econstor Make Your Publications Visible.

econstor Make Your Publcatons Vsble. A Servce of Wrtschaft Centre zbwlebnz-informatonszentrum Economcs Boer, Tto; Brücker, Herbert Workng Paper Mgraton, Co-ordnaton Falures and EU Enlargement DIW Dscusson Papers, No. 481 Provded n Cooperaton wth: German Insttute for Economc Research (DIW Berln) Suggested Ctaton: Boer, Tto; Brücker, Herbert (2005) : Mgraton, Co-ordnaton Falures and EU Enlargement, DIW Dscusson Papers, No. 481 Ths Verson s avalable at: http://hdl.handle.net/10419/18332 Standard-Nutzungsbedngungen: De Dokumente auf EconStor dürfen zu egenen wssenschaftlchen Zwecken und zum Prvatgebrauch gespechert und kopert werden. Se dürfen de Dokumente ncht für öffentlche oder kommerzelle Zwecke vervelfältgen, öffentlch ausstellen, öffentlch zugänglch machen, vertreben oder anderwetg nutzen. Sofern de Verfasser de Dokumente unter Open-Content-Lzenzen (nsbesondere CC-Lzenzen) zur Verfügung gestellt haben sollten, gelten abwechend von desen Nutzungsbedngungen de n der dort genannten Lzenz gewährten Nutzungsrechte. Terms of use: Documents n EconStor may be saved and coped for your personal and scholarly purposes. You are not to copy documents for publc or commercal purposes, to exhbt the documents publcly, to make them publcly avalable on the nternet, or to dstrbute or otherwse use the documents n publc. If the documents have been made avalable under an Open Content Lcence (especally Creatve Commons Lcences), you may exercse further usage rghts as specfed n the ndcated lcence. www.econstor.eu

Dscusson Papers Tto Boer* Herbert Brücker** Mgraton, Co-ordnaton Falures and EU Enlargement * Berln, February 2005 * Unverstà Boccon, Mlan, IGIER, IZA, Bonn ** German Insttute for Economc Research, DIW Berln, Aarhus School of Busness, IZA, Bonn

IMPRESSUM DIW Berln, 2005 DIW Berln Deutsches Insttut für Wrtschaftsforschung Köngn-Luse-Str. 5 14195 Berln Tel. +49 (30) 897 89-0 Fax +49 (30) 897 89-200 www.dw.de ISSN 1433-0210 (Druck) 1619-4535 (elektronsch) Alle Rechte vorbehalten. Abdruck oder verglechbare Verwendung von Arbeten des DIW Berln st auch n Auszügen nur mt vorherger schrftlcher Genehmgung gestattet.

MIGRATION, CO-ORDINATION, FAILURES AND EU ENLARGEMENT 1 Mgraton, Co-ordnaton Falures and EU Enlargement Paper presented at the 41 th Economc Polcy Panel n Luxembourg, 15/16 Aprl 2005 Tto Boer and Herbert Brücker 1 Unverstà Boccon, Mlan and IGIER; DIW-Berln and Aarhus School of Busness 1. INTRODUCTION EU leaders are well aware of the relevance of mgraton n the European polcy agenda. Thrty-two pages out of forty-eght of the Presdency Conclusons at the November 2004 European Councl were devoted to mgraton polces. Economc theory suggests that there s a strong case for polcy co-ordnaton n ths feld: relevant spllovers across natonal jursdctons, economes of scale and potental free-rdng n the enforcement of border controls. However, no delegaton of authorty to supra-natonal bodes s envsaged n the Presdency Conclusons n the feld of restrctons to legal mgraton. Whle Qualfed Majorty Votng s now accepted on measures tacklng llegal mgraton, decsons on restrctons to legal mgraton are envsaged only under unanmty rules. Un-coordnated natonal polces are gettng ncreasngly tough on mgrants. Border controls are tghtened or welfare access s prevented to foregners or both doors, work and socal assstance, are closed. Ths does not seem to prevent mgraton to occur. It dstorts ts geographcal orentaton, modfes ts skll composton and nflates the ranks of the nformal sector. Illegal mgraton s larger when restrctons to legal mgraton are tght. Illegal flows as a proporton of the populaton are about one fourth larger n Europe than n the US; at the same tme, legal flows are 25% larger n the US than n Europe. And the US has more realstc mgraton restrctons than most European countres. Countres are gettng tougher also because they fal to co-ordnate. The mplcatons of the falure of polcy co-ordnaton n terms of natonal restrctons came out very clearly wth the Eastern Enlargement. There was a race to the top of mgraton restrctons wth 12 out of the 15 Member States of the European Unon (EU) renegng on ther prevous 1 The authors wsh to thank Rchard Baldwn and two anonymous referees for comments on an ntal draft. They are also ndebted to Paola Mont for sklful research assstance.

