DOCUMENT DE TREBALL XREAP

Transcription

1 DOCUMENT DE TREBALL XREAP IMMIGRATION AND FIRM GROWTH: EVIDENCE FROM SPANISH CITIES Mercedes Teruel-Carrizosa (GRIT) Agusí Segarra-Blasco (GRIT)

2 Immigraion and Firm Growh: evidence from Spanish ciies Mercedes Teruel-Carrizosa ( ) Agusi Segarra-Blasco ( ) ABSTRACT This aricle analyses he effec of immigraion flows on he growh and efficiency of manufacuring firms in Spanish ciies. To dae, mos sudies have ended o focus on he effec immigrans have on labour markes a an aggregae level. Here, however, we underake an exhausive analysis a he firm level and repor conclusive empirical findings. Ten years ago, Spain began o regiser massive immigraion flows, concenraed above all on is mos dynamic and advanced regions. Here, herefore, raher han focusing on he impac his has had on Spain s labour marke (changes o he skill srucure of he workforce, increase in labour supply, he displacemen of naive workers, ec.), we examine he arrival of immigrans in erms of he changes his has mean o he srucure of he counry s ciies and heir ameniies. Thus, we argue ha he impac of immigraion on firm performance should no only be considered in erms of he labour marke, bu also in erms of how a ciy s ameniies can affec he performance of firms. Employing a panel daa mehodology, we show ha he increasing pressure brough o bear by immigrans has a posiive effec on he evoluion of labour produciviy and wages and a negaive effec on he job evoluion of hese manufacuring firms. In addiion, boh small and new firms are more sensiive o he pressures of such immigran inflows, while foreign marke oriened firms repor higher produciviy levels and a less marked impac of immigraion han heir counerpars. In his paper, we also presen a se of insrumens o correc he endogeneiy bias, which confirms he effec of local immigraion flows on he performance of manufacuring firms. JEL classificaion: L25, R12 Key words: firm growh, firm locaion, regional effecs ( ) agusi.segarra@urv.ca, mercedes.eruel@urv.ca Indusry and erriory research group Deparmen of Economics (Rovira i Virgili Universiy) Av. Universia, 1; Reus Tel Fax We are graeful o Verònica Gombau-Beromeu for her research suppor and we would like o acknowledge helpful commens from aendances o he "Sympoosium on Business Dynamics and Innovaion: he role of space" in Barcelona. 1

3 1. Inroducion Massive immigraion flows increase he supply of workers and aler he skill composiion of local labour markes. A large body of economic lieraure has ried o deermine he impac of immigraion on he equilibrium of he labour marke and, in paricular, is effecs on wages and employmen in hos counries (Borjas, 1989, 1994; Card, 2001). In general, hese empirical sudies conclude ha immigrans are complemenary o naive workers and he increase in he available workforce has lile effec on wages and income. Usually his lieraure adops a labour marke aggregae perspecive, bu inernaional immigraion flows are unequally disribued among counries and ciies. When he labour marke approach adops a spaial dimension, he correlaion beween he proporion of immigrans in a populaion and wages is low and someimes virually non-exisen. How do we accoun for wha is happening here? Immigraion is a complex phenomenon and commonly held beliefs are ofen incorrec. This paper analyses he consequences of immigran inflows ino manufacuring firms in Spanish ciies. In paricular, we are ineresed in deermining he effecs of immigraion on jobs, wages and produciviy in such firms. Spain is an excepional case in inernaional immigraion, since he foreign-born populaion has increased considerably over he las decade. However, he disribuion of immigrans is no homogeneous; here is a concenraion in highly dynamic regions. These regions experience a significan ransformaion in he social dimension religions, languages, culures- which affecs he local economy in erms of he diversiy of skills and capabiliies among is workforce. Today, many Spanish ciies have a high proporion of immigrans and have become veriable meling pos of peoples and culures. The increasing role of inernaional migraion in developed counries has given rise o a body of wide-ranging and ineresing heoreical and empirical research (Dusman and Glilz, 2005). Economic lieraure has focused mainly on he analysis of he economic impac of immigraion on he labour marke, emphasizing he wage differenial and is reducion over ime (Borjas, 1989, 2006; Oaviano and Peri, 2006). A furher wave of lieraure analyses he human capial effec of inernaional immigraion, in paricular examining is effec on he counries of origin, and porraying i as an engine ha drives he leading economies (Torben, 2007). Finally, a hird group of research focuses on immigraion policy in hos counries (Winer and Zweimüller, 1996; Lundborg and Segersromc, 2002). Previous empirical sudies have suggesed ha immigran populaions generally arrive wih few skills and considerable economic disadvanages, bu heir economic opporuniies improve rapidly over ime. And, wihin a decade or wo following heir arrival, immigran earnings converge wih hose of naive workers. There is also lile empirical evidence suggesing ha immigrans have an adverse impac on naive employmen 2

