J Ö N K Ö P I N G I N T E R N A T I O N A L B U S I N E S S S C H O O L JÖNKÖPING UNIVERSITY. The Impact of the EU GSP. Countries Trade Flows

Transcription

1 J Ö N K Ö P I N G I N T E R N A T I O N A L B U S I N E S S S C H O O L JÖNKÖPING UNIVERSITY The Impact of the EU GSP Agreement on the Andean Countries Trade Flows Paper within Economics Author: Frida Gabrielsson-Kjäll & Maria Ädel Tutors: Associate Professor Martin Andersson & Ph.D Candidate Mikaela Backman Jönköping January 2010

2 Abstract The purpose of this thesis is to analyze the impact of the Generalized System of Preferences (GSP) agreement on the export from the Andean Community (AC) to the European Union (EU) between the years 1995 to The GSP agreement enables developing countries to face lower- or no tariffs when exporting to developed countries. According to Ricardian theory, Heckscher-Ohlin theory, and New Trade theory decreased trade barriers tend to have a positive effect on trade. When analyzing the trade flow between these countries using the gravity model the outcome is found to be consistent with the theories i.e the results show that the GSP agreement implemented in 1995 has had a positive impact on trade. i

3 Table of Contents 1 Introduction Purpose and Outline Generalized System of Preferences (GSP) Andean Community Bolivia Colombia Ecuador Peru Venezuela Theoretical Background The Ricardian Theory Heckscher-Ohlin Theory New Trade Theory Empirical Analysis of the GSP Agreement Method Results and Analysis Conclusion References Appendix Model Summary Anova Residual Statitics Correlation Matrix Individual graphs ii

4 1 Introduction International trade has increased at an enormous speed in recent times and it is well known that trade has assisted many countries in their process of economic development. Some of the main international trade theories such as the Ricardian theory, Heckscher-Ohlin theory, and New Trade theory claim that there is a positive relation between reduced trade barriers and trade flows. However, it is a topic of constant debate, partly since the competition between developed and underdeveloped countries is very unequal when trading at the worldmarket. This is a result of large differences in economic development, technology know how and infrastructure, to mention some of the causes. Many countries apply trade barriers such as tariffs, quotas, and embargos to protect their own industries from international competition. The European Union (EU) is today one of the leading world economic blocs. The union consists of 27 nations and is also one of the most integrated economic alliences in the world. The EU is a customs union, meaning it is a free trade zone for its members and has common external tariffs for non-members (European Commission, 2009a). As an aid and to improve the trade relations with developing countries, the EU 1 and other developed economies established the Generalized System of Preference (GSP) in 1971 (European Commission, 2008). The preference enables developing countries to face lower- or no tariffs on some goods exported to developed countries. The GSP is an exception from the general applied rules regarding international trade within the World Trade Organization (WTO) (Feenstra, 2003). The EU introduced a GSP agreement with the Andean Community (AC) in The countries included in the AC at this point were Bolivia, Colombia, Ecuador, Peru, and Venezuela. Half of the Andean population live below the poverty line and this region has one of the largest income differences in world (Europeiska ekonomiska och sociala kommittén, 2006). These developing countries have suffered from both economic and political instability and they have an inadequate export sector (Piñeres et. al 2000). In 1995 EU s GSP agreement was evaluated and renewed which meant that the AC faced new terms for its export to the EU. The new agreement included a graduation system, a removal of quotas, and rules concerning the sensitivity of the goods exported (UNCTAD, 1999). This thesis investigates whether the reduction of tariffs, as the 1995 GSP agreement indicated, is consistent with international trade theory that a reduction of trade barriers leads to increased trade. The interesting aspect of a unilateral agreement such as the GSP is whether such a scheme actually has a significant impact on the trade flow of developing countries. From these results one could further analyze whether the agreement in the long run contributes to economic growth and reduced poverty, as was the original idea of the scheme. In theory the reduction of tariffs on goods would lead to a greater demand for these products, which in turn could induce a growth in national productivity - one of the most important components in economic development. On the other hand, one could question whether the agreement rather, at the expense of developing countries, is beneficial to the developed countries that issued them since they are in fact the ones deciding what goods are included in the preferences. However, due to limitations this study the effects of the 1995 GSP on the export from the AC to the EU is analysed. 1 At this time European Community (EC) 1

