The ETSG Conference – Brussels, 14 to 16 of September 2001
EU Enlargement and Trade Adjustments1 Ana Paula Africano-Silva∗° Paulo Teles*
Abstract The purpose of this paper is to analyse the progress of trade integration that Central and Eastern European Countries have experienced into the European Union, following the collapse of communist regimes and the implementation of preferential trade. By applying the gravity model to bilateral trade, it is possible to evaluate the level and progress in trade integration between Central and Eastern European Economies and the EU. This study considers 10 Central and Eastern European Economies, plus Turkey, all identified as candidate countries to EU membership. Based on estimations of the gravity model it is possible to compare, for pairs of countries, their level of trade integration with the average level of the whole group. To this end, we use intra-EU trade as the reference group against which the progress of trade integration is assessed. Finally, our results are compared to those of other studies.
1
Draft version (this paper is undergoing major changes) please do not quote Faculty of Economics and CEMPRE – University of Porto ° Corresponding author:
[email protected] ∗
1
1. Introduction
According to economic theory, the existing geographical and cultural proximity between the European Union (EU)
2
countries and the Central and Eastern European Countries
(CEECSs)3 should sustain a much higher level of bilateral trade than that observed at the end of the eighties. In a setting of political democratization and trade liberalization the natural expectation was for increased mutual trade. Such result was foreseen in early research by Hamilton and Winters (1992), and by Baldwin (1994). These expectations were further reinforced once trade policy measures were passed on to implement preferential trade between the EU and the CEECs under the so-called Europe Agreements. The main objective of this paper is to assess the degree of trade integration that these countries reveal into the EU market, taking as reference the degree of integration in intraunion trade of EU countries.
2. The Implications of Trade Liberalization: EU versus CEECs Western Europe has, since the 60’s, been the stage of the most interesting and successful process of regional trade liberalization among an increasing group of countries. Despite the long lasting record of economic integration among these trade partners, the nineties were one of the most formidable periods in this process as the ambitious (internal) goals of further economic and monetary integration were topped with new external demands from neighbouring Eastern European countries. At the internal level the setting was one for deeper and greater economic integration. The Single Market program was completed by early 1993 and aimed at reducing cross-border transaction costs through the elimination of technical, administrative and fiscal non-tariff trade barriers. In 1995, Austria, Finland and Sweden acquired full-membership. Finally, in January 1999 11 (of 15) EU countries entered in the last stage of Monetary Union eliminating, therefore, 2
EU stands for the 15 member countries unless stated otherwise. CEEC 10 stands for the following eastern countries: Bulgaria, Czech Rep., Estonia, Hungary, Latvia, Lithuania, Poland, Romania, Slovakia and Slovenia. 3
2
money conversion costs and exchange rates volatility. All these policies were expected to have a positive impact on intra-EU trade as each of them favors market integration of goods and services. At the external level, Western Europe had to face the political and economic challenges brought in by the democratization winds sweeping across Eastern Europe. The main concern was to provide a framework to support and facilitate the gradual economic and political integration with the West. The best solution the EU was able to offer was an old European policy: preferential trade relations were established under the so-called Europe Agreements in the early 1990s4. These were signed on a bilateral basis between the EU and each of the CEECs and despite their strong political motivation the economic content was crucial for most Eastern countries. The Europe Agreements provided a time schedule for asymmetric trade liberalization between signatories: with the EU reducing trade barriers faster than the Eastern countries. Tariff barriers on basic industrial imports were removed by early 1994 for Poland and Hungary, Czech Republic and Slovakia, by early 1995 for Romania and Bulgaria, and in 1997 for the Baltic States and Slovenia. Overall, by early 1999 most of CEECs exports of industrial goods – including sensitive products such as textiles and clothing and steel products - into the EU were free from duties and quantitative restrictions whereas similar conditions for exports from the EU into the CEECs were met by January 2000 (Piazolo, 2000). The consequences of those trade policy measures were immediate and immense, particularly so for the CEECs as we will see bellow. To illustrate such changes table I shows the average annual growth in EU trade for a period that covers most of the nineties. From these calculations it becomes evident that intra-EU trade grew in average faster than trade with the rest of the World. Trade with the CEECs was, nevertheless, a major exception. While intra-EU exports grew on average 8,5% a year, exports to the rest of the world increased at a slower pace of 7,5% a year. On both standards (intra and extra-union) EU exports to CEECs grew dramatically faster: values range from the lowest rate of 12,7% a year on exports to Bulgaria, to the highest rate of 52,5% a year on exports 4
