Forecasting EU-Romania Trade by Gravity Analysis

Transcription

1 Forecasting EU-Romania Trade by Gravity Analysis Anna Ferragina, Giorgia Giovannetti and Francesco Pastore Abstract # This paper attempts to forecast EU-Romania trade using the gravity approach developed in Ferragina, Giovannetti and Pastore (2005). The trade potential of Romania with five EU members (France, Germany, Italy, Spain and UK) is computed using an out-of-sample methodology for the period The coefficients are taken from panel estimators of the gravity equation relative to intra-eu15 trade. The analysis suggests the existence of an important unexploited trade potential with Romania, which, nonetheless, is not much greater in size than that of the new EU members of Eastern Europe. The potential to actual trade ratio ranges from 2.2 to 2.7 and is sharply declining, suggesting that further dramatic economic integration is to be expected in the near future. Romania s accession is likely to further push the process of economic integration. JEL Classification: Keywords: C23; F15; F17; P45; P52 Romania; Europe Agreements; Eastward Enlargement; Gravity Equation; Trade Potential; ISSM-CNR, Via Pietro Castellino 111, Tel.: , Fax: annamaria.ferragina@issm.cnr.it e ferragina@economia.uniroma2.it. University of Florence, Department of Economics, and Italian Foreign Trade Commission, Via Curtatone 1, Firenze, Italy, Tel.: , Fax: giovanne@dicea.unifi.it e c.giovannetti@ice.it. Corresponding author: Seconda Università di Napoli and IZA, Department of Law and Economics, Palazzo Melzi, Piazza Matteotti, 81055, Santa Maria Capua Vetere (Caserta), Tel./Fax: , Mobile: , fpastore@unina.it. # Previous versions of Ferragina, Giovannetti and Pastore (2005) have been presented at the ETSG, University of Nottingham; at a CNR Study Group on International Trade, Università Commerciale Luigi Bocconi; and in a seminar held at the Romanian Academy of Science in We thank Lucian Liviu Albu, Paolo Epifani, Rodolfo Helg, Lelio Iapadre, Paolo Malanima, Mariana Nicolae, Elena Pelinescu, Lucia Tajoli and one anonymous referee for useful comments. However, the usual disclaimer applies. 1

2 Introduction In the last decade, gravity models have been extensively used to forecast potential bilateral trade relations and integration effects between EU (or OECD) countries and the former CMEA members 1. The aim of this paper is to compare the degree of integration of Romania with a group of EU countries (France, Germany, Italy, Spain and UK), indirectly providing an assessment of the relative success of the Europe Agreement (EA) with Romanian, also in view of the coming EU accession of the country 2. This study uses gravity analysis, which is the best alternative when intertemporal extrapolation of trade patterns is not feasible, as it is the case of CEECs, due to their past economic isolation, distorted pricing structures and recent transition from a planned to a market economy. 1 Section two reports the results of the existing literature (see also, among others, Brenton and Gros, 1997). 2 The Europe Agreements (EAs) were initiated by the EU with each CEEC separately. The first agreements with Poland, Hungary and Czechoslovakia were signed in December 1991 and came into force only in On 1 st February 1993, similar agreements were signed with Bulgaria, Romania as well as the newly established countries of the Czech Republic and Slovakia. They came into force in In 1998, EAs were implemented with the three Baltic States, followed by Slovenia on 1st February The EAs were aimed at fostering trade integration, but also the political dialogue and cultural and economic cooperation between the partners, while providing a basic outline for the gradual EU accession of CEECs. Over the period before the agreements came into force, Interim Agreements provided for an anticipated and temporary application of trade provisions. Their aim was to establish a free trade area for industrial goods for ten years on a reciprocal, but asymmetric basis: the EU had to remove its trade barriers more quickly than the CEECs. This led to the total removal of all tariff barriers on industrial products from the EU on 1 st January However, for some sensitive industrial sectors a special discipline was created, in particular for textiles, iron and steel, car industry (only for Poland) and a much more gradual liberalization was applied to agricultural goods and fisheries. 2

