Tradable Refugee-Admission Quotas and EU Asylum Policy

Transcription

1 Tradable Refugee-Admsson Quotas and EU Asylum Polcy Jesús Fernández-Huertas oraga Hllel Rapoport CESIFO WORKING PAPER NO CATEGORY 1: PUBLIC FINANCE NOVEBER 2014 An electronc verson of the paper may be downloaded from the SSRN webste: from the RePEc webste: from the CESfo webste: Twww.CESfo-group.org/wpT

2 CESfo Workng Paper No Tradable Refugee-Admsson Quotas and EU Asylum Polcy Abstract The current EU Asylum polcy s wdely seen as neffectve and unfar. We propose an EUwde market for tradable quotas on both refugees and asylum-seekers coupled wth a matchng mechansm lnkng countres and mgrants preferences. We show that the proposed system can go a long way towards addressng the shortcomngs of the current system. We llustrate ths clam usng the recent problems regardng relocaton faced by the European Relocaton from alta EUREA program. JEL-Code: F220, F500, H870, I300, K330, O190. Keywords: mmgraton polcy, EU polcy, tradable quotas, refugee resettlement, asylum seekers, nternatonal publc goods. Jesús Fernández-Huertas oraga FEDEA adrd / Span jfernandezhuertas@fedea.es Hllel Rapoport Pars School of Economcs Unversté Pars 1 Panthéon-Sorbonne Pars / France hllel.rapoport@psemal.eu October 31, 2014 Ths paper was ntated at the graton Polcy Centre European Unversty Insttute as part of ts efforts to contrbute to the current debate on soldarty and asylum polcy n the EU. A non-techncal verson of ths paper appeared as RSCAS Workng Paper No. 2014/101 Rapoport and Fernández- Huertas oraga, We partcularly thank Phlppe De Bruycker for hs careful comments and suggestons, as well as Raner Baubock, Phlppe Fargues, Bernard Hoekman, Llan Tsourd, Patrca Van de Peer and Alessandra Venturn. We also thank Tm Hatton, Eko Thelemann, and partcpants at the CESfo Conference on Internatonal graton, unch, and at the graton Polcy Centre Semnar on Soldarty and Asylum Polcy of the European Unon at the EUI n Florence for ther comments and suggestons; and Suln Sardoschau for excellent research assstance. Jesús Fernández- Huertas oraga receved fnancal support from the ECO and ECO projects funded by the Spansh nstry for Economcs and Compettveness. Hllel Rapoport acknowledges support from the Pars School of Economcs Appel à projets 2013 and from CEPREAP.

3 1 Introducton The European Unon s currently experencng a refugee crss: thousands of refugees are accumulatng on ts shores Lampedusa, alta, Scly and land borders Greece, Bulgara, fleeng cvl war n Lbya or Syra, armed conflct or oppresson. Even f these numbers are small n comparson to those of refugees hosted by States of frst asylum such as Lebanon, Jordan or Turkey, ths puts pressure on countres of frst arrval, wth thousands of persons then wanderng n the EU-Schengen space and beyond. At the same tme, the exstng European asylum polcy s overwhelmngly judged as napproprate and s crtczed not just because of ts nablty to address the challenges posed by the volume of refugee flows but also due to the many legal defcences, poltcal nconsstences and economc neffcences that characterze the current asylum system. At a legal and poltcal level, the Common European Asylum System CEAS, launched n 1999, s ncreasngly under fre. For one thng, the so-called Dubln-system mplemented snce 1997 whereby an asylum seeker s manly under the responsblty of the country of frst-entry, s more and more regarded as ll-conceved e.g., De Bruycker et al., Under ths system, an asylum seeker who entered the EU say n Greece and got arrested for llegal stay say n Belgum could be transferred back to Greece. Such a system creates dsncentves for the Greek government to effectvely mplement the CEAS norms. Furthermore, the Dubln system prevents asylum seekers from choosng ther country of asylum wthn the European Unon, creatng ncentves for them to crcumvent legal restrctons to moblty before ther asylum clams have been examned. A second major ssue wth the current system s that of burden sharng. 1 Indeed, the current system places at least n theory a dsproportonate burden on the countres of frst entry such as alta, Greece or Italy that are responsble for many asylum seekers due to ther geographc poston. The Gudelnes adopted by the European Councl on 27 June 2014 underlne that the Unon needs an effcent and well-managed mgraton, asylum and borders polcy, guded by the Treaty prncples of soldarty and far sharng of responsblty, n accordance wth artcle 80 of the Treaty on the Functonng of the European Unon and ts effectve mplementaton gudelne 5 and that The Unon s commtment to nternatonal protecton requres a strong European asylum polcy based on soldarty and responsblty 1 Ths termnology s controversal but we keep t here because of ts wdespread use. Some authors, and the European Councl 2014, prefer to use the concept of responsblty sharng. 2

4 gudelne 7; European Councl, Under ths general process, many attempts of mprovements have been ntated: further harmonzaton of asylum law, creaton of a European agency as the European Asylum Support Offce EASO, contnuaton of EU fundng through the new Fund for Asylum, graton and Integraton, relocaton of refugees across recevng countres on the last ssue, see the alta example below. However, t s far to say that progress n practce has been rather lmted see Hatton, 2011, 2012, 2013; Thelemann et al., In ths paper we propose a new perspectve usng market-based solutons to deal wth the externalty/free rdng problem among EU ember States, takng advantage from the already exstng nsttutonal frameworks. We apply the dea of tradable mmgraton quotas proposed by Fernández-Huertas oraga and Rapoport 2014 to the case of refugees and asylum seekers n the EU context. Indeed, provdng refugee protecton and asylum s a well-recognzed nternatonal publc good, and there s an understandng that the costs ncurred should be shared more farly, at least wthn the EU. Buldng on ths dea, we explore the potental for a Tradable Refugee Quotas TRQs system to better coordnate natonal asylum polces. We supplement ths system wth a matchng mechansm allowng for takng refugees preferences over destnatons as well as countres preferences over refugees types nto account. Ths matchng component, adapted from the more generc model of Fernández-Huertas oraga and Rapoport 2014, s shown to address many of the ssues that have been rased aganst the current EU Asylum Polcy. The orgnal paper by Fernández-Huertas oraga and Rapoport 2014 already ntroduced an applcaton to the resettlement of long-standng refugees. The present paper extends the argument and shows that the same reasonng can be appled both to refugees and asylum-seekers arrvng at the EU. Refugees and asylum-seekers can be nterpreted as two dfferent mgrant types n the orgnal model. The rest of the paper s organzed as follows. In Secton 2 we provde the emprcal background for the dscusson as well as evdence of unfar burden sharng among EU ember States. Secton 3 provdes the polcy background n that t descrbes generally the evoluton of the EU asylum polcy over the last two decades and emphaszes the search for harmonzed polces. Ths s followed by an outlne of the exstng lterature on marketbased proposals n the feld of refugee polcy n Secton 4. In Secton 5 we present the model, frst ntroducng a model of tradable refugees quotas TRQs wth refugees beng ndfferent 3