In 1991 with Poland and Hungary; in 1993 with Bulgaria, Czech Rep., Slovakia and Romania; in 1995 with the Baltic States and with Slovenia in 1996. 3
to Estonia. Turning to imports we find a very similar pattern. Intra-EU imports grew faster than extra-EU imports - at 7,5% and 5,8% a year respectively – but EU imports from CEECs expanded much faster: imports from Poland grew at the slowest pace of 14% a year, whereas imports from Estonia grew the fastest at 53% a year. This is the first evidence that major changes in EU-CEECs trading relations have been taking place. As a result one can guess that the degree of market integration between eastern and western economies has improved – a matter addressed below. However, one should not neglect that such intensification of trade relations has been happening in a setting of increased market integration within the European Union itself. Still, from those figures it is not possible to conclude on the market dimensions of such trade adjustments, given that high rates apply to small values in the case of EU-CEECs trade, whereas small rates apply to big values in the case of intra-EU trade. To clarify this question table II shows how the trade changes reported above translate into market shares of EU imports and exports. It turns up that over that period CEECs have more than doubled their shares in EU total imports and exports, still the values at stake are very small. Overall these countries more than doubled their market-shares of EU imports and exports. By 1999, CEE countries were the origin of 3,75% EU imports (from 1,8% in 1992) and the destination of 4,50% of EU exports (from 2,15% in 1992). We also found out that trade expansion among EU members translated into market shares changes of similar size of those identified for CEECs: intra-EU trade gained also 2% points on total imports – from 59,5% to 61,7% - as well as on total exports - from 61,5% to 63,5%. Therefore, those trade adjustments that seem to have very different magnitudes are, in fact, of very similar dimension. Finally, from this perspective EU trade became, over the nineties, more biased towards the continent, which ought to be the result of new market integration with eastern economies under preferential trade relations and stronger market integration within union trade partners. Looking at the geography of EU-CEECs trade (table III) one realizes that behind the overall figures reported there is a very unbalanced reality as just few countries are responsible for most of the changes. Just three countries account for more than two thirds
4
of EU trade with the CEECs. The Czech Rep., Hungary and Poland are responsible for approximately 70% of EU imports from, and 68% of EU exports to the CEECs. It is also noticeable that CEECs trade with the EU is highly concentrated in Germany. Approximately 43% of CEECs imports from the EU came from Germany whereas 48% of CEECs exports to the EU go into the German market. Finally, all these numbers reveal a very asymmetric reality: whereas trade with the EU makes up for 50% to 70% of individual CEECs total trade; EU’s trade with the CEECs roughly amounts to 4% of total trade.
3. Method The gravity model of trade has been generally applied to the study of the determinants of bilateral trade volumes between pairs of countries or groups of countries. Basically, the gravity law from Physics is applied to trade by stating that bilateral trade flows between any two economies depends on (mass) their market size - given by the GDPs, the socalled gravity variable - and distance. Initially, the absence of a proper theoretical foundation5 was the reason for several criticisms of the usefulness of the model. However, it has been established that the simpler equation can be the reduced form of trade models based either on some special assumptions of the factor proportions model (Deardorff, 1998), or on increasing returns to scale and product differentiation models or based on special cases of both (Evenett and Keller (2000), and Shelburne (2000). The gravity equation has been widely used on empirical research on trade providing good insights about the determinants of trade volumes. It has also been useful to identify the extension of preferential trade amongst sub-groups of countries and to compare actual trade with potential trade - according to a certain benchmark. The specification of the gravity equation we use in this study considers that trade flows 5
For a comprehensive review of the debate on the gravity model theoretical foundations see also Bergstrand (1985) and Helpman (1998). 5
between any two economies are determined by three sets of variables: (1) those reflecting the potential demand for imports; (2) those reflecting the potential supply of exports; and (3) factors that favor, or obstruct bilateral trade it took the following form: X ij = β 0 + β 1GDPi + β 2 GDPi / POPi + β 3 GDPj + β 4 GDPj / POPj + β 5 DISTij + β 6 BORDij + ε where: •
Xij is the euro value of exports from country i to country j;
•
GDPi, GDPj are, respectively, the gross domestic product of exporting country i, and of importing country j in euros and PPP;
•
GDPi/POPi and GDPj/POPj are, respectively, per capita income of countries i, and j, in euros and PPP;
•
DISTij is the geographical distance between capitals of the exporting and importing countries, in kilometers;
•
BORD is a dummy variable that equals one for pairs of countries with a common border, or zero otherwise;
•
eij is a log normally distributed error term.
