Does Immigration Substitute for Offshoring?

Transcription

1 Does Immigraion Subsiue for Offshoring? William W. Olney March, 2009 Absrac: Research generally focuses on how immigraion affecs naive workers, while he impac of immigraion on domesic firms is ofen overlooked. This paper addresses his imporan omission by examining wheher firms respond o immigraion by adjusing he locaion of heir producion aciviies. Consisen wih he predicions of he model, he resuls sugges ha low skilled immigraion decreases offshoring while high skilled immigraion increases offshoring. This suggess ha a policy ha resrics immigraion will have imporan implicaions for he organizaional srucure of domesic firms. Furhermore, hese resuls explain why he impac of immigraion on he wages of naive workers is ofen found o be quie small. Keywords: offshoring, ousourcing, immigraion, firm srucure, esablishmen birhs and deahs JEL Codes: F16, F22, L2 Deparmen of Economics, Universiy of Colorado a Boulder, 256 UCB, Boulder, CO, William.olney@colorado.edu

2 1. Inroducion The impac of immigraion on domesic labor markes and specifically he implicaions for naive workers has ignied a fierce debae among economiss, poliicians, and he general public. However, he effec of immigraion on domesic firms and heir organizaional srucure is seldom sudied bu is of equal imporance. This paper examines wheher firms respond o immigraion by adjusing he locaion of heir producion aciviies. Specifically, does immigraion affec he firms decision o offshore jobs? 1 This paper consrucs a simple model ha highlighs he relaionship beween immigraion and offshoring and ess hese predicions using a comprehensive daa se of U.S. ciies. Offshoring and immigraion are forms of globalizaion ha are of grea concern o American workers. 2 However, he relaionship beween he wo is seldom examined. Boh offshoring and immigraion are mechanisms ha expand he effecive labor supply available o domesic firms. Thus, here is a naural radeoff beween foreign workers migraing ino he domesic counry o work or jobs being relocaed abroad o be performed by foreign workers in heir naive counry. While i is ofen assumed ha immigraion limis he need for domesic firms o locae producion aciviies abroad, here is lile empirical evidence supporing his hypohesis. This paper presens a simple model ha highlighs he ways in which immigrans can be absorbed ino a local labor marke. Eiher relaive wages can adjus or firms can respond by changing he locaion of heir producion aciviies. The fac ha research 1 Offshoring refers o he relocaion of domesic jobs o foreign counries. This includes movemens of jobs wihin he firm or o anoher foreign firm. 2 Public Says American Work Life is Worsening, Bu Mos Workers Remain Saisfied wih Their Jobs, Pew Research Cener,

3 shows ha he impac of immigraion on relaive wages is minimal (Card 2005, Borjas 1994, Friedberg and Hun 1995) suggess ha his laer componen may be paricularly imporan. The model generaes sharp predicions on how low skilled and high skilled immigraion will affec offshoring. In paricular, i is shown ha an influx of low skilled immigrans reduces he need for firms o offshore low skilled jobs. However, high skilled immigraion will increase offshoring since he relaively limied supply of domesic low skilled workers forces firms o shif low skilled producion aciviies abroad. The predicions of he model will be esed by exploiing differences in immigraion and offshoring across U.S. Meropolian Saisical Areas (MSAs or ciies). The endogeneiy of he immigran locaion decision is addressed by aking advanage of he fac ha curren immigrans ofen locae in ciies where previous immigrans from he same counry already live. Thus, a ciies hisorical share of foreign born residens is used as an insrumen for he curren selemen paerns of immigrans from ha paricular counry. This capures he variaion in immigraion ha is exogenous o local labor demand and allows he impac of immigraion on offshoring o be idenified. Consisen wih he predicions of he model, he resuls show ha low skilled immigraion increases he rae of ne esablishmen birhs (reduces offshoring) while high skilled immigraion decreases he rae of ne esablishmen birhs (increases offshoring). I also find ha low skilled immigraion and high skilled immigraion do no have a significan impac on relaive wages. Thus, his paper provides compelling evidence ha immigrans are absorbed ino local labor markes hrough changes in firms esablishmen locaion decisions raher han hrough changes in relaive wages. While he ne birh rae 4