4 opporuniies. In general, immigrans complemen naive workers (Borjas, 1994, Card, 2001). In spie of he fac ha skilled workers are relaively scarce in poor counries, he laer send skilled as well as unskilled workers. As Jones (1998) remarks Why, hen, does no skilled labour migrae from he Unied Saes o Zaire? As we will observe in he case of Spain, lowskilled workers predominae among he immigran populaion, while only a fracion of new immigrans are skilled workers wih prior experience in manufacuring indusries in heir counries of origin. In general, he empirical lieraure examining he impac of immigrans on labour marke and economic growh adops an aggregae perspecive. Bu if we observe he locaional paerns of immigrans we find ha new immigrans are concenraed in municipaliies wih more job opporuniies and more highly developed neworks of immigrans. When his fac is included in he analysis, he need o analyse firms responses o immigraion flows becomes apparen. For his reason, he main purpose of his aricle is o deal wih he impac of immigraion on firm growh a he municipal and firm level. We focus on he effec ha he share of immigrans locaed in each municipaliy has on a firm s performance. In he economeric specificaion he local share of immigran populaion is a dependen variable given ha migraion is a mobile facor ha seeks ou job opporuniies across ciies. Firm performance in erms of increased labour produciviy, employmen and wages play he role of he independen variable. How do firms respond o immigraion in erms of employmen, wages and produciviy? Are here any differences ha can be aribued o he age or size of a firm? Does exernal aciviy condiion he response of he firm? This aricle aims o idenify he deerminans of firm growh and he effec of immigran populaions on such growh. Our daabase compiles informaion on Spanish manufacuring firms locaed in ciies wih more han 1000 inhabians during he period , while our resuls emphasise he response of firms o immigraion from hree differen dimensions (employmen, wages and produciviy). There are hree main conribuing facors o consider. Firs, we analyse he effec of a firm s response o immigraion flows. The arrival of new coningens has increased he labour force and firms may have modified heir incenives o inves. Second, we consider he fac ha he arrival of new immigrans has also pu pressure on renal prices and has changed he appearance of municipaliies. Third, our analysis adops a municipal- and firm-based approach, in conras wih previous sudies on immigraion ha end o adop a macroeconomic or individual approach. The res of he paper is organized as follows: in he second secion we review recen lieraure examining he impac of immigraion flows on he labour marke and growh. In he hird secion we look a a simple derivaion of Roback s (1982) model ha inerpres local differences in wages and rens in 3

5 relaion o he pressure of immigrans on he labour marke and local facors. This secion presens he heoreical foundaions for our analysis of he effec of immigraion on firm performance. In he fourh secion, we se ou he characerisics of wo daa sources relaed o immigraion flows in Spanish ciies and Spanish manufacuring firms. The fifh secion presens various economeric mehodologies and our main hypohesis supporing he subsequen empirical analysis. In he sixh secion, we repor he main empirical resuls obained and, in he las secion we presen our main conclusions. 2. Immigraion and is effecs on firm performance The earh is no fla, bu barriers o mobiliy have diminished and inernaional immigraion has appeared as one of he main characerisics of he 21 s cenury. To a cerain exen, he world is becoming an open space where differences in lifesyle and income inequaliies persis beween ehnic groups, regions and counries. As a consequence, flows of inernaional immigraion are common nowadays. Afer he immigraion from Europe o he American coninen during he 18 h and 19 h cenuries, new immigraion flows began arriving in he Unied Saes and, more recenly, in Germany, France, he Unied Kingdom and oher European counries. Alhough flows of immigrans are a hisorical phenomenon, he inensiy of flows increased in he las decades of he 20 h cenury. Spain became a hos counry laer han mos, bu wih he same migraion inensiy of curren flows. Unil he sevenies, Spain was primarily a counry of origin and recorded flows of inernal migraion beween regions. However, in he las decade of he 20 h cenury, Spain became one of he main hos counries for inernaional immigraion. Because of his unusual circumsance, he inflow migraion paern in Spain oday is clearly an ineresing case o observe and sudy. The daa ha are available poin o he inensiy of he arrival of immigrans in Spain in recen years. In he las years of he 20 h cenury he foreign-born populaion in Spain was small compared wih ha of oher developed counries. The fracion of immigrans was jus 1.6 percen of he Spanish populaion in 1998, bu he immigran populaion increased o 7.0 percen in 2004 and rose o 11.3 percen in In a shor period, he foreign populaion in Spain has increased significanly. There were 637,085 immigrans in 1998 and his number increased o 5,220,577 in 2008 (see yearly evoluion in Table A-1). The recen wave of inernaional immigraion in Spain is inense and has had a significan effec on sociey and he economy. 1 The Spanish Saisics Insiue considers an immigran o be a foreign-born person who has lived in Spain for one year or more and who inends o live in Spain for a leas one year. I does no include minors of 16 years of age or younger, foreigners ha were born in Spain, or Spanish ciizens who were no born in Spain. 4

6 Wih his migraory phenomenon came economic effecs predominanly relaed o job opporuniies in he hos counry or he lack of hem in he counry of origin. Over he las few decades here has been a profusion of lieraure on he deerminans of inernaional flows. On he one hand, heoreical models have ackled immigraion as a "push" or "pull" phenomenon. In oher words, hey emphasise he facors ha explain migraion and hose ha deermine he araciveness of a counry or a region. On he oher hand, he empirical lieraure reveals several ineresing empirical facs (García and López, 2005). For one, here is a posiive relaionship beween immigraion and he differenial gap of income per capia beween he hos counry and he counry of origin; in general, economic growh in he hos counry has a posiive effec on migraory flows. The lower he fixed coss of migraion (policies of family regrouping, regularizaion of illegal immigrans, labour conracs a origin), he more inense he flows are. The exisence of previous immigrans in he hos counry diminishes he cos of immigraion and faciliaes neworks of recepion and solidariy. Finally, underground economies faciliae he arrival of illegal immigrans 2. Borjas (1999) also poins ou ha here are "welfare magnes" explaining he increase in immigran inflows in some regions. Apar from he differences beween counries, he lieraure has widely sudied he impac of immigrans on he labour marke of he hos counries. The empirical resuls can be summarized as follows: flows of immigrans have a slighly negaive impac on wages in he hos counry in he shor erm (Borjas, 2003) 3, bu naive workers have he same labour opporuniies because immigrans do jobs ha naives do no wan o do (Borjas, 1994). Naive workers also obain lower salaries in labour markes wih a high percenage of immigrans. However, in hose markes immigrans enjoy fewer labour opporuniies han naive workers (Borjas, 2006). In general, immigrans are less skilled han naives and heir salaries are lower, bu over ime heir salaries grow faser han hose of naive workers (Borjas, 1994). And finally, afer one or wo decades immigrans aain he same wage level as naive workers wih he same skills. In an aemp o decipher he effec of immigrans in hos counries, Peri and Sparber (2008) analyse he effecs of immigraion on he wages of 2 For 17 counries of he OECD beween 1980 and 2000, García and López (2005) show ha he main variables relaed o he "pull effec" of immigrans are income per capia (moneary income, no acual), growh rae and he exisence of neworks of immigrans in he hos counry. 3 During he period in he Unied Saes, empirical evidence suggess ha an increase in immigraion of 10 percen had he following effecs (Borjas, 2006): i reduced he wages of naive workers in ha same skill group by 3.5 percen; i reduced he wages of naive workers who had he same educaion bu who differed in heir experience by 0.7 percen; and i increased he wages of naive workers wih differen educaional aainmen by 0.5 percen. In he shor run, immigraion reduces he earnings of naive workers by 3.3%; bu in he long run, i increases hem by 0.1%. 5

