An Economic Assessment of the Accession of the Central and Eastern European Countries to the EU Single Market

Transcription

1 An Eonomi Assessment of the Aession of the Central and Eastern European Countries to the EU Single Market T H Edwards CSGR Working Paper No. 128/04 January 2004 Centre for the Study of Globalisation and Regionalisation (CSGR, University of Warwik, Coventry, CV4 7AL, United Kingdom. URL:

2 An Eonomi Assessment of the Aession of the Central and Eastern European Countries to the EU Single Market T.Huw Edwards 12 Centre for the Study of Globalisation and Regionalisation, University of Warwik CSGR Working Paper No. 128/04 January 2004 Abstrat In 2004 the European Union is due to inorporate 10 new members, mostly from the Central and Eastern European Countries (CECs. Trade between the EU and CECs urrently falls well short of that between EU ountries, and if we assume this pattern reflets both tariffs and a resoure ost due to regulatory differenes, then 1997 trade patterns would imply suh osts are 7-15% on trade between the EU and CECs. Elimination/harmonisation of remaining tariffs is likely to bring small welfare gains to new entrants. By ontrast entry to the Single Market looks far more signifiant: after both tariff union and entry to the Single Market total trade volumes between the EU and CECs ould rise by % (muh more in some ommodities, while welfare gains in the CECs ould be of the order of %, larger than the previous two studies have suggested. Welfare gains within the EU are around 0.4% of GDP, with all regions gaining but Germany gaining most. Gains are greater where apital is fully mobile. Keywords: European Union, Computable General Equilibrium, Transition, Trade JEL Classifiations: F12, F15, F17 Address for orrespondene T H Edwards CSGR, University of Warwik Coventry, CV4 7AL, UK T.H.Edwards@warwik.a.uk 1 This work was funded by the European Commission as part of the Framework 5 Projet on Eonomi and Politial Re-Integration in an Enlarged Europe (EPRIEE. 2 The author wishes to thank Paul Brenton at the World Bank and David Barlow and Roxana Radulesu, University of Newastle-upon-Tyne, for help in preparing this paper. Any errors are my own. 2

3 Non-Tehnial Summary This paper forms part of the researh projet on the Eonomi and Politial Re-Integration in an Enlarged Europe, funded by the European Commission. The paper is a numerial investigation assessing the possible benefits to the Central and Eastern European Countries (CECs, and to existing EU members, of an enlargement of the European Union. For data reason, the enlargement onsidered is aession of Poland, Hungary and an Other CEC region (Czeh Republi Slovakia, Slovenia, Romania, Bulgaria. The latter two are not inluded in the urrent aession round, but head the list of andidates for future enlargement. From a trade perspetive, muh of the integration proess has already taken plae. The Europe Agreements of 1997 were very important in this regard, removing tariffs and quotas on all areas of trade between the EU and CECs exepting agriulture. The tariff reform that remains to be done for aession ountries is therefore 1 removal of agriultural tariffs and 2 adoption of the Common External Tariff for trade with third ountries outside the CEC region. Arguably more important, however, is the general aession of the CECs to the EU s Single Market. This is a ompliated series of agreements dealing with the huge raft of regulations and standards whih ountries impose on industry for means of health, safety and onsumer protetion. There is onsiderable disagreement as to the real signifiane of suh regulations as barriers to trade. On the one hand, ertain osts to importers, suh as testing, ertifiation and relabelling of goods, are fairly easy to establish: on average these probably aount for an addition of between 1 and 2 per ent to the ost of imported goods into the European Union. However, a substantial shool of thought suggests that regulations and standards may in pratie be imposing far higher osts than this on ompanies engaging in trade, by foring them to alter the speifiations of the goods they produe in order to enter new markets. 3 Some eonomists, suh as Baldwin (2001 and Wallner (1998 arry this argument further, and suggest that domesti health, safety, environmental and other regulations may be 3 For disussion of this, see Maskus and Wilson (2001, whih ontains a number of papers linked to an ongoing World Bank study on regulatory barriers to trade. 3

4 systematially distorted by ountries for protetionist purposes. This is, however, ontroversial. 4 If we aept for the moment, however, the line that national regulations are impinging seriously upon international trade whether or not this is being done for protetionist motives then this suggests that international agreements on produt standards are entral towards opening markets for trade and ompetition. This ertainly was a motive for the institution of the Single Market by the European Union, whih relies primarily upon mutual reognition agreements to remove pereived barriers to trade among EU member states. 5 Membership of the Single Market is potentially of partiular value to the CEC states, as these are mostly small eonomies whih have inherited relatively monopolisti industrial strutures, and whih an potentially gain greatly from integration into a far more ompetitive single European market. 6 The issue addressed in this paper is to quantify suh potential benefits to the aession states, as well as the effet of entry upon existing EU members. This is done by means of a multiountry, multi-industry omputable general equilibrium (CGE simulation model. In this regard, I am following in the footsteps of earlier studies by Baldwin, Franois and Portes (1997 and LeJour et al (2001. However, the urrent model differs from the previous studies in a ouple of respets. Firstly, unlike LeJour et al, the urrent study utilises an imperfetlyompetitive eonomi model, whih potentially takes aount of the role of trade in enhaning ompetition within the hitherto sheltered CEC eonomies. Seondly, there is the issue of how the urrent regulatory barriers are estimated. As has already been explained, suh barriers are not easy to quantify. Baldwin et al (1997 simply applied an ad ho assumption that an additional resoure ost of 10 per ent was added to all goods traded between EU member and non-member ountries, due to the need to omply with different regulatory standards. By ontrast, LeJour et al (2001 inferred the size of regulatory barriers by omparing atual trade with predited trade using a gravity model. This is an ad ho eonomi model whih predits how muh two ountries will trade with 4 See Edwards, 2003 (1, for a septial view of this argument. 5 Mutual reognition is an agreement by whih ountry A agrees to aept without further testing any goods from ountry B whih meet ountry B s domesti standards, in return for ountry B aepting all goods from ountry A whih meet A s own regulatory standards. 6 For the basi eonomi arguments underlying this, see Baldwin and Venables (