MIGRATION, CO-ORDINATION, FAILURES AND EU ENLARGEMENT 2 commtment not to restrct worker flows from the New Members. And the few EU-15 countres that ultmately opened ther labour markets for workers from the New Member States at least partally restrcted nstead access to welfare by mgrants. Lack of coordnaton rases concerns among publc opnon that mgraton flows could be dverted to the most lberal countres, ncreasng pressures on ther welfare systems. Dverson of mgraton flows s, by tself, undesrable on economc grounds. It means that mgraton cannot fully play a spatal arbtrage functon, greasng the wheels of otherwse mmoble labour markets. Thus, natonal Governments seem to be caught nto a vcous crcle: they top-up mgraton restrctons enforced by other countres endng up for the most to ncrease llegal mgraton, whch tself rases concerns among publc opnon nducng a tough stance towards mgrants. Why s t so dffcult to co-ordnate mgraton polces at the EU level? Is t because of freerdng? Who gans and who loses from uncoordnated mgraton restrctons? How much do they dstort East-West mgraton both n terms of the geographcal orentaton of worker flows and the skll composton of mgraton? It may well be that the countres not located at the borders of the Unon and hence recevng less mgrants prefer to enjoy the benefts of stronger growth elsewhere n the Unon wthout havng to bear the assmlaton and fscal costs of mmgraton. Another explanaton s that for non-economc reasons (e.g., hstorcal and cultural factors) ctzens of dfferent EU countres have dfferent vews about mgraton and hence oppose any polcy co-ordnaton at the EU level. Else t s natonal poltcans who prefer to keep under ther jursdctons mgraton polces as they target a convenent poltcal scapegoat, namely the non-votng mmgrant. The purpose of ths paper s to shed some lght on these ssues, by drawng on lessons from the Eastern Enlargement epsode, usng prelmnary evdence on mgraton to countres wth dfferent types of restrctons, predctons from a computable general equlbrum model as well as publc opnon polls. The plan s as follows. At frst, n Secton 1, we succnctly revew recent evolutons of mgraton restrctons n the EU-15 and the race to the top occurred n the eve of the Eastern Enlargement. Moreover, we analyse whether and to what extent the transtonal arrangements result n the dverson and reducton of East-West mgraton flows. For ths purpose, we compare estmates of the mgraton potental wth recent mgraton patterns observed after May 1, 2004. Next, n Secton 2, we evaluate, based on a stylsed general equlbrum model, not only the costs for the EU of havng un-coordnated mgraton polces, but also potental explanatons of the falure to co-ordnate polces. Although the potental ncome gans from East-West mgraton exceed those from the further ntegraton of goods and captal markets, the uneven dstrbuton of gans and losses across recevng and sendng countres can create ncentves for closng-the-door polces and for free-rdng on lberal polces of other countres. Moreover, we analyse whether welfare benefts ncrease ncome n the regon by facltatng further mgraton and whether the dstrbuton of welfare benefts generates tself ncentves for co-ordnaton falures. In Secton 3, we turn to perceptons as to the costs and benefts of mgraton n general and from the East n partcular, as revealed by publc opnon polls n the West. We look at whether or not they pont to a dvergence of preferences on these ssues and what are the determnants of crosscountry dfference n the degree of acceptance of mgrants n relaton wth economc as well as non-economc factors. Fnally, n Secton 4 we summarse our results and draw mplcatons as to ways to overcome resstance of some governments to co-ordnate polces.

MIGRATION, CO-ORDINATION, FAILURES AND EU ENLARGEMENT 3 2. RECENT EVOLUTIONS IN NATIONAL IMMIGRATION POLICIES 2.1. The tghtenng of mmgraton restrctons Internatonal mgraton s the "great absentee" n the era of globalsaton. Whle the barrers to nternatonal trade and captal moblty have already been largely removed, labour markets are the most tghtly regulated area of economc actvty (Fan et. al., 1999). Governments regulate, among other thngs, the maxmum number of work permts to be granted wthn a gven perod of tme (usually a year), the crtera to be followed n rankng applcatons for vsas (e.g., sklls, lngustc capabltes, natonalty, famly lnks), the duraton of the work permts, the procedures to be followed n the renewal of vsas, the number of years requred before obtanng a permanent resdence permt, the type of resdence permts allowed (e.g., temporary vs. permanent), the nature and number of certfcates and guarantees requred for the admsson n the host country, the type and number of admnstratons nvolved n the processng of applcatons, and the procedures to be followed n case of famly reunfcaton. Not less regulated s asylum seekng mgraton, whch often respond also to economc factors. Regulatons are also frequently revsed, whch ncreases the uncertanty assocated wth the decson to mgrate. Ths may have perverse effects on the tmng of mgraton as there s an opton value n mgratng before borders are closed but certanly ncreases ts costs. Wthn the European Unon (EU) matters are, at least n prncple, dfferent. Snce the Rome Treaty, whch establshed the European Economc Communty (EEC) n 1957, the free moblty of labour has been acknowledged as one of the four fundamental freedoms of the Common Market. Free movement started n a communty of sx countres wth a jont populaton of 185 mllon people, and has been extended step-by-step to the 15 members of the 'old' EU and the three other members of the European Economc Assocaton (EEA) wth a jont populaton of 380 mllon people. Although many barrers to ntra-eu moblty of EU ctzens reman (e.g., n terms of portablty of prvate penson rghts, legal recognton of professons, nformaton about job opportuntes, etc.), the free moblty of labour, ncludng the equal treatment wth regard to welfare benefts, s n prncple guaranteed by the supranatonal legslaton of the EU. Legal mmgraton nto the EU from thrd countres s nstead regulated at the natonal level and the recent evoluton of these natonal mgraton polces n the EU nvolved tghter and tghter restrctons. Snce 1996 there have been 35 reforms n ths feld, that s, almost 4 reforms per year. Most of these reforms (80%) are margnal n that they adjust specfc provsons rather than revsng the overall regulatory framework. Furthermore, 7 reforms out of 10 tghten regulatons, e.g., by ncreasng procedural obstacles faced by those applyng for vsas, reducng the duraton of work permts or makng famly reunfcaton more dffcult. The trend n mgraton polces can also be charactersed n terms of ndexes for the man polcy areas. Fgure 2.1. draws on mmgraton polcy ndexes developed at Fondazone Rodolfo Debenedett (see www.frdb.org for detals) and an ndex of restrctons n asylum polces defned by Hatton (2004). Larger values of the ndexes denote tghter regulatons. As shown by Fgure 2.1., t s manly requrements to be fulflled for beng

MIGRATION, CO-ORDINATION, FAILURES AND EU ENLARGEMENT 4 granted an entry vsa and natonal quotas whch are gettng tghter. Some relaxaton s occurrng n terms of years requred to obtan ctzenshp and assmlaton polces are sometmes beng strengthened, but entry s becomng more and more dffcult. Importantly, countres tghtenng regulatons are often those whch had the most restrctve provsons to start wth. Ths s the vsual mpresson gven by Fgure 2.2., plottng the value of the aggregate polcy ndex obtaned by takng the average of the sx ndcators dsplayed n Fgure 2.1. n the ntal and fnal year for whch observatons are avalable. Not only are most countres above the bsectng lne through the orgn, denotng a tghtenng of regulatons, but also t s the countres whch ntally had the most lberal polces (perhaps because they were hstorcally emgraton countres) whch have lberalsed flows. Fgure 2.1 Trends n Mgraton Polces 160 Immgraton polcy ndexes 140 120 100 80 60 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 Admsson requrements Stayng requrements Number of admnstratons nvolved Asylum polcy Length of frst stay Years to obtan a permanent resdence permt Quotas There s also evdence that reforms are exertng spllover effects across jursdctons. For nstance, Fnland tghtened up ts regulatons n 2004, followng closely the most restrctve stance taken by Denmark n 2002. Portugal adopted more restrctve provsons n 2001, just after a lkewse restrctve reform mplemented by Span n 2000. And Ireland chose a more restrctve approach n 1999, after two reforms n the UK, whch tghtened up mgraton restrctons, respectvely n 1996 and 1998. Moreover, ndvdual countres closely montor developments n other countres. In the webste of the Brtsh Home Offce one can fnd several reports revewng the evoluton of mgraton polces n the other EU Members. Recently Denmark and the Netherlands protested aganst the large regularsaton process n Span, argung that t could have pushed more llegal mgrants nto the EU. In the publc debate, reference s often made to changes n mgraton polces occurred n other countres. The provocatve proposal of the leader of the Italan Northern League, Umberto Boss, to