7 naive workers wih low educaion and job raining levels. They found wo criical facors. The firs is immigrans aking jobs similar o hose of naive workers or aking differen jobs due o inheren comparaive advanages beween naive and foreign-born workers in performing paricular asks. The second is wheher naive workers respond o immigraion and adjus heir occupaional choices o shield hemselves from compeiion wih immigran labour. Immigrans wih lile educaion or job raining have a comparaive advanage in manual and physical asks, while naives of similar levels of educaion have a comparaive advanage in communicaion and language-inensive jobs. When immigraion generaes large increases in manual labour supply, he relaive compensaion paid for communicaion skills rises, hereby rewarding naives who progressively move owards language-inensive jobs. In general, flows of immigrans are a complemen o some groups of naive workers and produce an increase in low-skilled workers (Quispe and Zavodny, 2002), a moderaing effec on wage growh (Borjas 2003, 2006), an increasing asymmery in he levels of produciviy and efficiency across firms and across jobs and a posiive impac on economic growh. Thus, hese sudies adop an aggregae perspecive from he labour marke and do no approach he diversiy of he realiy in erms of local labour markes. In fac, he erriorial dimension has been negleced by several differen fields sudying indusrial organisaion and only in recen decades has here been any ineres in analysing is effecs. On his noe, he majoriy of sudies analysing firm performance have focused almos exclusively on firm variables (Foopoulos and Louri, 2000). This lack of ineres in erriory is more pronounced in he analysis of firm performance (Hoogsra and van Dijk, 2004). However, he locaion of a firm influences is behaviour and, as a consequence, influences pos-enry firm performance and firm survival. Alhough ineres in he effec of erriorial variables on firm performance has increased over he las decade, here are sill crucial aspecs which need furher sudy such as he effec of immigraion on firm performance. The impac of he increasing number of foreign-born workers on he labour marke is inense and affecs firm performance. The rapid economic growh experienced in recen decades is due o an increase in he number of workers. However, produciviy has remained unchanged. One canno help bu wonder wheher his increase in he number of workers in he labour marke has affeced firms decisions regarding growh via produciviy versus he employmen of more workers. To summarise, he lieraure analysing he effec of immigrans on firm performance is scarce and usually analyses he impac on economic growh a he counrywide level and, usually does no consider he subsiuion effec of capial inensiy and he labour marke. Thus, here is a gap which should be filled, given he social consequences and he economic impac of immigraion on our economies. 6

8 3. Theoreical argumens Recen empirical papers have observed he economic implicaions of immigraion a he meropolian, regional and local levels (Grossman, 1982; Borjas, 1987). Some papers repor a negaive correlaion beween he rae of immigraion and wages across ciies. These resuls show ha naive wages are lower in markes wih many immigrans and sugges ha immigrans reduce he opporuniies of naive workers in he labour marke. However, recen research has found no correlaion beween immigraion and wage levels in ciies, evidence ha immigraion has no effec on local labour markes. George Borjas (2006) offers wo inerpreaions of such resuls. On he one hand, immigrans may no be randomly disribued across labour markes. For example, in a case such as ha of Spain, if immigrans end o cluser in ciies wih hriving economies and high wages we obain a posiive correlaion beween immigraion and wages. Thus, we expec a posiive correlaion in he level of wages bu a negaive correlaion wih wage growh raes in local labour markes. On he oher hand, he pressure of immigrans on local labour markes may produce a displacemen effec for naive workers, especially in groups of naives ha compee wih immigrans for he same jobs. However, he effecs of immigraion should no be observed only from a macroeconomic perspecive bu also from he microeconomic dimension, i.e., a he business level, so as o ake ino accoun he differen pressures caused by he influx of immigraion in ciies. Analyses of he effecs of immigraion acquire much more ineres when a erriorial, raher han a macroeconomic, approach is adoped. In order o analyse he effecs of immigraion on hos ciies we can urn o Roback s (1982) model, which explains local differences in wages and rens as a resul of differences in ameniies. The formalisaion of he idea ha local differences in wages and rens compensae people and businesses for differences in desirable local ameniies is aribuable o Jennifer Roback (1982) and he seminal work by Sherwin Rosen (1974). The sudies of Roback (1982) and Rosen (1979) were pioneering in ha hey offered a framework in which o quanify hedonic prices for cerain urban characerisics. Roback s model offers a hedonic analysis of ciies and several aemps o develop qualiy-of-life indexes for ciies or meropolian areas. Of paricular noe for our purposes is he analysis of he effecs of immigraion on wages and housing prices in hos ciies. Today a large body of urban lieraure highlighs he imporance of ameniies in reaining workers and businesses in ciies (Florida, 2000a, 2000b, 2000c; Gyourko e al, 1999). In general, hese sudies sugges ha ameniies have no ye been fully capialized ino wages or rens. Here we would like o provide a simple heoreical framework for immigraion flows, wihin which rens, wages and ciy populaion can be inerpreed as implying ha immigraion o larger 7