3 The estimates of trade potential are based on the approach developed in Ferragina, Giovannetti and Pastore (2005), who use a panel data specification of the gravity model and an out-of-sample methodology. In other words, the parameters extracted from a gravity equation of intra-eu bilateral trade flows are used to predict trade between Romania and several EU members. This analysis aims at answering the following main questions: 1) What degree of trade integration could have been achieved if the trade elasticity with respect to economic and geographic variables (relative mass, physical distance, common language, common land border, colony links) had been like those achieved in intra-eu trade? 2) Has the EA with Romania induced a reduction of the gap between potential and actual trade via trade creation and diversion activated by the liberalisation process? 3) How much additional trade could be created if integration would be pushed further (trade enhancing effect), for instance, via EU accession? From an econometric point of view, this study differs from previous works on CEECs, which have hardly used panel analysis, and even more rarely coupled panel analysis with out-of-sample methods 3. The analysis provides evidence of substantial unexploited trade, which is in contrast with previous results relative to EU trade with CEECs: with the only exception of Baldwin (1994), who refers to pretransition data, most studies predicted that trade potential was almost fully exploited already in The differences between our and previous results might depend on three factors. This study a) focuses on intra-eu trade, rather than on trade with a larger, but less homogeneous group of countries, as a reference to estimate the 3 To our knowledge, only Baldwin (1994), Gros and Gonciarz (1996), Mathyas (1997) and Egger (2000; and 2002) use panel analysis, while Gros and Gonciarz (1996) use both panel analysis and an out-of-sample method. 3

4 elasticity of trade determinants; b) applies an out-of-sample method to compute potential trade; c) concentrates on a later period, , when GDP in CEECs in general and in Romania, in particular, was rapidly increasing. The results are remarkably stable across different specifications and estimation methods. The work is organised as follows. Section 2 contains a survey of the literature. Section 3 discusses the methodology adopted and the results of the gravity model. Section 4 provides measures of the ratio between potential and actual trade. Some concluding remarks follow. 2. A survey During the last decade or so, gravity models have been widely used to forecast EU-CEE trade potential, and to compare it with the actual level of trade. More specifically, gravity equations have been estimated relative to the already integrated EU15 to obtain the coefficients of the main determinants of bilateral trade, namely national income and population size of the countries involved, distance, common land border and common language. These coefficients have been used in similar equations, but with variables relative to EU-CEECs, in order to assess trade potential. The aim of these exercises has been to determine whether the integration process was already completed before accession, or whether one can expect further trade integration, which may continue to affect the labour markets in the two areas concerned. The results of such exercises available in the literature are mixed, and they depend closely on the period considered, specification and estimator, as well as on the computation method used to calculate trade potential. Wang and Winters (1992) 4

5 find that East-East trade was large in 1985, while East-West trade was only a fraction of what it would have been in an integrated Europe. Hamilton and Winters (1992) adopt a similar approach, finding that trade within the former Soviet Union and the Eastern Europe bloc (SUEE) was static or falling, while trade with Western Europe may increase by up to five times. Baldwin (1994) finds that potential EU12- CEE exports and imports are twice the actual 1989 exports and imports. All the studies which have used data on the early transition period in order to estimate trade potential support a different conclusion. Most of them suggest that a level of integration between Eastern and Western countries, which is high and above the potential level, has long since been reached. This indicates that adjustment is complete and that there is no need for special protection in Western countries (Gros and Gonciarz, 1996; Brenton and Di Mauro, 1999; Nilsson, 2000). For instance, Gros and Gonciarz (1996) correct Baldwin s estimates on the grounds that he used a GDP which was overvalued because it was calculated on pretransition data (the per capita GDP used by Baldwin was much higher than the 1992 GDP for CEECs). Combining the parameters from Baldwin (1994) with the 1992 data on GDP, Gros and Gonciarz end up with a downward revision of Baldwin s projections of CEEC-EU trade, and their results suggest that the adjustment is complete. Two recent studies on trade integration measured by gravity models (Egger, 2000; 2002) have cast doubt on the results of the above-cited literature. They make three main criticisms: 1) most of these results are based on cross-section gravity models which are mis-specified because they do not take account of exporter and importer effects, while only few authors make use of panel econometrics; 2) those 5