MIGRATION, CO-ORDINATION, FAILURES AND EU ENLARGEMENT 5 buld-up a wall at the border of Italy wth Slovena was based on the fact that Austra was closng border towards mgrants from the New Members states. And the current Dansh Prme Mnster has been actvely campagnng on mgraton referrng to polces carred out n Germany, UK and Sweden. Polcy spllovers are present also n the case of Asylum Polces, where a rush to the top occurred towards the end of the 1980s after the ntal decson of Germany and, later, the UK to tghten asylum legslaton (Hatton, 2004). In the case of polces for mmgratons of hghly sklled mgrants, spllover are nstead occurrng n the other drecton (OECD, 2001), wth a race n tryng to attract more hghly sklled mgrants. For nstance, the decson of France n 2004 to ntroduce tax deductons for hghly sklled mgrants was explctly motvated by the fact that Belgum had adopted polces encouragng nflows of hghly sklled workers. Smlarly., the Brtsh ad hoc scheme for hghly sklled mgrants was motvated referrng to the measures beng taken n Germany. Fgure 2.2. Convergence n mmgraton polces? Spllovers n mgraton polces were also evdent n the way n whch the EU-15 adjusted ther regulatons on work permts n the eve of the Eastern Enlargement, explotng the transtonal arrangements allowed for by the accesson treates. Ths event s dscussed below. 2.2. Eastern Enlargement and the Race-to-the-Top Ffteen years ago, the ron curtan was lfted for a regon of approxmately 400 mllon people n Central and Eastern Europe. Eght of these Central and Eastern European

MIGRATION, CO-ORDINATION, FAILURES AND EU ENLARGEMENT 6 countres 2 joned the EU on the 1 st of May, 2004. Another two countres, Bulgara and Romana, are expected to accede n 2006 or 2007. The jont populaton of the eght New Member States (NMS) from Central and Eastern Europe amounted to 72 mllons n 2004, the populaton of the two other accesson canddates number some 30 mllons. The accesson treates contan transtonal arrangements for the free labour moblty, whch allow to postpone the openng of labour markets up to a maxmum perod of seven years. 3 These transtonal provsons can only be adopted sequentally: at frst, n the two years followng accesson, all Member States can apply natonal rules on access to ther labour markets; at the end of ths two-year perod, each Member State can choose to apply natonal rules for another three years or mplement the Communty rules regulatng free labour moblty n the EU. If the countres decde to apply the Communty rules, they mantan a safeguard clause meanng that they can rentroduce work permts temporarly. There wll be an automatc revew before the end of the two-year perod and a further revew on request of each affected Member State, but the decson on the applcaton of transtonal perods s left to the natonal Governments. At the end of the fve year perod, Communty rules should be ntroduced under normal crcumstances, but the transtonal perods can be prolonged for another two years f the Member State can document that t experences (or s threatened by) serous dsturbances n ts labour market. Agan, the decson on the prolongaton of the transtonal perods s left to natonal Governments. Transtonal perods for the free moblty of labour have been agreed also n other Enlargement rounds: n case of the accesson of Greece a sx-year transtonal perod was agreed, and, at the accesson of Portugal and Span, a seven-year transtonal perod was ntroduced, later on reduced to sx years. However, what makes the present rules dfferent from those adopted n prevous Enlargement rounds s that ndvdual countres are let free to decde on whether or not adoptng the transtonal arrangements. Delegatng the decson on transtonal perods to the natonal level had mportant consequences: seven out of ffteen Member States of the EU, among them Austra and Germany, who attract about two-thrds of the mgrants from the accesson countres at present, declared from the begnnng that they planned to leave relatvely tght restrctons to the mmgraton of labour n place at least for the frst two years after Enlargement. 4 Governments of another fve countres Denmark, Ireland, the Netherlands, Sweden and the UK stated nstead that they planned not to restrct the access to ther labour market at that tme, whle the remanng countres (Greece, Italy, Portugal and Span) remaned undecded although the relevant Mnsters publcly stated that they were n favour of free moblty of workers from the NMS. The Governments n three out of the fve countres whch had formally stated ther ntenton to open-up ther labour market reneged on ths commtment and all adopted transtonal restrctons vs-à-vs workers from the NMS. In partcular, n Denmark, the government agreed wth the opposton to concede a work permt only to those ndvduals from the new Member States who can prove that they have a job whch meets regular standards wth regard to wage and workng condtons. If a mgrant looses her job, resdence permts are wthdrawn (Dansh Mnster of Employment, 2004). The Netherlands reversed the decson of the Kok II Government to open up the labour market completely and ntroduced nstead a quota of 22,000 employees untl May 2006. If the quota s not flled, the removal of the transtonal arrangement can be consdered (Kvst, 2004). Welfare access was also closed to mgrants. In Sweden, the mnorty 2 Czech Republc, Estona, Hungary, Latva, Lthuana, Poland, Slovak Republc and Slovena. 3 Free movement s granted to the ctzens from the new Member States n prncple, but the EU countres can restrct the access to ther labour markets durng the transtonal perods. 4 Beyond Austra and Germany ths was Belgum, Fnland, France, Greece and Luxembourg.

