Decomposing the EU membership trade
effect of the CEECs
ABSTRACT
This thesis examines the trade effect of the ten CEECs of joining the EU. Its aim is to decompose the overall EU trade effect into regional and country trade effects. I use an augmented gravity model for the econometric part of the thesis. Gravity models are the proven standard in the bilateral trade flow literature. First, I establish that the CEECs experienced a negative import and export effect after joining the EU. Second, I decompose this trade effect in east and west trade effects. Here, I find that the initial negative EU membership trade effect is driven by the effect on trade with the EU-17. Further, I find that the ten CEECs did not respond equally in terms of trade to joining the EU. Especially those countries with high degrees of international trade and high shares of trade with the EU-17 have highly negative trade effects with the EU-17. Thereafter, the results are supported by the regional trade analysis. I find that intra subgroup trade is positively and inter EU-17 is negatively affected by the selection into the EU. Thus, it is found that at least in the short run the CEECs benefited more from increase in intra-CEEC trade than from increase of trade with EU-17 as a result of EU entry. Additionally, I argue that cultural distance has a negative effect on trade, at least in the European context.
KEYWORDS
European Union, CEECs, gravity model, free trade agreements, trade effect, Cultural distance.
AUTHOR
W.A. Pool (1552503)
MSc International Business & Management, International Financial Management University of Groningen, Faculty of Economics and Business
Uppsala Universitet, Faculty of Economics and Business
THESIS SUPERVISORS
1st supervisor: G.J. Lanjouw 2nd supervisor: C.L.M. Hermes
TABLE OF CONTENTS
Table of contents ... 1
1. Introduction ... 3
2. Historical Background ... 4
2.1. The Council for Mutual Economic Assistance ... 4
2.1.1. Concluding remarks on trade prior to the collapse of the CMEA ... 6
2.2. Trade agreements ... 6
2.3. Trade patterns post CMEA ... 7
2.3.1. Concluding remarks on trade after the CMEA ... 8
2.4. Regional subgroups ... 9
2.4.1. Balkan ... 9
2.4.2. Baltics ... 10
2.4.3. Visegrád ... 11
2.5. Concluding remarks trade analysis ... 12
3. Trade theories ... 12
3.1. Regional trade theory ... 13
4. Theoretical foundation of the gravity model ... 14
4.1. A simplified derivation of the model ... 16
4.2. Determinants of bilateral trade ... 18
4.2.1. Income ... 18
4.2.2. Distance ... 19
4.2.3. Population and area ... 19
4.2.4. Cultural distance ... 20
4.2.5. EU membership ... 22
4.2.6. Individual country EU membership ... 23
4.2.7. Regional subgroups ... 23
5. Technical specification of the gravity model ... 24
5.1. Econometric issues ... 25
5.1.1. Omitted variables... 26
5.1.2. Simultaneity bias ... 26
5.1.3. Measurement error bias ... 26
5.1.4. Treatment effect... 27
5.1.5. Technical implications of using fixed effects ... 27
5.1.6. Zero trade flows ... 28
6. Data ... 28
6.2. Data description ... 29
7. Results and discussion ... 30
7.1. Difference between the exports and imports model ... 30
7.1. The basic gravity equation variables ... 31
7.2. Cultural distance and trade ... 33
7.3. The EU Membership trade effect ... 34
7.4. Intra-CEEC versus inter EU-CEEC trade effect... 36
7.5. The EU trade effect of the individual CEECs ... 37
7.6. The EU trade effect of the regional subgroups ... 39
8. Conclusion ... 42
8.1. Suggestions for future research ... 43
1. INTRODUCTION
In 2004, eight Central and Eastern European countries (CEECs) joined the European Union (EU). Thereafter, Bulgaria and Romania also entered into the EU in 2007. The principal economic argument for the EU-151 to expand their union is to broaden the free trade area (Venables, 2002). Consequently, they create a large single market, in which national production can exploit the benefits of economies of scale (Papazoglou, Pentecost and Marques, 2006). Prior to accession, candidates have to meet certain criteria and sign trade agreements with the EU. That trade increased for the EU-15 and the accession countries as a result of these trade agreements is well established (e.g. Spies and Marques, 2009; De Benedictis, De Santis and Vicarelli, 2005; Herderschee and Qiao, 2007; Adam, Kosma and McHugh, 2003; Martin and Turrión, 2001). However, several authors found small or even negative signs for the subsequent EU membership trade effect. They attribute this to prior trade agreements smoothening the trade effect and to econometric difficulties (e.g. Baier and Bergstrand, 2007; Bussiére, Fidrmuc and Schnatz, 2005; Caporale, Rault, Sova and Sova, 2008; Spies and Marques, 2009). Spies and Marques (2009) conclude that prior trade agreements have indeed supported and accelerated the CEECs‟ integration into the EU. Furthermore, Bussiére, Fidrmuc and Schnatz (2005) found that trade shares of the CEECs with the euro countries are high and stable, which according to them signals that trade is up to its full potential. In turn, several authors (e.g. Bussiére et al. 2005; Herderschee and Qiao, 2007; Spies and Marques, 2009) found that intra-CEEC trade is becoming more sizeable. This thesis builds upon these and more papers and proceeds to examine the academic issue. Therefore, this thesis investigates the effect of the actual subsequent EU membership of the ten CEECs on trade and the potential of inter EU-CEEC trade. For the CEECs, the inclusion into the EU has completely eliminated trade barriers and intensified competition. The empirical question remains whether joining the EU has led to an additional geographical restructuring of trade flows, involving trade creation and trade diversion. Further, after establishing the relevant variables that are of influence on the trade effect, I examine whether there is a differentiated effect among countries and regions. Thus, I decompose the trade effect from one total trade effect, into two east and west trade effects, three regional subgroup trade effects and 10 individual country trade effects. First, this thesis adds value to the existing literature by investigating the EU membership trade effect with new data. Second, prior studies examining the EU membership trade effect found small or also negative trade effects (e.g. Aristotelous, 2006; Belke and Spies, 2008; Bussiére, Fidrmuc and Schnatz, 2005; Herderschee and Qiao, 2007; Papazoglou, Pentecost and Marques, 2006; Martin and Turrión, 2001; Mchugh, 2003). However, whereas these studies focus mostly on the effect for the European Union as a whole or on the effect for the EU-15, this thesis focuses on CEECs. Third, this thesis adds value by decomposing the subsequent EU membership trade effect. This allows for a more thorough examination of what drives the trade effect and which regions or countries have benefited the most from EU membership.
