Beyond the Big-Bang Enlargement of the EU: Preferences and the Need for Flexibility

Transcription

1 Beyond the Bg-Bang Enlargement of the EU: Preferences and the Need for Flexblty Joachm Ahrens * and Martn Meurers ** The paper addresses the need for nsttutonal change n the European Unon after the envsaged eastward enlargement. Due to the ncreasng heterogenety wth respect to economc structures and performance, poltcal nterests, socal and cultural preferences, and fnancal constrants, governance problems n the enlarged Unon are expected to ncrease sgnfcantly. Based on a theoretcal model of fscal federalsm, the queston on whch level of government poltcal competences should be located s addressed for the case of a unon wth heterogeneous members. The model suggests a dstnct allocaton of competences n dfferent polcy felds dependng on the degree of country heterogenety and the sze of spll-over effects of government servces. In addton, the preferences of the ctzens n the current EU member states and the canddate countres wth respect to centralzaton and decentralzaton of polcy prerogatves are analyzed emprcally. It s nvestgated whether the preferences are consstent wth the theoretcal mplcatons, to what extent there are controverses wthn the two groups, and whether the scope for controverses changes wthn an enlarged unon. In a second step, the preference dstrbuton s used to smulate votng outcomes n the EU Councl wth respect to the queston of centralzaton vs. decentralzaton. Fnally, based on these mplcatons propostons are developed for a more flexble and poltcally feasble ntegraton strategy that mght help to avod a msallocaton of poltcal competences. Keywords: EU-Enlargement, Eurobarometer survey, flexble ntegraton, federalsm, consttutonal reform. JEL-codes: D71, D72, D73, H73, H77 1. Introducton As of May 2004, ten new member states wll jon the Unon, and more countres such as Bulgara and Romana wll follow quckly. 1 Ths bg-bang enlargement wll brng about a sgnfcant ncrease n heterogenety wthn the EU regardng economc structures and performance, poltcal nterests, socal and cultural preferences, and fnancal constrants. It mples that the tenson between deepenng and enlargng the EU wll rse and that governance problems of European polcy makng wll become much more acute. In essence, the EU s future wll crucally depend on ts ablty to cope wth growng conflcts due to dfferent natonal polcy objectves, preferences, and economc constrants. * European Busness School, Internatonal Unversty Schloß Rechartshausen, Oestrch-Wnkel, Germany; e-mal: joachm.ahrens@ebs.de ** fo Insttute for Economc Research, Munch, Germany; e-mal: meurers@fo.de 1 The countres jonng the EU n 2004 nclude Cyprus, the Czech Republc, Estona, Hungary, Latva, Lthuana, Malta, Poland, Slovaka, and Slovena. 1

2 Gven ths outlook, t s of major mportance to systematcally elaborate the nsttutonal framework as well as the poltco-economc modes of acton that shape European ntegraton. 2 Ths paper seeks to contrbute to such a research program. The central questons nclude: How should polcy-makng responsbltes be assgned to dfferent levels of governance n the EU? and How can a more flexble, economcally effcent, and poltcally feasble ntegraton strategy be developed that corresponds to the dverse nature of problems and preferences throughout the EU? The subsequent consderatons are structured as follows: Secton 2 provdes an economc ratonale for assgnng poltcal competences to dfferent levels of governance n the EU. Based on the Eurobarometer survey of EU ctzens preferences, Secton 3 emprcally explores the extent to whch enlargement ncreases the heterogenety n opnons about EU-based decsonmakng. Secton 4 emphaszes the need for a more flexble European ntegraton strategy and dscusses optons for nsttutonal change, whch could enhance the ncentve compatblty and thereby the effcency of European polcy-makng. Conclusons follow n Secton Assgnng poltcal responsbltes to dfferent levels of governance an economc ratonale At present, nether a precse road map nor a consstent method exsts that ams at clearly and consstently assgnng polcy competences or prerogatves among the supranatonal and natonal authortes as well as local governments n the EU (Alesna et al. 2001a). The exstng allocaton of polcy responsbltes has rather emerged spontaneously over tme wthout havng been subject to careful plannng (see Table 1). 2 In ths context, see, e.g., the works of Tsebels and Yataganas (2002), Scharpf (2003), Moberg (2002), and Steunenberg (2001). 2

3 Table 1: Polcy Responsbltes of the EU and ther Extent Extensve Shared Lmted I. Economc and Socal Areas 1. Competton X 2. Cultural Polcy X 3. Regonal Polcy X 4. Employment and Socal Polcy X 5. Enterprse Polcy X 6. Equal Opportuntes X 7. Industral Polcy X 8. Publc Health X 9. Soldarty/Welfare X 10. Consumer Polcy X 11. Monetary Polcy X 12. Educaton, Tranng and Youth X 13. Envronment X 14. Internal Market X 15. Research and Technology X 16. Trans-European Networks/Moblty X II. Sectoral Polcy 1. Agrculture X 2. Fsheres X 3. Transportaton X 4. Informaton and Telecommuncatons X 5. Audovsual Polcy X 6. Energy X III. External Polces 1. Common Foregn and Securty Polcy X 2. Development Polcy X 3. Humantaran Ad X 4. Common Trade Polcy X IV. Justce and Home Affars 1. Asylum, External Borders, Immgraton X 2. Judcal and Polce Cooperaton X 3. Drugs X 4. Trade n Human Bengs Source: Alesna and Waczarg (1999). However, elaboratng a ratonale n ths context s an ndspensable prerequste for mantanng and enhancng the EU s ablty to act wthn an ncreasngly complex, heterogeneous, and conflct-rdden fabrc consstng of 25 or more member countres. Agreeng on such a ratonale appears to be of partcular mportance at a tme when the Unon s enterng a fundamental consttutonal phase n ts development. The European Conventon under the leadershp of former French Presdent Valéry Gscard d Estang has been elaboratng a draft for a future European Consttuton. The draft versons and suggestons of the Conventon, whch became avalable to the publc n the course of 2003, proved to be dsappontng: hardly any nnovatve elements had 3