5 among the varous potental host countres and host countres beng ndfferent about the type of refugees they receve. We then ntroduce heterogenety n refugees preferences over ther destnatons and n host countres preferences over refugees types. Secton 6 serves as an llustraton for possble mplementaton buldng on the well-documented example of alta and concludes the paper. 2 The Numbers The EU receves around 300,000 asylum clams per year, about one thrd of the total regstered n the world see Fgure 1. ore than three quarters 79% of the stock of asylum seekers n the EU n 2012 accumulated n only sx destnatons countres: Germany, France, Greece, Austra, UK and Sweden. A smlar pcture emerges from the 2012 numbers on refugees n Europe see Fgure % of the 1.3 llon refugees n the EU are hosted by Germany, France, UK, Sweden, Italy and Austra and 44% of them resde only n Germany. Fgure 1: Source: own calculatons on end-of-year pendng applcatons from UNHCR Refugees are ether people who came to Europe as asylum seekers and were granted asylum, or people who were n refugee camps outsde of the EU, granted refugee status by the UNHCR and successfully appled for resettlement n one of the EU countres. 4

6 Fgure 2: Source: own calculatons from UNHCR 2013 The allocaton of mmgrants wthn the European Unon stll seems to present tself as a black box. The lack of nformaton on the moblty of non-eu natonals wthn the European Unon has made t dffcult to trace back the mechansms under whch refugees and asylum seekers locate themselves. For example, n a recent report, the European graton Network documents recent trends of ntra-eu oblty of thrd-country natonals European graton Network, However, those estmates are not able to capture non-eu ctzens that are n a state of transton regardng ther legal status. Often, those mmgrants vansh from the offcal fgures. Ths holds especally for rejected asylum seekers wllng to avod removal. For example, only about one fourth of the 22.5 thousand asylum applcaton decsons n Italy n 2012 were postve UNHCR, 2013, the rest beng rejected. Indeed, the accumulaton of mmgrants n certan locatons s dsproportonal to the capactes of the host countres. In ther Statstcal Yearbook, the UNCHR ranks host countres regardng certan ndcators: the refugee to GDP rato, refugee per 1000 nhabtants or per 1000 square klometers. Even among the bg sx host EU ember States there are tremendous dfferences regardng ther fnancal, demographc and geographc capactes. Whereas Germany s refugees to GDP PPP per capta rato amounts to $15.31, comparable countres such as France have a rato of only $6, or even as low as $0.57 n the case of 5

7 Belgum UNCHR, Compared to ther geographc and populaton sze Germany, Sweden and Swtzerland host a relatvely large number of refugees. These numbers reveal that the allocaton of refugees does not reflect the far burden sharng concept put forward by the European Unon, as we show n more detal n Secton EU Asylum Polcy The Dubln Conventon of 1990 establshed the prncple of asylum n the country of frst entry as a cornerstone of the European Asylum Polcy. Snce then, the Dubln System has been under attack. Its ntal am was to clearly allocate responsblty for the examnaton of asylum applcaton. However, the past two decades have shown that the resultng regulaton was unable to cope wth the strkng nequalty of refugee treatment across the EU, dd not mprove the ntegraton prospects of asylum seekers, and faled to mplement safeguards to protect them. In addton, the Dubln System nherently counteracts the far sharng perspectve and operates n a hghly neffectve and costly manner De Bruycker et al., In ths secton we brefly revew the evoluton of the EU Asylum Polcy of the last two decades or so, whch can be characterzed by a quest for elusve harmonzaton and no real nterest for far burden sharng procedures. 3.1 Harmonzaton In recognton of the shortcomngs of the Dubln system and n an attempt to harmonze EU polces, many reforms have been ntated n partcular snce Two generatons of drectves have been adopted n order to harmonze the European Asylum Polcy. The Asylum Procedures Drectve sets out rules on the whole process of clamng asylum, ncludng on: how to apply, how the applcaton wll be examned, what knd of assstance can be provded to asylum seekers, how to appeal and whether the appeal wll allow the person to stay on the terrtory or how to deal wth repeated applcatons De Bruycker et al., These nstruments try to ensure a coherent procedural system wthn the EU. The Recepton Condtons Drectve attempts to deal wth the dvergence n the practce of treatng asylum seekers. It ensures that applcants have access to housng, food, employment, as well as medcal and psychologcal care. The Qualfcaton Drectve specfes the grounds for grantng nternatonal protecton. 6

8 De Bruycker et al explan how dffcult t becomes to montor the mplementaton of these drectves by ndvdual countres. Exstng assessments are problematc due to lack of data but they tend to pont out at ember States crcumventng the sprt of the drectves by takng advantage from ts vagueness n crucal ponts. For example, regardng the Asylum Procedures Drectve, De Bruycker et al note as regards survvors of torture, the Drectves slence on the specal needs of ths category of applcants n combnaton wth provson allowng ember States to treat cases as manfestly unfounded and omt a personal ntervew, have the potental to produce errors upon the asylum decson makng. Addtonally, some of the shortcomngs of the Dubln System have been revsed not only to take nto account the nterests of asylum seekers but also to better address stuatons of partcular pressure of ember States recepton capactes and asylum systems. The Temporary Protecton Drectve was ntroduced n 2001 wth the ambton to create a common EU bass to manage n case of a large nflux of dsplaced persons fleeng ther country of orgn. However, ths drectve has never been mplemented n contrast to the smlar-n-sprt Temporary Protected Status n the Unted States, whch has been nvoked, for example, to prevent deportaton of undocumented mmgrants followng the earthquakes n El Salvador n 2001 or Hat n Fnally, the European Unon has set up systems to promote nformatonal exchange across ember States. The Eurodac regulaton regsters the fngerprnts of all asylum seekers to prevent multple clams. The Vsa Informaton System facltatng nformaton exchange across ember States and tracng back vsa applcatons can also be used for the purpose of the asylum polcy. 3.2 Burden Sharng In addton to attempts to harmonze EU refugee polces, there s also an ncreasng demand for far burden sharng. The EU has taken some ntatves to address ths ssue snce the early 2000s. In December 2011 the European Commsson ssued a communcaton on enhanced ntra-eu soldarty n the feld of asylum European Commsson, 2011 that has been followed by conclusons on a Common framework for genune and practcal soldarty towards ember States facng partcular pressures on ther asylum systems adopted by the Justce and Home Affars Councl n arch 2012 Councl of nsters of the European Unon, The European Refugee Fund covered the fnancal component of the dstr- 7

9 buton problem. Created n 2000, the fund hoped to ncrease the fnancal soldarty among member countres wth an annual budget of 105 llon Euros for Recently, the European Refugee Fund has been reorganzed for the years 2014 to 2020 and s now replaced by the Asylum, graton and Integraton Fund and the Internal Securty Funds whose annual budget amounts to approxmately 10.9 mllon Euros. In December 2013 the European Councl emphaszed the mportance of the EU Resettlement Program, followng an ntatve of the EU Parlament n Ths ntatve was created to encourage resettlement of refugees by EU ember States. Despte ts declared mportance, the EU has faled to mplement the program on a large scale. For example, up untl December 2013, only 0.54% of all Syran refugees benefted from the ntatve Ballex, Nevertheless, n 2012 about 4,500 refugees were able to beneft from the Resettlement Program see Table 1, whch meant just above 5% of the total number of refugees resettled n the world and 9% of the number of asylum applcants that were granted refugee status n the EU that year. 3 Table 1: Source: EASO Percentages are based on data from UNHCR 2013 for the total number of refugees resettled n the world and from EASO 2013 for the total number of asylum applcants granted refugee status n the EU n frst nstance 37,335 plus fnal decsons 13,345: 50,680. 8