6 authors who do use panel analysis to compute potential trade adopt a random effect model (the exception is Egger) which may be affected by the problem of correlation between the explanatory variables and the unobserved time invariant effects; 3. most analyses obtain information on trade potentials using the in-sample prediction approach; that is, the residuals of the estimated equation are interpreted as the difference between potential and actual bilateral trade relations, but this is in contrast with the fact that, in the case of proper specification, the estimators are consistent and efficient and therefore should exhibit white-noise residuals, rather than identifying large systematic differences between observed and in-sample predicted values among country groups. 3. Methodology This contribution combines a panel data analysis of intra-eu15 trade with an out-of-sample calculation of potential EU-Romania trade relative to the period In other words, we estimate the parameters of a gravity model for intra- EU15 trade 4 and then plug them into the regression of trade between Romania and her main European trade partners (Italy, Germany, France, UK, and Spain) to obtain potential trade. This potential or normal trade is then compared to actual trade volumes to assess the dimension of trade potential not exploited in the short run. As Ferragina, Giovannetti and Pastore (2005) discuss in more detail, the following specification has been used to analyse the determinants of intra-eu15 trade. 6

7 X ijt = α + β i + β CLB 6 1 POPit + β 2GDPPC it + β 3 POP jt + β 4GDPPC jt + β 5 Dij [1] ij + β CL 7 ij + ε ijt where: i are the countries of origin, j are the destination countries and t = is the period under examination 5 ; X ij are exports of country i to country j in real terms; α ι is the bilateral constant; POP it and POP jt are the populations at time t of country i and j respectively; GDPPC it and GDPPC jt are per capita GDP of country i and j at time t in real terms; D ij is the geographical distance in Km between the capital city of country i and of country j; CLB is a dummy equal to one, if the two countries share a common land border and/or have ex-colony links and zero otherwise. CL is a dummy for common language taking a value of one if the 2 trade partners speak the same language and zero otherwise and ε IID(, σ ) ijt 0 ε. All the variables in our equation, except for the dummies, are in natural logarithms and therefore the estimated parameters can be interpreted as elasticities. According to Feenstra et al (2001, p. 432) the constant term should represent the impact of world income on bilateral trade within the sample. Bilateral exports are expected to be positively influenced by: a) the importer demand and exporter supply, as proxied by their population (POP) and per capita income (GDPPC), respectively. A higher per capita income means a higher import demand and export supply, as it is a proxy of the economic development of the country. The effect of population is more ambiguous: a larger 4 We include 13 EU countries in the empirical analysis, since for Belgium and Luxembourg data were missing for some years. 5 Notice that while the gravity equation has been estimated over the period , potential trade is computed over the years This is because some observations for the year 2002 were missing. They have been substituted by average values relative to the previous three years. 7

8 population means a large domestic market, a higher degree of self-sufficiency and less need to trade. At the same time, a large population entails a deeper labour division and scale economies in production which are generally associated in the theoretical models with a larger need for trading. Therefore, the effects of this variable are ambivalent. b) dummies such as sharing a land or a sea border, ex-colony links, common language capture the geographical closeness, the better information, the lower cultural differences, the lower home bias and research and communication costs associated with proximity (familiarity with custom regime, institutions, legal system). Conversely, bilateral exports are expected to be negatively influenced by distance which can be expressed as geographical distance, but also by the surface of host s markets, the level of trade costs, the presence of a home bias effect and of time and search costs. As detailed in Ferragina, Giovannetti and Pastore (2005, Table 2), all the available test statistics support the adoption of a random effect model with exporting and importing countries and years dummies. This model has consistent coefficients like the fixed-effect model, while being more efficient. In the preferred specification, the variables have the expected sign and are highly significant. The explanatory power of the model is also high. Also the coefficients are reasonable. The statistical significance of distance in explaining the trade intensity of EU countries suggests that transport costs still have an important impact on the export performance. The estimated parameter for distance is stable across different specification and is in line with that typically 8