1EU-15: Austria, Belgium, Denmark, England, Finland, France, Germany, Greece, Ireland, Italy, Luxemburg,
In doing so, this thesis investigates whether there is a differentiated trade effect among the CEE regions and countries. Thus, it provides insight in the performances of regions and countries, which can be relevant for investors and policy makers. Fourth, this thesis provides additional academic relevancy by including a proxy for cultural distance instead of using the common dummies of cultural proximity. According to Felbermayr and Toubal, (2010) those cultural proximity variables clearly capture cultural familiarity but also reflect other trade-creating factors, such as the cost of communication. Therefore, Möhlmann, Ederveen, De Groot and Linders (2009) and Linders, Slangen and De Groot (2005) implement the same proxy of cultural distance, however, they find contradicting results. Finally, Aristotelous (2006) states that empirical literature that finds that the effect of a trade agreement on trade is positive, significant and large, focuses almost exclusively on the overall impact trade agreements have on trade while paying very little attention to whether there are significant differences across the individual countries involved. This thesis attempts to narrow the gap by examining whether the impact the EU has on trade is widespread across the CEECs. This thesis examines the effect of EU membership on bilateral trade of each CEE country while emphasizing the differences across them. I use panel data from 2000 to 2008 in the context of an augmented gravity model. Gravity equations are widely used to determine the superior trade theory (e.g. Feenstra, Markusen and Rose, 2001), for examining the influence of a diversity of policies concerned with trading agreements, currency unions and diverse trade distortions (e.g. Bougheas, Demetriades and Morgenroth 1999, De Grauwe and Skudelny 2000, Glink and Rose 2002, Costa-i-Font, 2010).
First, this thesis will provide a historical background of trade flows and trade agreements in Eastern Europe. Second, I discuss relevant theories and concepts, the theoretical foundation of the gravity model and empirical results. Next, I discuss the technical specification of the gravity model. Subsequently, the data chapter provides a clarification on data selection and a description of the data. Thereafter, I present and discuss the results. Lastly, I will conclude on my findings, present the limitations of my research and provide suggestions for future research.
2. HISTORICAL BACKGROUND
2.1. The Council for Mutual Economic Assistanceintegration. Economic integration in a socialist context implies laws and tendencies, which are aimed at gradual convergence of economic levels (Belayev, 1968). According to Köves (1985), the focus of the CMEA members prior to the formation was to produce internally. He estimates that in 1913 the total value of exports of the Soviet Union was 5,298 million roubles, whereas in 1940 this was only 1,066 million roubles.
Community (EEC) was openly recognized, relations with the West were developed and trade increased consequently. From 1965 onwards it is clearly shown that the share of trade with the West grew generally within the total trade of the CMEA countries (table A2.5; A2.7). However, when looking at the members share of total CMEA trade to the West from 1965 to 1983 (table A2.6) it can be seen that Soviet Union‟ trade to the west increases faster than that of the Eastern European countries. The total value of East-West trade increased tremendously from 1972 to 1975. However, this trade was rather moderate in real terms since the outstanding years coincide with the oil price explosion and with the inflationary peak of international trade. After moderate real and nominal trade growth in 1976 and 1977, the data again indicate considerable growth between 1978 and 1980. However, from 1981 to 1983 trade decreased significantly and its value in 1983 was 13 per cent below that of 1980 (Köves, 1985).
2.1.1. Concluding remarks on trade prior to the collapse of the CMEA
The most relevant characteristics of the CMEA are the bilateral trade relationship of the members with the Soviet Union, the dependency on the Soviet Union, the relative autarkic nature of CEEC‟ economies and the similar patterns of industrial expansion (supported by e.g. Baskos, 1993; Crabbé and Beine, 2009; Fisher 1971; Korbonski, 1990; Paas (2001). Only essential commodities that could not be produced domestically were traded. Prior to World War II, there was a strong internal focus in the CMEA (Köves, 1985). Thereafter, the Eastern European countries enlarged their foreign trade (table A2.1). Moreover, the trade share of the West increased considerably as well (table A2.5).
2.2. Trade agreements
CMEA members were depending on trade with each other (Crabbé and Beine, 2009). When the CMEA collapsed this perception was rather influential in convincing both western and eastern economists and policy makers that new trade arrangements needed to be formed. Many policy measures were implemented to ease the damage done by the collapse of the CMEA. Among which were trade agreements with the EU and regional preferential trading systems among former CMEA members (see figure 3 and table A2.9). Developing these kinds of free trade and preferential trade agreements is an important part of EU policy towards developing countries and neighboring European countries. Europe Agreements (EAs) are trade agreements between the EU and an individual European country, designed specifically for that country. Additionally, the EC creates free trade areas with neighboring countries. Essentially, the agreements are designed to liberalize trade and gradually cutback tariffs, quotas and other trade barriers (Bojkov, 2004). According to Brenton and Manchin (2002), the free trade agreements are a means of increasing economic integration through improved access to the EU market. In addition to raising economic integration, FTAs are also designed for political, foreign policy and security objectives (Brenton and Manchin, 2002). Furthermore, to avoid an overflow of CEECs cheap export products on EU market it is in the best interest of the EU that the CEECs as region develop and integrate economically. After the green light was given to negotiate trade agreements with Europe, the Eastern European countries shifted their focus swiftly to the West. The trade agreements gradually lowered trade barriers. The elimination of tariff barriers resulted in increasing total trade since prices for imports fall and profit margins for exports rise. Therefore, Papazoglou, Pentecost and Marques (2006) hypothesize that both EU and CEEC producers and consumers are potential beneficiaries. In contrast, in his econometric analysis of the CMEA, Holzman (1987) argues that the formation of the CMEA has led to trade creation in the case of the Soviet Union but he finds no welfare benefits for the Eastern European countries. The signing of trade agreements was the first step towards integration but the ultimate goal of these agreements has always been the accession of these countries. Next, the trade flows of this post CMEA period are discussed.
2.3. Trade patterns post CMEA
and Beine (2009) claim that after the collapse of the CMEA in June 1991, intra-CEECs trade came to an almost complete halt (see also phase one table A2.8). Furthermore, graph 1 shows two graphs from Crabbé and Beine (2009) that clearly show how the degree of trade with Russia has declined and the degree of trade with Europe has increased from 1989 to 2000 (see also table A2.8). Graphs 7 to 8 are extensions of these two graphs, they show the same graphs but now for export and import data for 2009. It can be seen that trade with other CEECs is rising and trade with Russia has declined significantly since 1989, but is stable in the last decade. Important to note when discussing imports from Russia is that oil dominates these trade flows. Because oil is a necessity for the CEECs, oil has a low demand elasticity and thus the traded volume is little affected by the economic situation. Nevertheless, the shares of trade with Russia are closely similar to the 2000 situation. Only Lithuania and Latvia trade relatively much with Russia. Furthermore, as can be derived from these graphs the Eastern European countries first reoriented their trade flows to the West, but then also more to each other. Table A2.8 shows trade flow alterations during four phases after the collapse of the CMEA and prior to the examination period of this study. From 1992 to 2000, intra-West trade increased with approximately 10%, in the same period intra-East trade increased with circa 50%. However, from 1989 to 2000 West-East trade increased with almost 500% (see also table A2.5; A2.7). The examination period of this thesis starts were Egger et al. (2007) and Crabbé and Beine (2009) stopped. Some relevant graphs are presented in the appendix. First, graph 2 to 5 show consistently higher total import values than total export values. Thus, as a region the CEECs are running a trade balance deficit. Second, most trade is with the other EU members. Third, the graphs clearly illustrate the economic downturn in 2009 and the effect it has on trade. Last, the graphs also show that prior to 2009 trade was increasing rapidly, especially with the EU-172 members. According to Bussiére et al. (2005), trade with the EU is now up to its full potential. They go even further by saying that it would be better if the CEECs experience a decline in their shares of trade with the EU (i.e. enlarge trade with others and consequently decrease dependency on the EU). They claim that the strong concentration of these countries‟ foreign trade with the EU indicates that they are “too heavily” oriented towards the EU. Bussiére et al. (2005) finds support for this hypothesis by the patterns of the Czech Republic and Hungary, which experienced a reduction of trade shares with the euro area in most recent years. Consistently, graph 5 shows a small decline in trade of the Eastern European countries with the EU-17 from 2000 to 2009. In general, this decline is compensated with an increase in intra-CEEC trade. Thus, continuing their line of reasoning it seems that now that trade with the EU is up to its potential the share of intra-CEEC is expected to increase relative to the share of trade with the EU-17.
