1 Middle Easern Assimilaion, Migraion and Trade o Europe and Norh America Hisham Foad March 2009 Absrac Why are immigraion and rade flows posiively relaed? While many sudies have documened his complemenary relaionship, we canno definiively say wha facors drive his relaionship. On one hand, migraion leads o he formaion of social neworks across borders, driving down rade coss. On he oher hand, immigrans may reain a preference for heir naive counry's producs, creaing a local marke for expors. This sudy examines hese wo compeing heories by esimaing he immigraion rade linkage separaely for migrans moving from he Middle Eas and Norh Africa (MENA) o boh Europe and Norh America. While hese wo groups originae in he same locaion and have similar preferences, hey are quie differen in erms of boh income and educaion, wih MENA migrans o Norh America ending o be less numerous bu more educaed. While he greaer degree of economic assimilaion for Norh American migrans should srenghen migran nework effecs, he fac ha hese migrans are more culurally assimilaed should weaken boh nework and preference effecs. I find ha he migraion rade link is sronger for migrans o Europe, wih he sronges effec for impors. The migraion rade link is sronger for differeniaed goods han for homogeneous goods, and sronges for differeniaed goods impors ino Europe. These resuls sugges ha while nework effecs maer, immigran preferences for naive counry goods are he key facor driving he migraion rade link. The resuls in his sudy also provide quaniaive evidence of weaker assimilaion among MENA migrans o Europe, a widely acceped resul ha has had lile empirical suppor in he exising lieraure. JEL: F1, F22 Keywords: Migraion and Trade, Middle Eas and Norh Africa * Deparmen of Economics, San Diego Sae Universiy, 5500 Campanile Drive, San Diego, CA Please direc all correspondence o he auhor a I would like o hank seminar paricipans a he SDSU Brown Bag Series for heir invaluable commens on his projec. All remaining errors remain my own.
2 1. Inroducion How are flows of labor and goods across naional borders relaed? Mundell (1957) argues ha free rade can lead o facor price equalizaion, hereby reducing he incenive for labor o move o counries wih high wages. This view was challenged by Markusen (1983), who finds a complemenary relaionship under alernaive assumpions o he Heckscher Ohlin rade model. A number of recen empirical sudies have suppored his laer view ha rade and migraion are complemenary. Furhermore, hese papers have also argued ha labor migraion leads rade. There are wo heoreical channels hrough which migraion can affec rade. Firs, migraion may induce rade because immigrans bring wih hem a reained preference for heir naive counry's goods and services. This channel applies mainly o impors of differeniaed goods from he immigrans' naive counries, as hese are he goods for which immigrans may no find suiable subsiues in heir new homes. 1 Second, migraion may help o reduce he ransacion coss associaed wih inernaional rade. Alhough migrans physically leave heir naive counry behind, hey do no necessarily sever all ies wih people living here, leading o he formaion of ehnic neworks across borders. These neworks can overcome informal rade barriers, such as weak inernaional legal sysems or a lack of informaion abou formal markes. Grief (1993) presens he hisorical example of he Maghribi raders, a disinc social group wihin he Jewish Diaspora living in he Islamic Medierranean in he 11h cenury. The raders shared informaion abou foreign markes and were willing o engage in collecive punishmen agains any ren seeking raders, hereby reducing risk. Though he exising empirical lieraure on his opic has done a good job documening he posiive relaionship beween migraion and rade, less aenion has been given o he specific channel hrough which migraion affecs rade. This is an imporan deficiency in he lieraure, no only from an academic perspecive bu also for policy implicaions. If migraion lowers informal rade barriers, hen we mus consider his posiive exernaliy when discussing he meris of immigraion. If reained preferences are driving his relaionship however, hen he exernaliy benefi accrues primarily o he sending counry. In his sudy, I aemp o resolve he quesion of how migraion affecs rade by looking a wo groups of migrans: hose moving from he Middle Eas and Norh Africa (MENA) o he European Union and hose moving o Norh America. Though hese migrans come from he same place, hey are quie differen in several imporan ways. MENA migrans o Norh America are more economically assimilaed, wih beer educaion and higher incomes han heir more numerous European counerpars. Anecdoal evidence suggess ha hese migrans o Norh America are also more 1 I is possible ha preferences can be ransmied hrough migran neworks, hus affecing expors o he immigran's naive counry. For example, knowledge abou a produc only available in an immigran's adoped counry may be ransmied back o he naive counry, creaing a demand for ha produc in ha naive counry. I ignore his channel in his sudy, bu i is an ineresing opic for furher research.