5 1.1 Purpose and Outline The purpose of this study is to analyze the impact of the GSP agreement on the export from the AC to the EU during the years 1995 to Furthermore, there will be a discussion on what other factors may have contributed to changes in the export from these countries. Freres et al. (2004) discuss the GSP agreement of the EU and its impact on reducing poverty in Latin America. According to the authors, the GSP-exports from the Andean countries to the world increased relatively more than the GSP-exports to the EU. The authors conclude that the EU GSP agreement did not have a major impact on the AC. However, somewhat contradictive the authors still argue the importance of keeping the preferences. Freres et al. (2004) are some of the critics to the graduation system with which the EU GSP can be removed for those sectors that have grown and reached a certain level of competitiveness. In line with what Panagariya (2002) argued that this could have a reverse effect and decrease the incentives of the producers to grow economically. The limitation of the GSP is emphasized since there are several other factors affecting the economic development of a country. According to the authors the GSP would be most beneficial if adjusted to the individual status of each country. Panagariya (2002) further mentions that preferences such as the GSP can have an adverse effect as the recipient countries do not have to liberalize their own trade since it is a non-reciprocal agreement. Meaning, the countries may be better off in the long run would they also remove trade barriers so as to grow competitively on the world market. Contrary to this Krugman et al. (2006) discuss the infant industry argument, which is in favour of temporary trade barriers in order to boost industries competitiveness. The argument is well known, and some of the largest economies such as Germany and the US started their industrialization process while applying protectionist measures. Wacziarg et al. (2000) argue in favour of trade liberalization when presenting empirical evidence of economic growth as trade liberalization has been present. According to the authors, countries that opened up their economies during 1950 to 1998 had average annual growth rates of 1.5% higher than before the liberalization. The decrease in trade barriers also had a positive impact on investment rates at an approximate increase of %, according to their study. This confirms earlier studies that liberalization supports growth partly by contributing to physical capital accumulation. There are however difficulties measuring the exact relationship between growth and liberalization, knowing that factors such as political instability can have a large impact on the overall economic growth. Colombia was one of the countries in the study that experienced negative growth, a result considered strongly connected to political instability. This study aims at contributing to the debate by analyzing if the GSP agreement has had an effect on the export by explaining the results of a tariff reduction within three main trade theories. The purpose is to suggest either continuing with the GSP agreement or advocate a different approach in favour of boosting export led economic growth in the Andean region. As the list of products included in the GSP agreement is too complex and extensive it is not included in the thesis. The analysis could have been more extensive would the products have been included. 2

6 The thesis has the following outline: continuing in chapter 1 is some background information regarding the GSP agreement and the AC. Chapter 2 describes three main international trade theories and effects of reduced trade barriers, and additionally the hypothesis is presented. In chapter 3 an empirical analysis of the result is provided. Finally chapter 4 entails a conclusion to the study and suggestions for further research. 1.2 Generalized System of Preferences (GSP): The idea of the concept GSP began in 1964 when it was proposed at the first United Nations Conference on Trade and Development. A report containing ideas and rules for trade between nations was evaluated, it approved rules such as lower tariffs for imports to developed countries from developing countries. However, since this proposal opposed the General Agreement on Tariffs and Trade (GATT) an exception was needed which subsequently lead to the first GSP program in 1971 (Jackson, 1997). After World War II the GATT started to reduce trade barriers such as tariffs for its member countries. According to GATT all member states had to have the same tariffs for all members, hence only multilateral reductions are allowed. This is formulated in the principle most favoured nation and it means that all member countries should be treated equally. The GSP is an exception to this principal, it is a nonreciprocal agreement that was accepted in contravention of the GATT (Feenstra, 2003). Within the WTO one decides what countries are classified as developing countries. The EU can thereafter decide what goods shall be concerned and thereafter set the tariff after the price sensitivity of the good (Europeiska Kommissionen, 2004). The EU implemented the GSP system with developing countries in 1971 (European Commission, 2008). The GSP cooperation with the AC instigated in 1990 (Comunidad Andina, 2001) and a new common 2 GSP agreement for a ten-year period started in 1995, after a revision of the previous agreement in For instance, the new agreement included additional decisions such as, adjustment of the tariffs after the sensitivity of the good, a graduation system, and a removal of quotas. The graduation system entailed an exclusion of the product sectors that had grown competitively on the market, since the goods were then considered to no longer be in need of the GSP. Every three years each country is evaluated to determine whether the agreement should continue or if certain goods or services should be excluded and/or included in the agreement. Normally there is an overall evaluation of the GSP system every ten years (Europeiska Kommissionen, 2004). After a period of ten years the 1995 agreement was evaluated again. Among the changes that were made for the following 2005 agreement was the implementation of a new preferential scheme; GSP+. This agreement allows decreasing tariffs further for the developing countries that are extra fragile. Today the four countries that are currently members of the AC 3 have a GSP+ agreement since 2005 (European Commission, 2005). This thesis analyzes the agreement that was implemented in For all developing countries under the EU GSP system 3 Venezuela left the AC in