9 ciies is primarily driven by producive ameniies (Krupka and Donaldson, 2007) 4. Suppose we have an economy wih a large number of ciies (M) where businesses and workers incur no coss for relocaion o oher ciies. Every worker may choose o look for work in he ciy wih he characerisics ha will provide he highes level of saisfacion for hem, given a uiliy funcion which feaures he characerisics of he ciy in which hey live and work. The vecor of characerisics only varies in erms of he level of ameniies (s) in a coninuous manner from s s (1..n). The residens of each ciy consume and produce a composie commodiy, X, whose price is fixed by global markes and will be aken as numeraire. Roback s (1982) framework presens a simple general equilibrium model in which boh capial and labour are assumed o be compleely mobile across ciies. In his conex he cos of changing residences or firm locaion are zero, bu inerciy commuing coss are high and we assume ha workers find jobs in heir ciies. In a sae of equilibrium in erms of he disribuion of workers among he ciies and firms, wage and ren differences can be characerized as funcions of he ameniies in he ciy (si ). The problem for he workers is, given a level of si in heir ciy, o choose quaniies of x he composie commodiy consumed, and l c he residenial land consumed- o saisfy a budge consrain, Max U(x, l c ; si) subjec o w + I = x + l c r Where w is an individual wage, I is a non labour income, and r is a house ren. From his equaion we can easily derive he indirec uiliy funcion, V. In condiions of marke equilibrium, he uiliy funcion for workers is given by he following formula, V (w, r, si ) = α Where α is a consan and wages and rens mus adjus in such a way ha he uiliy funcion of he workers is he same in all M ciies. The implicaion of his adjusmen is ha he workers have no incenive o move from one ciy o he nex. This in urn implies, for example, a higher level of he s 1 characerisics of he ciy, bu also lower wages so ha he level of uiliy remains he same. The model also assumes ha parial derivaives of he uiliy funcion, as he laer relaes o wages and ciy ameniies, are posiive, and ha he parial derivaives relaed o ren are negaive. So, Vw 0, Vr 0, Vsi 0 4 The lieraure classifies ameniies in hree caegories: producive, non-producive and unproducive. Producive ameniies increase uiliy and business profis; non-producive ameniies increase uiliy bu do no affec business profis, and unproducive ameniies increase uiliy bu reduce business profis. 8

10 Firms locaed in a paricular ciy produce commodiies according o a producion funcion wih consan reurns o scale, X = f(l p, N; si), where l p is he land used in producion, N is he oal number of he workers in he ciy and si is he local ameniies. The problem ha each firm locaed in a ciy M faces is o minimize coss subjec o he producion funcion. Under hese condiions, he businesses locaed in he various ciies are limied o adjusing real salaries o comply wih his condiion. So, C = C (w, r, si ) = 1, where Cw 0 and Cr 0 As usual he uni cos funcion increases in boh facor prices, bu he effec of local ameniies is more ambiguous. Ameniies can eiher be producive (cos reducing) or unproducive (cos enhancing), and when he naure and he dimension of local ameniies change, he effec on producion coss is indeerminae in sign. In a sae of equilibrium, workers and firms have no incenive o relocae o anoher ciy. A spaial equilibrium means ha he workers canno increase heir uiliy and businesses canno reduce coss by relocaing. Equilibrium of real salaries is obained hrough an equaliy of uiliy among he workers and an equal per uni cos of producion in firms across all ciies. The real salary of he workers is deermined by he ineracion of he condiions for equilibrium in such a way ha he economic effec of he differen levels of si in M ciies is seen in he differences in real salary beween he ciies. In equilibrium, we have: C (w, r, si ) = 1 and U 0 = V 0 = V (w, r, si ) (6) Noice ha real wages and rens are deermined by he ineracion of he equilibrium condiions of he wo sides of he marke land and labour markes, and wages and rens can be solved as funcions of he ciy ameniies, given a level of α. The resul of his srucure is he fac ha in he ciies where here is a higher level of he ciy variable si real wages are lower, and in ciies in which he variable is lower, real wages are higher. In his model he facor prices offer a balance beween he locaional preferences of he firms and hose of he workers. Firms prefer locaions wih low ren levels and low wages, while workers may accep lower real wages in ciies wih high rens as long as he ciy offers more ameniies ha serve o increase heir welfare. According o he above model, he impac of massive immigran inflows on local facor prices, labour marke and ameniies may vary. These siuaions are a paricular focus of he empirical work ha we underake below. In essence, massive immigraion affecs local prices and he evoluion of he firm in hree main ways, a) Immigraion can generae producive ameniies ha increase uiliy and business profis. Here, we expec a pressure on rens and a posiive impac on real wages, which in erms of he dynamics of firm evoluion, resuls in an increase in employmen in he locaion, an increase in he 9