9 found in earlier studies (- 0,7). Similar to previous studies, the coefficient for the per capita GDP of the exporting country is lower than that of the importing country. This is to be expected, considering that exports are more related to the income level of the importing, rather than of the exporting country. The coefficient of the per capita GDP of the importing country is slightly greater than 1, which is what one would expect if the propensity to import from EU countries was constant over time and the share of EU over total imports was constant. 4. Romania-EU trade potential A short discussion of the recent evolution of Romania-EU trade patterns seems useful to put the discussion about trade potential into context. Romania is not much different under this respect from other CEE countries. Asymmetric trade with the EU afflicts CEECs. In 2003 exports to CEECs as a share of total EU exports had become more than 12% from only 5% in 1990 and, for imports, the share was above 10%. The average share of exports and imports that CEECs devote to EU is more than 60 and 55% respectively. In few years exports to the EU as a share of total exports have become on average more than 60% of CEECs total exports (according to Eurostat they range from a minimum of 48% for Lithuania to a maximum of 70% for Poland with some impressive increases over the period from 4 to 70% in the case of Estonia, from 6 to 55% in the case of Bulgaria from 5 to 48% in the case of Lithuania). The average share was only 30% in Imports are on average more than 55%, starting from an average of 32% in 1990, with some peaks like Poland and the Czech Republic (more than 60%), followed by Romania (57%). 9

10 As already noted, to check whether Romania-EU trade approached its potential level in the period under consideration, we apply the estimated coefficients from the gravity equation of intra-ue15 trade to the same specification for EU-CEE trade flows. These parameters are used as a benchmark to estimate the potential integration that CEECs and Romania, in particular, might obtain if the elasticity of trade determinants were the same as those observed in the case of intra-eu15 trade. Trade volumes are then considered normal trade, which could be obtained with a deeper integration according to the predictions of the gravity model. We applied the same procedure to EU-CEE trade. Figure 1 contains the trends in the ratio between potential and actual trade between Romania and each of five EU main trade partners (France, Germany, Italy, Spain and UK) during the period A ratio of one suggests that potential trade equals actual trade. The higher is the ratio, the higher is the gap that has to be filled and therefore the possibility to create new trade. A decreasing (increasing) trend of this ratio over time suggests that trade is increasing (decreasing) and tends to approach its potential level. [Figure 1 about here] The analysis suggests the existence of important unexploited trade potential with Romania. The potential to actual trade ratio ranges from 2.2 to 2.7 from one country to the other, suggesting that further dramatic economic integration is to be expected in the near future. This is also confirmed by the clear downward trend of the potential to actual trade ratio. In other words, the actual trade is still less than 50 percent of its potential. 10

11 The decline is especially clear in the case of trade with Italy and Germany, which already had the lowest ratio at the beginning of the period considered. This suggests in turn that there is no catching up in the trade relationships between Romania and EU countries, but rather a strengthening of existing trade links. The projected/actual ratio of imports systematically exceeds that of exports. Only in the case of Romania-Spain trade, the two lines cross each other. This result is consistent with the EU trade surpluses with the CEECs. It indicates also that there is wide scope for an increase in imports more than in exports and might also suggest that the CEECs within the EAs have not benefited of a total and preferential opening for their exports to the EU. It is worth mentioning that the results contained in Figure 1 are robust to different estimation methods, with only small variations. As an example, Figure 2 provide the same ratio as in Figure 1, but based on coefficients drawn from a fixed effect model of intra-eu trade. Comparison of the two figures shows that the differences are negligible. [Figure 2 about here] The existence of large unexploited trade potential is typical also of other CEECs. Ferragina, Giovannetti and Pastore (2005) report similar results for other nine CEECs. The trend is marked by a large decline of the ratio in all cases: they start from a ratio of around 2 and further close the gap (especially in the case of countries which started from the worst positions such as the Baltic Republics and Bulgaria, which show the most dynamic trend). The trade potential between EU and CEECs is close to be exploited in 2002: the ratio is between 2 and

