The Gravity model and prediction of trade flows between Baltic States and their trade partners

Transcription

1 1 The Gravity model and prediction of trade flows between Baltic States and their trade partners Ieva Pravorne Natalija Skorohoda Arvis Strods Olegs Tkachevs The University of Latvia and Eurofaculty EuroFaculty Working Papers in Economics No. 17 August 2003

2 2 The Gravity model and prediction of trade flows between Baltic States and their trade partners 1 Ieva Pravorne 2, Natalija Skorohoda 3, Arvis Strods 4, Olegs Tkachevs 5 The University of Latvia, EuroFaculty August 2003 Abstract The paper analyzes Baltic Sea countries international trade flows using Gravity approach and then explores estimated exports and imports Gravity equations for Baltic Sea region to calculate predicted trade flows for Baltic States. The major issue of the paper is to find out with which countries Latvia, Lithuania and Estonia are over or under traded relative to the predicted trade flows by the Gravity models and therefore where the trade potential for Baltic States can be found. There are four major findings. First, Baltic Sea States international trade adjusts to the gravitational forces of income, population and distance. Regional effects are also strong in the Baltic Sea region. Second, Baltic States are still overtraded in exports with CIS and Baltic Sea region countries indicating that historical ties still matter for Latvia, Lithuania and Estonia. Third, Baltic States potential for exports is considerable in trade with non-baltic candidate countries for EU membership and more distant developed countries. Fourth, Baltic States are still under traded and therefore have potential for their exports to many non-baltic Sea members of the EU. Latvia and Estonia are under traded in exports with Russia. Based on these findings we argue that Baltic countries do not fully use their potential in international trade especially those preferences given by different preferential trade agreements with EU countries and EU candidates as well as big Russian market s possibilities. 1 The authors would like to thank Morten Hansen for supervising the research project. The authors have also benefitted greatly from comments and conversations with Mihails Hazans and Vita King. The paper was presented at the EuroFaculty semester conference held in Ligatne, Latvia, May 16-17, 2003 and at the International conference Enlargement of the European Union in the Baltic Sea Region: Social and Economic Challenges and Opportunities held at the University of Latvia, June 20, This paper is part of EuroFaculty Student Research Assistant programme. Any opinions expressed are those of the authors and not those of the EuroFaculty. 2 ieva_pravorne@yahoo.com 3 Natalja@medship.lv 4 arvisstrods@inbox.lv 5 otkachev@one.lv

3 3 1. Introduction The Baltic States regained their independence in the early nineties. Former Soviet Union republics: Latvia, Lithuania and Estonia before the collapse of the socialist block in Central and Eastern Europe actively participated in the COMECON (Community of Mutual Economic Assistance) trade relations and Soviet intratrade. The end of communism and disintegration of the Soviet Union opened western markets for the Baltic States. Similarly they themselves became open to western imports. The major steps on the way to international trade liberalization were signed by the Baltic States Europe Agreements in They meant opening of EU markets for Baltic States industrial products and implied free movement of services and capital in not so far future. Moreover the Baltic States were invited by EU to start accession negotiations (Estonia in 1998; Latvia and Lithuania in 1999) which were recently successfully completed. Since the beginning of the process of economic transition Baltic States international trade patterns have changed dramatically. Now the main trading partner of the Baltic States is the European Union. Taking into consideration the fact that before the regaining of independence Baltic States nearly had no relations outside the socialist block, the present export-import structure shows that the Baltic States have managed to reorient their economies. During the nineties Baltic States signed also a number of Free Trade Agreements with non-eu European countries to promote trade relations with them. Changes in Baltic States international trade patterns require very careful analysis of their future prospects. There exist several quantitative methods to estimate trade potential of a country. One of them is based on the commonly used in empirical analysis Gravity models. The main advantage of using Gravity models for this purpose is accessibility and reliability of data that is very important especially for former socialist countries experiencing now transition process in data collection methodology. Taking into account recent changes in Baltic States international trade patterns the paper aims to estimate trade potential of Baltic States. To reach this goal the following methodology is used in the analysis. Firstly the Gravity model is estimated for a group of Baltic Sea States using data for year It is assumed in this paper that Baltic Sea region as a whole is a good reference group for Latvia, Lithuania and Estonia as they also belong to this region and have intensive trade relations with other Baltic Sea countries. Historical and cultural links to this region also matter. Estimating Gravity model Russia is excluded to make reference group more homogeneous and appropriate for further predictions. Thus hypothetical coefficients of the exports and imports

4 4 Gravity equations are applied to Baltic States data for year 2001 to predict their trade flows in this year. Comparing actual and predicted trade values in year 2001 conclusions are made about which countries Latvia, Lithuania and Estonia are over or under traded with. It is argued that if Latvia, Lithuania or Estonia is under traded with some country (region) then it has trade potential in this country (region). The justification of using coefficients of Gravity models estimated for Baltic Sea region to predict trade flows of the Baltic States is similar to Byers et al (2000) who use the coefficients estimated for Scandinavian countries in the prediction of Baltic States trade. Accordingly to the goal of this paper it is organized as follows. In the second part Baltic States trade patterns in a historical perspective are analyzed. In the third part theoretical background of the Gravity model is discussed as well as a short overview of literature on using a gravity approach for modeling international trade flows is presented. The fourth part is devoted to the estimation of the export and import Gravity equations using Baltic Sea States data with a purpose described earlier. The fifth part presents the main results of prediction of Baltic States trade flows. And finally in the sixth part empirical results of the paper are summarized.