11 level of heerogeneiy of efficiency beween firms and an ambiguous effec on labour produciviy. In his case, indusries wih a large share of unskilled workers regisered a decrease in produciviy levels and generaed he majoriy of new local jobs, while firms ha operaed in indusries wih high skills and high echnological levels regisered an increase in labour produciviy, a moderaion in employmen growh and an increase in wages. b) Immigraion can generae non-producive ameniies in he ciy ha increase uiliy bu which do no affec business profis. Here, we expec a rise in housing rens and a negaive effec on wages ha lead o an increase in employmen and which have a negaive impac on labour produciviy. c) Finally, if immigraion generaes non-producive ameniies in he ciy ha increase uiliy bu which cause business profis o fall, hen we can expec an ambiguous effec on rens and a reducion in real wages ha lead o a decrease in he number of local workers and which have a negaive impac on labour produciviy. If a ciy receives massive immigraion flows, hen his hos ciy will be affeced in several ways. On he one hand, immigraion flows lead o an increase in oal populaion, affec local facor prices relaed o housing renals and produce an increase in he local labour supply. Rising immigran populaions and producive ameniies ha posiively affec wages increase housing demand, provide incenives for he real esae secor and increase rens (Glaeser and Gyourko, 2005). On he oher hand, immigraion flows produce ameniies and disameniies in he ciy. The former are relaed o he increasing cos of housing and changes in he composiion of he populaion. The laer are relaed o linguisic and culural diversiy which posiively affecs he creaiviy and iniiaive of he naive populaion. The ne effec of immigraion may be ambiguous; however, he recen lieraure sresses ha he posiive aspecs ouweigh he negaive (Oaviano and Peri, 2005). Inense pressure from immigran workers on a local labour marke may displace naive workers wih more experience, and force hem o move o new areas. The presence of significanly large groups of immigrans will, in he shor run, lead o salary inequaliies and less pressure o acquire more efficien echnologies. This resuls in a reducion in he capial o work raio and in produciviy in areas ha demand workers wih fewer qualificaions. Massive immigraion flows in ciies end o displace naive workers o indusries ha need workers wih more skills and iner-relaional capabiliies, and o displace naive workers o manufacuring indusries. Furhermore, echnological changes in manufacuring firms end o eliminae unskilled or semi-skilled jobs, while new jobs being creaed require he workers o have echnological experience (Lewis, 2005). In addiion, he migraion of skilled workers is a significan way of generaing 10

12 inerregional spillovers and faciliaing regional learning effecs (Faggian and McCann, 2006). 4. Daa 4.1. The daabase The daabase we use in our analysis is he Spanish Mercanile Regisry for he years 2001 o The daa include all manufacuring firms ha are required o declare he number of employees on heir books. Wha is crucial o our analysis is ha we are able o idenify where he firms are locaed and, moreover, we can obain addiional informaion abou he firms. Since he aim of his paper is o analyse he response of firms a he municipal and firm levels, our daabase consiues a useful ool for analysing he Spanish case. Daa relaed o populaion and immigraion a he municipal level are provided by he Spanish Saisics Insiue obained from he websie of he Anuario Económico de España (2007). This informaion has been crossreferenced wih he informaion on firms. The selecion process for he informaion was as follows. Firs, we considered only hose firms ha were in our daabase for a leas hree years, in order o avoid he appearance of firms ha had enered he marke as par of a financial sraegy. We considered only firms wih more han hree employees and we excluded all firms ha were ouliers. Specifically, we did no include any firms where he average wage was more han 80 housand euros, or any wih negaive invesmen, negaive produciviy, or where he produciviy per worker exceeded 300 housand euros per worker. Likewise firms wih excessive growh raios in erms of employmen, produciviy and wages were also excluded. Finally, he oal number of firms considered was 43,115 and he number of observaions was 119,564. Table 1 provides deailed informaion of hese Spanish manufacuring firms. I shows wheher a firm belongs o a region wih a higher or lower han average share of immigraion. In general, one hing was apparen: firms in regions wih a high percenage of immigrans are larger, hey have more sales, greaer value-added, increased labour produciviy, and hey pay higher wages and experience a higher degree of growh. However, sales growh, value-added produciviy and wages do no change as rapidly for firms in regions wih a low percenage of immigrans. Several differen paerns emerge depending on he individual characerisics of he firm. As expeced, new and small firms have more difficulies compeing han heir counerpars regardless of he region in which hey are locaed. In fac, he smaller hey are he lower he salesgeneraing capaciy, value-added and labour produciviy hey have. Furhermore, hey pay lower wages compared o he average of 21,010 and 11

13 23,430 euros. However, here are differences in growh raes since new firms perform beer han small firms, especially when we observe he rae of growh in erms of size, sales and value-added. Table 1. Saisical descripion of firm characerisics depending on immigran populaion in regions. Year Regions wih low share immigrans All firms Young Firms Small firms Expor- Impor Firms Non expor- Impor Firms Number of firms Size (employees) (87.90) (76.39) 6.30 (1.68) (176.48) (55.94) Gross Sales (housands euros) ( ) ( ) ( ) ( ) ( ) Value Added (housands euros) ( ) ( ) (576.87) ( ) ( ) Labour Produciviy (housands euros) (58.89) (19.99) (86.66) (30.94) (62.44) Wages (housands euros) (8.31) (7.60) (7.99) (9.02) (7.99) Growh Size (127.15) (122.63) 9.23 (44.05) 9.44 (272.95) (76.17) Growh Sales ( ) (720.45) (140.44) ( ) (363.03) Growh Value Added (131.60) (134.16) 9.77 (49.61) 9.82 (280.38) (80.15) Growh Labour Produciviy 1.77 (23.34) 3.53 (27.97) 2.45 (26.12) 1.07 (22.18) 1.89 (23.53) Growh Wages 2.92 (13.97) 3.77 (16.47) 3.61 (16.17) 2.50 (11.93) 2.99 (14.30) Regions wih high share of immigrans Number of firms Size (employees) (163.34) (94.18) 6.36 (1.69) (314.36) (84.65) Gross Sales (housands euros) ( ) ( ) (872.02) ( ) ( ) Value Added (housands euros) ( ) ( ) (150.45) ( ) ( ) Labour Produciviy (housands euros) (23.60) (19.47) (21.55) (31.37) (20.56) Wages (housands euros) (9.21) (8.70) (9.17) (10.13) (8.74) Growh Size (167.89) (176.18) 6.23 (41.76) 9.31 (207.06) (155.91) Growh Sales (302.13) (270.70) 9.11 (116.90) 9.15 (214.22) (321.34) Growh Value Added (161.85) (145.81) 6.65 (43.21) 8.35 (199.32) (150.41) Growh Labour Produciviy 1.56 (22.84) 2.97 (27.46) 2.57 (25.87) 0.81 (22.10) 1.76 (23.02) Growh Wages 3.05 (13.91) 3.64 (16.37) 3.74 (16.30) 2.92 (12.06) 3.08 (14.35) Noe: Young firms are less han six years old. Small firms have beween hree and en employees. Expor-Impor firms are hose ha declare hey expor and impor. Source: Spanish Saisics Insiue and Sabi (Sisema de Análisis de Balances Ibéricos) 12