7 1.3 The Andean Community The Andean Group was created when the Cartagena Agreement was signed in May The main goal with the cooperation was to create a counterpart to the larger economies in Latin America. The member countries of the Andean group were at this point Bolivia, Chile, Colombia, Ecuador and Peru. Venezuela that became a member in 1973, decided to leave the union in Chile left the union The members goals were to create a customs union and collaborate in areas such as transport, communication and industrial development (Xirinachs et al. 2001). The objectives were among other things to decrease unemployment, reach sustainable economic growth, and a strong regional integration (Adkinsson, 2003). The five countries (Bolivia, Colombia, Ecuador, Peru, and Venezuela) experienced progress as well as setbacks during the first years as members of the Andean Group. In 1979 the Andean Tribunal was created to interpret policies, agreements, and solve disputes in the common market. This was intended to facilitate regional integration and enhance cooperation. The tribunal still is, and has been, an active international court. However, it has experienced difficulties due to lack of sufficient institutions and other regulatory frameworks that could have helped strengthen the collaboration for the trading partners. During the 1990 s several measures were taken towards a more integrated cooperation as well as to further liberalize and open up the economies in the Andean countries. The free trade zone among the members was created in 1993 and common external trade barriers were set in place two years later. Peru joined the free trade zone in 1997 and was fully integrated in 2005 (Loayza et al. 2005). Other important milestones in the development of the Andean Group were the Trujillo and the Sucre Protocol signed in 1996 and 1997 respectively. These agreements were meant to make important necessary institutional changes that would eventually lead to a supranational authority. By signing the protocol the Andean group subsequently changed their name to the Andean Community (Loayza et al. 2005). The community has experienced large problems with the integration among the members, and it has been argued that not all members can benefit equally from the trade arrangements due to large differences in economic size and trade flows. Because of this there have also been problems with implementing policies that could have been highly efficient for the community as a whole, but that have been turned down due to indifferences between the individual states. When commenting on the overall distribution problems, Adkinsson wrote: Although Bolivia has experienced some impressive gains in this area, it remains the weakest exporter in the bloc, which has no doubt contributed to its general dissatisfaction with the distribution of benefits within the integrative program (Adkinsson p ). With this in mind, the community did in fact prove to be one of the more successful cooperations concerning integration among Latin American countries, at least in its earlier years. The intra trade, as a share of the AC s total trade, showed an increase from 4.1 % in 1990 to 14.2% in 1998, indicating a progress in integration between the countries (Europeiska ekonomiska och sociala kommittén, 2006). One of the main problems for the completion of an efficient internal market has been the lack of a functioning transport- and communication system (Avery, 1983). Adkinsson 4

8 (2003) suggests that the lack of an efficient supranational authority and further regional integration has slowed down the progress experienced in previous years. The individual Andean countries all suffered from the debt crisis in the 1980 s, from which their economies have recovered differently. During the late 1980 s and early 1990 s market oriented policies were implemented in all Andean countries. During this time the countries also liberalized their economies and opened up to international trade. The countries grew economically between the years 1993 to 1997 but were affected by different economic and political crises during the late 1990 s (Steinfatt et al. 2001). The EU is the AC s second largest trading partner after the United States. The main export products from the AC to the EU are agricultural products, fuels, and mining products (European Commission, 2009b). Graph 1.1 below shows which of the countries within the AC were the largest exporters to the EU as percentage of their total export to show how dependent the countries were of the trade with the EU during this time. Colombia was the largest exporter of the AC to the EU between 1991 to 2001 (Comunidad Andina, 2003). Both Colombia and Peru were steadily in the lead of the exports to the EU, while Venezuela and Ecuador show the smallest share of export to the EU during these ten years. Which is an expected result since Venezuela mainly trades with the United States. Graph 1.1 AC export to the EU15 as percentage of total AC export Bolivia Bolivia is one of the least developed countries in Latin America and also the poorest country in the AC. Bolivia suffered heavily from the debt crisis in the 1980 s and even if the economy started recovering during the 1990 s, when market oriented policies were implemented, it continued to show weak economic progress. The political instability and the lack of an efficient agricultural production have been some of the reasons for the country s weak development. Some of the main export commodities from Bolivia are natural gas, soybeans, and soy products. Among the AC countries, Bolivia and Ecuador are the main exporters of primary products to the EU (Comunidad Andina, 2003). The lack of investments in some of the country s strongest natural resources such as mining and natural gas has lead to a weak export sector (Europeiska ekonomiska och sociala kommittén, 2006). 5

9 1.3.2 Colombia Colombia is one of the economically largest and most developed countries in the AC. Due to a more enhanced and stable political and economic structure, Colombia had a faster recovery from the debt crisis in the 1980 s than many other Latin American countries. The relatively strong growth in GDP is mainly explained by a development towards a more diversified export sector (Piñeres et al. 2000). However, the country is constantly fighting the widespread drugtraffic and the organized crimes that come with it. During the 1990 s Colombia suffered from both political and economic difficulties, for example the banking crisis 1998 (Creamer, 2003). According to Sheahan (2006) Colombia had a weaker economic growth than the other AC countries during the 1990 s. Some of the main export commodities are petroleum, nickel, and coal (CIA World Fact Book, 2009a) Ecuador Ecuador had many drastic experiences in the 1990 s. In the beginning of this decade they took on liberalizing policies for the financial sector and the capital account, as well as trade reforms to increase international trade. The country suffered a severe economic crisis in 1999 that lead the economy to collapse, this in turn lead to a floating exchange rate and eventually a full dollarization in 2000 (Vos et al. 2003). Both the urban and the rural poverty increased over the years 1995 to The poverty increased more along the coast than the rest of the country between the years 1995 to 1998 due to the El Niño 4 (SIISE, 2001). The country s main export is petroleum which accounts for half of the national income from trade. Throughout the late 1970 s and early 1980 s the oil prices were in favour of the Ecuadorian economy. However, in the later 1980 s and 1990 s the opposite occurred and together with other relative commodity prices the oil price decreased, leading to a deficit in the balance of trade (Vos et al. 2003). Besides petroleum, other export commodities are bananas, cut flowers, shrimp, cacao, coffee, hemp, wood, and fish (CIA World Factbook, 2009b). Since the mid 1990 s migration from the country has increased, a result of this was an increase in remittances which is one of the main contributors to the national income (Vos et al. 2003) Peru Like many developing countries Peru is dependent on the export of primary goods which makes it highly sensitive to price- and demand changes in the world market. The poverty was high in Peru in 1991 but decreased over some years throughout the 1990 s. However, in 2000 the poverty rate had increased again and was measured to be 54.9%. Peru liberalized its trade by eliminating export subsidies through the 1990 s. Between the years 1987 to 1991 the country suffered a tough crisis due to high inflation and large decreases in income (Sheahan, 2006). Even though Peru had an extensive liberalization throughout the 1990 s it was, as earlier mentioned, the last AC country to fully enter the free trade zone among the members in 2005 (Loayza et al. 2005). Peru s main export commodities are copper, gold, zinc, petroleum, coffee, potatoes, asparagus, textiles, and fishmeal (CIA World Factbook, 2009c). 4 A wheather phenomena by the coast of Southamerica that has extensive regional effects on the fishing- and agriculture industries (Nationalencyklopedin 2010). 6