5 5 2. Baltic States Trade Patterns: Before and After Russian Crisis. At the very beginning of the nineties Latvia s international trade flows were still intra-soviet oriented, the same was true for Estonia and Lithuania either. Foreign trade with EU countries was of minor importance (see Tables 1 and 2). Still soon after regaining independence this trade structure changed entirely: the EU-15 countries became the main trading partners of Latvia, the same was true for other Baltic States although the development of bilateral trade structures reveals that the Baltic States were not a homogenous group. The relative changes in the structure of Estonia s trade were much more distinct compared to Latvia and Lithuania: very soon after regaining independence trade with EU countries covered more than half of Estonian exports and imports. Also Latvia and Lithuania made use of the free trade agreements with the EU (the far reaching Europe agreements were signed in 1995), but on a lower level. Not surprisingly, it can be observed that with no exception EU markets are more important for Baltic exports than for imports because Baltic goods face a tougher competition on world markets than on EU markets. Moreover, Baltic EU exports and imports are far from being evenly distributed: trade intensity with the more developed Northern EU countries is significantly higher, especially with Baltic rim countries, than with poorer European South. Although the EU share in Baltic States international trade became larger, significant trade relationships with former Soviet Union countries continued- especially with Russia. The still important trade with CIS indicates the intra-soviet division of labour and one can say that the period of Soviet occupation had sustainable effects on Baltic trade patterns. Despite the common feature of Baltic integration patterns, the single Baltic States do not share the same main trading partners: Latvia s exports concentrate on German and Russian markets, the same is true for Latvia s imports. Estonia s main trading partners are Finland and Sweden, in addition Russian imports are of major importance. Trade with Germany is most important for Lithuania with respect to both exports and imports, imports from Russia count for 20 p.c. If total trade is analysed, these partner countries keep their status as main trading partners, thus underlining the importance of the Baltic Sea region for the Baltic countries.

6 6 Table 1. Exports flows of Latvia, Estonia, Lithuania ( ) Exports Estonia Latvia Lithuania EU EU Baltic States Russia CIS Baltic Sea Region Europe Source: Central Statistical Bureau of Latvia, Foreign Trade of Latvia Quarterly Bulletin, Riga, various issues; Statistical Office of Estonia (2003), Foreign Trade: Yearbook 2002, Tallinn; Statistics Lithuania, Foreign Trade, Vilnius, various issues Table 2. Imports flows of Latvia, Estonia, Lithuania ( ) Imports Estonia Latvia Lithuania EU EU Baltic States Russia CIS Baltic Sea Region Europe Source: see Table 1

7 7 3. The Gravity Model as a Tool to Predict Trade Flows 3.1. The Theoretical Background of the Gravity Model The gravity model, used for modeling bilateral trade flows, is analogous to Newton s law that relates the gravity between two objects to their masses and the distance between them. Accordingly to the gravity approach, bilateral trade between two regions (countries) is directly related to their incomes and inversely related to the distance between them. The gravity equation ordinarily is specified as: M ij = α Y k β i Y γ j N ξ j N ε j d µ ij U ij where M ij is the trade flow between country i and j, α k is a constant, β, γ, ξ, ε, µ are coefficients, weighted geometric averages, Y i and Y j are incomes per capita in country i and j, respectively, N i and N j are population in i and j, and d ij is the distance between countries i and j. The U ij is a lognormal distributed error term with E(ln U ij )=0 Frequently dummy variables are also included in the model to take into account preferential trade factors between i and j. Although the gravity model has been widely recognized for its empirical success at predicting bilateral trade, initially it lacked a strong theoretical background. For example, it was claimed that the Heckscher-Ohlin (H-O) model of international trade was not capable to provide such a background, and perhaps was even theoretically inconsistent with the gravity equation. Tinbergen (1962) and Pöyhönen (1963) did the first econometric studies of trade flows based on the gravity equation, for which they gave only intuitive justification. Now it is no longer true that the gravity equation lacks theoretical foundation, since several of the same authors who noted an absence of the theoretical basis went on to provide one. Linneman (1966), Leamer and Stern (1970), Savage and Deutsch (1960) were the first ones who tried to provide theoretical explanation of the gravity equation. These contributions were followed by several more formal attempts to derive the gravity equation from models that assumed product differentiation. Anderson (1979) was the first to do so. He pointed out that the gravity equation can be derived from the properties of expenditure systems. In his interpretation it is an alternative method of doing cross-section budget studies, and one with potentially important efficiency properties. However Anderson s primary concern was to examine the econometric properties of the resulting equations, rather than to extract easily interpretable theoretical implications. Since Anderson it has been shown that the gravity equation can

8 8 be derived from the traditional and the new theories of international trade including Ricardian models, Hecksher-Ohlin (H-O) models, and increasing returns to scale (IRS) models. When consumers have both identical homothetic preferences and access to the same goods prices, a sufficient condition for obtaining a gravity equation is perfect product specialization, in the sense that each commodity is produced in only one country. Grossman (1998) argues that the assumption of the perfect product specialization is the key assumption that generates the empirical success of the equation as an exporting country increases the supply of its products, the importing country will increase its consumption proportionally raising the volume of trade between them. Evenett and Keller (1998) summarized three types of trade models that differ in the way product specialization is obtained in equilibrium: a) Technology differences across countries in the Ricardian model; b) Factor proportions outside the vector space-of-diversification in the H-O model; c) Increasing returns at the firm level in the IRS model. In addition, they demonstrated that, as far as their predictions for bilateral trade volumes are concerned, each of these perfect specializations models is a limiting case of a model with imperfect specialization. So they supposed that perfect product specialization is not the key assumption of the success of the gravity model. That is why Evenett and Keller (1995) derived the specific form of the gravity equation of trade for four models: 1)Multi-cone H-O: factor endowment differences lead to perfect product specialization; 2)Pure IRS: IRS leads to perfect product specialization; 3)Generalized H-O: with one good produced using IRS technology and perfectly specialized, the other good being produced with constant returns to scale (CRS) technology and not specialized 4)Uni-cone H-O: where both goods are produced using CRS technology and are not specialized. Evenett and Keller (1998) compared the accuracy of the perfect specialization versions of the Hecksher-Ohlin model and the Incresing Returns-based model, both of which predict the gravity equation. The empirical strategy of Evenett and Keller (1998) exploited two factors to identify which models might explain trade flows in a given sample. First, in the multi-cone H-O model large factor proportion differences are required to induce product specialization, whereas product specialization in the pure IRS model occurs for arbitrary differences in factor proportions. Secondly, there is no

