EU enlargement: Economic implications for countries and industries

Transcription

1 EU enlargement: Economic implications for countries and industries A.M. Lejour, R.A. de Mooij and R. Nahuis 1 Very preliminary, comments invited 1. I n t r o du ction This paper explores the economic consequences of the enlargement of the European Union (EU) with ten Central and Eastern European Countries (CEECs). In contrast to most earlier analyses, we do not focus on existing formal trade barriers. The reason is that, by the end of 2002, these barriers will be removed entirely in accordance with the Europe agreements -- at least for industry products. Instead, we focus on further steps in the integration process which primarily involve the accession to the internal market, the equalization of external tariffs and free movement of labour. The consequences of accession to the internal market have been analysed before using models for the world economy (see e.g. Baldwin et al, 1997 and the references therein). These studies usually simulate the accession to the internal market by means of an exogenous across-the-board reduction in trade costs. Our analysis deviates from this approach in two important ways. First of all, we take account of the sectoral variation in the reduction of trade costs. Indeed, enlargement of the internal market has disproportional effects on some industries. Second, rather than simulating a fixed reduction in trade costs, we estimate gravity equations to derive the size of the shock for various industries. More specifically, for 16 different industries, we estimate gravity equations to determine potential trade between the EU and the CEECs. The estimates provide an indication of trade flows when these countries are a full member of the EU. Comparing this potential trade with actual trade, we arrive at an estimate of the virtual non-tariff barriers that will be removed when countries accede to the EU. This nontariff trade barrier per sector is used as an input for our model simulations. The paper adopts a CGE model for the world economy, called WorldScan, to explore the implications of EU enlargement. The model, calibrated on the most recent version of the GTAP database, has a number of features that make it appropriate for analysing the impact of enlargement. First, the model makes an explicit distinction 1 Corresponding author: CPB Netherlands Bureau for Economic Analysis, P.O. Box 80510, 2508 GM The Hague, The Netherlands. Tel ; radm@cpb.nl. 1

2 between, on the one hand, six regions in the EU and, on the other hand, Poland, Hungary and the other accession countries. Second, the model distinguishes between 16 industries so that we are able to explore which industries will be most affected by EU-enlargement. Thus, combined with the gravity approach, the model does justice to the sectoral variation in the reduction in trade costs. Simulations indeed reveal substantial differences in the trade impact on different industries. The rest of this paper is organised as follows. Section 2 discusses the main features of the WorldScan model, its calibration and the baseline scenario. Section 3 demonstrates the shock of EU-enlargement in two dimensions: the accession to the internal EU market and free movement of labour. Section 4 analyses the implication of these shocks for both the EU and accession countries. Section 5 concludes. 2. T he Wo rl ds c a n m o d e l WorldScan is a computable general equilibrium model for the world economy. 2 The model is calibrated on the basis of the GTAP database, ver sion 5 (Purdue 2001) with 1997 as the base year. The database allows us to distinguish between a large number of regions and sectors. In particular, the EU is divided into six regions: Germany, France, UK, Netherlands, South EU (comprising Italy, Spain, Portugal and Greece) and Rest EU (comprising Austria, Belgium, Luxembourg, Ireland, Denmark, Sweden and Finland). The accession countries are divided into three regions: Poland, Hungary and Rest CEECs (comprising Czech Republic, Slovak Republic, Slovenia, Estonia, Latvia, Lithuania, Bulgaria and Romania). The rest of the world economy is divided into three other regions, namely, Soviet Union, rest OECD and rest of the world. For each region, we distinguish sixteen sectors. These consists of Agriculture, Raw Material, ten manufacturing sectors and four service sectors. The appendix (available upon request) provides more details. The heart of the model relies on neoclassical theories of growth and international trade. Sectoral production technologies are modelled as nested CES functions. At the lower nesting, value-added is produced by combining low-skilled labour, high-skilled labour, capital and, in some sectors, a fixed factor (land in the sector Agriculture and natural resources in the sector Raw materials), using a Cobb-douglas technology. At the lower nesting, the intermediate inputs are combined in a CES function with a substitution elasticity of 0.8. In principle, there exist sixteen intermediate inputs. However, there are only a few intermediate inputs important in the production process for most sectors. At the higher nesting value-added and intermediate inputs ae combined in a CES technology to yield final output with a substitution elasticities of 0.4. With respect to trade, WorldScan adopts an Armington specification, explaining two-way trade between regions and allowing market power of each region in the medium run. In the long run, trade patterns are determined by Heckscher-Ohlin mechanisms, i.e. based on factor endowments. On the capital market, WorldScan assumes imperfect capital mobility across borders. In particular, the model includes a portfolio mechanism in which capital owners distribute their investments over regions, depending on the rates of return and the preferences for asset diversification. Consumption patterns may differ across countries and depend on 2 See CPB (1999) for more details. 2