2.3.1. Concluding remarks on trade after the CMEA
The CEECs signed several trade agreements with the EU and with nearby countries to improve economic integration and market access. As a result, trade increased between the CEECs but increased
especially between the CEECs and Europe (respectively 50% and 500%). In sum, trade after the collapse of the CMEA to 2000 is characterized by a large increase in trade with the EU and a declining share of intra-CEEC trade. From 2000 to 2009, two general trade trends mirror those of the previous decade. First is the decline in trade as percentage of total trade with the EU-17. Second is the increase in trade as percentage of total trade with the CEECs. In addition to the general trade trends, I also examined trade patterns of the individual Eastern European countries and economic regions.
2.4. Regional subgroups
2.4.1. Balkan
34%) (see also table A7.3;A7.4). Consistently, in the case of Romania trade with the euro area was very low in 1993 but increased considerably (from 40% to 60%) in the following decade (Bussiére et al., 2005). On the other hand, in the case of Slovenia, trade with the EU was already high and was stable around 60% during the same period. Thus, Bussiére et al. (2005) conclude that for the individual CEECs inter EU-CEEC trade converge to a similar level. Graph 5 indeed shows a rather stable yet declining trade pattern of the CEECs with the EU-17 from 2000 to 2009. The trade trend of declining shares of trade with the EU and increasing trade shares with the CEECs from 2000 to 2009 are also apparent for the Balkan (graph 8). The only exception is Bulgarian export to the EU-17, which fluctuates around 50%. Further, given its proximity, the Balkan countries trade relatively little with Russia in 2009 (graph 6 and 7). Bulgarian‟s share of trade with the rest of the world is the largest of the CEECs. On average, the CEECs export 20% of their total export to the rest of the world, whereas for the Balkan countries this is on average nearly 30%. The bulk of this trade is with Turkey.
2.4.2. Baltics
declining rapidly and becoming less and less important as these countries grow, Russia‟s share in exports will remain significant in this region in the long-run perspective. Consistently, Bussiére et al. (2005) find that the CEECs‟ share of trade with Russia is around 5% on average, which is higher than that of the UK or the US. The share of trade is higher for the Baltic countries (above 10%), reflecting the importance of distance and cultural links among the determinants of trade. First, Egger and Larch (2010) find that for the Baltics between 1994 and 1999 intra-group trade decreased with 53.38%. Second, inter-EU trade increased with 67.48%. Whereas inter-group trade with Hungary and Poland, Romania, Bulgaria, Slovakia and Czech Republic and last Slovenia, respectively decreased with 56.77%, 55.18% and 64.11%. Trade with Russia decreased less compared relative to the other groups, namely 22.4% (table A7.3). Graph 6 and 7 show the percentage of trade of individual countries to EU-17, CEECs, Russia and the rest of the world in 2009. It can be seen that trade with Russia is amongst the highest of the CEECs. Further, graph 9 shows the percentage of export of the Baltic countries with EU-17 and with other CEECs. The trend is quite obvious and similar to the general trade patterns; for example, Estonian export to the EU-17 drops from circa 70% in 2000 to below 50% in 2006, whereas Estonian export to other CEECs rises from approximately 10% in 2000 to almost 20% in 2007.
2.4.3. Visegrád
hypothesis by the patterns of the Czech Republic and Hungary, which experienced a reduction of trade shares with the euro area in most recent years. Consistently, graph 5 shows a steady declining trade share with EU-17 from 2000 to 2009. Deriving a trade flow analysis for Visegrád from Egger and Larch (2010), similar to the trade flow overviews of the Balkan and Baltics, is difficult because its members are spread over multiple groups. Nevertheless, the results are presented in table A7.3. Most importantly, in the 90‟ trade with the EU rises over 50% whereas intra-group trade and trade with Russia decreased significantly. Moreover, Bussiére et al. (2005) find that trade between Slovakia and Czech Republic is very high because of strong cultural and historical links. For the Visegrád countries, similar to the Balkan and Baltic countries, the share of export to the EU-17 dropped and the share of export to the CEECs rose from 2000 to 2009 (graph 10). However, the change in trade flows seems to be more subtle. Further, the Visegrád group has the smallest inter-group trade with Russia compared to the Balkan and Baltic groups.
2.5. Concluding remarks trade analysis
In the decade before 2000, there was an overall increase in trade shares with Europe compensated by an overall decrease in intra-group trade shares and trade shares with others. Whereas in the decade thereafter, there was an overall decline of trade shares with the EU-17 and a rise in intra-group trade shares. Bussiére et al. (2005) suggests that the declining trade shares with the EU indicate that for the CEECs trade with the EU is almost up to its potential, whereas for intra-CEECs trade there is ample room for improvement. Furthermore, the trade agreements are successful in integrating and developing the region, which enlarges internal supply and demand. The general trade trends appear to apply to individual Eastern European countries as well. From the regional subgroups analysis it becomes apparent that the Eastern European countries moved in remarkably similar fashion when it comes to trade. However, Romania is the only exception. In contrast to the most CEECs, Romania traded little with the Soviet Union and lots with the EU from early on (table A2.5;A2.7). A note to this is that this is in percentage of total trade, whereas in total trade value Romania had the smallest foreign trade (table A2.4;A2.6).