4 of esablishmens is no a perfec proxy for offshoring, he resuls are sronger in indusries in which offshoring is more common and for firms ha are more likely o offshore. Furhermore, firms no only respond o immigraion by relocaing esablishmens bu hey also adjus employmen levels wihin esablishmens. Consisen wih he predicions of he model, he resuls indicae ha low skilled immigraion increases he rae of ne expansions (decreases offshoring) while high skilled immigraion decreases he rae of ne expansions (increases offshoring). Overall, his paper provides compelling evidence ha firms respond o immigraion by changing heir organizaional srucure and in paricular by adjusing heir offshoring decisions. Numerous careful sudies have shown ha immigraion has, a mos, a small impac on naive relaive wages. 3 Furhermore, here is lile evidence ha immigraion leads o ouflows of similarly skilled naive workers (Card and DiNardo 2000, Card 2001) or ha he indusry composiion wihin a ciy adjuss o accommodae immigrans (Lewis 2003, Card and Lewis 2005). Thus, i is unclear how local labor markes are able o absorb ofen large influxes of immigrans. This paper provides a plausible explanaion for why he impac of immigraion on relaive wages is found o be small. The resuls sugges ha immigrans are absorbed hrough changes in firms offshoring decisions wihou any change in relaive wages. Recen models have compared and conrased he impac of immigraion and offshoring on naive wages (Jones 2004, Grossman and Rossi-Hansberg 2008, Olney 2009). However, hese papers do no focus on wheher immigraion affecs offshoring. 4 3 See Card s (1990) seminal work on he Mariel Boalif and surveys of he lieraure provided by Borjas (1994), Friedberg and Hun (1995) and Card (2005). 4 There is lile evidence of immigraion impacing offshoring using he sae-indusry level daa from Olney (2009). 5

5 While he relaionship beween immigraion and offshoring is inuiively appealing, here are few empirical sudies ha invesigae he link beween he wo. Using firm daa from Ialy, Navarei, Berola, and Sembenelli (2008) show ha firms ha offshore are less likely o employ immigran workers. However, his paper views he causaion running in he oher direcion. Immigraion is due o facors largely exogenous o he firm and given hese labor supply condiions he firm hen makes a decision abou wheher o offshore. This paper is also relaed o he lieraure on how firms respond o changes in he local labor supply. These models ypically show ha he ypes of jobs (Acemoglu 1999) or he echnologies adoped (Acemoglu 1998, Beaudry and Green 2003) depend on he local facor supplies. Lewis (2005) finds evidence ha ciies wih less skilled labor supplies are more likely o adop less auomaed echnologies. This paper is similar in spiri o his research in ha i highlighs how firms decisions may respond o local labor supplies. However, none of hese papers invesigae wheher changes in facor supplies affec he producion locaion decisions facing domesic firms. This paper conribues o his lieraure by examining wheher immigraion affecs he decision o offshore. The remainder of he paper is organized as follows. Secion 2 presens wo simple models ha highligh he ways in which immigrans can be absorbed ino local labor markes. Secion 3 gives an overview of he daa used in his analysis and Secion 4 oulines he esimaion sraegy. Finally, he resuls are presened in Secion 5 and Secion 6 concludes. 6

6 2. Model This secion provides insigh ino he relaionship beween immigraion and offshoring. Firs, a simple model is presened in which offshoring is no possible. Thus, labor supply shocks due o immigraion are absorbed hrough changes in relaive wages. While inuiively appealing his framework is no suppored by recen empirical evidence. The model is hen exended o allow for he possibiliy ha firms may choose o offshore low skilled jobs. Thus, labor supply shocks due o immigraion are absorbed by changes in offshoring raher han changes in relaive wages. This generaes esable predicions for he implicaions of low skilled and high skilled immigraion on offshoring. 2.1 No Offshoring Suppose a perfecly compeiive firm uses low skilled labor (L) and high skilled labor (H) o produce a good (Y). Domesic low skilled and high skilled labor consiss of naive and immigran workers. Condiional on skill, naive and immigran workers are equally producive in he producion process. For he ime being, i is assumed ha offshoring is no possible. Thus, he firm faces he following profi maximizing decision (aking he price of good Y as he numeraire): α 1 α (1) max π = A ( L + L ) ( L + L ) w ( L + L ) w ( H + H ) N I N I L where he subscrip N indicaes naive workers, subscrip I indicaes immigran workers, w L is he wage paid o low skilled workers, and w H is he wage paid o high skilled workers. The sandard firs order condiion is: N I H N I (2) 1 α ( L α ( H N N + L ) I + H I ) = w w H L 7

7 Toally differeniaing (2), assuming ha naive labor supplies do no change, generaes he following relaionship beween immigraion and relaive wages: (3) wˆ H wˆ L = dl L I dh H I Equaion (3) highlighs he exbook relaionship beween wages and immigraion. An increase in low skilled immigrans should increase he relaive wage, while an increase in high skilled immigrans should decrease he relaive wage. Thus, labor supply shocks due o immigraion are fully absorbed by changes in he relaive wage. While inuiively appealing, his hypohesis has found lile empirical suppor. Numerous careful sudies have deermined ha, a mos, immigraion has a small impac on naive wages. Thus, i is unclear how local labor markes are able o absorb hese labor supply shocks. Offshoring may provide an answer. 2.2 Offshoring Assume ha i is now possible for domesic firms o offshore low skilled jobs, while all high skilled jobs mus be performed domesically. 5 Therefore, domesic firms have he opion of hiring a naive low skilled worker, hiring an immigran low skilled worker, or relocaing he job abroad o be performed by a low skilled foreign worker. Thus, he firm maximizes profis: (4) α 1 α * max π = A ( L + L + L ) ( H + H ) w ( L + L ) ( w + β )( L ) w ( H + H ) N I O N I L N I L O H N I 5 While his is a odds wih he recen aenion paid o he offshoring of whie collar jobs, i is generally agreed ha his phenomenon is sill quie small (Amii and Wei 2006, Blinder 2005) 8