3 per capita income. If welfare levels converge, these consumption patterns also converge towards a universal pattern. We assume that the labour markets for low-and high-skilled workers clear. In the baseline, labour does not migrate. Table 1: Characteristics of the CEEC in 1997 population GDP savings/gdp investment (in millions) (billion US$) ratio /GDP ratio Hungary Poland Rest CEEC EU CEEC in % EU-15 26% 3.7% source: UN (1998); Purdue (2001) Table 2: Trade shares (in % of GDP) and regional trade patterns (1997) EU ROW CEEC total Hungary - export import Poland - export import Rest CEEC - export import Tables 1 and 2 provide some background information about the CEEC economies. Table 1 reveals that EU enlargement implies an increase in the EU population by around 28%, while GDP will rise only by a mere 4%. Table 2 presents the export and import shares of the CEECs. Especially Hungary is a relatively open as it exports more than half of its GDP, a feature that is common for smaller countries. The CEEC s trade patterns are geared to the EU. Intra CEEC trade is only marginal. Trade with the rest of the world mainly concerns the Rest OECD. All CEECs experience a trade deficit in Indeed, table 1 reveals that investment/gdp ratio is higher than the saving/gdp ratio. A large share of investment is thus financed by foreign capital, particularly in Poland. Hence, the trade deficit is associated with a net inward FDI flow. 3

4 Table 3: Sectoral output shares (in % of total value added) in Hungary Poland Rest CEEC EU Agriculture Raw Materials Food Processing Textiles and Leather Non metallic Minerals Energy-intensive Products Other Manufacturing Metals Fabricated Metal Products Machinery and Equipment Electronic Equipment Transport Equipment Trade Services Transport and communication Financial services Other services source: Purdue (2001). Table 3 shows the sectoral output shares for all sixteen sectors distinguished in WorldScan, both for the EU and the CEECs. It shows that Agriculture, Food Processing and Textiles contribute relatively more to GDP in the CEEC s than in the EU. Electronic and Transport Equipment and Other Services contribute relatively little to GDP in these countries. Within the CEEC, there exist some remarkable differences. In Hungary Agriculture, Food Processing, Other Manufacturing and Metals are less important than in Poland and the Rest CEEC. These sectors are typically low-skilled labour intensive. Other Services is more important in Hungary than in the other regions. In the Rest CEEC, manufacturing contributes much to GDP, especially Textiles and Other Manufacturing. Services are less important, especially Trade and Other Services. Table 4 presents the trade shares of the various sectors in terms of value-added by that sector. The table reveals that the trade shares of the CEECs are large in most manufacturing sectors, relative to agriculture and services sectors. Hungarian sectors are typically more open than their counterparts in the other CEEC regions. To illustrate, Hungarian production of Textiles and Leather is almost entirely exported, while the 4

5 corresponding export share in Poland and Rest CEEC is much smaller. This suggests that Hungary engages much more in intra-industry trade than the other regions. Table 4 Sectoral trade (as % of sectoral output) in the CEECs in 1997 Hungary Poland Rest CEEC export import export import export import Agriculture Raw Materials Food Processing Textiles and Leather Non metallic Minerals Energy-inten. Products Other Manufacturing Metals Fab. Metal Products Mach. and Equipment Electronic Equipment Transport Equipment Trade Services Transport and commu Financial services Other services source: Purdue (2001), exports are a ratio of sectoral production, imports are a ratio of domestic absorption s e l i n e Ba In exploring the economic impact of EU-enlargement with WorldScan, we compare economic variables in 2020 with the results in a baseline scenario. In the baseline, GDP growth in Eastern Europe, the Former Soviet Union and the Rest of the World are based on long-term projections of the Worldbank (2000). They have constructed projections until 2010 for all developing regions and extrapolated these projections to Economic growth in Hungary and Poland is 4.6%, which is a bit higher than in the Rest of Eastern Europe (4.3%) because we assume that the pace of reform in Bulgaria and Romania is relatively slow. For Western Europe and the rest of the OECD, GDP grows by about 2.1%, while in the Rest of the World it is nearly 5%, especially due to high growth in Asia and in particular China. In the baseline, there are no further agreements on global trade liberalization so that the degree of openness remains at a stable level in the scenario period. 5