9 9 intra-industry trade in the multi-cone H-O model, whereas there is intra-industry trade, and possibly only intra-industry trade, in the pure IRS model. Evenett and Keller (1998) concluded with three main findings: 1) Little production is perfectly specialized due to factor proportions differences, making the perfect specialization version of the H-O model an unlikely candidate to explain the empirical success of the gravity equation. Overall, there is little evidence to support the notion that the multi-cone H-O model plays a role in explaining why the gravity model works well empirically. As far as the success of the gravity equation has something to do with the H-O model, it appears to be related to increased specialization within the cone of diversification, not outside of it. 2) Increasing returns are important causes for perfect product specialization and the gravity equation, especially among industrialized countries. 3) Models of imperfect specialization better explain the variation of bilateral trade flows than perfect specialization models. Accordingly to the Evenett and Keller research the model which incorporates imperfect specialization performs better to explain success of the gravity model. Factor proportion differences are important determinants of trade flows within the context of imperfect specialization models only, whereas there is evidence that increasing returns is a cause of product specialization along the lines of both trade models with imperfect as well as with perfect specialization of production. Additionally to the traditional and new trade theories also economic geography and its models support the explanation of trade flows and trade pattern thus partly explaining the empirical success of the gravity equation. Role of distance in determining the pattern and direction of trade is essentially a topic of economic geography models. Despite of many discussions on the theoretical foundation for the gravity equation it is widely used in the empirical analysis of the bilateral trade flows because of its ability not simply explain (what actually trade theory does) why countries trade in different products, but show why some countries trade more extensively than other countries. Deardorff (1995) suspected that just about any plausible model of trade would yield something very like the gravity equation, whose empirical success is not evidence of anything, but just a fact of life. Accordingly to Deardorff (1998) all that the gravity equation says, after all, aside from its particular form, is that bilateral trade should be positively related to the two countries incomes and negatively related to the

10 10 distance between them. Transport costs would surely yield the latter in just about any sensible model, and the dependence on income would also be hard to avoid. (p.8)

11 The Gravity Approach for Modeling International Trade Patterns of Economies in Transition: Review of Literature Since the beginning of 90-s gravity models were widely used for estimating trade relations between Eastern and Western European countries, especially for predicting and modeling trade relations between European Union, Central and Eastern European countries and Baltic countries. Among the most important studies are those of Wang and Winters (1991; 1994), Baldwin (1993, 1994, 1997), van Beers and Biesen (1996), Iversen (1998), Cornett and Iversen (1998). In the beginning of 90-s Wang and Wniter (1991) used gravity model in order to estimate foreign trade flows between European Free Trade association and Central and Eastern European countries. In their study they used data for 76 non- CEE countries for the time period from The analysis results indicate that the potential trade turnover between EFTA and CEE countries is larger than the actual one. The gravity model constructed by Wang and Winter was updated by R.E. Baldwin (1993,1994). According to the Baldwin model bilateral trade flows from the countries of origin to their trading partners depend on GDP level of these countries, their population, the distance between exporting and importing country and a set of dummies. Using this model the trade potential was estimated for Albania, Croatia, the Czech Republic, Bulgaria, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, Slovakia, Russia, Ukraine, Belarus etc. The model analysis was based on two different scenarios. The medium-run scenario assumed that Central and Eastern European countries (including Baltic States) will integrate into world trade. According to the longrun scenario CEE and Baltic countries per capita income level will achieve the poorer Western European countries income level. The model analysis based on above mentioned scenarios showed that CEE and Baltic countries exports have potential to increase in the direction to the EFTA countries. It was shown that CEE markets could account for a 17% share of EFTA countries exports to all Europe (Baldwin, 1994) Recent study of Van Bergeijk and Oldersma (1990) provides the results of the estimation of the gravity model for 49 countries (including 6 CEE countries), which is based on the data for population, income level, distances and dummy variables for East- West and East-East trade. The estimates of dummies show that all bilateral trade flows in 1985 between East and West were about 1/8 as large as the gravity model would

12 12 predict based only on income, population and distance. The model analysis results show that East-East trade s actual turnover was by 50% lower than potential. Among the first, who analyzed Baltic Sea region s trade patterns using gravity approach were Byers (2000) and Cornett and Iversen (1998). Byers used the hypothetical coefficients from recent trade data of the Scandinavian countries in order to predict future trade flows of the Baltic countries. Cornett and Iversen (1998) tried to estimate future trade flows in the Baltic Sea region taking into consideration bilateral trade relations between the European Union countries and Central and Eastern European accession to the EU countries. They control for different phases of integration in order to differentiate between various forms of trade barriers typical for the different forms of the bilateral trade links. Those studies successfully explain the present trade relations in Baltic Sea region using the data for income, population and proximity.

13 13 4. Estimation of the Gravity Model for Baltic Sea States International Trade The Gravity model introduced in the previous section is estimated here to calculate further Baltic States trade potential. As was described earlier we estimate hypothetical coefficients of exports and imports gravity equations using data of the Baltic Sea countries 6 for year 2000 in order to predict trade flows of the Baltic countries in Here we argue that Baltic Sea region is more or less homogeneous reference group with respect to foreign trade patterns as Latvia, Lithuania and Estonia constantly converge to the advanced Baltic Sea countries. The analysis is based on the gravity model with the following specification: X ij =b 0 +b 1 GDP i +b 2 POP i + b 3 GDP j +b 4 POP j + b 5 DIST ij +b 5+k DUM k +u where all the variables (excluding dummies) are expressed in natural logarithms and dependent variable X ij represents the exports from the country of origin i to the trade partner j in exports equation or imports from country j to i in imports equation. Independent variables include Gross Domestic Product per capita and Population in countries i and j respectively and distance between the capitals of the two countries. Dummies are used to control for different kinds of virtual distances and proximities (for the full description of the variables used see table 3). Table 3 Description of the variables used in the gravity models GDP i POP i GDP j POP j Variable Description of the variable GDP per capita of Baltic Sea country (excluding Russia) Population of Baltic Sea country GDP per capita of trading partner Population of trading partner DIST ij Distance between the capitals of Baltic Sea state and its trading partner DUM 1 (CB) =1, if the trading partner has a common boarder with corresponding Baltic Sea state; =0, if not DUM 2 (BALTSEA) =1, if the trading partner is the Baltic Sea state (excluding Russia); =0, if not DUM 3 (EU) =1, if the trading partner belongs to the EU, but is not included in BALTSEA (i.e. is non-baltic Sea member of the EU); =0, if not DUM 4 (CIS) =1, if the trading partner belongs to the CIS; =0, if not DUM 5 (CAND) =1, if the trading partner is the non-baltic candidate country for the EU membership (Malta, Cyprus, Czech Republic, Slovakia, Slovenia, Poland, Hungary, Romania and Bulgaria); =0, if not 6 Russia is excluded to make reference group more homogeneous