MIGRATION, CO-ORDINATION, FAILURES AND EU ENLARGEMENT 7 government proposed to ssue resdence and work permts only for those workers from the NMS who could prove that they had jobs meetng certan requrement wth respect to natonal wage agreements and excluded ndvduals from the CEECs from several welfare benefts. However, the proposal of the Swedsh government was overruled n Parlament, and Sweden s currently the only country where Communty rules for labour moblty apply at present. The Unted Kngdom and Ireland decded to open-up ther labour markets n prncple to ndvduals from the new Member States, but they left also certan restrctons n place. Work permts are only ssued for one year, and f mgrants lose ther jobs, the resdent permts can be wthdrawn. Agan, the access to welfare benefts remans restrcted (Home Offce, 2004a). Ths s a substantal change compared to the ntal announcement of the Brtsh Government to open labour markets mmedately after accesson. Fnally, the three undecded countres -- Italy, Portugal and Span opted for restrctve provsons. Italy, n partcular ntroduced a quota of 20,000 work permts for workers from the NMS, well below the projected mmgraton flows to ths country (Alvarez-Plata et al., 2003) whle Greece, Span and Portugal decded to leave ther mmgraton restrctons n place at least for the frst two-year perod. Table 2.1 Transtonal regulatons n the EU-15 Access to labour market Access to welfare benefts Austra Access to labour markets restrcted at least for 2 years, Restrcted. quotas for work permts. Belgum Access to labour markets restrcted at least for 2 years. Restrcted. Denmark General access to labour market, but oblgatons for Restrcted, resdence and work and resdence permts. Work permts ssued only work permts can be wthdrawn for 1 year (EU-natonals: 5 years). n case of unemployment. Fnland Access to labour markets restrcted at least for 2 years. France Access to labour markets restrcted at least for 2 years. Restrcted. Germany Access to labour markets restrcted at least for 2 years, Restrcted. prolongaton for further 3 years under dscusson. Greece Access to labour markets restrcted at least for 2 years. Ireland General access to labour market, but oblgaton to Restrcted, ncome support etc. regster for work and resdence permts. Work permts s granted only to ndvduals ssued frst for lmted tme. Safeguard clause apples. whch have a rght for a resdence permt. Italy Access to labour markets restrcted at least for 2 years, Restrcted. quotas for work permts. Luxembourg Access to labour markets restrcted at least for 2 years. Restrcted. Portugal Access to labour markets restrcted at least for 2 years, Restrcted. quotas for work permts. Span Access to labour markets restrcted at least for 2 years, Restrcted. blateral agreement wth Poland whch permts lmted number of Polsh natonals to work. Sweden Communty rule for free labour moblty apples. Equal treatment. Unted General access to labour market, but oblgaton to Restrcted, ncome support etc. Kngdom regster for work and resdence permts. Work permts s granted only to ndvduals ssued frst for lmted tme. Safeguard clause apples. whch have a rght for a

MIGRATION, CO-ORDINATION, FAILURES AND EU ENLARGEMENT 8 resdence permt. Sources: Collecton by the authors, based on Home Offce (2004); Kvst (2004) and natonal nformaton. Altogether, we observed a race to the top n mmgraton restrctons vs-à-vs workers from the NMS. The fnal outcome was lkely worse than had the restrctons beng agreed at the EU-level. It was certanly more restrctve than n the case of the prevous enlargement rounds, although ncome dfferences are n ths case unprecedented (Boer and Brücker, 2001). It s lkely that the ntal decson of the two largest mmgraton countres Austra and Germany to restrct mgraton from the new Member States fuelled fears that mgraton flows could be dverted nto smaller countres, rasng concerns about mgraton pressures there. Moreover, decsons to apply transtonal perods n ndvdual Member States have been carefully revewed by other Member States and affected decsons elsewhere. For nstance, the decson n the Netherlands to reverse the ntal decson of the Kok II Government nfluenced the decsons of the Swedsh government (Kvst, 2004) to renege on ts prevous commtments. Smlarly, the Dutch Government s presently proposng to extend the transtonal perod for another two years and a key argument beng used before the Parlament s that Germany s gong to do the same. Sgnfcant press coverage of decsons made n neghbourng countres was provded n these countres. The race to the top ended wth four dfferent transtonal regmes: frst, a restrctve mmgraton regme, whch provdes natonals from the new Member States no further rghts than ctzens to non-eea countres. Ths mples that work permts are only ssued n exceptonal crcumstances when t can be proved that nether natves nor other EU-natonals can fll the poston. The man channel of entry s n these countres famly reunfcaton. Ths regme apples to Belgum, Fnland, Germany, Greece, France, Luxembourg and Span. The second regme adopts bascally the same rules than the frst one, but t opens the labour market beyond that by a quota for natonals from the new Member States (Austra, Italy, Netherlands, Portugal). Thrd, we have a number of countres whch admt generally the access of natonals from the new Member States to ther labour markets, but work and resdence permts are only ssued f certan requrements wth regard to tarff wages, workng condtons etc. are met. Moreover, the access to welfare benefts s lmted and resdence permts can be wthdrawn n case of unemployment (Denmark, Ireland, UK). Fnally, we have one country, Sweden, where due to the falure of the government to fnd a majorty n Parlament, the rules of the Communty for the free movement of labour apply (Table 2.1.). 2.3. The consequences of the race-to-the top : mgraton reducton and mgraton dverson When the Berln Wall broke down n 1989, many observers expected a mass mgraton wave from the East to the West. Indeed, the ncome gap between the East and the West n Europe s substantal. The per capta GDP of the new Member States amounts n purchasng power partes to 49 per cent of ncomes n the EU-15 (Eurostat, 2003), and n the neghbourng Commonwealth of Independent States to around one-ffth of the EU-15 level (Fgure 2.3). At current exchange rates, the ncome gap s even larger. Nevertheless, wth few exceptons,.e. the cases of Albana and Eastern Germany, the removal of emgraton barrers for the people n the former Eastern Bloc caused only moderate mgraton so far: cumulatve net emgraton snce 1989 from the eght NMS and the two other accesson canddates from Central and Eastern Europe to the EU can be estmated at around 1,1 mllon people, whch