6 % & % ' ( % 3. S ho c k o f en larg t e m en This section discusses three shocks of EU-enlargement: (i) a gradual removal of formal barriers and the adoption of the common external tariff (CET), 3 (ii) accession to the internal market, and (iii) free movement of labour. In the experiments, we assume that Poland and Hungary enter the EU in 2004 and the Rest CEEC in The effects, evaluated in the year 2020, should be interpreted as the allocative implications of the three shocks and possible implications through capital accumulation. We do not analyse some other potential implications of enlargement such as an increase in political stability or the fact that accession to the EU may be followed by EMU membership after some time. Finally, we do not explore the implications of changes in the Common Agricultural Policies of the EU or the EU policies with respect to the Structural Funds. The rest of this section discusses how we determine the size of the latter two shocks (the first shock consists of removing the existing trade barriers and adoption of a CET, an appendix providing these numbers is available upon request) T ow a r d s a cus t oms u nio n Accession of the CEECs to the EU implies a move from an almost free-trade area towards a customs union. This means that all remaining bilateral formal trade barriers will be abolished. In addition, the external tariffs with respect to third countries will be set equal to the common external tariff (CET)of the EU. An appendix that demonstrates the magnitude of the existing export and import tariffs, both between the EU and the CEECs, and of these countries vis a vis third countries is available upon request.! 3. " 2 # $ Access io n to the in t % e r n al market We explore the implications of the accession of the CEECs to the internal market. The inclusion of the CEECs in the Single Market can affect the economies of the CEECs and current EU members via FDI, domestic investment and trade. Our focus is on the latter effect. The first key element of the CEECs accession to the Single Market is that administrative barriers to trade are eliminated or at least reduced to levels comparable to those between current EU members; think of reduced costs of passing customs at the frontier: less time delays, less formalities etc. Probably more important is the reduction in technical barriers to trade. The Single Market reduces these technical barriers by means of mutual recognition of different technical regulations, minimum requirements and harmonisation of rules and regulations. 4 Risk and uncertainty is a second impediment on trade that will be mitigated by the CEECs accession to the EU. One type of risk is the possibility that somewhere in the link from producer to consumer some agent defaults. This is especially important for goods moving from East to West, as export credit 3 As, by now, the formation of a customs union is completed we take in the remainder of the paper the Customs Union as a point of departure. 4 For a detailed discussion of these approaches and their effect on trade, see Brenton, Sheehy and Vancauteren (2000). 6

7 guarantees are less well developed in the CEECs. A second type of risk is political risk, a risk more relevant for goods moving from West to East (as insurance does not cover these risks and as democracies are thought to be less stable in the CEECs). To measure the economic consequences of accession to the internal market, we estimate gravity equations on the industry level to obtain potential trade flows between the CEECs and the current EU members, i.e. trade as if the CEECs were already EU members. The gravity m o d e l Reminiscent to the law of gravity in physics, the gravity model suggests that the trade flow between two countries depends positively on their size and negatively on the distance between them. In economic terms, trade flows between two countries depend on the importer s demand and the exporters supply and on the cost of trade. The latter is proxied by distance and specific characteristics of the bilateral country relation, like sharing a language or having a common border. The importer s demand and exporter s supply is proxied for by their outputs and per capita incomes. The early contributions applying the gravity approach (e.g. Tinbergen, 1962), did not provide a theoretical motivation for the model. Nevertheless, the model evolved to becoming the workhorse model of empirical international trade. Helpman and Krugman (1985) show, however, that the basic gravity equation is simply derived from a trade model with differentiated goods. Deardorff (1995), moreover, demonstrates that the gravity equation is consistent with the Heckscher-Ohlin theory of international trade. Though this consistency of the gravity model with different trade models hinders the gravity model s usefulness for model testing and selection, it increases the confidence in its use as a predictor for (potential) trade patterns. Previous studies on the potential trade between CEECs and the current EU members largely restrict themselves to an economy-wide perspective. Without questioning the value of the insights delivered by the economy-wide perspective, it is evident - given the relative size of the CEECs to the current EU - that an industry-level analysis is required to shed light on potentially painful adjustment problems and promising opportunities. The gravity model is rarelyused on the industry level. There are a few exceptions, however (ref.). Bergstrand (1989) derives a gravity equation for a multi-industry world on that basis of a 2 x 2 x n Heckscher-Ohlin model, where n is the number of countries. The model allows for intra-industry trade. Extensions to a higher industry dimension turn out to be cumbersome (see Bergstrand, 1989). Estimation of this gravity equation an one-digit SITC industry level data of the 1960s and 1970s yield plausible results. We follow Bergstrand (1989) in estimating the following gravity model (see Box): M J (1) 7