14 14 Dummies are chosen to capture the impact of different preferential trade agreements between countries that substantially reduce costs connected with distance or to take into account common past of some countries belonging to the Baltic Sea region and some third countries. It is usually argued that historical ties can serve as an explanatory variable for current trade flows 7. And of course dummy for common boarder is a contiguity dummy that must be included as trade between neighbors is usually less impeded by transaction costs. The data used when modeling exports and imports gravity equations are obtained from the following sources: - Exports and imports data of the Baltic Sea countries for year 2000 obtained from Direction of Trade Statistics: Yearbook 2002 issued by IMF. Exports are valued at free-on-board (FOB) prices and imports at cost-insurance-freight (CIF) prices. Both exports and imports are expressed in millions of US dollars and are collected according to the concept of special trade. - GDP per capita and Population data of the Baltic Sea countries and their trading partners are taken from the Worldbank online database - A matrix of distances as a crow flies between the capitals of the countries is obtained from the Bali Indonesia Travel Portal The relationship between trade flows on the one hand and the explanatory variables on the other is estimated by the Weighted Least Squares (WLS) method to get over the problem of heteroscedasticity usually presented in the gravity models estimated by OLS. Only those trading partners were considered whose exports and imports values exceeded 1 million USD and for whom other necessary data were available. Looking at the regression results for exports (see Table 4) it can be seen that all the coefficients have the expected signs and almost all of them are highly significant (except for the coefficient of the dummy for non-baltic candidate countries for the EU membership - DUM 5 (CAND)). The coefficients of the dummies for non-baltic Sea members of the EU - DUM 3 (EU) - and for CIS countries - DUM 4 (CIS) - are significant at 5%, but the coefficients of the remained dummies are significant even at 1%. Export flows from Baltic Sea countries are positively correlated with GDP per capita and population in both country of origin and country of destination, while distance between them is an export reducing factor. So as it is predicted by theory bilateral trade flows between Baltic Sea States and their trading partners comply with gravitational force. If a 7 For details see Laaser, C.-F., K. Schrader (2002)

15 15 trade partner belongs to the Baltic Sea region, non-baltic Sea part of the EU, CIS or has a common boarder with corresponding Baltic Sea country of origin then it has positive effect on the Baltic Sea state s exports to this country thus indicating close ties between these countries. All signs are as expected and all the coefficients are statistically significant at 1% level in the regression for imports (see Table 5). Again, GDP per capita and population show positive coefficients. But gravitational forces of income and population are less important for imports than for exports. Another difference between the two models is that the coefficients of the dummies are slightly bigger in the imports equation than in the exports one. Furthermore the coefficient of the dummy for non- Baltic candidate countries is statistically significant in the imports model. So being a part of one of the mentioned regional arrangements is more significant for imports than for exports. Moreover the absolute value of the coefficient for the distance, not virtual one reflected by dummies, but real distance between Baltic Sea State and its trading partner is lower in the imports gravity equation indicating that real distance on the other hand has less influence on the Baltic Sea countries import flows than on their export flows. Therefore the Baltic Sea States import flows are dominated more heavily by regional trade than their export flows. It can be also shown that both models are statistically significant at very high significance level (see Tables 4 and 5) and adjusted R squares (0,845 for exports equation; 0,791 for imports equation) are comparable to other gravity regressions and show that the independent variables explain on average 80% of the variation in the dependent variables in both equations.

16 16 Table 4 Regression results for exports gravity equation Variable # B t-value GDP i 0,992 27,089 *** POP i 0,856 31,108 *** GDP j 0,902 34,301 *** POP j 0,767 38,407 *** DIST ij -0,877-15,266 *** DUM 1 (CB) 0,889 5,280 *** DUM 2 (BALTSEA) 0,626 3,769 *** DUM 3 (EU) 0,293 2,052 ** DUM 4 (CIS) 0,330 2,504 ** DUM 5 (CAND) 0,107 0,761 Constant -9,825-15,600 *** Adjusted R Square = 0,845 F= 511, 337 signif F= 0,000 Number of observations= 939 Notes: *** denote statistically significant coefficients at 1% level ** denote statistically significant coefficients at 5% level # Dependent variable: logarithm of exports from i to j; Independent variables (except for dummies) are expressed in natural logarithms Source: authors calculations Table 5 Regression results for imports gravity equation Variable # B t-value GDP i 0,661 15,569 *** POP i 0,860 26,116 *** GDP j 0,873 26,766 *** POP j 0,812 30,646 *** DIST ij -0,478-6,908 *** DUM 1 (CB) 1,362 6,869 *** DUM 2 (BALTSEA) 1,515 7,629 *** DUM 3 (EU) 0,908 5,383 *** DUM 4 (CIS) 0,937 5,389 *** DUM 5 (CAND) 0,693 4,200 *** Constant -10,205-13,346 *** Adjusted R Square = 0,791 F= 319, 281 signif F= 0,000 Number of observations= 932 Notes: see Table 4 # Dependent variable: logarithm of imports from j to i; Source: authors calculations