3 culurally assimilaed, wih higher raes of ciizenship, local language proficiency, and preferences more in line wih naives. The unique characerisics of hese wo immigran groups allow us o explore in greaer deail he channels hrough which migraion affecs rade. The nework effecs of migraion should be increasing wih economic assimilaion, as educaed immigrans are beer able o lower rade coss hrough ehnic neworks. As culural assimilaion rises however, he nework effec may weaken since more culurally assimilaed migrans have weaker ies wih heir naive counries. The preference channel hrough which migraion affecs rade will be driven by culural assimilaion, wih more assimilaion leading o a drop in expors from he immigrans' naive counries. The immigran preference channel should no affec expors back o he immigrans' naive counries. I find ha he immigraion rade link is sronger for European migraion han for Norh American migraion. This difference is larges for impors from he MENA, suggesing ha he immigran preference channel is dominan. Immigraion does have a weaker, ye sill significan effec on expors o he MENA region, suggesing ha nework effecs do maer. Disaggregaing he level of rade, I find ha migraion induces rade mos srongly for differeniaed goods, lending furher suppor for he preference channel as he mechanism hrough which immigraion affecs rade. 2. A Profile of MENA Migraion Tables 1 and 2 presen a clearer picure of migraion from he Middle Eas and Norh Africa. Looking a Table 1, we see ha MENA migraion o Norh America and Europe has subsanially increased beween 1990 and 2000, rising from 3.9 million emigrans o nearly 5.5 million. By far, Europe ges more MENA migrans han Norh America, wih over 80% of oal MENA migrans o hese regions going o Europe. While MENA migrans o Europe are much more numerous, migrans o Norh America end o be beer educaed. In 2000, 65.1% of MENA migrans o Norh America held eriary degrees, compared wih only 12.6% of migrans o Europe. This reflecs he relaively lower migraion coss o Europe, ensuring ha only he migrans wih he highes expeced earnings would be willing o move o Norh America. For relaively low skilled migrans, he marginal increase in wages ha hey ge following migraion may be enough o overcome he low migraion coss o Europe, bu no he high coss of moving o Norh America. Table 2 presens evidence of he link beween migraion and rade. We see ha MENA counries generally have disproporionaely high rade linkages wih he counries ha receive mos of heir emigrans. For example, he op desinaion for Algerian emigrans is France. In 1990, 22.5% of Algeria's rade was wih France, a remarkable number given ha France only accouned for 6.9% of global rade flows. Though France's share of Algerian rade fell o 15.9% by 2000, his sill represened a
4 share well above France's presence in global rade. The same paern holds rue for mos MENA counries and for all bu one of he major senders. 2 One of he goals of his sudy is o explore he channels hrough which migraion affecs rade. As saed in he inroducion, his issue can be examined in he conex of economic and culural assimilaion. From Table 1, we see ha MENA migrans o Norh America are more economically assimilaed, wih higher educaion and herefore greaer income. All else being equal, he immigran nework effec on rade should be higher for Norh American migraion. However, we also have o consider culural assimilaion. Even hough MENA migrans o Norh America are beer educaed, have hey mainained close ies wih heir naive counries? Do hey sill crave he goods and services of heir naive lands? If migrans o Europe are more culurally assimilaed, hen hrough boh sronger nework effecs and immigran preferences, we should see a larger migraion rade link Anecdoal evidence suggess ha MENA migrans o Europe end o be less culurally assimilaed. There is significanly less quaniaive daa o suppor his asserion, however. Two ways o measure culural assimilaion are by looking a an immigran's proficiency wih he local language and wheher or no he immigran has become a nauralized ciizen of heir new home. Table 3 presens some evidence in suppor of weaker culural assimilaion for MENA migrans o Europe. From he 2000 US Census, 64% of migrans from he Middle Eas repored ha hey spoke English Very Well, wih lile or no difficulies. Van Tuebergen and Kalmijn (2008) find ha he language proficiency of Moroccan and Turkish migrans o he Neherlands is less han half his level. Dusmann (1994) finds similarly low local language proficiency for Turkish migrans living in Germany. While greaer language proficiency among migrans o he US may simply reflec a greaer worldwide proficiency wih English as compared o Duch or German, he conclusion regarding assimilaion is he same. Do immigrans move o a counry for only a few years hen go back o heir naive land or are hey esablishing a permanen residence here? One way o measure his is wih he fracion of migrans ha have become nauralized ciizens. The second column of Table 3 shows ha 54% of MENA migrans o he US are nauralized ciizens, a fracion mached by MENA migrans o France and in excess of he ciizenship shares of Ausria and Spain. While incomplee, hese figures sugges ha here is greaer permanence in MENA migraion o he US. If his is he case, here may very well be a sronger nework effec for Europe, since MENA migrans o Europe are going back and forh beween heir adoped and naive counries. 2 The excepion is Iran, for which he US is he op emigran desinaion. Iran is clearly a special case as here was a wave of migraion o he US in 1979 following he overhrow of he Shah and he subsequen rade embargo by he US has severely limied rade flows beween hese naions. As his rade embargo migh skew he resuls when comparing he differenial effecs of Norh American and European migraion, I esimae he empirical model wihou Iran. The resuls do no qualiaively change and ineresingly, he conclusion ha European migraion has a larger rade effec is quaniaively larger when we omi Iran.