10 1.3.5 Venezuela Venezuela is one of the richest Andean countries and has an economy highly dependent on petroleum. Since the early 1990 s the export of petroleum has driven the Venezuelan economy, and this high oil-dependence has made the country vulnerable to changes in the world market (Piñeres et al. 2000). Venezuela is the country in the AC that has received the most Foreign Direct Investment (FDI) from the EU during the 1990 s, with the overall highest values in 1993 and 1997 (Comunidad Andina, 2003). The Venezuelan economy has suffered from several economic crises, such as the banking crisis during the 1990 s (Creamer, 2003). During the beginning of this decade the poverty and the inequality differences in the country increased substantially (Sheahan, 2006). Venezuela left the AC in 2006 when Colombia and Peru decided to create a Free Trade of Americas Agreement with the United States (Comunidad Andina, 2003). Except for petroleum some of the main export commodities for Venezuela are bauxite, aluminum, steel, chemicals, agricultural products and basic manufacturing products (CIA World Factbook, 2009d). 2 Theoretical Background This chapter emphasizes how a tariff reduction affects trade between countries from a theoretical point of view. Three main international trade theories are presented. A trade barrier is a policy implemented by a government that restricts free trade between countries. Tariffs, quotas, embargos, and sanctions are examples of trade barriers (Law et al. 2008). Cultural and lingual differences are other factors that can hinder international trade since such differences make an international exchange more complicated. To analyze the effects and possible outcomes of international trade one has to approach the theories from the perspective of liberalization of trade and reduction of trade barriers. The trade barrier that is affected by the 1995 GSP-agreement is tariffs. A tariff is a tax levied on goods imported to a country to protect a domestic industry. It can be either an ad valorem which is a percentage of the goods imported or it can be a fixed amount (Bowen et al. 2003). Bowen et al. (2003) illustrate the different effects of adding a tariff on imports to a smalland a large country by using a two-good model. A tariff creates an increase in the domestic relative price of the imported good (Good 1), which subsequently leads to a lower domestic production of the good exported (Good 2) and an increase in production of the good sold on the domestic market (Good 1). This change of production in turn affects the country s GDP negatively. The main difference between a small country and a large country is that the latter is able to affect world prices. If a large country like the United States would implement a tariff on a good that has a large elasticity of foreign export supply, the domestic relative price of the imported good rises and thus lowers demand for this good. The reason that relatively larger countries can affect the world price is due to their relatively large domestic demand. How much the world price is affected can be interpreted with the elasticity of the foreign export supply - the less elastic the export supply, the greater is its effect on the world price. 2.1 The Ricardian Theory According to the Ricardian framework the main welfare gains from international trade are that countries are able to reach a higher income and consumption. When a country opens 7

11 up to trade, and hence international competition, it strengthens the incentives to specialize according to its comparative advantage and its possibility to reach a higher production possibility frontier. The economy will have a cost advantage for any good that has a relative productivity that is higher than the relative wage. The good for which the economy has the so called comparative labour productivity advantage will be exported. Reallocation of production and changes in relative prices are further determined by the supply and demand in the world market. The gains from trade are either represented in gains from exchange or gains from specialization and the benefits from trade are generally larger if the trading countries have a difference in labour productivities (Bowen et al 2003). Bhagwati et al. (1998) show the effects of a tariff using the Ricardian model. The production possibility curve (A-B in figure 2.1) shows to what extent the country can produce good 1 and good 2. The country specializes in one of the two goods according to Ricardian theory. In the scenario below the country specializes in good 1 (not producing a single unit of good 2). Looking at the figure one can see that under free trade the country consumes at C F and therefore they import G-C F of good 2 and exports from G-B units of good 1, since the production curve stretches until point B. Adding a tariff on good 2 (without raising the price enough to induce the country to produce good 2) affects the country to consume at point C T (still on the same terms of trade line). Therefore the imports decrease to D-C T of good 2 whilst exporting from D-B of good 1. Hence, the international trade has decreased due to the tariff. The distance between B-E is revenue from the tariff which is presumably returned to the consumers in a so called lump-sum transfer. The world price under the tariff is equal to the slope B-C T, meaning D-C T of good 2 is valued to D-B of good 1. The domestic price under the tariff is the slope E-C T, meaning for D-C T of good 2 one pays D-E of good 1. 0-B = National income (in terms of units of good 1), 0-E = National expenditure (in terms of units of good 1). If instead the tariff is reduced this would lead to an increase in international trade seeing as the consumption would move back towards its original point (C F ). Since under free trade the demand for good 2 is as large as the figure shows, the imports would increase as the tariff decreases. 8