5 one another from the size of their eonomies and the distane between them (with a few other adjustments for tariffs or ommon membership of eonomi blos. To the extent that trade between the EU and the CECs in 1997 fell short of what the gravity model would predit, this was assumed to be explained by unquantified regulatory barriers. The approah in this paper is based on similar assumptions to LeJour et al (2001. However, the general equilibrium model used here is somewhat more sophistiated, being based upon the variety goods approah of Dixit and Stiglitz (1978. It is assumed that a larger eonomy will be produing a greater variety of goods, and sine onsumers in all ountries like variety it will, onsequently, apture a larger share of export markets, other things equal, than a smaller eonomy. This produes a very similar basi model to the gravity model, but whih is diretly derived on theory and an be alibrated diretly from the same data set used for the general equilibrium simulations, one data for tariffs and transport osts and a few assumptions about prodution osts have been inorporated (see Appendix I. Like LeJour et al (2001 I find that there is a onsiderable shortfall in trade at present between the EU and the aession states, and I then assume that 1 this trade shortfall represents the effets of a real resoure ost of regulatory differenes and 2 this will be retified one the ountries join the Single Market. On these assumptions, the model simulations in this paper uphold the results of the two earlier studies: namely that aession to the Single Market ould greatly boost trade between the EU and the CECs, that this will lead to greater trade in both diretions in many industries and that the effets of inreased ompetition, as well as of better utilisation of omparative advantage, lead to sizeable welfare gains for the aession states, while having no measurable detrimental effet on existing EU eonomies. The results of these simulations are summarised in Tables 7 to 11. AGGREGATE Trade volumes between the EU and aession states ould potentially inrease by per ent, with Poland experiening the largest inreases. These trade volume inreases are in both diretions, though given the relatively small sale of the CECs ompared to the existing EU the signifiane is far greater for the former group. Table 9 shows output hanges: light manufaturing grows by over 20 per ent following aession in Poland, Hungary and the Other CEC region, while Hungarian heavy manufaturing also grows by over 20 per ent. Table 7 indiates that GDP is likely to inrease in all regions of the CECs and EU following 5

6 aession, with the main effets oming from entry to the Single Market rather than harmonisation of tariffs. Polish GDP ould gain by 20 per ent in this simulation, while Hungary and the Other CECs gain somewhat less. Table 10 indiates that CEC aession is likely to lead to substantial upward pressure in pries in those regions (refleting higher fator demand, while there is a small downward pressure on pries in the existing EU regions. All fator pries rise sharply in the aession ountries (Table 11, while they fall slightly in the existing EU. The net result is a rise of perhaps per ent in real wages in the aession ountries, with little net hange in the existing EU. The overall piture painted by these simulations is a highly optimisti one for the aession ountries, indiating large potential gains in trade and real inomes, at virtually no ost to the existing EU members. However, it is worth remembering that this model has been based upon an assumption that there are high regulatory barriers to trade before aession, and that these are refleted in real trading osts. These assumptions may not be aurate, and onsequently the figures in this paper may be regarded as painting a relatively optimisti piture of the potential gains from enlargement. A good deal of further work is required to examine in detail the real qualitative and quantitative nature of regulatory barriers to trade and the effets of the Single Market initiative: this is a subjet to whih eonomists have been slow to turn attention. 6

7 1. Introdution This paper represents a study modelling the osts and benefits of EU enlargement into some or all of the Central and Eastern European Countries (heneforth referred to as CECs. In partiular, I look at the effets of aession to the EU s Single Market, with its harmonisation and mutual reognition agreements whih eliminate many non-tariff barriers to trade between EU members. This is investigated using a omputable general equilibrium model based upon an imperfetly ompetitive variety goods model using the Dixit-Stiglitz framework, whih has been shown (eg Baldwin and Venables, 1995 to generate muh more signifiant potential welfare gains from trade than the more traditional Armington models. The layout of the paper is as follows. Setion 2 ontains an in-depth disussion of the literature to date, outlining the methodology and assumptions of three major previous studies of EU aession by Brown et al (1995, Baldwin et al (1997 and, LeJour et al (2001. Setion 3 summarises the modelling approah and data used in this paper: in partiular the novel model-onsistent alibration proedure to estimate ountry bias effets. In setion 4 I outline the results of simulations of simulations of aessions of Poland, Hungary and an Other CEC Region into the UK, and of aessions of the rest of Central and Eastern Europe. 7 The simulations are based upon an intermediate-term Dixit-Stiglitz model (whih inorporates sale and variety effets, but does not allow firm numbers to hange, whih would happen in a longer-term model. The final setion disusses the findings of these simulations. If we assume that trade between the EU and aession states moves into line with that between existing EU members, then the effets on the aession states, partiularly Poland, ould be very large, with sizeable inreases in trade with the EU and inome gains of up to 20 per ent. However, the assumptions upon whih this onlusion is based are rather ontentious: in partiular our urrent theoretial understanding of the working of mutual reognition agreements and of the relationship between non-tariff barriers and trade volumes is rather skethy. 7 Due to data limitations, I am unable to arry out simulations on the preise aession list of The Other CEC region omprises the Czeh Republi Slovakia, Slovenia, Romania and Bulgaria. The latter two are not on the 2004 EU aession list, whereas the 3 Balti States, as well as Cyprus and Malta are. 7