5 3. The Migraion Trade Link One of he firs empirical sudies o documen complemenariies beween migraion and rade was Gould (1994), who insers he lagged immigran sock from a range of counries ino a graviy equaion esimaing expors and impors separaely for he Unied Saes. He finds ha a 10% increase in immigraion o he US raises expors o he immigrans' naive counry by 4.7% and impors from he naive counry by 8.3%. Tha impors are more sensiive o immigraion reflecs he fac ha immigrans reain some preference for naive counry goods. Head and Ries (1998) follow up on Gould's work using Canadian daa and esimae he relaionship wih a Tobi model o accoun for he large number of zeros in bilaeral rade daa. Though hey also find a posiive relaionship beween immigraion and rade, heir elasiciy esimaes are lower. Head and Ries argue ha he Canada's rade is primarily in commodiies or in US bound auomoive goods which do no significanly benefi from he migraion induced reducion in ransacion coss. Girma and Yu (2002) find addiional evidence for he ransacion coss explanaion by examining immigraion and rade beween he UK and 48 rading parners. They find ha he migraion rade link is significanly posiive only for counries ha are no former Briish colonies. They argue ha he rade promoing effecs of immigraion (conrac enforcemen, more symmeric informaion, ec.) do no maer so much for former colonies, as hese naions already share many similariies wih he UK (such as in he legal sysem) and informaion abou foreign markes is pleniful. Several sudies have looked a he composiion of rade, finding ha immigraion mos srongly promoes rade in differeniaed goods. Rauch and Trindade (2002) show ha he presence of an ehnic Chinese nework in a counry significanly increases differeniaed goods rade wih China. Furher evidence of he relaively sronger effec on differeniaed goods is given by Dunleavy and Huchinson (1999) looking a immigraion and rade in he lae 19h cenury as well as Blanes (2005) and Whie (2008) who look a he effecs of immigraion on inra indusry rade for Spain and he US respecively. These sudies presen evidence ha labor migraion induces inernaional rade, mos noably for differeniaed goods. My paper adds o his lieraure in wo ways. This sudy is he firs I am aware of o look a how differing desinaions of he same migran group affecs he migraion rade link. Holding he source counry consan and looking a where migrans acually go gives us greaer insighs ino why migraion affecs rade. The second conribuion of my paper is o use rade linkages o examine he degree of assimilaion of MENA migrans o Europe and Norh America. I is generally assumed ha MENA migrans o Europe are less assimilaed, bu here is lile daa o suppor his noion. I find quaniaive evidence ha MENA migrans o Europe mainain sronger rade linkages wih heir naive counries han heir American counerpars. This effec is sronges for impors ino Europe, suggesing ha i is he reained preferences of immigrans from he MENA ha drive he migraion rade link.