10 A lot of attenton has been pad to the external borders of the European Unon. In addton to drect measures of border control see below the EU tres to support ember States that are under partcular pressure. For example, the European Asylum Support Offce EASO started operatng n alta n ost of the nstruments that promote soldarty are n fact means to help guardng ember States at the external border of the European Unon, such as Italy or Greece. In 2004, FRONTEX was created as an operatonal support mechansm for ember States n ther management of border controls. FRONTEX has a budget of 90 mllon Euros. Addtonally, the EU Border Survellance System EUROSUR created n 2013 offers a system of organzed operatonal nformaton exchange n order to reduce the number of mgrants enterng the EU llegally and to prevent cross border crmnalty Ballex, oreover, the Smart Borders Package, whch s stll under dscusson, serves as an addtonal mean to desgn border controls n a more effcent manner. Thus, border controls have been at the center of attenton over the last decade. Ultmately, those measures may tackle the ssues at the external borders of the EU but they do not deal wth the queston of nternal soldarty regardng the allocaton of refugees. All those measures may be a frst step towards a more soldary unon, but they do not suffce to algn the needs of all, especally not the preferences of the asylum seekers themselves. In any event, mprovng the balance across the EU frst requres a proper assessment of the burden carred by each ember State. And ndeed, a number of attempts have already been made to assess the current extent of fnancal burden sharng between EU countres wth respect to EU Asylum Polcy. ost promnently, Thelemann et al estmate the total amount to be dstrbuted n by the European Refugee Fund to represent only 14% of the total asylum costs for the EU-27 for the sngle year These costs nclude recepton, accommodaton, admnstratve procedures, deportaton, ntegraton measures, and so on. They do not however nclude the opportunty cost of these expensve and neffcent procedures, whch would drve up the cost even more, gven that t would not be hard to thnk of more productve uses for these resources. Also, the European Commsson 2010 reports that the average cost per relocated refugee can amount to 8,000 Euros, of whch selecton and travel s just over 1,000 Euros and the rest s accommodaton and other support costs. 4 Ths s double the actual per refugee budget allowed for relocatons under the ERF. Thelemann et al even suggest that, under dfferent rules, equal burden 4 Ths s calculated from a plot study of transfers between alta and France. 9

11 sharng mples transferrng 33 to 40 percent of the asylum seekers currently wthn the EU to other countres, a large share gong to the new ember States. In contrast to the UNHCR capacty assessment, Thelemann et al create a combned capacty ndex that ams to capture whether a country takes n a suffcent amount of asylum seekers accordng to ts potental ablty to accommodate them. They show that there s a tremendous dscrepancy across EU members wth respect to ther actual v. far share ntake of asylum seekers. Beyond ther quanttatve analyss, Thelemann et al compare EU countres Internal Dspersal Systems wth those of the Unted States. The Unted Kngdom, for example, agrees on the allocaton of asylum applcants wth local authortes, dependng on varous parameters, ncludng asylum seekers per person. Conversely, Germany decdes upon the dspersal of asylum seekers on the federal level. Regonal governments, the Länder, are not nvolved n the decson makng process, whch s manly populaton-based. In Fnland, the man authortes are the muncpaltes that decde how many asylum seekers they are voluntarly wllng to accept. As compensaton they receve a small lump-sum that however does not cover the actual costs that are ncurred by takng n an addtonal asylum seeker. Sweden and France take an approach that s more asylum seeker orented. There, asylum seekers choose freely where to stay. State funds are subsequently transferred to regonal governments, accordng to the total number they host. Compared to the Unted States, EU ember States base ther decson on governmental drectves, may they be federal, regonal or at the muncpal level. In the US however, NGOs 9 agences plus the State of Iowa decde how to dsperse the resettled refugees across the States. In general, the polcy ntatves of the European Unon seem to only allevate the symptoms of the refugee recepton problem, drectng ther man efforts towards border controls and lump-sum transfers, whch do not stand n proporton to the real cost occurrng n the host country. Addtonally, the burden-sharng assessment has so far only made use of drect costs of refugees, lke accommodaton or admnstratve efforts but they do not reveal the true cost drect costs, ndrect costs and opportunty costs and above all not the underlyng benefts, such as the potental economc effcency gans from recevng mmgrants n general or the welfare ncrease generated by the grantng of protecton postvely valued by publc opnon n host countres. These polces thus fal to tackle the problem at the root. In order to adequately assess the cost of refugees and to reveal the preferences of ember States and refugees equally, there s a need for a market-based approach that wll effcently and at a 10

12 low cost lead to the desred socal optmum. 4 Related Lterature Ths paper s closely related to two strands of recent lterature on the neffectveness of the current EU Asylum Polcy, mostly to Hatton 2010, 2011, 2012, for economcs, and Thelemann et al. 2010, for poltcal scence; and on market-based proposals to address the refugee ssue. Hatton 2012 nvestgates ths queston by examnng the bass upon whch a jont EU polcy can be justfed and by askng whether a more effcent outcome can be acheved by harmonzaton alone or a more centralzed decson makng processes s necessary. He does so by chartng the progress of harmonzaton and burden-sharng n the Common European Asylum System, explorng ts costs and benefts, and also analyzes the evoluton of publc atttudes towards refugees. He also develops a model where asylum s a locally provded publc good, thus n ts nature underprovded unlaterally. Consequently, ths creates the need for a jont polcy that ensures an optmal dstrbuton of asylum seekers across ember States. In ths case however, a harmonzed polcy s not suffcent snce t does not account for refugees preferences that may be dsproportonately based towards one destnaton country. In order to avod accumulaton of refugees n mmgraton darlngs, there needs to be a jont EU polcy that matches supply and demand and prevents oversupply n some and under-provson n other countres. Hatton proposes two dfferent tools to reach or at least move towards the socal optmum: frst, an asymmetrc subsdy dependng on lack or excess supply of refugees n certan countres, and secondly, an actve redstrbuton of refugees. As already ndcated n secton 3.2., Thelemann et al calculate far burden sharng rules accordng to populaton sze, populaton densty and current stocks of refugees hosted. Any of these suggest a dsproportonate dstrbuton of burdens across ember States. In contrast to Hatton 2012, the authors do not suggest a radcal shft n the paradgms of asylum polcy but, rather, an expanson and enhancement of exstng polcy tools. Based on surveys askng ember States about ther preferences for possble soldarty mechansms, Thelemann et al call for 1 a harmonzaton of the costs for asylum-seekers, 2 a fnancal compensaton for over-burdened countres, and 3 a voluntary movement of asylum 11