8 2. Bakground and Literature Survey While early studies of the eonomis of EU enlargement into the former Soviet blo (eg Brown et al, 1995 onentrated on the effets of removal of tariffs and formal non-tariff barriers (NTBs, suh issues are no longer at the entre of the enlargement debate, exept in agriulture, sine the Europe Agreements have removed most formal barriers. Nevertheless, later studies (eg Baldwin et al, 1997, or LeJour et al, 2001 assume EU membership for CECs would still have large effets upon trade with the EU, due to aession to the single market. It is known that trade between ountries falls well short of trade within a single ountry, 8 and that signifiant ountry bias also exists for members of trade blos vis-àvis non-member ountries. Some eonomists, notably Baldwin, have attributed muh of this to differenes in produt standards, trade law and other informal barriers to trade, whih supposedly onstitute regulatory protetion, 9 in many ases larger than formal trade barriers. The existene and sale of suh non-tariff barriers are generally imputed from the empirial gravity modelling literature. It has long been believed that trade between two ountries is usually roughly proportional, eteris paribus, to the produt of size of the eonomies, orreted for distane and formal trade barriers. However, there are residual differenes, whih we shall heneforth all residual ountry bias. Most notable is home bias: the preferene of onsumers in one ountry for produe of that ountry, rather than any other, whih seems empirially to be very strong, even when a ountry joins the single market. Beyond that is a lesser, but still signifiant bias towards produe of other ountries within the same trade blo. These biases are usually piked up in gravity equations of the form: X (1 g, g 1g. 2s TE g, D s, ijs EU 2g d 1s Y 2s y 3s Y 4s y d ds Dd 1s TM g, 8 MCallum (1995 found a national border effet reduing trade between Canadian provines and neighbouring US states by a fator of 22 ompared to between Canadian provines. 9 For the arguments on regulatory protetion, see, eg, Baldwin, 2001, Maskus and Wilson, 2001 and Wallner However, for ounter-arguments see Moenius, 1999 and Edwards, 2003 (1. 8

9 where all variables are in logs. X g, is exports from ountry to in industry g, Y is GDP and y is GDP per apita, d is distane between apitals of the ountries and. Dd is a set of dummies for border effets, with D EU set to 1 if both and are EU members, otherwise set to zero. T g, is the import tariff on imports of s from ountry to ountry. T g, is the export tariff levied by ountry on ountry. LeJour et al (2001 estimated the following residual ountry bias dummies for EU membership: Table 1: EU dummies in gravity equations (LeJour et al, 2001 EU dummy Trade inrease % Setor Agriulture 1.25 * 249 Raw materials -0.1 Food proessing 0.66 * 94 Textiles and leather 0.85 * 134 Non-metalli minerals 0.73 * 107 Energy-intensive produts 0.13 Other manufaturing 0.08 Metals -0.1 Fabriated metal produts 0.44 * 56 Mahinery and equipment 0.31 * 37 Eletroni equipment 0.58 * 79 Transport equipment 0.66 * 94 Trade servies 0.76 * 113 Transport and ommuniation 0.03 Finanial servie Other servies 0.27 * 31 The rather strong assumption made by both the LeJour et al and the Baldwin et al 10 studies is that this residual border effet of EU membership orresponds to an unspeified set of trade osts (henforth referred to here as the residual border trade ost, whose assumed removal or redution by joining the single market is a sizeable soure of potential eonomi benefit for the CECs. 11 However, it must be stressed that there are plenty of alternative explanations 10 In their earlier paper, Baldwin et al (1997 assume rather arbitrarily that joining the single market would mean a redution in trade osts (real terms osts, assumed to be measured as an ieberg ost a loss of value of all goods traded between exporting ountry and ountry of fration of 10 per ent aross the board. 11 The eonomi interpretation LeJour et al attah to these dummies stems from the well-known link (following Bergstrand, 1989 with a general equilibrium trade framework based upon an imperfetly ompetitive framework (though in fat LeJour et al use an Armington rather than Dixit-Stiglitz model for simulations. Though Deardorff (1998 points out that a fritionless Heksher-Ohlin model and an Armington model (whih may be the result of ountries speialising on sub-ategories within the measured goods ategories of a H-O model will also produe gravity relationships. 9

10 other than ieberg-style trade osts for residual border effets in empirial models (suh as differenes in onsumer preferenes ompanies produing goods suited, say, to Frenh rather than British taste would be more likely to set up in Frane rather than the UK or historial searh-related fators 12. The two above studies both found signifiant welfare and trade gains from EU enlargement. Baldwin et al s (1997 simulation results, based on an assumed 10% ieberg ost on trade between the EU and CECs, are shown in Table 2 (below: Table 2: Baldwin et al (1997 simulation results for EU enlargement. Real inome' hanges from EU enlargement (% hange on base. Conservative ase' Less onservative' ase trade gains only with risk premium redued CEC EU EFTA Ex-USSR Le Jour et al (2001 also find substantial benefits for the aession ountries, partiularly Poland, though not as sizeable as in Baldwin et al (1997. This is not surprising sine LeJour et al use an Armington model, whih does not model all of the benefits (partiularly those linked to inreased ompetition whih a Dixit-Stiglitz model aptures. Both studies are agreed that enlargement involves few osts for existing EU members, though LeJour et al imply Frane may have lost slightly from the 1997 tariff hanges. 12 See e.g. Rauh, 1997, Edwards, 2003 (2. 10