4 been ncorporated nto the draft verson; nether a clear assgnment of responsbltes nor mechansms n favor of more flexblty and hence problem-solvng capacty of the European ntegraton process were taken nto account. Economc theores of (fscal) federalsm suggest that publc goods ought to be suppled at the lowest level of government that s able to provde these goods cost-effectvely and effcently. 3 For the case of the EU as a federaton consstng of heterogeneous soveregn states, recent works of Alesna and Waczarg (1999) and Alesna et al. (2001b,c), dentfy a trade-off resultng from benefts based on the explotaton of scale economes or the nternalzaton of (cross-country) external effects on the one hand and preference costs on the other. These costs arse from the fact that explotng economes of scale or nternalzng external (.e., European wde) effects necesstates a harmonzaton of polces. A centralzaton of polcy makng, however, makes t ncreasngly dffcult for poltcal authortes to account for dfferent poltcal preferences, economc problems, and socal needs wthn a heterogeneous unon of countres. Ths trade-off between centralzaton and decentralzaton can be llustrated by usng and extendng a model of Alesna et al. (2001b). Consder a group of N countres wth equal populaton that could potentally delegate poltcal powers to a supranatonal level. For each country the representatve utlty functon s gven by: (1) U = c + α H ( g + β g j ), N j= 1, j where c s a prvate consumpton good and g s the total level of per-capta government spendng on a publc good n country. The publc good can be vewed as nfrastructure or publc nvestment but can also have a multlateral character such as custom and monetary unons or free trade agreements. 4 1 θ x The functon H ( x) = captures the so-elastc utlty from the publc good. 1 θ Thus, wth 0< θ < 1, H > 0, H < 0 the regular condtons for decreasng postve margnal x xx utlty are satsfed. The parameter α measures the ndvdual valuaton of the publc good and s restrcted to le between 0 and 1. In the model t serves to express the heterogenety n country 3 See Oates (1999) for an overvew of the key arguments. 4 See Alesna et al. (2001b), pp. 7. 4

5 specfc preferences. The parameter β : β (0,1] captures the spllover effects from other countres government spendng on country. For convenence, the prces of both goods are normalzed to unty. Furthermore, t s assumed that the representatve ndvdual s taxed for the government expendture n a drect manner t = g, such that the government s budget s balanced. 5 Wth constant ncome y ths mples that we can rewrte the utlty as: (1a) U = y g + α H ( g + β g ). N j j= 1, j The queston of decentralzaton versus centralzaton n a unon mples a bpolar decson between ether selectng the optmal provson of g by an ndvdual utlty maxmzaton of countres or by a central-plannng decson that sets a unque level of g = g for every country n order to maxmze the total utlty of the unon. As Alesna et al. (2001b) pont out, ndvdual maxmzaton leads to a Nash-equlbrum that s characterzed by the followng condtons: N g + β g j j= 1, j = α 1/ θ, whch can be solved for the ndvdual levels N g : (2) g N α β = 1 (1 ) 1/ θ 1/ θ α, β β z wth z = 1 + β ( N 1) as a measure for the sze of the spllover. The mportant result s that because of the term varous α /θ 1 the quantty of the publc good N g s a convex functon n the α. Hence, the more heterogeneous the countres are n terms of the varance of 1 gven a constant mean of preferences α µ = α, the lower s the level of publc servces N provded by each country. Obvously, the relatvely hgh supply of the publc servces by some α, 5 The model could be extended to addtonally capture scale economes. One smply has to assume that the publc servce s non-rval, whch would reduce the per capta cost of the publc good to g / P, where P s the populaton of the th country. 5

6 countres wth parameters α close to 1 more than offsets the small amount provded by the countres wth parameters α, close to 0. Ths nduces, e.g., the countres wth average preferences to free-rde on the postve externaltes of others. The centralzed maxmzaton of the unon s addtve utlty (W) s gven by the soluton to max W = α H ( zg) Ng. g Ths leads to a decson at the supranatonal level to provde (3) 1/ θ (1 θ ) / θ α µ g C = z. Note that the level of the publc good chosen at the central level s ndependent from the varance of preferences n the unon. The centralzed decson mples the followng margnal utlty for the representatve ctzens of each country: α α µ U '( g C ) = 1. Ths underlnes that countres wth hgher preference for the publc good than the average α µ would be better off f the quantty provded by the central authorty were larger, and vce versa. In order to calculate the welfare change of each country s ctzens f one swtched from the centralzed to the decentralzed provson of publc servces one can smply take the dfferences n the resultng utltes: 1/ θ C N C C N (4) U ( g ) U ( g ) = α [ H ( α ) H ( zg )] ( g g ) α = 1 θ ( 1 θ ) / θ (1 θ ) / θ N C ( α ( zα ) ) ( g g ) µ. Consequently, for the ndvdual country there s a trade off n swtchng to a natonal decson about the provson of the publc good. If a country s preference for the publc good s relatvely 6

7 low compared to the average of the unon ( α < α µ ), then t looses from the reduced amount of spllovers (frst brackets) but t saves natonal expendture as N C g < g (second brackets), and vce versa. Thus, e.g., n case of a hgh degree of heterogenety n α and a relatvely small spllover effect, countres n general would prefer to have a natonal provson of the publc good. However, under the assumpton that the wnners can compensate the losers t s more approprate to evaluate the decson for decentralzaton from an overall welfare perspectve. Therefore, we need to calculate the average welfare dfference between the two optons: 1 N C 1 C (5) U g U g = 1 1/ θ N z N N g g N ) (1 θ ) / θ 1 1 ( ) ( ) α ( α µ α 1 θ 1 = 1 θ 1/ θ (1 θ ) / θ 1 1/ θ (1 θ ) / θ ( α α ) α α z ω µ θ = z 1 θ z 1 θ z ω µ. z 1 1 1/ θ (1 θ ) / θ α ω + α µ, 1/ θ where α ω = α. Because of the convexty of ths expresson n α, t can be nterpreted as a measure for the degree of heterogenety of preferences gven a constant average of the countres preferences. That s, heterogenety ncreases wth α w. Settng expresson (5) to zero leads to the followng trade-off curve: (6) 1/ θ 1/ θ θα α = z µ ω. z (1 θ ) Fnally, takng the partal dervatve of ths expresson wth respect to β (note that z s a functon dα of β), t can be shown that along the trade-off curve ω > 0 dβ (see appendx). Ths leads to the followng proposton: 7