8 where he subscrip O indicaes offshored jobs, w* L is he wage paid o foreign low skilled workers and β is he coss associaed wih offshoring. The firs order condiion is now: (5) 1 α ( LN + LI + LO ) α ( H + H N I ) = w w H L Following he empirical evidence, I assume ha immigraion has lile impac on wages and focus on he implicaions for offshoring. Toally differeniaing (5) yields he following relaionship beween changes in immigraion and changes in offshoring: (6) dl L O dl = L I dh + H I Equaion (6) highlighs he ways in which he firms decision o offshore depends on differen labor supply shocks. An increase in low skilled immigraion (dl I >0) leads o a reducion in offshoring (dl O <0). Domesic firms no longer need o locae producion aciviies abroad in order o access he low skilled foreign labor supply. These foreign low skilled workers are migraing o he firms. Thus, low skilled immigraion and offshoring are subsiues in he producion process. However, an increase in high skilled immigraion (dh I >0) leads o an increase in offshoring (dl O >0). The relaively limied supply of domesic low skilled workers forces firms o shif low skilled producion aciviies abroad. As a resul high skilled immigraion and offshoring are complemens in he producion process. This simple framework idenifies a way in which local labor markes can absorb immigrans wihou any change in relaive wages. Domesic firms adjus he locaion of heir producion aciviies in response o immigraion. In oher words, offshoring provides a mechanism for domesic firms o adap o changes in local labor supplies. As 9

9 a resul, relaive wages are lef unchanged which is consisen wih much of he empirical evidence. This simple model generaes clear predicions for he relaionship beween immigraion and he firms organizaional srucure. High skilled immigraion will increase offshoring while low skilled immigraion will decrease offshoring. The remainder of he paper will examine wheher offshoring adjuss o immigraion in he manner prediced in (6) Daa The daa se used in his analysis spans 192 U.S. Meropolian Saisical Areas and 7 years ( ). Using an MSA as he uni of analysis is appealing because i represens a reasonably closed labor marke while inroducing a subsanial amoun of variaion. Curren Populaion Survey (CPS) daa on individuals beween he age of 18 and 65 is obained from he Inegraed Public Use Microdaa Series (IPUMS). From his he share of he populaion ha is foreign born by skill level, MSA, and year is calculaed. Immigrans are caegorized as low skilled if hey have a high school degree or less and caegorized as high skilled if hey have more han a high school degree. The op wo panels in Figure 1 show he disribuions of low skilled and high skilled immigran shares along wih he op five and boom five MSAs in each caegory. There is subsanial variaion in he share of low skilled and high skilled immigrans across MSAs. No surprisingly, he share of low skilled immigrans is highes in Miami, FL and in ciies locaed in California and Texas while he share of high skilled immigrans is 6 While his framework focuses on shifing jobs abroad, firms may also respond o immigraion by shifing jobs from one domesic ciy o anoher. However, from he perspecive of a hos ciy i is inconsequenial wheher hose jobs are ransferred o anoher domesic ciy or a foreign counry. Furhermore, jobs ha are shifed from one domesic ciy o anoher are he jobs ha are mos a risk of being offshored in he fuure (Jensen and Klezer 2006). 10

10 paricularly large in high ech ciies such as San Jose, CA (Silicon Valley), San Francisco, CA, and Samford, CT. The CPS daa is also used o calculae a variey of oher demographic characerisics, such as he age, gender, race, and educaional aainmen of he naive populaion in each MSA. Daa on offshoring by MSA is unavailable, however he U.S. Census Bureau s Saisics of U.S. Businesses (SUSB) provides daa on he universe of esablishmens. An esablishmen is defined as a single locaion a which business, services, or indusrial operaions are conduced. A firm may own and operae many esablishmens. Thus, an esablishmen represens he smalles uni of producion aciviy for which daa is available. Offshoring will lead o a decrease in he number of esablishmens as domesic firms relocae hese producion aciviies abroad. The possibiliy ha firms may offshore paricular jobs wihin an esablishmen will be examined in Secion 5.3. One paricularly appealing aspec of his daa is ha he SUSB consrucs a longiudinal daa se by linking esablishmens across years. Alhough daa on specific esablishmens and firms is wihheld for confidenialiy reasons, he SUSB does provide aggregae daa on esablishmen birhs, deahs, expansions, and conracions by couny, year, esablishmen indusry, and firm size. Counies are hen assigned o he appropriae MSA which alleviaes he problem ha MSA definiions have changed over he sample. The rae of ne esablishmen birhs is calculaed in he following manner: ne _ birh _ rae es _ birhs es _ deahs = where c indexes MSAs and indexes es years. This measure conrols for he naural churning of esablishmens by aking he difference beween birhs and deahs and accouns for he fac ha some MSAs are larger han ohers by aking he raio of ne birhs o he number of esablishmens a he 11