13 seekers from more to less affected states, avodng costly forced movements. 5 Overall, the lterature has not pad enough attenton to market-based solutons. Schuck 1997 and Hathaway and Neve 1997 were frst to dscuss a system of blateral negotatons over tradable refugee resettlement quotas. In the words of Schuck 1997, p. 248, the proposal conssts of two man elements. Frst, a group of states would... arrange for an exstng or newly-establshed nternatonal agency to assgn to each partcpatng state a refugee protecton quota.... Second, the partcpatng states would then be permtted to trade ther quotas by payng others to fulfll ther oblgatons. through blateral exchange Schuck, 1997, pp However, Schuck 1997 fell short of proposng a system of tradable quotas as he mostly envsoned the possblty for countres to trade refugees for money on a strctly blateral bass. Bubb, Kremer, and Levne 2011 supplement ths system of blateral exchange wth a screenng devce to separate refugees from economc mgrants. However, the blateral nature of these proposals lmts the scope for trade and, thus, mples an neffcent outcome. Fnally, Fernández-Huertas oraga and Rapoport 2014 develop a multlateral system of Tradable Immgraton Quotas TIQs whch ncludes a mechansm to match mmgrants to destnatons. One of the man applcatons they envson for ther proposed system s the resettlement of long-standng refugees. Ths would seem a partcularly well-suted context for mplementng a TIQs system, for a number of reasons. Frst, refugee protecton s a paragon example of nternatonal publc good, wth refugee protecton beng recognzed as a moral and legal oblgaton by the countres sgnatores of the 1951 Geneva Conventon and ts 1967 extended Protocol - that s, vrtually all the countres of the world. Second, there s also an obvous ncentve to free-rde, and not less obvous evdence of ensung under-provson of refugee resettlement. Thrd, t s qute common among polcymakers to consder refugee protecton as a burden and there s a general understandng that ths burden s not farly shared at the nternatonal level see, e.g., UNHCR, And fourth, the nsttutonal framework for nternatonal polcy coordnaton s already largely n place, wth wdely ratfed treates such as the Geneva Conventon and exstng well-establshed 5 They emphasze how forced relocatons of refugees or asylum seekers are much more expensve than voluntary relocatons ten tmes more expensve accordng to a UK report on return that they quote and ths s why they propose that any relocaton should be voluntary. From the same report, an example on how to promote voluntary movements s by better promotng the optons avalable to those due for removal and by establshng better contacts wth communty groups. 12

14 nternatonal organzatons such as UNHCR. All these reasons would seem all the more true n the European context, as demonstrated n the prevous secton. In what follows, therefore, we adapt Fernández-Huertas oraga and Rapoport 2014 to the context of the EU Asylum Polcy by proposng an EU Tradable Refugee-admsson Quotas TRQs System that takes nto account the specfcs of the EU context and addresses many of the neffcences of the current system. It bulds on exstng understandngs that far burden sharng rules should be adopted - ths s analogous, n our model, to the stage of ntal quotas determnaton - and shows how a TRQs system can allow for truthful revelaton of the country-specfc costs of accommodatng asylum seekers and refugees, explotaton of dfferences n such costs among EU countres to generate cost-effectve outcomes that s, mnmzng total cost for a gven number of refugees or maxmzaton the number of refugees for a gven budget constrant, and takng refugees preferences over destnatons nto account, therefore reducng ther ncentves to move between countres. Last but not least, the mechansm also allows for takng countres preferences over refugees types nto account. Ths can be understood as preferences over refugees characterstcs such as language, sklls, country of orgn, etc. as well as over refugees legal status such as asylum seekers n wat for a decson, refugees whose asylum request has been accepted, or nternatonally resettled refugees, allowng for compensaton among countres across refugees types. 5 The model Our model s a smple extenson of the one presented n Fernández-Huertas oraga and Rapoport 2014 for the case of both asylum seekers and refugees n the context of the European Unon. The model has two buldng blocks. On the one hand, provdng protecton to refugees and to asylum seekers wth vald clams s consdered an nternatonal publc good. On the other hand, provdng protecton both to refugees and to asylum seekers, wth or wthout vald clams, s costly so that t can be consdered a burden, whch European countres wll be sharng through the market. We proceed n three steps. Frst, we ntroduce a basc model for tradable refugee quotas when refugees and asylum seekers preferences are not taken nto account. Second, we 13

15 add refugees and asylum seekers preferences through a matchng model. Fnally, we also consder recevng countres preferences over the number the refugees and asylum seekers. 5.1 Tradable Refugee Quotas wthout matchng Suppose each EU country decdes how many refugees and asylum seekers to receve gven the net perceved total cost or beneft of recevng them: c r, a, where r denotes the number of refugees hosted by country and a refers to the number of asylum seekers. We assume that t s convex and dfferentable and that t has an nteror postve mnmum n both arguments. Ths cost refers to a set of drect costs, such as: Recepton and accommodaton; Admnstratve procedures; Potental removal; Potental ntegraton measures, etc. There are also ndrect net costs assocated wth the admsson of refugees and asylum seekers. These nclude the mmgraton surplus assocated wth refugees and asylum seekers effcency gans from mgraton, ther net fscal contrbuton, ther socal and poltcal costs, etc. The nternatonal publc good aspect s ntroduced by consderng that the nhabtants of the EU recevng countres or at least ther governments care about refugees and asylumseekers hosted by other countres through the functon g R, A, wth g R > 0, R = j r j and A = j a j. Country would maxmze the followng welfare functon: g A 0, The frst order condtons would be: max r,a g R, A c r, a 1 c r NC r, a NC c = 0 = r NC a, a NC 2 14

16 where N C stands for the non-cooperatve soluton. The number of refugees and asylum seekers hosted by one EU country generates a postve externalty for the other EU destnaton countres. Ths makes the non-cooperatve equlbrum dfferent from the optmal levels of refugees and asylum seekers that should be accepted from the vewpont of the EU as a whole. We denote these levels as R GO, A GO. They are the soluton to the problem: max {r,a } N =1 N [g R, A c r, a ] 3 =1 Wthout loss of generalty, we can assume that the functons g are such that R GO, A GO can also be obtaned as the soluton to the dual problem: 6 s.t. mn {r,a } N =1 N c r, a 4 =1 N r + a R GO + A GO =1 The frst order condtons would be: c r GO, a GO c = λ = r GO, a GO = 1...N 5 r a where λ s the Lagrange multpler assocated to the constrant. It s easy to see that: R NC + A NC N =1 r NC + a NC < N =1 r GO + a GO R GO + A GO 6 If there s no coordnaton and EU countres act unlaterally, each of them wll equalze the margnal cost of acceptng one addtonal refugee or asylum seeker to zero. However, more refugees and asylum seekers would be hosted f countres took nto account how other countres care about provdng nternatonal protecton. The optmal soluton would be to equalze the margnal net cost of hostng one addtonal refugee or asylum seeker across countres but not to zero. In a sense, refugees would be hosted where t s cheapest to host them from the pont of vew of cooperatve recevng countres. In the presence of 6 See Fernández-Huertas oraga and Rapoport 2014 for a formulaton that solves the maxmzaton problem drectly. 15