8 where X ijs stands for (log of) exports from country i to j in industry s. D EU is a dummy that equals unity if i and j are currently members of the EU (else zero), d measures distance between the countries i and j and Y(y) is the (log of) GDP (per capita) of the exporting and importing countries. The set D d denotes the other dummies included. In estimating (1), we use data from 1998 for 16 industries and 26 countries. 5 The economic data are derived from GTAP (see section 2; McDougall, 1998; and CPB, 1999); the distance data are the Great Circle distances between capital cities. Further details about the data are provided in Nahuis (2001). We estimate equation (1) with OLS. The estimation results are reported in table 5, where t-statistics are reported below the parameter estimates. 6 Before turning to the EU-dummies, we take a closer look at the other parameters. The distance variable is negative and significant in all but one industry (Transport and Communication). The size of the estimated coefficient is, however, notably lower for service sectors. This indicates that, if the service sector s products are tradeable, distance matter less; a result that is intuitively clear once one thinks about financial services for example. The exporter and importer income coefficients are estimated precisely and are all positive. Eyeballing at these coefficients for a moment learns that all but two are less than unity, a common finding in the literature. This implies that trade rises less than proportionally with size. Or, saying the same thing in a more familiar way: small countries tend to be more open. The export s per capita income term relates to the capital intensity of production. Without wanting to dwell or these results, some high-tech sectors (e.g. Electronic Equipment, Machinery and Equipment) do appear with a positive and significant sign, whereas the labour-intensive service industries (Textiles and Leather; Trade Services) have negative signs. Hence, rich countries are typically more high-tech. The negative signs for importer per capita income in all but one industry (Transport and Communication) are slightly puzzling, as these indicate that the Transport and Communication sector produces the only luxuries in the consumption basket. The adjacency dummy is significant and positive for all non-service sectors except for Electronic Equipment. This indicates that sharing a common border lowers trade costs. 5 Some countries are aggregates of countries, the accompanying choices for the distance variable are made explicit in Appendix (available upon request) ]. 6 An extensive robustness analysis is provided in Nahuis (2001). 8

9 Dummy Distance Exporter Exporter Importer Importer Dummy Constant R-square Potential trade EU Income Income p. c. Income Income p. c. Adjacency increase % Agriculture 0,90-0,47 0,80-0,51 0,77-0,29 1,06-12,97 0, ,11 6,58 19,17 8,03 18,00 4,63 5,09 17,23 Raw Materials -0,44-0,39 0,75-0,63 0,55 0,03 0,96-11,35 0,31 0 1,80 3,19 10,35 6,33 7,75 0,33 2,94 9,41 Food Processing 0,69-0,51 0,62-0,02 0,93-0,29 0,97-9,44 0, ,60 9,72 18,45 0,44 27,29 5,62 5,48 15,44 Textiles and Leather 0,94-0,77 0,87-0,55 0,82-0,21 0,60-10,26 0, ,30 15,88 27,96 11,53 26,38 4,45 3,70 18,23 Non metallic Minerals 0,53-0,71 0,85-0,14 0,85-0,20 1,07-9,93 0, ,77 16,33 32,08 3,60 32,27 5,20 8,04 21,66 Energy-intensive Products 0,25-0,88 0,90 0,12 0,87-0,22 0,71-5,91 0, ,48 21,10 33,44 2,98 32,45 5,29 5,05 12,05 Other Manufacturing 0,22-0,82 0,84 0,02 0,91-0,05 0,83-6,67 0, ,01 16,87 27,23 0,32 29,70 1,17 5,25 12,10 Metals 0,16-0,89 0,80-0,23 0,94-0,05 0,97-7,51 0,66 0 1,26 14,68 21,27 4,21 25,34 0,84 5,26 11,71 Fabricated Metal Products 0,26-0,83 0,87 0,04 0,80-0,19 1,02-7,75 0, ,58 18,05 30,18 1,02 28,07 4,50 7,19 15,61 Machinery and Equipment 0,18-0,81 0,96 0,40 0,83-0,18 0,59-5,06 0,78 0 1,71 17,87 32,77 8,82 28,44 4,09 3,84 9,52 Electronic Equipment 0,62-0,79 1,06 0,26 0,82-0,19-0,02-7,44 0, ,15 11,78 25,28 4,08 19,76 2,99 0,09 10,03 Transport Equipment 0,42-0,80 1,23 0,17 0,87-0,13 0,78-8,78 0, ,13 13,45 32,53 2,98 23,31 2,32 4,09 13,28 Trade Sevices 0,79-0,11 0,79-0,10 0,82-0,07-0,12-12,92 0, ,44 3,11 34,76 2,87 36,11 1,94 1,00 31,25 Transport and Communication 0,05-0,03 0,82-0,10 0,91 0,12 0,02-12,05 0,91 0 0,89 1,41 53,89 4,14 59,29 5,17 0,23 43,13 Financial Services -0,13-0,24 0,88-0,01 0,87-0,13-0,04-13,03 0,75 0 1,29 5,52 32,05 0,26 31,64 2,99 0,26 26,20 Other Services 0,32-0,27 0,84-0,02 0,81-0,06 0,09-9,87 0, ,75 9,46 46,53 0,80 44,93 2,23 0,93 29,86 Now, turn to the EU dummies. In eleven out of sixteen industries, the dummy has a positive and significant coefficient. Hence, in these sectors, bilateral trade between two countries is higher if these countries are both members of the EU. The dummies measure thus measure the impact of the non-existence of implicit and explicit barriers to trade between countries of the EU. Intuitively clear - think about the CAP policy - the dummies for both Agriculture and Food Processing are among the largest. The dummy for Raw Materials is negative but insignificant; this may be due to oil being intensively traded between EU members and nonmembers alike. Textiles has a high and significant dummy. For Metals. and Machinery&Equipment the EU dummy is insignificant. In the service sectors the dummies are low and/or insignificant (an exception is Trade Services). An explanation is that the internal market has not progressed much so that EU membership does not lead to more intense trade patterns. How to interpret these numbers? For industries with an insignificant dummy, we assume that accession to the internal market has no impact on trade. For other sectors, the dummy is used to calculate the potential trade increase. In particular, we assume that EU membership implies that the dummy would change from zero to one 9