11 beginning of he year. The boom panel of Figure 1 shows he disribuion of ne birh raes and liss he MSAs wih he highes and lowes raes. No surprisingly, MSAs in he wes and in Florida have paricularly high esablishmen birh raes while Midwesern MSAs experienced a decrease in esablishmens. 4. Esimaion Sraegy The models discussed in Secion 2 presen wo alernaive heories on how local labor markes can absorb immigrans. Immigraion may lead o changes in he relaive wage, changes in firms organizaional srucure, or some combinaion of he wo. While no he focus of his paper, I begin by examining he impac of low skilled and high skilled immigraion on relaive wages. Specifically, he following equaion is esimaed: (7) relaive _ wage = α 0 + α1l _ img + α 2H _ img + α 3 ' X + e The dependen variable is he difference beween he log median wage of high skilled and low skilled workers. The independen variables of ineres are he share of low skilled immigrans in he oal populaion (L_img) and he share of high skilled immigrans in he oal populaion (H_img). X is a vecor of conrol variables including he age, gender, race, and educaional aainmen of he naive populaion. Equaion (3) suggess ha low skilled immigraion increases he relaive wage (α 1 >0) and high skilled immigraion decreases he relaive wage (α 2 <0). However, if previous research is any indicaion, i is likely ha his effec is small. While research generally assumes ha immigraion will primarily affec he wages of naive workers, he model oulined in Secion 2.2 provides an alernae heory. Immigraion may lead o changes in he organizaional srucure of domesic firms raher 12

12 han affecing naive workers. Specifically, he goal of his analysis is o es wheher offshoring responds o immigraion in he manner prediced in (6). To es hese predicions he following equaion will be esimaed: (8) ne _ birh _ rae = β 0 + β1l _ img + β 2H _ img + β3 ' X + e Given he predicions of he model in (6), we expec ha β 1 >0 because low skilled immigraion reduces offshoring and hus increases he rae of ne esablishmen birhs. Conversely, we expec ha β 2 <0 because, according o he model, high skilled immigraion increases offshoring and hus reduces he rae of ne esablishmen birhs. One legiimae concern is ha immigraion and he dependen variables in (7) and (8) are correlaed wih ciy characerisics ha are no conrolled for. For insance, a MSA wih a rapidly expanding economy would likely experience a faser ne esablishmen birh rae and may also arac immigrans looking for jobs. To alleviae hese concerns, I use hisorical selemen paerns of immigrans as an insrumen for curren selemen paerns of similar immigrans. 7 The prediced share of immigrans is consruced by assigning acual immigrans in he curren year o he ciies where heir counrymen were locaed in This insrumen akes advanage of he fac ha curren immigrans ofen sele in ciies where previous immigrans from he same counry already live (Barel 1989). New immigrans likely find i appealing o sele in ciies wih enclaves of residens who share heir language, religion, or culure. Furhermore, in order o reunie families, he Unied Saes mainly awards visas o applicans who have relaives already residing in he counry. 7 This insrumen is similar o he one used by Card (2001) and Lewis (2005). 13

13 The prediced shares of low skilled and high skilled immigrans are calculaed in he following manner: (9) pred _ L _ img = r img L _ img r, * img pop r,1980 r,1980 pred _ H _ img = r img H _ img r, * img pop r,1980 r,1980 where c indexes MSAs, indexes years, and r indexes 8 regions of he world. 8 The firs erm in he numeraor of he righ hand side of he equaions in (9) is he oal number of foreign born low skilled and high skilled residens from region r in year. This oal is hen assigned o MSAs using he share of oal immigrans from region r ha resided in ciy c in 1980 (he second erm in he numeraor of he righ hand side of he equaions in (9)). Thus, for each year he acual number of foreign born residens from region r is disribued across ciies based on where immigrans from he same region of he world were locaed in This produc is hen summed over regions r and divided by he populaion o obain a prediced immigran share for boh skill groups. Using he prediced share of immigrans as an insrumen will capure variaion in immigraion ha is driven by family and culural reasons raher han by labor demand facors. This miigaes endogeneiy concerns and allows he impac of immigraion on offshoring o be idenified. 8 The regions are U.S. Areas, Norh America, Cenral America, Souh America, Europe, Asia, Africa, and Oceania. 14

14 Figure 2 plos he acual immigran share agains his prediced immigran share. The correlaion beween acual and prediced immigraion is srong. The correlaion coefficien for low skilled immigrans is 0.73 while he correlaion coefficien for high skilled immigrans is However, here are many MSAs whose acual immigran shares are bigger or smaller han wha curren immigraion and hisorical selemen paerns predic. These deviaions from a 45-degree line sugges ha local labor demand facors influence which ciies immigrans choose o live in and imply ha he IV esimaion sraegy will be imporan. Table 1 repors he firs sage regressions resuls. The insrumens do a remarkable job of predicing acual immigran shares. The prediced low skilled immigran share has a large, posiive, and significan effec on acual low skilled immigraion. Similarly, he prediced high skilled immigran share has a large, posiive, and significan effec on acual high skilled immigraion. The F-sa in boh regressions is above 200. These resuls indicae ha hisorical immigran enclaves are useful in predicing immigran selemen paerns more han 20 years laer. Furhermore, given he lengh of his lag, i is unlikely ha he insrumen is correlaed wih curren local labor demand condiions. 5. Resuls This secion includes four ses of empirical resuls. Firs, he impac of immigraion on naive relaive wages is examined. While no he focus of his paper, hese resuls provide a poin of reference for he analysis ha follows. Second, I esimae he effec of immigraion on he organizaion srucure of firms. In paricular, he impac 15

15 of low skilled and high skilled immigraion on he rae of ne esablishmen birhs is examined. Third, ne esablishmen birhs are broken down by indusry and firm size. If offshoring is leading o he observed changes in he ne birh rae hen he resuls should be sronger in indusries in which offshoring is more prevalen and among firms ha are more likely o offshore. Finally, he impac of immigraion on esablishmen employmen is examined. This accouns for he possibiliy ha offshoring may lead o changes in employmen wihin an esablishmen (inensive margin) in addiion o changes in he number of esablishmens (exensive margin). 5.1 Relaive Wages The OLS esimaes of he impac of low skilled and high skilled immigraion on naive relaive wages (Equaion 7) are repored in he firs column of Table 2. The resuls indicae ha high skilled immigraion has a posiive and significan effec on relaive wages. This conradics he predicions of he model where high skilled immigraion leads o a reducion in relaive wages. However, high skilled immigrans may be relaively mobile and hus move o MSAs ha are paying higher wages. This would lead o a spurious posiive correlaion beween high skilled immigraion and relaive wages and may explain he resuls in he firs column. To conrol for hese labor demand facors, I insrumen curren immigran shares wih he prediced immigran shares consruced using hisorical selemen paerns. These IV resuls are repored in he second column. Neiher low skilled immigraion nor high skilled immigraion has a significan effec on naive relaive wages. These findings are consisen wih previous research bu are somewha puzzling. How are local labor markes able o accommodae 16

16 influxes of immigrans wihou local relaive wages changing? One poenial answer is ha firms esablishmen locaion decisions may respond o immigraion paerns. 5.2 Ne Birhs As oulined in Secion 2.2, firms may respond o immigraion by adjusing heir organizaional srucure. Specifically, immigraion may lead o changes in offshoring wih no effec on local relaive wages. To es his hypohesis, he impac of immigraion on he ne esablishmen birh rae is esimaed (Equaion 8). In he firs column (OLS) of Table 3, low skilled and high skilled immigraion have a posiive impac on he ne birh rae. However, hese resuls should be inerpreed wih cauion due o he endogeneiy of he immigran locaion decision. The second column of Table 3 presens he IV resuls. Low skilled immigraion has a significan posiive effec and high skilled immigraion has a significan negaive effec on he ne esablishmen birh rae. A one percenage poin increase in he share of low skilled immigrans leads o a 0.11 percenage poin increase in he ne birh rae while a one percenage poin increase in he share of high skilled immigrans leads o 0.26 percenage poin decrease in he ne birh rae. The difference beween he OLS resuls and he IV resuls indicae ha he OLS esimaion sraegy is plagued by endogeneiy. A ciy wih a growing economy may experience faser esablishmen growh and may arac high skilled immigrans, which would explain he upward bias in he OLS coefficien on high skilled immigraion. Conversely, low skilled immigrans may be araced o more affordable ciies wih a less robus economy and slower esablishmen growh, which would lead o he downward bias in he OLS coefficien on low skilled immigraion. Table 3 emphasizes he 17

17 imporance of focusing on a componen of immigraion ha is unrelaed o local labor demand facors. The IV resuls in Table 3 suppor he asserion ha firms respond o immigraion by adjusing he locaion of heir producion aciviies. Togeher he resuls in Table 2 and 3 sugges ha immigrans are absorbed ino local labor markes hrough changes in he number of esablishmens raher han hrough adjusmens in relaive wages. Furhermore, hese resuls are consisen wih he predicion of he model from Secion 2.2. As low skilled immigraion increases, he firm has less need o offshore jobs and hus he esablishmen birh rae increases. As high skilled immigraion increases, he firm offshores more low skilled jobs and hus he esablishmen birh rae decreases. The resuls in Table 3 are consisen wih hese predicions. Specifically, low skilled immigraion increases he ne esablishmen birh rae while high skilled immigraion decreases he ne esablishmen birh rae. However, o more carefully examine wheher offshoring is explaining he changes in he ne birh raes observed in Table 3, i is useful o compare he relaionship beween immigraion and he ne birh rae across groups in which offshoring is more and less common. 5.3 Esablishmen Indusry and Firm Size Ne esablishmen birhs can be summed over a group of indusries in which offshoring is more prevalen and a group of indusries in which offshoring is less prevalen. 9 If offshoring is indeed leading o he observed changes in he ne 9 Indusries are broken ino offsoring and non-offshoring groups based on which indusries had he larges share of foreign affiliae employmen (BEA). Those indusries in which offshoring is relaively more common include: Mining; Uiliies; Manufacuring; Wholesale Trade; Informaion; Finance and Insurance; Professional, Scienifi and Technical Services; Adminisraive and Suppor and Wase Managemen and 18

18 esablishmen birh rae, hen immigraion should have a larger effec in hose indusries in which offshoring is more common. The resuls in Table 4 confirm his hypohesis. The effec of low skilled and high skilled immigraion on he ne birh rae is smaller and less significan in indusries ha have relaively lile offshoring (Column 2). Since relocaing producion faciliies abroad is less common in hese indusries, immigraion has a smaller impac on he number of esablishmens. However, he effec of low skilled and high skilled immigraion on he ne birh rae is larger and more significan in hose indusries in which offshoring is relaively more common (Column 3). The comparison beween hese wo ses of resuls suggess ha offshoring is a key componen in explaining he relaionship beween immigraion and changes in he number of esablishmens. Ne esablishmen birhs are also grouped according o he employmen size of he paren firm. Since larger firms are more likely o engage in mulinaional aciviy (Bernard and Jensen 2007), hese firms are prone o offshore jobs in response o immigraion. If offshoring is explaining he changes in he ne birh rae, hen he resuls should be sronger among larger firms who are more likely o relocae producion aciviies abroad. Table 5 compares he effec of immigraion on he ne esablishmen birh rae for small (employmen<20), medium (20 employmen<500) and large (500 employmen) firms. The resuls in Table 5 verify ha immigraion has he sronges effec on he ne esablishmen birh rae of he larges firms. These are he firms ha are mos likely engaged in mulinaional aciviy and ha have he resources Remediaion Services; and Accommodaion and Food Services. The non-offshoring group of indusries includes: Agriculure, Foresry, Fishing and Huning; Consrucion; Reail Trade; Transporaion and Warehousing; Real Esae and Renal and Leasing; Managemen of Companies and Enerprises; Educaion Services; Healh Care and Social Assisance; Ars, Enerainmen, and Recreaion; and Oher Services. 19

19 and experise o offshore producion aciviies. These differences across firm size offers furher evidence ha offshoring is a key facor in explaining he relaionship beween immigraion and changes in he ne birh rae. 5.4 Ne Expansions So far, I have assumed ha offshoring consiss of an enire esablishmen being relocaed abroad. Implici in his assumpion is ha a cerain number of producion jobs need o be carried ou in close proximiy o one anoher. Thus, in order o relocae jobs abroad a firm would need o offshore he enire esablishmen. However, i is also possible ha offshoring akes a more coninuous form in which cerain jobs wihin an esablishmen are relocaed abroad bu he esablishmen iself does no close. This form of offshoring may be imporan bu would no be capured by he ne birh rae variable. Forunaely, SUSB also provides daa on he number of esablishmens who expand and conrac employmen in a given year. Thus, i is possible o consruc he ne esablishmen expansion rae: ne _ expansion_rae es _ expanding = es _ conracing es Equaion (8) will be re-esimaed wih his new dependen variable. The model predics ha low skilled immigraion reduces he need o offshore and hus will lead o an increase in he ne expansion rae. High skilled immigraion should increase offshoring and hus reduce he ne expansion rae. The resuls in Table 6 are consisen wih hese predicions. In he IV regressions (Column 2), we see ha low skilled immigraion leads o an increase in he ne expansion rae while high skilled immigraion leads o a decrease in he ne expansion rae. A one percenage poin increase in he share of low skilled 20

20 immigrans leads o a 0.21 percenage poin increase in he ne expansion rae while a one percenage poin increase in he share of high skilled immigrans leads o 0.43 percenage poin decrease in he ne expansion rae. Like he analysis using he ne birh rae, here is a sharp conras beween he OLS and IV resuls which suggess ha endogeneiy is biasing he OLS resuls. Overall, he resuls in Table 6 are consisen wih he predicions of he model and provide furher evidence ha firms offshoring decisions depend in par on local labor supplies. 6. Conclusion Research generally focuses on how immigraion affecs naive workers, while he impac of immigraion on domesic firms is ofen overlooked. This paper presens a simple model ha highlighs how immigrans are absorbed ino local labor markes eiher hrough changes in he relaive wage or hrough changes in he firms organizaional srucure. The empirical analysis provides hree imporan resuls. Firs, consisen wih previous findings, his paper shows ha immigraion has a minimal impac on relaive wages. Second, immigraion has a significan effec on he organizaional srucure of firms boh a he exensive and inensive margins. Low skilled immigraion increases he esablishmen birh rae and expansion rae while high skilled immigraion decreases he esablishmen birh rae and expansion rae. Togeher hese resuls provide compelling evidence ha labor supply shocks due o immigraion are absorbed hrough changes in he firms organizaional srucure raher han hrough adjusmens in relaive wages. Third, offshoring is a key facor in explaining he relaionship beween immigraion and he ne esablishmen birh rae. Admiedly he ne birh rae is no a perfec proxy for 21

21 offshoring, however addiion resuls show ha he impac of immigraion on he ne birh rae is sronges in indusries in which offshoring is more common and among firms ha are more likely o offshore. The resuls of his paper improve our undersanding of immigraion and offshoring. Firs, hese resuls may explain why research finds ha immigraion has a relaively small impac on he wages of naive workers. Raher han relaive wages changing in response o immigraion, firms may adjus he locaion of heir producion aciviies in order o absorb labor supply shocks. Second, here is evidence ha low skilled immigraion and offshoring are subsiues. Thus, a policy ha resrics low skilled immigraion may lead o domesic firms offshoring more jobs. Eiher foreign workers migrae o he U.S. o perform hese jobs domesically or he jobs are relocaed abroad o be performed by foreign workers in heir naive counry. A he very leas any immigraion policy should ake ino consideraion he impac hese regulaions may have on he organizaional srucure of domesic firms. 22

22 REFERENCES: Acemoglu, Daron (1998), Why Do New Technologies Complemen Skills? Direced Technical Change and Wage Inequaliy, Quarerly Journal of Economics, 113, pp Acemoglu, Daron (1999), Changes in Unemploymen and Wage Inequaliy: An Alernaive Theory and Some Evidence, American Economic Review, 89, pp Amii, Mary and Shang-Jin Wei (2006), Service Offshoring and Produciviy: Evidence from he Unied Saes, NBER Working Paper No. W Barel, Ann P. (1989), Where Do he New U.S. Immigrans Live? Journal of Labor Economics, Vol. 7, No. 4, pp Beaudry, Paul and David A. Green (2000), Wages and Employmen in he Unied Saes and Germany: Wha Explains he Differences? American Economic Review, 93, pp Bernard, Andrew B. and J. Bradford Jensen (2007), Firm Srucure, Mulinaionals, and Manufacuring Plan Deahs, The Review of Economics and Saisics, Vol. 89, No.2, pp Blinder, Alan S. (2005), Fear of Offshoring, CEPS Working Paper No Borjas, George J. (1994), The Economics of Immigraion, Journal of Economic Lieraure, December 1994, pp Card, David and John DiNardo (2000), Do Immigran Inflows Lead o Naive Ouflows? American Economic Review Papers and Proceedings, Vol. 90, No. 2, pp Card, David and Ehan G. Lewis (2005), The Diffusion of Mexican Immigrans During he 1990s: Explanaions and Impacs, NBER Working Paper No , Augus Card, David (1990), The Impac of he Mariel Boalif on he Miami labor Marke, Indusrial and Labor Relaions Review, January 1990, pp Card, David (2001), Immigran Inflows, Naive Ouflows, and he Local Marke Impacs of Higher Immigraion, Journal of Labor Economics, Vol. 19, No. 1, pp

23 Card, David (2005), Is he New Immigraion Really so Bad? The Economic Journal, vol. 115 (November), pp. F300-F323. Friedberg, Rachel M. and Jennifer Hun (1995), The Impac of Immigrans on Hos Counry Wages, Employmen and Growh, Journal of Economic Perspecives, Spring 1995, pp Grossman, Gene M. and Eseban Rossi-Hansberg (2008), Trading Tasks: A Simple Theory of Offshoring, NBER Working Paper #W12721 (forhcoming in AER), January 22, Hanson, G., K. Scheve, M. Slaugher, and A. Spilimbergo (2002), Immigraion and he U.S. Economy: Labor-Marke Impacs, Illegal Enry, and Policy Choices, Boeri, T., Hanson, G., McCormick, B. (Eds), Immigraion Policy and he Welfare Sysem, Oxford Universiy Press. Jensen, J. Bradford and Lori G. Klezer (2006), Tradable Services: Undersanding he Scope and Impac of Service Offshoring, in Susan M. Collins and Lael Brainard, eds., Brookings Trade Forum 2005, Offshoring Whie-Collar Work. Jones, Ronald W. (2005), Immigraion vs. Ousourcing: Effecs on Labor Markes, Inernaional Review of Economics and Finance, pp Lewis, Ehan G. (2003), Local, Open Economies Wihin he US: How Do Indusries Respond o Immigraion? Federal Reserve Bank of Philadelphia Working Paper No. 04-1, December Lewis, Ehan G. (2005), Immigraion, Skill Mix, and he Choice of Technique, Federal Reserve Bank of Philadelphia Working Paper # Navarei, G. B., G. Berola, and A. Sembenelli, (2008), Offshoring and Immigran Employmen: Firm-level Theory and Evidence, CEPR Discussion Paper no Olney, William W. (2009), Offshoring, Immigraion, and he Naive Wage Disribuion, Universiy of Colorado mimeo. Ruggles, Seven, Mahew Sobek, Tren Alexander, Caherine A. Fich, Ronald Goeken, Paricia Kelly Hall, Miriam King, and Chad Ronnander. Inegraed Public Use Microdaa Series: Version 4.0 [Machine-readable daabase]. Minneapolis, MN: Minnesoa Populaion Cener [producer and disribuor],