17 cooperaton, there would be no need for a race to the bottom n terms of each country unlaterally adoptng excessvely strct polces from the pont of vew of the EU. Ths optmal soluton can be replcated by creatng a market for tradable refugee quotas. Suppose each country s assgned an ntal quota q 0 that can be flled both wth refugees and asylum seekers. 7 These quotas can be traded n a market n whch the prce receved for acceptng one addtonal refugee or asylum seeker nto a country s denoted by p. The ntal dstrbuton of quotas must be agreed upon at the EU level and be such that: N R + A = q 0 7 Three ponts must be emphaszed. Frst, ths would be a non-tradtonal market n the sense that partcpaton would be restrcted, at least ntally, to EU governments. Second, the market would not apply to all refugees or asylum seekers at the doors of the EU but only to a predetermned number that ember States would need to agree upon. Thrd, the system presupposes that the ntal dstrbuton of quotas must also be agreed upon at the EU level through some commonly accepted burden-sharng rules. Ths would certanly be a poltcally senstve ssue and focusng on one partcular rule, such as the ones enumerated by Thelemann et al. 2010, s out of the scope of ths paper. We only descrbe brefly below subsecton some desrable propertes for ths ntal dstrbuton of quotas. Then, the market would operate n the followng fashon: Supply of vsas demand for quotas. At a gven vsa prce hgher than ther margnal cost, some countres would be wllng to get pad to receve refugees n excess of ther quota. Demand for vsas supply of quotas. At a gven vsa prce lower than ther margnal cost, some countres would be wllng to pay to receve less refugees than ther quota. A smple example wth two destnaton countres can llustrate how ths market would work. Suppose Australa and New Zealand agree on hostng a gven quantty of refugees from Krbat, denoted by the dstance O A O NZ n Fgure 3. They also agree ntally that Australa wll host Q refugees whle New Zealand wll take care of the rest. Fgure 3 also 7 We can thnk of alternatve formulatons n whch there would be one quota and one prce for each of the categores: refugees and asylum seekers. =1 16

18 depcts the margnal net perceved costs of both countres on the vertcal axs, wth Australa countng refugees from left to rght, so that the net margnal cost ncreases n the number of receved refugees and New Zealand countng them n the opposte drecton. If a market for tradable quotas opened between both countres, they would reach an equlbrum quota prce at the ntersecton of both margnal cost curves. At the prce marked by ths ntersecton, Australa s wllng to get pad to receve more refugees than ther allocated quota whle New Zealand s wllng to pay not to receve them. Fgure 3: The market wll be a Pareto mprovement over the non-cooperatve soluton as long as R NC + A NC < R + A R GO + A GO Fernández-Huertas oraga and Rapoport, It s assumed that the cost functons are expressed n monetary unts and that the market s compettve so that all countres behave as prce-takers. Indvdual countres would then need to solve: mn c r, a p r + a q 0 8 {r,a } The second term means that the cost wll be just c r, a as long as the quota s flled: r +a = q 0. If country hosts more refugees and asylum seekers than ts quota r +a > q 0, 17

19 then ts cost wll be reduced by ths extra amount multpled by the prce p. If, on the contrary, t hosts less than ts quota r + a < q 0, then ts cost wll be ncreased by ths extra amount multpled by the prce p. If the market s compettve, the frst order condtons are: c r r, a c = p = r a, a = 1...N 9 The margnal costs of acceptng one addtonal refugee or one addtonal asylum-seeker are equalzed across destnatons through the market, whch must clear, so that:: R + A = N q 0 = =1 N r + a 10 =1 Proposton 1 The market soluton to 4 s effcent. The proof comes from establshng that p = λ Fernández-Huertas oraga and Rapoport, The ntal dstrbuton of quotas does not affect the effcency of the mechansm and t only has redstrbutve consequences as long as the market s compettve. As dscussed n Fernández-Huertas oraga and Rapoport 2014, ths can be acheved through a computerzed contnuous double aucton mechansm, whch converges to compettve equlbrum outcomes even when there are very few buyers and sellers thanks to the Bertrand-type competton t promotes Casella, 1999; Fredman and Ostroy, Takng partcpaton constrants nto account In addton to ts effcency propertes, the market can also be made ndvdually ratonal for every country through the manpulaton of ntal quotas. Indvdual ratonalty would ensure that every country has an ncentve to partcpate n the market, snce t can acheve a better result than by stayng out of t. The general formulaton of the problem n whch the countres partcpaton constrants are satsfed s: max {q 0 } N =1 N =1 [ g R, A c r, a ] 11 18

20 N s.t. q 0 = R + A =1 g R, A c r, a + +p r + a q 0 g R NC, A NC c r NC, a NC = 1...N r, a = arg mn {c r, a p r + a q 0 } = 1...N Denotng by q P C 0 the soluton to ths problem, the frst order condtons are: N =1 µ pπ = 0 = 1...N q P C 0 R A = 0 π [g R, A c r, a + p r + a q0 P C g R NC, A NC + c r NC, a NC ] = 0 = 1...N π 0 where µ s assocated to N =1 q 0 = R + A and π s assocated to g R, A c r, a + p r + a q 0 g R NC, A NC c r NC, a NC. We are usng the fact that r j q 0 = a j q 0 = 0 = p q 0, j snce the soluton to the market problem does not depend on the ntal allocaton of quotas. From the frst set of condtons, we have: π = µ p > 0 = 1...N 12 Ths leaves us wth a rule to allocate ntal quotas satsfyng: whch mples: g R, A c r, a + c r NC g R NC, A NC + p r + a q0 P C, a NC = 0 = 1...N 13 19

21 q0 P C = r + a + g R, A g R NC, A NC p c r, a c r NC p, a NC = 1...N 14 The countres benefttng the most from the externalty hgher g R, A g R NC, A NC should get hgher ntal quotas whereas those who devate most from ther ndvdually optmal allocatons because of the market hgher c r, a c r NC, a NC should get lower ntal quotas. Of course, the equalty n 14 ensures that the dstrbuton of quotas wll obtan the maxmum level of total welfare for a gven sze of the market R + A. ore generally, partcpaton can be obtaned wth any q 0 q P C 0. It could be argued that the nformatonal requrements for establshng such a quota are unrealstc so that countres could be unwllng to partcpate. However, partcpaton constrants are even less clear n the case of sgnatores of the 1951 Geneva Conventon or ts 1967 Protocol for the protecton of refugees and stll 145 countres have sgned the former and 146 the latter. They are less clear because they mply countres must host as many ndvduals as they arrve n ther sol as long as they are n danger of persecuton n ther orgn countres. As a result, the potental cost of sgnng the Conventon s not bounded. In our model, the potental cost s bounded by the actual number of refugees and asylum seekers to whch the market s appled. Hence, t looks lke the externalty derved from the provson of ths publc good s large enough for countres to be wllng to enter nto ths type of multlateral agreements. 5.2 Takng Refugees Preferences nto Account Takng the preferences of refugees and asylum seekers nto account s a straghtforward task. From a theoretcal pont of vew, the problem s analogous to assgnng houses to tenants wth exstng rghts, studed, among others, by Abdulkadroglu and Sonmez The exstng rghts can ether refer to the country of frst asylum n the EU or to a refugee camp. Techncally, the problem s to assgn ndvsble tems rghts for a refugee or an asylumseeker to enter a gven destnaton country, or vsas to agents refugees or asylum-seekers takng nto account ther preferences. 20