8 Proposton 1: The decson to centralze the provson of a publc good s subject to a trade off between losses stemmng from heterogenety n preferences opposed to benefts resultng from the nternalzaton of postve spllovers,.e. the more heterogeneous the countres are wth respect to the varance of ther preference parameter α, the larger the spllovers β have to be for centralzed decsonmakng to be welfare mprovng. Thus, economc theory provdes a straghtforward ratonale regardng the assgnment of poltcal responsbltes to dfferent levels of governance wthn a federaton of countres. Accordngly, the EU should be exclusvely nvolved n polcy areas n whch scale economes are relatvely large and the nternalzaton of external effects appears to be mportant. The EU should leave those polcy areas to natonal governments or sub-natonal authortes, n whch the costs assocated wth heterogeneous poltcal preferences, economc structures or vulnerabltes, and socal needs are predomnant relatve to the aforementoned benefts (Alesna et al. 2001a; Ahrens and Hoen 2002). Alesna et al. (2001) dstngush nne domans of polcy makng n the EU, whch are essental for the European ntegraton process (see Table 2). 6 Based on the cost-beneft ratonale outlned above, they dentfy three dmensons n whch polcy prerogatves should be located at the EU level (or even at the global level). These nclude the conduct of nternatonal trade polces, polcng the Common Market, and Non-sectoral Busness Relatons,.e. mantanng compettve markets essentally through ant-trust polcy. Another three polcy dmensons should be covered by shared responsbltes of natonal authortes and the EU: These nclude the realms Internatonal Relatons, the Envronment 7, and Money and Fnance (monetary polcy assgned to the ECB and subsdary polces such as fnancal market regulaton, payments systems, and bank supervson should be centrally undertaken, whereas fscal polcy should bascally be decded upon by natonal governments). Fnally, three polcy domans reman whch should fall nto the exclusve jursdcton of natonal or local governments: Educaton, Research, 6 See Alesna et al. (2001a) for a detaled analyss and dfferentated arguments underpnnng the subsequent classfcaton of polcy domans. 7 Envronmental polces should be assgned accordng to the spatal effects of potental negatve external effects. If externaltes affect only regons wthn one country, the underlyng problem should be addressed natonally; f external effects nvolve two or more EU-countres, supranatonal rulng s adequate (ether by the affected countres governments or by the EU; f negatve externaltes have an mpact at the global level (e.g., global warmng), the soluton needs to be found by the nternatonal communty. 8

9 and Culture, Sectoral Busness Relatons (.e., agrculture and fshng, ndustry, and transportaton), and Ctzen and Socal Protecton (ncludng home affars, cvl rghts, health, labor relatons, justce, consumer protecton, etc.). Table 2: Polcy Domans and Recommended Poltcal Prerogatves Preference Devoluton of Polcy Polcy Domans Externaltes Asymmetry Prerogatves Internatonal Trade Hgh Low EU/Global Common Market Hgh Low EU Busness Relatons (Non-Sectoral) Hgh Unclear EU/Global Internatonal Relatons Medum/Hgh Low Natonal/EU Envronment Medum/Hgh Hgh Natonal/EU/Global Money & Fnance Medum/Hgh Unclear Natonal/EU Educaton, Research and Culture Low Hgh Local/Natonal Busness Relatons (Sectoral) Low Hgh Natonal Ctzen and Socal Protecton Low Hgh Local/Natonal Source: Alesna et al. (2001a). The foregong theoretcal consderatons reveal that the actual allocaton of poltcal competences n the EU (see Table 1) s nconsstent wth the normatve crtera. In some cases such as "Agrculture and Fshng" EU nvolvement has been too extensve, whereas n other cases such as the Envronment EU acton appears to be too lmted. The next secton ams at analyzng the preferences of the EU15 s ctzens and those of the accesson countres wth respect to the assgnment of poltcal prerogatves to dfferent levels of government. Subsequently, t wll be analyzed how the allocaton of poltcal competences would look lke f natonal governments decde accordng to ther ctzens preferences. 3. Emprcal analyss of the desred allocaton of poltcal powers by the ctzens n the EU15 and EU27 The goal of ths emprcal secton s to analyze the atttudes towards EU centralzed decsonmakng of the current EU15 countres and the 10 new members of May 2004 plus Bulgara und 9