12 Figure 2.1 Effects of a tariff, Ricardian Model GOOD 2 C F C T A 0 G D B E GOOD 1 Source: Bhagwati et al The effects of a tariff reduction can have different outcomes for the welfare of a country. A large country can benefit from an optimal tariff if it increases the country s terms of trade. A country s terms of trade is its external relative price i.e. its price of exports over its price of imports (Rivera-Batiz et al. 2003). As mentioned earlier, the shift in production after adding a tariff has a negative impact on the GDP of a small country, i.e. a country that is not large enough to affect world prices. Concluding the Ricardian theory, the reduction of a tariff leads to increased specialization and a reallocation of production as countries trade according to their comparative advantage. The Ricardian theory has received much criticism due to lack of empirical research and its simplicity. The predictions made about strong specialization have not yet been the case in international trade and the exclusion of income distribution as a gain from trade is missing in the theoretical model. Nor does the Ricardian theory account for increasing returns to scale (IRS) which would help explain the rather large trade between countries similar in resource endowments (Krugman et al. 2006). 2.2 Heckscher-Ohlin Theory The Heckscher-Ohlin theory also referred to as the Factor Proportions theory discusses countries differences in resources as the main argument for international trade. Technologies are assumed equal in the trading countries and the difference in comparative advantage is a result of differences in factor prices. The factor price is determined by the countries relative abundant factor, i.e. land or labour. If a country has a higher ratio of land to labour the comparative advantage lies in its land intensive good and vice versa. Countries will ex- 9

13 port the commodity that is relatively more intensive in the factor for which the country is factor endowed. The greater the difference in economic structure, the greater are the benefits and the volume of world trade when free trade is present. Assuming two countries, Home and Foreign, Home is a labour abundant country and produces a higher ratio of cloth to food. Foreign is land abundant and produces a higher ratio of food to cloth. Notice that abundance always is defined in relative terms, meaning the ratio of labour to land in both countries is compared so that no country can be considered abundant in both land and labour. In autarky the relative price of cloth would be lower in Home than in Foreign (see equilibrium point 1 in figure 2.2). At this point the domestic relative supply (RS) crosses the relative demand (RD) for country Home. The equivalent equilibrium for Foreign would be at point 3, a result of the country s comparative advantage in land. Free international trade leads to equilibrium point 2, where relative prices are converged (Krugman et al. 2006). Figure 2.2 Convergence of prices Source: Krugman et al Free trade causes one single relative price of cloth and food to emerge on the world market as a consequence of equalized land rents and wages when international competition is present. The convergence inclines the distribution effect that, the owner of the factor for which the country is abundant gains from trade whiles the owner of the scarce resource loses. An economy would gain the most from trading according to its resource endowments during free trade (Bowen et al. 2003). Summarising, the implications of the theory are that the reduction of tariffs would lead to the previously mentioned distribution effects on income, and factor price equalization. With this said, the main criticism forwarded by among others Krugman et al. (2006) argues that factor price equalization has not been observable in international trade. The main reasons for this are according to the authors the wide differences in resources, transportation costs, and other barriers to trade, as well as wide international differences in technologies. 10

14 2.3 New Trade Theory Krugman (1980) argued that the former theories of trade failed to explain many of the actual trade patterns on the international arena. He meant that the extensive trade between industrialized countries and the two-way trade in differentiated products were some of the factors left unspoken. He therefore developed the New Trade theory where increasing returns to scale (IRS), product differentiates, and imperfect competition were the main explanations for international trade. Here, the similarity in production between countries, on the contrary to earlier theories, gives an incentive for producers to specialize in differentiated products. The so called monopolistic competition implies that countries are fully engaged in this type of intra industry trade. International trade enables countries to specialize in a limited amount of goods without sacrificing the amount of verified goods available. Firms choose to locate where the demand is large so that scale economies can be realized and transportation costs minimized. Countries tend to be the net exporter of goods for which they have the largest domestic demand and an increased welfare would be the result of a greater variety of goods: Gains from trade will occur because the world economy will produce a greater diversity of goods than would either country alone, offering each individual a wider range of choice (Krugman, 1980 p. 252) Figure 2.3 illustrates how firms experiencing access to a larger market, other things equal, can produce at a lower cost since the average cost decreases. This is presented by the shift from point 1 to point 2. At point 2, there is a larger amount of firms (n) on the market, firms can produce at a lower cost and a larger variety of goods are available for the consumers from the increased amount of firms. Figure 2.3 Increasing Returns to Scale Source: Krugman et al