22 The soluton proposed by Abdulkadroglu and Sonmez 1999 s the use of the top tradng cycles mechansm: 1. Each refugee/asylum seeker ranks all potental destnaton countres, specfyng those to whch she would not want to go at all. 2. An orderng of refugees and asylum seekers s randomly chosen from a gven dstrbuton of orderngs. A queston remans whether the EU would want to prortze refugees over asylum seekers. 3. For any gven rankng of countres done by the refugees and orderng of refugees, the outcome s obtaned usng the followng algorthm: a Assgn the frst refugee from the orderng obtaned n step 2 her top choce, the second refugee her top choce among the remanng vsas, and so on, untl someone requests a vsa for whch the quota resultng from the market s flled. It s as f the frst refugee wth a vsa n that quota s requested to exchange her vsa. b If at that pont, the refugee whose vsa s requested has already chosen before, then go to the second refugee n that quota. If ths one has also chosen, go to the thrd and so on. If the quota s flled wth refugees who have already chosen before, then do not dsturb the procedure there s no room for Pareto mprovement. Otherwse, modfy the remander of the orderng by nsertng the refugee who dd not choose yet to the top of the lne and go on wth the procedure. c Smlarly, nsert any refugee who s not already served at the top of the lne once her vsa to stay n her frst asylum country s requested. d If at any pont a loop forms, t s formed exclusvely by refugees wth a vsa each of them requestng the vsa of the refugee who s next n the loop a loop s an ordered lst of refugees j 1, j 2,..., j k where refugee j 1 requests the vsa of refugee j 2, refugee j 2 requests the vsa of refugee j 3..., refugee j k requests the vsa of refugee j 1. In such cases, remove all refugees n the loop by assgnng them the vsas they request and contnue the procedure. A key ngredent of ths mechansm s that a refugee whose vsa s requested s upgraded to the frst place at the remanng of the lne before her vsa s allocated. As a result, 21

23 the top tradng cycles mechansm s ndvdually ratonal, as t assures every refugee a vsa that s at least as good as the possblty of stayng n her frst-asylum country or her refugee camp. It s also ncentve compatble no refugee has an ncentve to msrepresent her preferences whatever the strateges others use 8 and Pareto effcent. Ths s a drect applcaton of Abdulkadroglu and Sonmez 1999 followng drectly the exposton n Chen and Sonmez 2002 and substtutng word by word house for vsa and refugee for tenant. The relevant pont for the case of refugees studed here s the possblty that the fnal allocaton determned by the market mght not be acheved. We follow Fernández-Huertas oraga and Rapoport 2014 to show what the central planner s problem looks lke n ths case: mn {r,a } N =1 N =1 c r, a 15 s.t. N =1 r + a R + A r = F r 1, r 2,..., r N ; a 1, a 2,..., a N = 1...N a = G r 1, r 2,..., r N ; a 1, a 2,..., a N = 1...N The matchng mechansm s embedded n the last two lnes. The sequence {F } N =1 of functons F : [0, R] N [0, A] N [0, R] transforms an allocaton of vsas {r, a } N =1 decded by the central planner as f countres were homogenous from the pont of vew of refugees and asylum seekers nto an allocaton of vsas for refugees { r that does take nto account refugees preferences through the matchng mechansm denoted. Smlarly, the sequence {G } N =1 of functons G : [0, R] N [0, A] N [0, A] transforms the allocaton of } N =1 vsas {r, a } N =1 nto an allocaton of vsas for asylum seekers { a account asylum seekers preferences through the matchng mechansm. } N =1 that does take nto We can approxmate both sequences by dfferentable functons so that t s smple to wrte the frst order condtons of the problem: 8 Ths follows drectly from ndvdual ratonalty. Gven that an ndvdual whose vsa s rejected s upgraded to the frst place at the lne, there s no ncentve to msrepresent preferences snce t can only result n a worse outcome for whoever does t. 22

24 N [ Fj j=1 r N [ Fj j=1 a cj r j r j cj r j r j, a j λ + G j cj r, a j λ + G j cj a r j a j r j a j ], a j λ = 0 = 1...N 16 ], a j λ = 0 = 1...N 17 where λ s the multpler assocated wth the frst constrant. The equalzaton of margnal costs across countres s just one possble soluton. In partcular, t wll be the soluton whenever the matchng mechansm does not dstort the planner s allocaton. The TRQ s problem can be formulated as follows: mn c r r,a, a p r + a q 0 + p r + a r a 18 s.t. r = F r 1, r 2,..., r N ; a 1, a 2,..., a N a = G r 1, r 2,..., r N ; a 1, a 2,..., a N The term p r + a r a s a penalty unattractve countres would have to pay for not beng able to attract as many refugees and asylum seekers through the matchng mechansm as they would bd for n the market. In equlbrum, the penalty would always be zero but t s needed so that countres do not have ncentves to become unattractve from the pont of vew of refugees and asylum seekers. In practce, the EU could be n charge of collectng ths penalty n case of some off-equlbrum behavor. The objectve functon can be smplfed to c r, a p r then easy to wrte the frst order condtons of the problem: F c r F c a r r r r The followng proposton holds:, a p + G r, a p + G a c r a c r a + a q 0. It s, a p = 0 19, a p =

25 Proposton 2 Let 18 represent the structure of the problem solved by country partcpatng n a market for tradable refugee quotas. In partcular, the setup of the market s such that country pays s compensated for the actual number of refugees and asylum seekers receved r so that p r + a r + a rather than by the number bd n the market r + a, can be consdered as a penalty assocated to the outcome a of the matchng mechansm, denoted by r G r 1, r 2,..., r N ; a 1, a 2,..., a N. = F r 1, r 2,..., r N ; a 1, a 2,..., a N and a = Under ths defnton of a market for tradable refugee quotas, at least one of the market solutons to 15 s effcent, specfcally margnal cost equalzaton across countres. Agan, the proof follows smply from showng p = λ Fernández-Huertas oraga and Rapoport, It turns out that the addton of the matchng mechansm to the market for TRQs descrbed n the prevous secton does not alter ts effcency propertes as long as t s properly desgned. If partcpatng countres were compensated on the bass of the number of refugees and asylum seekers they bd for n the market, they would have an ncentve to bd for a large quota and later on dscourage refugees and asylum seekers from gong there. Ths way, they would be compensated by the market n addton to not actually ncurrng the cost of hostng the refugees and asylum seekers, who would use the matchng mechansm not to go to an undesrable destnaton. In order to prevent ths perverse ncentve from happenng, the soluton s to make countres be compensated on the actual numbers of refugees and asylum seekers they host rather than on those they bd for. Ths amounts to forcng destnaton countres to pay the market prce for the unflled part of ther quotas. Ths s a penalty unattractve countres would have to pay for not beng able to attract as many refugees and asylum seekers through the matchng mechansm as they would bd for n the market. In equlbrum, the penalty would always be zero but t s needed so that countres do not have ncentves to become unattractve from the pont of vew of refugees and asylum seekers. In practce, the EU could be n charge of collectng ths penalty n case of some off-equlbrum behavor. In terms of enforcement, the penalty would generate ncentves for countres to abde by ther agreements and actually host the number of refugees they accept to host. Of course, collectng the penalty would be an addtonal enforcement ssue but we do not thnk t dfferent from the enforcement problems assocated wth the collecton of other payments at 24