6 4. Empirical Specificaion The exising lieraure suggess wo channels hrough which immigraion may affec rade: nework effecs lower ransacions coss and reained preferences creae a marke for impors from immigrans' naive counries. The ransacions cos explanaion suggess ha immigraion should increase boh expors and impors beween he sending and receiving counries, while he reained preference explanaion only holds for impors ino he receiving counry. To es hese predicions, I gaher daa on immigraion and rade from a number of sources. Bilaeral migraion daa is from Docquier and Marfouk (2007) and covers he number of migrans born in a paricular sending counry now residing in a given receiving counry. The daabase covers he foreign born populaion in 1990 and 2000 for 19 sending counries in he MENA region and 19 receiving counries in Norh America and Europe. Bilaeral rade daa is aken from he UN Commodiy Trade Saisics. To disinguish goods as differeniaed, homogeneous or reference priced, I use he Rauch (1999) classificaion across goods disaggregaed a he 4 digi SITC Revision 2 level. I hen re aggregae across he hree caegories o ge oal bilaeral rade in homogeneous, reference price, and differeniaed goods across he sample for 1991 and These years are chosen for he rade saisics so ha we are measuring he effecs of he lagged immigran sock on rade, an aemp o conrol for simulaneiy in rade and migraion. 3 The immigraion rade relaionship has generally been esimaed wih graviy equaions. These empirical models have had remarkable success in predicing rade flows and may be derived from a general equilibrium model wih imperfec compeiion as in Bergsrand (1985). The baseline model o be esimaed is: Trade Mig *exp g * Lang i, j a 1 Y b Y c h* Colony d Yi, / Pop Y / Pop u i, j i, j, e Dis f j (1) We log linearize his o: ln Trade e* ln a*lnmig 1 b *ln Y Y / Pop f *lndis j g * Lang j h* Colony j u c * ln Y d* ln Y / Pop (2) In his specificaion, bilaeral rade depends on oal income in boh counries, on per capia income in boh counries, and a series of variables ha represen rade coss: disance, wheher or no he counries share a common language, and any hisory of colonial ies ha may indicae a common legal sysem or esablished rading relaionship. Migraion is prediced o reduce rade coss (and herefore increase rade) as well as increase expors from he sending o he receiving counry due o reained 3 Admiedly, his is a relaively crude mehod given ha we are using he immigran sock which should be relaively sable from one year o he nex.
7 immigran preferences. To assess wheher or no he migraion effec on rade is sronger for Europe, define a dummy variable equal o 1 if he receiving counry is in Europe and 0 oherwise hen inerac his wih log migraion: ln Trade b *ln Y f *lndis a * lnmig c * ln Y j 1 g * Lang 1 d* ln j a Y / Pop e * ln 2 * lnmig h* Colony j 1 u * EUj a3euj Y / Pop (3) The elasiciy of migraion o rade is hus a 1 for Norh America and a 1 +a 2 for Europe. Esimaes of his relaionship for 1990 and 2000 immigraion across expors and impors are given in Table 4. Equaion 3 is esimaed wih boh OLS and Tobi o accoun for he relaively high number of zeros in bilaeral rade daa. The migran sock variable may suffer from endogeneiy. Greaer rade beween wo counries may lead o increased familiariy and warmer poliical relaions beween hese naions. This lowers he cos of migraion, suggesing ha causaliy may indeed be running in he opposie direcion. To address his issue, I insrumen he migran sock following Javorcik e al (2006) and McKenzie (2005). The migran sock in 2000 is insrumened wih he migran sock in 1990, he cos of obaining a passpor in he sending counry as a percenage of gross naional income, bilaeral elephone raffic beween he sending and receiving counries, populaion densiy in he sending counry, wo dummy variables equal o 1 if he sending counry has culural resricions on women raveling alone and legal resricions ha require residens o ge governmen approval before raveling abroad, and all explanaory variables in 3. The prediced migran sock from his firs sage esimaion is hen insered back ino 3 and re esimaed. As I only have migraion daa for 1990 and 2000, he IV esimaion can only be done for 2001 rade flows. 5. Discussion The immigran sock in 1990 only has a significan effec on rade beween he MENA and Europe, while he migraion rade elasiciy is no significanly differen from zero for Norh American migraion, using eiher expors or impors as he dependan variable. I esimae ha a 10% increase in migraion from a MENA counry o a European naion increases expors o ha MENA naion by 1% and impors ino he European naion by 1.25%. Looking a resuls for 2000, I find ha migraion o Norh America only significanly affecs expors o he MENA region and his effec drops ou when conrolling for poenial endogeneiy in migraion. European migraion has an excess effec only on impors from he MENA region. I esimae ha a 10% increase in migraion from a MENA counry will raise expors o ha MENA counry by 0.62% 1.72% for Norh America, while he European expor migraion elasiciy is slighly larger bu no significanly differen. This paern is reversed for impors, wih a 10% increase in