15 Feenstra (2003) argued that one valid example of the benefits of scale economies was when the small economy of Canada signed a free trade agreement with United States in 1989 and Canadian firms were able to decrease their average cost curve and become significantly more efficient. On the contrary, scale economies can also implicate closure of small companies due to fierce competition. Intra industry trade is more common between countries at similar stage of economic development and less observable among countries trading raw materials and goods within more traditional sectors. Considering external economies of scale based on industry level, once already established on the market an industry can have an advantage over a newcomer because the already established industry is able to charge a lower price than a newer industry within the same market. This is the case even though the newer industry potentially has a lower average cost than the established one. In this sense countries industries could be worse off when opening up to trade, since the price would be higher than before when the country produced the good itself (Krugman et al. 2006). Concluding the New Trade theory one can argue that as trade barriers such as tariffs are reduced, the production will be located close to where the demand is large, i.e. where the market is large and increasing returns to scale can be exploited. However, not all industries are positively affected if free trade is present. Newer industries have a harder time surviving as already established industries have an advantage on the market. With these three theories above as background, the following hypothesis will be tested in this thesis: The GSP agreement implemented in 1995 had a significant impact on the trade flow from the AC to the EU15. The following analysis is based on the international trade theories presented in this chapter and aims at thoroughly investigate whether a reduction of tariffs has the expected result of increased trade. 3 Empirical Analysis of the GSP Agreement This chapter will explain the method, the results from the empirical testing, and the analysis of the results. 3.1 Method The trade flow between the AC and the EU15 will be analyzed with a gravity model. The model is commonly used amongst economists and has gained worldwide recognition for its robustness. Due to the adequate formulation of the model, it is simple to apply on bilateral trade flows and it facilitates the inclusion of different variables in the model (Sanso et al. 1993). With the gravity model one can analyze the impact of the 1995 GSP agreement on the export from the AC to the EU15. The concept of the model comes from Newton s universal law of gravity and states that similar size in GDP and small distances lead to an increase in trade. According to Krugman et al. (2006) a country s trade is proportionate to its economic size. The larger the market, the larger is GDP and hence the larger is the country s expenditure on imports. Thereby the trading partner is encouraged to trade, due to the increase in demand. 12

16 Tinbergen (1962) was first with introducing the factors of distance and transport costs in the analysis of when and to what extent countries will trade with one another. The factors that affect the amount of trade between two countries are; size of the country s GDP, the size of the market of the receiving country, and transportation costs. The gravity model appears as follows in a basic formulation: E ij = a 0 * Y i a1 * Y j a2 * D ij a3 (1) Where E ij = Exports of country i to country j, Y i = GDP of country i, Y j = GDP of country j, D ij = Distance between country i and country j. The exponents, a 1, a 2, and a 3 explain that there is no direct proportionality between the explanatory variables (Y i, Y j, and D ij) (Tinbergen, 1962). The gravity model can be motivated theoretically by several trade theories: the monopolistic competition approach is one example. Feenstra et al. (2001) write that a type of equation such as the gravity model is created when countries completely specialize in differentiated goods. They describe the model by using the monopolistic competition from New Trade theory. With help of the gravity model one can show how there is a home-market effect as a country has a higher export elasticity of a good in relation to its own income than to the foreign country s income. Hence, the home country is the net exporter of this certain good. Since demand is larger in the home country, firms will enter until profits are zero, leaving the country with an unbalanced amount of firms. The authors conclude that although the theoretical foundations of the model are quite broad, the usage of the gravity model has shown to be explicit. The gravity equation can be used both for homogenous- (which is common in developing countries) as well as differentiated goods. The gravity model used in this thesis is an extension of the original model and looks as follows: Ln(X ijt) =β 0 + β 1Ln(Y it) + β 2Ln(Y jt) + β 3Ln(D ij) + d(g ijt) + d(l i) + d(c ij) + d(t ij) + d(y it) * e ijt (2) Table 3.1 Variables included in the regression Variable Description Source Expected outcome Ln(X ijt) Export value from country i to country j UN Comtrade, value in current US dollar Dependent variable (Y) Ln(Y it) GDP of country i UN Comtrade, value in current US dollar Ln(Y jt) GDP of country j UN Comtrade, value in current US dollar + + Ln(D ij) Distance from country i to country j CEPPI, measured in kilometres - d(g ijt) GSP dummy for the Andean countries Com- European mission + 13