11 Table 3: LeJour et al (2001 simulation results of EU enlargement Volume of GDP effets of: Removal of 1997 bilateral Aession to the tariffs and adoption of Internal market Common External Tariff CEEC Hungary Poland CEEC EU Germany Frane United Kingdom 0 0 Netherlands South Europe Rest EU Third Countries 0 0 Rest OECD 0 0 Former Soviet Union 0 0 ROW Methodology Despite the doubts expressed above, this paper proeeds, upon the assumption that residual border effets do indeed reflet residual trade osts, as in the Baldwin et al and LeJour et al papers. Nevertheless, the approah here differs somewhat. First of all, in this paper I derive residual border effets by diret alibration of a theoretial Dixit-Stiglitz model, whih is then used for simulation. Unlike previous studies the alibration and simulation models are fully onsistent. Seondly, the alibration exerise alibrates residual border effets for imports and exports between eah pair of ountries (though averages are then onstruted for inter-eu trade using model-onsistent CES funtions for aggregation. Sine gravity studies typially use a muh more parsimonious set of dummies (eg just a home dummy and a seond dummy if both ountries are EU members they are effetively onstraining many residual border effets to be equal yet just by omparing two different alibrations we show that the hoies of whih prior restritions to make on border effets has a potentially very large effet on the putative impat of EU membership on ountries trade patterns. This is perhaps a topi not suffiiently explored in the gravity literature. 11

12 The third differene from standard gravity approahes is that more speifi aount is taken of the importane of relative output pries. While we do not know exatly what the relative osts of prodution in different ountries are (partiularly when quality is orreted for we an alibrate for revealed omparative osts one a ertain set of restritive assumptions has been made about border effets. However, the interrelationship between alibrated residual border effets and revealed omparative advantage is a lose one, and different restritions on border effets will greatly affet our piture of the underlying ompetitiveness of the CEECs in different industries. 3.1 Methodology for estimating border osts. Appendix I shows the derivation of the equations for estimating omparative osts and assumed border osts of trade in this study, and how they relate to more orthodox gravity studies. Basially, I estimate the border ieberg ost on trade in good g from ountry to ountry g, using the equation: (2 (1 g,.(1 g,, E g,.(1 g, 1.(1 t g, E,.(1 g,, 1.(1 t g,, H g,. H g, 1/(1 Where E g, is export value of g from to, is transport ost, t is the import duty rate, H g, is onsumption of home-produed g in ountry and is the onstant elastiity of substitution. Data on sales pries of eah ountry s goods at home, P, may not be easy to ome by (and anyway, pries may not be diretly omparable if quality varies. For that reason, it may be better to use a revealed omparative advantage approah, and atually inorporate alibration of P into the general alibration. This means we are alibrating the model for both pries and residual border osts. 3.2 Calibration For alibration, I start with equation (2. In the absene of better data, it may well be most sensible to assume initially average firm size is the same aross ountries: T =T =T for all. 12

13 If the only unknowns are the residual border trade osts,, then if we assume = 0 if = and 0 if we need only fit for the (probable non-zero elements of, where. To model the effets of the single market in terms of these border osts, I make the following assumptions about the struture of (non-tariff and non-transport border osts: g = home bias in ountry. This ost is applied to import of good g from any other ountry into (regardless of whether and are both members of the EU s Single Market. g,cec,eu = additional ost for imports from CEC ountries to EU members (ompared to imports from other EU members. This means that the total border ost for importing from a CEC ountry to an EU ountry is: g,, = g + g,cec,eu g,eu,cec = additional ost for importing from the EU to the CEC. The total border ost for imports from the EU to the CEC is therefore: g, = g, + g,eu,cec g,row,eu additional osts for imports from the rest of the World to either the CEC or EU ountries. To alibrate, I assume I,CEC,EU = I,EU,CEC = I. It is this ost on trade (whih I assume to be the same in both diretions between the EU and CEC whih is removed one the CEC ountry joins the Single Market. 3.3 The Model for simulations Simulations are arried out using a multi-ountry stati omputable general equilibrium (CGE model. Goods are produed using a Cobb-Douglas aggregate of intermediate inputs and 4 primary fators: unskilled labour, skilled labour, apital and land. Land is fixed setorally. Both types of labour are mobile between setors, but not between ountries. For 13

14 apital, I investigate two variants, one where it is fixed in total within a ountry, and one where it is internationally mobile 13. Intermediate inputs and final onsumption goods are CES aggregates of home prodution and imports from various soures. The elastiity of substitution between different soures of a good is set at 4 in all setors. There are also transport osts (modelled as ieberg osts, ieberg unspeified trade osts (see above and tariffs, as well as taxes/subsidies on output and use of a ommodity. Firms both at home and abroad are imperfetly ompetitive (ompeting with a Dixit-Stiglitz symmetrial CES funtion, and harge profit markups dependent on their market shares. For omputational reasons, the number of firms per industry and per ountry is however assumed to be fixed in simulations these an therefore be seen as intermediate-term variety goods simulations. The top level of the onsumption funtion, where different industries produts are aggregated, uses a Cobb-Douglas struture. 3.4 Data I use the GTAP version 5 database. This database has harmonised trade and input-output data for regions aross the world in GTAP potentially has a large number of goods and regions, so for pratial purposes I aggregate data into 8 goods and 10 regions, hosen for their relevane to the issue of enlargement. Goods: AG agriulture, forestry and fishing OP other primary FP food proessing IS iron and steel TX textiles MH heavy manufaturing ML light manufaturing SV servies. Regions*: PLD Poland HUN Hungary 13 In the latter ase, apital rents are equated aross the world at RBW. A ountry will then pay a rent at this rate to foreigners if it imports apital. This assumption, whih follows Fehr et al avoids some of the problems Rodrik notes in the Baldwin et al model s treatment of hanging apital stoks. 14