12 However, the gap should not be underrated. Without external positive shocks, the analysis carried out here suggests that much time is still necessary to Romania to close the gap. In fact, as shown in Table 1, assuming a constant population of the countries considered and taking for realistic the World Bank forecast of a 5% rate of growth of per capita GDP for Romania over the period , the EU export to Romania will increase by about 6.4 and the EU imports from Romania by 4.8 per year. Spain and the UK should perform slightly better, essentially because their projected GDP growth rate is higher. [Table 1 about here] According to the calculations reported in Table 1, Romania will perform not only better compared to Bulgaria, but also to several new EU members, with the only exception of the Baltic Republics. Table 2 reports the number of years, which are necessary according to the previous calculations for the actual to reach potential levels of trade. The figures have been obtained by dividing the potential to actual trade ratio in 2002 (multiplied by 100) for the projected export and import annual growth rates reported in Table 1. Inspection of the table suggests that even if Romania grows at the same rate also in the years after 2007, about 36 and almost 50 years are still necessary to close the gap in EU export to and import from Romania. Again, Romania seems to perform better than several other countries, including those that are already EU members. [Table 2 about here] This forecasting exercise suggests that economic integration of Romania, and other CEECs, with the EU has a very high potential, which cannot be fully 12

13 exploited unless the EU grows faster or some important policy factors that hinder trade expansion are removed. Next section turns to discuss such factors. 5. Policy factors that hinder trade expansion The EAs entailed a gradual liberalisation of tariffs for trade in industrial goods accompanied by a strong commitment to reform the institutional framework in exchange for technical and financial assistance from the EU budget. However, in that case the dismantling was scheduled for a closer deadline and was not only more rapid but also more effective. Also structural reforms obligations imposed to the CEECs were much stricter and challenging. But what made everything more effective was certainly the perspective of joining the EU, which made inevitable to adhere to the Acquis Communitaire. However, even though trade liberalization plays a pivotal role in fostering a high degree of openness, a number of factors other than trade policy measures have exerted a significant influence as well. The extent of trade openness could be linked to the characteristics of the industrial system, i.e. countries producing intermediate or finished goods matching a growing world demand generally trade more than countries producing goods associated with a less dynamic global demand. Moreover, in the CEE economies the achievement of rapid integration in international trade had been favoured by the application of programmes which promoted the private economy creating a favourable environment and institutional framework. 13

14 The implementation of reforms was also crucial to gain credibility and to attract foreign investment, which are an accelerator of trade expansion and have recently produced a deep penetration of the CEECs goods in EU markets. Given this scenario, the agreements have to produce a deepening and widening of EU relations not only by providing accession to the EU, but also by establishing a broad economic, financial, cultural and political co-operation including issues such as immigration, environment, law enforcement and infrastructure 6. This also entails provision of technical and scientific assistance and a great commitment on the ground of human resources training in the CEE economies. Only in this way, together with the static benefits of improved resource allocation, the dynamic benefits of liberalisation deriving from greater efficiency can be achieved. Concluding remarks The results of our gravity analysis confirm that there is still a large trade potential between EU and Romania, though the trend is towards a dramatic decline. If the World Bank optimistic predictions regarding Romania s economic growth will come true, the catching up will happen in not less than ten years. Considering also the low growth rate of the EU, a decline in Romanian growth is bound to cause also a decline in economic integration with the EU. These results, in turn, are worrisome if one thinks of the possible further adjustment process, which might 6 But it is common belief that to attain benefits the agreements need to be complemented by deeper structural reforms and have to be part of a governments overall development strategy. Previous experiences seem to suggest that successful regional trade initiatives tend to be an extension of domestic reforms rather than a catalyst that generates such reforms. 14

15 take place in the labour market of EU members and of associate countries in the next few years. The result that trade potential is still not exhausted is in line with Baldwin, but not with other studies. Gros and Gonciarz (1996), for instance, suggest that the trade potential had been already reached in But they base their analysis on a very negative period for the CEECs, unlike Baldwin. In the first half of the 1990s, the CEECs were in the descending part of the J-curve of transition. GDP was much lower than the current level suggesting that potential had already been exploited. Baldwin considers the much higher income level of the pre-transition period and obtains results quite similar to ours. In all CEECs, GDP almost doubled between 1992 and 2002, which translates into a progressive reduction in the potential and a trend towards convergence with actual trade. 15