24 Figure 1 Low Skilled Immigran Share ( Average) Fracion L_Img MSA wih Highes Share: L_Img Visalia-Tulare-Porerville, CA Miami, FL Salinas, CA Los Angeles, CA El Paso, TX MSA wih Smalles Share: L_Img Canon, OH Lafayee, LA Saginaw-Bay Ciy-Midland, MI Charleson, WV Jackson, MS High Skilled Immigran Share ( Average) Fracion H_Img MSA wih Highes Share: H_Img Miami, FL San Jose, CA For Lauderdale, FL San Francisco, CA Samford, CT MSA wih Smalles Share: H_Img Youngsown-Warren, OH Topeka, KS Lake Charles, LA York, PA Pueblo, CO Ne Esablishmen Birh Rae ( Average) Fracion Ne_Birhs MSA wih Highes Share: Ne Birhs Las Vegas, NV Provo-Orem, UT Fayeeville-Springdale, AR For Myers-Cape Coral, FL Orlando, FL MSA wih Smalles Share: Ne Birhs Flin, MI Youngsown-Warren, OH Saginaw-Bay Ciy-Midland, MI Decaur, IL Souh Bend, IN

25 Figure 2 Low Skilled Immigran Shares Acual Immigran Share Prediced Immigran Share High Skilled Immigran Shares Acual Immigran Share Prediced Immigran Share 26

26 Table 1 Firs Sage Regressions L_Img H_Img Prediced L_Img 0.404*** 0.090*** [0.038] [0.022] Prediced H_Img *** [0.071] [0.040] Observaions R-squared F-Sa, Insrumens Sandard errors in brackes. * significan a 10%; ** significan a 5%; *** significan a 1%. All regressions include he mean age of naives, he educaional aainmen of naives, and he share of he naive populaion ha is male, black, asian, and hispanic. Table 2 Naive Relaive Wages OLS L_Img IV [0.178] [0.856] H_Img 0.581** [0.245] [1.780] Observaions Robus sandard errors in brackes. * significan a 10%; ** significan a 5%; *** significan a 1%. All regressions include he mean age of naives, he educaional aainmen of naives, and he share of he naive populaion ha is male, black, asian, and hispanic. 27

27 Table 3 Ne Birh Rae OLS L_Img 0.019*** 0.113** IV [0.007] [0.051] H_Img 0.023* *** [0.013] [0.093] Observaions Robus sandard errors in brackes. * significan a 10%; ** significan a 5%; *** significan a 1%. All regressions include he mean age of naives, he educaional aainmen of naives, and he share of he naive populaion ha is male, black, asian, and hispanic. Table 4 Ne Birh Rae by Offshoring and Non-Offshoring Indusries (IV Regressions) All Indusres Non-Offshoring Indusries Offshoring Indusries L_Img 0.113** 0.077* 0.140** [0.051] [0.042] [0.063] H_Img *** ** *** [0.093] [0.076] [0.117] Observaions Robus sandard errors in brackes. * significan a 10%; ** significan a 5%; *** significan a 1%. All regressions include he mean age of naives, he educaional aainmen of naives, and he share of he naive populaion ha is male, black, asian, and hispanic. 28

28 Table 5 Ne Birh Rae by Firm Size (IV Regressions) Toal Employmen < Employmen < Employmen L_Img 0.113** 0.091* * [0.051] [0.049] [0.074] [0.076] H_Img *** ** ** ** [0.093] [0.090] [0.135] [0.144] Observaions Robus sandard errors in brackes. * significan a 10%; ** significan a 5%; *** significan a 1%. All regressions include he mean age of naives, he educaional aainmen of naives, and he share of he naive populaion ha is male, black, asian, and hispanic. Table 6 Ne Expansion Rae OLS L_Img 0.044*** 0.207** IV [0.013] [0.093] H_Img ** [0.027] [0.174] Observaions Robus sandard errors in brackes. * significan a 10%; ** significan a 5%; *** significan a 1%. All regressions include he mean age of naives, he educaional aainmen of naives, and he share of he naive populaion ha is male, black, asian, and hispanic. 29