All variables included are in logs6. Therefore, each parameter value can be seen as the elasticity of the dependent variable in relation to each of the independent variables. The variables GDPi and GDP,j reflect the economic size of the exporting country i and the importing country j and they capture the potential for export supply and import demand. These variables are expected to influence Xij positively. The per capita income variables (GDP/POP) do intend to capture the idea that rich countries tend to trade more densely than poorer countries. These variables are also expected to have a positive impact on Xij. The geographical distance variable captures transport costs, transaction costs and producer’s horizons. Transport costs increase with distance and affect negatively trade, and transaction costs stand for other factors hampering trade such as border procedures and other non-tariff barrierers to trade. Finally, the variable BORD captures the fact that countries with a common border tend to have some cultural and linguistic similarity, and 6
Each parameter value can be seen as the elasticity of the dependent variable in relation to each of the independent variables. 6
have better information on each other domestic markets, therefore with an expected positive impact on trade. Most studies estimate the gravity model for a large set of countries - studies on trade among developed economies normally include all bilateral trade between OECD countries. Trade specificities of particular sub-groups are than picked up (identified) through dummy variables. The estimated equation stands for the average (geographical) pattern of trade amongst the included countries, or the “normal” trade. A similar procedure has been used in several previous studies that have assessed the progress of CEECs trade integration into the EU economy. Wang and Winters(1992), Hamilton and Winters (1992), Gros and Gonciarz (1996) and Nilson (2000) have run the gravity equation for a sample including most of OECD countries. The estimated coefficients were than used to calculate the potential EU-CEECs bilateral trade and compared with their actual values. Thus the CEECs trade integration in the EU has been evaluated by taking as a benchmark the normal intra-OECD trade. In our study we depart from this approach for two reasons. First, because the EU is so unique in terms of trade regime/policy - its degree of economic integration is not matched by any other group of countries – making it plausible that the gravity determinants of intra-EU trade are different from those of intra-OECD trade. Second, if this argument is correct then a more appropriate procedure to evaluate the progress of trade integration of CEECs into the EU would be to take as a benchmark the so-called normal intra-EU trade - as given by the gravity determinants.
4. Data and Results In this context, two different regressions of the gravity equation were run. In one case, the gravity equation was applied to all bilateral trade flows amongst EU members. This is a highly homogeneous group of countries in terms of both economic regime and trade integration and the estimated coefficients stand for the gravity determinants of trade between countries that enjoy a common degree of trade integration. In the second case, the gravity equation was applied to bilateral trade between each of the 15 EU members
7
and each of the 11 candidate countries7 (CCs). It is, therefore, possible to compare the values that gravity determinants of trade take in the two sets of trade relations. We use 1998 and 1999 data which are the most recent available. In a similar fashion to what has been done in other studies8, average values for the two years were used in order to reduce the effects of temporary shocks. Trade data are from the EUROSTAT - Comext database, and are expressed in euros. The income variables are expressed in euros and adjusted for PPP and the source is the European Commission. 4.1 The Gravity Determinants of Intra-EU Trade Table 4 reports the ordinary least squares estimates of regression one. In this case, the gravity equation was run for intra-EU trade. The sample refers to bilateral trade between the 15 EU members, with Belgium and Luxembourg taken together. When there are n countries, there are nx(n-1) trade flows, thus we have 14x13 = 82 observations.
TABLE 4 - ESTIMATION RESULTS OF THE GRAVITY MODEL - EU15 Variable Intercept
GDPi GDPi / POPi GDPj
Estimate -23.273 0.788
t statistic -7.363 19.319
2.250
8.393
0.859
21.084
DISTij
-0.756
-7.897
BORD ij
0.538
3.712
R2
0.893
F statistic
295.1 (p-value = 0)
All estimates are significant (p-values = 0).
The results are basically consistent with the theoretical expectations. The exception is the coefficient for the per capita GDP of the importing country – this variable was statistically not significant and was thus excluded from the model. All the other estimates 7
The candidate countries include the 10 CEECs and Turkey.
8
are significant at the 1% level and have the expected sign. The explanatory power of the model is good as the included variables were able to explain 89% of trade variations between pairs of countries. The estimated coefficients of GDP variables have a greater magnitude for the importing than for the exporting countries, whereas the coefficient for per capita income of the exporting country is much bigger than has been reported by other studies. The main differences between our results and those observed on equivalent studies that consider larger samples of OECD countries are reported with some possible explanations: •
The magnitude of the coefficient of GDP of the exporting country i is lower than in other studies. This may reflect the fact that EU economies enjoy, by international standards, an above average market integration - the share of exports to EU markets range from 50% to about 85%. Therefore, the scope for additional exports to each other markets as a result of increased production is smaller.