17 17 5. Prediction of Baltic States Trade Flows Now the abovementioned models are explored to predict exports and imports of Latvia, Estonia and Lithuania to other countries in 2001 and calculate the ratios between actual and predicted trade flows. Therefore this section is devoted to the analysis of the ratios between actual and predicted trade flows of Baltic States to different regions. If the ratio between actual and predicted values for exports (imports) exceeds 1 then the correspondent country is overtraded in exports (imports) with the corresponding third country (region), and if it is less than 1, then it is under traded. Let turn to the results presented in the tables 4&7. Baltic States actual and predicted exports. As to the Latvia s exports, considering trade with all countries, Latvia s actual exports are less than the predicted, i.e. on the whole Latvia is under traded. Turning to the trade with different regions Latvia is clearly overtraded with CIS and candidate countries for the EU membership, but still has strong trade potential in non-baltic candidate countries (meaning that it is mainly overtraded with Baltic neighbors) as well as in EU countries and Baltic Sea states (excl. Lithuania and Estonia). It is also under traded with so called other countries, i.e. those that are not included in one of the regional arrangements which Latvia has historical links with. Detailed calculations (given in Appendix 1) show that being overtraded with CIS on the whole Latvia is nevertheless under traded with Russia and could under certain circumstances use this large market more intensively. As to Estonia s exports, due to high degree of Estonia s openness to international trade it is in total 1.46 times overtraded. Almost to each of the observed regions in general the predicted value of Estonia s exports is lower than the actual one. Thus Estonia is overtraded with EU countries, CIS countries, the Baltic Sea region and the candidate countries. In spite of being overtraded with EU countries almost 1.5 times Estonia s predicted trade flows for non-baltic Sea members of the EU exceed actual ones. As to non-baltic candidates the picture looks the same. Again the situation with trade with Russian Federation is the same as for Latvia. Estonia is under traded with Russia and still has there potential for its exports. Lithuania s total actual export is 21% higher than the predicted. Furthermore the results suggest that Lithuania is overtraded with EU (both including and excluding Baltic Sea states) and CIS countries, as well as with Baltic Sea countries and EU

18 18 candidates. However Lithuania is seriously under traded with non-baltic candidate countries and so called other countries and therefore has there potential for its exports. It should be mentioned here nevertheless that Lithuania is overtraded with non-baltic Sea members of the EU mainly due to intensive trade with Great Britain. Actual export flows to UK are 3.2 times higher than the predicted. Conclusions about trade potential with non-baltic Sea members of the EU change if UK is excluded. Baltic States actual and predicted imports. As regards imports Latvia s trade with all countries broadly matches the predicted values, while estimates suggest that Latvia is under traded with EU countries both including and excluding Baltic Sea States and other countries. Estonia is not only overtraded on the whole, but it is also overtraded with all the mentioned regions. Lithuania s value of actual imports does not differ from the predicted one. But, like its northern neighbor, it is still under traded in imports with EU countries especially non- Baltic Sea members of the EU. The value of Lithuania s actual imports from candidate countries and the Baltic Sea region excluding Russia is seriously behind the predicted. Like Latvia Lithuania is under traded in imports with other countries. Table 6 Ratios between actual and predicted Baltic States export flows 8 Trading partners Latvia Estonia Lithuania EU countries 0,91 1,44 1,13 - non-baltic Sea members of the EU 0,92 0,90 1,29 CIS countries 1,14 1,23 2,19 Baltic Sea countries 1,01 1,73 1,30 - excl. Russia 1,03 1,80 1,24 - excl. Baltic States 0,81 1,53 0,97 Candidate countries for EU membership (incl. 1,68 3,63 1,66 Baltic States) - non-baltic candidate countries for EU 0,51 0,84 0,59 membership Other countries 0,45 0,81 0,57 All countries 0,91 1,46 1,21 Source: authors calculations 8 Detailed calculations are given in Appendix 1, Appendix 2 and Appendix 3

19 Table 7 Ratios between actual and predicted Baltic States import flows Trading partners Latvia Estonia Lithuania EU countries 0,97 1,34 0,98 - non-baltic Sea members of the EU 0,69 1,12 0,86 CIS countries 1,49 2,08 3,21 Baltic Sea countries 1,32 1,48 1,01 - excl. Russia 1,37 1,45 0,66 - excl. Baltic States 1,09 1,39 0,99 Candidate countries for EU membership (incl. 2,21 2,07 0,39 Baltic States) - non-baltic candidate countries for EU 1,12 1,31 0,31 membership Other countries 0,43 2,23 0,89 All countries 1,06 1,57 1,00 Source: authors calculation 19

20 20 6. Conclusions The major issue of this paper was to find out with which countries Latvia, Lithuania and Estonia are over or under traded relative to the predicted trade flows by the Gravity models and therefore where the trade potential for Baltic States can be found. First of all exports and imports gravity equations were run using Baltic Sea States data for year 2000 and it was shown that Baltic Sea States international trade adjusts to the normal gravitational forces of income and large markets and is repelling effect of distance to trading partners. Thus Baltic Sea countries tend to have trade relations with rich and highly populated countries, but the larger is the distance between Baltic Sea States and their trading partners the smaller are the trade flows between them. Specific forces with regional effects are at work in Baltic Sea region. Belonging a trade partner to CIS, Baltic Sea region, non-baltic Sea part of the EU or a group of candidate countries for EU membership has positive effect on Baltic Sea countries exports and imports. Applying the hypothetical coefficients of the exports and imports gravity equations to Baltic States data for year 2001 and calculating the predicted trade flows of Latvia, Lithuania and Estonia to other countries and regions the following conclusions were maid: 1) Latvia is the only Baltic country that is under traded in exports, 2) Estonia s ratios between actual and predicted exports and imports values are the highest between Baltic countries, 3) Baltic States are under traded in exports with non-baltic Sea members of the EU (it is true for Lithuania only if exports to the UK are not taken into account), 4) Baltic States are seriously under traded in exports with non-baltic candidate countries for EU membership, 5) All Baltic States are overtraded in exports with CIS countries, but in spite of it Latvia and Estonia are under traded in exports with Russian Federation, 6) All Baltic countries have trade potential in so called other countries, countries that are not included in one of the preferential regional arrangements, 7) Baltic States are totally overtraded in exports with each other, indicating very close historical and cultural traditions, 8) All Baltic countries are overtraded in imports. Overall Baltic States have considerable potential for exports to non-baltic candidate countries for EU membership and the majority of non-baltic Sea members of

21 21 the EU. Baltic States have good prospects to develop trade relations with more distant developed countries. Russia s share in Estonia s and Latvia s exports has to be larger in future as well, as the value of their trade with Russia is relatively small. The privilege of geography, common past and the existence of cultural traditions has to be used. Baltic States trade relations with each other are remarkable. Regarding imports one important factor can explain Baltic States overtrade figures in them. And this factor is small Baltic States natural resource endowments. Latvia, Lithuania and Estonia are not endowed with mineral resources, oil and natural gas and therefore have to import these commodities.