3. TRADE THEORIES
increasing diversity of products. Furthermore, when the CEECs open up their economies for each other and the EU this will give rise to international trade, which according to Paas (2001), is the most effective economic factor in driving economies to transition and integration. According to Husted and Melvin (2004), the benefits from international trade are twofold. First are consumption gains, the CEECs will be able to purchase a larger variety of goods. Additionally, competition will drive down prices as well. Second are production gains, international trade induces production to be clustered in those sectors where their labor is relatively more efficient. In turn, this will increase output and thus real GDP, which enables citizens to enlarge the volume and variety of goods they purchase and thereby increase their collective standard of living. Moreover, a rise in international trade implies an increase in trade openness. Empirical studies on the effect of trade openness on GDP growth generally find a positive relationship between them (e.g. Sachs and Warner, 1995; Frankel and Romer, 1999; Dollar and Kraay, 2002; and Wacziarg and Welch, 2003). Furthermore, according to Aristotelous (2006) open countries are better able to reap larger benefits from lower transaction costs and enhanced competition resulting from improved trade conditions. A regional expansion of this magnitude (i.e. enlargement of EU with ten CEECs) will inevitably alter trade flows between the countries, in particular between the accession countries and the EU-15. Regional trade theories are used to formally analyze what effect such an enlargement has on welfare.
3.1. Regional trade theory
other hand are tailored to country specific needs and thus induces differentiation among treatment and negotiation of margins. Viner (1950) showed that gains from trade exist if not partial and discriminatory but only if all trade barriers are eliminated. Thus, according to the regional theory of Viner (1950) the trade effect of the EU (as a customs union) should be larger than that of the Europe agreements.
4. THEORETICAL FOUNDATION OF THE GRAVITY MODEL
The econometric analysis is based on the gravity model (see Greenaway and Milner, 2002 for a review of this model). Gravity equations are widely used for examining the influence of a diversity of policies considered with trading agreements, currency unions and diverse trade distortions (Bougheas, Demetriades and Morgenroth 1999, De Grauwe and Skudelny 2000, Glink and Rose 2002, Costa-i-Font, 2010). Further, gravity equations are used to determine the superior trade theory (Feenstra, Markusen and Rose, 2001). Leamer and Levinshohn (1995) stated that gravity models have provided „some of the clearest and most robust empirical findings in economics‟. However, in their book Van Bergeijk and Brakman (2010) include the most important developments of the gravity equation and state that at first, the lack of a convincing and unambiguous micro-economic foundation gave the gravity model a somewhat ambivalent reputation. It became known as perhaps useful as an empirical tool, but unsatisfactory from a theoretical point of view. The first and according to Cieslik (2009) and others not very successful attempts to provide a theoretical justification for the gravity equation were made by Linneman (1966), Leamer and Stern (1970) and Leamer (1974). However, in the last two decades the model has again become fashionable due to seminal contributions by Anderson (1979) and Bergstrand (1985). The inspiration for the gravity model comes from Newton‟s law of universal gravitation that states that the force of gravity between two objects is proportional to the product of the masses of the two objects divided by the square of the distance between them (Baldwin and Taglioni (2008):
(1)
To make equation (1) applicable for examination of trade flows the force of gravity is replaced with the value of bilateral trade between country i and j and the masses and are replaced with the trade partners‟ GDPs (in physics G is the gravitational constant):
(2) Where;
indicates bilateral trade between country i and j;
indicates the bilateral distance between the two countries.
Equation (2) explains bilateral trade using economic size and distance. Usually the model explains 70 to 80 per cent of the variance in bilateral trade flows (Van Bergeijk and Brakman, 2010; Baldwin and Taglioni, 2008). However, even though the model could explain a great deal of trade flows the “lack of theoretical underpinnings significantly weakens the credibility of a model, as it introduces a certain degree of subjectivity in the interpretation of the estimated coefficients” (Piermartini and Teh, 2005). Tinbergen (1962) even introduced his equation as “a turnover relationship in which prices are not specified”. In this sense trade is determined by supply potential (exporter GDP), market demand potential (importer GDP), and transportation costs (distance).
Anderson (1979) was the first to attempt to derive the gravity model directly from a theoretical model. His pioneering work showed how to derive a gravity model using the properties of a Cobb–Douglas expenditure system when each good is produced by one country only (Cieslik, 2009). Anderson (1979) provided a sound micro-economic foundation in which he assumes a (weakly) separable social utility function with respect to traded and non-traded goods, each country produces both types (Van Bergeijk and Brakman, 2010). First, the share of country j‟s income that is spent on traded goods ( ) can vary across countries and depends on income and populations size in country j. Second, country j maximizes a homothetic Cobb-Douglas utility function, which is identical across all countries. This implies that when ignoring price discrimination the share that country j spends on exports of tradable goods from country i ( ) is equal for all countries (i.e. varies only with i). Consequently, country j‟s imports from country i can then be expressed as:
(3)
Equilibrium on the traded goods market implies that equals ∑ . Solving for and substituting in equation (3) gives:
∑ ∑ ⁄
note that ∑ ∑ ∑
(4)
(1979) and Bergstrand‟s (1985, 1989, and 1990) work and introduce a method to deal with the complicated price index terms (which are already present in the appendix of Anderson, 1979 and Bergstrand, 1985, 1989, and 1990).
4.1. A simplified derivation of the model
Anderson and van Wincoop (2003) has become the basis for subsequent work constructed with the gravity model, as will it be in this study. In an attempt to write a simplified derivation of the model Baldwin and Taglioni (2006) designed six steps in which to arrive at an unbiased gravity equation.
Step 1: The expenditure share identity
The first step is the expenditure share identity that says that the value of trade flow from country i to j ( ) should equal the share country i has in expenditure of j (Van Bergeijk and Brakman, 2010):
(5)
Where;
= import price from i to j;
= quantity of good sold from i to j;
= share of i in j‟s expenditure; .
is country j‟s expenditure on goods that compete with imports (i.e. tradable goods) (Baldwin and Taglioni, 2006).
Step 2: The expenditure function: shares depend on relative prices
In microeconomics expenditure shares depend upon relative prices and income levels, however, for this first-pass presentation we assume the expenditure share is only depends upon relative prices (Anderson and van Wincoop, 2003; Baldwin and Taglioni, 2006). When adopting the CES demand function and assuming that all goods are traded, depends on the bilateral prices relative to a
price index (Van Bergeijk and Brakman, 2010):
( )
(6)
Where;
(∑ ( ) ) ⁄ is the price index associated with the CES demand structure;
is the elasticity of substitution between varieties;
N = the number of countries;
Note that varieties are defined symmetrically, which allows us to ignore a variety index. Step 3: Adding the pass-through equation
Step three is adding trade costs which are a crucial element in all gravity equations. If indicates
bilateral trade costs, the price in market j equals:
(7)
Where is the mill price of a variety in country I (note again that varieties are defined symmetrically; there is an absence of an index for varieties). After transportation the price in market j becomes . To keep things simple, Anderson and van Wincoop (2003), Baldwin and Taglioni (2006) and Van Bergeijk and Brakman (2010) agree to set =1 as in Dixit-Stiglitz monopolistic competition or perfect competition with Armington goods. Thus equation (7) becomes;
(8) Step 4: Aggregating across individual goods
To describe total trade between two countries with the gravity equation one has to aggregate across varieties. Thus far, the equations involved per-variety exports. In order to arrive at bilateral exports from country i to j, the expenditure share function is multiplied by the number of symmetric varieties that country i has to offer; (Baldwin and Taglioni, 2006).