8 migraion increasing impors o Europe by 1.84% 2.32% and having no effec on impors o Norh America. 4 For boh Europe and Norh America, he migraion rade linkage was significanly posiive for boh expors o and impors from he MENA region. This resul suggess ha nework effecs maer, as immigran preferences could only explain he linkage for impors from he MENA region. The sronger link beween migraion and rade for Europe implies one of wo possibiliies. Firs, MENA migrans o Europe mainain closer ies wih heir naive counries and exer sronger nework effecs. The fac ha he European migraion rade link is sronger for boh expors and impors in 1990 suppors his idea. Second, MENA migrans o Europe have sronger preferences for heir naive counry goods. The fac ha he European migraion rade link is sronger only for impors in 2000 suppors his view. We gain furher insighs ino his analysis by looking a how migraion affecs rade in differen caegories of goods. In paricular, do he effecs of migraion on rade vary when looking a differeniaed vs. homogeneous goods? The channels hrough which migraion can affec rade will boh be influenced by how differeniaed he goods being raded are. Homogeneous goods widely raded on organized exchanges will no benefi much from informaion flows hrough migran neworks. Trade in differeniaed goods on he oher hand is more likely o encouner he kind of informal rade barriers ha migran neworks reduce. Immigran preferences should only maer for rade in differeniaed goods, as homogeneous goods are, by definiion, he same regardless of where hey are produced or consumed. I caegorize goods rade by using he classificaion sysem found in Rauch (1999). Goods are classified ino one of hree caegories: Homogeneous, Reference Priced, or Differeniaed. Homogeneous goods are raded on organized exchanges and include such classificaions as crude peroleum (SITC Rev. 2 Code 3330) and coon yarn (SITC Rev. 2 Code 6513). Reference priced goods are no raded on organized exchanges, bu do have price daa referenced in rade publicaions. Examples include Insecicides (SITC Rev. 2 Code 5911) and Calf Leaher (SITC Rev. 2 Code 6113). Differeniaed Producs do no have a price lised eiher on organized exchanges or in rade publicaions. Examples include Jams and Marmalade (SITC Rev. 2 Code 0582) and Color Televisions (SITC Rev. 2 Code 7611). I gaher bilaeral rade daa a SITC 4 digi level of disaggregaion hen reaggregae bilaeral expors and impors ino hese hree caegories. Equaion 3 is hen esimaed for six differen dependan variables: expors and impors of homogeneous, reference price, and differeniaed goods. Table 6 gives OLS esimaes of equaion 3 across he six specificaions defined above for 2000 immigraion daa (2001 rade daa) using IV esimaion. The firs column looks a differeniaed goods expors from Europe or Norh America o MENA counries. The migraion rade link is significanly posiive, wih a 10% increase in migraion o hese regions raising expors o a MENA counry by 5.24%. The 4 The European migraion rade elasiciy was compued by adding ogeher he coefficien on log Migraion and he ineracion effec beween migraion and he Europe dummy variable.
9 European elasiciy is slighly higher (5.8%), bu is no significanly differen from zero. There is a significan difference for European migraion when looking a differeniaed goods impors, however. I esimae ha a 10% increase in European migraion will increase differeniaed goods impors ino European counries by 1.93%, while migraion o Norh America has no discernible effec. These resuls sugges ha nework effecs maer for expors o he MENA, wih migrans faciliaing rade by lowering informal he rade barriers. For differeniaed goods impors ino Norh America, nework effecs do no seem o maer, nor do immigran preferences make much of an impac. Tha migraion does affec impors ino Europe suggess ha eiher nework effecs are sronger here, or ha MENA migrans o Europe reain sronger preferences for naive counry goods. Similar paerns hold for reference price and homogeneous goods, hough he migraion rade link weakens as producs become less differeniaed. For reference priced goods expors o he MENA, he immigraion rade elasiciy is no significanly differen from zero for eiher Europe or Norh America, suggesing inconsequenial nework effecs for hese goods. Ineresingly, reference priced impors ino Norh America are significanly influenced by migraion. The European impor elasiciy is higher, hough no significanly differen from he Norh American elasiciy. Tha migraion only affecs impors of reference priced goods, suggess ha preferences are he driving channel here, and for hese goods he preference channel is he same beween he wo groups of migrans. I esimae no discernable difference beween Europe and Norh America in he migraion rade elasiciy for homogeneous goods expors o he MENA, wih a 10% increase in migraion raising homogenous goods expors by abou 2.5%. However, homogeneous goods impors are only influenced by migraion o Europe, wih an esimaed elasiciy of 0.21 compared o a saisically insignifican elasiciy for Norh America. Tha expors of homogeneous goods o he MENA are affeced by migraion suggess he imporance of nework effecs for homogeneous goods rade. This is a surprising resul, as one would expec homogeneous goods o benefi less from he rade barrier reducing effecs of migran neworks. While indeed he migraion rade elasiciy is larger for differeniaed goods, i is sill posiive for homogeneous goods, suggesing ha even for hese goods here are rade barriers. These rade barriers do no seem o maer as much when goods are coming ino Norh America, perhaps reflecing he greaer marke infrasrucure in his region. Migraion does affec impors ino Europe, which may be due o sronger nework effecs or immigran preferences. In he laer case, his suggess ha homogeneous goods are viewed as differeniaed goods by some migrans expressing a preference for naive counry producs. 