16 References Baldwin, R. E. (1994). Towards an Integrated Europe. London: Centre for Economic Policy Research. Brenton, P., and D. Gros (1997). Trade Reorientation and Recovery in Transition Economies. Oxford Review of Economic Policy 13(2): Brenton, P., and F. Di Mauro (1999). The Potential Magnitude and Impact of FDI Flows to CEECs. Journal of Economic Integration 14(1): Egger, P. (2000). A Note on the Proper Econometric Specification of the Gravity Equation. Economic Letters 66(1): Egger, P. (2002). An Econometric View on Estimation of Gravity Models and the Calculation of Trade Potential. The World Economy 25(2): Eurostat (various years), Statistics in focus, Bruxelles. Feenstra, R.C., J.R. Markusen, and A.K. Rose (2001). Using the Gravity Equation to Differentiate among Alternative Theories of Trade. Canadian Journal of Economics 34(2): Ferragina, A.M., G. Giovanetti and F. Pastore (2005), EU Actual and Potential Trade with Mediterranean and Central and Eastern European Countries. A Gravity Study, ICE, mimeo. Gros, D., and Gonciarz A. (1996). A Note on the Trade Potential of Central and Eastern Europe. European Journal of Political Economy 12(4): Hamilton, C.B., and A.L. Winters (1992). Opening Up International Trade with Eastern Europe. Economic Policy 14: Mathyas, L. (1997). Proper Econometric Specification of the Gravity Model. The World Economy 20(3): Nilsson, L. (2000). Trade integration and the EU economic membership criteria. European Journal of Political Economy 16(4): Wang, Z., and Winters A.L. (1992). The Trading Potential of Eastern Europe. Journal of Economic Integration 7(2):

17 Annex Table 1: Projected export and import annual growth rates (in %). Projected GDP per France Germany Italy Spain UK EU CEEC capita annual growth rates E 1 M 2 E M E M E M E M E M Bulgaria 8,2 9,7 6,4 9,4 5,7 9,6 6,1 10,2 7,3 10,0 7,0 9,8 6,5 Estonia 5,8 7,2 5,1 6,8 4,5 7,0 4,9 7,6 6,0 7,5 5,7 7,2 5,2 Hungary 3 3,9 5,1 4,1 4,8 3,4 5,0 3,9 5,6 5,0 5,5 4,7 5,2 4,2 Latvia 6,8 8,2 5,6 7,9 5,0 8,1 5,4 8,7 6,6 8,5 6,3 8,3 5,8 Lithuania 6,6 8,0 5,5 7,7 4,9 7,9 5,3 8,5 6,5 8,3 6,1 8,1 5,7 Poland 3,2 4,4 3,7 4,1 3,1 4,3 3,5 4,9 4,7 4,7 4,3 4,5 3,9 Romania 5 6,3 4,7 6,0 4,0 6,2 4,5 6,8 5,6 6,6 5,3 6,4 4,8 Slovenia 3 4,1 5,4 4,2 5,0 3,6 5,2 4,0 5,8 5,1 5,7 4,8 5,4 4,3 The Czech Republic 3 2,1 3,2 3,1 2,9 2,5 3,1 2,9 3,7 4,1 3,6 3,8 3,3 3,3 The Slovak Republic 3,8 5,0 4,0 4,7 3,4 4,9 3,8 5,5 5,0 5,4 4,7 5,1 4,2 Notes: 1 Projected growth rates of exports; 2 Projected growth rates of imports; 3 Based on projected average incomes growth rates Source: Own elaboration on World Bank projected annual growth rates of per capita GDP. Table 2: Number of years necessary for actual to approach potential trade France Germany Italy Spain UK EU CEEC E 1 M 2 E M E M E M E M E M Bulgaria Estonia Hungary Latvia Lithuania Poland Romania Slovenia The Czech Republic The Slovak Republic Notes: 1 Exports; 2 Imports; 3 Based on average incomes growth rates Source: Own elaboration. 17

18 Figure 1. Estimates based on a random effect model Romania-EU: Ratio of potential to actual trade (1995-'02) France Germany Italy Potential trade / Actual trade Spain UK Romania Export Import Graphs by eu Figure 2. Estimates based on a fixed effect model Romania-EU: Ratio of potential to actual trade (1995-'02) France Germany Italy Potential trade / Actual trade Spain UK Romania Esportazioni Importazioni Graphs by eu 18