( ) (9)
The second equality follows from combining equations (6) and (8) in the bilateral trade equation (9). Country i‟s general equilibrium condition is used to compensate for the absence of data on the number of varieties ( ) and producer prices ( ) (Baldwin and Taglioni, 2006).
Step 5: Using general equilibrium in the exporting nation to eliminate the nominal price
Since all goods are traded, the budget constraint states that total output of country i equals total sales to all destination countries j (including country i itself) (Van Bergeijk and Brakman, 2010).
∑ ∑ ( ) (10)
The second equality follows from combining (9) and (10). We can rewrite equation (10) as:
∏
where ∏ (∑ ( ))
Step 6: A first-pass gravity equation
In step six a gravity equation can be derived from inserting equation (11) into equation (9) which gives:
(∏ )
(12) Equation (12) is a microeconomic founded gravity equation (Baldwin and Taglioni, 2006). The main difference between equation (2) and equation (12) can be attributed to the price indices P and ∏ which are the so-called multilateral resistance terms. This implies that according to Anderson and Van Wincoop (2003) bilateral trade does not only depend on bilateral determinants related to these two countries alone but also on the resistance to trade with other countries and on their position relative to the world. For instance, considering multilateral trade resistance of a pair of countries in the EU compared to a similar country pair more isolated from the world economy. The EU country pair faces tougher competition from nearby countries which affects price indices and in turn ensures that trade between two countries in the EU is, ceteris paribus, smaller than trade between two countries that are more isolated (i.e. for which P and ∏ are higher) (Van Bergeijk and Brakman, 2010).
4.2. Determinants of bilateral trade
Initial studies on the gravity model postulated that the amount of bilateral trade is positively related to the incomes of both nations and negatively related to the distance between the trade partners. However, the model could only explain 70 to 80 per cent of the trade variation (Van Bergeijk and Brakman, 2010; Baldwin and Taglioni, 2008). Therefore, subsequent studies tried to expand the explanatory power of the gravity model by adding more factors that may affect bilateral trade (Cieslik, 2009). This study incorporates the established determinants of bilateral trade such as distance and several proxies for economic size such as income and geographical area. Additionally, instead of the more common dummies of cultural proximity, I include a measure that estimates the cultural distance between two trade partners. The next paragraphs describe the determinants, provide a brief theoretical explanation, hypothesize about the effect on bilateral trade and present relevant empirical results.
4.2.1. Income
4.2.2. Distance
Most authors agree that distance is an inadequate proxy for transportation costs. Nevertheless, the results achieved with incorporating distance as a proxy for transportation costs in gravity models have provided some of the most important contribution of the gravity model. For example, on the one hand Limao and Venables (2001) or Combes and Lafourcade (2005) show that by using real data on shipping cost, distance is an improper proxy for transportation cost. On the other hand, Disdier and Head (2008) perform a meta-study consisted of 103 publications using the gravity equation including the distance proxy. Their results indicate that the mean effect of distance on trade is around 0.9, indicating that a 10 per cent increase in distance results in a 9 per cent reduction of international trade flows. Interestingly, according to their results the distance effect has increased (with some increased variance) in the second half of the twentieth century. Thus, two conclusions can be arrived from this; first, distance is negative related to bilateral trade and second, the explanatory power of distance for determining international trade flows in gravity model is becoming stronger (Van Bergeijk and Brakman, 2010). In the case of the EU distance is also proven to be negatively related to bilateral trade (e.g. Aristotelous, 2006; Baier and Bergstand, 2007; Bussiére et al., 2005; Caporal et al. 2008; Linders et al. 2008; Papazoglou, Pentecost and Marques, 2006; Rault, Sova and Sova, 2008; Spies and Marques, 2009). More particularly, Egger at al. (2007) conclude that intra-East trade is significantly more sensitive to distance than intra-West trade. They state that improvements in infrastructure are likely to reduce distance-related trade frictions. In sum, I hypothesize that distance will have a negative effect on bilateral trade.
To calculate distance I use the haversine formula with data on longitudes and latitudes of economic centers of the particular nations involved. The formula is widely used for measuring great circle distance in particular for relatively small distances as in the context of the Europe (e.g. Spies and Marques, 2009; Belke and Spies, 2008).
̂
(13) ̂ √ ( ) ( )
Where is the difference in latitudes and is the difference in longitudes.
4.2.3. Population and area
(1973), Bikker (1987) and Sapir (1981) provide evidence for a negative effect of population on bilateral trade, whereas Brada and Mendez (1983) find prove for a positive relationship (Egger and Pfaffermayr, 2003). According to Egger and Pfaffermayr (2003), the population variable could have either sign from a theoretical point of view (Bussiére et al., 2005). Egger et al. (2003) first cite Bergstrand (1989) who states that a negative relationship between exports and population indicates that the exported products tend to be capital intensive, whereas a negative relationship between imports and population suggests that the traded goods consist of luxury goods. Secondly, Egger et al. (2003) cite Oguledo and MacPhee (1994) who believe that a large domestic market encourages the “division of labor and thus creates opportunities in a wide variety of goods’’, suggesting a positive relationship between population and bilateral trade. However, Papazoglou, Pentecost and Marques (2006) postulate and provide evidence that a country with a big home base indicates large domestic demand that moderates exports and increases imports. Additionally, Papazoglou et al. (2006) find that greater foreign demand (i.e. economic size of trade partner) has a positive effect on trade. Basically, a sizeable trade partner is able to offer and produce a wide variety of goods increasing imports and demands a vast volume of goods raising exports. However, the truly large countries, such as the United States, can produce a considerable portion of the demanded products themselves and thus do not need to import such a large degree of their income. However, because this thesis is from the perspective of the CEECs, which are all relatively small and less developed economies, this negative relationship is not relevant. Therefore, I hypothesize that greater domestic demand moderates export and raises import whereas greater foreign demand raises bilateral trade.
4.2.4. Cultural distance
makes interactions more difficult (Parkhe, 1991). According to Linders et al. (2005), one of the biggest complications arising from cross-cultural interactions is associated with understanding, and more specific with the way individuals perceive the same situation. Differences in perceptions make interactions difficult, increase the probability of failure and impede the development of trust. These are also the factors that Neal (1998) claims to lower the costs of trade. Therefore, large cultural distance is expected to reduce the amount of trade. However, Möhlmann, Ederveen, De Groot and Linders (2009) state that cultural distance can also have a positive effect on trade. They claim that when a firm is deciding how to enter a foreign market it makes a tradeoff between producing locally and producing at a single site to benefit from economies of scale. This is known as the proximity-concentration tradeoff. Among others, Brainard (1997) and Helpman et al. (2004) suggest that choosing between entry modes may result in a positive relationship between cultural distance and trade. Their reasoning is that cultural difference will raise the cost of foreign direct investment (FDI) more than the cost of trade, since FDI requires considerable investment and resource commitment in the local market. Thus, if the costs of a local presence are higher due to cultural distance, this might lead companies to substitute local presence by trade (Möhlmann et al., 2009). The total effect of cultural distance on trade then amounts to a direct negative effect and a positive substitution effect. Empirically, evidence on the relationship between cultural distance and bilateral trade shows negative and positive results, respectively by Möhlmann et al. (2009) and Linders et al. (2005). However, the positive substitution effect exists when cultural distance is so large that local presence is accompanied with high costs associated with cultural differences. Since within the CEECs cultural difference will be relatively small, the direct negative effect will dominate the indirect positive effect. Thus, I hypothesize that there is a negative relationship between cultural distance and bilateral trade.