22 22 References [1] Anderson, J. (1979), A Theoretical Foundation for the gravity equation, American Economic Review, 69(1): [2] Baldwin, R. (1994), Towards an Integrated Europe, London: Centre for Economic Policy Research [3] Byers, D.A., T.B. Iscan, B. Lesser (2000), New Boarders and Trade Flows: A Gravity Model Analysis of the Baltic States, Open Economies Review, 11(1): [4] Carrillo, C., and C.A. Li (2002), Trade Blocks and the Gravity Model: Evidence from Latin American Countries, University of Essex Working Paper, Colchester [5] Central Statistical Bureau of Latvia, Foreign Trade of Latvia Quarterly Bulletin, Riga, various issues [6] Cornett, A.P., S.P. Iversen (1998), The Baltic States in an European and Baltic Perspective: Trade and Transition in the Baltic Rim, Paper presented at the 5-th Nordic Baltic Conference in Regional Science 1-4 October 1998 in Parnu, Estonia (available via internet: [7] Deardorff, A.V. (1995), Determinants of Bilateral Trade: Does Gravity Work in a Neoclassical World?, NBER Working Paper # 5377, Cambridge MA [8] Evenett, S.J., and W. Keller (1998), On the theories explaining the success of the gravity equation, NBER Working Paper # 6529, Cambridge, MA [9] Head, K. (2003), Gravity for Beginners, University of British Columbia Working Paper, Vancouver [10] International Monetary Fund (2002a), Direction of Trade Statistics: Yearbook 2002, Washington [11] International Monetary Fund (2002b), International Financial Statistics: Yearbook 2002, Washington [12] Kalbasi, H. (1999), The Gravity Model and Global Trade Flows, University of Isfahan Working Paper, Isfahan [13] Laaser, C.-F., K. Schrader (2002), European Integration and Changing Trade Patterns: The Case of the Baltic States, Kiel Working Paper No. 1088, Kiel [14] Leamer, Edward E. and Robert M. Stern (1970), Quantitative International Economics, Boston: Allyn and Bacon. [15] Linnemann, Hans (1966), An Econometrsic Study of International Trade Flows, Amsterdam: North-Holland

23 23 [16] Paas, T. (2001) Gravity approach for modeling trade flows between Estonia and the main trading partners, University of Tartu Working Paper, Tartu, Estonia [17] Pöyhönen, Pentti (1963), A Tentative Model for the Volume of Trade Between Countries, Weltwirtschaftliches Archiv 90, pp [18] Savage, I. R. And K. W. Deutsch (1960), A Statistical Model of the Gross Analysis of Transactions Flows, Econometrica 28, (July), pp [19] Statistical Office of Estonia (2002), Foreign Trade: Estonia, Lithuania, Latvia 2001, Tallinn [20] Statistical Office of Estonia (2003), Foreign Trade: Yearbook 2002, Tallinn [21] Statistics Lithuania, Foreign Trade, Vilnius, various issues [22] Tinbergen, J. (1962), Shaping the World Economy: Suggestions for an International Economic Policy, New York: The Twentief Century Fund. [23] [24]

24 24 Appendix 1 Actual and predicted Latvia s exports and imports in 2001, mill.$ Country Exports Exports Ratio Imports Imports Ratio mill $ predicted actual/predicted mill $ predicted actual/predicted Austria 12,17 42,22 0,29 40,30 49,56 0,81 Belgium 25,31 32,47 0,78 64,18 46,26 1,39 Denmark 115,77 78,01 1,48 130,46 99,70 1,31 Finland 45,69 112,30 0,41 280,91 108,93 2,58 France 38,64 125,44 0,31 84,85 192,39 0,44 Germany 333,84 428,38 0,78 596,51 664,10 0,90 Greece 0,80 14,94 0,05 6,65 23,46 0,28 Ireland 22,94 16,11 1,42 19,79 22,87 0,87 Italy 36,65 99,20 0,37 144,13 158,16 0,91 Luxembourg 0,51 6,49 0,08 1,76 7,27 0,24 Netherlands 73,72 60,37 1,22 113,80 78,95 1,44 Portugal 0,96 9,90 0,10 4,84 18,24 0,27 Spain 13,77 42,29 0,33 41,53 77,58 0,54 Sweden 191,50 140,66 1,36 227,78 152,58 1,49 United Kingdom 313,10 136,24 2,30 84,58 207,23 0,41 Norway 28,10 71,18 0,39 44,08 95,20 0,46 Iceland 7,15 1,29 5,54 3,59 0,98 3,66 Switzerland 6,70 29,55 0,23 61,01 21,37 2,85 Czech Republic 15,00 12,48 1,20 49,74 14,42 3,45 Estonia 114,79 24,28 4,73 221,42 33,73 6,56 Hungary 3,39 9,01 0,38 35,78 11,73 3,05 Lithuania 162,33 46,25 3,51 296,72 66,00 4,50 Poland 38,36 95,07 0,40 175,42 220,98 0,79 Slovakia 9,52 5,07 1,88 25,59 5,95 4,30 Slovenia 1,79 4,35 0,41 8,04 5,22 1,54 Cyprus 1,02 1,56 0,66 0,35 2,26 0,15 Bulgaria 0,19 2,25 0,09 8,42 3,20 2,63 Romania 0,22 6,05 0,04 0,82 8,59 0,10 Croatia 0,10 3,70 0,03 1,62 2,61 0,62 Yugoslavia 0,36 1,91 0,19 1,50 1,47 1,02 Belarus 35,79 20,92 1,71 135,67 28,82 4,71 Moldova 0,95 0,58 1,64 1,92 0,75 2,55 Russian Federation 116,93 142,50 0,82 321,95 292,93 1,10 Ukraine 33,35 10,27 3,25 49,53 12,81 3,87 Armenia 0,00 0,46 0,00 0,00 0,74 0,00 Azerbaijan 0,63 0,99 0,64 0,46 1,70 0,27 Georgia 4,03 0,62 6,51 0,50 1,04 0,48 Kazakhstan 7,49 1,94 3,85 2,88 4,19 0,69 Kyrgyzstan 0,70 0,20 3,43 0,17 0,43 0,38 Tajikistan 0,01 0,14 0,07 0,59 0,31 1,91

