countries still have not liberalized their current account restrictions. Coming to the last variable, capital account restrictions, its coefficients are mostly insignificant at any conventional level indicating its negligible effect on intra-Arab trade flows. As for the intercept, it is mostly negative and mostly significant. R2 has increased over time reaching acceptable levels from 1990 onwards and also in the pooled sample, which indicates a goodness of fit of the model. Based on the regressions of column 6 (pooled 1985-2000) in Table 4.4, another table was prepared, Table 4.4b, which shows the effect on intra-Arab total bilateral trade flows of increases in the continuous and the dummy variables. As expected, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 138 Table 4.4b. Predictions based cm Table 4.4, column 6 A=Coef. B=Std. Dev. A*B exp(A*B)- 1 exp(A )- 1 gdp 1.1213 1.3308 1.4923 345% 345% gdppc -0.5448 1.7160 -0.9348 -61% -61% distance -1.7317 0.7820 -1.3543 -74% -74% areas -0.4608 1.4078 -0.6487 -48% -48% 11 (D) border 0.3002 0.3450 35% 35% language (D) regional (D) nation (D) colonizer 0.1668 0.3998 18% 18% colonial 0.0099 0.1025 1% 1% erv -2.2493 0.0764 -0.1719 -16% -16% cu (D) gov_ds 0.0915 2.2633 0.2071 23% 23% cpy (D) cpe -0.8064 0.5937 -55% * -80% ** -55% * m2 gdp -0 .0 0 0 1 2694.1580 -0.1382 -13% -13% current -0.3867 0.6839 -32% * -54% ** -32% * capital -0.1444 0.6834 -13% * -25% ** -13% * * change due to change in dummy varaible from either 0 to 1 or 1 to 2 ** change due to change in dummy variable from 0 to 2 GDP is at the top of the positive effects, where a one standard deviation increase in GDP increases total trade flows by 345 %, i.e. almost 31/2 times. This is of course a very small figure compared to the results in Table 3.1b (16 times). The other positive effects are associated with border, Gov_ds, colonizer, and colonial. However as just discussed in the interpretations of Table 4.4, border, colonizer, and colonial, because of their low significance levels, have negligible effects on trade flows, and thus their values in Table 4.4b should not be taken seriously. Intra-Arab trade could be promoted through enhancing Gov_ds. A one standard deviation increase in Gov_ds increases total trade by 23%. All the remaining variables are trade impeding, the largest negative effect associated to distance. Concerning the newly introduced Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 139 variables, with the exception of Gov_ds (just discussed), no other variable has a positive effect on intra-Arab trade flows. CPE and current account restrictions have negative effects, while the effects of capital account restrictions and M2GDP are negligible. Table 4.5 presents the regression results from estimating the gravity model for intra-Arab total bilateral trade flows for: each individual year, the whole pooled sample (1985-2000), and the pooled sample not including year 2000. This is carried out using OLS, where the null values in the dependent variable are assumed to be zeros. The objective of this exercise is to check the sensitivities of the different coefficients to such assumption. Notice that column 3 in Table 4.4 and 4.5 are exactly the same. This means that in 1995, there are no null values in the dependent variable - meaning that all trade data exists for the observations that are complete in all other variables. As in Table 4.4, GDP is positive and mostly highly significant. GDPCC’s effect on trade flows is mostly negative and mostly highly significant. Regarding distance, it has negative coefficients all through and is mostly highly significant. Areas has mostly negative coefficients that are significant. The impact on intra-Arab total trade flows of border is positive and significant only in the pooled data. This perhaps indicates that border has only a minor effect on trade flows (This finding is a little different than the negligible effect of Table 4.4). Concerning colonizer, colonial, and ERV, their coefficients are not significant at Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 4.5. Regression results from estimating the gravity model for intra-Arab total bilateral trade using OLS, where the null values in the dependent variable are assumed zeros. 1985 Coef. (R.Std.Er) 1990 Coef. (R.Std.Er) 1995 Coef. (R.Std.Er) 1997 Coef. (R.Std.Er) 2 0 0 0 Coef. (R.Std.Er) Pooled (1985-2000) Coef. (R.Std.Er) Pooled (1985-1997) Coef. (R.Std.Er) gdp 4.10 *** 2.04 ** 1 61 1.50 *** 0.33 1.85 *** 2 .2 0 *** (1.34) (0.83) (0 .2 0 ) (0.54) (0.43) (0.34) (0.40) gdppc -4.33 *** -0.58 -0 .8 8 *** 0.26 0.42 -0 97 *** -1.47 *** (1.51) (0.80) (0.33) (0.75) (0.87) (0.32) (0.39) distance -3.11 ** -1.56 ** -1.62 *** -0.98 * -1.54 *** -1.99 *** -2 09 *** (1.53) (0.73) (0.32) (0.52) (0.39) (0.34) (0.40) areas -2.67 ** -1.52 -0.76 *** -0.03 0.29 -1 18 *** -1.54 *** (1.32) (0.98) (0 .2 0 ) (0.40) (0.46) (0.34) (0.42) 11 (D) (D) (D) (D) (D) (D) (D) border 1.08 1 .1 0 0.63 0.97 0.29 0.90 * 1.14 ** (2.34) ( 1.1 2) (0.52) (0.78) (0.64) (0.47) (0.57) language (D) (D) (D) (D) <P) (D) (D) regional (D) (D) (D) (D) (D) (D) (D) nation (D) (D) (D) (D) (D) (D) (D) colonizer 1.37 -0.23 0.37 1.89 0.92 0.62 0.73 (1.89) (2.28) (0.48) (1.29) (1.07) (0.49) (0.60) colonial 0.07 1.94 -0 .8 6 1 .2 2 1.93 -0 .2 0 -0.64 (3.66) (2.53) (0.79) (2.36) (2.34) (1.08) (1.28) erv -51.57 1.72 -5.74 4.64 -4.29 0.63 -0.78 (73.71) (15.93) (4.44) (12.50) (5.48) (3.16) (3.45) cu (D) (D) (D) (D) (D) (D) (D) Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 4.5. Regression results from estimating the gravity model for intra-Arab total bilateral trade using OLS, where the null values in the dependent variable are assumed zeros (continued). 1985 Coef. (R.Std.Er) 1990 Coef. (R.Std.Er) 1995 Coef. (R.Std.Er) 1997 Coef. (R.Std.Er) 2 0 0 0 Coef. (R.Std.Er) Pooled (1985-2000) Coef. (R.Std.Er) Pooled (1985-1997) Coef. (R.Std.Er) gov_ds -1.31 0.26 0.30 0.84 0.59 * 0.19 ** 0 .1 1 (0.91) (0.45) (0 .2 2 ) (0.57) (0.31) (0.09) (0 .1 0) cpy <P) (D) (D) (D) (D) (D) (D) cpe -1.95 -1.24 -1 77 *** -1.09 0.78 - 1 .0 1 ** -1 28 *** (2.13) (0 .8 6 ) (0.51) (0 .6 8 ) (1.38) (0.47) (0.44) m2 gdp -0.00029 0.00019 -0 .0 0 0 2 2 ** -0 .0 0 0 1 2 -0 .0 0 0 0 2 -0.00007 -0.00008 (0.00078) (0.00019) (0 .0 0 0 1 0 ) (0.00015) (0.00013) (0.00008) (0 .0 0 0 1 1 ) current -5.57 ** -1.03 1.63 *** -1.63 * -0.82 ** -0.85 *** -I 13 *** (2.70) (2.09) (0.51) (0.84) (0.38) (0.31) (0.40) capital 0.93 (D) -1.62 *** (D) (D) -0.26 -0.48 (4.08) (0.57) (0.23) (0.31) _cons -47.06 -37.79 -2 1 .6 8 *** -50.86 1.64 -24.23 *** -26.84 *** (52.68) (31.21) (5.97) (37.37) (10.72) (8.16) (9.26) No.of Obs 91 78 105 105 105 484 379 R-squared 0.42 0.27 0.70 0.33 0.32 0.28 0.30 RMSE 6.34 4.68 1.45 3.92 2.98 4.28 4.52 Notes: 1) D stands for dropped. 2) Standard errors are beneath the coefficients. 3) ***, **, * denotes significance at 1, 5, and 10% respectively. 4 * . 142 conventional levels, suggesting their negligible effect on intra-Arab trade flows. Concerning ERV, this is different from the conclusion regarding Table 4.4 regression results, perhaps underlining the sensitivity of ERV to the different assumptions employed. Gov_ds' effect on intra-Arab total trade flows is positive (except in 1985). Its coefficients are highly significant in 2000 and the whole pooled sample. As you notice Gov_ds has been quite robust to the different assumptions since results are similar to those in Table 4.4. As for M2GDP, its coefficients are not significant at any conventional level suggesting its negligible effect on intra-Arab trade flows. The effect of CPE on intra-Arab trade flows is negative (except in 2000) and significant in the pooled years as well as in 1995. This indicates that CPE played a minor role in impeding trade flows - a finding not as strong as that in Table 4.4. Current account restrictions show mostly a negative effect on intra-Arab trade flows and is mostly highly significant. This finding is the same one reached in Table 4.4 indicating that current account restrictions is quite robust to the different assumptions. As for capital account restrictions, its coefficients are mostly not significant at any conventional level, indicating that it perhaps has a negligible effect Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 143 on trade between Arab countries.6 1 R2 indicates a relatively poor goodness of fit as compared to the R2 values in Table 4.4. To conclude this exercise, a number of points are due. First, except for 1995, where coefficients in Table 4.4 are exactly the coefficients in Table 4.5, most of the magnitudes of the coefficients in Table 4.5 are larger than their corresponding magnitudes in Table 4.4, again supporting the claim that OLS downwardly bias coefficients by excluding zero trade flows. Second, some variables are robust to the assumptions employed, where they did not experience significant changes (with very few exceptions) in Table 4.5 compared to Table 4.4 and thus there was no change in the conclusions regarding these variables’ effect on intra-Arab trade flows. These variables are: GDP, GDPPC, distance, areas, colonizer, colonial, Gov_ds, CPE, M2GDP, current and capital account restrictions. On the other hand, border and ERV are only relatively robust because they are sensitive to the assumptions made. Based on the regressions of column 6 (pooled 1985-2000), Table 4.5b is prepared to present the effect on intra-Arab total bilateral trade flows of the increases in the continuous and dummy variables. All percentages are bigger than their corresponding ones in Table 4.4b as expected. The largest effect is associated with GDP just as in Table 4.4b. A one standard deviation increase in Gov_ds increases 61 Further investigations revealed that in 1985, capital account restrictions has a positive coefficient because it is correlated to current account restrictions. When current account restrictions was dropped and the regression was run once again, capital account restrictions became negative (-3.03) but still insignificant at even 1 0 % . Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 144 Table 4.5b. Predictions based on Table 4.5 column 6 A=Coef. B=Std. Dev. A*B exp(A*B) - 1 exp (A )- 1 gdp 1.8469 1.3248 2.4468 1055% 1055% gdppc -0.9719 1.7209 -1.6725 -81% -81% distance -1.9921 0.7794 -1.5527 -79% -79% areas -1.1817 1.4051 -1.6604 -81% -81% 11 (D) border 0.9025 0.3413 147% 147% language (D) regional (D) nation (D) colonizer 0.6189 0.3976 8 6 % 8 6 % colonial -0.1999 0 .1 0 1 2 -18% -18% erv 0.6252 0.0764 0.0477 5% 5% cu (D) gov_ds 0.1861 2.2700 0.4224 53% 53% cpy C D ) cpe -1.0123 0.5915 -64% * -87% ** -64% * m2 gdp -0 .0 0 0 1 2677.1640 -0.1818 -17% -17% current -0.8464 0.6828 -57% * -82% ** -57% * capital -0.2556 0.6794 -23% * -40% ** -23% * * change due to change in dummy varaible from either 0 to 1 or 1 to 2 ** change due to change in dummy variable from 0 to 2 trade flows by 53%. If two countries decided to place current account restrictions at the same time trade will decrease by 87%, if one country only decided to do so or joined the other one lately trade decreases by 64%. With the exception of GDP, border, colonizer (insignificant and thus negligible) and Govjds, all other factors impede trade. ERV positively impacts trade here because it has a positive value in the regression, however it shouldn’t be taken seriously because it is insignificant. This brings us to an important point; not all predictions should be taken seriously, only the ones associated with significant coefficients because these are the only meaningful ones. Consequently the predictions of the variables impeding trade that should be Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 145 taken into account are the predictions associated only with GDPPC, distance, areas, CPE, and current account restrictions. In yet another stimulating exercise, the individual governance components: VA, PS, GE, RQ, RL, and CC are introduced as explanatory variables one at a time. The goal behind such exercise is to again explore the sensitivities of the variables to these specifications, and to try to determine which governance component is the most important. Table 4.6 displays the regression results from estimating the gravity model, with the individual governance components each at a time, for intra-Arab total bilateral trade using OLS for the whole sample (1985-2000), where the null values were assumed genuinely missing. This Table can be compared to column 6 in Table 4.4. Some variables were pretty robust to the different specifications used in Table 4.6 and as compared to Table 4.4. GDP, GDPPC, distance, areas, border, colonizer, colonial, ERV, CPE, M2GDP, and current account restrictions had no sign changes and only some of them had very minor changes in significance, which can be disregarded. Capital account restrictions with further investigations proved that it, too, is a fairly robust variable.6 2 Concerning the governance variable and its components. Gov_ds is positive and significant at less than 5% level in column 6 Table 4.4. Most of the individual 62 Capital account restrictions is correlated to current account restrictions and when the latter was dropped, capital account restrictions became negative but not significant (same as original conclusion in Table 4.4). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 4.6. Regression results from estimating the gravity model for intra-Arab total bilateral trade for pooled data (1985-2000) using OLS, where each of the six governance components is used separately and where the null values in the dependent variable are assumed genuinely missing. Coef. (R.Std.Er) Coef. (R.Std.Er) Coef. (R.Std.Er) Coef. (R.Std.Er) Coef. (R.Std.Er) Coef. (R.Std.Er) gdp 1.15 *** 1 1 0 *** I jj *** 1 12 *** 1.18 *** (0 . 1 0 ) (0 .1 1 ) (0 .1 0) (0 .1 0) (0 . 1 0) (0 .1 0 ) gdppc -0.51 *** -0.56 *** -0.57 *** -0.62 *** -0.56 *** -0.58 *** (0 . 1 2) (0 .1 0) (0 .1 0) (0 . 1 0) (0 . 1 0) (0 .1 1) distance -I 73 *** -I 72 *** -1 73 *** -1 69 *** -1.71 *** -1 72 *** (0.17) (0.17) (0.17) (0.18) (0.18) (0.17) areas -0.48 *** -0.43 *** -0.47 *** -0.54 *** -0.47 *** -0.52 *** (0 . 1 2) (0.13) (0 .1 1 ) (0 -1 2) (0 .1 2) (0 .1 2) 11 (D) (D) (D) (D) (D) (D) border 0.30 0.30 0.30 0.38 0.36 0.33 (0.31) (0.30) (0.31) (0.31) (0.31) (0.31) language (D) (D) (D) (D) (D) (D) regional (D) (D) (D) (D) (D) (D) nation (D) (D) (D) (D) (D) (D) colonizer 0.16 0.18 0.18 0 .1 0 0.16 0 .1 1 (0.23) (0.23) (0.23) (0 .2 2 ) (0.23) (0 .2 2 ) colonial -0.09 0.04 0.06 -0.31 -0 .1 1 -0.16 (0.53) (0.55) (0.63) (0.53) (0.54) (0.55) erv -2.30 -2.41 -2.32 -3.15 ** -2.59 * -2.64 * (1.52) (1.50) (1.53) (1.48) (1.53) (1.53) cu (D) (D) (D) (D) (D) (D) Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 4.6. Regression results from estimating the gravity model for intra-Arab total bilateral trade for pooled data (1985-2000) using OLS, where each of the six governance components is used separately and where the null values in the dependent variable are assumed genuinely missing (continued). Coef. (R.Std.Er) Coef. (R.Std.Er) Coef. (R.Std.Er) Coef. (R.Std.Er) Coef. (R.Std.Er) Coef. (R.Std.Er) va_d 0.06 * (0.03) ps_d 0.04 ** (0 .0 2 ) ge_d 0.07 ** (0.03) rq_d -0.03 (0.03) rl_d 0.08 * (0.04) cc_d 0.06 (0.07) cpy (D) (D) (D) (D) (D) (D) cpe -0 79 *** -0.80 *** -0.84 *** -0 79 *** -0.76 *** -0.82 *** (0 .2 0 ) (0.19) (0.19) (0 .2 0 ) (0 .2 0 ) (0.19) m2 gdp -0.00007 * -0.00004 -0.00005 -0.00006 -0.00005 -0.00006 (0.00004) (0.00004) (0.00004) (0.00004) (0.00004) (0.00004) current -0.41 *** -0.42 *** -0.35 ** -0.54 *** -0.41 *** -0.50 *** (0.15) (0.15) (0.16) (0.17) (0.15) (0.16) capital 0 .0 0 -0 .1 1 -0.36 ** 0.05 -0 .1 0 0 .0 0 (0 .1 2) (0 . 1 2) (0.18) (0.15) (0.13) (0.13) _cons -9.53 *** -7.78 ** -7.06 ** -9.46 *** -7.98 ** -9.59 *** (3.43) (3.44) (3.37) (3.54) (3.45) (3.59) -J Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 4.6. Regression results from estimating the gravity model for intra-Arab total bilateral trade for pooled data (1985-2000) using OLS, where each of the six governance components is used separately and where the null values in the dependent variable are assumed genuinely missing (continued). Coef. (R.Std.Er) Coef. (R.Std.Er) Coef. (R.Std.Er) Coef. (RStd.Er) Coef. (R.Std.Er) Coef. (R.Std.Er) No. of Obs 472 472 472 472 472 472 R-squared 0.52 0.52 0.52 0.52 0.52 0.52 RMSE 1.73 1.72 1.72 1.73 1.73 1.73 Notes: 1) D stands for dropped. 2) Standard errors are beneath the coefficients. 3) ***, **, * denotes significance at 1, 5, and 10% respectively. 4 5 - o o 149 governance components show the same thing; voice and accountability, political stability, government effectiveness and rule of law all have positive coefficients and are significant, suggesting that pairs of Arab countries enjoying high scores in any of these components trade more with each other than countries that have low scores. Once again, as in Table 3.5, rule of law seems to have the highest coefficient in comparison to the other governance components, reinforcing its importance. Regulatory quality is insignificant at conventional levels, however, a negative sign perhaps suggests that better regulatory quality in the one hand makes domestic firms more competitive and thus less keen to export and on the other hand makes foreign firms less able to enter and compete with domestic firms and thus less keen to import, hence the negative effect on trade flows of regulatory quality. Concerning control of corruption, it is also not significant at any conventional level. CC might have a negligible effect on trade flows because, with the exception of Iraq, Lebanon and Somalia in some years, the rest of the Arab countries do not have very different corruption scores (2-4 according to ICRG Tables) that choosing between trading partners within the Arab world on the basis of control of corruption might not be worth while - they all have close scores - moderately corrupt. Moreover, the fact that Arab countries enjoy a common language, and have a high degree of cultural similarity allows traders to deal better with corruption, thus rendering control of corruption’s effect on trade flows negligible. Tables 4.7 and 4.8 present the results of estimating the gravity model separately for energy and other intra-Arab trade. Once again one would expect the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 150 gravity considerations to be quite a bit weaker for energy trade than other trade. To some extent this is the case. R2 is considerably lower for the energy trade in Table 4.7 than that of the other trade in Table 4.8. Tables 4.7 and 4.8 display respectively the regression results from estimating the gravity equation for energy and other intra-Arab trade for the pooled sample not including year 2000 with and without the Arab regional trade agreements - GCC and AMU. For the results not including Arab RTA’s, they were carried once where the null values are assumed to be genuinely missing and once where some of the missing values were assumed to be genuinely missing while the others were assumed to be zeros when they were found zeros in all the other years. Only the latter assumption was used when results included GCC and AMU. In all cases the zero values were excluded. For Table 4.7, the different assumptions yielded the same column (column 1) indicating that there were perhaps no complete observations that had null values in the dependent variable, or that they didn’t have zeros in all the other years. Consistent with expectations and earlier explanations, the coefficients of GDP are all positively signed and significant in both tables, while those of GDPPC and CPE are negative and significant. As for the coefficients of border, GCC, ERV, and capital account restrictions, they are insignificant at any conventional level in both tables indicating their negligible effect on intra-Arab trade flows. All the other left out variables reveal considerable differences in their results and perhaps interpretations in both tables. In Table 4.8, as expected, distance and Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 151 Table 4.7. Regression results from estimating the gravity model for intra-Arab energy trade flows using OLS for pooled data (1985-1997). Two Null assumptions Coef. (RStd.Er) Some null values are assumed genuinely missing while others are assumed zeros and including GCC & AMU Coef. (R.Std.Er) gdP 0.66 ** 0.69 ** (0.33) (0.34) gdppc -0.86 ** -0.97 ** (0.33) (0.40) distance -0.08 0.12 (0.45) (0.56) areas -0.07 -0.13 (0.42) (0.42) 11 (D) C D ) border 1.05 1.13 (0.79) (0.84) language (D) (D) regional (D) (D) gcc 0.74 (1.10) amu 0.09 (1.00) nation (D) (D) colonizer 1.12 * 1.09 (0.61) (0.72) colonial -7.25 *** -7 39 *** (1.18) (1.32) erv -3.32 -3.58 (5.19) (5.22) cu (D) (D) gov_ds -0.09 -0.10 (0.13) (0.13) cpy (D) C D ) cpe -1.63 *** -1.57 *** (0.53) (0.55) m2 gdp -0.0003 *** -0.0002 ** 0.0001 0.0001 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 152 Table 4.7. Regression results from estimating the gravity model for intra-Arab energy trade flows using OLS for pooled data (1985-1997) (continued). Two Null assumptions Coef. (RStd.Er) Some null values are assumed genuinely missing while others are assumed zeros and including GCC & AMU Coef. (R.Std.Er) current 0.03 0.05 (0.47) (0.46) capital -0 .2 0 -0.17 (0.41) (0.41) _cons 0.40 -0.25 ( 1 1 .0 1 ) ( 1 1 .0 0 ) No.of Obs 140 140 R-squared 0.27 0.27 RMSE 2.65 2 .6 6 Notes: 1) D stands for dropped. 2) Standard errors are beneath the coefficients. 3) ***, **, * denotes significance at 1, 5, and 10% respectively. areas have negative coefficients that are highly significant, and current account restrictions has mostly negative and significant coefficients. Also as expected, Gov_ds is positive and significant. As for the coefficients of colonizer, colonial, and M2GDP, they are insignificant at conventional levels indicating their negligible effect on intra- Arab other trade flows. AMU has negative and significant effect on intra-Arab other trade flows indicating that other trade between two AMU member countries is less than two non-AM U countries implying that AMU may have not really liberalized trade amongst each other rendering this regional agreement negligible and/or they have similar factor endowments. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 153 Table 4.8. Regression results from estimating the gravity model for intra-Arab other trade flows using OLS for pooled data (1985-1997). Null values are assumed genuinely missing Coef. (R.Std.Er) Some null values are assumed genuinely missing while others are assumed zeros Coef. (R.Std.Er) Some null values are assumed genuinely missing while others are assumed zeros and including GCC & AMU Coef. (RStd.Er) gdp 0 9 7 *** 1.13 *** 1 j g *** (0 .1 2 ) (0.13) (0.13) gdppc -0.34 *** -0.51 *** -0.52 *** (0 .1 2 ) (0 .1 2 ) (0.13) distance -1.56 *** -1 72 *** -1.81 *** (0 .2 0 ) (0 .2 1 ) (0.26) areas -0.42 *** -0.57 *** -0.58 *** (0.15) (0.15) (0.16) 11 (D) ( D ) ( D ) border 0.54 0.52 0.64 (0.39) (0.38) (0.41) language (D) (D) (D) regional (D) (D) (D) gcc amu nation (D) (D) -0.29 (0.45) -0.96 * (0.57) (D) colonizer 0.14 0 .2 1 0.33 (0.28) (0.27) (0.31) colonial 0 .6 6 0.18 0.04 (0.61) (0.71) (0.73) erv -2.37 -0.58 -0.95 (1.96) (1.74) (1.79) cu (D) (D) (D) gov_ds q \2 * * * 0.08 * 0.09 * c p y (D) (0.05) (D) (0.05) CD) cpe -0.57 *** -0.63 *** -0 .6 8 *** (0 .2 2 ) (0.23) (0.23) m2 gdp 0.000025 0.000003 -0.000013 (0.00005) (0.00005) (0.00006) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 154 Table 4.8. Regression results from estimating the gravity model for intra-Arab other trade flows using OLS for pooled data (1985-1997) (continued). Some null values are Null values are assumed genuinely missing Coef. (RStd.Er) Some null values are assumed genuinely missing while others are assumed zeros Coef. (R.Std.Er) assumed genuinely missing while others are assumed zeros and including GCC & AMU Coef. (R.Std.Er) current capital cons No.of Obs R-squared RMSE 0.06 (0.17) -0.14 (0.16) -6.83 * (3.87) 289 0.54 1.51 -0.32 * (0.18) -0.04 (0.16) -8.70 ** (3.96) 329 0.51 1.75 -0.31 (0.18) -0.05 (0.16) -9.01 (4.01) 329 0.51 1.75 Notes: 1) D stands for dropped. 2) Standard errors are beneath the coefficients. 3) ***, **, * denotes significance at 1, 5, and 10% respectively. With respect to energy trade flows, both distance and areas have insignificant coefficients indicating that for energy trade, it really all depends on which countries are energy endowed irrespective of their distances from other countries or the sum of their geographical areas. The effect of a common colonizer on intra-Arab trade is positive and almost significant for energy trade. As for colonial, its coefficients are negative and highly significant. But this is expected when realizing that Egypt and Sudan are both net energy importers suggesting that in that resource, neither of them has much to offer the other and so they trade less. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 155 Concerning the regional variables, GCC and the AMU, they have an insignificant effect on energy trade flows, indicating that being a member of a regional trading agreement does not necessarily mean more energy trade than amongst non members. Concerning GCC this might be true because they have similar factor endowments and thus might not need to trade with each other. This also indicates that these oil-rich countries have low or limited intra-industry trade in energy resources. Regarding governance, its coefficients are insignificant as far as energy trade is concerned, perhaps indicating that governance has a negligible effect on energy trade flows. Energy exporting really depends on which countries own energy resources irrespective of their governance score, and it is a vital good that energy importers might not think twice about their partner’s governance scores. Concerning M2GDP, its coefficients are negative and significant with respect to energy trade perhaps indicating that Arab countries that trade in energy resources are themselves not financially developed. As for current account restrictions, its coefficients are insignificant suggesting the negligible effect of current account restrictions on energy trade. Now, let’s move to Arab-World trade results leaving the summary of both intra-Arab and Arab-World regression results to the end of the chapter. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 156 4.2.2 Arab-World Trade Results This section explores Arab-World trade, where again the Gini variable was excluded. Concerning the trading pairs, one of them is an Arab country, while the other is a non- Arab country. Table 4.9 and 4.10 present the regression results from estimating the gravity model for Arab-World total bilateral trade flows for: each individual year, the whole pooled sample (1985-2000), and the pooled sample not including year 2000. These were carried out using OLS, where once the null values in the dependent variable were assumed to be genuinely missing thus resulting in Table 4.9, and once they were assumed to be zeros thus resulting in Table 4.10. Consistent with the expectations of the gravity model, the impact of the product of GDPs on total bilateral trade flows is positive and highly significant throughout both tables. Similarly, colonizer, Gov_ds, and M2 GDP have positive coefficients that are mostly highly significant in both tables. The magnitudes of all these variables’ coefficients in Table 4.10 are larger than their corresponding ones in Table 4.9 as expected and as explained earlier. Concerning the product of GDPPCs, the coefficients are negative and mostly highly significant throughout the two tables with magnitudes in Table 4.10 greater than magnitudes in Table 4.9. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 4.9. Regression results from estimating the gravity model for Arab-World total bilateral trade using OLS, where the null values in the dependent variable are assumed genuinely missing. 1985 Coef. (R.Std.Er) 1990 Coef. (R.Std.Er) 1995 Coef. (R.Std.Er) 1997 Coef. (R.Std.Er) 2 0 0 0 Coef. (R.Std.Er) Pooled (1985-2000) Coef. (R.Std.Er) Pooled (1985-1997) Coef. (R.Std.Er) gdp 1 .2 0 *** 1.35 *** 1.36 *** 1.33 *** 1.36 *** j [ 29 *** 1.26 *** (0.07) (0.06) (0.05) (0.05) (0.05) (0 .0 2 ) (0.03) gdppc -0.17 ** -0 39 *** -0.19 *** -0.06 -0 .1 1 ** -0.16 *** -0.16 *** (0.08) (0.06) (0.06) (0.05) (0.05) (0.03) (0.03) distance -1 23 *** -1.58 *** -1 75 *** -1.57 *** -1 40 *** -1.53 *** -1 57 *** (0.17) (0.14) (0 .1 0) (0 .1 1) C O . l i ) (0.05) (0.06) areas -0.52 *** -0.47 *** -0 .1 0 -0 .1 2 * -0.17 ** -0.19 *** -0.19 *** (0.09) (0.09) (0.06) (0.07) (0.07) (0.03) (0.04) 11 -0.14 -0.26 -0 78 *** -0.45 ** -0 49 *** -0.63 *** -0.60 *** (0.36) (0.30) (0.18) (0.18) (0.18) (0 .1 0) (0 . 1 1) border 0.43 -0.64 -0.39 1.05 * 0.09 0 .0 1 0.07 (0.77) (0.64) (0.89) (0.62) (0.80) (0.37) (0.41) language 1 .2 0 I 13 *** -4.53 *** -5.50 *** -3 28 *** -4.01 *** -4.29 *** (0.84) (0.44) (1.03) (0.83) (0.56) (0.67) (0.79) regional (D) (D) (D) (D) (D) (D) (D) nation (D) (D) (D) CD) CD) (D) (D) colonizer 0.83 *** 0.49 * 0.49 ** 0.71 *** 1 03 *** 0.62 *** 0.52 *** (0.29) (0.26) (0 .2 1 ) (0 .2 2 ) (0.24) (0 .1 1) (0 .1 2) colonial (D) (D) C D ) (D) (D) (D) (D) erv -3.83 ** -0.29 0.04 0.39 6.45 *** -0.09 -1.29 *** (1.79) ( 1.1 0) (0.85) (0.85) (1.26) (0.45) (0.48) cu (D) (D) (D) (D) (D) (D) (D) e/i Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 4.9. Regression results from estimating the gravity model for Arab-World total bilateral trade using OLS, where the null values in the dependent variable are assumed genuinely missing (continued). 1985 Coef. (R.Std.Er) 1990 Coef. (R.Std.Er) 1995 Coef. (R.Std.Er) 1997 Coef. (R.Std.Er) 2 0 0 0 Coef. (R.Std.Er) Pooled (1985-2000) Coef. CR.Std.Er) Pooled (1985-1997) Coef. (R.Std.Er) gov_ds 0.18 *** 0.24 *** 0.24 *** q 3 2 *** 0.25 *** 0.08 *** 0.08 *** (0.06) (0.05) (0.04) (0.05) (0.04) (0 .0 2 ) (0 .0 2 ) cpy 1.16 ** (D) (D) (D) (D) 1 04 *** 0.84 ** (0.51) (0.39) (0.39) cpe -0.82 *** -0 .6 8 *** -0 .0 2 -0.43 *** -0.62 *** -0.38 *** -0.36 *** (0.24) (0.18) (0.15) (0 .1 2) (0 .1 1) (0.06) (0.08) m2 gdp 0.0003 *** 0 .0 0 0 1 *** 0 .0 0 0 1 *** 0 .0 0 0 1 *** 0 .0 0 0 1 *** 0 .0 0 0 1 *** 0 .0 0 0 2 *** (0.00008) (0.00004) (0.00003) (0 .0 0 0 0 2 ) (0 .0 0 0 0 2 ) (0 .0 0 0 0 1 ) (0 .0 0 0 0 2 ) current 0.14 0 .1 1 0.04 -0.14 -0.23 ** 0 .0 1 0.13 * (0 .2 0 ) (0.23) (0.15) (0 .1 1) (0 .1 1 ) (0.06) (0.07) capital -0 .6 8 *** -1.35 *** -0.57 *** 0.76 *** 0.96 *** -0.51 *** -0.60 *** (0.24) (0.25) (0.13) (0.27) (0.25) (0.06) (0.07) _cons -21.43 *** -22.17 *** -31.26 *** -35.29 *** -37.05 *** -28.45 *** -26.90 *** (3.25) (2.48) (1.70) (1.90) (1.81) (0.87) ( 1.0 1 ) No.ofObs 690 747 1128 1106 1162 4833 3671 R-squared 0.51 0.58 0.67 0 .6 6 0.63 0.59 0.59 RMSE 2.41 2 .2 0 2.06 2.06 2 .1 1 2.23 2.24 Notes: 1) D stands for dropped. 2) Standard errors are beneath the coefficients. 3) ***, **, * denotes significance at 1, 5, and 10% respectively. U \ 00 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 4.10. Regression results from estimating the gravity model for Arab-World total bilateral trade using OLS, where the null values in the dependent variable are assumed zeros. 1985 Coef. (R.Std.Er) 1990 Coef. (R.Std.Er) 1995 Coef. (R.Std.Er) 1997 Coef. (R.Std.Er) 2 0 0 0 Coef. (R.Std.Er) Pooled (1985-2000) Coef. (R.Std.Er) Pooled (1985-1997) Coef. (R.Std.Er) gdp 3.76 *** 4.06 *** 3.40 *** 3 y] *** 3.20 *** 3.43 *** 3 .4 7 *** (0 .2 1 ) (0.18) (0.16) (0.16) (0.17) (0.08) (0.08) gdppc -1.70 *** -1.57 *** -1 16 *** _0 4 9 -0.56 *** - 1 .0 2 *** -1 13 *** (0.26) (0.23) (0.19) (0.17) (0.19) (0.09) (0 .1 0) distance -4 33 *** -4.27 *** -4 12 *** -3.52 *** -4.03 *** -4.08 *** -4.09 *** (0.55) (0.42) (0.34) (0.35) (0.47) (0.18) (0.19) areas -1 37 *** -1.52 *** -0.91 *** -0 87 *** -0 .8 6 *** -1.03 *** - 1 .0 2 *** (0-31) (0.27) (0 .2 1 ) (0 .2 2 ) (0.23) (0 .1 1) (0 .1 2) 11 -2 89 *** -0 .8 8 -0 .1 1 -0.43 -0.72 -1 1 2 *** -1.06 *** (1.04) (0.96) (0.57) (0.63) (0.65) (0.32) (0.37) border 3.95 1.17 -1.82 -0.18 2.70 0.75 0.36 (2.92) (2.82) (3.06) (3.26) (2.72) (1.31) (1.49) language -26.55 *** -25.00 *** -22.36 *** -17.03 *** -26.66 *** -23.55 *** -22.96 *** (1.99) (3.14) (3.20) (3.15) (3.23) (1.38) (1.51) regional <D) (D) (D) (D) (D) (D) (D) nation (D) (D) (D) (D) (D) (D) (D) colonizer 2.17 ** 2 3 2 *** 2 09 *** I 9 9 *** 2.67 *** 2 12 *** P96 *** (0 .8 6 ) (0.78) (0.62) (0.70) (0.83) (0.34) (0.37) colonial (D) CD ) (D) (D) (D) (D) (D) erv2 -13.35 ** 3.67 0.38 -3.02 2.70 -1.07 -3.03 ** (6.06) (3.48) (2.73) (2.73) (4.23) (1.44) (1.54) cu (D) (D) (D) (D) (D) (D) (D) Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 4.10. Regression results from estimating the gravity model for Arab-World total bilateral trade using OLS, where the null values in the dependent variable are assumed zeros (continued). 1985 Coef. (R.Std.Er) 1990 Coef. (R.Std.Er) 1995 Coef. (R.Std.Er) 1997 Coef. (R.Std.Er) 2 0 0 0 Coef. (R.Std.Er) Pooled (1985-2000) Coef. (R.Std.Er) Pooled (1985-1997) Coef. (R.Std.Er) gov_ds 2 25 *** 1 28 *** 1.07 *** 1 .0 2 *** 0 .8 6 *** q 7 3 *** 0 .8 6 *** (0 .2 0 ) (0.18) (0.14) (0.14) (0.15) (0.06) (0.06) cpy 2.57 (D) (D) (D) (D) 0.84 0.64 (1.84) (1.84) cpe -3.62 *** -3.16 *** -0.98 ** -0.83 ** -0.70 - 1 .2 2 *** -1.62 *** (0.87) (0.62) (0.48) (0.39) (0.46) (0 .2 2 ) (0.25) m2 gdp 0.0003 0.0004 *** 0 .0 0 0 2 * 0 .0 0 0 2 *** 0.0003 *** 0.0003 *** 0.0003 *** (0.00028) (0 .0 0 0 1 2 ) (0.00008) (0.00007) (0.00007) (0.00004) (0.00005) current -0.91 2 ,9 4 *** 0 .8 8 * -0.27 -1 2 0 *** -0.08 0.37 * (0.69) (0.74) (0.48) (0.36) (0.39) (0.19) (0.23) capital -0.62 -3 71 *** -2.62 *** 1.93 ** 0.61 -1.72 *** -1.78 *** (0.83) (0 .8 6 ) (0.45) (0.94) (0.99) (0 .2 1 ) (0.24) _cons -82.74 *** -97.45 *** -83.67 *** -96.10 *** -90.81 *** -86.37 *** -86.64 *** (10.32) (8.17) (5.64) (5.99) (6.63) (2.84) (3.16) No.of Obs 994 1027 1350 1320 1470 6161 4691 R-squared 0.42 0.47 0.49 0.46 0.36 0.43 0.46 RMSE 9.43 8.77 7.37 7.58 8.97 8.51 8.30 Notes: 1) D stands for dropped. 2) Standard errors are beneath the coefficients. 3) ***, **, * denotes significance at 1, 5, and 10% respectively. 1 6 1 Also consistent with expectations, the effects on Arab-World total bilateral trade flows of distance, areas, LL, and CPE are negative and mostly highly significant in both tables (with few exceptions for LL). In all four variables almost all the magnitudes of the coefficients in Table 4.10 are larger than their corresponding ones in Table 4.9. Regarding border, its coefficients are insignificant indicating that in Arab- World trade, a common border does not play a significant role. This is not hard to accept because Arab countries that lie in Africa border African countries that might lack the type of goods, especially industrialized goods that Arab African countries might want to import and vice versa. Concerning Arab countries that lie in Asia, they mostly border each other. For the few that border other countries (Iraq-Iran) (Iraq- Turkey) (Syria-Turkey) (Arab countries bordering Israel), Iraq was at war with Iran after which it engaged in war with Kuwait after which it experienced economic sanctions definitely adversely affecting its trade with Iran and Turkey. With the exception of Egypt and Jordan, Arab countries bordering Israel do not trade with it. Thus the fact that border has a negligible effect on Arab-World trade is not after all surprising. Remember also that Table 4.1 revealed that Arab countries trade the most with industrialized countries followed by Asian countries, with which they do not have common borders. Concerning language, Chad is the only non-Arab country that has Arabic as an official language. Djibouti is the only Arab country that has another language (French) as an official language. It is thus expected that a common language might Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 162 have a negligible effect on Arab-World trade flows. If Israel is included as a country that has Arabic as an official language, which it does, the language variable is expected to even have a negative effect on trade because of many Arab countries’ non existent trade relations with Israel.6 3 Tables 4.9 and 4.10 support this conclusion. In Table 4.9 the coefficients of the language variable are mostly negative, and highly significant. When including zero trade flows, the magnitudes of the coefficients dramatically jump upwards, and as you can see all language coefficients are negative and highly significant in Table 4.10. Regarding ERV, Table 4.9 shows mixed results for E R V s coefficients. ERV proved sensitive to the zero values inclusion in the dependent variable as shown in Table 4.10, but still showed mixed results. ERV is not a reliable variable: for instance, in the pooled years (1985-2000) it was insignificant in both tables. On the other hand, in pooled years not including year 2000, it showed negative coefficients that are highly significant. Both results are theoretically and empirically supported, as discussed in section 3.3. This makes reaching a conclusion about the significance of this variable and its effect on trade very hard. CRY shows a positive and significant effect on Arab-World trade flows in Table 4.9. Note however that the only Arab country that was centrally planned until 1985 was Yemen. So this variable captures the trade relations with Yemen and other centrally planned countries in 1985. CPY is a little sensitive to the assumption 63 Arab countries that do not trade with Israel are: Algeria, Bahrain, Comoros, Lebanon, Libya, Saudi Arabia, Somalia, Sudan, Syria, and Yemen. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 163 employed in Table 4.10; as it shows an insignificant coefficient indicating that CPY probably has a negligible effect on Arab-World trade flows. The impact on Arab-World total trade flows of current account restrictions is negligible. In Tables 4.9 and 4.10, current account restrictions’ coefficients have positive results because of their correlation with capital account restrictions. When the latter was dropped, the coefficients of current account restrictions became negative but were still insignificant at conventional levels indicating that current account restrictions has negligible effects on Arab-World trade flows. As for capital account restrictions, its effect on Arab-World trade flows is negative and highly significant, except in 1997 & 2000. Further investigations revealed that capital account restrictions in these two years became negative and significant when current account restrictions was dropped. This indicates that capital account restrictions have an adverse effect on Arab-World trade flows. As for the intercept, it is negative and highly significant in both tables also consistent with expectations. R2 ranges from 0.5 to 0.67 while RMSE ranges from 2.06 to 2.41 in Table 4.9, which provides a strong support for the goodness of fit of the model. The goodness of fit fell substantially in Table 4.10 as compared to Table 4.9 where R2 ranges from 0.36 to 0.49 while RMSE ranges from 7.37 to 9.43. Based on regressions of column 6 (pooled 1985-2000) in Tables 4.9 and 4.10, Tables 4.9b and 4.10b display the effect on Arab-World total trade flows of the increases in the continuous and dummy variables. Predictions that are based on Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 164 Table 4.9b. Predictions based on Table 4.9 column 6 A=Coef. B=Std. Dev. A*B exp(A*B) - 1 exp(A )-1 gdp 1.2885 2.1858 2.8163 1572% 1572% gdppc -0.1617 1.9713 -0.3187 -27% -27% distance -1.5338 0.6663 -1.0219 -64% -64% areas -0.1902 1.4276 -0.2715 -24% -24% 11 -0.6327 0.3654 -47% * -72% ** -47% * border 0.0050 0.0906 1% 1% language -4.0112 0.0658 -98% -98% regional (D) nation (D) colonizer 0.6231 0.3509 8 6 % 8 6 % colonial (D) erv -0.0862 0.0842 -0.0073 - 1 % - 1% cu (D) gov_ds 0.0789 2.0085 0.1584 17% 17% cpy 1.0424 0.0673 184% * 704% ** 184% * cpe -0.3837 0.5858 -32% * -54% ** -32% * m2 gdp 0 .0 0 0 1 2616.7470 0.3321 39% 39% current 0.0145 0.7131 1% * 3% ** 1 % * capital -0.5076 0.5939 -40% * -64% ** -40% * * change due to change in dummy varaible from either 0 to 1 or 1 to 2 ** change due to change in dummy variable from 0 to 2 insignificant coefficients should be ignored because they will be misleading. The two tables are self-explanatory. Notice for instance that Table 4.9b shows that a one standard deviation increase in GDP increases trade by 1572%, i.e. almost 16 times, which is comparable to the value in Table 3.1b (16 times). Two countries that are centrally planned traded in the same year 704% (7 times) more than if none of them were centrally planned. Other than language (which is a very specific case because it more or less examines Arab-Israeli trade relations), distance has the highest negative effect on trade flows. A one standard deviation increase in distance reduces trade by 64%. Gov_ds affect trade flows but only to a small extent (17%). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 165 Table 4 .10b. Predictions based on Table 4.10 column 6 A=Coef. B=Std. Dev. A*B exp(A*B) - 1 exp(A) - 1 gdp 3.4251 2.2706 7.7771 238441% 238441% gdppc -1.0226 1.9621 -2.0064 -87% -87% distance -4.0793 0.6634 -2.7064 -93% -93% areas -1.0299 1.4235 -1.4661 -77% -77% 11 -1.1230 0.3756 -67% * -89% ** -67% * border 0.7483 0.0915 1 1 1 % 1 1 1 % language -23.5525 0.1075 - 1 0 0 % - 1 0 0 % regional (D) nation (D) colonizer 2.1171 0.3542 731% 731% colonial (D) erv -1.0706 0.0911 -0.0976 -9% -9% cu (D) gov_ds 0.7333 2.1003 1.5401 367% 367% cpy 0.8410 0.0673 132% * 438% ** 132% * cpe -1.2160 0.5743 -70% * -91% ** -70% * m2 gdp 0.0003 2485.8800 0.6632 94% 94% current -0.0838 0.7116 -8 % * -15% ** -8 % * capital -1.7159 0.5868 -82% * -97% ** -82% * * change due to change in dummy varaible from either 0 to 1 or 1 to 2 ** change due to change in dummy variable from 0 to 2 As for Table 4.10b a one standard deviation increase in Gov_ds increases trade by almost 4 times. As expected GDP is at the top of the list of factors positively affecting trade flows; while distance is at the top of the list of factors negatively affecting trade flows (as mentioned earlier, language here is a very specific case). Notice that a one standard deviation increase in GDP increases trade by 238,441%. This is a huge figure and it is even bigger than its corresponding value 99,676%, in Table 3.2b, indicating that perhaps the openness of Arab economies to trade is relatively higher than many other world countries such that only one standard deviation increase in the product of GDPs has this huge effect on total trade flows. Many Arab countries are low to medium income countries suggesting that may be this Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 166 low-medium income is what preventing them from more trade, and that as GDP increases, trade will increase even more. Table 4.11 presents the regression results from estimating the gravity model, with the individual governance components each at a time, for Arab-World total bilateral trade flows using OLS for the whole pooled sample (1985-2000), where the null values were assumed to be genuinely missing. This Table can be compared to column 6 in Table 4.9. Regarding current account restrictions, it experienced sign changes and further investigations revealed that when capital account restrictions was dropped, all current account restrictions coefficients were negative and mostly highly significant. Concerning the coefficients of the individual components of governance, they are all positive. VA, RQ, RL, and CC are highly significant, GE is significant at 10% while PS is not significant at any conventional level. Control of corruption has the highest magnitude, it is followed by RL and VA, which are then followed by the other governance components. Control of corruption might be more important in Arab-World trade than it is in intra-Arab trade as explained earlier because it might be the case that Arab countries, having very close CC scores and sharing the same language and having a high degree of cultural similarity, know how to deal with such corruption, while other less or highly corrupt non-Arab countries might be unaware of the appropriate practices when dealing with the Arab World’s corruption to a degree that makes them Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 4.11. Regression results from estimating the gravity model for Arab-World total bilateral trade for pooled data (1985-2000) using OLS, where each of the six governance components is used separately and where the null values in the dependent variable are assumed genuinely missing. Coef. (R.Std.Er) Coef. (R.Std.Er) Coef. (R,Std.Er) Coef. (R.Std.Er) Coef. (R.Std.Er) Coef. (R.Std.Er) gdp 1 29 *** Y 29 *** 1 29 *** 1 29 *** I 29 *** 1 30 *** (0 .0 2 ) (0 .0 2 ) (0 .0 2 ) (0 .0 2 ) (0 .0 2 ) (0 .0 2 ) gdppc -0.15 *** -0 17 *** -0 17 *** -0.17 *** -0.16 *** -0.16 *** (0.03) (0.03) (0.03) (0.03) (0.03) (0.03) distance -1.54 *** -1.53 *** -1.54 *** -1 54 *** -1 52 *** -1.53 *** (0.05) (0.05) (0.05) (0.05) (0.05) (0.05) erv -0.24 -0.27 -0.19 -0.03 -0 .2 1 -0.24 (0.44) (0.45) (0.45) (0.45) (0.45) (0.45) border 0 .0 0 -0.06 -0.06 -0.06 -0.03 -0.06 (0.37) (0.37) (0.37) (0.37) (0.37) (0.37) language -4 10 *** -4 00 *** -4 01 *** -4 Oi *** -4.03 *** _ < j 09 (0.67) (0.67) (0.67) (0.67) (0.67) (0 .6 6 ) colonial (D) (D) (D) (D) (D) (D) colonizer 0.65 *** 0.59 *** 0.58 *** 0.59 *** 0.60 *** 0.62 *** (0 .1 1) (0 . 1 1) (0 .1 1 ) (0 . 1 1) (0 . 1 1) (0 . 1 1) nation (D) (D) (D) (D) (D) (D) regional (D) (D) (D) (D) (D) (D) cu (D) (D) (D) (D) (D) (D) m2 gdp 0 .0 0 0 1 *** 0 .0 0 0 1 *** 0 .0 0 0 1 *** 0 .0 0 0 1 *** 0 .0 0 0 1 *** 0 .0 0 0 1 *** (0 .0 0 0 0 1 ) (0 .0 0 0 0 1 ) (0 .0 0 0 0 1 ) (0 .0 0 0 0 1 ) (0 .0 0 0 0 1 ) (0 .0 0 0 0 1 ) o \ - j Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 4.11. Regression results from estimating the gravity model for Arab-World total bilateral trade for pooled data (1985-2000) using OLS, where each of the six governance components is used separately and where the null values in the dependent variable are assumed genuinely missing (continued). Coef. Coef. Coef. Coef. Coef. Coef. (R.Std.Er) (R.Std.Er) (R.Std.Er) (RStd.Er) (R.Std.Er) (R.Std.Er) va_d 0.05 *** (0 .0 1 ) ps_d 0 .0 1 (0 .0 1 ) ge_d 0 .0 2 * (0 .0 1 ) rq_d 0.04 *** (0 .0 1 ) rl_d 0.05 *** (0 .0 2 ) cc_d 0 .1 2 *** (0 .0 2 ) cpy 0.94 ** 0.84 ** 0 .8 8 ** 0.97 ** 0.93 ** 0.94 ** (0.38) (0.39) (0.39) (0.39) (0.39) (0.38) cpe -0 37 *** -0.34 *** -0.34 *** -0.35 *** -0.35 *** — Q 40 (0.06) (0.06) (0.06) (0.06) (0.06) (0.06) areas -0.19 *** -0.19 *** -0 .2 0 *** -0 .2 0 *** -0.19 *** -0 .2 0 *** (0.03) (0.03) (0.03) (0.03) (0.03) (0.03) 11 -0.62 *** -0.62 *** -0.62 *** -0.63 *** -0.63 *** -0.58 *** (0 .1 0 ) (0 .1 0 ) (0 .1 0 ) (0 .1 0 ) (0 .1 0 ) (0 .1 0 ) current 0 .0 1 -0 .0 2 -0 .0 1 0 .0 1 0 .0 0 2 -0.07 (0.06) (0.06) (0.06) (0.06) (0.06) (0.06) capital -0.45 *** -0 44 *** -0.50 *** -0.50 *** -0.46 *** -0 39 *** (0.06) (0.06) (0.07) (0.07) (0.06) (0.06) _cons -28.82 *** -28.54 *** -28.31 *** -28.40 *** -28.56 *** -29.08 *** (0 .8 8 ) (0.87) (0 .8 8 ) (0.87) (0.87) (0 .8 8 ) os 00 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 4.11. Regression results from estimating the gravity model for Arab-World total bilateral trade for pooled data (1985-2000) using OLS, where each of the six governance components is used separately and where the null values in the dependent variable are assumed genuinely missing (continued). Coef. (R.Std.Er) Coef. (RStd.Er) Coef. (R.Std.Er) Coef. (R.Std.Er) Coef. (R.Std.Er) Coef. (RStd.Er) No.ofObs 4833 4833 4833 4833 4833 4833 R-squared 0.59 0.59 0.59 0.59 0.59 0.59 RMSE 2.23 2.23 2.23 2.23 2.23 2.23 Notes: 1) D stands for dropped. 2) Standard errors are beneath the coefficients. 3) ***, **, * denotes significance at 1, 5, and 10% respectively. as SO 170 prefer to trade with countries with high scores of control of corruption, where corruption is low, to avoid any confusions or misunderstandings. Rule of law has been a common important variable in Tables 3.5,4.6, and 4.11 perhaps rendering it the most important governance component for promoting trade. Other than these previously discussed variables, all remaining variables have the same signs and significance levels, with very minor exceptions, in both Tables (4.9 column 6 and 4.11). In general, the variables are quite robust to the inclusion of the different governance components. Tables 4.12 and 4.13 present the regression results from estimating the gravity equation separately for Arab-World energy and other trade flows, for the pooled sample not including year 2000 once where the null values were assumed to be genuinely missing, and once where some of the missing values were assumed to be genuinely missing while the others were assumed to be zeros when they have been zeros in all the other years. In both cases the zero values were excluded. GDP and colonizer, have positive coefficients that are highly significant in both Tables 4.12 and 4.13, while distance, LL, ERV, and capital account restrictions all have negative coefficients that are highly significant. As for CPY, it is insignificant at any conventional level in both levels. The coefficients of GDPPC, areas, and language are negative and highly significant for other trade flows as expected and as explained earlier in Table 4.9. They are, however, not significant at any conventional levels for energy trade flows. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 171 Table 4.12. Regression results from estimating the gravity model for Arab-World energy trade flows using OLS for pooled data (1985-1997). Null values assumed genuinely missing Coef. (RStd.Er) Some null values assumed genuinely missing while some others assumed zeros Coef. (RStd.Er) gdp 0.67 *** q 7 0 *** (0.08) (0.08) gdppc 0.13 0 .1 2 (0.09) (0.09) distance -0.58 *** -0.60 *** (0.17) (0.17) areas -0.16 -0.17 (0 .1 1) (0 .1 1) 11 -1.65 *** -1.60 *** (0.39) (0.39) border 2.23 *** 2.24 *** (0.85) (0.85) language -1.64 -1.67 (1.99) (1.99) regional (D) (D) nation (D) (D) colonizer 1 .0 0 *** 1 04 *** (0.38) (0.38) colonial (D) (D) erv -8.74 *** -8.69 *** (1.77) (1.78) cu (D) (D) gov_ds -0 .1 0 -0 .1 0 (0.06) (0.06) cpy 0.58 0.62 (1.98) (1.99) cpe Q 7Q *** 0.65 *** (0.25) (0.25) m2 gdp -0.0003 *** -0.0003 *** 0.00004 0.00004 current -0.16 -0.13 (0.24) (0.24) capital -0.82 *** -0.82 *** 0 .2 0 0 .2 0 _cons -10.48 *** -11.56 *** (2.91) (2.84) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 172 Table 4.12. Regression results from estimating the gravity model for Arab-World energy trade flows using OLS for pooled data (1985-1997) (continued). Some null values assumed genuinely missing while some others assumed zeros Null values assumed genuinely missing Coef. (RStd.Br) Coef. (RStd.Er) RMSE No. of Obs R-squared 850 0.24 3.21 857 0.24 3.21 Notes: 1) D stands for dropped. 2) Standard errors are beneath the coefficients. 3) ***, **, * denotes significance at 1, 5, and 10% respectively. On the other hand, Gov_ds shows a positive and highly significant effect on other trade flows but an insignificant effect on energy trade flows. Border in Table 4.13 shows an insignificant effect on other trade flows. On the other hand, it shows a positive and highly significant effect on energy trade flows. This is easily understood because non-Arab countries that want to trade in energy will prefer trading with an Arab country that is endowed with energy resources if they have shared borders than trading with another relatively far away energy-endowed country, especially that some of these countries have energy pipeline arrangements, like the oil pipeline between Iraq and Turkey. The effect of M2 GDP on other trade flows is positive and highly significant as expected, while its effect on energy trade is negative and highly significant. This probably indicates that Arab countries that trade in energy are themselves not financially developed. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 173 Table 4.13. Regression results from estimating the gravity model for Arab-World other trade flows using OLS for pooled data (1985-1997) Null values assumed genuinely missing Coef. CR.Std.Er) Some null values assumed genuinely missing while some others assumed zeros Coef. (RStd.Er) gdp 1 17 *** 1 23 *** (0.03) (0.03) gdppc -0.19 *** -0 .2 1 *** (0.03) (0.03) distance -1.33 *** -1 42 *** (0.06) (0.06) areas -0 .2 0 *** -0 .2 0 *** (0.04) (0.04) 11 -0.33 *** -0.36 *** (0 .1 1) (0 .1 2) border -0.18 -0.36 (0.47) (0.45) language -4.53 *** -4.12 *** (0.70) (0.77) regional CD ) (D) nation (D) (D) colonizer q 7 7 *** 0.67 *** (0 .1 2) (0 .1 2) colonial C D ) (D) erv -1.07 ** -1.06 ** (0.47) (0.50) cu (D) (D) gov_ds 0.08 *** 0.08 *** (0 .0 2 ) (0 .0 2 ) cpy 0.37 0.46 (0.41) (0.37) cpe -0.50 *** -0.52 *** (0.08) (0.08) m2 gdp 0 .0 0 0 2 *** 0 .0 0 0 2 *** 0 .0 0 0 0 2 0 .0 0 0 0 2 current 0.16 ** 0.14 ** (0.06) (0.07) capital -0.60 *** -0.59 *** 0.07 0.07 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 174 Table 4.13. Regression results from estimating the gravity model for Arab-World other trade flows using OLS for pooled data (1985-1997) (continued) Null values assumed genuinely missing Coef. (RStd.Er) Some null values assumed genuinely missing while some others assumed zeros Coef. (R.Std.Er) _cons -23.99 *** -25.54 *** (0.96) (0,98) No.ofObs 2793 3170 R-squared 0.61 0.60 RMSE 1.85 2 .0 2 Notes: 1) D stands for dropped. 2) Standard errors are beneath the coefficients. 3) ***, **, * denotes significance at 1, 5, and 10% respectively. CPE shows a negative and highly significant impact on other trade flows as expected. On the other hand, the CPE coefficients are positive and highly significant for energy trade flows. This could be due to the fact that Iraq and Russia orchestrated a lot of their energy trade flows with oil and natural gas pipelines. As for current account restrictions, its effect on energy trade flows is insignificant. For other trade flows, current account restrictions show positive and highly significant effect on trade flows. However, further investigations revealed that when capital account restrictions was dropped, the coefficients of current account restrictions became negative but insignificant, which indicates that current account restrictions has a negligible effect on Arab-World trade flows. As expected, R2 is lower and RMSE is higher in energy trade than they are in other trade, indicating that gravity considerations are a bit weaker for energy trade than they are for other trade, just as expected. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 175 4.2.3 Intra-Arab and Arab-World Results Comparison With the results of regression on energy trade flows set aside because it is of a very special character, Table 4.14 shows that GDP and Gov_ds are the only two variables that have positive and significant effects on intra-Arab trade flows and that have been quite robust to the different assumptions and specifications. Variables that have a negative and significant effect on intra-Arab trade flows, which have also been quite robust to the different assumptions and specifications are: GDPPC, distance, areas, CPE, and current account restrictions. Other than the above-mentioned variables, GCC, colonizer, colonial, M2GDP, and capital account restrictions have, too, been quite robust but in showing that they were largely insignificant. Border and AMU have been relatively robust, where the former showed changes only to the assumption of including zero trade flows, while the latter showed changes only to the other trade flows specification. As for ERV, it has been sensitive to both the assumptions used as well as to one of the specifications (that of other trade flows). Concerning governance’s individual components, rule of law and government effectiveness seem to be the most important. Concerning Arab-World trade, again setting energy trade flows aside, Table 4.15 shows that variables that have positive and significant effects on Arab-World trade and that have been quite robust to the different assumptions and specifications are: GDP, Colonizer, Gov_ds and M2GDP. While GDPPC, distance, areas, LL, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 176 Table 4.14. A summary for Intra-Arab trade regression results Variable/ Table no. 4.4 4.5 4.6 4.8 Result gdp -t-ve, sig -t-ve, sig -t-ve, sig R (+ve, sig) gdppc -ve, sig -ve, sig -ve, sig R (-ve, sig) distance -ve, sig -ve, sig -ve, sig R (-ve, sig) areas -ve, sig -ve, sig -ve, sig R (-ve, sig) 11 border insig -t-ve, sig insig RR (insig) language regional GCC AMU colonizer insig insig insig insig insig -ve, sig insig R (insig) RR (insig) R (insig) colonial insig insig insig R (insig) erv -ve, sig insig insig S cu gov_ds +ve, sig -t-ve, sig -t-ve, sig R (+ve, sig) cpy cpe -ve, sig -ve, sig -ve, sig R (-ve, sig) gini m2 gdp insig insig insig R (insig) current -ve, sig -ve, sig -ve, sig R (-ve, sig) capital insig insig insig R (insig) Notes: 1) Blank cells in column 4.6 indicate no significant changes due to including gov. components. 2) VA, PS, GE &RL are all positive and significant, while RQ & CC are insignificant. RL has the highest magnitude. language, CPE, and capital account restrictions have negative and significant effects on Arab-World trade and have been quite robust to the different assumptions and specifications. Other than those mentioned variables, border, and current account restrictions are robust in showing insignificance, while CPY and ERV are sensitive to the different assumptions and the different specifications. Concerning governance’s individual components, control of corruption and rule of law seem to be the most important components. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 177 Table 4.15. A summary for Arab-World trade regression results Variable/ Table no. 4.9 4.10 4.11 4.13 Result gdp +ve, sig -t-ve, sig +ve, sig R (+ve, sig) gdppc -ve, sig -ve, sig -ve, sig R (-ve, sig) distance -ve, sig -ve, sig -ve, sig R (-ve, sig) areas -ve, sig -ve, sig -ve, sig R (-ve, sig) 1 1 -ve, sig -ve, sig -ve, sig R (-ve, sig) border insig insig insig R (insig) language -ve, sig -ve, sig -ve, sig R (-ve, sig) regional colonizer +ve, sig -t-ve, sig -t-ve, sig R (+ve, sig) colonial erv mixed S -ve, sig S cu gov_ds +ve, sig +ve, sig -t-ve, sig R (+ve, sig) cpy +ve, sig insig insig S cpe -ve, sig -ve, sig -ve, sig R (-ve, sig) gini m2 gdp -t-ve, sig -t-ve, sig -t-ve, sig R (+ve, sig) current insig insig insig R (insig) capital -ve, sig -ve, sig -ve, sig R (-ve, sig) Notes: 1) Blank cells in column 4.11 indicate no significant changes due to including gov. components. 2) VA, GE, RQ, RL & CC are positive and significant. PS not significant. CC has highest magnitude followed by RL & VA The following chapter is the concluding chapter, where summary of all previous results, policy implications, and further research ideas will be provided. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 178 5. CONCLUSION This chapter is composed of three sections. Section one will present a summary of this study. Section two will discuss some policy implications, and section three will offer some ideas regarding further research. 5.1 Summary Good governance has multiple bases and aspects. It describes a situation in which governments hear and are responsive to citizens’ concerns; internal conflicts are minimal; the government is able to perform its declared programs and is held accountable for failures in design or implementation; bureaucratic institutions are strong and policies need not be revised each time governments change; the risk to investment is minimal; the law is strong, impartial and widely observed; and in which corruption is controlled. While given such characteristics, good governance should be considered a “good” in and of itself, its relationship to economic and social development, as mentioned in the introduction, is also clear. Kaufmann, Kraay, and Zoido-Lobaton (1999b) estimated for 150+ countries a series of very parsimonious regressions of the log-level of per capita income, the infant mortality rate, and the adult literacy rate one at a time on each of: voice and accountability, political stability, government effectiveness, regulatory quality, rule of law, and control of corruption. They Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 179 corrected for reverse causation, omitted variable bias, and measurement error in the governance variables and still found that good governance mattered a great deal for economic outcomes. Specifically, they found that a one-standard deviation increase in any of the six individual components of governance caused per capita incomes (infant mortality) to increase (decrease) between two-and-a-half and four-fold, and literacy to increase by 15 to 25 percent. The World Bank’s report on governance (which investigated the impact in MENA countries of an integrated package of policy realignments that improves the business and the investment climate for the private sector and promotes integration with the world economy) using similar international evidence of well performing countries, suggested that enhancing the institutions of accountability and public administration in MENA countries could boost per capita output growth by 0.8 and 1.3 percent a year (World Bank 2003b: 10). Given the importance of good governance and given that its impact on the level of trade had rarely been examined, the main purpose of this study was to utilize a gravity-type model to investigate the effect of domestic governance on trade flows, with special emphasis on intra-Arab trade. Chapter 1 provided an introduction to the paper. Chapter 2 provided evidence for the theoretical foundations of the gravity model and underlined its empirical success. The gravity equation is an empirical model that in its simplest form models the flow of trade between a pair of countries as being proportional to their national incomes and inversely proportional to the distance between them. Besides physical Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 180 distance, several other resistance-enhancement factors (some of the commonly used and others rarely used in existing research) have been used. The gravity model has a remarkably consistent history of success as an empirical tool; it is considered the workhorse for empirical studies of trade patterns. The gravity model is not a theoretical orphan; theoretical aspirants have claimed the singular empirical success of the gravity model. The question of which theoretical model best describes the empirical findings of the gravity model is a matter of dispute and others like Baier and Bergstrand (2001) showed that these theoretical foundations are complementary to one another and not substitutes. Chapter 2 also presented a comprehensive literature review of some of the important studies that have applied the gravity model to bilateral trade flows. These studies have examined the effect on bilateral trade flows of: RTAs, exchange rate volatility, exchange rate regimes, currency unions, current and capital account restrictions, tariffs, income growth and convergence, alliances, democracy, conflict and cooperation, foreign policy orientation, defense pacts, immigration, infrastructure, intellectual property rights, common border, and common language. In addition, chapter 2 reviewed the less than handful studies that utilized the gravity model to investigate the impact of a number of variables on intra-Arab trade flows. Lastly, chapter 2 highlighted the purpose of the study and the contribution to the literature summarized in: investigating the effects of domestic governance, as well as some other new and other rarely used variables, including CPY, CPE, Gini, M2GDP, and current and capital account restrictions, on bilateral trade flows; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 181 including data for 2000 - a year no previous gravity study has included; and placing special emphasis on intra-Arab trade flows. Chapter 3 presented the model in great detail along with the sources of the data. It also discussed the expected signs of the coefficients of the variables used in the model based on theoretical foundations and previous empirical findings. The signs of the coefficients of some variables were expected to be positive and thus to promote trade. These are: GDP, border, language, regional, nation, colonizer, colonial, CU, Gov_ds, and M2GDP. Other variables like distance, areas, LL, CPE, Gini, and current and capital account restrictions were expected to have an impeding effect on trade flows and thus to have negative coefficients. The expected signs of the coefficients of the remaining variables were ambiguous because there are theoretical and/or previous empirical findings that support both. These variables are: GDPPC, ERV, and CPY. Governance, as used in this study, is composed of six individual components, four of which were expected to have positive coefficients namely: political stability, government effectiveness, regulatory quality, and rule of law, while for the other two - voice and accountability and control of corruption - the expected signs were ambiguous. Since governance is the deviated standardized weighted sum of the six components, and since the six components have almost equal weights, it was expected that the four positive components would outweigh the two ambiguous ones whatever their signs, resulting in an overall positive coefficient as mentioned above. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 182 Chapter 3 then displayed the results with their interpretations from estimating the gravity model on World bilateral trade flows under a number of assumptions and using different estimation techniques as well as employing different specifications. The aim of these assumptions, techniques, and specifications was to examine the sensitivity/robustness of the results. As clear from Table 5.1, column 1 (which is also last column in Table 3.8), some variables have been persistently robust to the different assumptions employed (missing and zero null values), to the different estimation techniques used (OLS and Tobit), and to the different specifications modeled (total trade flows, individual governance components, and other trade flows6 4 ). These variables are: GDP, language, colonizer, CU, and Gov_ds, all of which showed mostly positive effects on trade flows and were mostly significant at the 1% level. Other parameters that have also been persistently robust include distance, areas, LL, ERV, and capital account restrictions, where these showed mostly negative effects on trade flows and were mostly significant at the 1% level. There are eight variables that in general have been relatively robust. These are border, regional, colonial, M2 GDP (showing mostly positive and significant effects), CPY, Gini (showing mostly insignificant effects), CPE, and current account restrictions (showing mostly negative and significant effects), which showed changes due to only one sensitivity analysis (assumption, technique, or specification). 64 Lagged dependent variable and Energy trade flows are not included because they are very special cases. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 183 Table 5.1. Summary & comparison for World, Intra-Arab and Arab-World trade regression results World Flows Intra-Arab Arab-World 3.1 4.4 4.9 gdp R (+ve, sig) R (+ve, sig) R (+ve, sig) 1 09 *** 1 22 *** 1.29 *** (0.01) (0.10) (0.02) gdppc S R (-ve, sig) R (-ve, sig) 0.02 * -0.54 *** -0.16 *** (0.01) (0.11) (0.03) distance R (-ve, sig) R (-ve, sig) R (-ve, sig) -1.38 *** -1 73 *** -1.53 *** (0.03) (0.17) (0.05) areas R (-ve, sig) R (-ve, sig) R (-ve, sig) -0.14 *** -0.46 *** -0 19 *** (0.02) (0.12) (0.03) 1 1 R (-ve, sig) R (-ve, sig) -0.31 *** (D) -0.63 *** (0.04) (0.10) border RR (+ve, sig) RR (insig) R (insig) 0.76 *** 0.30 0.01 (0.11) (0.31) (0.37) language R (+ve, sig) R (-ve, sig) 0.61 *** (D) -4.01 *** (0.05) (0.67) regional RR (+ve, sig) 1.34 *** (D) (D) (0.13) GCC R (insig) AMU RR (insig) colonizer R (+ve, sig) R (insig) R (+ve, sig) 0.59 *** 0.17 0.62 *** (0.07) (0.23) (0.11) colonial RR (+ve, sig) R (insig) 0.63 * 0.01 (D) (0.34) (0.57) erv R (-ve, sig) S S -0.06 -2.25 -0.09 (0.23) (1.54) (0.45) cu R (+ve, sig) 1.54 *** (D) (D) (0.24) gov_ds R (+ve, sig) R (+ve, sig) R (+ve, sig) 0,09 *** 0.09 ** 0.08 *** (0.01) (0.04) (0.02) cpy RR (insig) S 0.14 (D) 1.04 *** (0.18) (0.39) cpe RR (-ve, sig) R (-ve, sig) R (-ve, sig) -0.06 -0.81 *** -0.38 *** (0.04) (0.19) (0.06) gini RR (insig) -0.003 * (0.00) m2gdp RR (+ve, sig) R (insig) R (+ve, sig) 0.0001 *** -0.0001 0.0001 *** (0.000) (0.0000) (0.00001) current RR (-ve, sig) R (-ve, sig) R (insig) -0.01 -0.39 ** 0.01 (0.03) (0.15) (0.06) capital R (-ve, sig) R (insig) R (-ve, sig) -0.30 *** -0.14 -0.51 *** (0.03) (0.13) (0.06) Notes: 1) D stands for dropped. 2) Standard errors are beneath the coefficients. 3) ***, **, * denotes significance at 1, 5, and 10% respectively. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 184 The only variable that displayed the highest sensitivity in terms of showing different results in two or more of the three sensitivity analyses - different assumptions, techniques, and specifications employed - was GDPPC. All of governance’s six components were positively signed and highly significant, with rule of law and control of corruption having the highest magnitudes, indicating that they might be the most important components. Of the seven new variables that represent this study’s contribution to the literature, Gov_ds and capital account restrictions seem to be the most important - significant - and the most robust as far as World trade flows is concerned. Chapter 4 aimed to investigate the effect of domestic governance on intra-Arab trade flows, and so the gravity model was estimated for intra-Arab trade flows but after excluding the Gini variable. The gravity model was also estimated for Arab- World trade flows to check if trade impediments faced by Arab countries in their dealings with each other are greater than those facing them when they are trading with the rest of the world. In other words, to investigate whether intra-Arab trade has specific features that characterize trade only between Arab countries and that perhaps act as trade impediments. The main objective of studying intra-Arab trade and factors affecting it was to evaluate the extent of and prospects for Arab economic integration where, whatever the form or stage of a regional arrangement/integration, increased intra-regional trade ranks high among the priorities and is the yardstick against which the success or failure of regionalism is measured. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 185 The importance of creating an FT A among Arab countries is attributable to a number of factors. First, after the proliferation of FTAs worldwide since the 1980s, Arab countries feared that their access to world markets might be curtailed considerably and thus were led to consider the regional option, which might be their only way to meet the challenges posed by developments of FTAs in the Americas, Europe, and Asia. Second, many analysts believe that greater integration among the Arab countries would bring important benefits, including greater efficiency, deepening of capital markets, fostering FDI, and greater product variety. Greater product variety is expected to enhance intra-industry trade by weakening the monopoly power of local producers. Third, many industries in the Arab countries could not be economically viable because domestic sales are insufficient for economies of scale to be realized. Regional economic integration could bring about major economic gains through the realization of such economies of scale, which are otherwise unattainable because of the small size of the Arab countries’ national markets. Fourth, the Arab FT A and other Arab bilateral agreements can reduce the “hub-and-spokes” effects that are likely to take place as a result of the EU-Med agreements. Fifth, Arab economic development, which could be achieved through economic integration as mentioned above is crucial to the attainment of peace in the region. Miniesy, Nugent, and Yousef (2004) argued that a full fledged FTA between Arab countries is expected to more than double the levels of trade that would be expected on the basis of gravity model considerations, thus fostering economic development. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 8 6 Chapter 4 then discussed Arab regionalism in some detail. It covered the history of and reasons behind past integration failures; prospects for Arab FTA; GAFTA; recent trends in intra-Arab trade; reasons for the low intensity of intra-Arab trade/regionalism; and empirical studies investigating whether intra-Arab trade is low. Chapter 4 then presented the results of the estimations carried. Whether in intra-Arab trade or Arab-World trade, regressions run on energy trade flows seemed to dramatically affect the signs and significance of the variables. As discussed earlier, energy trade is a very special case and thus results’ summaries are given for all estimations except for those of energy trade flows. Concerning intra-Arab trade flows, results of regressions on energy trade flows set aside, Table 5.1, column 2 (which is also the last column in Table 4.14) shows that GDP and Gov_ds are the only two variables that have positive and significant effects on intra-Arab trade flows and that have been quite robust to the different assumptions and specifications. Variables that have a negative and significant effect on intra-Arab trade flows, which have also been quite robust to the different assumptions and specifications are: GDPPC, distance, areas, CPE, and current account restrictions. Other than the above-mentioned variables, GCC, colonizer, colonial, M2GDP, and capital account restrictions have, too, been quite robust but in showing that they were largely insignificant. Border and AMU have been relatively robust, where the former showed changes only to the assumption of including zero trade flows, while the latter showed changes only to the other trade flows’ specification. As for ERV, it has been Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 187 sensitive to both the assumptions used as well as to one of the specifications (that of other trade flows). Concerning governance’s individual components, the coefficients of voice and accountability, political stability, government effectiveness and the rule of law were all positively signed and significant, with the last two being the most important and thus had the largest magnitudes. On the other hand, regulatory quality and control of corruption were insignificant. Of the seven new variables that represent this study’s contribution to the literature, Gov_ds, CPE and current account restrictions seem to be the most significant and the most robust as far as intra-Arab trade is concerned. Concerning Arab-World trade, again setting energy trade flows aside, Table 5.1, column 3, (which is also the last column in Table 4.15) shows that variables that have positive and significant effects on Arab-World trade and that have been quite robust to the different assumptions and specifications are: GDP, Colonizer, Gov_ds and M2GDP. While GDPPC, distance, areas, LL, language, CPE, and capital account restrictions have negative and significant effects on Arab-World trade and have been quite robust to the different assumptions and specifications. Besides these variables, border, and current account restrictions are robust in showing insignificance, while CPY and ERV are sensitive to the different assumptions and the different specifications. Concerning governance’s individual components, the coefficients of all six components were positively signed and significant except political stability, which Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 188 was insignificant. The coefficients of control of corruption and rule of law had the largest magnitudes and thus seem to be the most important components. Of the seven new variables that represent this study’s contribution to the literature, Gov_ds, M2GDP, CPE, and capital account restrictions seem to be the most significant and the most robust as far as Arab-World trade is concerned. Comparing intra-Arab trade results to Arab-World trade results and to World trade results from chapter 3, Table 5.1 shows that of the seven newly introduced variables that represent this study’s contribution to the literature, Gov_ds is the only common variable that is the most significant and the most robust and that has a positive effect on trade flows. CPE seems to be a significant impeding trade factor in Arab trade whether with other Arab countries or with the rest of the world but not for World trade in general. Current account restrictions seems to play an important role in reducing trade flows between Arab countries may be because some Arab countries have no current account restrictions, while still others have not yet liberalized restrictions on current transfers and payments.6 5 However, it is not robustly significant when trade is between the Arab countries and the rest of the world or World trade in general as shown in chapter 3 because most of the countries in the world have no current account restrictions or have recently liberalized current accounts from restrictions. Capital account restrictions on the other hand is an 65 With the exception of Djibouti, the GCC countries, and Lebanon, the rest of the Arab countries have current account restrictions. However, a number of them have lately liberalized these restrictions such as Egypt, Jordan, and Yemen. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 189 impeding factor for Arab-World trade and World trade because many countries in the world have not liberalized capital account restrictions. Concerning M2GDP, it does not seem to robustly affect intra-Arab trade. It actually proved insignificant, may be because many Arab countries are to some extent similarly not financially developed. To explore whether certain variables played a more effective role in intra-Arab trade as opposed to World and Arab-World trade flows, the magnitudes of the coefficients of the variables that significantly affected intra-Arab trade flows are compared to those of World and Arab-World trade flows. The magnitudes of GDP and Gov_ds for intra-Arab trade seem to be quite close to those of World and Arab-World trade flows, suggesting their almost similar effect on all trade flows. Regarding GDPPC, its effect on intra-Arab and Arab-World trade flows is negative while its effect on World trade flows is positive, perhaps indicating that the impact on trade of population is negative as far as Arab countries are concerned, meaning that countries with smaller populations traded less with each other or that countries with larger populations traded more with each other as explained earlier. Concerning distance and areas, their magnitudes for intra-Arab trade seem to be significantly bigger than those for World and Arab-World trade flows, indicating that transportation costs between Arab countries seem to have a bigger negative effect than on World and Arab-World trade. This could be justified, as explained earlier, by the difficult geographic terrain between some Arab countries, the lack of transportation networks, the presence of minimal paved roads and rail links across Arab countries. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 190 Coming to CPE, its magnitude for intra-Arab trade is significantly greater than for Arab-World trade, which is also significantly greater than for World trade, perhaps suggesting that in Arab countries, the centrally planned economic systems, once followed, caused structural problems and damages that impeded trade more than in other centrally planned countries. Current account restrictions seemed to play an impeding effect only between Arab countries because most of them still have restrictions, while these restrictions were abolished in many countries in the rest of the world. In short, the variables that generally had positive effects on trade flows had relatively similar effects on intra-Arab trade as they did for World and Arab-World trade. On the other hand, the variables that generally had negative effects on trade flows had more deleterious effects on intra-Arab trade than they did on World and Arab-World trade. 5.2 Policy Implications According to the findings of chapter 4, intra-Arab trade can be fostered6 6 through either increases in GDP and Gov_ds, or reductions in GDPPC, distance, areas, CPE, and current account restrictions. Since this study is primarily concerned with the effect of governance on intra-Arab trade, governance will be the main focus of policy implications. However, a few, though trivial, points are mentioned below. 6 6 This is irrespective of its impact on welfare, which is not the issue here. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 191 Concerning GDP, increasing it is not as easy as it sounds. After all, this is what specifically developing countries have been trying to achieve for at least as long as the term ‘developing’ has been attached to them. Reduction in distance and areas of course do not refer to their actual physical reduction but rather to lowering transportation costs, for which they are both proxies. Concerning lowering GDPPC, this of course refers to the population component and not the GDP component where GDPPC can be reduced through increasing population. This applies only to a number of oil-rich GCC Arab countries and there are policies already implemented to pursue this population-increase goal. As for CPE, what is in the past is in the past, nothing can be done to reverse the fact that a country was at one point in time centrally planned. However, the consequences of being centrally planned can be reversed through for instance, lowering protectionist measures, making trade policies more transparent, and reducing governments involvements. Regarding current account restrictions, liberalizing these controls would be an effective way to foster intra-Arab trade. As discussed before, current account restrictions has a negligible effect on Arab-World and World trade flows because it has largely been liberalized worldwide as Tamirisa (1999) argued. Concerning governance, the concept of good governance evolved from a set of fairly specific policy recommendations for structural adjustment and economic reform to a broad range of related ideas and sets of policies incorporating the following: economic liberalization and the creation of market friendly environments; transparency and accountability regarding economic and political decision making; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 192 political liberalization, in particular democratic reforms; rule of law and the elimination of corruption; the promotion of civil society; the initiation of fundamental human rights guarantees, especially concerning political rights such as freedom of expression, freedom of assembly and freedom from arbitrary imprisonment; and the adoption of policies to protect long term interests like education, health, and the environment (Najem 2003). For the Arab countries, the main governance challenges are derived from weaknesses in inclusiveness and public accountability. Inclusiveness reflects the notion that everyone who has a stake in governance processes and wishes to participate in them can do so on equal basis with all others. Accountability, as discussed earlier, draws on the principal of proper representation, where those chosen to act in the name of the people are answerable to the people (World Bank 2003b: 8). Weaknesses in inclusiveness are reflected in: limited public services in rural areas; high levels of illiteracy, especially among peoples of low and middle-income countries; gender inequalities in voice and participation in society as well as different treatment under the law; nepotism, patronage, or money determining who gets public services and who does not, as well as who gets access to lucrative business opportunities and who does not (World Bank 2003b: 8). Weaknesses in accountability are reflected in limited and reluctant transparency and contestability. No Arab country ensures citizens the right to access government information, and freedom of the press is closely monitored and circumscribed in most countries. Contestability - the ability to debate, question Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 193 choice, and have competition among alternative representatives and policies - is also rare in most Arab countries (World Bank 2003b: 9). The weaknesses in inclusiveness and public accountability in the Arab world are attributed to the Arab countries’ historical model of development, which was based on state-led development and social policies designed for redistribution and equity. This development strategy paid large dividends in the beginning, but had significant costs. Early dividends included: fast growth, large government revenues from oil fVI which supported vast welfare and social service systems, large education gains, improved health indicators, and dramatically falling poverty rates. Underlying this state-led development model, were some significant costs including: stifled demands for accountability, where limited taxation and mechanisms for redistribution (both outcomes of oil revenues) muted most demands for accountability; highly regulated labor markets; swelling public sectors; and heavily protected and inward looking economies (World Bank 2003b: 11-16). Flawed from the start, the policies that were put in place could not meet the challenges that came later. The state-led development model came under examination starting in the 1980s as the oil price collapse led Arab countries to suffer from macroeconomic imbalances, declining growth, and severe employment challenges. 6 7 This applies to both oil producing and non-oil producing Arab countries. For oil producers, the argument is straightforward: oil revenues permitted the creation of welfare systems that redistributed the oil wealth to citizens. In non-oil producers, remittance incomes (form people working in oil- producing countries) boosted household consumption. Moreover, loans, grants, and other forms of assistance from oil to non-oil producers increased government revenues and sustained redistributive commitments. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 194 Half-hearted reform efforts were put in place to try to secure economic recovery. Such reforms led to temporary macroeconomic stabilization, however other problems remained, as growth rates lagged, private investment failed to materialize, and unemployment doubled (World Bank 2003b: 17-21). For at least two decades, several factors have intensified the need for reforms. These include increasing labor market pressures, rising expectations for improved standards of living, lagging public service delivery as compared to the rest of the world, and rising competition for global markets. Meeting all these challenges demands a number of economic reforms, which will require fundamental changes in the role of the state through a broad agenda of governance reforms (World Bank 2003b: 25-28). As mentioned in the introduction, good governance complements and reinforces other economic reforms because it improves the capacity and incentives within governments and fosters a larger role for civil society in governance. Good governance, though cannot guarantee optimal economic policies, is nonetheless essential to guard against persistently poor policies and to ensure that the policies set to meet economic challenges are implemented faithfully. Indeed, good governance is essential for balanced, sustained growth. Just as weak governance has contributed to low growth rates in Arab countries, good governance may be expected to spur growth. In one of the World Bank’s studies, it was found that, if MENA had matched the average quality of administration in the public sector for a number of good-performing Southeast Asian countries (Indonesia, Malaysia, the Philippines, Singapore, and Thailand), its growth rates would have been Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 195 higher by around one percent point yearly (World Bank 2003a: 8-9). Poor governance has contributed to weak growth through its shackling of the business environment, and its limiting of the quality of public goods and services. By improving inclusiveness and accountability, good governance can help growth in three ways: by diminishing the scope for persistently arbitrary or distorted policies, by enhancing bureaucratic performance and thus decreasing the uncertainties and costs of doing business, and by improving the delivery of public services for businesses to be productive (World Bank 2003a: 10-12 & 77-78). But if reforming governance is that critical why hasn’t it already been initiated? Such a reform process is long, uncharted, and difficult. All Arab states are authoritarian (if to varying degrees) and hence are unwilling or at best reluctantly willing when under outside pressure to implement governance reforms. This is because such changes reduce the power in the hands of the decision makers and the groups that support them. A number of other factors have played a particularly pernicious role in perpetuating poor governance in the Arab world: geopolitics, conflict, and oil. The Arab world’s strategic location at the intersection of three continents has made it a target of great foreign powers, whose interest has been domination and control, thus setting a bad example for governance and actively depressing voice and accountability within the region. Moreover, even today, as external forces call for better governance in the region, some nonetheless find a strong convergence of interests with authoritarian regimes, which provide secure access to strategic assets and which offer Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 196 convenient alliances. Such strong foreign involvement has often restrained the development of more accountable and inclusive governance systems (World Bank 2003a: 66-67). Regarding the threat of conflict, over the past decades almost every Arab country has been directly involved in some form of interstate conflict of varying intensity, from full-scale war to the threat of conflict. Over the years both factors led to the concentration of power in the hands of the (often militarily-based) executive, hence strengthening repressive governance and building up coercive organizations (World Bank 2003a: 68). As for oil, recent empirical studies have revealed that after controlling for incomes and population size, oil rents had a strong explanatory power in accounting for weaker governance in MENA, while variables measuring Islamic and Arab culture did not. This finding could be explained by the fact that substantial revenues from natural resources relieved a government from the need to tax and allowed it to redistribute a significant share of its oil revenue. Both factors helped mute demands for accountability (World Bank 2003a: 68). Given the many challenges complicating the implementation of good governance measures, the most important of which is the unwillingness of most Arab leaderships to undertake these measures, four different sources of external and internal pressures might play an important role, as they have previously proven effective in other developing countries. First, external pressure could come from international aid donors, when they establish a more or less - well-defined set of good governance Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 197 reforms as a precondition for financial assistance. However, this kind of pressure might work only with non oil-rich countries. A second factor could be international trade, since organizations such as the WTO require a commitment to good governance as a prerequisite membership. Concerning sources of internal pressure, these could come from: a crisis of legitimacy, which many Arab authorities are facing due to the spread of chronic poverty and other social problems that have resulted from worldwide economic conditions; or from the emergence or activation of civil society elements (students, domestic NGOs, scholars, journalists, etc) (Najem 2003). The last kind of pressure is perhaps the one that should be fostered, since it would come from within and would thus meet less opposition from Arab societies as opposed to the external pressures coming from outside forces. Many developing countries were able to have democratic transitions and to even consolidate their democracies through the efforts of their civil societies as in the Philippines, South Korea, Eastern European as well as some African countries (Diamond 1999). Other countries and international NGOs promoting inclusiveness and accountability might also play a role in encouraging Arab countries to follow through with governance reforms, if the latter would be willing to accept help, through grants and other incentives such as sharing expertise, and providing training for potential democratic proponent cadres. But what should Arab countries or their civil societies do to improve governance? According to the World Bank (2003a), the program to enhance governance could be elaborated along five pathways: Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 198 1) Measures to enhance inclusiveness, such as mandating universal suffrage for all elected posts, encouraging broad-based civil-society organizations, and reducing discrimination in laws and regulations. 2) National actions to strengthen external accountability, such as encouraging independent and responsible media, inviting public debate on policies by representative civil-society groups, and mandating greater freedom of information and public disclosure of government operations. 3) Local actions to strengthen external accountability, such as introducing feedback mechanisms from clients to providers and publishing results, and increasing competition among public service agents - and with private providers. 4) National checks and balances to strengthen internal accountability, such as increasing oversight authority and capability of parliaments over the executive, ensuring greater independence of the judiciary, enhancing professional capacity of parliaments and the judiciary, and empowering other independent oversight agencies and mandating reviews by them. 5) Administrative reforms to improve internal accountability, such as strengthening the resources and capacity of local agencies to design, adapt, and deliver public services, ensuring independence of regulatory agencies, and reforming the civil service to improve its service orientation and professional competence (World Bank 2003a: 18-23). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 199 Governance has six components as discussed earlier with policies specifically addressing how to augment some of them being more difficult to set and implement than others. However, their reform can come under the big umbrella of governance reforms just discussed. Still, we can try to address each one of them separately. Policies aimed at augmenting voice and accountability and political stability were the main focus of the governance reforms extensively discussed above. Still a number of crucial points must be highlighted. First, if put in place, inclusiveness and accountability can take a long time to be effectively implemented, especially since the political culture itself needs to be changed. Embedded within the political culture in most of the Arab world is the idea of the ‘one man show’, whereby, a president/King/Prince is the final decision maker. To attain political culture change, civil society organizations should exert a substantial effort, a major part of which is not only increasing literacy rates in Arab societies but also spreading political awareness in terms of both rights and responsibilities. Second, there is no one size of democracy that fits all. Democracy a la West might not always work. Each country has specific conditions and the democracy promoted must be fitted to each country’s conditions. In order to avoid the mistakes encountered with the one sized structural adjustment and economic reform, outside pressures deciding to get involved in Arab countries’ governance reforms - mainly as mentors - must have this idea vividly clear in their minds. Third, other countries should not dictate Arab countries’ political reforms. The USA’s war on Afghanistan has not brought about democracy as it claimed, nor will its Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 200 war on Iraq. In both cases, the USA opened the gate for civil warring among different ethnic groups. Regarding government effectiveness, reforms here should concentrate on two things. First, to involve citizens as much as possible in the setting of policies so that citizens not only see to the implementation of these policies but also support their governments even when these policies impose harsh conditions, such as those accompanying structural adjustment policies. This could be done through spreading political awareness among the citizens and perhaps allowing citizens to vote on important policies, and above all, by increasing the literacy rate. Second, policies should be set with the future in mind, meaning that policies set should be far-sighted. They should also be transparent. Specific institutional bodies should be formed to see that these policies are effectively implemented and to make sure that these policies are not just changed solely because of changes in government, but rather, only if genuinely needed. As for regulatory quality, it could be augmented by building trust between authorities and investors, be they local or foreign. This involves transparent policies as well as their fair enforcement. Investors must know what to expect and must be assured that no matter what happens, their problems will not be handled arbitrarily. To attract investments, especially FDI, which as many scholars suggested fosters international trade, policies set should involve clear clauses about contract enforcement, profits repatriation and payment delays, perhaps through rules enforcing the first, allowing a part of the second, and limiting the third. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 201 Coming to rule of law and control of corruption, they are the most important governance variables affecting not only trade flows but also FDI (thus affecting also regulatory quality discussed above) as shown in chapter 3 and 4. As Powell put it: Markets need good government, functioning institutions, transparent policy making and above all the rule of law. Capital is coward; money flees uncertainty and corruption. To entice capital in and then keep it in, governments must recognize private property rights, deeds of trust, and the sanctity of contract, and they must enforce these rights transparently and fairly [...] [if] I got a dollar bill [, and] I want to invest it somewhere where it is going to be safe [...] I am not going to send it to some place where there is no rule of law, where if I had a problem [...] there is no court I can take it to, there is no justice. I am not going to invest it in a place where corruption is rampant (Collin Powell in Ridgway and Talib 2003). Since efficient trade necessitates reliable enforcement of the agreements governing the exchange, rule of law reforms need to be carried out. There are three types of rule of law reforms. Type one focuses on rewriting laws, particularly commercial and criminal law. Type two involves strengthening the law-related institutions through increases in salaries of judges and court staff; training police, prosecutors, public defenders and prison staff; enhancing legal education; and strengthening legislation and local government. Type three aims at increasing government’s compliance with the law (Carothers 1998). These three types of reforms are needed in the Arab world, but they vary in terms of difficulty of implementation with rewriting laws being relatively the easiest compared to the ‘heavy lifting’ of institutional reforms and to the hardest type of reform - subjecting leaders to the rule of law. Funding for rule of law programs, though modest, has become the fastest growing type of international aid. Those Arab countries that decide to go through with Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 202 these reforms can be assured of external support. They can also benefit from the expertise of other governments, foundations, etc. But it is important to note again that there are no one size fits all solutions - fundamental differences between legal cultures must be well noted and taken into consideration. Regarding corruption, besides being unethical, it can hinder economic development, as many scholars have argued and as the level of its control has proved in this study to affect international trade flows. A number of scholars have gone into lengths in designing anti-corruption policies such as Rose-Ackerman (1978), Klitgaard (1988), and Bardhan (1997). Bardhan, for instance, mentioned that the importance of moral exhortations in anti-corruption campaigns must not be minimized but focused on the incentive structures that might induce even opportunists to forego corrupt practices. He argued that regulations and bureaucratic allocation of scarce public resources breed corruption and the immediate task would be to get rid of them through, for instance, legalizing activity that was formerly prohibited or controlled. Klitgaard (1988) noted, for instance, that when Singapore allowed more imported products to enter duty free, corruption in the customs area fell. There are some regulations, however, that serve social objectives as well and that cannot be simply eliminated. In such cases there will be a trade-off between these social objectives and the reduction of corruption through deregulation. In this case, reducing the monopoly power of the bureaucrat herself/himself might diminish bureaucratic corruption. As Rose-Ackerman (1978) suggested, instead of allowing Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 203 each official a clearly defined sphere of influence over which he has monopoly control, officials should be given competing jurisdictions so that when a client is not well-served by one official he can go to another. Bribes will be driven to zero because collusion among several officials is relatively difficult. But competitive pressure might increase theft from the government at the same time as it reduces bribes, as Shleifer and Vishney (1993) argued. In that case, intensive monitoring and auditing should accompany competition in the provision of government services to prevent theft. Rose-Ackerman (1994) has suggested that multiple officials with overlapping jurisdictions may also help in such cases. Still, in some cases, bureaucratic competition through overlapping jurisdictions is not feasible; as in the case of large defense contracts or expensive single items like aircraft. Bardhan (1997) argued that periodic anticorruption campaigns might be effective on the condition that they are both credible and sustained. Short-lived campaigns together with repeated amnesties to offenders create cynicism about the next campaign and give the wrong signals. Thus Bardhan highlighted the importance of institutionalizing various kinds of accountability mechanisms.6 8 In short, all the above could be summarized in the need for inclusiveness, external and internal accountability, transparency and contestability. 6 8 These can be found in Bardhan (1997:1338). Note that controlling corruption is country-specific and even case-specific and it is not the intention of this study to investigate it in details but only to shed light on some ways that can reduce corruption. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 204 5.3 Further Research Regression results for intra-Arab trade flows were all carried out using the OLS estimation technique. Further research might utilize a number of alternative estimating techniques, such as fixed effects, random effects, and others, to explore the sensitivity or robustness of the findings, especially those of governance, with respect to the techniques used. Based on the regression results obtained from estimating the gravity model for intra-Arab trade flows, it is clear that with the exception of the governance variable, all the newly introduced variables have adverse effects on intra-Arab trade flows, although some are more significant than others. This would perhaps cause the predicted intra-Arab trade values to be lower than their corresponding values in other studies that did not include these new variables. Accordingly it will be quite interesting to investigate the claim that intra-Arab trade is low in the light of this study’s augmented gravity model and especially that the data run up and include year 2000. Given that trade flows in goods are not the only kind of flows between Arab countries, perhaps it is interesting to explore how flows of services, labor and capital affect intra-Arab trade in goods. Flows in labor services can be captured through a variable that would have the number of civilian workers residing in an Arab country who are also citizens of another Arab country, while capital flows can be captured Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 205 through a variable that represents the amount of FDI, aid, or grants from one Arab country to another in any given year. Another interesting area of research would be carrying this same study but after disaggregating commodities to see which ones are mostly affected by governance and which are not. In conclusion, governance does affect World, intra-Arab, and Arab-World trade flows, especially its rule of law and control of corruption components. It could also play a major role in attaining both economic and social development in the Arab countries. Thus governance reforms, even if not implemented for their own intrinsic good, should nevertheless be aggressively pursued for their positive impact on development. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 206 BIBLIOGRAPHY Abouyoub, H. (1992) “Trade Policies and Economic Integration Among the Arab Maghreb Union Countries.” In Said El-Naggar ed., Foreign and Intratrade Policies of the Arab Countries. Washington, D.C.: IMF. Aitken, N. (1973) “The Effect of the EEC and EFT A on European Trade: A Temporal Cross-Section Analysis.” American Economic Review, vol.63 (Dec.): 881-892. Aitken, N. and R. Obutelewicz (1976) “A Cross-Sectional Study of EEC Trade with the Association of African Countries.” The Review of Economics and Statistics, vol.58 (Nov.): 425-433. Akhtar, M. and R. Hilton (1984) “Exchange Rate Uncertainty and International Trade: Some Conceptual Issues and New Estimates from Germany and the U.S.” Federal Reserve Bank of New York, Research Paper no 8403. Al-Atrash, H. and T. Yousef (2000) “Intra-Arab Trade: Is It Too Little?” A Working Paper of the International Monetary Fund, no 00/10. Allum, P. (1998) “Inter-Arab Trade: Constraints and Prospects” Paper for the Sectoral Meeting on Trade and Development Between the League of Arab States and the United Nations, Cairo, June 8-11. Anderson, J. (1979) “A Theoretical Foundation For the Gravity Equation.” American Economic Review, vol.69: 106-116. Anderson, J. and D. Marcouiller (2002) “Insecurity and the Pattern of Trade: An Empirical Investigation.” The Review of Economics and Statistics, vol. 84 (2): 342-352. Anderson, J. and E. van Wincoop (2001) “Gravity with Gravitas: A Solution to the Border Puzzle.” NBER Working Paper Series no. 8079. Aristotelous, K. (2001) “Exchange-Rate Volatility, Exchange-Rate Regime, and Trade Volume: Evidence from the UK-US Export Function (1889-1999).” Economics Letters, vol. 72: 87-94. Asseery, A. and D. Peel (1991) “The Effects of Exchange Rate Volatility on Exports— Some New Estimates.” Economics Letters, vol.37: 173-177. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 207 Baier, S. and J. Bergstrand (2001) “The Growth of World Trade: Tariffs, Transport Costs, and Income Similarity.” Journal of International Economics, vol. 53: 1- 27. Bailey, M., G. Tavlas and M. Ullan (1986) “Exchange Rate Variability and Trade Performance: Evidence for the Big Seven Industrial Countries.” Weltwirtschaftliches Archiv, vol. 122: 466-477. Bardhan, P. (1997) “Corruption and Development: A Review of Issues.” Journal of Economic Literature, vol. 35 (3): 1320-1346. Barkoulas, J ., C. Baum and M. Caglayan (2002) “Exchange Rate Effects on the Volume and Variability of Trade Flows.” Journal of International Money and Finance, vol. 21: 481-496. Baron, D. (1976) “Fluctuating Exchange Rates and the Pricing of Exports.” Economic Inqiry, vol. 14 (3): 425-438. Bayoumi, T. and B. Eichengreen (1997) “Is Regionalism Simply a Diversion? Evidence from the EU and the EFTA.” In T. Ito and A. Krueger ed., Regionalism Vs. Multilateral Trade Arrangements. Chicago: University of Chicago Press. Bergstrand, J. (1985) “The Gravity Equation in International Trade: Some Microeconomic Foundations and Empirical Evidence.” The Review of Economics and Statistics, vol. 67 (August): 474-81. ________ (1989) ‘The Generalized Gravity Equation, Monopolistic Competition and the Factor-Proportions theory in International Trade.” The Review of Economics and Statistics, vol.71: 143-153. ________ (1990) “The Heckscher-Ohlin-Samuelson Model, the Linder Hypothesis and the Determinants of Bilateral Intra-industry Trade.” The Economic Journal, vol. 100 (Dec.): 1216-1229. Bhagwati, J. (1993) “Regionalism and Multilateralism: An Overview.” In Jaime De Melo and Arvind Panagariya ed., New Dimensions in Regional Integration. Cambridge: Cambridge University Press. Bliss, C. and R. Di Telia (1997) “Does Competition Kill Corruption?” Journal of Economic Literature, vol. 35 (Sep.) 1320-1346. Bliss, H. and B. Russet (1998) “Democratic Trading Partners: The Liberal Connection, 1962-1989.” The Journal of Politics, vol. 60 (4): 1126-1147. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 208 Bougheas, S., P. Demetriades and E. Morgenroth (1999) “Infrastructure, Transport Costs, and Trade.” Journal o f International Economics, vol. 47: 169-189. Brada, J. and J. Mendez (1985) “Economic Integration among Developed, Developing and Centrally Planned Economies: A Comparative Analysis.” The Review of Economics and Statistics, vol. 67 (4): 549-556. ________ (1988) “Exchange Rate Risk, Exchange Rate Regime and the Volume of International Trade.” KYKLOS, vol. 41: 263-280. Bremer, S. (1992) “Dangerous Dyads.” Journal o f Conflict Resolution 36(2): 309-341. Carothers, T. (1998) “The Rule of Law Revival.” Foreign Affairs, April-March. Cho, G., I. Sheldon and S. McConiston (2002) “Exchange Rate Uncertainty and Agricultural Trade.” American Journal o f Agricultural Economics, vol. 84(4): 931-942. Chowdhury, A. (1993) “Does Exchange Rate Volatility Depress Trade Flows? Evidence from Error-Correction Models.” The Review of Economics and Statistics, vol. 75(4): 700-706. Clark, P. (1973) “Uncertainty, Exchange Risk, and the Level of International Trade.” Western Economic Journal, vol.ll: 302-313. Cushman, D. (1983) “The Effects of Real Exchange Rate Risk on International Trade.” Journal of International Economics, vol. 15 (1-2): 45-63. ________ (1986) “Has Exchange Rate Risk Depressed International Trade? The Impact of Third Country Exchange Risk.” Journal of International Money and Finance, vol. 5(3): 361-379. Deardorff A. (1984) “Testing Trade Theories and Predicting Trade Flows.” In R. Jones and P. Kenen ed., Handbook o f International Economics. Amsterdam: North-Holland. Deardorff, A. (1995) “Determinants of Bilateral Trade: Does Gravity Work in a Neoclassical World?” NBER Working Paper no. 5377. Cambridge, MA. Deininger, K. and L. Squire (1997) “Measuring Income Inequality: A New Database.” Working Paper, Harvard: Institute for International Development. Dell’Ariccia, G. (1999) “Exchange Rate Fluctuations and Trade Flows: Evidence from the European Union.” International Monetary Fund Staff Papers, vol. 46(3): 315-334. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 209 De Melo, J. and A. Panagariya (1993) “Introduction.” In Jaime De Melo and Arvind Panagariya ed., New Dimensions in Regional Integration. Cambridge: Cambridge University Press. Dhar, S. and A. Panagariya (1999) “Is East Asia Less Open than North America and the EEC? No.” In J. Piggott and A. Woodland ed., International Trade Policy and the Pacific Rim. London: Macmillan. Diamond, L. (1999). Developing Democracy: Toward Consolidation. Baltimore: Johns Hopkins University Press. Dixon, W. and B. Moon (1993) “Political Similarity and American Foreign Trade Patterns.” Political Research Quarterly, vol. 46(1): 5-25. Doyle, M. (1986) “Liberalism and World Politics.” American Political Science Review, vol. 80(4): 1151-69. Eaton, J. and S. Kortum (2001) Technology, Geography, and Trade. Boston University Economic Research Forum (1998) Economic Trends in the MENA Region. Cairo: The Economic Research Forum for the Arab Countries, Iran and Turkey. ________ (2000) Economic Trends in the MENA Region. Cairo: The Economic Research Fomm for the Arab Countries, Iran and Turkey. Ekholm, K., J. Torstensson, and R. Torstensson (1995) “Prospects for Trade in the Middle East and North Africa: An Econometric Analysis,” paper prepared for an Economic Research Forum Conference on Liberalization of Trade and Foreign Investment, Bogazicu University, Istanbul. ________ (1996) “The Economics of the Middle East Peace Process: Are There Prospects for Trade and Growth?” The World Economy. Oxford: Blackwell Publishers Ltd. El-Erian, M. and S. Fischer (1996) “Is MENA a Region? The Scope for Regional Integration.” Working Paper no. 96/30, IMF. El-Naggar, Said (1992) “Foreign and Intratrade Policies of the Arab Countries.” In Said El-Naggar ed., Foreign and Intratrade Policies of the Arab Countries. Washington, D.C.: IMF. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 210 Feenstra, R. (2000) “World Trade Flows, 1980-1997, with Production and Tariff Data.” Davis, CA: Institute of Governmental Affairs, University of California, Davis. Feenstra, R. and J. Kendall (1999) “Exchange Rate Volatility and International Prices.” NBER Working paper no. 3644, Cambridge, MA. Finger, J., M. Ng and I. Soloaga (1998) Trade Policies in the Caribbean Countries: A Look at the Positive Agenda. Washington, DC.: Caribbean Group for Co- operation on Economic Development. Fischer, S. (1993) “Prospects for Regional Integration in the Middle East.” In Jaime De Melo and Arvind Panagariya ed., New Dimensions in Regional Integration. Cambridge: Cambridge University Press. Franke, G. (1991) “Exchange Rate Volatility and International Trading Strategy.” Journal of International Money and Finance, vol. 10: 292-307. Frankel, J. (1993) “Is Japan Creating a Yen Bloc in East Asia and the Pacific?” In J. Frankel and M. Kahler ed., Regionalism and Rivalry: Japan and the United States in Pacific Asia. Chicago: University of Chicago Press. Frankel, J. and A. Rose (2002) “An Estimate of the Effect of Common Currencies on Trade and Income.” The Quarterly Journal o f Economics, vol. 117(2): 437- 466. Frankel, J., E. Stein and S-J. Wei (1995) “ Trading Blocs and the Americas: the Natural and Unnatural and the Super-Natural.” Journal of Development Economics, vol. 47: 61-95. Frankel, J.A. and S-J. Wei (1993) “Trade Blocs and Currency Blocs.” NBER Working Paper no. 4335. Cambridge, MA. Gaile, G. and R. Grant (1989) “Trade, Power, and Location: The Spatial Dynamics of the Relationship between Exchange and Political-Economic Strength.” Economic Geography, vol. 65(4), Trade Theories, Scale, and Structure: 329- 337. Galal, Ahmed (2000) “Incentives for Economic Integration in the Middle East.” In Bernard Hoekman and Hanaa Kheir-El-Din ed., Trade Policy Developments in the Middle East and North Africa. Washington, D.C.: World Bank. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 211 Galal, A. and B. Hoekman (1997) “Overview.” In Ahmed Gala! and Bernard Hoekman ed., Regional Partners in Global Markets: Limits and Possibilities of the Euro-Med Agreements. Cairo: ECES. Geraci, V. and W. Prewo (1977) “Bilateral Trade Flows and Transport Costs.” The Review of Economics and Statistics (Feb.): 67-74. Glick, R. and A. Rose (2002) “Does a Currency Union Affect Trade? The Time-Series Evidence.” European Economic Review, vol. 46: 1125-1151. Gotur, P. (1985) “The Effects of Exchange Rate Volatility on Trade.” IMF Staff Papers, 32(Sep.): 475-512. Gould, D. (1994) “Immigrant Links to the Home Country: Empirical Implications for U.S. Bilateral Trade Flows.” The Review o f Economics and Statistics, vol. 76: 302-316. Gowa, J. and E. Mansfield (1993) “Power Politics and International Trade.” American Political Science Review, vol. 87(4): 408-420. Greenaway, D. (2000) “Multilateralism, Minilateralism, and Trade Expansion.” In D. Das ed., Asian exports. Oxford: Oxford University Press. Greenaway, D. and C. Milner (2000) “Regionalism and Gravity.” Scottish Journal of Political Economy, vol. 49(5): 574-585. Gurr, T., K. Jaggers and W. Moore (1989) Polity II Codebook. Boudler, CO: Center for Comparative Politics, Department of Political Science, University of Colorado. Hamilton, C. and L. Winters (1992) “Opening up International Trade with Eastern Europe.” Economic Policy, vol. 14: 77-116. Hassan, M. (2001) “Is SAARC a Viable Economic Bloc? Evidence from Gravity Model.” Journal of Asian Economics, vol. 12: 263-290. Havrylyshyn, O. (1997) A Global Integration Strategy fo r the Mediterranean Countries: Open Trade and Market Reforms. Washington, D.C: IMF. Head, K. and J. Ries (1998) “Immigration and Trade Creation: Econometric Evidence from Canada.” The Canadian Journal of Economics, vol. 31(1): 47-62. Helliwell J. (1996) “Do National Borders Matter for Quebec’s Trade?” Canadian Journal of Economics, vol. 29: 507-522. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 212 _________ (1998) How Much Do National Borders Matter? Washington: Brookings Institute. Helpman, E. and P. Krugman (1985) Market Structure and Foreign Trade. Cambridge, MA: MIT Press. Hunter, L. (1991) “The Contribution of Nonhomothetic Preferences to Trade.” Journal of International Economics vol.30: 345-358. Hunter, L. and J. Markusen (1988) “Per-Capita Income as a Determinant of Trade.” In R. Feenstra ed., Empirical Methods for International Trade. Cambridge: MIT Press. Huntington, S. (1968) Political Order in Changing Societies. New Haven: Yale University Press. Hutchinson, W. (2001) “Linguistic Distance as a Determinant of U.S. Bilateral Trade.” Working Paper no. 01 -W-25.Vanderbilt University Department of Economics Hutchinson, W. (2002) “Does Ease of Communication Increase Trade? Commonality of Language and Bilateral Trade.” Scottish Journal o f Political Economy, vol. 49(5). International Monetary Fund (1986) Annual Report on Exchange Arrangements and Exchange Restrictions. Washington,DC.: IMF. Kaufmann, D., A. Kraay and P. Zoido-Lobaton (1999a) “Aggregating Governance Indicators.” Policy Research Working Papers no. 2195, World Bank. _________ (1999b) “Governance Matters.” Policy Research Working Papers no. 2196, World Bank. _________ (2002) “Governance Matters II.” Policy Research Working Papers no. 2772, World Bank. Kaufmann, D., G. Mehrez, and S. Schmukler (1999) “Was the East Asia Crisis Predictable?” Manuscript, The World Bank. Kaufmann, D. and S-J. Wei (2000) “Does ‘Grease Money’ Speed Up the Wheels of Commerce?” IMF Working Paper WP/00/64. Keynes, J.M. (1936) The General Theory of Employment, Interest, and Money. New York: Harcourt, Brace and World. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 213 Klitgaard, R. (1988) Controlling Corruption. Berkeley: University of California Press. Koray, F. and W. Lastrapes (1989) “Real Exchange Rate Volatility and U.S. Bilateral Trade: A Var Aproach.” The Review of Economics and Statistics, vol. 71(4): 708-712. Kroner, K. and W. Lastrapes (1993) “The Impact of Exchange Rate Volatility on International Trade: Reduced Form Estimates Using the GARCH-in-mean Model.” Journal of International Money and Finance, vol. 12 (3): 298-318. Krueger, A. (1974) “ The Political Economy of the Rent-Seeking Society.” American Economic Review, vol. 64 (3): 291-303. Learner, E. E. and J. Levinsohn (1995) “International Trade Theory: The Evidence.” In Grossman, G. and K. Rogoff ed., The Handbook of International Economics, vol. Ill, Amsterdam: Elsevier-North Holland. Learner, E. and R. Stem (1970) Quantitative International Economics. Boston: Allyn and Bacon. Leff, N. (1964) “ Economic Development through Bureaucratic Corruption.” The American Behavioral Scientist, vol. 8 (2): 8-14. Limam, I. and A. Abdalla (1999) Inter-Arab Trade and the Potential Success of AFT A. Kuwait: Arab Planning Institute. Limao, N. and A.Venebles (2001) “Infrastructure, Geographical Disadvantage, Transport Costs, and Trade.” World Bank Economic Review, vol. 15 (3): 451- 479. Linder, S. (1961) An Essay on Trade and Transformation. Stockholm: Almqvist and Wicksell Linnemann, H. (1966) An Econometric Study of World Trade Flows. Amsterdam: North Holland Publishing Co. Long, A. (2003) “Defense Pacts and International Trade.” Journal of Peace Research, vol. 40 (5): 537-552. Lui, F. (1985) “ An Equilibrium Queuing Model of Bribery.” Journal of Political Economy, vol. 93(4): 760-781. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 214 Mansfield, E. and R. Bronson (1997a) “Alliances, Preferential Trading Arrangements, and International Trade.” The American Political Science Review, vol. 91(1): 94-107. ________ (1997b) “The Political Economy of Major-Power Trade Flows.” In E. Mansfield and H. Milner ed., The Political Economy of Regionalism. New York: Columbia University Press. Maoz, Z. and N. Abdolali (1989) “Regime Type and International Conflict, 1816- 1976.” Journal of Conflict Resolution, vol. 33: 3-35. McCallum, J. (1995) “National Borders Matter: Canada-US Regional Trade Patterns.” American Economic Review, vol.85 (3): 615-623. McKenzie, M. and R. Brooks (1997) “The Impact of Exchange Rate Volatility on German-U.S. Trade Flows.” Journal of International Financial Markets, Institutions, and Money, vol. 7: 73-88. Miniesy, R., J. Nugent and T. Yousef (2004) “Intra-Regional Trade Integration in the Middle East: Past Performance and Future Potential.” In H. Hakimian and J. Nugent ed., Trade Policy and Economic Integration in the Middle East and North Africa: Economic Boundaries in Flux. London: RoutledgeCurzon. Morrow, J., R. Siverson and T. Tabares (1998) “The Political Determinants of International Trade: The Major Powers, 1907-90.” The American Political Science Review, vol. 92(3): 649-661. Mundell, R. (1961) “Optimum Currency Areas.” American Economic Review, vol.51: 509-517. Nabli, M. and A. DeKleine (2000) “Managing Global Integration in the Middle East and North Africa.” In Bernard Hoekman and Hanaa Kheir-El-Din ed., Trade Policy Developments in the Middle East and North Africa. Washington, D.C.: World Bank. Najem, T. (2003) “Good Governance: the Definition and Application of the Concept.” In T. Najem and M. Hetherington ed., Good Governance in the Middle East Oil Monarchies. London: RoutledgeCurzon. Nilsson, L. (2000) “ Trade Integration and the EU Economic Membership Criteria.” European Journal of Political Economy, vol. 165: 807-827. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 215 Nilsson, K. and L. Nilsson (2000) “Exchange Rate Regimes and Export Performance of Developing Countries.” The World Economy. Oxford: Blackwell Publishers Ltd.. Nitsch, V. (2000) “National Borders and International Trade: Evidence from the European Union.” The Canadian Journal of Economics, vol. 33(4): 1091- 1105. _________ (2002) “Honey, I Shrunk the Currency Union Effect on Trade.” The World Economy. Oxford: Blackwell Publishers Ltd. Penubarti, M. and M. Ward (2000) “Commerce and Democracy.” Working Paper no. 6. Center for Statistics and the Social Sciences, University of Washington. Peree, E. and A. Steinterr (1989) “Exchange Rate Uncertainty and Foreign Trade.” European Economic Review, vol.33: 1241-1264. Pollins, B. (1989) “Conflict, Cooperation, and Commerce: The Effect of International Political Interactions on Bilateral Trade Flows.” American Journal of Political Science, vol. 33(3): 737-761. Potter, D. (1997) Democratization. Cambridge,UK: Blackwell Publishers. Pozo, S. (1992) “Conditional Exchange-Rate Volatility and the Volume of International Trade: Evidence from the Early 1900s.” The Review of Economics and Statistics, vol. 74 (2): 325-329. Primo Braga, C. and C. Fink (1998) “Reforming Intellectual Property Rights Regimes: Challenges for Developing Countries.” Journal of International Economic Law, vol. 1 (4): 537-554. Rauch, J. (1999) “Networks Vs. Markets in International Trade.” Journal of international Economics, vol.48 (1): 7-35. Refaat, A. (2000) “Egypt: An Assessment of Recent Trade Policy Developments.” In Bernard Hoekman and Hanaa Kheir-El-Din ed., Trade Policy Developments in the Middle East and North Africa. Washington, D.C.: World Bank. Ridgway, D. and M. Talib (2003) “Globalization and Development - Free Trade, Foreign Aid, Investment and the Rule of Law.” California Western International Law Journal, vol. 33 (2): 325- 344. Rose, Andrew (2000) “One Money, One Market: Estimating the Effect of Common Currencies on Trade.” Economic Policy, April. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 216 Rose-Ackerman, S. (1978) Corruption: A Study in Political Economy. New York: Academic Press. _________ (1994) “Reducing Bribery in the Public Sector.” In D. Trang ed., Corruption and Democracy: Political Institutions, Processes and Corruption in Transition States in East-Central Europe and in the Former Soviet Union. Budapest: Institute for Constitutional and Legislative Policy. Rose, A. and E. Van Wincoop (2001) “National Money as a Barrier to Trade: The Real Case for Monetary Union.” American Economic Review, Papers, and Proceedings vol. 91(2), 386-390. Samuelson, P. and W. Nordhaus (1995) Economics. New York: McGraw-Hill. Sanso, M., R. Cuairan and F. Sanz (1993) “Bilateral Trade Flows, The Gravity Equation, and Functional Form.” The Review of Economics and Statistics, vol. 75(2): 266-275. Sapir, A. (2001) “Domino Effects in Western Europe Regional Trade, 1960-1992.” European Journal of Political Economy, vol. 17: 377-388. Secru, P. and C. Van Hulle (1992) “Exchange Rate Volatility, International Trade, and the Value of Exporting Firms.” Journal o f Banking and Finance, vol. 16: 155- 182. Sercu, P. and R. Uppal (2003) “Exchange Rate Volatility and International Trade: A General-Equilibrium Analysis.” European Economic Review, vol. 47: 429-441. Shafik, N. (1998) “Prospects for Middle Eastern and North African Economies: An Overview.” In Nemat Shafik ed., Prospects for Middle Eastern and North African Economies: From Boom to Bust and Back? Houndmills: Macmillan Press Ltd. Shartna, S. and S. Chua (2000) “ASEAN Economic Integration and Intra-Regional Trade.” Applied Economic Letters, vol. 17: 165-169. Shleifer, A. and R. Vishny (1993) “Corruption.” Quarterly Journal of Economics, vol. 108(3): 599-617. _________(1994), “Politicians and Firms.” Quarterly Journal of Economics, vol. 109. Soloaga, I. and A. Winters (1999) “ How Has Regionalism in the 1990s Affected Trade Policy?” Research Working Paper no. 2156. Washington, DC.: World Bank, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 217 Srivastava, R. K. and R. T. Green (1986) “Determinants of Bilateral Trade Flows.” Journal of Business 59 (Oct.): 623-640. Summary, R. (1989) “A Political-Economic Model of U.S. Bilateral Trade.” The Review of Economics and Statistics, vol. 71(1): 179-182. Summers, L. (1991) “Regionalism and the World Trading System.” In L.H. Summers, Policy Implications of Trade and Currency Zones. Wyoming: Federal Reserve Bank of Kansas City, M-295-303. Svedberg, P. (1981) “Colonial Enforcement of Foreign Direct Investment.” Manchester School of Economic and Social Studies 49 (Mar.): 21-38. Tamirisa, N. (1999) “Exchange and Capital Controls as Barriers to Trade.” IMF Staff Papers, vol. 46(1). International Monetary Fund. Thurow, L. (1992) Head to Head: The Coming Battle Among Japan, Europe, and America. New York: William Morrow. Thursby, J. and M. Thursby (1987) “Bilateral Trade Flows, the Linder Hypothesis, and Exchange Risk.” The Review of Economics and Statistics, vol. 69 (3): 488-495. Verdier, D. (1998) “Democratic Convergence and Free Trade.” International Studies Quarterly, vol. 42(1): 1-24. Wei, S-J. (1996) “Intra-National Versus International Trade: How Stubborn are Nations in Global Integration?” NBER Working Paper no. 5531. Wilson, R (1994) “The Economic Relations of the Middle East: Toward Europe or Within the Region?” Middle East Journal, vol. 48(2): 268-287. World Bank (1994) World Development Report; Infrastructure for Development. Oxford University Press. ________ (2003a) MENA Development Report; Better Governance for Development in the Middle East and North Africa: Enhancing Inclusiveness and Accountability. Washington, DC.: World Bank. ________ (2003b) Jobs, Growth, and Governance in the Middle East and North Africa: Unlocking the Potential for Prosperity. Washington, DC.: The World Bank. Yeats, A. (1996) “Export Prospects of Middle Eastern Countries: A Post-Uruguay Round Analysis.” Policy Research Working Papers no. 1571, World Bank. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 218 Yeyati, E. (2003) “On the Impact of a Common Currency on Bilateral Trade.” Economics Letters, vol. 79: 125-129. Zarrouk, J. (1992) “Intra-Arab Trade: Determinants and Prospects for Expansion.” In Said El-Naggar ed., Foreign and Intratrade Policies of the Arab Countries. Washington, D.C.: IMF. ________ (2000) “The Greater Arab Free Trade Area: Limits and Possibilities.” In Bernard Hoekman and Jamal Zarrouk ed., Catching Up with the Competition: Trade Opportunities and Challenges for Arab Countries. Ann Arbor: University of Michigan Press. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 219 APPENDICES Appendix A: List of Countries and Summary Statistics Table A.I. List of Countries. Afghanistan Comoros Albania Congo Algeria Congo, Dem. Rep. Of (Zaire) Angola Costa Rica Argentina Cote D'Ivoire Armenia Croatia Australia Cuba Austria Cyprus Azerbaijan Czech Republic Bahamas Czechoslovakia Bahrain Denmark Bangladesh Djibouti Barbados Dominican Rp Belarus Ecuador Belgium-Luxembourg Egypt Belize El Salvador Benin Eq. Guinea Bermuda Estonia Bhutan Ethiopia Bolivia Falkland Isl Bosnia And Herzegovina Fiji Brazil Finland British Indian Ocean Territory Fm East Germany Brunei Fm USSR Bulgaria Fm Yugoslavia (Includes Croatia, Slovenia) Burkina Faso France Burundi French Guiana Cambodia Gabon Cameroon Gambia Canada Georgia Cayman Islds Germany Central Afr. Rep. Ghana Chad Gibraltar Chile Greece China Greenland Colombia Guadeloupe (Includes Martinique) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table A .I. List of Countries (continued). Guatemala Guinea Guinea-Bissau (Includes Cape Verde) Guyana Haiti Honduras Hong Kong Hungary Iceland India Indonesia (Including Macao) Iran Iraq Ireland Israel Italy Jamaica Japan Jordan Kazakhstan Kenya Kiribati (Includes Tonga) Kuwait Kyrgyz Republic Laos Latvia Lebanon Liberia Libya Lithuania Macedonia Madagascar Malawi Malaysia Maldives Mali Malta Mauritania Mauritius Mexico Moldova Mongolia Morocco Mozambique Myanmar (Burma) Nepal Neth Antilles Netherlands New Caledonia (Includes French Polynesia and Vanutu) New Zealand Nicaragua Niger Nigeria North Korea Norway Oman Pakistan Panama Papua N.Guinea Paraguay Peru Philippines Poland Portugal Qatar Reunion Romania Russia Rwanda Saudi Arabia Senegal Serbia And Montenegro Seychelles Sierra Leone Singapore Slovak Republic Slovenia Solomon Islds Somalia South Africa South Korea Spain Sri Lanka St Kitts Nev (Includes Dominica, St Lucia,St Vinct& Grena, Grenada) St Pierre Miqu St.Helena Sudan Surinam Sweden Switzerland Syria Taiwan Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 221 Table A. 1. List of Countries (continued) Tajikistan Tanzania Thailand Togo Trinidad-Tobago Tunisia Turkey Turkmenistan Turks Caicos Islands Uganda Ukraine United Kingdom Untd Arab Em Uruguay USA Uzbekistan Venezuela Vietnam Western Sahara Yemen Zambia Zimbabwe Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 222 Variable Obs Mean Std. Dev. Min Max totaltrade 76145 5.400184 13.04579 -11.51293 26.75217 energytrade 57662 -6.965439 10.00887 -11.51293 23.87519 othertrade 57662 2.392658 13.56567 -11.51293 26.44751 gdp 58152 46.48911 3.147089 36.44554 59.10682 gdppc 58152 14.99546 2.213378 9.151914 21.30394 distance 75917 8.274606 0.7440462 4.017474 9.422191 areas 76145 13.1325 1.634479 4.091006 17.28909 11 76145 0.3409285 0.5312208 0 2 border 76145 0.0176111 0.1315341 0 1 language 76145 0.1188522 0.3236166 0 1 regional 76145 0.0126338 0.1116886 0 1 nation 76145 0.0024952 0.0498903 0 1 colonizer 76145 0.0796507 0.2707534 0 1 colonial 76145 0.0057784 0.0757967 0 1 erv 57196 0.089882 0.1137208 0 1.445983 cu 76145 0.0078534 0.0882715 0 1 gov_ds 35407 0.0182123 2.330399 -8.013136 18.56942 cpy 76145 0.0420907 0.20462 0 2 cpe 76145 0.4022588 0.5681495 0 2 gini 33639 80.85396 13.90207 38.36 128.9 m2 gdp 42609 1653.952 1965.633 5.359096 42917.96 current 64344 1.031005 0.7162096 0 2 capital 64640 1.665408 0.5422688 0 2 erv'85 57196 0.0083587 0.0323139 0 0.426402 erv_'90 57196 0.015277 0.0563455 0 0.6600248 erv_'95 57196 0.0278781 0.0850334 0 1.445983 erv_'97 57196 0.0238137 0.0749437 0 0.9946942 colonialj 76145 0.0115569 0.1858429 0 4 barder_'85 76145 0.0031519 0.0560534 0 1 border_'90 76145 0.0031519 0.0560534 0 1 barder_'95 76145 0.0037691 0.0612778 0 1 border_’ 97 76145 0.0037691 0.0612778 0 1 lagged bilat 42557 5.049489 13.04734 -11.51293 26.37619 va 40819 14.44867 4.437643 0 24 ps 40819 17.37307 4.360792 0.25 24 ge 40819 19.20227 4.84989 2 30.41667 rq 40819 12.73283 3.106689 1 22.16667 rl 40819 7.456003 2.270393 1 1 2 cc 40819 6.512643 1.929329 0 12.16667 gov_s 39207 0.0363035 3.000921 -9.765629 22.7647 gov_w 39207 32.05532 7.87892 5.856679 90.83852 gov_dw 35407 0.0581288 6.191391 -21.57695 48.12693 gcc 76145 0.000985 0.0313689 0 1 amu 76145 0.0005253 0.0229138 0 1 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 223 Appendix B: Data Manipulation General Comment on Countries used Composite countries are: Belgium-Luxembourg, Guadeloupe including Martinique, Indonesia including Macao, Kiribati including Tonga, New Caledonia including French Polynesia and Vanuatu, and St. Kitts and Nevis including Dominica, St. Lucia, Grenada, and St. Vincent and the Grenadines. - Disaggregated Countries are countries that used to exist until 1990 but were disaggregated afterwards, such as Czechoslovakia, Soviet Union, and Yugoslavia. From 1995 onwards, the components of the disaggregated countries themselves start to exist like Czech and Slovakia, 15 ex-USSR countries, and the 5 ex- Yugoslavian countries. - East Germany exists until 1990, Germany until 1990 refers to W. Germany, while from 1995 onwards it refers to United Germany. Trade - The values of trade are in nominal US dollars. - Bilateral trade flows are simply the addition of exports from i to j and from j to i (or imports of j from i and of i from j) in Feenstra et al (2000). This is because exports from i to j are the same as imports of j from i. In IMF data however, there are variations between recorded exports from i to j and imports Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 224 of j from i and so we took the bigger value and added it to the bigger value of exports from j to i / imports of i from j to represent bilateral trade flows in that specific year. Trade flows are also divided to energy (mainly oil and gas), which are denoted by all SITC codes starting with “3” and non-energy or other, which is total trade flows minus energy flows. Since IMF’s DOTS CD-ROM does not have disaggregated data, we don’t have energy and other trade in 2000 (only total trade flows). Missing data left, which is not complemented from IMF for 1985, 1990, 1995, and 1997, and missing data in 2000 are considered once as nulls, i.e. genuinely missing, where the OLS estimation technique is used, and once as zeros, i.e. no trade between the two countries, where both OLS and Tobit are used each at a time. The function is loglinear and we use Ln. But since Ln (0) is undefined, we assumed a very small number (0.00001) to represent the absence of trade flows. In composite countries, we followed the same steps in point 2 and 3 above. Afterwards, we considered them as one country and added their trade with each trading partner together to get bilateral trade flows. Countries that existed until 1990 were included along with their trade data in 1985 and 1990. Afterwards, they cease to exist in the same name (e.g. Czechoslovakia, East Germany, Soviet Union and Yugoslavia). Starting in 1995, the new countries exist (Czech Republic, Slovakia, United Germany, 15 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 225 ex-Soviet republics, and 5 countries from the former Yugoslavia) and we have trade data for them. GDP, GDPPC, M2GDP The values of GDP are in nominal prices. - For composite countries, their GDPs were simply added. They were then multiplied by the trading partner’s GDP. - The complemented data from the UN source are mainly GDP for Guadeloupe, Martinique, French Guyana, and Reunion for 1985 and 1990. - GDPPC was calculated by dividing GDP in current prices by population for each country for each year. - For GDPPC of composite countries, the sum of their GDPs was divided by the sum of their populations. The resultant GDPPC was then multiplied by the trading partner’s GDPPC. - The UN source was used to get populations of Czechoslovakia, Soviet Union, and Yugoslavia in 1985&1990 and French Guyana, Gibraltar, Guadeloupe, Martinique, Reunion, St. Helena, St. Pierre and Miquelon, Taiwan, Turks and Caicos Islands, Tuvalu, and Western Sahara in 1985-2000. - For M2GDP of composite countries, the average M2GDP is multiplied by the trading partner’s M2GDP. If data for one component country is missing, the average of the rest is taken. So for Indonesia and Macao, in all years except 2000, only M2GDP of Indonesia is taken, for 2000, the average is taken. For Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 226 Kiribati, only M2GDP of Tonga is taken. For St. Kitts and Nevis, an average of all M2GDP values is taken. Areas, LL, border, Language, Nation, Colonizer, and Colonial Areas are calculated in square kilometers. Since, the 2002 World Fact Book did not provide East Germany’s area, we took it from Microsoft Encarta Encyclopedia 1999. - For Czechoslovakia, Soviet Union, and Yugoslavia, the areas of the former component republics that are now independent were added together. For composite countries, we added their individual areas together. - Composite countries are assumed to be not landlocked if at least one component country is not landlocked. For instance, although Luxembourg is landlocked, together Belgium and Luxembourg are considered not landlocked. A composite country is assumed to have a common border with all the countries bordering each of its component countries. For instance, Belgium has a common border with Netherlands but Luxembourg does not. We assumed that together Belgium and Luxembourg have a common border with Netherlands. - For Common Language, only official languages are taken into consideration. A composite country is assumed to have the official languages of all of its component countries. For instance, Belgium has French, Dutch, and German as official languages, while Luxembourg has French and German only. We Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 227 assumed that together Belgium and Luxembourg have the three languages as official. - For common nation, common colonizer, and colonial relationship, again the union of the component countries’ data values is used as in the border and language variables before.. - If a country is now an overseas territory of another country (e.g. French Polynesia and France), it is not considered as being colonized by France. Therefore, it will have the colonial relationship variable set to false while the common nation variable will be set to true. Distance - Distance is calculated in miles. - For composite countries, the coordinates of the primary country were used. For instance, for Belgium-Luxembourg, Belgium’s coordinates are used. Regional - Only regional agreements that were adopted and significant were used. A composite country is assumed to have RTAs with all the countries that have RTAs with each of its component countries. Exchange Rate Volatility - Exchange rate data was available from IFS at www.imf.org. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 228 Each country’s exchange rate is obtained with respect to United States dollars. The bilateral exchange rate is then obtained by dividing the two countries’ exchange rates. Afterwards, the first difference log of the bilateral exchange rate is calculated by subtracting the monthly log bilateral exchange from the previous month. The exchange rate volatility is then calculated as the standard deviation of the subtraction result for all months. - No data was available for disaggregated countries (Czechoslovakia, Soviet Union, and Yugoslavia) or East Germany. Data was available for West Germany in 1985-1990 and then for Unified Germany afterwards. - For composite countries, the average of their bilateral nominal exchange rates for every month is obtained and then the calculation continues as usual i.e. taking natural logs, then first difference, and finally standard deviation. If data on one component country is not available, the average of the other component countries is used. Currency Union - A composite country is assumed to have a common currency with all the countries that have a common currency with each of its component countries. For instance, Kiribati uses the Australian Dollar but Tonga does not. We included that Kiribati, which includes Tonga, has a currency union/common currency with Australia. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 229 Gini - The UNU/WEDER - UNDP World Income Inequality Database contains within it the Deininger and Squire (1997) Database. - As much as possible, the criteria used to choose Gini coefficients was through choosing “Income” with all its criteria rather than “Expenditure”. Next, “Household” rather than “Individual” and other criteria, then all areas (rural or urban), all population, all age groups. If all existed and still we have a number of Gini’s for the same year, we choose the Best rating (5 Nook or 1 OKIN). If the criteria have been used but still we had missing observations, the criteria limits were extended and used starting from the beginning, meaning that the best rating would be ignored then all age groups would be ignored where whatever age group we find we take the data, then all population until we reach income. Income includes all its criteria but still expenditures were not used. Afterwards, data is intrapolated and extrapolated throughout the entire sample to try to fill in missing data as much as possible. Intrapolating taking average for in between years (e.g. 1990 is calculated as average of 1985 and 1995 if both exist). Extrapolation taking previous year or latest year available. Current and Capital Account Restrictions - For each year, the data was collected from the following year’s Annual Report. In other words, for 1985, 1990, 1995, 1997, and 2000, Annual Reports of 1986, 1991, 1996, 1997, and 2001 were used. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 230 - For composite countries, if the number of countries with restrictions is greater than or equal to the number of countries without restrictions, the result would be that restrictions exist, i.e. the value for current or capital account will be true for this specific country of the trading pairs, the final value placed after the condition of the other trading country is evaluated. Governance - Governance data was derived from variables in Table 3B of the International Country Risk Guide (ICRG) because it had the most coverage of countries and years. - Averages of the monthly indices for military in politics added to democratic accountability, internal conflict, government stability added to bureaucratic quality, investment profile, law and order, corruption were obtained to form early values for voice and accountability (VA), political stability (PS), government effectiveness (GE), regulatory quality (RQ), rule of law (RL), and control of corruption (CC), respectively. - For composite countries, the average of the component countries’ values is calculated. If one of the component countries had missing data, the average of the remaining countries is used. - For Germany, data corresponds to West Germany’s data in 1985 and 1990 and to United Germany afterwards. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 231 - For Yugoslavia, data corresponds to Yugoslavia (former Yugoslavia) in 1985 and 1990 and to Serbia/Montenegro afterwards. The 6 governance components were used to form several composite governance variables that were experimented with before deciding on which one to use. Namely: Gov_w, Gov_s, Gov_dw and Gov_ds. The choice of the governance variable that was used in all regressions (Gov_ds) was based on the fact that it has no high collinearity with GDPPC thus excluding Gov_w and Gov_s, and that it is more theoretically appealing and thus makes more sense than Gov_dw. Note that in Appendix C some of the sample regressions using the four governance choices are presented to prove that the choice of Gov_ds was the best and that nevertheless other variables experienced only very minor changes. Detailed steps for calculating the 4 governance variables 1) We obtained the 6 different components of governance for each country. Those being 1) VA (Voice and Accountability), 2) PS (Political Stability), 3) GE (Government Effectiveness), 4) RQ (Regulatory Quality), 5) RL (Rule of Law), and 6) CC (Control of Corruption) as mentioned in the main text. These components were then analyzed using Principal Components Analysis to obtain statistical weights for each component (to see in what proportions they should be added). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 232 2) To get Gov_w (weighted governance) for each country, weights from point 1 above were multiplied each with its appropriate governance component. To get Gov_s, the standardized values for each component were calculated. Each was then multiplied by its respective weight from point 1. These were then added up to form Gov_s, which is a standardized weighted sum variable. 3) Then separately, Gov_w and Gov_s of the individual countries were regressed on GDPPC for each country. 4) The predict command was then used to get the residual, which in this case is the difference between the actual Gov_w (Gov_s) and the predicted Gov_w (Gov_s). Note, we didn't use percentages or absolute values. We got the difference as it is, be it +ve or _ve difference. 5) Again separately, pairing what we got from point 4 above for every pair of countries we got Gov_dw (Gov_ds). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 233 Appendix C: Experimentation with the Four Governance Variables In the main text, the variable Gov_ds is used all through. The objective of this appendix is to show that using any of the other three governance variables (Gov_w, Gov_s, and Gov_dw) would not have caused any major changes in the signs of almost all the coefficients or in their significance (except with Gov_w or Gov_s and GDPPC for reasons mentioned earlier and that’s why they were excluded). As you can see, columns 2 and 6 in Tables C.1& C.2 are the same as the last four columns in Table 3.1, while the rest of the columns are just experimenting with the different governance variables. Notice that there are no changes in the signs of any of the variables (except GDPPC because of its high collinearity with Gov_w and Gov_s). There are some but very minor changes in the coefficients and the significance of the other variables, which can be disregarded. However, in Table C.2, the last four columns show relatively significant changes in the Gini variable’s coefficient and significance under Gov_w and Gov_s. But this difference can also be disregarded for two reasons: First, Gini has shown sensitivity to almost all the assumptions, estimation techniques and specifications experimented with and thus it is expected that whatever governance variable is used the Gini variable might show different results. Second, the Gini coefficient is not even significant at the 44/45% level (under Gov_w and Gov_s) or at the 92/97% level (under Gov_dw and Gov_ds), which might tell us that its effect on trade flows is negligible after all. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 234 Similarly, columns 2 and 6 in Tables C.3 & C.4 are the same as the last four columns in Table 3.2, while the rest of the columns are again experimenting with the different governance variables. Notice that the changes that occurred to GDPPC this time are in its significance and not its sign. There are significant changes concerning the border variable; however, these changes can be disregarded because the border variable shows no significance even at the 70% level, which means that its effect on trade is negligible and so it does not really matter whether its coefficients show positive or negative signs. The same logic applies to CPE. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table C. 1. Regression results for estimating the gravity model for World total bilateral trade using OLS, where null values in the dependent variable are assumed genuinely missing, and where different governance variables are used. Pooled (1985-1997) Coef. (RStd.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) gdp I 0 7 *** 1.07 *** 1.06 *** 1.06 *** 1.08 *** 1 08 *** I 0 7 *** 1 Q 7 *** (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) gdppc 0.03 ** 0.03 ** -0 .0 2 -0 .0 2 0.03 ** 0.03 ** -0.05 *** -0.05 *** (0 .0 2 ) (0 .0 2 ) (0 .0 2 ) (0 .0 2 ) (0 .0 2 ) (0 .0 2 ) (0 .0 2 ) (0 .0 2 ) distance -1 41 *** -1 41 *** -1.40 *** -1.40 *** -1.42 *** -1.42 *** -1.42 *** -1.42 *** (0.03) (0.03) (0.03) (0.03) (0.03) (0.03) (0.03) (0.03) areas -0.15 *** -0 15 *** -0.15 *** -0.15 *** -0.15 *** -0.15 *** -0.16 *** -0.16 *** (0 .0 2 ) (0 .0 2 ) (0 .0 2 ) (0 .0 2 ) (0 .0 2 ) (0 .0 2 ) (0 .0 2 ) (0 .0 2 ) 11 -0.34 *** -0.34 *** -0.38 *** -0.38 *** -0.33 *** -0.33 *** -0.39 *** -0 40 *** (0.04) (0.04) (0.04) (0.04) (0.04) (0.04) (0.04) (0.04) border 0.57 *** 0.57 *** 0.61 *** 0.61 *** 0.55 *** 0.55 *** 0.60 *** 0.60 *** (0 .1 2) (0 .1 2) (0 . 1 2) (0 .1 2) (0 .1 2) (0 .1 2) (0 .1 2 ) (0 .1 2) language 0.58 *** 0.58 *** 0.57 *** 0.57 *** 0.55 *** 0.55 *** 0.53 *** 0.53 *** (0.06) (0.06) (0.06) (0.06) (0.06) (0.06) (0.06) (0.06) regional 1 33 *** 1 33 *** 1.36 *** 1 36 *** 1.30 *** 1.30 *** 1.33 *** 1.33 *** (0.14) (0.14) (0.14) (0.14) (0.14) (0.14) (0.14) (0.14) nation (D) (D) (D) (D) (D) (D) CD) (D) colonizer 0.57 *** 0.57 *** 0.539 *** 0.54 *** 0.58 *** 0.58 *** 0.54 *** 0.55 *** (0.08) (0.08) (0.08) (0.08) (0.08) (0.08) (0.08) (0.08) colonial 0.62 * 0.63 * 0.61 * 0.62 * 0.59 0.60 * 0.57 0.58 (0.36) (0.36) (0.36) (0.36) (0.36) (0.36) (0.36) (0.36) erv -0.63 *** -0.63 *** -0.63 *** -0.64 *** (0.24) (0.24) (0.24) (0.24) erv_'85 I 97 *** 2 01 *** 2 .2 2 *** 2.25 *** (0.63) (0.63) (0.64) (0.64) Pooled (1985-1997) w. erv interaction to U ) cn Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table C. variable 1. Regression results for estimating the gravity model for World total bilateral trade using OLS, where null values in the dependent are assumed genuinely missing, and where different governance variables are used (continued). Pooled (1985-1997) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) erv_'90 0.10 0.09 0.22 0.19 (0.35) (0.35) (0.36) (0.35) erv_'95 -1.01 *** -1.05 *** -1.04 *** -1.09 *** (0.30) (0.30) (0.30) (0.30) erv_'97 -2.36 *** -2.28 *** -2.59 *** -2.48 *** (0.46) (0.45) (0.47) (0.46) cu 1.82 *** 1.82 *** I gj *** 1.80 *** 1 3 7 *** 1.86 *** 1.86 *** 1.85 *** (0.22) (0.22) (0.22) (0.22) (0.22) (0.22) (0.22) (0.22) gov_dw 0.04 *** 0.05 *** (0.004) (0.004) gov_ds 0.10 *** 0.13 *** (0.01) (0.01) gov_w 0 Q4 *** 0.05 *** (0.01) (0.01) gov_s 0 .11 *** 0 1 4 *** (0.01) (0.01) cpy 0.19 0.21 0.20 0.22 0.08 0.10 0.09 0.11 (0.18) (0.18) (0.18) (0.18) (0.19) (0.19) (0.19) (0.19) cpe -0.11 ** -0.12 ** -0.08 * -0.09 * -0.06 -0.07 -0.03 -0.03 (0.05) (0.05) (0.05) (0.05) (0.05) (0.05) (0.05) (0.05) gini -0.0014 -0.0014 -0.0003 -0.0002 0.0011 0.0012 0.0031 * 0.0031 * (0.002) (0.002) (0.002) (0.002) (0.002) (0.002) (0.002) (0.002) m2gdp 0.0002 *** 0.0002 *** 0 .0 0 0 1 *** 0 .0 0 0 1 *** 0.0002 *** 0.0002 *** 0 .0 0 0 1 *** 0 .0 0 0 1 *** (0.00001) (0.00001) (0.00001) (0.00001) (0.00001) (0.00001) (0.00001) (0.00001) current 0.06 * 0.06 * 0.07 ** 0.08 ** 0.02 0.02 0.05 0.05 (0.03) (0.03) (0.03) (0.03) (0.03) (0.03) (0.03) (0.03) Pooled (1985-1997) w. erv interaction 236 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table C. 1. Regression results for estimating the gravity model for World total bilateral trade using OLS, where null values in the dependent variable are assumed genuinely missing, and where different governance variables are used (continued). Pooled (1985-1997) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (RStd.Err) Coef. (RStd.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) capital -0.41 *** -0 40 *** -0.40 *** -0.40 *** -0 38 *** -0 37 *** -0.38 *** -0 37 *** (0.04) (0.04) (0.04) (0.04) (0.04) (0.04) (0.04) (0.04) _cons -22.21 *** -22.26 *** -22.39 *** -21 14 *** -22.80 *** -22.86 *** -23.08 *** -21.36 *** (0.52) (0.52) (0.52) (0.55) (0.53) (0.53) (0.53) (0.55) No. of Obs 9776 9776 9776 9776 9776 9776 9776 9776 R-squared 0.71 0.71 0.71 0.71 0.71 0.71 0.71 0.71 RMSE 1.90 1.90 1.90 1.90 1.89 1.89 1.90 1.89 Pooled (1985-1997) w. erv interaction Notes: 1) D stands for dropped. 2) Standard errors are beneath the coefficients. 3) *, * denotes significance at 1, 5, and 10% respectively. 237 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table C.2. Regression results for estimating the gravity model for World total bilateral trade using variable are assumed genuinely missing, and where different governance variables are used. OLS, where null values in the dependent Pooled (1985-1997) w. colonial interaction Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Pooled Coef. (R.Std.Err) (1985-1997) Coef. (R.Std.Err) w. border interaction Coef. (R.Std.Err) Coef. (R.Std.Err) g d p gdppc distance areas 1 1 border border_'85 border_'90 borderJ95 border_'97 language regional nation I 07 * * * (0.01) 0.03 ** (0.02) -I 41 *** (0.03) -0.15 *** (0.02) -0.34 *** (0.04) 0.57 *** (0.12) 0.58 *** (0.06) 1 33 *** (0.14) (D) 1.07 *** (0.01) 0.03 ** (0.02) -1.41 *** (0.03) -0.15 *** (0.02) -0.34 *** (0.04) 0.57 *** (0.12) 0.58 *** (0.06) 1.33 *** (0.14) (D) 1.06 *** (0.01) -0.02 (0.02) -1 40 *** (0.03) -0.15 *** (0.02) -0 38 *** (0.04) 0.61 *** (0.12) 0.57 *** (0.06) 1.36 *** (0.14) (D) 1.06 *** (0.01) -0.02 (0.02) -1 40 *** (0.03) -0.15 *** (0.02) -0.38 *** (0.04) 0.61 *** (0.12) 0.57 *** (0.06) 1.36 *** (0.14) (D) I 07 *** (0.01) 0.03 ** (0.02) -141 *** (0.03) -0.15 *** (0.02) -0.33 *** (0.04) 0.87 (0.21) -0.23 (0.37) (D) -0.37 (0.29) -0.47 * (0.26) 0.58 *** (0.06) 1.33 *** (0.14) (D) ^ 07 *** (0.01) 0.03 ** (0.02) -1 41 *** (0.03) -0.15 *** (0.02) -0.34 *** (0.04) 0.87 (0.21) - 0.22 (0.37) (D) -0.38 (0.29) -0.47 * (0.26) 0.58 *** (0.06) 1.33 *** (0.14) (D) 1.06 *** (0.01) -0.02 (0.02) -1 40 *** (0.03) -0.15 *** (0.02) -0.38 *** (0.04) 0.89 (0.21) -0.23 (0.37) (D) -0.35 (0.28) -0.45 * (0.26) 0.57 *** (0.06) 1.35 *** (0.14) (D) 1.06 *** (0.01) -0.03 (0.02) -1 40 *** (0.03) -0.15 *** (0.02) -0.38 *** (0.04) 0 89 * * * (0.21) -0.23 (0.37) (D) -0.35 (0.28) -0.44 * (0.26) 0.57 *** (0.06) 1.36 *** (0.14) (D) to U J 00 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table C.2. Regression results for estimating the gravity model for World total bilateral trade using OLS, where null values in the dependent variable are assumed genuinely missing, and where different governance variables are used (continued). Pooled (1985-1997) w. colonial interaction Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (RStd.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (RStd.Err) colonizer 0.57 *** 0.57 *** 0.54 *** 0.54 *** 0.57 *** 0.57 *** 0.54 *** 0.54 *** (0.08) (0.08) (0.08) (0.08) (0.08) (0.08) (0.08) (0.08) colonial 1 .1 2 ** 1 .1 2 ** j j 7 ** 1.18 ** 0.62 * 0.63 * 0.61 * 0.62 * (0.52) (0.52) (0.50) (0.50) (0.36) (0.36) (0.36) (0.36) colonial_i -0.33 -0.33 -0.38 -0.37 (0.30) (0.30) (0.29) (0.29) erv -0.63 *** -0.63 *** -0.63 *** -0.64 *** -0.64 *** -0.64 *** -0 64 *** -0.64 *** (0.24) (0.24) (0.24) (0.24) (0.24) (0.24) (0.24) (0.24) cu 1 82 *** 1.82 *** 1 81 *** 1 80 *** 2 gj *** 1.81 *** 1.80 *** 1.80 *** (0 .2 2 ) (0 .2 2 ) (0 .2 2 ) (0 .2 2 ) (0 .2 2 ) (0 .2 2 ) (0 .2 2 ) (0 .2 2 ) gov_dw 0.038 *** 0.038 *** (0.004) (0.004) gov_ds 0 jo *** 0 2 1 *** (0 .0 1 ) (0 .0 1 ) gov_w 0.039 *** 0.040 *** (0.005) (0.005) gov_s q j j * * * q jj ** * (0 .0 1 ) (0 .0 1 ) cpy 0.19 0 .2 1 0 .2 0 0 .2 2 0.19 0 .2 2 0 .2 0 0 .2 2 (0.18) (0.18) (0.18) (0.18) (0.18) (0.18) (0.18) (0.18) cpe -0 .1 1 ** -0 .1 2 ** -0.08 * -0.09 * -0 .1 1 ** -0 .1 2 ** -0.08 -0.09 * (0.05) (0.05) (0.05) (0.05) (0.05) (0.05) (0.05) (0.05) gini -0.0014 -0.0014 -0 .0 0 0 2 -0 .0 0 0 1 -0.0014 -0.0013 -0 .0 0 0 2 -0 .0 0 0 1 (0.0018) (0.0018) (0.0018) (0.0018) (0.0018) (0.0018) (0.0018) (0.0018) m2 gdp 0 .0 0 0 2 *** 0 .0 0 0 2 *** 0 .0 0 0 1 *** 0 .0 0 0 1 *** 0 .0 0 0 2 *** 0 .0 0 0 2 *** 0 .0 0 0 1 *** 0 .0 0 0 1 *** (0 .0 0 0 0 1 ) (0 .0 0 0 0 1 ) (0 .0 0 0 0 1 ) (0 .0 0 0 0 1 ) (0 .0 0 0 0 1 ) (0 .0 0 0 0 1 ) (0 .0 0 0 0 1 ) (0 .0 0 0 0 1 ) Pooled (1985-1997) w. border interaction 239 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table C.2. Regression results for estimating the gravity model for World total bilateral trade using OLS, where null values in the dependent variable are assumed genuinely missing, and where different governance variables are used (continued). Pooled (1985-1997) w. colonial interaction Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) current 0.06 * 0.06 * 0.07 ** 0.08 ** 0.06 * 0.06 * 0.07 ** 0.08 ** (0.03) (0.03) (0.03) (0.03) (0.03) (0.03) (0.03) (0.03) capital -0.41 *** -0 40 *** -0 40 *** -0 39 *** -0 41 *** -0 40 *** -0 40 *** “0 40 *** (0.04) (0.04) (0.04) (0.04) (0.04) (0.04) (0.04) (0.04) _cons -2 2 .2 2 *** -22.27 *** -22.40 *** -21.15 *** -22.23 *** -22.28 *** -22.41 *** -21 14 *** (0.52) (0.52) (0.52) (0.55) (0.52) (0.52) (0.52) (0.55) No. of Obs 9776 9776 9776 9776 9776 9776 9776 9776 R-squared 0.71 0.71 0.71 0.71 0.71 0.71 0.71 0.71 Root MSE 1.90 1.90 1.90 1.90 1.90 1.90 1.90 1.90 Pooled (1985-1997) w. border interaction Notes: 1) D stands for dropped. 2) Standard errors are beneath the coefficients. 3) ***, **, * denotes significance at 1, 5, and 10% respectively. 240 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table C.3. Regression results for estimating the gravity model for World total bilateral trade using OLS, where null values in the dependent variable are assumed zeros, and where different governance variables are used. Pooled (1985-1997) Pooled (1985-1997) w. erv interaction Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) gdp 2.48 *** 2.48 *** 2,47 *** 2.47 *** 2 49 *** 2,49 *** 2,47 *** 2 4 7 *** (0.05) (0.05) (0.05) (0.05) (0.05) (0.05) (0.05) (0.05) gdppc -0.06 -0.05 -0.25 *** -0.23 *** -0.07 -0.06 -0.29 *** -0.26 *** (0.05) (0.05) (0.06) (0.06) (0.05) (0.05) (0.06) (0.06) distance -2 90 *** -2 89 *** -2.87 *** -2 .8 6 *** -2.92 *** _2 91 *** -2 .8 8 *** -2 .8 6 *** (0 .1 0) (0 .1 0) (0 .1 0) (0 . 1 0) (0 -1 0) (0 .1 0) (0 .1 0) (0 . 1 0) areas -0 93 *** -0 92 *** -0.94 *** -0 93 *** -0 92 *** -0 92 *** -0.93 *** -0 93 *** (0.07) (0.07) (0.07) (0.07) (0.07) (0.07) (0.07) (0.07) 11 -0 72 *** -0 72 *** -0 .8 8 *** -0 .8 8 *** -0 .6 8 *** -0.69 *** -0.87 *** -0 87 *** (0.15) (0.15) (0.15) (0.15) (0.15) (0.15) (0.15) (0.15) border -0.14 -0.13 0 .0 2 0.03 -0.18 -0.16 0 .0 1 0 .0 2 (0.38) (0.38) (0.38) (0.38) (0.38) (0.38) (0.38) (0.38) language 2.03 *** 2 03 *** 2 .0 2 *** 2 .0 2 *** 1 99 *** 2 .0 0 *** 1 93 *** 2 .0 0 *** (0 .2 0 ) (0 .2 0 ) (0 .2 0 ) (0 .2 0 ) (0 .2 0 ) (0 .2 0 ) (0 .2 0 ) (0 .2 0 ) regional 0.74 * 0.75 * 0.84 * 0 .8 6 * 0.69 0.71 0.82 * 0.84 * (0.44) (0.44) (0.45) (0.45) (0.44) (0.44) (0.45) (0.45) nation (D) (D) (D) (D) (D) (D) (D) (D) colonizer 2 .1 0 *** 2 .1 2 *** 2 03 *** 2.05 *** 2 .1 1 *** 2.13 *** 2.03 *** 2.05 *** (0.27) (0.27) (0.27) (0.27) (0.27) (0.27) (0.27) (0.27) colonial -2.99 ** -2.95 ** -3.07 ** -3.03 ** -3.03 ** -2.98 ** -3.10 ** -3.05 ** (1.30) (1.30) (1.31) (1.31) (1.30) (1.30) (1.31) (1-31) erv -5.44 *** -5.49 *** -5.63 *** -5.69 *** (0 .8 8 ) (0 .8 8 ) (0.89) (0 .8 8 ) erv_'85 -7 07 *** j-j j 2 *** -7.58 *** -7 83 *** (2.23) (2.24) (2.26) (2.27) to 4 ^ Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table C. variable 3. Regression results for estimating the gravity model for World total bilateral trade using OLS, where null values in the dependent are assumed zeros, and where different governance variables are used (continued). Pooled (1985-1997) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (RStd.Err) Coef. (R.Std.Err) Pooled (1985-1997) w. erv interaction Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) erv_'90 erv_'95 erv_'97 cu gov_dw gov_ds gov_w gov_s cpy cpe gini m2 gdp current 4 92 * * * (0.67) 0.18151 *** (0.02) -2.15 ** (0.84) -0.09 (0.17) - 0.02 * * * (0.01) - 0.0001 (0.00004) 0.24 ** (0-11) 4 39 *** (0.67) 0.469 (0.04) -2.14 ** (0.85) - 0.10 (0.17) - 0.02 * * * (0.01) -0.0001 (0.00004) 0.23 ** (0 .11) 4.82 *** (0.67) 0 14 *** (0.02) -2.43 *** (0.85) 0.14 (0.17) - 0.02 * * * (0.01) - 0.0002 (0.00004) 0.24 ** (0.12) 4.80 *** (0.67) 0.352 *** (0.05) -2.46 *** (0.85) 0.15 (0.17) - 0.02 * * * (0.01) - 0.0002 (0.00004) 0.22 * (0.12) -2.74 ** ( 1.20) -6.09 (1.18) -9.27 (1.61) 4 9| *** (0.67) 0.20439 *** (0.02) - 2.01 * * (0.85) -0.08 (0.18) - 0.02 * * * (0.01) -0.0001 (0.00004) 0.21 * (0.12) -3.02 ** ( 1.20) - 6.22 * * * (1.18) - 8.66 * * * (1.60) 4.88 *** (0.67) 0.518 *** (0.04) - 2.00 * * (0.85) -0.10 (0.18) - 0.02 * * * (0.01) -0.0001 (0.00004) 0.20 * (0.12) -3.52 *** ( 1.22) -6.07 *** (1.19) -8.40 *** (1.65) 4.81 *** (0.67) 0.16 *** (0.02) -2 27 *** (0.85) 0.13 (0.17) - 0.02 * * * (0.01) - 0.0002 (0.00004) 0.24 ** (0 .12) -3 92 *** (1.22) -6 14 *** (1.19) -7.65 *** (1.63) 4.78 *** (0.67) 0.386 (0.05) -2.32 (0.86) 0.14 (0.17) - 0.02 (0.01) - 0.0002 (0.00004) 0.22 (0.12) 242 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table C.3. Regression results for estimating the gravity model for World total bilateral trade using OLS, where null values in the dependent variable are assumed zeros, and where different governance variables are used (continued). Pooled (1985-1997) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) capital -I 38 *** -1 33 *** -1.31 *** -1.26 *** -1 33 *** -1.29 *** -1 27 *** -1.23 *** (0.13) (0.13) (0.13) (0.13) (0.14) (0.14) (0.14) (0.14) _cons -64.15 *** -64.48 *** -65.56 *** -61.46 *** -64.52 *** -64.79 *** -65.83 *** -61.17 *** (1.82) (1.82) (1.83) (1.92) (1.85) (1.85) (1.87) (1.93) No. of Obs 11432 11432 11432 11432 11432 11432 11432 11432 R-squared 0.48 0.48 0.48 0.48 0.48 0.48 0.48 0.48 Root MSE 7.22 7.22 7.24 7.24 7.21 7.21 7.24 7.24 Pooled (1985-1997) w. ctv interaction Notes: 1) D stands for dropped. 2) Standard errors are beneath the coefficients. 3) ***, **, * denotes significance at 1, 5, and 10% respectively. 243 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table C.4. Regression results for estimating the gravity model for World total bilateral trade using variable are assumed zero, and where different governance variables are used. Pooled (1985-1997) w. colonial interaction Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) gdp 2,49 *** 2.48 *** 2.47 *** 2.47 *** 2 4 9 *** 2 .4 9 *** 2 47 2 47 *** (0.05) (0.05) (0.05) (0.05) (0.05) (0.05) (0.05) (0.05) gdppc -0.06 -0.05 -0.25 *** -0 23 *** -0.06 -0.05 -0.25 *** -0 23 *** (0.05) (0.05) (0.06) (0.06) (0.05) (0.05) (0.06) (0.06) distance -2 90 *** -2.89 *** -2 87 *** -2 .8 6 *** -2 90 *** -2.89 *** -2.87 *** -2 .8 6 *** (0 .1 0 ) (0 .1 0) (0 .1 0) (0 . 1 0) (0 .1 0 ) (0 .1 0 ) (0 .1 0) (0 .1 0) areas -0 93 *** -0.92 *** - 0 9 4 *** -0 93 *** -0 93 *** -0.92 *** -0.94 *** -0 93 *** (0.07) (0.07) (0.07) (0.07) (0.07) (0.07) (0.07) (0.07) 11 -0 72 *** -0.72 *** -0 .8 8 *** -0 .8 8 *** -0 71 *** -0.72 *** -0 .8 8 *** -0 .8 8 *** (0.15) (0.15) (0.15) (0.15) (0.15) (0.15) (0.15) (0.15) border -0.14 -0.13 0 .0 2 0.03 0.98 0.95 0.97 0.91 (0.38) (0.38) (0.38) (0.38) (0.85) (0.85) (0.85) (0.85) border_'85 -0.29 -0.26 -0.24 -0 .2 0 (1.43) (1.43) (1-41) (1.41) border_'90 (D) (D) (D) (D) border_'95 -1.38 -1.35 -1.14 -1.09 (0.97) (0.97) (0.96) (0.96) border_'97 -2.13 ** -2 .0 2 ** -1.80 * -1.67 * (0.95) (0.95) (0.94) (0.94) language 2.03 *** 2.03 *** 2 .0 2 *** 2 .0 2 *** 2 .0 2 *** 2 03 *** 2 .0 1 *** 2 .0 1 *** (0 .2 0 ) (0 .2 0 ) (0 .2 0 ) (0 .2 0 ) (0 .2 0 ) (0 .2 0 ) (0 .2 0 ) (0 .2 0 ) regional 0.74 * 0.75 * 0.85 * 0 .8 6 * 0.72 * 0.74 * 0.83 * 0.85 * (0.44) (0.44) (0.45) (0.45) (0.44) (0.44) (0.44) (0.45) nation (D) (D) (D) (D) (D) (D) (D) (D) OLS, where null values in the dependent Pooled (1985-1997) w. border interaction 244 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table C.4. Regression results for estimating the gravity model for World total bilateral trade using OLS, where null values in the dependent variable are assumed zero, and where different governance variables are used (continued). Pooled (1985-1997) w. colonial interaction Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) colonizer 2 .1 0 *** 2 .1 2 *** 2.03 *** 2.05 *** 2 .1 0 *** 2 1 2 *** 2 03 *** 2.05 *** (0.27) (0.27) (0.27) (0.27) (0.27) (0.27) (0.27) (0.27) colonial -1.59 -1.58 -1.44 -1.44 -2.99 ** -2.95 ** -3.07 ** -3.03 ** (2.04) (2.04) (1.98) (1.98) (1.30) (1.30) (1.31) (1.31) colonial_i -0.94 -0.92 -1.08 -1.06 ( 1.1 1) ( 1. 1 0) (1 .1 0) ( 1.1 0) erv -5 44 *** -5.48 *** -5.63 *** -5.69 *** -5.46 *** -5.50 *** -5.65 *** -5.71 *** (0 .8 8 ) (0 .8 8 ) (0.89) (0 .8 8 ) (0.89) (0.89) (0.89) (0.89) cu 4.92 *** 4.89 *** 4.83 *** 4.80 *** 4.88 *** 4.86 *** 4 7 9 *** 4 7 7 *** (0.67) (0.67) (0.67) (0.67) (0.67) (0.67) (0.67) (0.67) gov_dw 0.18 *** Q Jg *** (0 .0 2 ) (0 .0 2 ) gov_ds 0.47 *** 0.47 *** (0.04) (0.04) gov_w q 24 *** 0.15 *** (0 .0 2 ) (0 .0 2 ) gov_s 0.35 *** 0.36 *** (0.05) (0.05) cpy -2.15 ** -2.14 ** -2.43 *** -2.46 *** -2.16 ** -2.14 ** -2.43 *** -2 47 *** (0.84) (0.85) (0.85) (0.85) (0.84) (0.85) (0.85) (0.85) cpe -0.09 -0 .1 0 0.13 0.15 -0.08 -0 .1 0 0.14 0.15 (0.17) (0.17) (0.17) (0.17) (0.17) (0.17) (0.17) (0.17) gini -0 .0 2 *** -0 .0 2 *** -0 .0 2 *** -0 .0 2 *** -0 .0 2 *** -0 .0 2 *** -0 .0 2 *** -0 .0 2 *** (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) (0 .0 1 ) m2 gdp -0.0001 *** -0.0001 *** -0 .0 0 0 2 *** -0 .0 0 0 2 *** -0.0001 *** -0.0001 *** -0 .0 0 0 2 *** -0 .0 0 0 2 *** (0.00004) (0.00004) (0.00004) (0.00004) (0.00004) (0.00004) (0.00004) (0.00004) Pooled (1985-1997) w. border interaction to Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table C.4. Regression results for estimating the gravity model for World total bilateral trade using OLS, where null values in the dependent variable are assumed zero, and where different governance variables are used (continued). Pooled (1985-1997) w. colonial interaction Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) Coef. (R.Std.Err) current 0.24 ** 0.23 ** 0.24 ** 0 .2 2 * 0.24 ** 0.23 ** 0.24 ** 0 .2 2 * (0 .1 1) (0 .1 1) (0 .1 2) (0 .1 2) (0 .1 1) (0 .1 1) (0 .1 2) (0 .1 2) capital -1 37 *** -1.33 *** -1 30 *** -1.26 *** -1 38 *** -1.33 *** -1 30 *** -1.26 *** (0.13) (0.13) (0.13) (0.13) (0.13) (0.13) (0.13) (0.13) _cons -64.18 *** -64.50 *** -65.59 *** -61.48 *** -64.25 *** -64.57 *** -65.65 *** -61.48 *** (1.82) (1.82) (1.83) (1.92) (1.82) (1.82) (1.83) (1.92) No. ofObs 11432 11432 11432 11432 11432 11432 11432 11432 R-squared 0.48 0.48 0.48 0.48 0.48 0.48 0.48 0.48 Root MSE 7.22 7.22 7.24 7.24 7.22 7.22 7.24 7.24 Pooled (1985-1997) w. border interaction Notes: 1) D stands for dropped. 2) Standard errors are beneath the coefficients. 3) ***, **, * denotes significance at 1, 5, and 10% respectively. 246 247 Appendix D: Some Important Aspects of GAFTA Concerning liberalization of industrial products, member states are permitted to draw up a list of exceptions from tariff reductions during the first years of the program. The purpose of the exceptions is to allow local industry to restructure and improve its competitiveness before having to face competition from other GAFTA countries’ imports. However, the member state requesting such exceptions should submit a report showing in full the economic impact on the national economy and the duration for each requested good to be excepted from the tariff and tariff-like reductions. With respect to agriculture, the executive program offers members the opportunity to suspend tariff reductions on some produce during the peak harvest seasons. Each GAFTA country is permitted to propose ten produce items for suspension, with a total exemption for all the items of 45 months. The most serious shortcomings of the GAFTA program are the loopholes in the safeguard measures that give member states the right to exclude certain industrial and agriculture products from immediate liberalization as discussed above. (ERF 1998 and 2000; and Zarrouk 2000). Regarding rules of origin, the program offers rules of origin for duty-free treatment. The GAFTA value added requirement is set at 40%. There are two methods for calculating origin. The first is based on the local value added approach, while the other one is based on the net cost approach, which subtracts specified imported expenses from the transaction price to determine the base for calculating the Reproduced w ith permission of the copyright owner. Further reproduction prohibited without permission. 248 ratio of foreign to domestic content. An important characteristic of the program is the ongoing scheme for the elaboration of detailed preferential rules of origin for GAFTA- made products. The program also promotes the need for harmonization of preferential rules of origin to comply with the Euro-Mediterranean free trade agreements underway (ERF 2000; and Zarrouk 2000). The executive program calls for the private sector - chambers of industry and commerce in Arab countries - to monitor the implementation of the different stages of the program. This is an innovation, which aims at enhancing the transparency of the Arab FTA (ERF 1998 and 2000; and Zarrouk 2000). GAFTA is primarily concerned with freeing trade in goods. This is not sufficient to ensure that GAFTA members will be able to confront competition on world markets. Deeper integration is thus required. Worldwide, integration at the regional level is constantly extending beyond the trade of goods; in particular trade in services is becoming a well-established pattern of integration. Arab countries should consider liberalizing trade in services among all of them. As a matter of fact, the ERF (1998) argued that the liberalization of trade in services is in many ways fundamental to the smooth functioning of a regional integration agreement. The ERF also found it advisable to give the issue of integration of the service sectors a higher priority from the beginning of the regional negotiations rather than postponing it, since services, in particular financial services, telecommunications and transport, are depicted as the backbone of economic activity. The GAFTA members also need to extend liberalization to service - related movement of natural persons, and should also Reproduced w ith permission of the copyright owner. Further reproduction prohibited without permission. 249 consider mutual recognition of regulatory regimes relating to mandatory standards - for product safety, professional certification, pmdential regulation, and so on. Reproduced w ith permission of the copyright owner. Further reproduction prohibited without permission.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