Cultural distance is computed with the formula developed by Kogut and Singh (1988) which makes Hofstede‟s (1980) dimensions of national culture (i.e. power distance, uncertainty avoidance, masculinity vs. femininity and individualism vs. collectivism) applicable because it provides a single comparative measure based on the differences between two countries in multiple dimensions. It is constructed by taking a weighted average of the squared difference in each dimension divided by the total number of dimensions:
∑* ( ) ⁄ +⁄
measurement in the gravity model by Linders et al. (2005), it will be, to my knowledge, for the first time used to assess cultural differences in Europe and the effect thereof on bilateral trade flows.
4.2.5. EU membership
measurement bias can lead to an underestimated FTA coefficient. Moreover, the difference in empirical findings (see also table A7.1) can be attributed to different specifications of the gravity model, varying estimation techniques, country samples and time spans (e.g. Baier and Bergstrand, 2007; Caporale, Rault, Sova and Sova, 2008; Spies and Marques, 2009). These and more econometric issues are discussed in the following chapter. In sum, the enlargement of the EU with the CEECs will theoretically shift the sources of supply to a lower cost point, which, according to regional trade theory, creates trade. However, there are econometric difficulties when measuring a gradual trade effect. Moreover, successful EA´s have improved trade conditions and competition prior to the CEECs joining the EU. Nevertheless, I expect a positive trade effect for subsequent EU membership. Moreover, according to Bussiére et al. (2005) the potential in EU-CEEC trade growth is limited. Consequently, I expect the intra-CEEC trade effect to be larger than the inter EU-CEEC effect.
4.2.6. Individual country EU membership
Aristotelous (2006) states that empirical literature that finds that the effect of a trade agreement on trade is positive, significant and large, focuses almost exclusively on the overall impact trade agreements have on trade while paying very little attention to whether there are significant differences across the individual countries involved. This thesis attempts to narrow the gap by examining whether the impact the EU has on trade is widespread across the CEECs. In the historical trade overview, I found that the individual countries were in line with the general trade trends for the CEECs. Therefore, the hypotheses formed for the CEECs as a region are similar to the hypotheses for the individual CEECs. Thus, I expect to find positive EU trade effects on intra-CEEC as well on inter EU-CEEC trade. However, similar to the overall EU trade effect, due to econometric conditions and EU-CEEC trade that is up to its full potential (e.g. Baier and Bergstrand, 2007; Bussiére et al. 2005; Caporale et al., 2008; Spies and Marques, 2009) I expect the EU, especially the inter EU-CEEC, trade effect to be small. Further, following the line of reasoning of Bussiére et al. (2005), I expect that countries with high and stable trade shares with the EU-17 will have the lowest inter-EU trade effect.
4.2.7. Regional subgroups
subgroups (e.g. Bröcker and Herrmann, 2001; Cornett and Iversen, 1998; Paas, 2001 and 2003). Therefore, they might experience a trade effect larger than the others, when they join the EU. However, according to Bussiére et al. (2005) the Baltic and Visegrád depend too heavily on trade with the EU. Therefore, not mot potential for growth exists. On the other hand , Bulgaria and Romania were the last to join the EU and were seen as the least performing of the CEECs (e.g. Crabbé and Beine, 2009; Herderschee and Qiao, 2007; Brenton and Manchin, 2002; Bojkov, 2004). Further, Spies and Marques (2009) found that imports from Bulgaria and Romania rose up to 80%. Thus, most potential for trade growth exists in the Balkan.
5. TECHNICAL SPECIFICATION OF THE GRAVITY MODEL
In this chapter, I will build upon the equations and determinants of bilateral trade from the previous chapter. Bergstrand (1985) is one of the first authors to derive a theoretically founded gravity equation. His equation was the basis for almost all subsequent gravity model studies and is written as follows:
( ) ( ) ( ) ( ) ( ) ( ) (15)
Where:
is the value of exports from country i to country j,
( ) is the value of GDP in country i (j);
( ) is the population in country i (j);
is the distance between country i and country j;
is a dummy variable which represents a trade agreement between the countries i and j;
is the residual term.
The next step involves creating linear equation by taking logarithms of Bergstrand‟s (1985) variables in equation (15). First, adopting the log-linear form assures that any constant term is cancelled out (Anderson, 1979). Second, since the model is now linear, it is readily estimable using OLS (Westerlund and Wilhelmsson, 2006). Further, I add the cultural distance measurement along with dummies for adjacency, EU membership, CEEC trade, EU-17 trade, three subgroups and the individual country dummies as in Aristotelous (2006) (see table A6.1 for a description of the variables and dummies). The equation now consists of standard determinants ( ), a cultural distance measure ( ) and dummies ( ):
Note that for the import model is replaced with , which naturally denotes real bilateral imports
from country i to j. Due to high levels of correlation the population variable is replaced with a geographical size variable (for explanation see chapter 6.2; for description see subparagraph 4.2.3). The equation without dummies is then as follows:
( ) ( ) ( ) ( ) ( ) ( )
(17)
Several dummies are designed to estimate and decompose the effect on trade of joining the EU. First, the EU membership dummy is split into an intra-CEEC and an inter EU-CEEC dummy. Thereafter, in a similar method as Aristotelous (2006) I designed the individual country dummies. For example, the dummy EU/BGR in model 3 is one when bilateral trade is between Bulgaria and another CEEC and both have joined the EU (see also table A6.1). For the export (import) equation, the result is a trade elasticity that explains the trade effect on Bulgarian exports (imports) with fellow CEECs of joining the EU. The full export equation used for model 3 then becomes:
( ) ( ) ( ) ( ) ( )
(18)
The results of the intra-CEEC trade effect (model 3) are compared to the inter EU-CEEC trade effect (model 4) and to the EU membership trade effect (model 5). In order for this comparison, the country dummies of equation (18) are changed to denote trade between the EU-17 and the CEECs with EU membership (model 4) and between EU members (model 5). To estimate the regional subgroup trade effect as a result of subsequent EU membership, the regional subgroup dummies are changed to denote intra subgroup trade whilst both trade partners being EU members (model 6).