25 25 Turkmenistan 0,52 0,80 0,65 0,73 1,53 0,48 Uzbekistan 5,95 1,04 5,73 2,28 2,30 0,99 Algeria 20,90 3,76 5,56 0,00 3,93 0,00 Egypt 12,50 5,34 2,34 0,21 6,08 0,03 Ghana 0,02 0,25 0,07 0,01 0,37 0,02 Kenya 0,01 0,40 0,02 0,01 0,63 0,01 Mauritania 0,01 0,08 0,06 1,20 0,11 11,43 Nigeria 4,65 1,28 3,62 0,01 2,02 0,01 Senegal 0,02 0,25 0,06 0,00 0,35 0,00 Argentina 0,00 4,13 0,00 0,98 7,70 0,13 Brazil 0,62 7,03 0,09 2,41 13,34 0,18 Canada 4,63 17,65 0,26 8,41 24,13 0,35 Chile 2,81 1,22 2,31 1,77 2,26 0,79 Costa Rica 3,69 0,54 6,83 0,01 0,85 0,01 Dominican Republic 0,00 0,72 0,00 0,00 1,11 0,00 Ecuador 0,83 0,48 1,73 0,38 0,84 0,45 Guatemala 3,95 0,59 6,75 0,03 0,98 0,03 Mexico 0,93 9,39 0,10 0,63 16,79 0,04 Panama 2,03 0,05 42,92 0,11 0,08 1,39 Peru 3,16 1,09 2,91 0,71 2,01 0,35 Trinidad and Tobago 0,00 0,40 0,00 0,00 0,55 0,00 United States 51,93 138,65 0,37 64,72 214,82 0,30 China 1,65 17,14 0,10 27,07 30,64 0,88 Hong Kong 4,33 4,91 0,88 2,22 6,82 0,33 India 0,81 9,80 0,08 4,83 16,06 0,30 Indonesia 0,00 2,28 0,00 0,94 4,52 0,21 Iran 16,49 6,30 2,62 0,38 7,16 0,05 Israel 1,34 8,79 0,15 9,73 8,21 1,19 Japan 12,35 62,51 0,20 4,05 98,65 0,04 Lebanon 0,02 1,75 0,01 0,05 1,63 0,03 Malaysia 0,07 2,17 0,03 1,65 3,52 0,47 Philippines 0,00 1,52 0,00 0,10 2,73 0,04 Republic of Korea 0,54 9,77 0,05 7,14 14,68 0,49 Singapore 0,12 2,62 0,05 1,41 3,77 0,37 Sri Lanka 0,19 0,58 0,33 1,07 0,90 1,18 Thailand 0,28 2,71 0,10 1,97 4,46 0,44 Turkey 2,95 11,28 0,26 17,60 10,70 1,65 Viet Nam 0,17 0,88 0,19 1,53 1,51 1,02 Yemen 1,36 0,47 2,92 0,00 0,64 0,00 Australia 0,89 5,09 0,17 4,74 9,66 0,49

26 26 Appendix 2 Actual and predicted Estonia s exports and imports in 2001, mill.$ Country Exports Exports Ratio Imports Imports Ratio mill $ predicted actual/predicted mill $ predicted actual/predicted Austria 16,34 26,50 0,62 40,42 31,80 1,27 Belgium 22,38 26,33 0,85 81,77 34,15 2,39 Denmark 116,45 52,13 2,23 108,37 66,21 1,64 Finland 1121,84 739,89 1,52 777,24 606,05 1,28 France 36,08 87,76 0,41 104,98 131,02 0,80 Germany 229,84 269,71 0,85 469,92 426,97 1,10 Greece 3,42 10,12 0,34 4,87 15,70 0,31 Ireland 12,54 11,90 1,05 43,70 16,04 2,72 Italy 32,37 66,82 0,48 141,58 105,50 1,34 Luxembourg 0,19 4,43 0,04 2,54 4,88 0,52 Netherlands 91,95 42,07 2,19 105,84 53,65 1,97 Portugal 4,34 7,16 0,61 10,99 12,65 0,87 Spain 9,26 30,16 0,31 38,41 53,39 0,72 Sweden 465,67 122,26 3,81 396,48 116,97 3,39 United Kingdom 139,56 97,39 1,43 101,46 142,79 0,71 Norway 92,68 57,02 1,63 45,96 69,81 0,66 Iceland 3,95 1,02 3,88 0,62 0,71 0,88 Switzerland 20,42 20,03 1,02 39,98 14,31 2,79 Czech rep 6,62 7,83 0,85 51,36 9,25 5,55 Hungary 31,03 6,64 4,67 44,22 8,21 5,38 Latvia 227,84 23,52 9,69 96,15 31,66 3,04 Lithuania 99,04 7,82 12,67 111,34 8,62 12,92 Poland 24,22 49,75 0,49 96,80 128,43 0,75 Slovakia 0,93 3,16 0,29 7,13 3,81 1,87 Slovenia 0,29 2,83 0,10 6,01 3,42 1,76 Cyprus 0,88 1,08 0,81 0,60 1,53 0,39 Bulgaria 0,36 1,47 0,25 3,84 2,10 1,83 Romania 0,37 3,91 0,10 2,35 5,60 0,42 Yugoslavia 0,11 1,23 0,09 0,78 0,95 0,81 Belarus 5,36 4,35 1,23 46,96 4,21 11,14 Moldova 2,62 0,37 7,14 1,78 0,49 3,66 Russia 90,66 104,92 0,86 349,55 205,13 1,70 Ukraine 47,43 6,27 7,56 43,38 8,10 5,35 Armenia 0,14 0,34 0,41 0,08 0,53 0,16 Azerbaijan 0,07 0,75 0,10 0,58 1,21 0,48 Georgia 0,22 0,44 0,49 0,15 0,72 0,20 Kazakhstan 0,68 1,48 0,46 6,98 2,99 2,34 Kyrgyzstan 0,01 0,15 0,06 0,57 0,31 1,85 Tajikistan 0,00 0,11 0,00 4,73 0,22 21,69 Turkmenistan 0,48 0,59 0,81 3,52 1,07 3,28 Uzbekistan 0,18 0,78 0,23 13,39 1,63 8,23 Algeria 3,24 2,64 1,23 0,00 2,69 0,00 Egypt 8,97 3,75 2,39 0,71 4,15 0,17 Ghana 0,96 0,18 5,27 1,43 0,26 5,57