15 OCEC Other CECs (Cz Rep, Slovakia, Slovenia, Romania, Bulgaria UK United Kingdom GER Germany OEUN Other EU Northern OEUS Other EU Southern (Italy, Spain, Portugal, Greee FSU Former Soviet Union ODX Other OECD exluding EU and CECs LDC rest of the world (mostly less developed ountries *note GTAP version 5 has only 3 CEC regions. For trade and protetion I use 4 prinipal data series from GTAP for these ountries and regions: VXMD exports at market (ie domesti pries, VXWD exports at world pries, VIWS imports at world pries, VIMS imports at market pries (ie sales pries in the importing ountry before indiret tax. The differene between VXWD and VIWS is taken to be the transport ost margin. VXWD VXMD is a value for net export tax/subsidy, and the GTAP estimates of the tariff equivalent of some quantitative trade restritions whose revenue arues to the exporting ountry. VIMS VIWS is the value for net import tax/subsidy and the tariff equivalent of remaining NTBs. Corretion is made for some data errors in the GTAP Version 5. In partiular, I have removed tariffs on trade between the EU and CECs other than in agriulture and food proessing, as these had been abolished under the Europe Agreements. 4. Results of the alibrations for border osts Table 4 shows the formal trade barriers (tariffs and tariff equivalent of NTBs in existene between the EU and CECs in These are CES weighted averages over the various EU omponent regions (UK, GER and OEU. As an be seen, imports from the CECs into the EU faed sizeable barriers in agriulture and food proessing, but barriers elsewhere had been removed by 1997 under the Europe Agreements. 15

16 Table 4: Net formal trade barriers (tariff equivalent INDUSTRY OCEC EU INTO HUN INTO EU INTO PLD EU INTO PLD INTO EU OCEC EU HUN INTO EU AG OP FP TEX IS MH ML SV However, even when ountry size, transport osts and these formal trade barriers are taken into aount, there is still a onsiderable shortfall in imports ompared to domesti produe in all ases: our model attributes this home bias to an ieberg ost of trade, g,. Table 5: Calibrated relative prodution pries and home/ountry bias oeffiients, POLAND INDUSTRY RELATIVE PLD PRICE INTER-EU HOME BIAS EU V PLD PLDC V EU AG OP FP TEX IS MH ML SV HUNGARY INDUSTRY RELATIVE HUN PRICE INTER-EU HOME BIAS HUN V CEC HUN V EU AG OP FP TEX IS MH ML SV OTHER CECs INDUSTRY RELATIVE OCEC PRICE INTER-EU HOME BIAS EU V OCEC OCEC V EU AG OP FP TEX IS

17 MH ML SV Table 5 (above shows the alibrated omparative osts and ountry bias based on the alibration assumptions in this paper. In this ase, average exess EU bias against CEC goods has been set the same as average CEC bias against EU goods. This alibration suggests the CECs are low-ost produers ompared to the EU in almost all industries, espeially servies 14, agriulture, and light and heavy manufatures. Hungary is low-ost in textiles, while the OCEC region is high-ost in iron and steel. The average ieberg osts of trade in both diretions varies from slightly negative (for Polish food proessing only to around 15% for Polish manufatures, 10-13% for other CEC manufatures and 9-14 per ent for Hungarian manufatures. For agriulture they are around 7-10%. 4.1 Gravity equivalent. It is also possible to onvert the ieberg trade osts into equivalent gravity dummies (- 1.ln(1-. The extra dummies for imports from the EU into CECs and from CECs into the EU (whih are both zero or negative in almost all ases are as follows for alibration 3: Table 6: Gravity dummy equivalents of alibrated residual border effets. INDUSTRY OCEC INTO EU EU INTO OCEC HUN INTO EU EU INTO HUN PLD INTO EU EU INTO PLD AG OP FP TEX IS #MH ML SV The dummies for trade between the EU and CEC are broadly of a similar order of magnitude to those found by LeJour et al s (2001 gravity model study, whih estimated an EU trade dummy of 1.25 for muh of agriulture and around 0.7 for most industrial setors Comparative osts in servies would, of ourse, be expeted to be lower in poorer ountries (see Balassa (1962. However, it seems that, at least for Poland, the low relative osts apply to all setors. Only for the Other CEC region does there seem to be lear evidene supporting the Balassa-Samuelson relationship. 15 The reversal of signs in LeJour et al s formulation ompared to mine is just due to the set-up of the model. 17

18 4.2 Enlargement simulations. The simulation runs are arried out on the CGE model, assuming the number of firms per setor in eah ountry does not vary. The welfare effets are probably smaller than would be expeted in a fully long-run model where sale and variety effets of altering firm numbers were inluded. Table 7 (below shows the effets on onsumer welfare in eah region resulting from 1 ustoms union (the removal of the remaining tariffs on agriulture and foodstuffs between the EU and CEC regions and harmonisation of the CEC s external tariffs with those of the EU and 2 assumed abolition of ieberg unspeified trading osts I when ountries join the EU single market. These simulations are arried out for ases where apital is immobile between ountries and where it is assumed to be mobile. Customs union has only small simulated welfare effets, though these generally benefit the aession states by 0-2 ½% while having no signifiant effet on existing EU members. The former effet is not surprising given the fat that most tariffs have already been abolished, while the latter reflets the small size of the CEC eonomies relative to the existing EU. Under 2 the CEC trade shares with the EU, and the EU trade shares with the CEC are inreased to reflet the supposed removal of trade osts when the CEC ountries join the single market. Sine it is assumed these osts are real resoure osts, it is possible in this ase for all ountries to gain, and this does indeed seem to be the ase. The biggest benefiiaries are the CEC ountries, where welfare rises by 10-20% ompared to 1997 base. Gains to the existing EU members are small, typially around ½%. While Germany gains most, even the poorer EU ountries in the South experiene gains of 0.4%, so that the benefits of expansion of trade outweigh the heap-wage ompetition effets even for these ountries. And the Former Soviet Union and LDCs also see small welfare gains, so that trade diversion effets are outweighed for them by the effets of the overall expansion of the EU and CEC eonomies. 18