5.1. Econometric issues
over- or underestimated due to the potential endogeneity of this variable. Baier and Bergstrand (2007) confirmed these findings and state that the regional agreement variable is not exogenous. Therefore, due to unobserved heterogeneity and omitted variables, using cross-section data will result in biased findings. The consensus in gravity model literature is that in order to take endogeneity into account, panel data with fixed effects methods were shown to be best suitable (e.g. Baier and Bergstrand, 2007; Spies and Marques, 2009; Rault, Sova and Sova, 2008). Endogeneity of variables can fall under any one of three categories: omitted variables, simultaneity and measurement error (Wooldridge, 2002). Even though all three categories may contribute to endogeneity caused by FTA, Baier and Bergstrand (2007) argue that the most important source is the omitted variables bias. The categories of econometric issues along with the treatment effect will be discussed next.
5.1.1. Omitted variables
Baier and Bergstrand (2004) and Caporale, Rault, Sova and Sova (2008) present a model of economic determinants of free trade agreements. They find evidence that pairs of countries that have FTAs tend to share economic characteristics. Subsequently, theory suggests that these economic characteristics increase the net welfare gains of an FTA. For example, according to Baier and Bergstrand (2007) the probability a country pair has signed a FTA increases the more similar and larger their GDP, the closer they are located to each other geographically, the more remote the pair is from the rest of the world and the greater the difference in their factor endowments. However, these factors also tend to explain bilateral trade flows. Baier and Bergstrand (2007) find that the best way to treat this econometric issue of endogeneity is by using panel data and fixed effects.
5.1.2. Simultaneity bias
GDP is measured as a function of net exports and is thus potentially endogenous to bilateral trade flows (e.g. Frankel and Romer, 1999). However, Baier and Bergstrand (2007) feel this can largely be ignored since the exports mentioned in this literature are concerned with total exports as opposed to bilateral exports, which tend to be a small fraction of any country‟s multilateral exports. Moreover, while GDP is related to net exports, GDP's connection to gross exports is much less direct. Lastly, among others, Frankel (1997) accounted for potential endogeneity of national incomes econometrically using instrumental variables and reported insignificant changes.
5.1.3. Measurement error bias
5.1.4. Treatment effect
In econometric literature, the treatment effect is concerned with methodological issues regarding a binary variable (FTA) on a continuous variable (bilateral trade flows). The treatment effect suggests that bilateral trade flows depend on whether a country pair has signed a FTA. However, as Baier and Bergstrand (2007) rightfully state the fundamental econometric dilemma is that one can observe only one situation or the other. Therefore, Magee (2003) approached the relationship between bilateral trade and FTA as a simultaneous-equations system. He uses 2SLS to estimate the effect of FTAs on trade flows. Nevertheless, according to Baier and Bergstrand (2007) this does not solve the endogeneity bias issue. It is by now well established that the results of the gravity model with cross-sectional data are unreliable. However, many authors agree that strong and reliable conclusions can be drawn from the gravity model when applied to panel data. When using panel data one can use either fixed effects or random effects. Wooldridge (2002) states that when the number of time periods exceeds two, the fixed-effects estimator is more efficient under the assumption of serially uncorrelated error terms. Nonetheless, according to Rault, Sova and Sova (2008) the choice between using fixed effects (FEM) or random effects (REM) depends on two important issues, namely its economic and econometric relevance. They state that from an economic perspective, there is a preference for FEM when there are unobservable time invariant random variables that are hard to quantify and which might affect some explanatory variables and trade volume simultaneously. However, from an econometric perspective Rault et al. (2008) base their reasoning on Baier and Bergstrand (2005), they suggest that FEM is preferable to REM because rejecting the assumption of no correlation of the unobservable characteristics with some explanatory variables is less likely. Rault et al. (2008) conclude by saying that FEM allows for unobserved or misspecified factors that simultaneously explain trade volume between two countries and lead to unbiased and efficient results. Thus, following the criteria of first, Wooldridge (2002), second Rault et al. (2008) and third Baier and Bergstrand (2005) all suggest implying fixed effects. Fourth, as suggested by Brooks (2008), I conducted a Hausman test to determine whether random effects are to be preferred over fixed effects. The test resulted in a p-value of less than 1% (table A6,2), which indicates that the random effects model is not appropriate and thus the fixed effects model is preferred. Fifth, recent relevant papers in the context of this study (e.g. Aristotelous, 2006; Baier and Bergstrand, 2005; Belke and Spies, 2008; Van Bergeijk and Brakman, 2010; Rault et al., 2008; Spies and Marques, 2009) are all in consensus of the superiority of FEM over REM.
5.1.5. Technical implications of using fixed effects
( ⁄ ) (19) Where;
is a dummy that is one when i is the exporter and zero otherwise;
is one when j is the importer and zero otherwise;
is the error term.
Furthermore, the income terms are added to function as correction factors for the bilateral trade flows, additionally the coefficients and measure the multilateral resistance terms (i.e. ∏
and ( ) ) (Van Bergeijk and Brakman, 2010). A drawback of FEM is that the estimates cannot be used to measure comparative-static effects concerning alterations in trade costs. However, this disadvantage does not affect the focus of the thesis and can therefore be neglected.
5.1.6. Zero trade flows
As a last econometric issue, Van Bergeijk and Brakman (2010) mention dealing with zero trade flows. In the sample of this thesis Malta, Cyprus and Luxemburg are the only countries with zero trade flows. These could be the result of rounding errors, missing observations or truly zero trade flows. However, due to their inconsiderable existence in the database and non-effect on results the issue is resolved using the standard procedure of Linnemann (1966). This procedure implies adding a small constant to all trade flows in order to estimate a log-linear equation.
6. DATA
6.1. Data selectionMoreover, in order to generate great circle distance between a country pair I used the haversine formula (13). The collection of these variables allows for a standard gravity equation examination. Additional to the standard gravity model variables this thesis‟ model also includes a measurement of cultural distance. Specifically, instead of using cultural familiarity proxies such as common languages and colonial ties I compute cultural distance using formula (14) and data on Hofstede´s dimensions of national culture. Unfortunately, Hofstede did not study Cyprus, Slovenia, Latvia and Lithuania. I resolved the issue by taking estimates of other researchers for these nations (namely Heuttinger (2007) for Latvia and Lithuania; and Bertoncelj and Kovač (2008) for Slovenia). The initial time span of 10 years was chosen to have data on sufficient years after and prior to EU entry. However, as can be seen from the trade graphs, 2009 shows an extreme case of economic turmoil and is therefore eliminated from further examination to ensure that it does not influence the robustness of the results. Graphs 11 to 13 show how the economic situation has affected the real GDP of the CEECs compared to the EU-15. The export figures are f.o.b. (i.e. free on board) whereas the import figures are c.i.f. (cost insurance freight). F.o.b. is a term of sale under which the price invoiced by a seller includes all charges up to the port of departure, while c.i.f. is a term of sale that includes all charges up to the port of destination. Thus, the difference lies in who pays the cost of insurance and transportation in between the port of departure and destination. Consequently, the larger the distance between the two ports the larger the difference between f.o.b. exports and c.i.f. imports. Nevertheless, using both export and import data creates a complete image that allows for unbiased interpretation of results.