6. Conclusion This sudy addresses wo issues: why does migraion affec rade and are MENA migrans o Europe less assimilaed han heir Norh American counerpars? Migraion can lower informal rade barriers hrough he formaion of neworks across borders
10 linking a migran's adoped and naive counries. In his case, migraion should have a pro rade effec on impors ino he migran's adoped home as well as expors back o heir naive counry. I do find evidence ha expors o he MENA are posiively affeced by migraion, suggesing ha nework effecs maer. These effecs may be paricularly imporan when looking a rade wih developing counries ha people in Europe and Norh America have lile informaion abou. When looking a impors ino Europe and Norh America, I find ha migraion generally only has an effec for he former. Nework effecs may be less imporan for impors, as informaion abou European and Norh American markes is pleniful. However, immigran preferences may maer a grea deal for impors. If immigrans reain a preference for heir naive counry's goods and services, hey consiue a ready made marke for MENA counries o expor o. This appears o be happening in Europe, bu no in Norh America. I has been widely acceped ha MENA migrans o Europe end o be less assimilaed han heir Norh American counerpars and he resuls in his sudy seem o confirm his. While his sudy does provide good insighs ino boh he channels hrough which migraion affecs rade and he differenial experiences of migrans o he MENA and Norh America, here are sill several unanswered quesions. One such quesion is o wha exen are Norh American neworks wih he MENA weaker han European neworks? If we could hold hese nework effecs consan, hen any remaining difference would have o be due o he immigran preference channel. One possible soluion is o gaher daa on he lengh of ime a migran has lived in a counry, wih he idea ha immigrans who have been away from heir naive counry for a long ime have weaker ies. While here is informaion on immigran enure, i is no available for all counries in his sudy. Despie his shorcoming, he resuls in his sudy are relevan o boh academics and policymakers. From an academic perspecive, hese resuls confirm ha boh nework effecs and immigran preferences are viable channels hrough which migraion induces rade, hough he preference channel appears o be much more imporan for less assimilaed migrans o Europe. This paper yields one imporan policy implicaion. Immigraion creaes posiive exernaliies for receiving counries ha need o be considered. Receiving counries benefi from increased informaion abou foreign markes (nework effecs) while sending counries are able o capure a leas some of heir emigrans' consumpion hrough increased expors.
11 7. References  Bergsrand, J. The Graviy Equaion in Inernaional Trade: Some Microeconomic Foundaions and Empirical Evidence. The Review of Economics and Saisics, 67: ,  Blanes, J. Does Immigraion Help o Explain Inra Indusry Trade? Evidence for Spain. Review of World Economics, 141: ,  Combes, P. P., M. Lafourcade, and T. Mayer. The Trade Creaing Effecs of Business and Social Neworks: Evidence from France. Journal of Inernaional Economics, 66:1 29,  Docquier, F. and A. Marfouk. The Brain Drain in Developing Counries. World Bank Economic Review, 21: ,  Dunlevy, J. A. The Influence of Corrupion and Language on he Pro Trade Effec of Immigrans: Evidence from he American Saes. The Review of Economics and Saisics, 88: ,  Dunlevy, J.A. and W.K. Huchinson. The Impac of Immigraion on American Impor Trade in he lae Nineeenh and early Twenieh Cenuries. The Journal of Economic Hisory, 59: ,  Dusmann, C. Speaking Fluency, Wriing Fluency, and Earnings of Migrans. Journal of Populaion Economics, 7: ,  Girma, S. and Z. Yu. The Link beween Immigraion and Trade: Evidence from he U.K. The Review of World Economics, 138: ,  Gould, D.M. Immigraion Links o he Home Counry: Empirical Implicaions for U.S. Bilaeral Trade Flows. The Review of Economics and Saisics, 76: ,  Greif, A. Conrac Enforceabiliy and Economic Insiuions in Early Trade: The Maghribi Traders' Coaliion. American Economic Review, 83:47 62,  Head, K. and J. Ries. Immigraion and Trade Creaion: Economeric Evidence from Canada. Canadian Journal of Economics, 31:47 62,  Javorcik, B. C. Ozden, M. Spaareanu, and C. Neagu. Migran Neworks and Foreign Direc Invesmen. World Bank Policy Research Paper No. 4046,  Markusen, J.R. Facor Movemens and Commodiy Trade as Complemens. Journal of Inernaional Economics, 14: , 1983.
12  McKenzie, D. Paper Walls are Easier o Tear Down: Passpor Coss and Legal Barriers o Emigraion. World Bank Policy Research Paper No. 3783,  Mundell, R.A. Inernaional Trade and Facor Mobiliy. American Economic Review, 47: ,  Rauch, J. Neworks vs. Markes in Inernaional Trade. Journal of Inernaional Economics, 48:7_35,  Rauch, J. and V. Trindade. Ehnic Chinese Neworks in Inernaional Trade. The Review of Economics and Saisics, 84:116_130,  Van Tuebergen, F. and M. Kalmijn. Language Proficiency and Usage among Immigrans in he Neherlands: Incenives or Opporuniies? European Sociological Review, Forhcoming,  Whie, R. Exploring a U.S. Immigran Inra Indusry Trade Link. Easern Economic Journal, 34:252_262, 2008.
13 Table 1: MENA Migraion o Europe and Norh America, 2000 Emigraion NA Share EU Share Teriary NA Teriary EU Algeria 605, % 96.1% 79.5% 11.3% Bahrain 4, % 60.8% 71.8% 34.9% Djiboui 1, % 84.2% 64.0% 32.4% Egyp 221, % 42.1% 78.3% 33.3% Iraq 224, % 39.1% 72.2% 38.4% Iran 499, % 59.4% 45.6% 32.4% Israel 133, % 23.3% 64.8% 33.6% Jordan 56, % 24.4% 61.2% 37.0% Kuwai 21, % 25.4% 80.8% 33.0% Lebanon 246, % 38.7% 56.9% 35.9% Libya 19, % 57.7% 81.0% 36.9% Morocco 1,093, % 95.3% 67.6% 10.1% Oman 1, % 56.0% 98.1% 33.4% Qaar 1, % 38.7% 95.6% 32.7% Saudi Arabia 16, % 28.2% 76.7% 34.7% Syria % 44.2% 52.8% 33.9% Tunisia 263, % 96.3% 71.3% 12.6% Turkey 1,940, % 96.0% 56.4% 6.5% UAE 2, % 41.4% 98.1% 32.7% Yemen 20, % 39.6% 32.6% 35.4% Toals 5,482, % 80.0% 65.1% 12.6% * Emigraion is he oal number of people born in he Sending counry ha are residing in Europe or Norh America in NA Share is he share of emigrans living in Norh America, while EU share is he share of emigrans living in Europe. Teriary NA refers o he percenage of migrans living in NA from each sending counry who hold a eriary degree while Teriary EU gives he share for migrans living in Europe. Migraion daa is from Docquier and Marfouk (2007).
14 Table 2: Immigraion and Trade Top Receiver Emigrans Toal Trade Trade Share World Share Algeria France 511,971 $28, % 4.7% Bahrain UK 2,335 10, % 4.7% Djiboui France 1, % 4.7% Egyp USA 96,660 18, % 15.7% Iran USA 250,785 42, % 15.7% Israel USA 89,385 62, % 15.7% Jordan USA 39, % 15.7% Kuwai USA 12,505 26, % 15.7% Lebanon Canada 55, % 4.0% Libya USA 7,024 16, % 15.7% Morocco France 425,096 18, % 4.7% Oman UK , % 4.7% Qaar USA % 15.7% Saudi Arabia USA 10, , % 15.7% Syria USA 47,660 35, % 15.7% Tunisia France 184,603 14, % 4.7% Turkey Germany 1,272,000 82, % 8.1% UAE USA 1,322 84, % 15.7% Yemen USA 11, % 15.7% Toals Germany 5,482, , % 8.1% * Top receiver is defined as he counry wih he larges foreign born populaion from he sending counry. Emigrans is he oal number of emigrans from he sending counry residing in he op receiver. Toal Trade is he oal value of expors and impors in millions of dollars for he sending counry. Trade Share is he share of oal rade wih he Top Receiver, while World Share is he Top Receiver's share of global rade. Immigraion daa comes from Docquier and Marfouk (2007), while rade daa is from he UN Commodiy Trade Saisics.