26 the EU level, whose dscusson s out of the scope of ths paper. 5.3 Takng Countres Preferences nto Account Not only do refugees and asylum seekers have preferences over the countres to whch they can go, but recevng countres also have preferences both between hostng refugees or asylum seekers and possbly about the type natonalty, skll level, etc. of refugee or asylum seeker they wll be hostng as well. Satsfyng these preferences can make hostng refugees and asylum seekers more attractve for host countres. Ether the market could be expanded or the same market sze could acheve a more cost-effectve allocaton. There are at least two ways to ntroduce countres preferences nto the model. The frst and less nterestng one would mply creatng one market for each type of refugee and asylum seeker there s. For example, f countres only had preferences between refugees and asylum seekers, the EU would just need to create a market for refugee quotas and a market for asylum seekers quotas. A second possblty, whch we followed n the prevous secton, s to group refugees and asylum seekers nto the same market even f they are heterogeneous. Suppose there are many dfferent types of refugees or asylum seekers over whch countres can have preferences n terms of, for example, ther natonalty or ther skll level. We ndex these dfferent types by k and redefne the total cost functon as c r, a where r and a are vectors of K elements types. We assume that the cost functon s convex n each of the elements of the vectors and such that the global optmal soluton mples hgher levels of refugees and asylum seekers than the non-cooperatve soluton. Ths way we can use total cost mnmzaton as an alternatve to utlty maxmzaton. The total mnmum cost problem would then be: mn { {r k,a k } K k=1 } N =1 N =1 c r, a 21 s.t. R + A N =1 K k=1 r k, + a k, r k, = F k r 1, r 2,..., r N ; a 1, a 2,..., a N = 1...N; k = 1...K a k, = G k r 1, r 2,..., r N ; a 1, a 2,..., a N = 1...N; k = 1...K 25

27 For the soluton, we would have N K 2 frst order condtons: [ N K F l j j=1 N l=1 r k [ K F l j a k j=1 l=1 c j r l j c j r l j λ λ + Gl j r k + Gl j a k c j a l j c j a l j λ λ ] ] = 0 = 1...N; k = 1...K 22 = 0 = 1...N; k = 1...K 23 Stll, margnal cost equalzaton across mgrant types and across countres remans a soluton to the problem n cases where the matchng mechansm does not affect the market outcome. If the EU sets up a market for tradable refugee quotas n whch both countres and refugees and asylum seekers can express ther preferences over each other, the formulaton of the problem for an ndvdual country would be: mn {r k,ak } K k=1 c r, a K K p r k + a k q 0 + p k=1 k=1 r k + a k r k, a k, 24 s.t. r k, = F k r 1, r 2,..., r N ; a 1, a 2,..., a N k = 1...K a k, = G k r 1, r 2,..., r N ; a 1, a 2,..., a N k = 1...K The frst order condtons assocated wth ths problem are: K l=1 K l=1 [ F l r k [ F l a k c r l c r l p + Gl r k p + Gl a k c a l c a l The followng proposton can be establshed: ] p = 0 k = 1...K 25 ] p = 0 k = 1...K 26 Proposton 3 Let 24 represent the structure of the problem solved by country partcpatng n a market for tradable refugee quotas. In partcular, the setup of the market s such that country pays s compensated for the actual number of refugees and asylum seekers receved K k=1 r k, + a k, rather than by the number bd n the market K k=1 r k + a k, 26

28 so that K k=1 r k + a k r k, a k, outcomes of the matchng mechansm. can be consdered as a penalty assocated to the Under ths defnton of a market for tradable refugee quotas, at least one of the market solutons to 21 s effcent, specfcally margnal cost equalzaton across countres and types. Once more, the proof follows drectly from showng p = λ Fernández-Huertas oraga and Rapoport, Addng a matchng mechansm that assgns both destnatons to ther preferred refugees and refugees to ther preferred destnatons to the market for TRQs has no effect on the effcency propertes of the market. across countres would stll be optmal. argnal cost equalzaton across mgrant types and The choce of the approprate matchng mechansm s a dfferent ssue. In ths case, the preference over one partcular mechansm s not that clear. The problem s smlar to the allocaton of students to colleges Gale and Shapley, 1962; Roth 1985 although the market and the penalty make t a bt dfferent. Stll, Fernández-Huertas oraga and Rapoport 2014 establsh the mpossblty of havng a stable Pareto-effcent matchng mechansm n whch countres colleges reveal ther preferences truthfully over the type of refugees students they want, followng Roth s 1985 result for the college admssons problem. 9 Among the many possble matchng mechansms that would be compatble wth the market for tradable refugee quotas, we follow Fernández-Huertas oraga and Rapoport 2014 n argung for the country-proposng deferred acceptance mechansm, on the bass that t s less manpulable than the refugee-proposng one, as defned by Pathak and Sonmez The way the mechansm works s by countres submttng ther preferences over refugees and asylum seekers frst. Refugees and asylum seekers accept ther most preferred vsa among the countres wllng to accept them and reject the unacceptable ones countres where they would not want to go. Rejected countres would then offer vsas agan to ther preferred refugees and asylum seekers among those who had not rejected them yet. Refugees and asylum seekers wth several vsa offers would then hold to ther most preferred one and reject ther unacceptable ones. The process would repeat untl no country would have vsas left to offer. 9 Azevedo and Budsh 2013 argue that manpulaton ncentves dsappear n deferred acceptance mechansms as the market grows, though. 27

29 Even though ths mechansm attans the most preferred stable matchng for countres, 10 takng nto account countres preferences could ntroduce a trade-off. On the one hand, cost-effcency s ncreased Fernández-Huertas oraga and Rapoport, On the other hand, the matchng mechansm can generate some addtonal uncertanty over ts outcome. However, t must be noted that the frst effect can be proved whle the second would be an emprcal queston. 6 The alta Example Ths secton makes use of a plot program by the European Unon that amed to mplement the prncple of soldarty and burden sharng n the context of refugee and asylum polcy. Partcularly, a European Councl Concluson ntated the EUREA European Relocaton from alta program n 2009 at the ntatve of Commssoner Barner n order to allevate the dsproportonate burden accrung to the Island of alta. In fact, alta was under great pressure regardng the large nflux of refugees and other asylum seekers n 2009, as well as n 2011 and The European Asylum Support Offce EASO, whose man objectve s to coordnate ember States polces n vew of the development of the Common European Asylum System, was naugurated n alta n 2011 as a response to these developments and wth the objectve of evaluatng the EUREA program. EASO 2013 states that: The altese Offce of the Refugee Commssoner receved a total of 2,114 applcatons for nternatonal protecton n % of applcatons receved by the Offce of the Refugee Commssoner n 2012, as n prevous years, were lodged by thrd country natonals TCNs who entered alta rregularly by sea. In vew of the sgnfcant pressures facng alta, and followng an nter-nsteral pledgng conference organsed by the European Commsson n ay 2011, relocaton of protected persons from alta to other ember States took place durng Relocaton actvtes were organsed ether as part of the EU plot project on Intra-EU relocaton EUREA, or through blateral projects. EUREAs success n 2009 led to the extenson of the program n 2011 EUREA II. EASO dvdes EUREA nto Phase I, ntated n 2009 and ultmate relocaton n 2011, and 10 Azevedo and Leshno 2013 show that t would also be the preferred mechansm for refugees and asylum seekers wth a large number of partcpants. 28

30 Phase II, brought to lfe n 2011 and relocaton partly stll n process. Tables 2 and 3 show the partcpatng ember States, the pledges, and the fnal dstrbuton of asylum seekers. Table 2: Source: EASO 2012 In Phase I, Germany and France relocated most of the mmgrants: 197 out of 227, even more than they had pledged ntally. In contrast, other countres eventually dd not commt to the number of relocatons they had pledged at the begnnng. In Phase II only 217 out of 306 pledged allocatons had effectvely taken place by January Table 3 shows that partcpaton extended to more countres n ths second phase, although ths was acheved by allowng some of the new partcpatng countres to sgn ther own blateral agreements to relocate refugees and asylum seekers. The relevance of the EUREA program for ths study s three-fold. Frst, t sets the context for possble applcatons of the matchng model proposed above. Second, the program apples a smplfed matchng procedure between ember States and asylum seekers and thus uncovers possble shortcomngs and challenges faced by ember States and admnstrators. 29

31 Table 3: Source: EASO 2013 Thrd, t reveals how the proposed matchng technque can deal wth problems that occur f the relocaton s uncoordnated and at whch ponts t s more effcent. For ths purpose, the lessons learned from the EUREA plot are of major mportance. EASO 2012 analyzes the results n a fact-fndng exercse and descrbes partcpaton, matchng procedure, and feedback of the ember States. Partcpaton n EUREA was voluntary and, stll, 12 EU countres agreed to partcpate at some pont snce 2009, when the need of some measure to assst alta was frst brought to the attenton of the European Commsson. The beneft for partcpatng countres conssted of fundng for relocaton actvtes as well as access to the expertse from the desgners of the EUREA framework. Ths level of partcpaton would be a good sgn for a smlar plot experence wth a system of tradable refugee quotas lke the one presented above. In addton to the 12 European ember States partcpatng through EUREA, Phase II also allowed ndvdual countres, both ember States or Assocated Countres such as Swtzerland, to reach blateral agreements wth alta for the relocaton of refugees or asylum 30

32 seekers. Eght addtonal countres chose ths alternatve. The selecton of potental benefcares that would be relocated was made n two steps. Frst, UNHCR counseled potental benefcares through a regstraton exercse that took place n In a second step, partcpatng countres sent mssons to alta to make the fnal selecton. The matchng part of the market for tradable refugee quotas could follow a smlar strategy. Durng the regstraton exercse, nformaton could also be gathered about the preferred destnatons of all of the potental benefcares. Then, countres could express ther preferences about types of refugees and asylum seekers and some authorty, ether UNHCR or EASO, could perform the matchng applyng the country-proposng deferred acceptance mechansm to both lsts of preferences: those of countres and those of potental benefcares. Generally, the fact-fndng exercse of EASO reveals that the approach s rather ember- States orented. Not only has there been no nqury about the personal preferences on destnatons by refugees and asylum seekers but there s also no or very few nformaton about the satsfacton level of the relocated persons wth the EUREA program. In contrast, there s detaled nformaton about the selecton crtera and demands of ember States. Among ths detaled nformaton, there s a long lst of challenges to be addressed as dentfed by partcpatng countres. We enumerate a selecton of them below, whle argung n whch way our market for tradable refugee quotas could have a useful role n addressng them: Tme constrants relatng to dentfcaton of canddates, n partcular as some requred several rounds of counselng before confrmng ther nterest n relocaton. The matchng mechansm would prevent ths type of problem from happenng. If we collect the preferences of potental benefcares durng the regstraton drve, there would be no need n general to confrm the nterest n relocaton. Lmtatons and constrants of the partcpatng States selecton crtera, n order to match the profle of benefcares of nternatonal protecton n alta. Agan the matchng mechansm, would gather the requred nformaton beforehand so that ths would no longer be a problem. Partcpatng states could express ther lmtatons and constrants when statng ther preferences over types of refugees and asylum seekers. Assessng the wllngness and sutablty of potental benefcares to beng relocated. Ths challenge s drectly related to the frst one and, as n that case, the matchng 31

33 mechansm would drectly take care of t. Target group composton refugees, subsdary protecton benefcares, asylum seekers. In the same way, ths challenge s related to the second one so that the part of the matchng mechansm that takes care of countres preferences would address t. Settng up clear crtera concernng relatves, especally consderng famly reunfcaton. Ths feature s not specfcally consdered n the current verson of the matchng mechansm but t can be easly ncorporated. For example, Roth 2002 explans how classcal matchng mechansms can be modfed to take nto account the assgnment of couples to resdency postons n the US. Lack of wll by some canddates to commt to relocaton offers by new EU ember States where there are few mgrant communtes. Agan, ths s a matchng problem on the sde of potental benefcares that could be solved by collectng ther preferences at the tme of regstraton. ost of these selected challenges are related to matchng ssues that can be addressed by the matchng mechansm that we ncorporate to the market for tradable refugee quotas. Ths quote from EASO 2012 makes t even clearer emphass added: Some of the selecton crtera dd not match the characterstcs of the benefcares of nternatonal protecton n alta, makng t dffcult to carry out the relocaton to some of the partcpatng States... Ths sometmes led to a msmatch between the crtera and the pool of canddates... Other partcpatng States requested that the benefcares be refugees... The last sentence reveals key dfferences between the admsson crtera of several countres. Some preferred already recognzed refugees whle others were wllng to accept asylum seeker watng for a decson. Fgure 4 shows that there was a great deal of heterogenety n the selecton crtera of partcpatng countres. Out of the ten selecton crtera lsted, only one of them language was mentoned by the ten surveyed countres whle two of them refugee status and not beng a threat to publc order were only mentoned by one country Romana. All n all, ths heterogenety n preferences s good news snce t allows explotng the gans from trade n the market. We end ths secton wth another quote, ths tme from EASO 2013: 32

34 Fgure 4: Source: EASO there s room for dscussng and developng the nstrument of ntra-eu relocaton n the future, as part of a range of ntra-eu soldarty measures. The Commsson created the scope for Unon co-fnancng of such actvtes n the Asylum and graton Fund proposal, whch wll facltate acton by ember States wllng to engage n voluntary projects, wth the EASO takng a coordnatng role as establshed n ts foundng Regulaton. The man advantage of the proposed matchng mechansm s that t deals wth the heterogenety n the preferences of the ember States most effcently by explotng the comparatve advantage of each partcpant. Consequently and contrary to conventonal EU mmgraton polces, the matchng model embraces heterogenety rather than tryng to fnd a one sze fts all soluton. As a general concluson, we argue that, gven the alta example wth EUREA, our proposed market for tradable refugee-admsson quotas combned wth a matchng mechansm would be a perfect nstrument for ntra-eu relocaton of refugees and 33