MIGRATION, CO-ORDINATION, FAILURES AND EU ENLARGEMENT 9 equals one per cent of ther populaton. No doubt, these moderate mgraton flows reflect also tght mmgraton restrctons n Western Europe. But also suggest that mgraton pressures are not as strong as to bypass natonal restrctons. The transtonal arrangements for mmgraton from the new Member States whch we observe n the EU snce the 1 st of May, 2004, affect the allocaton of mgraton flows and stocks. They have two effects: on the one hand, they dvert mgraton flows away from the countres whch have decded to pursue restrctve mmgraton polces durng the transtonal perods to those whch decded to open ther labour markets, at least partally. On the other hand, transtonal arrangements reduce mgraton stocks and flows relatve to the scenaro wth free labour moblty, snce the cost of mgraton ncrease wth the dstance from the country that would have been chosen otherwse. Dstance s meant here to be a measure whch captures not only geographcal dstance, but also the cultural, lngustc and socal dstance from the preferred destnaton. 5 45 000 Fgure 2.3 The ncome gap n Europe, 2002 PPP-GDP per capta (EURO) 40 000 35 000 30 000 25 000 20 000 15 000 10 000 5 000 LUX SWI DK UK European Economc Area (EEA) + Swtzerland FRA GER ITA ESP CEEC-10 SLO CZE HUN POL ROM TK CIS ALG MOR 0 0 100 200 300 400 500 600 700 800 900 populaton (mllons) Table 2.2 dsplays the number of mgrants from the CEEC-10 n the EU-15. It clearly documents that geographcal dstance s a key factor explanng the allocaton of mgrants from the CEEC-10 across the old EU Member States: the countres borderng the CEECs,.e. Austra, Germany, Greece and Italy, absorbed more than 80 per cent of the mgrants from the accesson countres before the Enlargement. The share of mgrants from the CEEC- 10 n the populaton of these countres vares -- wth the excepton of Italy, whch shares only a small border wth Slovena between 0.7 and 1.0 per cent, whle t s only 0.3 per cent at the average of the EU-15. Ths regonal mgraton pattern has been rather stable 5 See Hansen (2003) for a formal exposton of ths argument.

MIGRATION, CO-ORDINATION, FAILURES AND EU ENLARGEMENT 10 durng the 1990s and early 2000s, although the share of CEEC-10 mgrants gong to Germany has slghtly declned n the wake of the economc stagnaton n the recent years. Note that all these countres attractng sgnfcant flows from the CEEC-10 mposed tght restrctons for mmgraton from the new Member States durng the transtonal perods. Among the Bg-Four n the old EU only the UK decded to open ts labour market partally. It s too early to assess emprcally whether and to what extent the transtonal arrangements actually reduced mgraton flows and dverted mgraton away from the tradtonal destnatons of mgrants from NMS. We provde below just some frst scattered evdence. Before dong that, n Secton 2.3.1 we provde the counter-factual,.e. a projecton of the mgraton potental under the hypothetcal assumpton of free labour moblty for all countres n the enlarged EU. Then, n Secton 2.3.2 we wll present reported mgraton patterns after Enlargement as devatons from ths projecton. Table 2.2 Regonal break-down of mgrants from the CEEC-10 across the EU-15 resdents as a percentage year from CEEC-10 of populaton as a percentage of total CEEC-10 mgrants Austra 2001 78,886 1.0 7.3 Belgum 2001 13,208 0.1 1.2 Denmark 2004 11,596 0.2 1.1 Fnland 2001 13,639 0.3 1.3 France 1999 51,942 0.1 4.8 Germany 2003 614,094 0.7 57.0 Greece 2001 71,742 0.7 6.7 Ireland 1) 2002 12,235 0.4 1.1 Italy 2) 2001 102,105 0.2 9.5 Luxembourg 2001 1,547 0.3 0.1 Netherlands 2004 17,538 0.1 1.6 Portugal 2001 963 0.0 0.1 Span 3) 2003 17,104 0.0 1.6 Sweden 2003 24,295 0.3 2.3 UK 2001 45,858 0.1 4.3 EU-15 1,076,752 0.3 100.0 1) Only Latva, Lthuana, Poland and Romana.- 2) Only Poland and Romana.- 3) Only Bulagra and Poland. Sources : Authors' calculatons based on fgures reported by natonal statstcal offces and populaton censuses. 2.3.1. Lookng for the counter-factual: East-West mgraton under free movement A number of studes tred to estmate the long-run mgraton potental from the CEECs. Although most of these studes estmate the long-run mgraton potental from the Central and Eastern European countres n the EU-15 at between 3 and 4 per cent (e.g. Layard et al., 1992; Bauer and Zmmermann, 1999; Boer and Brücker, 2001; Alvarez-Plata et al., 2003; Kreger et al., 2003), there exst also studes whch obtan sgnfcantly lower (Fertg, 2001;

MIGRATION, CO-ORDINATION, FAILURES AND EU ENLARGEMENT 11 Fertg and Schmdt, 2001; Dustmann et al., 2003) or hgher estmates (Snn et al., 2001, Flag, 2001). The dfference between the estmates can be largely traced back to dfferent econometrc estmaton procedures. The methodologcal problems assocated wth estmates of the mgraton potental from the CEECs are dscussed n some detal n Box 1.

MIGRATION, CO-ORDINATION, FAILURES AND EU ENLARGEMENT 12 Box 1 Methodologcal problems n estmatng mgraton potentals Startng wth the semnal contrbuton of Layard et al. (1992), numerous studes have tred to estmate East-West mgraton potental. Bascally we can dstngush three man approaches n the lterature: representatve surveys, extrapolatons from South-North to East-West mgraton, and forecasts based on econometrc estmates of macro-mgraton models. Each approach has ts advantages and dsadvantages. Representatve surveys allow deep nsghts nto mgraton ntentons and the human captal characterstcs of potental mgrants (see Fassmann and Hntermann, 1996; IOM, 1999, Kreger et al., 2003). However, there are three problems whch make t almost mpossble to derve quanttatve forecasts of the mgraton potental from them: frst, we do not know how serous mgraton ntentons revealed n surveys are,.e. whether mgraton ntentons result n actual movements. Second, surveys capture only the supply sde and gnore demand sde factors such as job opportuntes and the avalablty of housng. Thrd, surveys cannot mrror the temporary dmenson of mgraton approprately: snce only few mgrants stay permanently abroad, a large number of ndvduals whch mgrate once n ther lfetme can concde wth a small fracton of the populaton whch stays at a certan pont of tme abroad. Most careful surveys of mgraton ntentons make therefore adjustments, whch scale down mgraton ntentons of 10 per cent of the populaton or more to an actual estmate of the mgraton potental of between 2 and 4 per cent of the populaton (see e.g. Kreger et al., 2003). Another strand of the lterature extrapolated the number of South-North mgrants n the 1960s and early 1970s to East-West mgraton (Layard et al. 1992; Bauer and Zmmermann, 1999). Note that the ncome gap between the Southern and the Northern European countres n the 1960s was smlar to the gap between the EU-15 and the accesson countres today (Maddson, 1995). In general, these extrapolaton studes fnd a long-run mgraton potental of around 3 per cent of the populaton. However, n stark contrast to the condtons for South-North mgraton n the early 1960s and 1970s, the condtons for East-West mgraton today are affected by mbalances n both the labour markets of the recevng and sendng countres, ncomplete recovery from the transton shock, and close geographcal proxmty. Thus, extrapolaton studes can provde no more than a hnt at plausble orders of magntude. The majorty of the forecasts of East-West mgraton are based on econometrc estmates of macro mgraton models, whch explan mgraton flows or mgraton stocks by economc varables such as the ncome dfferental, (un-)employment rates n the sendng and recevng countres, and some nsttutonal varables. Although most studes employ the same set of explanatory varables, the estmates of the parameters, and, hence, of mgraton potentals dffer consderably n the lterature. The man problem of the econometrc estmates can be traced back to the fact that they have to transfer parameter values whch are estmated n another hstorcal context and n another country sample are used for projectons n the Central and Eastern European countres (Alecke et al., 2001; Fertg and Schmdt, 2001; Dustmann et al., 2003). The Central and Eastern European countres cannot be ncluded n the orgnal sample, snce the ron curtan prevented effectvely emgraton. Thus, the projectons rely mplctly on the assumpton that the estmated parameter values reman constant not only across tme, but also across space. Unfortunately, mgraton behavour dffers largely across countres due to dfferences n geography, language, culture, etc. Ths affects both the estmates of the ntercept terms whch capture all factors whch have a tme-nvarant mpact on mgraton and the slope parameters.