The outcomes of trade conflicts and negotiations between the United States and South Korea, 1980--2000

PDF

Political economy of economic reform, quality of reform and economic performance: Reform matters, but consistent incentives matter more

PDF

The implementation of endangered species policy: An analysis of the 'takings' issue in specific contexts

PDF

Carrots or sticks: The Taiwanese government's contrasting policy approaches to banking sector liberalization

PDF

An empirical analysis of foreign aid and the provision of international public goods

PDF

Analysis of the role of government in Taiwan's industrialization and economic development

PDF

The Japanese main bank system: A transaction cost approach

PDF

Three essays on democracy and economic policy

PDF

Before and after reform: The effects of structural adjustment on women's employment in Latin America

PDF

Political economy of financial liberalization in emerging markets: A comparative study of South Korea and Taiwan in the 1990s

PDF

The political economy of procyclical fiscal policy in Mexico, 1970--1988

PDF

Trends in mortality in Turkey, 1960--1995

PDF

The role of coffee in the creation of institutions in Brazil: 1830--1930

PDF

The political economy of Turkish Islamism and secularism: The role of preference falsification in the establishment of Kemalist secularism and the polarization of Turkish public opinion on the pu...

PDF

Dependence on foreign labor, quality of education and unemployment in the GCC countries: In search of solutions

PDF

The fortunes of fishes and fishers: The political economy of innovation in Atlantic resource management

PDF

Essays in environmental policy

PDF

Understanding Federal Reserve enforcement behavior: Bureaucracy and regulatory structure in the United States banking industry

PDF

Globalization and the decline of the welfare state in less developed countries

PDF

By sea: The port nexus in the global commodity network (the case of the West Coast ports)

Asset Metadata

Creator Miniesy, Rania Samir (author)

Core Title The gravity model at work: An empirical study on the effects of governance on bilateral trade flows with special emphasis on intra-Arab trade

Degree Doctor of Philosophy

Degree Program Political Economy and Public Policy

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag Economics, General,OAI-PMH Harvest,political science, general,political science, international law and relations

Language English

Contributor Digitized by ProQuest (provenance)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c16-410538

Unique identifier UC11335561

Identifier 3145247.pdf (filename),usctheses-c16-410538 (legacy record id)

Legacy Identifier 3145247.pdf

Dmrecord 410538

Document Type Dissertation

Rights Miniesy, Rania Samir

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA