A Simple General Equilibrium Model with International Labour Market Linkages

Transcription

1 Jean Mercener & Ncolas Schmtt Unverst of Cerg Pontose Unverst of Geneva and 10 Smon Fraser Unverst A Smle General Equlbrum Model wth Internatonal Labour Market Lnkages INTRODUCTION A N IMPORTANT ASPECT OF GLOBALIZATION s the growng mortance of nternatonal labour moblt. Although, n overall volumes, mgratons reman largel tradtonal (.e. non-sklled labour from non-oecd countres to OECD countres; see Gross and Schmtt, 2002), there s growng ntra-oecd countr labour moblt and there s a erceton that a substantall hgher level of nternatonal labour moblt ma be just around the corner. Current ntra-oecd labour moblt ehbts a relatvel hgh mean and varance wth resect to skll level as comared to worldwde nternatonal labour moblt. In ths regard, Canada and the Unted States are robabl at the forefront of that evoluton due n large art to the relatvel hgh degree of labour moblt at the natonal level and to the natural ressure thus created on cross-border moblt under current mgraton olces. In Euroe, deste legall free moblt wthn the Euroean Unon, mgraton flows, though rsng, reman modest. For nstance, t s rumoured that well over 250,000 French ctzens have recentl moved to London and that a hgh share of these mgrants are oung, hghl sklled and entrereneural ndvduals fleeng the French bureaucrac n favour of Brtsh lassez-fare. The number of cross-border commuters s growng steadl n border regons (for nstance between Swtzerland and neghbourng countres). Wth radl agng oulatons, German, Ital, Austra and Swtzerland are adotng olces amed at attractng and facltatng the mgraton of sklled ndvduals. 1 However, even f moblt wthn the Euroean Unon were to ncrease dramatcall, t s unlkel that t wll ever be able to contan 1

2 MERCENIER & SCHMITT the costs of a radl agng oulaton. Hence, mgraton wll have to come from outsde the Euroean Unon (Fertg and Schmdt, 2002). What are the economc costs and benefts of nternatonal labour moblt n a modern econom? Should t be encouraged? Should t be encouraged for some sklls but not for others? Is nternatonal labour moblt of sklled workers an mortant factor for secalzaton? Although these questons are nterestng, the requre new tools. In artcular, the estng lterature largel treats sklls as homogeneous and dentcal across countres. If nternatonal labour moblt wth homogeneous sklls s otentall mortant because of ts scale effect, such treatment of nternatonal moblt and ts mact on modern economes s obvousl oversmlfed. For hgh-value-added manufacturng and servce sectors, we need to dfferentate sklls and, more mortantl, assgn a sgnfcant role to hghl sklled ndvduals n the roducton of goods and servces. The man urose of ths stud s to suggest a modelng aroach to hghlght the otental economc effects of mgraton when dfferentatng labour b sklls. One could magne buldng an aled general equlbrum model wth these features, and our stud could be consdered a tentatve frst ste n ths drecton. The stud s organzed as follows. In the secton enttled Trade Lberalzaton and Internatonal Labour Moblt, we revew the man channels through whch nternatonal labour moblt affects the econom. In the net secton, enttled A Model of Trade, Wage Dstrbuton and Internatonal Labour Moblt, we focus on one artcular mechansm based on trade and ncome nequalt. We develo a model where ndvduals choose to mgrate wherever ther sklls suort the hghest wage. In the followng secton, we conduct smulaton eercses usng fcttous data n order to hghlght the basc mechansms at work. TRADE LIBERALIZATION AND INTERNATIONAL LABOUR MOBILITY THE TRADITIONAL HECKSCHER-OHLIN APPROACH to nternatonal trade and factor moblt redcts that nternatonal factor moblt wll not occur wth free nternatonal trade of roducts snce all effcenc gans can be realzed b nternatonal trade alone. Ths classc result due to Mundell (1957) thus suggests that trade and nternatonal labour moblt are substtutes. Because the assumtons underlng the neoclasscal aroach are qute strngent, t s eas to fnd models where ths substtuton does not hold and, therefore, where the ressures to mgrate ncrease wth freer trade. Such comlementart between trade and labour moblt can be obtaned n a varet of cases (for surves and a dscusson, see Fan, de Melo and Zmmermann, 1999; and Venables, 1999). In fact, man cases have now been dentfed and t s robabl 2

3 A SIMPLE GENERAL EQUILIBRIUM MODEL far to sa that, toda, economsts lace a greater emhass on such comlementart than on the more classc substtuton between trade and factor moblt. Ths s mortant because comlementart between trade and factor flows mles that the economc effects of ntegraton can be ver dfferent wth or wthout nternatonal factor moblt, and thus wth or wthout ntegraton of dfferent labour markets. We frst revew some of the man redctons about the relatonsh between trade and factor flows comng from the workhorse models used n the feld. MAIN TRADE MODELS TODAY, THE MOST COMMONLY USED TRADE MODELS are wthout doubt the Heckscher-Ohln model, the secfc factor model, the ntra-ndustr trade monoolstc cometton model, and the core-erher model of economc geograh. In a world wthout dstorton where the same factors of roducton are used to roduce ever good, the Heckscher-Ohln model redcts substtuton between trade and factor flows. Essentall, n ths world, smaller nternatonal rce dfferences for a roduct lead to smaller dfferences n factor rces. Thus, f there s no ncentve to move factors across borders before trade lberalzaton, freer trade certanl does not create nternatonal factor moblt. And, f there s an ncentve to move factors across borders before trade lberalzaton, ths ncentve necessarl decreases wth freer trade. In other words, the Heckscher-Ohln aroach leads to the classcal substtuton between trade and factor moblt. Ths result does not necessarl hold wth the secfc factor model. Ths ont s mortant because factor secfct s robabl a ke element of modern economes. It s ndeed far to sa that, wth globalzaton, there s an overall ncrease n factor secfct (ncludng labour secfct) through factor secalzaton. To see wh factor secfct could lead to comlementart between trade and nternatonal labour moblt, suose that catal or natural resources are secfc and that trade lberalzaton decreases the rce of the mortable wthout ncreasng the rce of the eortable. If labour s the ntersectoral moble factor n ths econom, the nomnal wage wll fall. However, the real wage ma rse (n terms of the mortable) or fall (n terms of the eortable) wth ambguous effects on the ncentve to mgrate. Suose now that labour s secfc to an mort-cometng sector. B lowerng ts real return, trade lberalzaton ncreases the ncentve to mgrate to another countr. The mortant ont here s that trade lberalzaton does not necessarl lead to a lower ncentve to mgrate, as n Mundell s stor, because factor secfct adds a rch set of ossble nteractons between factors. As a result, factor flows accomanng trade lberalzaton ma make economes more dfferent. 2 3

4 MERCENIER & SCHMITT Comlementart between trade and factor flows also arses n models of merfect cometton and ncreasng returns to scale. Suose for eamle that one sector s characterzed b Dt-Stgltz monoolstc cometton, whle the factors of roducton used n that sector conform to the Heckscher- Ohln model (see Venables, 1999, for such a model). Consumers bu all the dfferentated roducts from ths sector, wth constant elastct of substtuton (CES) between roducts. Ths aroach adds several mortant elements. Frst, an absolute advantage now matters snce a larger countr mles a greater number of dfferentated roducts and a more comettve envronment. When the dfferentated roducts are used as ntermedate nuts n other sectors, then the more comettve envronment amounts to a cost reducton n these sectors, translatng nto hgher returns to other factors used n roducton. Second, frms tend to locate n the larger countr because consumers send more on the dfferentated roducts. Ths means a hgher demand for factors of roducton n the larger countr, and thus hgher factor rces. Ths combned wth a lower cost of lvng (snce there s more cometton n the larger countr) also contrbute to rase factor rces n terms of the numerare. Not surrsngl, factor moblt could be destablzng n ths model. Full agglomeraton occurs onl n the unlkel case of erfect moblt of all factors. Even f not ever factor s moble (sa one of two are not moble), trade lberalzaton ma lead the smaller econom to become even smaller, resultng n a coreerher te of structure (Krugman, 1991). Whether ths occurs n a contnuous or a dscontnuous fashon deends on the secfct of the model. In artcular, for ths to occur, at least one factor of roducton must be erfectl moble. Hence, there s comlementart between trade and labour moblt snce free trade does not elmnate the ncentves to mgrate. As can be easl magned, the economc mact of relang nternatonal mgraton rules could be qute sgnfcant n such an envronment, 3 wth ossble core-erher te of outcomes. 4 Other mechansms leadng to comlementart between trade and factor flows could be magned, but the would not necessarl be relevant for develoed economes lke those of Canada, the Unted States or Western Euroean countres. It should be clear, for nstance, that mechansms based on dfferences n technolog among countres are robabl not good canddates sml because t s unlkel that substantal technologcal dfferences est among Western countres. Other avenues are not ver romsng ether. The nclude mgraton as a resonse to adjustment lags (e.g. n nvestment) and comlementart due to the resence of mgraton networks. These networks ml a otental ostve renforcement effect between trade and mgraton. Although ths elanaton certanl has valdt for countres lke the Unted States or 4

5 A SIMPLE GENERAL EQUILIBRIUM MODEL Canada wth resect to the rest of the world, 5 t s robabl not ver sgnfcant between two develoed economes. Factor secfct, merfect cometton, and geograh and trade seem the most relevant buldng blocks to cature mortant elements of modern economes and the effects of labour moblt n a freer trade envronment. However, two mortant roblems must be addressed. Frst, the geograh and trade aroach s notorousl senstve to assumtons and qute comle to al. Integratng such a buldng block nto an aled general equlbrum model s not an eas task. Second, although t s straghtforward to consder sklled labour as the moble factor and unsklled labour as the mmoble factor, nothng n the trade and geograh aroach actuall lnks these factors to labour. In other words, the man mechansms at work are ndeendent of the functonng of labour markets or of the role of workers. For these reasons, we turn our attenton to trade models characterzed b labour heterogenet. MODELS OF LABOUR HETEROGENEITY AND TRADE AS SHOULD BE CLEAR FROM ABOVE, t s ossble to re-nterret the monoolstc cometton/geograh and trade te of model as a model of nternatonal moblt of sklled labour (see Commander, Kangasnem and Wnters (2002) for more detals). We want to deart from ths model b ntroducng more structure on the frm s roducton sde to account for labour heterogenet. The man advantage s to have dfferentaton at the nut level, secfcall among workers, ntroducng the ossblt of nternatonal movements of workers wth well-defned characterstcs. B dong so, one can address secfc concerns, such as: Would a countr lke Canada lose frms (or ndustres) usng ver hghl sklled labour f nternatonal labour moblt was made easer? What would be the mact on other sectors and on wages? Could Canada become a erheral regon for actvtes where market sze and sklls are mortant features? Models of ths te work as follows. The roducton technolog requres sklled ndvduals (let s call them entrereneurs) along wth labour and/or catal n order to roduce goods. These models have two ke asects. Frst, sklled labour s a comlementar nut to other factors of roducton. Hence, there are non-convetes n the model, so that a gven level of talent (skll) can have a ver dfferent mact deendng on the sze of the market. Ths s a Rosen suerstar te of effect (Rosen, 1981). Second, the mact of talent s modeled n one of two was, namel as an effcenc effect or as a demand effect. Consder an entrereneur eercsng hs talent through a frm. When ths talent translates nto an effcenc effect or a roducton effect, the total roducton derved from a gven set of nuts s hgher the more the entrereneur s talented. Whether t s due to organzatonal or marketng talent, a more able 5

6 MERCENIER & SCHMITT entrereneur s sml able to roduce more than a less able entrereneur. Ths roductvt effect s useful because t can be ncororated nto a model wth homogeneous roducts. Hence, n such a model, the effcenc effect translates nto a frm-sze effect and a roft effect (gven a fed sul of entrereneurs). The set-u s artcularl smle n a model where ndvduals choose the role the want to la, such as the choce between actng as an entrereneur (hrng other workers and/or catal) or as a worker. In ths case, the number and the sze dstrbuton of frms become endogenous. Not surrsngl, the most talented ndvduals act as entrereneurs whle less talented ndvduals act as workers. Consequentl, f (workers ) wages rse for an reason, t mles that, ceters arbus, the number of frms (or entrereneurs) wll decrease. The basc model s due to Lucas (1978) and t has been used b Murh, Shlefer and Vshn (1991), Rauch (1991) and Schmtt and Souberan (2002). For nstance, Rauch (1991) nvestgates the connecton between the attern of trade and the attern of mgraton n a Heckscher-Ohln framework, whle Schmtt and Souberan (2002) use ths framework to nvestgate theoretcal asects of the nternatonal moblt of brans n a two-countr envronment. 6 An alternatve wa to model the mact of talents s through a demand effect. In ths case, the entrereneur s ablt s not assocated wth her roducton ablt but wth how she (or her roduct or servce) s vewed b consumers. Hence, there s a demand-volume effect assocated wth hgher sklled entrereneurs (thnk about entertaners). Manasse and Turrn (2001) model ths case b assumng a corresondence between the entrereneur s skll and roduct qualt. The then assume a corresondence between the qualt of a roduct and the volume of demand for t. Ths means that roducts are dfferentated at least along the qualt sectrum. 7 It also mles that merfect cometton s the more natural envronment n whch to set ths demand effect of sklled entrereneurs. So, what are the economc effects of nternatonal labour moblt nferred b ths aroach? To llustrate wh labour heterogenet ma ml sgnfcant economc effects comared to a more standard aroach, consder the followng case. A standard aroach wth homogeneous labour tcall assumes that labour and catal are substtutes n roducton (as n the trade and geograh aroach; see Venables, 1999). Hence, nternatonal labour moblt essentall mles some substtuton awa from catal for the host countr and scale effects n roducton. When sklled entrereneurs and labour and/or catal are requred to roduce dfferentated goods, the ntroducton of nternatonal labour moblt of sklled entrereneurs has a drect mact on the number of roducts and, thus, the number of frms n an ndustr, as sklled entrereneurs are requred comlementar nuts to other factors of roducton (whether the entrereneurs effect oerates through effcenc or demand). When a countr 6

7 A SIMPLE GENERAL EQUILIBRIUM MODEL loses sklled entrereneurs, there are non-trval sectoral and general equlbrum consequences for t. 8 Ths ma suggest that the out-mgraton of sklled ndvduals has manl negatve economc effects. But t s not alwas the case. Several arguments suggest that out-mgraton can favour growth n the countr losng sklled labour. Such s the case, for nstance, when human catal formaton s boosted b the rosects of emgraton. Here, emgraton rovdes more oortuntes for sklled labour than the confnes of the domestc market, thus offerng a otentall hgher return on human catal nvestment. The average level of human catal ma be hgher followng an out-mgraton of sklled labour than n the absence of moblt (Wldasn, 2002; Stark, Helmensten and Prskawetz, 1998). 9 In the same ven, arguments can be made about ncreased labour market ntegraton as a mean to enhance the fleblt and functonng of natonal labour markets (Wldasn, 2000). These effects are dnamc n nature and wll not be consdered further here. However, another otental ostve effect of out-mgraton can be taken nto account n the roosed aroach. It s the effect lnked to the ossblt that ndvduals ma not be able to eercse full ther talent n ther countr of orgn but could do so n the countr of mmgraton. Whether ths s due to dfferences n market sze or to the absence of comlementar nuts n the countr of orgn does not matter. The mortant ont s that new knowledge, roducts or servces ma emerge from mgraton, beneftng both the countr of emgraton and the countr of mmgraton. Another wa to ut t s to recognze that t s effcent for the world that ndvduals can mgrate to a countr where market sze or comlementar nuts allow them to create new roducts, servces or knowledge that the would not have created otherwse. 10 An mortant queston s, of course, whether ncororatng labour heterogenet and the role of entrereneurs nto an aled general equlbrum model s relevant to the Canada-U.S. case. Frst, t must be clear that such an aroach makes more sense for hgh-value-added servce and manufacturng sectors (lke hgh technolog, botechnolog and the lke) than for more tradtonal manufacturng sectors. In other words, t s relevant for knowledge-based sectors more than for an other sector. Second, the fact that hghl sklled ndvduals are (or could be) entrereneurs n one countr or another s clearl a statc long-term effect. Consequentl, lke most aled general equlbrum analss, ths one has essentall a long-term horzon. Thrd, the fact that these ndvduals are entrereneurs should not necessarl be taken lterall. It sml means that hghl sklled entrereneurs are essental roducton nuts. Whether the are lterall the resdual clamant s not crucal to the stor. 7

8 MERCENIER & SCHMITT A MODEL OF TRADE, WAGE DISTRIBUTION AND INTERNATIONAL LABOUR MOBILITY T HE COMPARISON BETWEEN CANADA AND THE UNITED STATES over the last two decades or so generall leads to three ke observatons (Harrs and Schmtt, 2003): () there s ncreased wage nequalt across skll grous n both Canada and the Unted States; () there s a roductvt ga between the two countres; and () there was sgnfcant growth n trade and foregn drect nvestment followng trade lberalzaton n North Amerca. The model roosed below can take nto account the frst two henomena. B takng nto account, we do not necessarl mean an endogenous determnaton of all the relevant effects, but a model that can endogenze the frst element and that could be calbrated to account for the second. The model develoed n ths secton bulds on Manasse and Turrn (2001). The man urose of ther stud s to lnk labour heterogenet n a countr and nternatonal trade (or technologcal change). Ther am s not to nvestgate the economc mact of the nternatonal movement of sklled workers, but rather to elan ncome nequalt through trade, technologcal change and globalzaton. We consder ths lnk as one of the man causes of the nternatonal movement of sklled workers among develoed countres. Sml ut, f t s true that, everthng else beng equal, trade (or technologcal change) creates ncome nequalt wthn a countr and across countres, then t must also create ncentves for ndvduals to move across borders n order to take advantage of these nequaltes because t lkel means that the return to a gven skll level s becomng ncreasngl dfferent among countres. An eas wa to cast the comlementart between trade and nternatonal labour flows mled b ths model s to sa that trade tends to create ncome nequaltes wthn and across countres (n the resence of ostve transort costs) and that nternatonal labour moblt tends to mtgate these nequaltes. The rncle here s essentall the same as n more tradtonal models of nternatonal labour moblt, ecet that t works not onl at the countr level but also at the ndvdual level. In a standard model of nternatonal labour moblt wth a homogeneous labour force, the necessar ngredent for nternatonal moblt s a dfference n (unform) countr real wages. Wth labour heterogenet, the average wage ma dffer not onl across countres but across ndvdual wages. Ths means that, for an ndvdual s skll level, the wage dfference between two countres can be ostve, negatve or nl, and that f nternatonal trade creates more nequaltes n the absence of nternatonal labour moblt, then trade lberalzaton and nternatonal moblt ma nduce certan categores of workers to emgrate to a artcular countr and other categores to mmgrate. It s the economc consequences of ths artcular lnk 8

9 A SIMPLE GENERAL EQUILIBRIUM MODEL between trade lberalzaton, nternatonal moblt and the locaton of frms and entrereneurs that we want to nvestgate. Consder two countres ndeed, j wth two sectors ndeed,. Sector (called hereafter Hgh Tech) roduces dfferentated goods (that are merfect substtutes), whle sector (called hereafter Low Tech) roduces a homogeneous good. Each countr has two factors of roducton: entrereneurs and raw labour. Unsklled labour s homogeneous, n fed overall sul and comettvel rced. In contrast, an entrereneur s a secfc factor used onl n the Hgh Tech sector. The domestc sul of entrereneurs s also n fed overall sul, but ths factor s dfferentated accordng to sklls n such that n [ n, n ]. The entrereneur s the resdual clamant n the frm where she eercses her talent. The roducton of each dfferentated good requres one unt of entrereneur and a varable amount of raw labour roortonal to the quantt roduced. The entrereneur s skll mroves the qualt of the roduct and thus the quantt demanded. Hence, roducng dfferentated goods has both a horzontal comonent n the Dt-Stgltz tradton and a vertcal comonent snce qualt matters. In the absence of nternatonal moblt of entrereneurs, and unlke the tcal Dt-Stgltz model, there s a fed number of frms determned b the sul of entrereneurs. 11 However, each frm earns zero roft snce the entrereneur catures the entre quas-rent generated b her talent. The roducton n the Low Tech sector s ver standard: frms are rce takers and the roducton functon s Rcardan for raw labour. In the current verson of the model, we assume ths homogeneous roduct to be a urel non-traded good. Internatonal trade of goods roduced n Hgh Tech nvolves two secfc costs (both eressed n terms of raw labour): a varable cost, whch can be nterreted as a transort cost (of the ceberg te), and a fed cost nterreted as a cost to access a foregn market (for nstance, to establsh a sales network). Ths fed cost of trade s crtcall mortant because t arttons frms nto two subsets: those who are able to eort and those who are confned to the domestc market. The demand sde comes from a standard Dt-Stgltz sub-utlt functon, ecet for the role of qualt. Hgher erceved roduct qualt ncreases utlt and thus the demand for the roduct. Manasse and Turrn adot a relatvel ad hoc secfcaton snce the sml assume the estence of a functon T mang technolog and the entrereneur s skll nto a qualt comonent as erceved b consumers. Hence, the reresentatve consumer s utlt functon, over the set of roducts N avalable to hm, takes the general CES form: σ 1/ σ 1/ σ ( σ 1)/ σ ( ) (1) U = T a d, σ>1, n N n n 9

10 MERCENIER & SCHMITT where d n s the quantt consumed of roduct n and T n (a) s the qualt of the roduct as evaluated b the reresentatve consumer and whch deends on the entrereneur s skll n used to roduce good n and on the technolog of roducton reresented b the arameter a. It s mortant to underlne here two characterstcs of the model. Frst, f raw labour s used n roorton of the quantt roduced of one dfferentated good, one unt of talent s used whether the market s small or large. Second, onl talent (for gven technolog) adds qualt to a roduct. In other words, even f horzontal dfferentaton needs both the rmar factors and one entrereneur, vertcal dfferentaton requres the skll of an entrereneur onl. It s these non-convetes n roducton and n consumton that gve a suerstar - Rosen-te flavour to the model. Gven (1), countr s demand for a varant takes the followng form: E E,,n j,,n = T,n(a ) c = Tj,n(a j ) c. j / τ j, Hence, the demand for one varant deends on ts qualt T,n (a ), T j,n (a j ), on ncome devoted to the consumton of dfferentated roducts,, and on the rce aggregator over avalable dfferentated roducts ndvduall rced and j adjusted for the (ceberg-te) barrer to trade τ j,. These demands deend on the entrereneur s skll through two channels: the qualt erceved b consumers and the rce of the roduct. Because the utlt functon s CES, frms wth the same technolog have the same mark-u. Thus, hgher roduct qualt sml translates nto a hgher volume of sales and not nto a hgher rce. Snce the entrereneur s the resdual clamant, her wage s equal to the oeratng roft. Ths roft s dfferent deendng on whether the frm eorts or not. If t does not trade, then the wage of the non-trader-entrereneur low w,n s sml equal to: c low,n [ ],, n (2) w = v E, 10

11 A SIMPLE GENERAL EQUILIBRIUM MODEL where v s the varable unt cost as well as the rce of raw labour (Rcardan technolog). Of course, the roft of the entrereneur deends on her skll. If the entrereneur s tradng, her wage s equal to: E,j,n w,n = v E + γ φ,,n v, n, τ,j hgh η (3) [ ] where the last term s the fed cost of eortng. 12 Ths cost s assumed to declne (η<0) wth the number of te-n eorters (enetraton of foregn markets s easer when man roducers are wllng to sell abroad). The man dfference wth the non-trader-entrereneur s that the entrereneur s skll allows for sales at home and abroad. In other words, talents gan from market sze. In ths case, the trader-entrereneur s wage ncreases more than roortonatel wth skll. Because of the dfference n market sze, the general relatonsh between the entrereneur s wage and skll s llustrated n Fgure 1. FIGURE 1 RELATIONSHIP BETWEEN ENTREPRENEUR S WAGE AND SKILL w (s) s z s s 11

12 MERCENIER & SCHMITT Entrereneur n chooses to be a trader or a non-trader deendng on whether eortng rovdes a hgher oeratng roft or not. At low levels of n, the non-trader entrereneur s wage ncreases wth skll. As soon as the frm trades, her wage ncreases more than roortonatel as a gven level of entrereneur s talent now reaches a much larger market. The eortng frm necessarl generates a hgher gross roft than a non-eortng frm. However, because eortng nvolves a fed cost, onl the most talented entrereneurs artcate n the eort market. Of course the skll level z of the entrereneur who s just ndfferent between tradng and not tradng s endogenous. Therefore, at z, we have low,z w = w hgh,z. Trade lberalzaton, whether through a lower transort cost or a lower fed access cost, boosts the wage of the traderentrereneur (snce the foregn market can now be more easl accessed) and thus decreases the level of the crtcal skll necessar to artcate n the eort market. Trade lberalzaton creates wage nequalt nsofar as the relatve earnngs of a trader-entrereneur rses comared to the earnngs of a nontrader entrereneur. It s mortant to note that, n the absence of mgraton, the total number (and the range of roduct qualt) of goods roduced (and thus of entrereneurs) n each countr s eogenousl fed. Ths s not the case for consumton snce the number of goods consumed deends on the number of roducts that are traded. 13 Let s now ntroduce nternatonal labour moblt. Suose frst that onl sklled entrereneurs can move between the two countres. A sklled entrereneur can move across the border and be ether a trader- or a non-trader entrereneur n the other countr. In other words, beng of one te n one countr does not re-determne the te of entrereneur n the other countr. 14 Snce we have just establshed that, n one countr, an ndvdual wth skll n chooses to be a trader or a non-trader accordng to low w,n w Ma, con hgh,n con, ths entrereneur wll mgrate from countr to countr j f and onl f (4) low w,n w Ma, con hgh,n con w < Ma j w, ρ j,, low hgh jn, jn, con con j 12

13 A SIMPLE GENERAL EQUILIBRIUM MODEL where ρ, j s the fed cost of movng to countr j and are the consumer rce ndces n each countr. Smlarl, an entrereneur mgrates from countr j to countr f and onl f con, con j (5) low hgh w j,n w j,n w Ma, con con < Ma j j w low hgh n, n,, con con ρ j,. It s of course qute ossble that for some range of sklls, sklled ndvduals wll want to move from one countr to another, whle for other ranges, the wll have no ncentve to mgrate or wll have an ncentve to mgrate n the other drecton. In other words, t s qute ossble n ths model that the attern of mgraton of sklled ndvduals would be a two-wa mgraton. Obvousl, even wthn a sngle sector, the ossble mgraton atterns are much rcher than n more tradtonal models of nternatonal labour moblt. Consder now the ossblt that unsklled labour mgrate across the border. Snce v s the wage of unsklled labour, workers wll move from countr to countr j f (6) v con v < ρ,j, j con j where ρ, j s the (statc) cost of movng across the border for unsklled labour. Unsklled labour moves n the other drecton f (7) v con v ρ j, >. j con j Not surrsngl, onl a one-wa mgraton attern s ossble wth nternatonal moblt of unsklled labour. The man dffcult wth havng nternatonal mgratons (of sklled or unsklled labour) s that t affects the demands for the roducts snce workers and entrereneurs are also consumers. In the case of entrereneurs, we need to take nto account the dstrbuton of entrereneurs and truncate t accordng to who s leavng and who s stang. Smlarl, these mgrants affect the dstrbuton of entrereneurs n the host countr. In addton, we need to know whch role (trader or non-trader) the la n each countr. In the net secton, we develo a smle general equlbrum model where, usng artfcal data, we nvestgate the senstvt of trade and labour flows to eogenous shocks. 13

14 MERCENIER & SCHMITT SIMULATION EXERCISES W E NOW USE A TWO-COUNTRY VERSION OF THE ABOVE MODEL n order to nvestgate the model s senstvt to nternatonal labour moblt. The equatons used n ths model are resented n the Aend. We roceed n two stes. Frst, we set the model n a free-mgraton envronment (mgraton costs are set to zero) and n a quas free-trade envronment (the er unt transort cost s equal to zero but the fed eort cost s ostve), and we sml ask: Gven a secfc asmmetr between the two countres, what are the effects of ntroducng free nternatonal moblt of entrereneurs? Assume that the two countres are strctl dentcal. There s clearl no ncentve to mgrate (for an skll level, the wages are the same across countres). But there s nonetheless trade between the two countres (at least as long as the ceberg transort cost s not rohbtve) snce roducts are dfferentated both horzontall and vertcall. Not surrsngl, the model bols down to a standard ntra-ndustr trade model wth two factors of roducton (sklled labour and raw labour). In order to nvestgate the effects of nternatonal labour moblt, we must ntroduce asmmetres between the two countres. We consder four dfferent tes of asmmetres: () Countr 1 has a larger endowment of sklled ndvduals than Countr 2 (but the same endowment of raw labour); () Countr 1 has a larger endowment of unsklled labour than Countr 2; () Frms eortng from Countr 1 faces a hgher fed eort cost than frms eortng from Countr 2, and (v) The technolog n the Low Tech sector s more roductve n Countr 1 than n Countr 2. For each of these cases, we comare the equlbrum wth and wthout nternatonal moblt of sklled labour (entrereneurs). We dect the results wth two grahs: one reresentng the dstrbuton of entrereneur s wage er skll (smlar to Fgure 1), the other dslang the dstrbuton of frms n each countr er skll (and thus qualt). We set the model n such a wa that the ntal dstrbuton of frms wthout nternatonal moblt (and thus the ntal dstrbuton of sklls) s the same and s unform n both countres. Consder the frst case, where Countr 1 has 20 ercent more sklled labour than Countr 2 but both countres have the same endowment of raw labour. As Fgure 2b llustrates, ths dfference s unforml dstrbuted over the range of entrereneur s sklls (Countr 1 has a unform denst of entrereneurs equal to 1.2, whle Countr 2 has a unform denst equal to 1). To understand how the model works, we start from the ntal smmetrc nternatonal equlbrum and mlement a 20-ercent ncrease n Countr 1 s endowment of entrereneurs. We frst assume no mgraton and an eogenous skll level z that searates eortng entrereneurs from non-eortng entrereneurs. At gven rces, the resource constrant mles that the ncrease n φ n n nduces a 20-ercent 14

15 A SIMPLE GENERAL EQUILIBRIUM MODEL FIGURE 2a SKILLED LABOUR ASYMMETRY Fgure 2a: Sklled Labor Asmmetr Entrereneur's Wage Sklls Countr 1 Countr 2 Wth Moblt FIGURE 2b SKILLED LABOUR ASYMMETRY Denst Frm Dstrbuton Countr 1 Countr 2 Wth Moblt 15

16 MERCENIER & SCHMITT decrease n the ndvdual frm s sze for each te of good n; hence, both low w 1 hgh w 1 and hgh w 1 fall. Because of the fed eort cost, the downward shft of the curve s more mortant and the ntersecton between the two curves moves to the rght. However, a hgher φ n mles more dverst for the Dt- Stgltz household, both domestc and foregn: substtuton n consumton follows at the eense of the comettvel roduced good: c j j rses. The ndvdual domestc frm s sales ncrease both n the local ( n ) and n the foregn ( n >z ) market so that both w 1 and w 1 now ncrease. However, cometton for raw resources nduces v to rse, whch hurts the eorters rofts relatvel more (through the fed cost) than other roducers so that the low w 1 uward shft of the curve eceeds the other; the new ntersecton between the two curves s moved further to the rght. Endogenzng z thus results n an ncrease n the mnmal talent level requred for eortng. Not surrsngl, gven the ntal ncrease n φ 1n, at the resultng equlbrum an ndvdual entrereneur s real earnng s lower n Countr 1 than n Countr 2, for all skll levels. Allowng for mgraton re-establshes nternatonal smmetr. The relatve abundance of entrereneurs now ncreases the lowest skll level requred for roftable eorts n both countres. Table 1, Eerment 1 summarzes some aggregate results of the two equlbra wth and wthout nternatonal moblt of talents. Wth the ntroducton of nternatonal moblt of entrereneurs, the welfare (of the reresentatve consumer) decreases n Countr 1 and ncreases n Countr 2 b about 1 ercent (Table 1). Ths s due, among other thngs, to the change n roduct varet for consumers n each countr. Wthout moblt, consumers of Countr 1 beneft from a larger offerng of roducts than consumers n Countr 2 snce there are more entrereneurs n Countr 1 and some of the addtonal roducts are non-traded. The moblt of entrereneurs reequlbrates the number of roducts avalable to consumers rresectve of ther locaton. The welfare effect s relatvel small sml because two forces act n ooste drectons. Wth moblt, some non-traded roducts are no longer avalable to consumers of Countr 1, but Countr 2 eorts a larger fracton of the total number of ts dfferentated roducts than t dd wthout nternatonal moblt. Consder now the second case, where the asmmetr s wth resect to the endowment of unsklled (or raw) labour: Countr 1 s endowed wth a 20 ercent larger number of unsklled workers than Countr 2. Agan, we trace the effects of ths asmmetr b usng the ntall smmetrc nternatonal equlbrum wth no mgraton and z fed as a startng ont. Snce the number of entrereneurs s the same n both countres, so s the number of frms n the Hgh Tech sector low hgh 16

17 A SIMPLE GENERAL EQUILIBRIUM MODEL TABLE 1 THE EFFECTS OF FOUR SIMPLE ASYMMETRIES COUNTRY 1 COUNTRY 2 NO MOBILITY WITH MOBILITY NO MOBILITY WITH MOBILITY EXPERIMENT 1: +20% ENTREPRENEURS IN COUNTRY 1 Welfare = v c c Po PoE EXPERIMENT 2: + 20% UNSKILLED LABOUR IN COUNTRY 1 Welfare = v c c Po PoE EXPERIMENT 3: +20% EXPORT FIXED COST IN COUNTRY 1 Welfare = v c c Po PoE n the absence of nternatonal moblt of entrereneurs. However, more abundant and, thus, cheaer raw labour mles that the scale of these frms s larger n Countr 1 (comare, for nstance, v between the two countres wthout 17

18 MERCENIER & SCHMITT moblt; see Table 1, Eerment 2). Though rofts of both domestc eorters and non-eorters are boosted, the mact on the former s larger because of the fed eort cost. The ntersecton between the two wage rofles moves to the left, ndcatng that less-talented entrereneurs are now able to undertake roftable eort actvtes. Not surrsngl, entrereneurs real wages are hgher n Countr 1 for all sklls (Fgure 3a). Costless mgraton nduces entrereneurs to move to Countr 1 n order to take advantage of cheaer resources and a larger market. Interestngl, there s a strong comoston effect n ths mgraton snce those who are massvel movng to Countr 1 are the non-trader entrereneurs. As Fgure 3 ndcates, ths mgraton of non-trader entrereneurs ncreases wth skll level n Countr 1. Ths s sml due to the fact that the ncentve to mgrate ncreases wth the skll level snce t s drectl related to the dfference n real earnngs between the two countres. However, the same s not true for tradng frms as there s a two-wa flow of medum/hgh sklls: some ntermedate sklls are attracted nto the home countr, whle there s an outflow of hgh talents. Ths s not surrsng: because z 1 <z 2 at the no-mgraton equlbrum, movng nto Countr 1 makes t ossble for some foregn ntermedate-skll entrereneurs to become eorters and ncrease ther rofts, everthng else beng equal. Ths motvaton to mgrate s obvousl not shared b the most talented who are eorters ndeendentl of ther geograhc locaton. More mortantl, the balance-of-trade equlbrum requres that f there s a net nflow of ntermedate-skll eorters, there must be due to terms-of-trade changes a net outflow of (a smaller number of) hgh-skll eorters. There s also a second effect. The large nflow of entrereneurs n Countr 1 rases the cost of roducng Hgh Tech goods domestcall. Indeed, v ncreases b 3.6 ercent whle remanng unchanged abroad (see Table 1, Eerment 2). Snce the fed cost of eortng s eressed n terms of raw labour, eortng from Countr 1 would be more costl, everthng else beng equal, than t was wthout moblt. However, there are now more eorters n Countr 1 than before, whch makes enetraton of foregn markets less costl from the home countr, thanks to the eort eternalt. 15 Hence, the fed cost of eortng ma actuall be lower n Countr 1 deste the dfference n. Obvousl, ths effect s artcularl v crucal for the margnal traders snce the eort cost s most sgnfcant n the decsons of these frms. Here too, ntermedate-skll entrereneurs have a secfc ncentve to move nto Countr 1 that the most talented ones do not share: for the latter, access s ensured to all markets, domestc and foregn. And agan, the balance-of-trade equlbrum requres that f there s a net nflow of ntermedate-skll eorters, there must be a net outflow of hgh-skll eorters. 18

19 A SIMPLE GENERAL EQUILIBRIUM MODEL FIGURE 3a UNSKILLED LABOUR ASYMMETRY Entrereneur's Wage Sklls Countr 1 Countr 2 Wth Moblt FIGURE 3b UNSKILLED LABOUR ASYMMETRY Denst Frm Dstrbuton Countr 1 and 2 (no moblt) Countr 1 Countr 2 19

20 MERCENIER & SCHMITT In ths case, allowng for entrereneur moblt ncreases welfare b about 2 ercent n Countr 1 and decreases t b 2.1 ercent n Countr 2 (Table 1). Ths sgnfcant welfare effect s not surrsng gven the mrovement n the terms of trade of Countr 1 and the substantal net ncrease n the number of dfferentated goods avalable to ts consumers (esecall non-traded goods). We see from the two revous eerments that the sze of the relatve sklled- versus unsklled-labour endowments matters a lot n ths model. Deendng on where the asmmetr les, the ntroducton of nternatonal entrereneur moblt has ver dfferent qualtatve and quanttatve effects. When the endowment of sklled entrereneurs n Countr 1 s greater than n Countr 2, entrereneurs move to the latter, whereas when the unsklled labour endowment s larger n Countr 1, the domnant flow s that of entrereneurs movng to Countr 1. These asmmetres also result n dfferent roducton secalzaton and tradng atterns. In the frst two eerments, more Hgh Tech roducts (traded and non-traded) are roduced n Countr 1 wthout labour moblt than wth moblt. Ths means that fewer resources are beng used n the Hgh Tech sector, and Countr 1 eorts relatvel less Hgh Tech roducts wth moblt than wthout moblt. In the second asmmetr, the ooste occurs. Wth moblt, there s a relatve secalzaton n roducton toward the Hgh Tech sector at the eense of the Low Tech sector n Countr 1 but, qute nterestngl, t leads to relatve secalzaton along the qualt dmenson wthn the Hgh Tech sector. Indeed, moblt leads Countr 1 to secalze manl n low and ntermedate qualt varants, whle Countr 2 tends to secalze n hgh qualt eort roducts. The quanttatve effects are also qute dfferent. In the frst eerment (+20 ercent sklled workers n Countr 1), the flow of sklled entrereneurs movng from Countr 1 to Countr 2 s roughl equal to half the dfference n endowment (reresentng 10 ercent of the ntal number of entrereneurs n Countr 2). In the second case (+20 ercent raw labour n Countr 1), the total number of sklled entrereneurs mgratng between the two countres reresents about 34 ercent of the ntal endowment of entrereneurs n each countr (for a net change equal to 27 ercent of the ntal endowment). Ths s qute a sgnfcant effect. Consder now the thrd case, where the γ set-u cost of eortng Hgh Tech roducts from Countr 1 s 20 ercent hgher. Non-tradng frms are onl ndrectl affected b the cost of the rmar factor. As s to be eected, n the absence of nternatonal labour moblt, the entrereneurs wages are ver smlar n the two countres (Fgure 4a). Gven such small dfferences n terms of oeratng rofts, ntroducng nternatonal moblt of entrereneurs cannot change the entrereneur s wage much. Nonetheless, the effect on the mgraton of entrereneurs s sgnfcant, esecall for those workng n tradng frms, 20

21 A SIMPLE GENERAL EQUILIBRIUM MODEL FIGURE 4a EXPORT COST ASYMMETRY 0.3 Entrereneur's Wage Sklls Countr 1 Countr 2 Wth Moblt FIGURE 4b EXPORT COST ASYMMETRY Denst Frm Dstrbuton Countr 1 and 2 (no moblt) Countr 1 Countr 2 21

22 MERCENIER & SCHMITT where some entrereneurs mgrate from Countr 1 to Countr 2, whle others mgrate from Countr 2 to Countr 1 (Fgure 4b). The hgher fed cost of eortng leads to lower raw labour cost n Countr 1 wth or wthout labour moblt. Indeed, moblt reduces the rce of labour even further, whch nduces a decrease n the fed cost of eortng n Countr 1. At the equlbrum wth moblt, Countr 1 ends u wth more traders than Countr 2 deste the 20 ercent hgher fed cost of eortng. Hence, at equlbrum, gven the number of eorters and the rce of labour n Countr 1, the fed cost s lower n that countr than n Countr 2. It s aarent that, even f the two countres are smlar ecet for the fed cost of eortng, man tradng entrereneurs wll mgrate across the border. Moblt brngs greater secalzaton along the qualt dmenson among Hgh Tech roducts: Countr 1 secalzes relatvel n low qualt traded roducts, whle Countr 2 secalzes relatvel n hgh qualt traded roducts. Indeed, at each end of the qualt range of traded roducts, there s almost comlete secalzaton n each countr. At the ver hgh end, nearl all tradng frms roduce n Countr 2 and, at the other end (but stll among the tradng frms), nearl all tradng frms roduce n Countr 1. Deste the mgraton of a sgnfcant fracton of entrereneurs, welfare onl margnall changes n each countr (wth a slght decrease n Countr 1 and a slght ncrease n Countr 2; see Table 1). In the last eerment, factors are more roductve n Countr 1 than n Countr 2 n the Low Tech sector: +20 ercent total factor roductvt n Countr 1 s Low Tech sector. In the absence of nternatonal labour moblt, resources wll be heavl used n the Low Tech sector as the rce of the Low Tech roduct s low. Because the consumton functon takes a Cobb-Douglas form over the Low Tech and Hgh Tech roducts and because the Low Tech roduct s a non-traded good, the effect of ths hgher roductvt shock s confned to the Low Tech sector. Movements of entrereneurs stll take lace when moblt s allowed snce the real value of the quas rents n the Hgh Tech sector s now hgher n Countr 1. Mgraton follows from Countr 2 to the home countr, wth effects on non-tradng as well as on tradng frms. The number of tradng frms ncreases n Countr 1, wth a smlar eternalt effect and, through t, smlar roducton secalzaton (at least for tradng frms) and trade attern effects, as n revous smulatons. It should be noted that, n ths case, moblt affects manl tradng frms because the dfference n real quasrents between the two countres ncreases wth skll level. Hence, there s a much stronger ncentve to move for tradng frms than for non-tradng frms. Of course, the balance-of-trade condton revents that these entrereneurs move n one drecton onl. 22

23 A SIMPLE GENERAL EQUILIBRIUM MODEL FIGURE 5a TECHNOLOGICAL ASYMMETRY 0.35 Entrereneur's Wage Sklls Countr 1 Countr 2 Wth Moblt FIGURE 5b TECHNOLOGICAL ASYMMETRY Denst Frm Dstrbuton Countr 1 and 2 (no moblt) Countr 1 Countr 2 23

24 MERCENIER & SCHMITT CONCLUSIONS W HAT TO MAKE OF THESE SIMULATIONS? Of course, we have used artfcal data so that none of these smulatons drectl al to the Canada- U.S. case. We have also comletel dsregarded how we mght make the model oeratonal wth real data. However, the artcular model used here rases a number of ssues about the meanng of sklls, the corresondence between sklls and frms, not to menton the corresondence between sklls and roduct qualt. Though secfc, we beleve that ths te of model has great otental for a number of reasons. Frst, there s a lnk between trade, earnng nequalt and labour moblt. The lnk between trade and earnng nequalt comes from two sources. Snce trade affects the wages of unsklled and sklled workers (entrereneurs), and wages among entrereneurs, trade-nduced ncreases n nequalt ma mean here that the wages of sklled workers ncrease wth resect to the wages of unsklled workers, and t can mean that the wages of hghl sklled workers (n tradng frms) ncrease wth resect to the wages of less hghl sklled workers (n non-tradng frms). Snce these nequaltes occur not onl wthn a countr but across countres, the create ncentves to mgrate across the border n order to take advantage of earnng dfferentals. Second, sklls matter n ths model and the returns to sklls are ostve, gvng the model a Rosen suerstar flavour. Thrd, because ndvduals (at least sklled workers) are dfferentated, mgraton decsons are otentall dfferent among ndvduals. Introducng nternatonal labour moblt leads to changes n secalzaton at the roducton level across sectors and wthn sectors roducng dfferentated goods. In turn, ths leads to changes n trade atterns. Ths s an nterestng result because the changes n trade atterns across and wthn sectors are not due to trade lberalzaton er se but are b-roducts of lberalzng nternatonal labour mgraton. In order to llustrate wh ths last ont ma make ths model relevant for the Canada-U.S. case, consder the followng. In a stud commssoned b the EU Commsson, Fontagné, Freudenberg and Pérd (1998) have uncovered nterestng changes n trade atterns wthn the Euroean Unon between 1980 and The frst dvded trade between ever ar of EU member countres nto nter- and ntra-ndustr trade, and further dvded ntra-ndustr trade nto horzontal (Helman-Krugman te) and vertcal (Shaked and Suttonqualt te) trade. To ths end, the sml comared the rce of eort wth the rce of mort. If, at the level of the varants, the rce of eort s roughl smlar to the rce of mort, then trade n ths varant belongs to the horzontal ntra-ndustr trade categor. If there s a sgnfcant dfference (ostve or negatve) between the two rces, then trade n ths varant belongs to the vertcal ntra-ndustr trade categor. Aggregatng these trade shares, the 24

25 A SIMPLE GENERAL EQUILIBRIUM MODEL dscovered that f, as eected, the share of overall ntra-ndustr n total trade ncreased over the erod, t s not due to an ncrease n the share of horzontal ntra-ndustr trade but to an ncrease n the share of vertcal ntra-ndustr trade. In other words, based on ther methodolog, some knd of secalzaton has taken lace n Euroe at the countr level deste the resence of smlar countres (sa France and German) and of trade lberalzaton. Nobod has et offered a good elanaton for ths henomenon. Alng a smlar methodolog to the Canada-U.S. case over the erod, Andresen, Harrs and Schmtt (2001) fnd smlar changes but on a much smaller scale than n Euroe. Wh? We do not reall have a good answer ether. However, our smulatons suggest that ths ma have a ver smle elanaton: the dfference between the results obtaned for Euroe and for North Amerca mght sml be assocated wth dfferences n the degree of economc ntegraton. The 1992 Unque Market (decded well before 1992) ma well have trggered locaton decsons among frms (for nstance) focused at servng Euroe as a whole rather than at servng onl a secfc Euroean countr, whereas NAFTA dd not have the same mact on frms locaton decsons (or on ndvduals locaton decsons) because of ts more lmted focus. 16 Ths suggests that, n evaluatng the effects of a deeer ntegraton between Canada and the Unted States and, n artcular, measures of ntegraton dealng wth nternatonal labour moblt, an mortant asect would be the consequences of these measures on nternatonal trade and on the atterns of trade. The above advantages are urel statc. Needless to sa, n a dnamc envronment, other elements could be added to a model of ths te, for eamle human catal formaton, the mact of an agng oulaton, or qualtladder/endogenous growth comonents. ENDNOTES 1 For nstance, German s adotng a Canadan-ont-te sstem to evaluate mmgrants. A blateral agreement between Swtzerland and the Euroean Unon s now n force snce June 1, After a fve-ear transton erod, t s suosed to free u labour moblt between Swtzerland and the Euroean Unon. The man reason for these olces seems to be lnked to labour shortages n secfc hghl sklled actvtes. 2 See Markusen (1988) for a model belongng to ths categor and secfcall addressng ostve and normatve asects of human catal formaton, sklled labour and the bran dran. 25