10 for bilateral trade patterns between an EU and the CEECs. Thus, potential trade can be calculated as exp( s ), where s denotes the estimated coefficient for the EU dummy in (1). To illustrate, the coefficient for the EU dummy in Food Processing is equal to 0.69 so that the potential trade is e , i.e. twice as high as the actual trade between CEECs and EU members. The potential trade increase is therefore 99%. The potential increase per sector in bilateral trade between the EU and the CEECs can be used to calculate the aggregate trade increase. To that end, we multiply the existing trade shares of the corresponding sectors with the potential trade increases... Is this increase large? Probably not! First of all, borders tend to be important impediments to trade. For example, a typical Canadian province trades 22 times more with another Canadian province than with a comparable neighbouring US state (see McCallum, 1995 and Helliwell, 1996). For countries that are more different than Canada and the US, a border is likely to matter even more This suggests that a reduction in trade cost by joining the EU (a removal of economic borders) may have substantial effects on bilateral trade. Second, our estimates may still provide a lower bound on the trade effects because joining the EMU is a logical next sept for the CEECs. 7 Frankel and Rose (2000) find that joining a free trade area triples(!) trade and that joining a currency union triples trade once more! Though their study uses a broader sample of countries and, hence, is not entirely comparable, these results are suggestive. Finally, other studies using gravity models have also reported substantial trade effects from EU membership. To translate the potential trade increases due into non-tariff barriers, we first re-calibrate the model so as to replicate trade volume predicted by the gravity model. This yields an alternative set of preference parameters for goods from different countries. Using these alternative parameters, we again re-calibrate a set of alternative prices in order to replicate the actual trade data again. These alternative prices consist of market prices, tariffs and virtual tariffs that can be interpreted as non-tariff barriers associated with non-membership of the internal market. Hence, these non-tariff barriers (NTB s) are comparable to actual tariffs. If we abolish the NTB s in the model, we will arrive at trade levels that correspond to the predictions from the gravity model. The appendix (available upon request) gives a more formal exposition of the calibration of the NTB s. In principle, the predicted trade increases in table 5 capture the impact of both a reduction in formal trade barriers and the accession to the internal market. Table 6 presents the corresponding virtual tariffs that reflect the impact of accession to the internal market. Together with the formal trade barriers, these NTB s reflect the trade effect estimated by our gravity equation. 7 We do not intend to indicate anything about the likeliness or desirability of such a step. 10