10 Romana, whch are lkely to follow soon afterwards (C12). 8 The central questons are 1) whether the preferences for centralzaton n dstnct polcy felds are n lne wth the clear-cut allocaton suggested by the theory of fscal federalsm n the prevous secton; 2) whether despte ths clear gudelne controversal atttudes reman among the countres' ctzens; and 3) whether enlargement changes the scope for controverses and also the outcomes of poltcal decsons wthn the EU's decson framework.in order to systematcally obtan answers, we frst nvestgate the preferences of the two groups separately. Subsequently, we draw conclusons for the jont decson process n an enlarged EU of 27 members. Ths task ncludes the dentfcaton of polcy felds n whch clear majortes exst for ether centralzaton or decentralzaton, and to hghlght the polcy felds where controverses and an mplct manfestaton of country specfc heterogenety can be observed. Beyond the polcy felds n queston, heterogenety s characterzed by the devaton of each group member's preference from the majorty poston about centralzaton versus decentralzaton n the sample, and also by the uncertanty of the ctzens n each country n ther atttudes towards the locaton of decsons n each polcy doman. Ths uncertanty s dsplayed n the varablty of ther answers n the Eurobarometer poll over the past years. Addtonally, we elaborate the possble determnants of the dstrbuton of preferences we observe. Anderson and Kaltenthaler (1996), e.g., found that the length of EU-membershp, the tmng of entry, and natonal economc condtons are the man determnants of the support for European ntegraton. There are probably more plausble factors, lke hstorcal, geographcal, socal or economcal reasons, that mght explan the dfferences across countres. Therefore, f the groups of countres wth shared preferences pont to certan specfc factors that mght be responsble for the preference dstrbuton, we dscuss ths matter explctly. Even though we do not subject our hypotheszed factors to explct statstcal testng, the observed regulartes can offer a frst step towards a deeper understandng of preference heterogenety n the EU15 and the EU27. In addton, they can provde for a startng pont for developng a more flexble nsttutonal framework for an enlarged EU. Our fnal ntenton s to confront the observed preferences and the mpled votng behavor n the Councl of the European Unon wth an economcally justfable degree of centralzaton n the respectve polcy domans. As Alesna et al. (2001a) show, at least for the EU15, the revealed preferences n the poll can be regarded as hghly consstent wth the predctons of a broad range 8 The followng emprcal analyss draws on Ahrens and Meurers (2003). 10

11 of models for an optmal trade off between benefts of scale and the cost of heterogenety. 9 As our results show, the case s dfferent for the C12 countres. Based on our observatons, we provde some reasons for the obvously much more EU optmstc poston of the C12. Most mportant n ths context seems to be the weakness of natonal nsttutons, whch can be emprcally supported by the World Bank Governance ndcators that were compled by Kaufmann, Kraay, and Mastruzz (2003). However, we treat ths matter rather as a sde queston, snce our prmary focus concerns the mplcatons whch the ncreased controverses about the locaton of poltcal powers have for the poltcal decson-makng process wthn an EU27. It turns out that votng n an enlarged EU yelds some majortes wth respect to the degree of centralzaton that clearly conflct wth economc crtera. Thus the emprcs support the need of reformng the EU s nsttutons. 3.1 Preference ndces and the extent of controversal opnons The Eurobarometer poll conssts of a very comprehensve questonnare about EU related poltcal ssues, household decsons, economc expectatons, and overall lfe satsfacton. For the purpose of ths paper we concentrate on a sngle block of questons that refers to 25 polcy areas n whch the EU admnstraton s currently nvolved. The questonnare asks whether ctzens prefer poltcal decsons n each polcy feld at the EU-level or natonal level. 10 For the current member countres the poll s conducted on a bannual bass, and the responses are avalable for a long tme perod startng n Snce we are bascally nterested n the current vews about EU polces, we take the start of the EURO n 1999 as a crucal date, from whch onwards ctzens have lkely become more responsve to European poltcs. Beyond the avodance of too much data mnng, ths relatvely short tme perod mght be reasonable n order to avod havng a tme trend of atttude n the sample, lke Anderson and Kaltenthaler (1996). Thus the data we use n the analyss covers the eght bannual surveys from sprng 1999 to 9 See n addton Alesna and Waczarg (1999) as well as Alesna et al. (2001b). 10 Each survey conssts of approxmately 1000 face-to-face ntervews per Member State (except Germany: 2000, Luxembourg: 600, Unted Kngdom 1300 ncludng 300 n Northern Ireland). It s conducted between 2 and 5 tmes per year, wth reports publshed twce yearly. The exact queston text s: For each of the followng areas, do you thnk that decsons should be made by the (NATIONALITY) government, or made jontly wthn the European Unon. 11 The results of the polls can be downloaded from 11

12 autumn By the tme of wrtng, the Eurobarometer poll has been conducted for the twelve canddate countres (C12) only twce n 2001 and Ths tme frame of two years for the C12 can be suspected to be too short to measure stable preferences. But as ponted out n the emprcal analyss, the standard devatons of the preference ndces, that we calculated, are most of the tme even smaller than those of the EU15 sample. We begn our analyss of the dfferences n EU preferences wth respect to the ndvdual polcy felds (J = 25). For each of these polcy felds the Eurobarometer data provdes the percentage of respondents n each country who prefer natonal competences (%NAT) and who prefer EU competences (%EU). In order to characterze the preference of the voters of each member and canddate country (=1,..,15, and =1,..,12, respectvely) over the polcy domans (j=1,..,j), we calculate a smple ndex: 13 1 X, j = 100 (% NAT, j, t % EU, j, t ), t = 1,..., T, T t where the number of observatons s T = 8 for the EU15 and T = 2 for the C12 countres. The correspondng standard devaton S,j s then: S = Var(% NAT, j, t ) + Var(% EU, j, t ) 2Cov(% NAT, j, t,% EU, j, )., j t A frst ndcator for domnatng preferences wthn the EU15 and the C12 sample can be obtaned by countng the number countres for whch the ndces X,j are sgnfcantly below or above zero accordng to a smple t-test. From ths we obtan a frst crude mpresson about the polcy felds wth a clear majorty of countres for ether centralzaton or decentralzaton. The dfference D of sgnfcant pro EU votes and sgnfcant pro natonal votes provdes for a sutable comprehensve ndcator n the further analyss. A second measure for the preferences n the EU15 and the C12 group, respectvely, s the lkelhood of a 50% majorty L50 for centralzed decson-makng f each country had one vote. 12 C12 nclude the ten countres, whch wll jon the EU n May 2004 plus Bulgara and Romana. Turkey s also ncluded n the survey. Nevertheless, we dd not consder Turkey n our analyss snce t s far from havng a concrete tme schedule for entry n the EU. 13 The latest three Eurobarometer waves also nclude questons about the polcy felds terrorsm and ageng. Snce these questons were not part of the canddate poll, they are not consdered n the analyss of ths paper. 12