MIGRATION, CO-ORDINATION, FAILURES AND EU ENLARGEMENT 13 There exst a number of alternatve econometrc models whch mpose dfferent restrctons on the ntercept, the slope parameters and the error terms. As an example, pooled OLS models, whch are wdely appled n the emprcal lterature, assume that both the ntercept and the slope parameters are homogenous across countres, whle fxed effects estmators allow for dfferent ntercepts, but assume that the slope parameters are homogeneous. Fnally, there exst a number of heterogeneous estmators, whch allow also the slope parameters to dffer across countres and use for forecasts averages of the estmated parameters. Each estmaton procedure has ts trade-offs; the choce of the adequate estmaton procedure s largely an emprcal queston. A large number of dfferent estmaton procedures have been tested on ther out-of-sample forecastng performance n the context of nternatonal mgraton (Alvarez-Plata et al., 2003; Brücker and Slverstovs, 2004). The key results of these studes are that () fxed effects estmators, whch allow for country-specfc ntercepts, but assume that the slope parameters are constant across countres, outperform all other estmators, () the forecastng errors of pooled OLS models, whch assume that both the ntercept and slope parameters are homogenous across countres, are around twce as hgh as those of fxed-effects models, and () that heterogeneous estmators whch allow both the ntercept and the slope parameters to dffer across countres are clearly outperformed by standard panel estmators. Thus, we chose for the estmaton of the mgraton potental here a standard fxed effects estmator. However, the reader should keep n mnd that any mgraton forecast for the Central and Eastern European countres reles on a number of arbtrary assumptons, n partcular the assumpton that we can transfer the parameter values obtaned from another country sample to the Central and Eastern European countres. All forecasts are therefore subject to a good deal of uncertanty and should be nterpreted carefully. The mgraton forecast s based here on the estmaton of a macro model for the mgraton to Germany from a panel of European source countres for the perod 1967-2001 (see Appendx A). Germany has been chosen because t s the man destnaton for mgrants from the CEECs, and t reports stocks and flows of mgrants at relatvely long seres, unlke many other EU countres. Table 2.3 Germany: potental mgraton from the accesson countres, 2004-2030 2004 2005 2006 2007 2008 2009 2010 2015 2020 2025 2030 net mgraton (persons) CEEC-10 225 453 258 201 239 719 203 173 163 623 127 916 98 420 29 379 16 195 12 716 10 449 CEEC-8 155 561 169 031 155 228 132 003 107 644 85 798 67 691 23 551 13 284 9 750 7 197 CEEC-2 69 892 89 171 84 491 71 169 55 979 42 118 30 729 5 828 2 911 2 965 3 252 foregn populaton (persons) CEEC-10 824 202 1 082 404 1 322 123 1 525 295 1 688 918 1 816 834 1 915 254 2 158 985 2 257 596 2 327 059 2 383 958 CEEC-8 628 065 797 096 952 324 1 084 327 1 191 971 1 277 768 1 345 459 1 527 200 1 608 334 1 663 647 1 704 652 CEEC-2 196 137 285 308 369 799 440 968 496 947 539 066 569 795 631 785 649 262 663 412 679 306 CEEC-10: Bulgara, Czech Republc, Estona, Hungara, Latva, Lthuana, Poland, Romana, Slovaka, Slovena. - CEEC-8: CEEC-10 wthout Bulgara and Romana. - CEEC-2: Bulgara, Romana. - See text for assumptons of the projecton. The projectons of the mgraton potental n Table 2.3 are based on the assumpton that the GDP per capta between the new Member States and the EU-15 converges at a rate of 2 per cent p.a. Ths speed of convergence s consstent wth a number of studes for the EU and other European market economes (e.g. Barro and Sala--Martn, 1991; 1995). It mples that halvng the ntal gap n per capta ncome levels wll take 35 years. Growth rates n the new Member States snce the end of the transtonal recesson ft pretty well nto ths