17 d(l i) d(c ij) exports to EU15 Dummy for countries without coastline Dummy for countries sharing colonial history CEPPI - CEPPI + d(t ij) Trade dummy con - trolling for missing trade data d(y it) Dummy controlling for missing year data - + As in the original model the variables included are: Export, GDP, and Distance. The variables are logged to facilitate the analysis. The error term controls for the omitted variables that still has an affect on the dependent variable Export, t stands for time since time series data is used. The dependent variable Export consists of export data to and from all countries included in this particular analysis. The expected effect of the GDP variables is positive in accordance to what Krugman et al. (2006) stated that the larger the GDP the larger the amount of trade. Hence, size of GDP represents one of the contributing factors to the trade flow. All else equal, as GDP changes by one percent, the dependent variable Export will change by β 2 percent. The distance data shows the distance in kilometres between the centres of each trading country s capital. The variable is expected to have a negative impact on the dependent variable, since the larger the distance between the countries, the less are the countries expected to trade due to transport- and time costs. A GSP dummy is included to control for the export from the AC to EU15 between the years 1995 to This binary variable is expected to have a positive impact on the dependent variable since it entails that tariffs have been decreased and according to international trade theory reduced trade barriers leads to an increase in trade. The colonial- and the landlocked dummy have been included in accordance with what Rose et al. (2001) argued in their study, that both these variables have an impact on trade. The authors argue that when a country is landlocked i.e. lacks a coastline, its trade will be limited, since being surrounded by land hinders transportation. The share of a colonial history has a positive impact on trade since having similar cultural, political- and legal institutions facilitate trade (CEPII, 2006). The year dummies are included to control for the years that are missing in the data, and a trade value dummy is included for missing trade value data. Time series data have been obtained for the years 1980 to These years are chosen with the intention of analyzing whether the GSP implementation in 1995 had a significant impact on the trade flow from the AC to EU15. The countries included in the study are the EU15 5 : Austria, Belgium-Luxembourg 6, Finland, France, Germany, Greece, Ireland, Italy, 5 Denmark is exlcuded due to lack of data 14

18 the Netherlands, Portugal, Spain, Sweden, and the United Kingdom and the AC countries during this period of time: Bolivia, Ecuador, Colombia, Peru, and Venzuela. 146 countries are included in the analysis as the data of interest could be obtained for these countries. The regression showed results of heteroscedasticity when performing a White s test. Hetersocedasticity is a common problem when one has many binary variables in the regression since equal variance among the dummy variables can not be assumed. Since this is a violation against the assumption about normality i.e. the variables in the regression are not BLUE 7, a normal regression does not give reliable results (Gujarati 2003). A robust regression was performed to correct for heteroscedasticity in the error term through the use of Weighted Least Squares. Table 3.2 Descriptive Statistics from regression Variables Minimum Maximum Mean Median Std. Deviation Skewness Export value GDP i GDP j Distance GSP dummy The results give a description of the difference between the four main variables used in the regression and the variable of interest GSP. Export value has the largest dispersion in its minimum- and maximum values. This is an expected result as there are great differences in countries export. The minimum value here is zero because in some cases there is no trade value between countries. Since the means of all the variables (except for the binary variable) lie closer to their maximum values, one can draw the conclusion that these observations contain more high values. Since the majority of the countries that were excluded from the regression (due to lack of data) had a relatively low GDP this is an expected result. The mean and the median have for the variables, Export value, GDP i, GDP j and Distance similar values which indicate that there is an equal spread of the observations. The low values for the standard deviation in all variables indicate that the data points do not deviate much from the mean. Moreover, all variables (except for the binary variable) have a negative skewness in the distribution: this further strengthens the fact that there are relatively more high values than low values. As the GSP dummy has positive high values of skewness in comparison, this indicates that the distribution of the data is skewed to the right, which is due to the small amount of data given from the dummy. Therefore one can say that the mean is higher than the median for the GSP variable, because these data points have relatively more low values (i.e. more zeros than ones) than high values. 6 Belgium and Luxembourg are treated as one country in the analysis 7 Best Linear Unbiased Estimate 15

19 3.2 Results and analysis Table 3.3 Results from regression Explanatory Variables Coefficient estimate t-statistic GDP i GDP j Distance GSP dummy Landlocked dummy Colonial dummy Missing trade value dummy Dependent variable: Export value N = The outcome of the regression shows good results since all variables are significant at the 95% level of confidence. One can also confirm the signs of the values for the variables from the expectations in the previous section. The regression shows an adjusted R 2 of hence, explaining 63.4 % of the model (see table 5.1 in appendix). The coefficient estimates for the GDP variables are positive and significant, indicating that they have a positive impact on the dependent variable. This is consistent with theories of the gravity model that a country s GDP has a positive relation to trade. The explanatory variable Distance, also originating from the gravity model, has a significant and negative coefficient estimate which is consistent with theory that distance has a negative relation to trade. As previously mentioned the coefficient estimate for the landlocked dummy was expected to be negative in accordance to what Rose et al. wrote that the lack of access to sea hinders transportation and therefore trade. The authors further argued that a shared colonial history has a positive relation to trade. As evident by international trade flows, countries with similar language, culture, and institutional structures trade to a greater extent since these similarities facilitate the international exchange of goods and services. The regression results are consistent with the theory and one can conclude that the coefficient estimate is significant and has a positive impact on trade. The coefficient estimate of the missing trade value dummy is also in accordance with previous expectations, since it indicates when export data is missing for a certain year. Continuing with the variable of interest one can see that the results from the regression show a significant positive value for the GSP coefficient estimate given significant values for GDP, Distance, and the control dummies. One can therefore accept the hypothesis that the GSP agreement from 1995 had a significant impact on the dependent variable: Export value. This is in accordance with theory that reduced trade barriers have a positive impact on trade. 16