6.2. Data description
A6.3). Furthermore, as stated the import data do not mirror the export data because f.o.b. priced goods do not include transportation costs between the port of departure and destination. The implication of this difference is that the imports values are not only higher than the export values, the difference also becomes greater as distance increases. Hence, the differential of minima is lower than the differential of maxima (table A6.3). Table A6.4 provides evidence of the asymmetry i.e. the correlation between real (nominal) import and real (nominal) export is 0.9 (0.90). The difference has an important implication for this research; since the export and import equations do not present mirror results they cannot be used as a robustness check. Another consequence from the correlation results is concerned with population and geographical size. Specifically, to ensure the results are not biased, one of these highly correlated variables needs to be excluded from further examination. I exclude population because it is highly correlated with geographical size and GDP. Geographical size is preferred since it shows lower correlation with GDP than population (Table A6.4). In those studies that apply gravity models authors include variables as either one variable that is computed as the logarithm of country i times country j or as two separate variables that present the logarithm of country i and the logarithm of country j (e.g. GDP(ij) versus GDP(i) and GDP(j)). When modeling the bilateral trade flows of the CEECs, it became apparent that equations with two separate variables instead of one result in substantial higher R-squared and t-values. Furthermore, employing separated variables has the advantage of allowing for a more specific interpretation of the variables. Therefore, the models will include variables that are separated for exporting and importing countries. Moreover, the results are derived using fixed effects and data from 2009 is excluded from the database for previous explained reasons.
7. RESULTS AND DISCUSSION
The nine models differ in the way they estimate the trade effect as a result of subsequent EU membership. The first model is a standard gravity equation that allows for a sound comparison with other studies. The model measures the EU membership trade effect with a single dummy. The second model makes a distinction between intra-CEEC and inter EU-CEEC trade. The third model in turn replaces the EU-CEEC dummy with country specific dummies, whereas the fourth model replaces the EU-17 dummy with country specific dummies. The fifth model includes country dummies denoting trade between EU members. The results are presented in table 1 and discussed below. The sixth to ninth models analyze the trade effect for the regional subgroups more specifically (table 2).
7.1. Difference between the exports and imports model
are more subject to variation in distance than exports values. Naturally, this results in the distance coefficient having a higher weight in the imports model than in the exports model. Further, for the Balkan and Baltic countries distance and thus transportation costs (i.e. the difference between exports and imports values) are relatively small, resulting in small differences between exports and imports coefficients. However, for the Visegrád countries, distance plays a more important role and thus the difference between the exports and imports coefficients is greater (for a map see figure 3). Furthermore, countries that share a common border have lower transportation costs and thus this dummy has a smaller weight in the imports models. In sum, the differences between the export and import equations can be attributed to the difference in which they are priced.
7.1. The basic gravity equation variables
TABLE 1: Results Model 1
Overall EU effect
Model 2 Inter vs. Intra effect
Model 3 Country intra effect
Model 4 Country inter effect
Model 5 Country EU effect
Variable Exports Imports Exports Imports Exports Imports Exports Imports Exports Imports
Thus, the results in table 1 are in accordance with theory and empirical findings. Furthermore, Egger at al. (2007) show that intra-CEEC trade is significantly more sensitive to distance than intra-EU-15 or inter EU-CEEC trade. Consistently, the distant coefficients in table 1 are indeed higher than distant coefficients in studies that focus more on the West (e.g. Aristotelous, 2006; Bussiére et al., 2005; Papazoglou et al., 2006). Fourth, adjacency indicates; strong historical ties, cultural familiarity and low trade cost. Therefore, adjacency induces higher bilateral trade. Consistently, adjacency is found to be positively and significantly related to bilateral trade throughout all models which is also in accordance with literature (e.g. Aristotelous, 2006; Egger et al., 2003; Bussiére et al., 2005; Papazoglou et al., 2006).. In sum, I can conclude that the standard gravity equation variables are consistent with the hypotheses. The most important thing that can be derived from the results being in line with the expectations is that it signals the strength of the models. Particularly, even though this thesis does not attempt to add value by estimating these standard variables correctly is does indicate the explanatory power of the models.
7.2. Cultural distance and trade
costs of producing locally increase faster than the costs of trade because it requires closer interactions with a wide variety of local stakeholders such as employees, unions, suppliers, and government agencies. Furthermore, an increase in cultural distance raises the differences in organizational and management practices (Kogut and Singh, 1988). These differences cause FDI in culturally dissimilar environments to be difficult and costly (Gómez-Meija and Palich, 1997). Therefore, companies expanding into such regions tend to opt for entry modes that require relatively little resources, such as exporting (Dunning, 1993). However, this indirect positive effect starts to be of considerable effect when the countries are culturally highly distant. Therefore, the limited cultural and physical distances in Europe might explain the lack of a substitution effect found in the results. Egger et al. (2007) and Cieslik (2009) confirm that trade determinants alter in importance as the tested sample changes. In sum, in the sample of this thesis, which has relatively little cultural dissimilarities, no positive substitution effect exists. Thus, this thesis supports the hypothesis of Möhlmann et al. (2009) that cultural distance induces an increase in trade costs and thereby decreases bilateral exports as well as imports. Consequently, the results reject the theoretical hypothesis of Linders et al. (2008), Brainard (1997) and Helpman et al. (2004) of a positive relationship between cultural distance and bilateral trade, at least in the context of Europe.
7.3. The EU Membership trade effect
7.4. Intra-CEEC versus inter EU-CEEC trade effect
turn, some authors argue that this potential in intra-CEEC trade is because intra-CEEC trade has been kept artificially low (Bojkov, 2004) and that improved trade conditions adjusted trade quickly towards their natural levels. However, I argue if this were so it had to be a legacy of the CMEA period and gravity equations estimating trade in subsequent periods would have shown more potential (e.g. Economic bulletin for Europe, 1993). A more logical explanation is the difference in development of economies between the CEECs and the EU-15. For example, Egger at al. (2007) argue that the CEECs and the EU-15 will remain responding differently to distance-related trade frictions. They state that investments that improve infrastructure (i.e. decrease transportation costs and hence the importance of distance) will lead to a far greater increase in trade for the CEECs than for the EU-15. Taking in consideration; first, the arguments of Bussiére, et al. (2005) of trading with the EU-15 being up to its full potential. Second, the relatively low trade shares of intra-CEEC trade (table A2.6; graph 5 and 8 to 10). Third, the fact that these economies are growing at a faster rate than the EU-15 (graph 11 to 13) and last, the results showing higher trade effect coefficients for intra CEEC- than inter EU-CEEC trade (table 1), I agree with Bussiére et al. (2005) that intra-CEEC trade has more potential than inter EU-CEEC trade. However, the main finding of this model is that the trade effect for the EU-CEECs as a result of subsequent EU membership is mainly due to declining trade with the EU-17.
7.5. The EU trade effect of the individual CEECs