13 Ths s obtaned by smulatng 10,000 random draws from a Bernoull dstrbuton over 1 for centralzed, and 0 for decentralzed decsons, where the probabltes are derved from hypotheszed normally dstrbuted preferences of each country based on the prevously calculated mean ndex X,j and ts standard devaton. Subsequently, t s checked for every draw whether a majorty of countres exsts votng n favor of centralzaton. These majortes are counted and L50 s fnally the percentage of majortes for centralzed polces of the total number of experments (10,000).As an addtonal pece of nformaton for an evaluaton of the robustness of these two ndcators, for each polcy feld j we compute the average over the ndces X,j for the two groups ( ( S j ). 14 X j ), respectvely, and also the standard devaton of the ndces wthn the groups Beyond the preference for the level of decson-makng n certan polcy domans, the extent of controverses mght be reflected n further dmensons: frst, n the overall atttude of the ndvdual countres towards the EU, whch can be quantfed smply by the number of polcy felds for whch country 's ctzens would opt for the EU level ( X 0 ), and second, n the devatons of an ndvdual country from the preferences of the ctzens of the other countres (.e., X, j has a dfferent sgn than k j, j < X, for a country k ). To obtan a quanttatve measure for a country's preferences n relaton to the others, we count the number of countres wth the same preference as country (#G,j ) and relate t to the number of countres wth the majorty poston (#M,j ). The degree of accordance wth the majorty poston can then be summarzed n the followng ndex: A = 1 J j # G, j + 1. # M j 14 It wll become evdent from Tables 3 and Table 5 that t would be much more dffcult to nfer the scope of potental majortes by only lookng at the average of the preference ndces. Especally the lkelhood measure for a 50% majorty provdes a useful tool n drawng borderlnes. 13

14 3.2 Preferences of ctzens wthn the EU15 We start the dscusson of our emprcal results wth the analyss of the preferences wth regard to the 25 polcy felds. The rankng n Table 3 s compled by sortng the polcy felds accordng to preference measures gvng prorty to D, L50, and then to the mean of the country specfc ndcators. In order to better understand the motves for the observed dstrbuton of preferences, t s also useful to look at the par-wse correlatons of the ndcators for dstnct polcy felds, e.g., between X,j and X,k for j k, k J. 15 In some cases, the close relatonshp between certan polcy felds mght offer a frst tentatve ndcaton of economc or poltcal consderatons that are responsble for the artculated preferences. Another nterestng aspect s, whch coaltons of countres preval especally for the polcy felds wthout a clear majorty. Ths can gve us some gudance about geographcal factors or past poltcal experences, whch mght cause common preferences of certan country groups. Remarkably, for 11 of the 25 polcy felds a clear majorty of countres favors a decson process located at the EU level. The desre for centralzaton s the strongest wth respect to polces aganst the Explotaton of Human Bengs. Lookng at the correlatons of the preferences for ths polcy feld wth those for the other polces, we observe a hgh correlaton wth Currency, Info about EU, Foregn Polcy, Organzed Crme and Drugs that are also strongly preferred at the EU level. Ths agglomeraton shows that the preference for centralzaton mght be a reflecton of a multtude of nterrelated concerns about human rghts, economc wealth and securty. Beyond the correspondence wth the normatve consderatons about an economcally meanngful assgnment of responsbltes, ths result mght also represent a common desre for protecton and coordnated decsons to cope wth the uncertantes created by a qucker pace of global changes. The almost unanmous preference for EU-led Regonal Ad seems to be uncorrelated wth preferences over other domans. Obvously, a great majorty of countres regards the past progress of some of the poorer regons (Portugal, Greece, Span) as a persuasve argument for the contnuaton of such a polcy. 16 In lght of the latest poltcal controverses n connecton wth the mltary acton n Iraq, the strong desre for a common Foregn Polcy seems noteworthy. We postpone a tentatve nterpretaton of ths result to the 15 The complete correlaton matrx s shown n the appendx. 16 Alesna et al. (2001a, p. 10) nterpret ths result as a bas n the answers of the poll by the presumpton that transferrng polcy responsblty to the EU may result n net benefts for those countres. In other words, the 14

15 dscusson of the preferences for defense polces. As ndcated by the correlaton matrx, the preferences for EU coordnaton over Envronment, Humantaran Ad, and Poverty and Socal Excluson are all nterrelated. Hence, the EU perspectve can possbly be attrbuted to the fact that ctzens recognze potental coordnaton falures and free rder problems of natonal ntatves. Research seems to be a smlar matter. Its strong correlaton wth currency mght reflect economc consderatons that EU coordnated research mght provde for the explotaton of scale effects. Table 3: Preference ndces over polcy felds n the EU15 sample Polcy doman D L50 X j S j Explot. human bengs % Informaton EU % Regonal ad % Foregn Polcy % Organsed crme % Research % Drugs % Humantaran ad % Poverty/socal excluson % Currency % Envronment % Mean EU % Agrc.& Fshng % Poltcal asylum % Immgraton % Defence % Acceptng refugees % Unemployment % Mean undecded % Juvenle crme % Urban crme % Cultural polcy % Meda % Justce % Health and socal welfare % Educaton % Polce % Mean Natl % Mean Total % response n the questonnare may reflect the percepton of personal or country gans rather than a far judgement about the optmalty of allocatve crtera. 15