19 Table 7: Summary of results hange on 1997 base, alulated onsumer utility. 1. EU-CEC ustoms union 2. CEC trade shares shift in line with intra-eu trade a National b Capital a National b Capital apital mobile apital mobile stoks fixed internation stoks internationally ally fixed Poland 1.87% 2.44% 15.27% 19.39% Hungary 0.17% 0.17% 14.62% 17.56% Other CEC 1.03% 1.21% 11.46% 13.25% UK -0.01% 0.00% 0.16% 0.14% Germany 0.01% 0.00% 0.64% 0.71% Other EU North 0.00% 0.00% 0.42% 0.44% Other EU South 0.01% 0.00% 0.37% 0.38% EU total 0.01% 0.00% 0.42% 0.44% Europe total 0.06% 0.06% 0.94% 1.09% Former Soviet Union 0.07% 0.08% 0.04% 0.09% Other OECD 0.00% 0.00% 0.00% 0.00% LDCs 0.00% -0.01% 0.03% 0.01% Global total 0.02% 0.02% 0.26% 0.29% Table 8 shows the hange in trade volumes: these are typially of the order % between the EU ountries and CECs on aession. Table 8: hanges in trade volumes with trade share shifts and mobile apital assumed. Total trade volumes Before After % hange Pld to EU % Hun to EU % OCEC to EU % EU to Pld % EU to Hun % EU to OCEC % Table 9, whih summarises hanges in output by industry shows that gains in output are spread widely aross all industries in the CEC region, though the biggest gains are to agriulture, food produts and manufatures. Within the EU there appear to be few losers, though agriulture and heavy manufatures deline marginally in the UK. 19

20 Table 9: simulated hange in output by ountry and industry ompared to 1997 base. INDUSTRY AG OP FP TEX IS MH ML SV PLD % 19.52% 10.15% 9.40% 16.51% 26.28% 13.25% % HUN % 9.08% 18.63% 18.36% 12.80% 26.61% 24.15% 11.76% OCEC % 0.85% 18.54% 8.54% 7.75% 12.49% 23.10% 7.33% UK -0.03% 0.25% 0.07% 0.45% 0.67% -0.03% 0.03% 0.07% GER 0.95% 0.68% 0.72% 2.03% 1.41% 0.90% 0.86% 0.47% OEUN 0.41% 0.96% 0.49% 1.02% 1.35% 0.68% 0.37% 0.32% OEUS 0.59% 0.33% 0.30% 0.62% 0.97% 0.16% -0.11% 0.21% FSU 0.76% 0.36% 1.23% -0.22% -0.89% -0.15% -0.15% 0.01% ODX -0.07% -0.07% -0.02% 0.10% -0.03% -0.13% -0.22% 0.00% LDC -0.02% -0.03% 0.05% -0.15% 0.15% 0.03% -0.06% 0.01% Table 10 shows that output pries in the EU generally fall as a result of the saving in osts of inputs (the unskilled wage in Germany is set to 1 in this model, to at as a numeraire. However in Poland output pries generally rise (and the same is true to a lesser degree of some setors in other parts of the CEC region as pries rise towards Western European levels. Table 10: simulated hange in output prie by ountry and industry ompared to 1997 base AG AG OP FP TEX IS MH ML SV PLD 5.20% 3.39% 6.29% 1.07% 3.13% 4.57% 5.44% 3.89% HUN -2.74% 7.07% -3.07% -1.31% 2.32% 3.26% 1.01% 1.42% OCEC 2.71% 1.01% 3.01% -0.33% -0.25% 1.28% 0.99% 2.22% UK -0.79% -0.49% -0.85% -0.71% -0.69% -0.72% -0.79% -0.55% GER -0.82% -0.40% -0.82% -0.66% -0.53% -0.91% -0.92% -0.49% OEUN -0.78% -0.36% -0.88% -0.76% -0.75% -0.91% -0.90% -0.51% OEUS -0.67% -0.55% -0.89% -0.77% -0.58% -0.80% -0.88% -0.51% FSU -0.10% -0.12% 0.09% -0.31% -0.20% -0.15% -0.23% -0.18% ODX -0.48% -0.46% -0.52% -0.57% -0.59% -0.54% -0.54% -0.47% LDC -0.47% -0.44% -0.55% -0.51% -0.61% -0.57% -0.60% -0.46% Table 11, summarising hanges in fator returns, indiates that relative skilled/unskilled wages do not hange greatly in any ountry, though there are sizeable gains to both types of labour in Poland in partiular. The wages quoted are all relative to the German unskilled wage: the slight wage falls in some EU ountries are only relative to this (onsumer goods pries fall more sharply. The lak of distributional hanges between types of labour may partly be beause of the Cobb-Douglas prodution funtion struture, and partly beause the 20

21 presene of a fixed fator (land in two setors absorbs muh of the effets of hanges in output pries. Table 11: simulated hanges in fator returns with trade share shifts and mobile apital assumed. Land Unskilled Skilled Capital Ag Other Prim PLD 18.95% 18.25% -0.29% 23.16% 1.68% HUN 14.71% 14.15% -0.29% 8.30% 12.70% OCEC 12.36% 11.08% -0.29% 16.80% 1.57% UK -0.51% -0.50% -0.29% -0.82% -0.24% GER 0.00% -0.01% -0.29% 0.11% 0.21% OEUN -0.23% -0.22% -0.29% -0.37% 0.48% OEUS -0.34% -0.34% -0.29% -0.09% -0.16% FSU -0.12% -0.16% -0.29% 0.66% 0.25% ODX -0.54% -0.54% -0.29% -0.55% -0.53% LDC -0.47% -0.46% -0.29% -0.49% -0.47% 5. Conlusions In this paper, I have extended the modelling approah of Baldwin et al (1997 and LeJour et al (2001, introduing a model-onsistent framework of alibration and simulation to estimate the likely effets of EU enlargement to inorporate the CECs, assuming that the 1997 residual ountry bias against the CECs in EU trade (and vie-versa reflets resoure osts of regulatory differenes whih an be eliminated by entry to the single market. Sine this observed ountry bias is large (EU ountries trade far more with eah other than with the CECs, even orreting for size and differene, the regulatory barriers whih would be needed to explain suh differenes would also be signifiant of the order of 7 to 15 per ent on most goods. It follows that entry into the EU s Single Market would have large effets in terms of inreasing trade between the EU and CECs in fat simulations suggest that Poland s trade with the EU ould double, while Hungary and the other CECs would also see large trade inreases. Suh trade inreases would produe sizeable gains, not just from the elimination of the resoure ost of trade and from trade more aurately refleting omparative advantage, but also due to the effets of inreased ompetition and sale utilisation, partiularly within the CECs. As a result, welfare in the CECs ould rise by per ent (Poland being the largest gainer while the existing EU ountries would not lose from enlargement. Output would rise almost aross the board for all industries in the CECs. 21