MIGRATION, CO-ORDINATION, FAILURES AND EU ENLARGEMENT 14 projecton, although the growth experence dffers from country to country. Unemployment rates are hard to predct. We assume therefore that unemployment rates reman constant durng the projecton perod at a level whch s equal to the average rate n the respectve countres durng the last fve years. Ths s consstent wth the observaton of jobless growth n the CEEC-10 (Boer and Garbald, 2005). The projecton for Germany s dsplayed n Table 2.3. The scenaro predcts under the hypothetcal assumpton of a removal of the mgraton barrers n 2004 a net mgraton of 156,000 persons to Germany from the eght new Member States (CEEC-8) and of 225,000 persons for all ten accesson countres (CEEC-10). Net mgraton acheves ts peak one year later nvolvng around 170,000 and 260,000 persons from the CEEC-8 and the CEEC-10, respectvely, and then declnes to attan sx years later about one-thrd of ths level. The long-run mgraton stock s attaned 25 years after the ntroducton of the free movement at a foregn populaton of 1.7 and 2.4 mllon persons from the CEEC-8 and the CEEC-10, respectvely. The smulaton results also demonstrate that transtonal arrangements have no mpact on the long-run mgraton potental, snce the convergence of per capta ncome levels s relatvely slow (not dsplayed here). Of course, all these results rely on a number of arbtrary assumptons and can provde no more than a rough gudance as to the actual magntudes nvolved. Table 2.5 EU-15: potental mgraton from the eght NMS, 2004-2030 net mgraton foregn populaton 2004 2005 2010 2020 2030 2004 2005 2010 2020 2030 persons Austra 19,983 21,714 8,696 1,706 924 80,681 102,394 172,837 206,605 218,978 Belgum 3,346 3,636 1,456 286 155 13,508 17,144 28,938 34,592 36,664 Denmark 2,937 3,192 1,278 251 136 11,860 15,052 25,406 30,370 32,189 Fnland 3,455 3,754 1,503 295 160 13,949 17,703 29,883 35,721 37,860 France 13,158 14,297 5,726 1,124 609 53,124 67,421 113,803 136,038 144,185 Germany 155,561 169,031 67,691 13,284 7,197 628,065 797,096 1,345,459 1,608,334 1,704,652 Greece 18,174 19,747 7,908 1,552 841 73,374 93,121 157,184 187,895 199,147 Ireland 3,099 3,368 1,349 265 143 12,513 15,881 26,806 32,044 33,963 Italy 25,865 28,105 11,255 2,209 1,197 104,428 132,533 223,709 267,417 283,431 Luxembourg 392 426 171 33 18 1,582 2,008 3,389 4,052 4,294 Netherlands 4,443 4,827 1,933 379 206 17,937 22,764 38,425 45,933 48,683 Portugal 244 265 106 21 11 985 1,250 2,110 2,522 2,673 Span 4,333 4,708 1,885 370 200 17,493 22,201 37,474 44,796 47,479 Sweden 6,154 6,687 2,678 526 285 24,848 31,535 53,230 63,629 67,440 UK 11,617 12,623 5,055 992 537 46,901 59,524 100,473 120,104 127,296 EU-15 272,761 296,378 118,689 23,292 12,619 1,101,249 1,397,627 2,359,127 2,820,052 2,988,936 Sources : Authors' calculatons. See text for assumptons of projecton and extrapolaton. What are the mplcatons of these projectons for the other EU-15 countres? Although tme-seres does not exst for all EU-15 countres, avalable data suggest that the regonal dstrbuton of mgrants across the EU-15 was relatvely stable durng the 1990s and early 2000s. Ths can, nter ala, be traced back to network effects, whch renforce regonal mgraton patterns once they have been establshed. Under the strong assumpton that the regonal dstrbuton of mgrants across the EU dsplayed n Table 2.2 remans constant over tme, we can extrapolate the German results to remanng EU countres. Ths has been done n Table 2.5. Assumng then that all EU-15 ntroduce free movement n 2004, we would get an ntal net mgraton of 270,000 persons from the eght NMS n 2004, and of around 300,000 persons one year later. The long-run mgraton potental s acheved at around 3.0 mllon persons roughly 30 years later. Note that projectons whch are based on

MIGRATION, CO-ORDINATION, FAILURES AND EU ENLARGEMENT 15 econometrc estmates for other countres yeld smlar results: For nstance, Dustmann et al. (2003) estmated n a study for the Home Offce the ntal net mgraton for the Unted Kngdom at between 4,900 and 12,600 persons p.a., whch s n the range of our projectons (11,600 persons). 2.3.2. Is there mgraton dverson after Enlargement? At present, there s lmted nformaton on mgraton trends snce the 1 st of May, 2004. Avalable evdence suggests that the transtonal arrangements resulted n both, the dverson and the reducton of mgraton flows relatve to the case of free labour moblty. Frst, the Home Offce n the Unted Kngdom reports that more that 130,000 natonals from the new Member States have regstered for work between May and December 2004, of whch 40 per cent have been already n the UK pror to Enlargement (Home Offce, 2004b). If these fgures mply that 80,000 persons from the eght new Member States mgrated to the UK n 2004, ths would be more than fve tmes the mgraton potental of the projectons provded n Secton 2.3.1. However, the fgures publshed by the Home Offce cover also temporary mgrants lke seasonal workers, and not all ndvduals who regster do later actually take-up jobs, so that actual mmgraton mght be lower. Yet, gven that the offcal number of natonals from the CEECs has been below 50,000 persons at the begnnng of the 1990s, the fgures publshed by the UK Home Offce pont to a substantal ncrease n mmgraton. For Ireland, a country whch pursues the same mmgraton polces as the UK durng the transtonal perod, we have contradctory nformaton. On the one hand, 7,500 work permts were ssued to natonals from the NMS from January 1 to October 31, 2004, down from 20,000 n the 12 months from January 1 and December 31 n 2003. On the other hand, the Irsh government reports that 31,000 personal publc servce numbers (certfcates whch are necessary for a work permt 6 ) were ssued to natonals from the NMS n the fve months from May 1 st to October 31 st, 2004, pontng to a substantal ncrease n the flows from CEECs relatve to 2003. Thus, t s possble that mgraton flows nto Ireland have been several tmes hgher than predcted n the counterfactual scenaro. Some dverson of mgraton flows from CEECs was also observed n the Nordc countres. In Sweden, the only EU country wthout transtonal arrangements, the number of work permts doubled from 2,097 n 2003 to 3,966 n 2004. However, ths fgure s below the predctons n the counterfactual scenaro (6,200). In Norway, whch partally opened ts labour market and s boomng because of the ol prce hke, the number of released work permts ncreased from 18,170 n 2003 to 25,325 n 2004. 7 Meanwhle the Nordc countres tghtly restrctng mgraton from the CEECs experenced modest or declnng mgraton flows. In Denmark, 2,048 work permts were ssued n 2004. Comparable fgures for 2003 are not avalable here. However, the number s pretty low relatve to the predcted nflow of 3,000 persons. In Fnland, work permts dropped from 6,747 n 2003 to 2,169 n 2004. 8 However, the Fnnsh Mnstry of Labour reports that the number of posted workers has ncreased substantally snce Eastern Enlargement. 6 We are grateful to Frank Berry and Gerry Hughes who provded nformaton on work permts and personal publc servce numbers n Ireland. 7 We are grateful to Jon Erk Dølvk from the Fafo Insttute for Labour and Socal Research for the provson of the data for the Nordc countres. 8 However, the drop n Fnsh fgures can be at least partally explaned by the fact that no work permts are requred for seasonal work below three month snce May 2004.