16 There are altogether only sx polcy felds n whch the EU15 members' preferences are controversal. As far as the nterrelated felds Poltcal Asylum, Acceptng Refugees and Immgraton are concerned, there are only four countres n favor of a EU decson level, namely, Belgum, Span, Italy, France (wth respect to poltcal asylum and acceptng refugees). The other end of the poltcal spectrum wth countres opposed to centralzaton s represented by Denmark, Fnland, Sweden, the U.K., Austra (wth respect to mmgraton and acceptng refugees) and Ireland (only wth respect to acceptng refugees). For polces related to Agrculture and Fshng, the proponents of EU decsons are manly located n Belgum, Germany and the Netherlands, whereas the opposton conssts manly of the Scandnavan countres, Fnland and Sweden, and the U.K. In the past, the former group benefted from EU polcy measures when t came to common market rules, qualty standards and control (e.g. preventng the spread of anmal dseases). In contrast to ths, the two Scandnavan countres and the U.K. mght fear that such polces are used as measures of protectonsm and could potentally harm the natonal agrcultural sector. Most nterestng are the preference dstrbutons over a common Defense Polcy and EU measures aganst the Unemployment problem. Intutvely, foregn polcy and defense polcy should be strongly connected. Ths s underpnned by the correlaton coeffcent of the preference ndces of 85%. The preferences over foregn polcy, however, are strongly correlated wth those of currency, justce, humantaran ad, and Info about the EU. Thus, beyond a pure mltary dmenson, the clear preference for a common foregn polcy seems to reflect a rather general desre for a strong poston of the EU n the globalzaton process. Furthermore, t should be consdered that defense s already part of the NATO framework, and the open debate about ts future role mght contrbute to the heterogeneous atttudes wthn the EU countres. 17 The queston of common measures aganst the unemployment problem can be regarded as even more controversal. 18 Only Italy s clearly n favor of EU polces. Denmark, Fnland and the U.K. are obvously more satsfed wth natonal measures. 17 Interestngly, the aggressors of World War II (Germany and partly Italy) and ther mmedate neghbors Belgum, Luxembourg, Netherlands are relatvely commtted to a EU defense polcy. On the other hand, countres that were nvolved n natonal and nternatonal mltary conflcts snce the War, lke Ireland, the U.K., and Greece favor natonal decsons n ths doman. 18 Despte a majorty of three sgnfcant votes aganst EU level decsons, there exsts a hgh probablty for a majorty of countres n favor EU admnstraton. Ths can be explaned, by the fact that ctzens of only 5 countres have a dstnct (sgnfcant) poston regardng ths ssue. 16

17 [nsert Fgure 1 about here] The correlaton between the preference for EU decsons and the ten-years average of the unemployment rate n the member countres n Fgure 1 suggests that the falure of natonal polces n the past s lkely to account for ths dstrbuton of preferences. The outlers Span and Fnland can be explaned by the fact that despte persstent hgh unemployment, people seem to acknowledge the past success of natonal authortes n reducng unemployment sgnfcantly. 19 The example of Fnland underlnes the preference of Scandnavan countres for subsdarty that becomes apparent n connecton wth the other polcy felds and that s a lkely result of the natonal tradton of decentralzed admnstraton. 20 From the observaton of the country specfc dfferences n opnons we can draw the followng ntermedate concluson: Beyond Anderson s and Kaltenthaler s (1996) result that the tme of EU membershp s mportant (whch mght account for the Scandnavan countres' reservatons about EU based decsons), the preference for EU nsttutons seems to depend also on the geographcal condtons to protect natonal borders (as ndcated by the controverses about mmgraton, etc.), past costs and benefts of EU polces (agrculture and fshng), and fnally the trust n the capablty of EU nsttutons to handle problems more effectvely than natonal authortes (labor market polces). 21 The eght domans wth clear preferences for natonal decson-makng essentally represent ssues of natonal dentty, lke Cultural Polcy, Meda, and Educaton or can be clearly regarded most effcent when decentralzed decsons are taken, lke Polce, Urban and Juvenle Crme, and Justce, because local coordnaton of actons and nformaton gatherng are requred. Health and Socal Welfare can be regarded as matchng both crtera. Correspondng to the separaton of polcy felds nto the three groups, the standard devaton of the country preferences s hghest on average for those polcy felds where a clear majorty does not preval. If the standard devaton s nterpreted as an addtonal measure for heterogenety, 19 In the early 1990s, Span had unemployment rates of around 20%; subsequently s was reduced to roughly 10% n In Fnland unemployment was reduced from a peak of 17% n 1994 to roughly 9% n The strong overall preference for respectng subsdarty s hghlghted by the European Economc Advsory Group; see EEAG Report (2003), p The U.K., Sweden, Denmark and Austra are at the lower end when t comes to Trust n the EU nsttutons and bodes accordng to the latest Eurobarometer survey n autumn 2002; see European Commsson (2002), p

18 then, ndependent from the majorty stuaton, one can characterze the scope for controverses that are stll present among the EU15 members. Notwthstandng the clear majortes, preferences seem stll to be most dsparate about a common currency and health and socal welfare polcy, as s ndcated by relatvely hgh standard devatons of 30 and 25 ndex ponts. The standard devatons wll be an mportant element n the comparson of the heterogenety between the EU15 and the EU27 sample that s dscussed n the next secton. Now, we turn to the analyss of heterogenety n the dmenson of country specfc EU optmsm or skeptcsm and of the countres ctzens' preference relatve to the majorty wthn the respectve EU15 group. The majorty of countres n the EU15 sample s n favor of centralzng 13 polcy domans (the frst 11 domans of Table 3 plus agrculture and fshng as well as unemployment) and of decentralzng the rest. Table 4: General atttude of the EU15 towards centralzaton Polcy felds Devatons Accord ndex wth EU from majorty S preference poston GER IRL P DK L A B F FIN UK S NL E GR I Mean Mn Max Table 4 shows that the center preference s that of Germany and Ireland whch both prefer an allocaton of powers dstnct from the majorty poston only n a sngle polcy feld (Germany wth respect to Defense Polcy and Ireland wth respect to Unemployment). Due to ts poltcal capacty, Germany can be consdered as the EU' s poltcal gravty center. Closely attached wth ther preferences are the four smaller countres Ireland, Portugal, Luxemburg, and Austra. 18