22 These alulations suggest that, on the assumption that observed ountry bias reflets resoure osts to trade, and that entry to the single market ould eliminate those osts, existing studies have, if anything, been quite autious in their optimisti assessment of EU enlargement. However, it must be borne in mind that these onlusions are highly dependent on the above assumptions. There are a number of important autions: There is onsiderable room for unertainty over the omparative osts of prodution of different industries in different regions, and over the assoiated residual ountry biases, interpreted as ieberg trade osts, depending on the prior assumptions made in order to arry out the alibration (ie that g,cec,eu = g,eu,cec = g. It is probable that these prior assumptions are also important in gravity studies. Gravity modellers typially measure residual ountry bias with a set of trade dummies (eg a dummy set to 1 if both ountries are members of the EU and 0 otherwise. The number of these dummies is typially muh less than the number of alibrated oeffiients in our study, meaning that the gravity modellers are making far more restritions on the relative sizes of different ountry bias effets. Whether aession to the EU would in fat lead to the elimination of the fitted bias against CEC imports into the EU ompared to the produe of other EU ountries is not ertain. Indeed, gravity studies of the single market (Brenton and Verauteren 2001 ast doubt on the effets to date of institution of the single market. It is possible that the use of transport osts alone may underestimate the effets of distane upon trade (a weakness of the diret alibration approah ompared to standard gravity models.an extra regression of estimated oeffiients on distane might be worthwhile, to see if there are additional distane effets at work. The assumption that the residual ountry bias represents unmeasured trade osts,, rather than, say, differene in tastes and that these osts would be redued or removed by ountries joining the EU, is a strong one. For one thing, reorientation of prodution and onsumption is unlikely to be ostless. Estimates of the savings from double-testing and frontier ontrols due to the Single Market are more in the order of 1-2% saving on the ost of traded goods, rather than the 15% typially inferred by omparing trade shares. Whether the remainder of the 15 % atually represents other trading osts suh as the effets of different produt standards, labelling proedures, legal and guarantee systems et is hard to tell. Whether the harmonisation of produt standards in CEC ountries to onform to the existing EU standards would benefit the aession ountries, or would 22

23 impose unwanted osts on produers and onsumers in the aession ountries, is a very important point. This will be investigated further, as will the effets of the two different prinipal types of mehanism involved in the Single Market (harmonisation and mutual reognition. It is also possible, indeed likely, that over time CEC onsumers and produers may beome more oriented towards trade with the EU even if the ountries do not formally join the single market, so the oeffiients might well redue over time anyway. It is also likely that in 1997 the CEC ountries were far from in equilibrium: real exhange rates and trade barriers would have hanged very substantially in just the preeding 4 years. For that reason, export and import volumes might well not be at an equilibrium level relative to pries and trade barriers. For all these reasons, the figures in this study should be taken with a degree of aution. 23

24 Bibliography Balassa, B. (1964: The Purhasing Power Parity Dotrine: a Reappraisal. Journal of Politial Eonomy 72 (De 1964 pp Baldwin, R.E. and A.J. Venables (1995: Regional Eonomi Integration, Paper in Handbook of International Eonomis (III pp Elsevier, Amsterdam. Baldwin, R.E. (2001: paper in Maskus and Wilson (op it. Baldwin, R.E., Franois, J.F. and Portes, R. (1997: EU enlargement: Small osts for the west, big gains for the east.eonomi Poliy April 1997, pp Also omments by Rodrik following. Bergstrand, J.H. (1989: The Generalised Gravity Equation: Monopolisti Competition, and the Fator-Proportions Theory of International Trade. Review of Eonomis and Statistis 71 pp Brown, D. Deardorff, A., Djankov, S. and Stern, R. (1995: An eonomi assessment of the integration of Czehoslovakia, Hungary and Poland into the European Union.In S.W. Blak Europe s Eonomy Looks East Ch 2. CUP. Deardorff, A. (1998: Determinants of Bilateral Trade: Does Gravity Work in a Neolassial World? In J.A.Frankel (ed: The Regionalization of the World Eonomy, University of Chiago Press. Edwards, T.Huw (2003(1: Quality Standards under Classial Oligopoly and Trade: Regulatory Protetion of Just Over-Regulation? CSGR working paper 126/03 Nov 2003, Warwik UK. Edwards, T.Huw (2003(2: History Matters for Import Demand: a Math-Searhing Model of Inter-Firm Trade. Mimeo, CSGR Warwik. LeJour, DeMooij and Nahuis (2001: EU Enlargement: Impliations for Countries and Industries. Report CPB Netherlands Bureau for Eonomi Poliy Analysis. MCallum, J. (1995: National Borders Matter: Canada-US Regional Trade Patterns, Amerian Eonomi Review, 85: Maskus, K.E. and J.S.Wilson (eds 2001: Quantifying the Impat of Tehnial Barriers to Trade. Mihigan University Press. Chs 1 and 2 by the editors. Moenius, J. (1999: The role of standards in trade Mimeo, University of California, San Diego. Rauh, J.E. (1996: Trade and Searh: Soial Capital, Sogo Sosha and Spillovers. NBER Working Paper w Wallner, K. (1998: Mutual reognition and the Strategi Use of International Standards. Stokholm Shool of Eonomis Disussion paper. 24