14 There are also several differences beween hose firms ha repor exporimpor aciviy and hose wihou. In general, firms wih exernal aciviy are larger and perform beer in erms of sales, value-added, labour produciviy and wages. However, firms wihou exernal aciviy obain larger growh raes in erms of size, value-added, labour produciviy and wages. This provides evidence ha, in general, foreign marke oriened firms mus reach a criical size in order o compee in inernaional markes Spanish immigraion evidence Immigran inflows from developing counries o Spain were inense in he period from 1998 o In his decade, he majoriy of immigrans were new, regardless of gender, had few qualificaions, had difficulies speaking Spanish, and came from underdeveloped counries in Africa, Souh America, Asia and Eas European counries. Recenly, family reunificaion processes have led o he migraion of wives/husbands, children and, less ofen, parens. The main consequence of his large migraion has been a rise in he populaion of Spain, alhough here is an unequal disribuion across regions. When we observe he disribuion of immigrans across Spanish regions, we find large differences beween rural and urban regions. According o he municipal regisry, as of he 1s January 2008 in Spain here was a foreign populaion of 5,220,577 inhabians, or 11.3% of he oal populaion. The disribuion of immigrans is heerogeneous among Spanish regions (Map 1). Regions wih an immigran share exceeding he Spanish average form wo differen groups. In he firs, we have wo ses of islands where he presence of immigrans is high due o he fac ha hey are radiional desinaions for reired immigrans from Norhern Europe. In he Balearic Islands, immigrans make up 20.8% of he populaion and in he Canary Islands, 13.6%. In he second group we find regions such as Madrid which are among he mos economically dynamic areas of he Spanish economy. These areas include he Medierranean arc (Murcia %, Valencia- 16.7%, and Caalonia %) and he Corridor of he Ebro (La Rioja %, and Aragón %). In our analysis, we only considered regions on he mainland peninsula. We did no consider he Canary Islands or he Balearic Islands since a major share of immigrans here are ouriss who say for many monhs a a ime. In general, he locaional paerns of immigrans is relaed o he economic dynamism and he manufacuring specializaion of he regions, wih he excepion of Navarra (10.4%) and he Basque Counry (5.4%) which have low percenages of immigrans. 13

15 Map 1. Disribuion of immigrans in Spanish ciies (wih more han 1,000 inhabians). Source: Anuario Económico de España 2007 (La Caixa) The Spanish case seems o be highly illusraive of he riddle: Do firms follow people or do people follow firms? Immigrans end o sele in dynamic regions ha offer more labour opporuniies and which have a higher proporion of foreign populaion (he pull effec) (Map 2). The proporion of immigrans differs in regions and ciies. Immigrans concenrae in dynamic ciies wih diversified indusrial-mixes ha have a diversiy of labour alernaives. Obviously, he concenraion of inflows of immigrans in some ciies has an immediae effec on he demand for housing and in he labour marke. Apar from he regional disribuion, anoher ineresing way o examine he phenomenon of immigraion in a erriory is o analyse he locaion of immigrans by ciy size. Table 2 shows he disribuion of populaion and immigrans according o wheher he municipaliy is in a region wih a higher or lower proporion han average of immigrans. In order o demonsrae he large influx of immigrans in numerical erms, we show informaion from 2001 o We found considerable differences in immigraion raes beween ciies locaed in indusrialized and hose in more rural regions. For example, in he firs group he share of immigrans is high (Madrid %, Barcelona % and Valencia %), while in less indusrialized regions he presence of immigrans is more scarce (Seville - 3.5%, Valladolid - 4.3% and Vigo - 3.8%). Here, he case of he Basque 14

16 Counry is worh menioning, as i is a dynamic, indusrialized region wih low immigraion raes (Bilbao - 5.5%, Vioria - 6.3% and Donosia - 4.5%) 5. Table 2 shows ha of he 1,992 municipaliies, he percenage of immigrans was 1.71%, while in 2006 his percenage had increased o 5.06%. The oal growh of he populaion during his period was 5.35%, while he increase in he percenage of immigrans was %. A he same ime, Table 2 shows ha he percenage of immigrans in 2001 was 4.64%, while in 2006 his percenage had increased o 13.12%. The oal growh rae during he period was 12.60% while he growh rae of immigrans was %. Map 2. Disribuion of employees in Spanish ciies (wih more han 1,000 inhabians). Source: Sisema de Análisis de Balances Ibéricos Thus, hree differen paerns describing he flow of populaion o Spanish ciies emerge. Firs, populaion growh in ciies wih fewer han 5,000 inhabians was negaive. This expulsion of populaion from large meropolian areas reflecs he disadvanages suffered by large meropolian areas in comparison wih less densely populaed areas. Second, for ciies wih beween 20,000 and 99,999 inhabians he growh rae in erms of populaion and immigrans was higher han in larger or smaller ciies. Third, here is an inverse U-shaped relaionship beween ciy size and percenage of immigrans. This is he resul of he capaciy of large meropolian areas o arac young, highly-qualified populaion ha can compensae for he large number of immigrans. 5Anuario Económico de España,