20 Freres et al. (2004) compared GSP-exports from the Andean countries to the EU with GSP-exports to the world, since it is not controlled for in this particular study one cannot agree with nor contradict their result. One can however, from the regression results in this particular paper, argue the importance of the agreement. With the coefficient estimates for the GSP dummy in table 3.3, one can see how the dependent variable Export value is affected. All else equal, should the dummy assign the value of 1, the dependent variable would increase on average by [ (exp( )-1) *100 ] = % than would the dummy assign the value 0. For a more thorough analysis, a correlation matrix has been included (see table 5.4 in appendix). One can compare the values within the matrix to analyze whether the variables are strongly correlated to the dependent variable. According to the correlation matrix, all variables in the regression have a relationship with the dependent variable, however the GDP and Distance shows the strongest correlation to the dependent variable. From graph 3.1 below one can see that the export from the AC to the EU15 followed a similar pattern as the export from the AC to the world up until When analyzing the export performance of the AC one has to take into account that these countries are relatively sensitive to changes in the world prices. Partly because they foremost export primary goods that have many fluctuations in price and also because they do not have a fully diversified export sector which make them very vulnerable for fluctuations in the world market. This is in accordance to what Moreno-Brid et al. (2004) found, that many of the Latin American countries did not have an enough diversified trade and were thus vulnerable for the increased competition when they opened up their economies during the 1990 s. Also worth mentioning is that the Andean countries have a large internal dispersion when it comes to their export performance during 1990 to 2000 as can be seen when looking at the individual graphs (see graph in appendix), this makes the analysis of the AC as an export unity further complex. Graph 3.1 AC export to the EU and to the world Source: UN Comtrade The fact that the overall AC export to the EU decreased in the early 1990 s comes as no surprise since the two largest exporters to the EU (Colombia and Venezuela) had a deterioning export performance during this time. Venezuela and Peru had decreasing export to the EU15 during the years 1990 to 1993 (see graph 5.4 and 5.5 in appendix). In Venezuela, 17

21 the decrease in export can be due to an increased poverty rate during this time. As earlier mentioned, the poverty augmented in Venezuela during the 1990 s and Sheahan (2006) argues that Venezuela s poor economic performance during this time was partly a result of trade policies that failed, he wrote: Trade liberalization was not helpful, because successive governments failed to correct the overvalued currency (Sheahan, 2006 p. 126) Peru had a severe crisis between the years 1987 to 1991, which affected the country badly. The crisis can be an explanation for the decrease in export from Peru during this time. Colombia s export decreased between the years 1991 to Colombia had a weaker economic performance during the 1990 s than did other countries within the AC. Sheahan (2006) suggests that there are three possible reasons for this; continuing liberalization of the economy, unsuccessful monetary policies, and increased political conflicts. In 1993 there is an increase in trade both to the world and to the EU15. The AC experienced an average annual growth of 23.3% of FDI over the years 1993 to 1997 (Creamer, 2003). The same author argues that the main reasons for the increase in FDI were several beneficial measurements taken by the region to attract foreign investments. Since the 1990 s 26% of the total FDI to the AC has come from the EU (Comunidad Andina, 2003). Hejazi and Safarian (1999) is a few of many authors that claim a relation between FDI and trade. They mean that FDI encourages the domestic production in a country through technological spillover. With this in mind, one could argue that the export increase from 1993 to 1997 was partly a result of increased domestic capital formation and productivity. Since the export pattern from the AC to the world and the EU was similar until 1998 (see graph 3.1), the question is raised as to what happened after 1998 to make the export decrease to the EU? In 1997 the FDI decreased substantially due to political and financial problems in the Andean region. This could be one of the explanations for the decrease in trade to the EU15. Moreno-Brid et al. (2004) argued that an aftermath of the Asian crisis in 1997 was a fear of investing in developing countries as the uncertainty was more evident. Another possible argument for the decreased trade with the EU could be the increasing intra-regional trade during the 1990s. However, Creamer (2003) argues that this increased intra-trade did not increase at the expense of the external trade. One of the possible outcomes of the graduation system that was introduced with the GSP agreement in 1995 was that several goods exported would lose its preferential status to the EU15 and be excluded from the scheme. Hence, it could have proven more beneficial for the AC to export these goods to other countries, depending on where they would have the highest demand. Freres et al. (2004) wrote the following about the graduation system: The mechanism of graduation defeats the purpose of the GSP, because countries that manage to gain competitiveness in a particular sector is subsequently punished with the loss of the preference (Freres et al p.29). By this the authors indicate how the graduation system is contradictive as it could restrain the incentives to grow competitive. Like mentioned earlier Pangariya (2002) also brought up the same issue of lost incentives to become a competitive exporter due to the risk of losing preferential access to the EU market. With this in mind, one explanation for the de- 18