19 Belgum and France consttute another powerful poltcal pole of jont preferences, however, wth a much stronger emphass on EU centralzaton than the Germany group, especally n the aforementoned felds of Immgraton, Poltcal Asylum, and Refugees. Dutch ctzens are closer to ths group than to the German Group. Only wth respect to the unemployment problem, they prefer natonal decsons, whch can be easly explaned by the relatve success of natonal labor market polces. Skeptcsm about the EU seems to be most apparent n the U.K. and n the Scandnavan countres Denmark, Sweden and Fnland, where ctzens prefer natonal decsons n the majorty of polcy ssues. A fourth group s consttuted by Span, Greece, and Italy. These countres are agan qute optmstc about the hgher effectveness of EU-located decsonmakng, but seem to prefer centralzed decsons n many polcy felds, where the other two gravty centers, despte ther broad EU commtment, would prefer natonal admnstraton. Ther common characterstc s the preference of EU polces wth respect to juvenle crme, whch can be seen as an ndcaton that natonal authortes do not succeed n ths doman and consequently the hopes for an mprovement are attrbuted to the EU. Ths lne of reasonng receves support from the specfc case of Italan ctzens, who seem to be overly EU optmstc n preferrng natonal decson competence only for ssues such as polce, urban crme and the meda. 22 Emprcal support for ths argument can be obtaned through a consultaton of the World Bank's Governance Indcators. 23 Fgure 2 shows the scores for three ndcators that can be consdered to represent the capacty and competence of natonal governments. 24 Concurrent wth the above reasonng, the governance scores for Italy, Greece and also Span are at the bottom end of the EU15 sample. 22 Ths strong protrudng support for a shft of poltcal responsbltes to the EU-level was already present n the early 1990s ahead of the preparatons for the common market; see Kuntze (1990), p The ndcators are compled as percentle ranks ndcatng the share of countres n a worldwde perspectve that are rated below a partcular country. See Kaufmann, Kraay, and Mastruzz (2003), 24 For a dscusson of the measurement problems of these and other governance ndcators, see, e.g, Ahrens and Meurers (2002). 19

20 Fgure 2: Natonal governance n the EU15 countres AUSTRIA BELGIUM DENMARK FINLAND FRANCE GERMANY GREECE IRELAND ITALY LUXEMBOURG NETHERLANDS PORTUGAL SPAIN SWEDEN UNITED KINGDOM EU15 Government Effectveness Regulatory Qualty Control of Corrupton Average Governance Source: Kaufmann, Kray, Mastruzz (2003) Furthermore, Table 4 reveals that there s a negatve correlaton between accordance wth the majorty and the duraton of membershp n the EU, e.g., f one looks at the ndcators for Greece, Austra, and the Scandnavan countres. The latent EU- and Euro-skeptc Brtsh certanly provde an excepton from ths rule. Regardng the uncertanty n the preferences of the countres' ctzens, the proposed ndcator S, whch s an average of the S, for all polcy felds, seems to be ndependent of overall EU support or the extent of devaton from the majorty poston. The only regularty that one mght recognze s that smaller countres lke Portugal, Greece, Austra, and the Netherlands tend to have a greater varablty n the preferences of ther ctzens. Fnally, note that the number of devatons from the majorty and the accord ndex lead to almost dentcal conclusons. j 20

21 3.3 Comparson wth the preferences of the new member states Table 5 summarzes the preference ndcators of the C12 countres wth respect to the 25 polcy felds. Obvously, the canddate countres have a much stronger preference for EU-level decsons than the current member countres. All sx polcy-felds that are controversal among the EU15 are clearly preferred at the central level. Moreover, the four polcy domans Table 5: Preference ndces over polcy felds n thec12 sample Polcy doman D L50 X j S j Explot. human bengs % Drugs % Organsed crme % Informaton EU % Research % Humantaran ad % Regonal ad % Foregn Polcy % Poverty/socal excluson % Unemployment % Envronment % Currency % Defence % Juvenle crme % Health and socal welfare % Immgraton % Poltcal asylum % Educaton % Acceptng refugees % Agrc.& Fshng % Urban crme % Mean EU % Justce % Polce % Mean undecded % Meda % Cultural polcy % Mean NAT % Mean Total %

22 Health and Socal Welfare, Juvenle Crme, Urban Crme and Educaton, that are overwhelmngly preferred at natonal levels by the EU15, are also ncluded n the set of polcy felds that C12 countres would lke to have admnstered by European nsttutons. The parallel to the preferences of Italan, Greek and partly Spansh ctzens ndcates that the skeptcsm about the competence and capactes of natonal authortes to provde suffcent securty and transparency are the lkely causes for ths supranatonal tendency. Agan, ths can be underpnned by a consultaton of the World Bank Governance Indcators. In comparson wth the data presented n Fgure 2, the governance ndcators for the C12 countres are sgnfcantly below the EU-average of roughly 92, and only Malta and Cyprus le sgnfcantly above the average of ndcators for Greece, the country wth the weakest nsttutonal performance n the current EU. Fgure 3: Natonal governance n the C12 countres Level of Greece Government Effectveness Regulatory Qualty Control of Corrupton Average BULGARIA CZECH REPUBLIC ESTONIA HUNGARY LATVIA LITHUANIA POLAND ROMANIA SLOVAK REPUBLIC SLOVENIA MALTA CYPRUS Source: Kaufmann, Kray, Mastruzz (2003) Controverses among the canddate countres reman about the assgnment of Polce and Justce. Unfortunately, the exstng allances wth Malta, Estona, and Bulgara n favor of natonal decsons and Romana, Slovaka, and Slovena n favor of an EU locaton do not provde for a startng pont to nfer any reason for ths preference dstrbuton. Fnally, the canddate countres share the preferences wth the current members for decentralzaton only wth respect to Meda related and Cultural Polces. 22