15 Table 3: Culural Assimilaion of MENA Migrans Desinaion Language Proficiency Ciizenship Ausria a 30% Nauralized Ciizens France b 54% Nauralized Ciizens Germany c Neherlands d 26.2% of Turkish men and 9.5% of Turkish women have Good or Very Good proficiency in German 27% of Moroccan and 21% of Turkish immigrans have lile o no difficuly speaking Duch Spain e 26.6% Nauralized Ciizens USA f 64% speak English Very Well 54.2% Nauralized Ciizens a 2001 Ausrian Naional Census b 1990 French Naional Census c Dusmann (1994) d Van Tuebergen and Kalmijn (2008) e 2001 Spanish Naional Census f 2000 US Naional Census
16 Table 4: Migraion and Aggregae Trade Flows, 1990 Expors Impors OLS Tobi OLS Tobi Migraion [0.249] [0.159] [0.158] [0.107] Mig*EU [0.005] [0.003] [0.000] [0.000] EU [0.971] [0.975] [0.007] [0.000] Y i [0.000] [0.000] [0.000] [0.000] Y j [0.000] [0.000] [0.000] [0.000] Y i /Pop i [0.000] [0.007] [0.000] [0.000] Y j /Pop j [0.000] [0.000] [0.000] [0.000] Disance [0.000] [0.000] [0.001] [0.002] Language [0.111] [0.066] [0.052] [0.014] Colony [0.917] [0.953] [0.740] [0.750] Adj. R Censored Obs * OLS and Tobi esimaes of equaion 3 in he ex. Migraion daa for 1990 and 2000 are from Docquier and Marfouk (2007). Expors refer o expors from European or Norh American counries o MENA counries, while impors originae in MENA counries. Trade daa is from UN Comrade, Income daa is from he IFS, while disance, language, and colonial linkage daa is from he CEPII Geodesic Disance Daabase. P values are given in brackes.
17 Table 5: Migraion and Aggregae Trade Flows, 2000 Expors Impors OLS Tobi IV OLS Tobi IV Migraion [0.020] [0.085] [0.500] [0.592] [0.723] [0.565] Mig*EU [0.464] [0.697] [0.727] [0.000] [0.000] [0.001] EU [0.394] [0.368] [0.060] [0.000] [0.000] [0.029] Y i [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] Y j [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] Y i /Pop i [0.088] [0.218] [0.459] [0.006] [0.004] [0.012] Y j /Pop j [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] Disance [0.000] [0.000] [0.000] [0.010] [0.013] [0.000] Language [0.034] [0.000] [0.001] [0.271] [0.065] [0.061] Colony [0.390] [0.374] [0.219] [0.648] [0.804] [0.955] Adj. R Censored Obs * OLS, Tob and IV esimaes of equaion 3 in he ex. For he IV esimaion, he migran sock in 2000 is insrumened wih he migran sock in 1990, he cos of obaining a passpor in he sending counry, elephone raffic beween he sending and receiving counries, populaion densiy in he sending counry, culural and legal resricions on female ravel in he sending counry as well as all explanaory variables in equaion 3. Migraion daa for 1990 and 2000 are from Docquier and Marfouk (2007). Expors refer o expors from European or Norh American counries o MENA counries, while impors originae in MENA counries. Trade daa is from UN Comrade, Income daa is from he IFS, while disance, language, and colonial linkage daa is from he CEPII Geodesic Disance Daabase. P values are given in brackes.
18 Table 6: Migraion and Aggregae Trade Flows, 2000 Differeniaed Reference Price Homogenous Expors Impors Expors Impors Expors Impors Migraion [0.000] [0.144] [0.396] [0.012] [0.001] [0.856] Mig*EU [0.402] [0.000] [0.971] [0.141] [0.313] [0.000] EU [0.501] [0.062] [0.033] [0.003] [0.850] [0.035] Yi [0.000] [0.000] [0.000] [0.000] [0.018] [0.000] Yj [0.016] [0.000] [0.000] [0.000] [0.004] [0.128] Yi/Popi [0.666] [0.000] [0.226] [0.941] [0.835] [0.007] Yj/Popj [0.000] [0.000] [0.000] [0.003] [0.000] [0.000] Disance [0.009] [0.011] [0.000] [0.000] [0.000] [0.017] Language [0.918] [0.013] [0.013] [0.184] [0.906] [0.004] Colony [0.602] [0.287] [0.681] [0.837] [0.696] [0.224] Adj. R * See he noes for Table 5 for a descripion of he daa. All esimaes are from he IV specificaion described he empirical secion of he ex. Bilaeral rade a he 4 digi SITC Revision 2 level is colleced hen aggregaed ino Differeniaed, Reference Price, and Homogeneous goods rade according o he classificaion given by Rauch (1999). P values are given in brackes.