27 27 Morocco 2,61 1,48 1,76 1,43 1,65 0,87 Nigeria 1,73 0,93 1,86 1,62 1,41 1,15 RSA 0,92 1,81 0,51 1,82 2,93 0,62 Togo 2,35 0,06 38,51 0,00 0,08 0,00 Argentina 0,12 3,08 0,04 1,18 5,43 0,22 Brazil 0,48 5,24 0,09 3,48 9,40 0,37 Canada 14,51 13,55 1,07 7,38 17,29 0,43 Chile 2,22 0,91 2,43 1,49 1,60 0,94 Colombia 0,01 1,31 0,00 1,27 2,18 0,58 Costa Rica 0,00 0,41 0,00 0,73 0,60 1,22 Ecuador 0,00 0,36 0,00 2,89 0,60 4,85 Mexico 3,11 7,18 0,43 2,18 11,99 0,18 Panama 6,24 0,04 174,05 1,19 0,06 21,48 USA 60,18 106,07 0,57 98,24 153,60 0,64 Bangladesh 0,12 0,98 0,12 2,01 1,52 1,32 China 14,73 13,20 1,12 375,89 21,99 17,09 Hong Kong 5,93 3,75 1,58 10,93 4,87 2,24 India 1,62 7,35 0,22 8,66 11,36 0,76 Indonesia 1,21 1,72 0,70 14,49 3,20 4,52 Iran 0,35 4,56 0,08 0,45 4,97 0,09 Israel 1,85 6,18 0,30 2,49 5,61 0,44 Japan 33,12 48,23 0,69 192,50 70,87 2,72 Malaysia 6,12 1,64 3,74 27,57 2,49 11,05 Pakistan 0,02 1,65 0,01 7,86 2,22 3,54 Philippines 0,59 1,16 0,51 4,02 1,95 2,07 Rep of Korea 21,49 7,53 2,85 26,43 10,54 2,51 Singapore 0,56 1,98 0,28 6,20 2,68 2,31 Srilanka 0,83 0,44 1,91 1,38 0,64 2,16 Syria 1,11 1,07 1,03 0,19 1,08 0,18 Thailand 2,24 2,05 1,09 10,98 3,17 3,46 Turkey 6,06 7,69 0,79 20,25 7,18 2,82 Vietnam 0,15 0,67 0,23 5,46 1,07 5,08 Yemen 2,00 0,34 5,89 0,00 0,44 0,00 Australia 1,52 3,86 0,39 1,24 6,87 0,18 New Zealand 0,07 0,69 0,10 0,71 1,22 0,59

28 28 Appendix 3 Actual and predicted Lithuania s exports and imports in 2001, mill.$ Country Exports Exports Ratio Imports Imports Ratio mill $ predicted actual/predicted mill $ predicted actual/predicted Austria 11,80 71,29 0,17 47,05 77,45 0,61 Belgium 65,53 55,67 1,18 122,66 72,90 1,68 Denmark 200,98 104,07 1,93 181,76 137,00 1,33 Finland 62,96 104,87 0,60 144,52 123,21 1,17 France 149,01 186,29 0,80 222,92 280,33 0,80 Germany 560,69 649,32 0, ,89 978,49 1,05 Greece 0,86 24,71 0,03 6,54 36,26 0,18 Ireland 9,04 22,85 0,40 25,07 32,50 0,77 Italy 86,98 159,27 0,55 248,86 240,47 1,03 Luxembourg 0,42 9,73 0,04 7,22 10,65 0,68 Netherlands 132,31 87,20 1,52 145,10 113,30 1,28 Portugal 5,96 14,79 0,40 14,40 26,66 0,54 Spain 59,35 63,65 0,93 76,34 113,87 0,67 Sweden 167,62 143,97 1,16 188,85 181,45 1,04 United Kingdom 629,97 196,80 3,20 209,31 297,41 0,70 Norway 59,18 87,15 0,68 55,91 124,84 0,45 Iceland 1,03 1,78 0,58 5,16 1,37 3,76 Switzerland 26,10 45,56 0,57 52,59 31,78 1,65 Czech Republic 20,86 20,22 1,03 81,86 22,03 3,72 Estonia 137,66 8,40 16,38 68,38 9,24 7,40 Hungary 11,46 18,78 0,61 52,50 20,56 2,55 Latvia 537,88 48,16 11,17 95,01 66,42 1,43 Poland 276,66 469,41 0,59 290, ,32 0,19 Slovakia 4,10 8,66 0,47 15,30 9,36 1,63 Slovenia 0,63 7,14 0,09 24,84 8,04 3,09 Cyprus 2,17 2,54 0,86 0,57 3,47 0,17 Bulgaria 2,54 3,86 0,66 12,93 5,04 2,57 Romania 1,17 10,71 0,11 2,88 13,77 0,21 Albania 0,01 1,25 0,01 0,00 0,93 0,00 Bosnia and Herzegovina 0,31 1,60 0,19 0,17 1,13 0,15 Croatia 0,42 6,16 0,07 4,58 4,06 1,13 FYR of Macedonia 2,54 1,21 2,11 0,11 0,85 0,13 Yugoslavia 1,49 3,30 0,45 1,12 2,32 0,48 Moldova 9,82 0,77 12,69 2,81 0,49 5,78 Russian Federation 406,76 234,08 1, ,70 450,97 3,46 Ukraine 128,35 20,81 6,17 88,04 22,12 3,98 Armenia 4,48 0,67 6,64 2,27 1,08 2,10 Azerbaijan 0,71 1,57 0,45 3,25 2,57 1,27 Georgia 2,12 1,00 2,12 0,74 1,59 0,46 Kazakhstan 17,04 2,93 5,83 6,91 6,15 1,12 Kyrgyzstan 3,39 0,31 11,04 0,10 0,63 0,16