•
The GDP per capita of the importing country j has no impact on exports from country i to j. This is possibly explained by the smaller dispersion of income per capita among EU countries than that observed in larger samples. Also high per capita income is associated with intra-industry trade rather than with inter-industry trade. The similarity of per capita incomes may result in similar patterns of import demand for differentiated goods across EU countries.
•
Finally, the estimated coefficient of the GDP per capita of exporting countries is much larger than usual. This means that, within the EU, increased income per capita has a higher positive impact on intra-EU exports than usual. This could also stand for the fact that intra-EU trade is mostly intra-industry trade of differentiated goods. This type of trade is bigger among highly developed countries as the scope for the demand and the supply of differentiated goods is stronger.
Note: These results do suggest that the gravity determinants of intra-EU trade are somehow different from those exhibited by OECD trade which is the basis of our 8
Bayoumi and Eichengreen (1995), Nilsson (2000), Hellvin and Nilsson (2000). 9
argument to take this relation as the benchmark to evaluate EU-CEEC trade. 4.2 The Gravity determinants of trade between the EU and the Candidate Countries When considering trade between the EU and the candidate countries there is a striking change in the sample of trade partners included in the study. They are far less homogeneous in terms of economic development. All candidate countries, except Turkey, are transition economies and for that reason have less efficient markets than EU economies. It was therefore plausible that gravity determinants could affect trade differently from that found for intra-EU trade. In such case, it was than possible that the gravity determinants of trade would differ depending on which stand point trade was being considered: the EU, or the candidate countries. In order to investigate these hypotheses three different regressions were run for the gravity equation, and the ordinary least squares estimations are reported in table 5: columns I, II and III. TABLE 5 - ESTIMATION RESULTS OF THE GRAVITY MODEL- EU15 VS. CANDIDATE COUNTRIES
Variable Intercept
GDPi GDPi / POPi GDPj DISTij BORD ij
I Trade
II Exports EU15-
III Exports
EU15/Cand.Count. Estimate t statistic -4.829 -3.133 0.779 17.863
Cand.Count. Estimate T statistic -11.399 -2.766 0.754 12.459
Cand.Count.-EU15 Estimate t statistic 0.811
14.168
0.503
4.188
1.250
3.225
0.839
20.210
0.840
15.719
0.868
16.161
-0.979
-10.094
-1.047
-10.572
-1.124
-11.792
0.426
1.677*
R2
0.750
0.784
F statistic
180.7 (p-value = 0)
135 (p-value = 0)
8710 (p-value = 0)
All estimates are significant (p-values = 0, except * with a p-value = 0.095, thus significant at the 10% level).
In regression I we looked for the gravity determinants of trade between the EU economies and the candidate countries. In this case, the gravity equation was run for bilateral trade between each of the EU9 members and each of the eleven candidate 9
Belgium and Luxembourg were taken together resulting in a sample with 14x11x2=308 trade flows. 10
countries considered in this study. The regression results reveal that the coefficient of the income per capita of the importing country (j) is not significant, and, as in the previous case, it was excluded. All the estimates of the included variables have the expected sign and are significant at the 1% level, except that attached to the Border variable that is significant at 10%. The explanatory power of the model is good as the included variables were able to explain 75% of trade variations between pairs of countries. Comparing the magnitude of the gravity coefficients of this regression with those of the intra-EU trade there are some noteworthy differences: •
First, the positive impact that per capita income of the domestic economy has on exports to trade partners is considerably smaller than that observed on the intra-EU trade. A possible explanation is that this set of economies is more heterogeneous in terms of economic development. Hence, in average the scope for intra-industry trade of differentiated goods is lower than in the EU case.
•
Second, the geographical distance has a stronger negative effect on bilateral trade between the EU and the candidate countries than on intra-EU trade. This may be explained by the fact that the average distance between pairs of countries is bigger in this sample. It may also reflect the lower degree of market integration between the EU and the candidate countries, and therefore, the higher transaction costs involved on trade compared with those observed on intra-EU trade.
•
Finally, the border variable has in this case a lower positive impact than in the intraEU trade.