MIGRATION, CO-ORDINATION, FAILURES AND EU ENLARGEMENT 16 No nformaton s avalable as yet on mgraton to the tradtonal destnaton of mgrants from the CEECs, namely Austra, Germany and Italy, although accordng to statements of Government offcals t would seem that mgraton from the CEECs has been stable after the Enlargement. Overall, the scattered nformaton avalable at the tme of wrtng pont to some dverson of flows from countres tghtly closng borders to countres wth more lberal rules wth respect to mgraton from the NMS. Ths s partcularly true for the Englsh speakng countres, where mgraton fgures exceed by far those of the mgraton projectons. The Eastern Enlargement epsode so far suggests that asymmetres n mgraton restrctons affect the geographcal orentaton of mgraton flows. These dverson effects may become over tme more mportant as networks of ctzens from the CEECs are establshed n the new destnatons, although they are unlkely to become as marked as n the case of dfferences n the enforcement of controls across the US-Mexco border (Boer, Hanson and McCormck, 2001) because there are language barrers n Europe. 3. THE ECONOMIC COSTS OF MIGRATION RESTRICTIONS 3.1. Is mgraton needed n open economes? Insofar as asymmetres n transtonal perods affect the destnaton of flows from the CEECs, they are also lkely to reduce ther magntude. In ths Secton we evaluate these scale effects and provde some estmates of the overall costs of un-coordnated mgraton restrctons. Old and new Member States dffer markedly n terms of factor endowments and factor productvtes. The book value of the physcal captal stock per capta n the NMS s currently about one-tenth of the level n the EU-15. Internatonal mgraton s just one of the potental channels leadng from these dfferences n factor endowments to a new allocaton of resources n the EU-25. The other two channels, trade and captal movements, have already been operatng snce the begnnng of transton to a market economy. The economc mpact of mgraton depends nter ala on whether and to what extent nternatonal mgraton substtutes or complements trade and captal moblty. If mgraton substtutes trade and captal movements, an solated analyss of the effects mgraton tends to overstate ts mpact, snce some of the effects of trade and captal moblty are lost. If t complements trade and captal moblty, an solated analyss of mgraton tends to understate ts effects, snce we have to consder the effects of addtonal trade and captal movements as well. Unfortunately, economc theory does not provde clearcut answers to the queston whether trade and factor moblty are substtutes or complements (see Box 2).

MIGRATION, CO-ORDINATION, FAILURES AND EU ENLARGEMENT 17 Box 2 Does mgraton substtute trade and other factor flows? In the standard Heckscher-Ohln-Samuelson (HOS) framework, trade and factor moblty are substtutes as ponted out by Robert Mundell (1957) almost 50 years ago. Consder the smple case where two countres produce two goods wth two factors and dentcal technologes. If barrers to trade are removed, the captal abundant country wll export goods whch use captal ntensvely n producton, and mport goods whch use labour ntensvely n producton. The prce for the captal ntensve goods wll rse and that for the labour ntensve good wll declne, whle profts wll ncrease and wages fall. The reverse holds for the labour abundant country. Under certan assumptons on technologes and preferences, factor prces wll eventually equalse (Samuelson 1949; Lerner 1952). Hence, no ncentves for factor moblty reman. Analogously, the openng of captal or labour markets wll result n the movement of the abundant factor nto the country where ths factor s scarce, leadng to factor (and good) prce equalsaton as well. Thus, factor moblty reduces the ncentves for trade n ths framework: t s suffcent to open any of the three channels trade, captal or labour moblty to acheve factor prce equalsaton. However, the predctons change sgnfcantly f we relax some of the most restrctve assumptons of the HOS model. The assumptons on technologes are partcularly mportant. If there s a productvty gap between countres, whch holds across all sectors, then trade wll equalse only relatve factor ncomes (Trefler, 1987). Ths means that ncentves for factor moblty wll reman even f all barrers to trade are removed. If nstead cross-country productvty gaps vary across sectors, trade and factor moblty can be complements: Assume for nstance that two countres have dentcal factor endowments, but that one the two has a (Hcks-neutral) productvty advantage n the captal ntensve sector. In ths case ths country wll export captal ntensve goods and mport labour ntensve goods, whch wll rase profts and reduce wages at home, whlst n the other country wages ncrease and profts fall. It follows that trade ncreases the ncentves for labour to mgrate from the country whch has a comparatve advantage n the captal ntensve good to the country whch has a comparatve advantage n labour ntensve producton (Markusen, 1983). Moreover, trade theory has dscussed a number of other cases where trade and factor movements are complements rather than substtutes: In case of specfc factors,.e. factors whch are not moble across sectors, trade may nduce more factor moblty, makng economes more dssmlar, and thus ncreasng even further the trade volume (Venables, 1999; Collns et al., 1999). In trade models wth ncreasng returns to scale and monopolstc competton, factor moblty allow countres to attan economes of scale, ncreasng the real return to the factors of producton at home, and, va ths channel, nduce further factor nflows. Thus, trade lberalsaton n these models stmulates larger mgraton flows (Krugman, 1991; Venables 1999). As economc theory does not provde unambguous predctons, emprcal evdence may offer some gudance. In the ffteen years snce the fall of the Berln Wall, East-West trade and captal flows ncreased dramatcally, wthout reducng dfferences n ncome per capta and factor prces. Between 1988 and 2003, trade between the EU-15 and the NMS grew approxmately by a factor of 6, 9 and the stock of foregn drect nvestment ncreased from vrtually zero to 142.2 bllon Euro (roughly one fourth of GDP n the NMS) by 2003 (Hunya, 2005). Nevertheless, economc convergence s slow: from the through of the transtonal recesson, GDP per capta are recoverng n the regon at an average annual rate 9 Authors calculatons based on the Drecton of Trade Statstcs of the IMF.