11 Ta b l e 6 E s timated s e ctoral non -tariff barriers (a v e r a ge for all countries ) Agriculture 32.6 Nonmetallic minerals 22.2 Energy-intensive Products 9.4 Other Manufacturing 8.6 Electronic Equipment 25.0 Machinery and Equipment 0.0 Fabricated Metal Products 9.8 Other Services 17.1 Food Processing 37.4 Raw materials 0.0 financial Services 0.0 Transport and communication 0.0 Metals 0.0 Trade Services 48.3 Transport Equipment 16.8 Textiles 41.1 source: own calculations 3. 3 F r ee movement of labour Regarding the impact of EU-enlargement on migration, we rely on a study conducted by Boeri et al. (2000). They use historical immigration figures for Germany to estimate migration as a function of wage differentials, employment differentials and a set of dummy variables. By substituting current wages, employment levels and assuming free movement of labour from the first day of accession, the authors compute the likely implications of EU-enlargement on German immigration from the CEECs. These figures are then extrapolated to the other EU-countries on the basis of historical migration patterns between the CEECs and respective EU countries. Assuming accession in 2002 for the ten candidate member states, Boeri et al. predict an inflow of immigrants in the first year after accession towards the EU. This flow gradually declines in subsequent years. In 2030, the stock of migrants in the current EU countries will have grown to 4 million, which is approximately 4% of the total population in the CEECs. A nice feature of the study by Boeri et al. is that it gives an indication of the origin and destination of migrants. For instance, it suggests that 30% of all migrants originate from Poland, 7,5% from Hungary and the other part from the Rest CEECs. These shares depend not only on the size of countries, but also on the incentives for migration, determined by wage levels and employment rates. As the income gaps of Poland and Hungary with the EU are smaller than for the other CEECs, migration shares are somewhat lower for Hungary and Poland than their population sizes might suggest at first glance. From the migrants of the CEECs, 65% will move to Germany, 2,5% to France, 4% to the UK, 1% to the Netherlands, 7,5% to Southern Europe and 20% to the rest of Europe. Of this latter group, approximately 12% of the immigrants will go to Austria. We use these figures in constructing our migration experiment in Worldscan (see table 7). In particular, we simulate the implications of an exogenous migration impulse and do not take into account endogenous feedback effects on migration, e.g. due to wage convergence. Our impulse in Worldscan differs from the migration flows reported by Boeri et al. in two important ways. First, we assume accession in 2004 for Poland and Hungary and 11

12 $ % # " & 2007 for the other CEECs. We therefore adjust the aggregate figures derived by Boeri et al. according to our differentiated accession pattern. Secondly, we evaluate the implications for migration in the year 2020 whereas the estimates by Boeri et al. suggest that migration will continue until Hence, we do not capture the entire migration impulse reported in their study. Thus, we arrive at a total stock of immigrants in the EU of 2.4 million in 2020 (see table 7). This figure is consistent with the consensus estimate reported by Bauer and Zimmerman (2000).On the basis of a literature review and some own calculations, they estimate the migration effect of EU enlargement at around 3 million people after 15 years. Ta b l e 7 Migrants by source and destination (in 1000 persons and % of the population) Poland Hungary Rest CEEC CEECs % 2% 3.5% 2.3% Netherlands Germany Rest-EU EU % 2% 0.3% 0.6% Source: Boeri et al. and own calculations 4. E c o n o m i c i m p a c t o f en larg e m en t We now explore the economic implications of the three experiments with the WorldScan model A ccess i on to th! e inte " r n # al market Macro-economic effects We now explore the implications of accession to the internal market. To that end, we simulate the gradual abolishment of the NTB s presented in table 6. In particular, we reduce the NTB s linearly between 2000 and The channels through which NTBs affect the economies in WorldScan are equivalent to those in case of a reduction in formal trade barriers, i.e. it exerts trade creation and causes changes in the terms-of-trade. There is one important difference between the abolishment of the NTB s and the elimination formal trade barriers analysed in the previous section, however. Whereas the formal barriers are disproportionately imposed by the CEECs relative to the EU, the NTB s refer to the EU and the CEECs alike. The macroeconomic effects of accession to the internal market are presented in Table 8. We see that the CEECs experience a terms-of-trade gain at the expense of a terms-of-trade loss in the EU countries. The terms-oftrade effect is determined by the sectoral trade pattern of countries. Indeed, if a country specializes its export in sectors that are affected substantially by accession to the internal market, the removal of NTB s is likely to 12