4.2.1 - The Gravity determinants of EU exports to the candidate countries In Regression II we looked for evidence on the gravity determinants of EU exports to the CCs. In this case, the gravity equation was run for exports from each EU country to each candidate country, resulting in a sample with 14x11 trade flows; The first results revealed that per capita income of importing country, and the binary 11
variable Border were not significant and thus they were excluded from the model. All the estimates of the included variables are significant at the 1% level and have the expected sign. The explanatory power of the model is good as the included variables are able to explain 78% of trade variations between pairs of countries Here we focus on similarities and differences between these estimations and those found using the bilateral trade between the EU members and the CCs: •
The GDPs coefficients are of similar magnitude in the two regressions;
•
Yet the per capita income of the exporting countries reveals a stronger positive impact on EU exports to the CCs than that found out on bilateral trade between the two sets of countries, still the magnitude of the coefficient is half of that in intra-EU trade.
•
Finally, distance has a stronger negative effect on these trade flows
4.2.3 - The Gravity determinants of CCs exports to the EU In Regression III we looked for evidence on the gravity determinants of CCs exports to the EU and for that purpose the gravity equation was regressed on exports from each candidate country to each EU10 country, resulting in a sample with 11x14 trade flows. The regression results reveal that only the GDPs, of both exporting and importing countries, plus distance (DISTij) between markets are significant in explaining the volume of exports from the CCs into the EU markets, and that all the estimates are significant at 1% level. This is a very interesting result as it suggests that the simplest form of the gravity equation - with only market size and distance - is able to explain trade flows from the CCs countries to the EU. Comparing the estimations with those on regressions I and II the conclusions are:
10
Belgium and Luxembourg were taken together resulting in a sample with 14x11=154 trade flows. 12
•
The GDPs of both the exporting and importing countries have a stronger positive effect on exports from CCs to EU markets than that observed on their bilateral trade or on exports from the EU to the CCs markets;
•
We also observe that distance has a stronger negative impact on these trade flows compared to all other trade flows studied in previous regressions. Somehow it shows up that, in average, producers in CCs face greater transaction costs in reaching western markets for their products than the opposite. Because the variable distance catches all sorts of transaction costs from transport, to information, non-tariff barriers, etc, it is difficult to argue which particular factor contributes the most for this result.
NOTE: The results observed are the evidence that trade relations between the EU and CCs exhibit some differences when compared to intra-EU trade but also when a comparison is made between the two groups of countries – namely in the way they relate to the other group market. The next step in this research is the calculation of potential EU-CCs bilateral trade taking the coefficients of intra-EU trade.
13
References: Baldwin, R.E. (1994). Towards an Integrated Europe. CEPR, London. Bayoumi, T. and Eichengreen, B. (1995). Is regionalism simply a diversion? Evidence from the evolution of the EC and the EFTA, NBER Working Paper: No. 5283. Cambridge. Bergstrand, J.H., 1985. The gravity equation in international trade: some microeconomic foundations and empirical evidence. The Review of Economics and Statistics 67, 474-481. Brenton, P., Groz, D. (1997). Trade Reorientation and Recovery in Transition Economies, Oxford Review of Economic Policy, vol.13, 2, 65-76. Deardoff, A.V. (1995). Determinants of Bilateral Trade: Does Gravity Work in a Neoclassical World? NBER Working Paper: No.5377. Cambridge. Evenett, S.J., Keller, W. (2000). On Theories Explaining the Success of the Gravity Equation. Gros, D., Gonciarz, A., (1996). A note on the trade potential of Central and Eastern Europe. European Journal of Political Economy 12, 709-721. Hamilton, C.B., Winters, L.A., (1992). Opening up international trade with Eastern Europe. Economic Policy 14, 77-116. Hellvin, L. and Nilsson, L. (2000). Trade Flows Between Trading Blocs: The Case of the EU’s Trade with Asia and Nafta. Ministry for Foreign Affairs. Stockholm, Sweden. Nilsson, Lars (2000), Trade integration and the EU economic membership criteria, European Journal of Political Economy, Vol.16, 807-827. Piazolo, D. (2000). Eastern Europe between Transition and Accession: An Analysis of of Reform Requirements Kiel Working Paper: No. 991. Kiel Institute of World Economics. Shelburne, R.C. (2000). Intra-Industry Trade, the Gravity Model and Similarity in Endowments and Country Size, Papers and Proceedings of the International Trade and Finance Association, San Diego State University, 2000. Presented at the Ecole Superieure de Commerce de Montpellier, France, June 2000. Wang, Z. K., Winters, A. (1992), The Trading Potential of Eastern Integration 7(2), 113-136.
Europe. Journal of Economic
14