25 APPENDIX I: Derivation of border osts and omparative prodution osts from trade data, based on an assumed Dixit-Stiglitz framework. In priniple it is possible to estimate border effets diretly by alibration of a general equilibrium model, rather than relying on indiret methods suh as estimation of a gravity model. This is most learly seen in the ase of a Dixit-Stiglitz model. The theoretial relationship between a Dixit-Stiglitz model, with monopolisti ompetition between differentiated goods g, eah produed in one ountry only, is well-established sine Bergstrand (1989. For simpliity, onsider a D-S model where goods are onsumed in ountries 1..C yielding onsumer utility. Consumption of good g in ountry is Q g,. Total onsumer utility in ountry is assumed to reflet the funtion: U. Q g -(1 /( 1 1 / 1 g, Where is the elastiity of substitution between goods varieties, and is an ieberg ost reduing by a fixed proportion the usable value of all goods from ountry onsumed in. Differentiating (1 and setting the marginal utility of onsumption of g equal to its relative prie yields: P ( 1 / 1/ ( 1.(1 t /.(1. U / Qg, -(2 Where is the proportionate transport ost between ountry and, and t is the net ontribution of import and export tariffs, subsidies and the tariff equivalents of NTBs 16. P is the selling prie of goods from ountry at the point of export (ie prior to trade osts and tariff. is an aggregate onsumer prie index for ountry. 16 One possible inonsisteny in the urrent analysis is that the tariff equivalent of NTBs has been taken diretly from the GTAP database, and so may not neessarily be onsistent with the substitution elastiities elsewhere in this paper. 25

26 , ( 1 We an rearrange this equation as: Q (..(1. / (1.(1 1 / g, U P t -(2a The next step is to rewrite the equation in terms of observable variables. The nominal value of exports from to, E is the number of goods varieties produed in ountry n, times the volume of sales per good, Q g, (g, (upsaled by (1+ to take aount of the assumed ieberg transport ost times the export prie P. We an also replae U with total expenditure in ountry, Y divided by the aggregate prie index. E ( 1.( 1. n.(1. P.( Y /.( / P.(1 (1 t -(3 Next we an replae n with the value of output in ountry X, divided by the size of turnover of a representative firm T. E E -( ( 1, ( 1 1.( 1. X. Y. T P..(1 (1 t. X This equation an of ourse be rewritten in logarithmi form: ln E (1.ln(1 -(4a.ln ( 1.ln(1.ln(1 t ln X lny ln T (1.ln P ( 1.ln It should be lear that this is a very similar funtional form to the equations estimated by gravity modellers, but with various parameter restritions imposed in order to ahieve onsisteny with the general equilibrium Dixit-Stiglitz framework. This is even learer if we hoose to model transport osts as a funtion of distane d : ln( 1 ln -(5 a b d Substituting from (5 into (4a, we essentially have a gravity model, but unlike the eonometrially estimated gravity models the oeffiients on industry output in ountry and on demand in ountry are onstrained to equal 1, while prodution pries are introdued as exogenous data (rather than being proxied by per apita inome, as in many gravity studies, and it is worth noting that the tariff term is ln(1+t not ln(t as in many gravity models. 26

27 The number of fitted residual border ost oeffiients,, is far greater than the number of dummies estimated in a gravity model. Effetively the gravity modeller is rewriting these as = DUM + f(, where DUM is whatever ombination of ountry dummies happens to apply to trade between ountries and, and is the estimated equation residual. Beause there are more oeffiients to estimate in our version, there are fewer degrees of freedom, making alibration more appropriate than eonometri estimation. Sine we are partiularly interested in the fitted residual border ost oeffiients,, we rearrange equation (4: ( 1 1 E.( 1. X. Y. T P..(1 (1 t To eliminate the onsumer prie indies, the easiest way is to say that for = we an replae E with H (home use. For H we have a simpler version of equation (4, sine, =t, =0: H ( ,. X. Y. T P..( 1 -(4b And dividing (4 by (4b gives us: E ( / H ((1 /(1,.( X / X.( T / T.( P / P (1 (1 t -(6 We an rearrange this to put (1- on the LHS, and if we assume = 0 if = we an simplify somewhat: (1 -( /(1 {( E / H.( X / X.( T / T.( P / P (1 (1 t } An interesting result is found if we multiply together these expressions for trade in both diretions between a pair of ountries, and, sine a lot of terms an then be eliminated: (1 -(8 or 1 1.(1, E.(1.(1 t E,.(1,.(1 t, H. H 1/(1 (1.(1, E. E, H. ~ ~ H 2 /(1 27

28 -(8a where the tild represents exports adjusted for the effets of tariffs, NTBs and transport osts. Effetively, if the geometri average volume of trade between two ountries, one tariffs and transport osts have been orreted for, is signifiantly smaller than the geometri mean of home-based onsumption in the two ountries, then the model implies there must be residual border osts present. One an elastiity of substitution, has been hosen, all the other terms on the RHS of (8 are eonomi variables whose value is known. Effetively, for given data sets, the values for (1- and (1-, are not independent: their produt an be written: (1- x (1-, =K, -(9 This means that for a given data set, the higher the value of the trade ost for imports from to,, the lower will be the trade ost in the other diretion,,. 28