17 Table 2. Immigran locaion according o ciy populaion and regional raes of immigraion Regions wih low relaive immigrans rae N Toal Populaion Toal Immigrans % N Toal Populaion Toal Immigrans % Growh (%) Populaion More han inhab. 28 6,155,386 99, ,330, , Beween inhab ,752, , ,507, , Beween inhab ,652,386 72, ,797, , Fewer han 5000 inhabians 1,356 3,096,937 40, ,020, , Toal 1,992 18,656, , ,992 19,655, , Regions wih high relaive immigrans rae Growh (%) immigraion N Toal Populaion Toal Immigrans % N Toal Populaion Toal Immigrans % Growh (%) Populaion Growh (%) immigraion More han inhab. 24 9,412, , ,462,391 1,375, Beween inhab ,461, , ,413, , Beween inhab ,861, , ,185, , Fewer han 5000 inhabians 649 1,381,134 51, ,339, , Toal 1,078 18,117, , ,078 20,400,700 2,675, Noe: Regions wih low relaive immigraion raes: Andalucia, Asurias, Canabria, Casilla y León, Casilla-LaMancha, Exremadura, Galicia, Navarra and Basque Counry. Regions wih high relaive immigraion raes: Madrid,Murcia, Valencia, Caalonia, La Rioja and Aragón Sources: Spanish Saisics Insiue 16

18 When immigrans sele in a counry hey do no disribue hemselves homogeneously hroughou he erriory. Indeed, immigrans are heerogeneous groups made up of individuals ha are moivaed o uproo hemselves for differen reasons; some wish o leave a place behind, ohers wish o be somewhere new. Of paricular ineres here are daa ha disinguish beween emigraion flows from counries wih higher or lower income levels han hose of he counry of desinaion. According o he municipal regisry office, as of he 1 s January 2008, 18.4% of immigrans came from counries wih higher income levels (he UK - 6.7%, Germany - 3.5%, Ialy - 3.0%, and France - 2.2%) while he remaining 71.6% came from less developed counries (Rumania %, Morocco %, Ecuador - 8.0% and Colombia - 5.4%). Of equal ineres is a deerminaion of he main secors of aciviy ha employ foreign-born immigrans. In general, men find largely permanen jobs in he consrucion, commercial and ouris secors, while women work in domesic service, and less frequenly, in he ouris and commercial secors. The manufacuring secor is an opion for a smaller number of immigrans. According o he Encuesa de Inmigración (immigraion census) in 2007, a oal of 2,269,092 foreign-born immigrans have jobs. The secors wih he highes percenages of immigrans are consrucion (19.3%), he hoel and caering indusry (13.1%), he commercial secor (12.6%) and manufacuring (11.0%). In 2007, Spanish manufacurers employed 249,857 immigrans. Alhough he percenage of immigrans in manufacuring jobs sands a 10% of oal employees in Spanish manufacuring, hey apply considerable pressure on he labour marke because he oal volume of employmen has fallen moderaely and access o jobs for immigrans in he indusrial secor is a recen phenomenon 6. The Encuesa de Inmigración offers informaion abou he work immigrans performed in heir counry of origin. In Spanish manufacuring indusry, 44.3% of immigrans have worked previously in manufacuring jobs in heir counries of origin and 55.7% have had work experience in oher secors (agriculure %, consrucion - 8.6%, he hoel and caering indusry - 7.4% and he commercial secor - 6.7%). The lack of experience and qualificaions of immigran workers increases he coss of adjusmen and adapaion in manufacuring jobs. This siuaion hinders job efficiency and lowers job produciviy. 5. Economeric mehodology and variables 5.1. Economeric mehodology In keeping wih he main premise underlying Roback s model, we propose hree equaions for analysing he effecs of immigraion on a firm s 6 In 2000 here were 2,577,929 employees in manufacuring jobs and 2,511,279 in Encuesa Indusrial de Empresas, Spanish Insiue of Saisics. 17

19 performance. Thus, our main purpose is o analyse he impac of immigraion on he dynamics of he firm. However, we also analyse his impac on levels (Tables A.2 and A.3) and we obain similar resuls 7. Since our main purpose is o analyse he response of firms o he percenage of immigraion, we use Gibra s Law, which assumes ha he firm growh (gi,) ln ): S i, of firm i in period is independen of firm size( ( ) ( Si, 1 ) ui g i, = α + β1 ln +, Since some firms are more likely o be efficienly organized because of indusry-specific differences in fixed coss or because hey are locaed in cerain regions wih access o specialised raw maerials, we include firmspecific variables and erriorial variables ha influence firm behaviour, and obain he following equaion: ( Si, 1 ) + β 2 ln( Agei, ) + β 3Densi, + 4 mig i, ui, g i, = α + β1 ln β Im + (Eq. 1) Where gi, is he dependen variable calculaed as he difference beween firm size i in year and period -1, ln ( S i, ) is firm size, ln ( Age i, ) is firm age, Dens i, is he municipaliy densiy, Im mig i, is he share of immigrans and u, is an error erm. i Deparing from Gibra s Law (Equaion 1), in order o analyse wheher flows of immigraion in each Spanish municipaliy influence paerns of firm growh in erms of employees (Equaion 2), wages (Equaion 3) and produciviy (Equaion 4) we calculae he following equaions: growhe growhw i, i, growhlp = ln = α = ln = α + i, ( Ei, ) ln( Ei, 1 ) + β1 ln( S i, 1 ) + β 2 ln( Agei, ) + β Densi, + β 4 Im mig i, + ui, = ln = α + ( Wi, ) ln( Wi, 1 ) β1 ln( S i, 1 ) + β 2 ln( Agei, ) + β 3Densi, + β 4 Im mig i, + ui, ( LPi, ) ln( LPi, 1 ) β1 ln( S i, 1 ) + β 2 ln( Agei, ) + β 3Densi, + β 4 Im mig i, + ui, 3 (Eq. 2) (Eq. 3) (Eq.4) In addiion, hese equaions are esimaed for boh small and new firms and according o he exernal aciviy of firms. 7 Our resuls confirm our hypohesis: a higher percenage of immigraion has a negaive impac on job performance, bu a posiive impac on he level of wages and labour produciviy. 18