13 improve its terms of trade since export prices tend to rise. The results in table 8 suggest that, indeed, the CEECs have a comparative advantage in sectors that are substantially affected by accession to the internal market (such as Textiles and Leather) There is also a substantial difference in the magnitude of the terms-of-trade effects. This is because of the large trade-share of the CEECs that is affected by the NTBs, relative to the trade share for the EU. Compared to the formal trade barriers, the virtual trade barriers are relatively large, especially on the side of the EU. Accordingly, trade creation is more substantial while also the magnitude of the terms-of-trade gain is relatively large. This boosts the value of the return to capital so that capital accumulation in the CEECs expands. Table 8 shows that GDP in 2020 is 10.4% higher in Hungary, 5.1% higher in Poland and 3.4% higher in Rest CEEC. The macroeconomic effects for the EU countries are relatively small due to the small trade share with the CEECs. All EU countries benefit in terms of GDP because of the gains from trade creation. The effects on consumption are typically smaller due to the terms-of-trade losses in the EU. b l e 8 M a cro- e c on om ic Ta f effects o access i o n to the internal market (e ffects relative to the ba s e l i n e in ) volume of GDP volume of consumption terms of trade Average CEEC - Hungary Poland Rest CEEC Average EU - Germany France United Kingdom Netherlands South Europe Rest EU source: WorldScan Trade effects Table 9 presents the trade effects of accession to the internal market. 1. Effecten zijn veel groter dan je mag verwachten op grond van tabel 9. Wat gaat er mis? 2. Vergelijk handelseffect met potentiele handel uit Gravity. 13

14 G Ta b l e 9 C h a nge s in exports a nd imports d u e to access i on to the internal market percentage of D P in ) source/destination EU ROW CEEC world Hungary Poland Rest Eastern Europe s o l u te changes as ( ab source: WorldScan. Sectoral effects At a sectoral level, the changes due to accession to the internal market are much larger. The largest changes take place in the sectors Textiles, and Electronic Equipment. The changes in the sector Transport industries are also very large for Hungary and Rest CEEC, see table

15 Ta b l e 1 0 S ectoral effects of access i on to the internal market (r e l a t i ve changes compared to the bas e l i ne in ) Rest South Rest Hung. Poland CEEC Germ. France UK NL EU EU Agriculture Raw Materials Food Processing Textiles and Leather Non metallic Minerals Energy-intensive Products Other Manufacturing Metals Fabricated Metal Products Machinery and Equipment Electronic Equipment Transport Equipment Trade Services Transport and communication Financial services Other services source: WorldScan 4. 2 M i g r a t i on res p o ns es Macro-economic effects Table 11 reports the simulation results of the migration inflows. It reveals that the migration flow between 2004 and 2020 reduces the population in Hungary and Poland with about 2% in 2020 and more than 3% in Rest CEEC. As imposed by the experiment, Germany attracts most migrants: its population will increase with about 2%. In the Rest EU population size will increase with 1%, while in the inflow in the other countries is much lower. The size of the migration flows determines the economic effects. 15

16 b le Ta 11 E c f onomic effects o migr a t i o n relative changes compared to the ba s e l i n e in GDP per volume of volume of terms of population average wage capita GDP consumption trade Hungary Poland Rest CEEC Germany France United Kingdom Netherlands South Europe Rest EU source: WorldScan results. The results for the other three regions are not depicted because these are negligible. Table 11 shows that wages in the CEECs increase because of the reduced supply of labour in these countries. In particular, since capital is not perfectly mobile across countries, the lower supply of labour will not drive out capital until its rate of return is recovered. Indeed, capital owners bear the burden of lower labour supply in terms of a lower rate of return. The higher capital/labour ratio is associated with increased wages. For the same reason, wages in Germany and Rest EU decrease. In the other EU countries, immigration has a negligible impact on wages. Consequently, GDP per capita rises in the emigration countries and falls in the two main immigration countries. The total volume of GDP, however, drops in all CEECs by about 2%. In Germany it increases by that size while the rise in GDP in the other countries is somewhat smaller. The effects on consumption are smaller than those on GDP because of changes in trade patterns. In particular, the lower wages in Germany and Rest EU have a downward pressure on producer prices and the terms of trade, while higher wages in the CEECs renders the effects on the terms of trade effect positive. Accordingly, the trade deficit of the CEECs vis a vis the EU increases, thereby mitigating the adverse consumption effects in the CEECs. The small effects of migration on wages is consistent with empirical evidence. In particular, Bauer and Zimmermann (2000) present a survey of the literature and conclude that immigrants have only a negligible negative impact on native wages. Indeed, the elasticity is somewhere in the range between zero and 0.3, i.e. a 1%-point increase in the share of migrants in the total population reduces 16