23 The only major dfference between the correlaton matrces for the EU15 and the C12 sample can be found for Agrculture and Fshng. The EU15 assocate these polces wth envronmental ssues and defense, whch seems to stress the potental role of agrcultural und fshng polces n preservng the qualty of nature. The C12, however, regard ssues of agrculture and fshng to be closely lnked wth ther economc well-beng as the hgh correlatons wth unemployment, health and socal welfare, and poverty and socal excluson suggest. Comparng the standard devatons of the preference ndces of the two groups (last column n Tables 3 and 5), one cannot fnd a notable dfference of the mplct country specfc heterogenety of the preferences wthn the two groups. There s even a slght tendency for tghter atttudes of the canddate countres. Interestngly, n both groups the preferences seem to be most dsparate wth respect to a common Currency. The hgh wthn-group heterogenety of the EU15 sample wth respect to Defense and Immgraton s not shared by the sample of canddate countres. Table 6: General atttude of the C12 towards centralzaton Polcy felds wth EU preference Devatons from majorty poston Accord ndex Czech Republc Cyprus Poland Latva Lthuana Hungary Slovaka Estona Romana Slovena Bulgara Malta Mean Mn Max S The smlar degree of wthn-group heterogenety between the EU15 and the C12 s also reflected n the country specfc preferences of the C12. The majorty poston of ths group s n favor of centralzng the frst 21 polcy domans of Table 5 and decentralzng the 4 domans Justce, 23

24 Polce, Meda, and Cultural Polcy. Ths poston s not represented by the ctzens of a specfc country, but, accordng to the number of devatons, the Czechs and Cyprots come closest to t (Czechs addtonally prefer ssues of Justce at the EU level, whereas Cyprots addtonally prefer meda-related polces by the EU). The ctzens of the other countres down the lst to Romana are n close vcnty of these postons. They prefer the one or the other polcy feld more or less at the natonal or the EU level, wthout any sgnfcantly dfferent overall EU confdence. Romana and Slovena consttute the hgh end wth complete commtment to EU nsttutons. A more skeptc vew about the EU seems to persst n Bulgara and Malta, snce these countres lke to have a much hgher number of polcy felds kept wth natonal authortes. Comparng the summary statstcs, the smlarty of wthn-group heterogenety to the EU15 sample s confrmed. The mnmum and the maxmum number of devatons from the majorty poston are wth 3 to 8 only slghtly smaller than n the EU15 sample, and the accordance ndces have exactly the same mean, wth almost dentcal upper and lower bounds. We can summarze the results of the emprcal analyss as follows: The canddate countres are clearly more enthusastc about centralzed EU decsons than the current member states. Ths s reflected n the canddates preference for EU decson-makng n 21 of the 25 polcy felds n comparson to the much smaller set of 11 polcy domans favored by the EU15 countres. A natural explanaton for ths result s the canddates desre to have certan ssues admnstered by EU nsttutons, where natonal authortes do not have the capacty yet. Ths also ncludes polcy felds, lke Health and Socal Welfare, Educaton, as well as Juvenle and Urban Crme, for whch decentralzed admnstraton s clearly more effcent. Obvously, the ctzens n the canddate countres prefer rather a slghtly neffcent EU admnstraton of polces for securty and an mprovement of lvelhood than relyng on ther natonal poltcs and bureaucracy. As ponted out by Alesna et al. (2001a), the preferences of the current EU members are on average n lne wth an economcally ratonal allocaton of poltcal domans to natonal and supranatonal decson levels. In addton, our analyss reveals that the preferences possbly reflect a demand for common EU polces and a jont EU poston n response to rapd global changes, whch s ndcated by the correlaton of the preferences over a mxture of polcy domans that ncludes economy, securty and nternatonal relatons. Controverses among the EU15 are most dstnct wth respect to the polcy cluster of Immgraton, Asylum and Refugees, Defense and Labor market polces. The reasons for ths preference dstrbuton can be manly 24

25 seen n geographcal characterstcs, past poltcal experences and the perceved falure of natonal polces. Man poltcal poles can be seen n France, Belgum, Italy and Span, wth ther clear preference for centralzed decsons regardng the "mmgraton cluster" n opposton to the U.K and the Scandnavan countres, whch, n addton, prefer decentralzed decsons wth respect to labor market, agrculture and fshng, and defense polces. Thus, n spte of possble constrants by the assgnment of votng powers, the ncluson of the C12 countres wth ther clear bas towards centralzaton tends to ncrease the dffculty for the latter fracton to nfluence EU poltcs accordng to ther ctzens' preferences. A fnal pont beyond the smple bpolar queston of centralzed versus decentralzed decsons s the mplct country specfc heterogenety that s reflected n the ndcators S j n Tables 3 and 5. Notably the hgh standard devaton of the preference ndces for Currency n both groups could be an ndcator for a persstent heterogenety n the preferences regardng an approprate monetary polcy from each country's perspectve. Therefore, especally n ths doman, jont EU decsons are lkely to become more dffcult to acheve n an enlarged EU. 3.4 Outlook for future decson-makng n the EU An mportant queston that arses from the former analyss s whether the ncreased potental of controverses n an enlarged EU leads to a smaller scope for majortes wthn the actual decsonmakng nsttutons. In the Unon's current nsttutonal framework, the the Councl of the European Unon can be consdered as the preemnent platform for legslatve decsons. Evdently, the queston of deepenng EU-admnstraton n certan polcy felds, whch has been at the center of the prevous dscusson, wll not be drectly subjected to a Councl vote. Nevertheless, there wll be numerous decsons on specfc matters that could brng about more EU nvolvement and centralzaton. Stll, as ponted out by Garett and Tsebels (1996), the focus on Councl votng s only a crude smplfcaton of the power structure wthn the EU. In realty, a much more complcated nteracton takes place between the European Parlament, the European Commsson, and the Councl. Therefore, we consder our results not as a defnte characterzaton of decson processes but rather a possble ndcaton for outcomes conflctng wth economc ratonalty. 25