20 Our economeric mehodology involved he applicaion of panel daa wih fixed effecs (GLS, Generalised Leas Squares) o conrol for individual characerisics ha may affec firm performance. Hausman ess were performed o confirm is suiabiliy for hese esimaions. The variables applied were he following: Variables Descripion Daabase Dependen variable growhei, Firm growh equal o he logarihmic difference of SABI employees (E). growhwi, Firm growh equal o he logarihmic difference of SABI wages (W). growhlpi, Firm growh equal o he logarihmic difference of labour produciviy (LP) SABI Explanaory variables ln ( S i, ) Logarihm of firm size measured in number of employees SABI ln Logarihm of firm age SABI ( ) Age, i Terriorial variables Dens i, Densiy of populaion where he firm locaes INE Immigraion i, Share of immigrans in he municipaliy INE Source: auhors own The relaionship beween firm growh and firm size measured in erms of he number of workers allows us o conduc our analysis using Gibra s Law, which assumes ha firm growh follows a random pah. However, a large body of empirical lieraure repors a negaive relaionship and, hus, refues his hypohesis (for Spain, see Fariñas and Moreno, 2000; Peña, 2004; Calvo, 2006; Teruel, 2008). Two facors explain his negaive relaionship. Firs, new firms end o be smaller han hose ha have been operaing for a longer ime, and so are unable o exploi scale economies in he secor. Second, he lieraure poins o he exisence of a minimum efficien size. Thus, firm growh should favour he likelihood of a firm s survival. A firm s age ranslaes as marke experience (Jovanovic, 1982). Alhough, in general, here is a negaive relaionship beween firm growh and age, a diversiy of resuls are, in fac, o be found. On he one hand, new firms are usually small and seek o grow o achieve he minimum efficien size. On he oher hand, hese new firms suffer from a lack of experience in he marke and are unaware of heir level of efficiency. Cabral and Maa (2003) repor ha new firms presen a more asymmeric disribuion, which approaches normaliy over ime. According o hese auhors, such differences reflec he financial resricions wih which new firms have o conend. Recenly, his hypohesis has been confirmed by Barrios e al. (2005) while Fagiolo and Luzzi (2006) fail o find empirical evidence of greaer financial consrains among new firms. 19

21 Keeble and Walker (1994) mainain ha populaion densiy represens he exisence of agglomeraion economies or diseconomies. These can be relaed o labour coss, knowledge spillovers, marke demand, accessibiliy and congesion. Populaion densiy can have wo differen effecs on a firm s performance. On he one hand, i may have a posiive impac hanks o he effecs of a compeiive environmen and because of he ameniies ha are available in culurally diverse ciies 8. Bu, on he oher hand, ciies wih high populaion densiies may ac o eliminae negaive impacs on firm performance, if firms do no benefi in some oher way. Higher wages, congesion problems and higher land prices are among he drawbacks a firm faces if choosing o locae in a densely populaed ciy. Recenly, Duranon and Puga (2001) repored a posiive relaionship beween populaion and locaion when firms use highly skilled workers and he firms are knowledge inensive, while firms based on scale economies and which are land inensive end o locae in less densely populaed ciies. Here, we considered he number of immigrans as a proporion of he oal populaion. There is a wide body of evidence suggesing ha immigrans are unskilled and occupy jobs wih low levels of produciviy. Kim (2007) considers he share of foreign-born populaion (FB/(FB+NB)) o be a useful measure of he relaive supply of unskilled o skilled workers (L/H). Some auhors, such as Salerian, (2006) propose ha he arrival of immigrans has an effec on labour skills. Skilled workers are more producive and, hus, are paid higher wages han unskilled workers. As a consequence, he arrival of immigrans negaively affecs average wages, produciviy and skill levels. However, Oaviano and Peri (2005, 2006) repored a posiive impac of immigrans on wages and employmen. This posiive impac occurs via produciviy, since a higher number of low-skilled immigrans implies an increase in skilled workers, which has a compensaing effec on he produciviy average. Wheher here is labour displacemen or compensaion in he labour marke, i is unlikely ha he percenage of immigrans in his marke is direcly affeced. However, he foreign-born populaion rae is subjec o wo significan sources of measuremen error: (1) a considerable number of foreign-born workers in manufacuring indusries are skilled and (2) no all naive-born workers are skilled. Due o hese measuremen errors, he share of foreign-born immigrans mus be inerpreed carefully. An addiional reason for cauion is he fac ha no all immigrans paricipae in he labour marke, paricularly following an inense process of family regrouping in recen years. Moreover, immigrans who paricipae in he 8 Agglomeraion economies can have five differen effecs, hey migh: i) increase knowledge spillovers (Morrison and Siegel, 1999); ii) generae economies of localizaion because of producion inpus ha are shared (Quigley, 1998); iii) exploi specialisaion and scale economies in local ranspor neworks (Ciccone and Hall, 1996); iv) reduce ransacion coss (Quigley, 1998); v) increase he adopion of new echnologies due o he higher capaciy for R&D (Keller, 2002). 20