17 wages of natives by 0.3% at the maximum. 8 Our simulation results are within this range, thus confirming the modest implications of immigration for native wages in the EU. 5. C o n c l us ïo ns The analysis leads to the following sets of conclusions: On methodology Modelling accession to the internal market as a uniform trade-cost reduction is inappropriate, the shock differs for different industries. What is EU enlargement Accession to the internal market if by far the most important effect of EU enlargement. The impact of free movement of labour is small. On winners and losers On the level of countries the enlargement of the EU knows only winners. Different industries are affected differently; it is important to acknowledge that on the industry level losers are inevitable. A corollary of the first two conclusions is that redistribution of the gains might be necessary (especially when specific factors are important). The accession countries gains are large (up to 1% a year). 8 The elasticity generally differs between low-skilled and high-skilled wages. In particular, migration typically reduces low-skilled wages of natives because migrants usually enter low-skilled jobs. Many studies report positive implictions for native wages of the high-skilled. 17

18 R e f e r en c es Baldwin, R. E., J.F. Francois, and R. Portes, 1997, Costs and Benefits of Eastern Enlargement: The Impact on the EU and Central Europe, Economic Policy: A European Forum, 0(24), Bergstrand, J. H., 1989, The Generalized Gravity Equation, Monopolistic Competition, and the Factor- Proportions Theory of International Trade, Review of Economics and Statistics, 71, CPB, 1999, WorldScan: the Core version, The Hague. Deardorff, A.V., 1998, Determinants of Bilateral Trade: Does Gravity Work in a Neoclassical World? in J. A. Frankel, Ed., The Regionalization of the World Economy, University of Chicago Press. Helliwell, J., 1996, Do National Borders Matter for Quebec s Trade? Canadian Journal of Economics, 29, Helpman, E and P. Krugman, 1985, Market Structure and Foreign Trade, The MIT Press. McCallum, J., 1995, National Borders Matter: Canada-U.S. Regional Trade Patterns, American Economic Review, 85, McDougall, R.A., A. Elbehri, and T.P. Truong, 1998, Global Trade Assistance and Protection: The GTAP 4 data base, Center for Global Trade Analysis, Purdue University. Nahuis, R., 2001, One size fits all? Accession to the internal market, an industry level assessment of EU enlargement, mimeo, CPB. Purdue, 2001, The GTAP 5 data base (prerelease), Center for Global Trade Analysis, Purdue University. Tinbergen, J., 1962, Shaping the World Economy - Suggestions for an International Economic Policy, The Twentieth Century Fund. United Nations, 1998, A ppendix C a l i b r a t i on To translate the estimated potential trade volumes into non-tariff barriers we actually take four steps (one is not essential). The starting point is the demand equation following from the Armington demand system. The first step in the calibration, is the standard procedure. This gives the parameters with a superscript C1 (calibration step 1). For each industry, we drop the industry subscript, we get: J (2) where x is the trade from country i to country j. Preferences are given by s. X is total demand for the good produced in the industry. Prices (p) are determined by another procedure in the calibration procedure, therefore the bar indicates them as exogenous. The price-index P is a function of the (given) prices. Formal tariffs are denoted as t. Superscript D denotes data. Hence the preference parameter is used to calibrate the model such that it replicates the bi-lateral trade data. Second, we calculate the preferences required to produce the cet. par. trade volume predicted by the gravity model: 18

19 J (3) where, the potential trade increase. X j is calculated such that it is consistent with the new bi-lateral trade flows. This gives the preference parameter s C2 that are consistent with the potential trade flows that would materialise if there where no non-tariff barriers. In the last step, we use this new preference parameter to calculate the non-tariff barriers. We re-calibrate the model to replicate the actual data again. For this we adjust the prices in the Armington demand system; we introduce a that is the non-tariff barrier. This gives a set of prices and parameters that re-produces the data and will produce the estimated trade flows if the tariff and nontariff barriers are reduced to zero. J ( 4) Not essential, an intermediate step is taken; step three. To determine the price index in the last calibration step(4) -- note that obviously only relative prices matter -- we use the fact that on domestic goods the NTB is equal to zero by definition, hence J ( 5) We use this to pin down the nominal prices. To be precise equations (3) and equation (4) make a system of equations that is to be solved simultaneously. 19