SALARY DIFFERENTIALS BETWEEN NATIVE-BORN AND FOREIGN-BORN FACULTY

Transcription

1 Salary Dieretials betwee Native-bor ad Foreig-bor Faculty SALARY DIFFERENTIALS BETWEEN NATIVE-BORN AND FOREIGN-BORN FACULTY Recai Aydi, Police Academy, Akara, Turkey ABSTRACT Numerous studies have ivestigated the earig dieretials ad potetial salary discrimiatio with respect to geder, race ad uio status across academic aculty i the Uited States. However, there are very ew studies i the literature ivestigatig the compesatio structures o ative-bor aculty versus oreig-bor aculty. This paper uses the most recet survey o the Natioal Study o Post Secodary Faculty (NSOPF) coducted by the Natioal Ceter or Educatio Statistics i 2004 amog over 26,000 aculty ad sta members. The widely used Blider-Oaxaca decompositio shows that there is a 3.5% total salary dieretial i avor o oreig-bor aculty while characteristics accout or +7.1% ad uexplaied actors, which is geerally attributed to discrimiatio, accout or 3.6 % (whe the decompositio is based o ative-bor aculty salary structure). This implies that the oreig-bor aculty are uderpaid as compared to atives. Further ivestigatio o the data reveals that this gap is maily due to relatively better characteristics o oreig-bor but aturalized aculty who represet over 55% o the overall oreigbor aculty sample. JEL Classiicatios: J3, J7, Q4 Keywords: Earig dieretials, Faculty salaries, Foreig-bor aculty INTRODUCTION The Uited States has bee a coutry o immigrats ad practiced a ope door policy rom coloial times util the late 19 th cetury with ew exceptios to certai ethicities i some periods. I 1909, the U.S. Immigratio Commissio Survey o over a hal-millio wage ad salary earers reported that 60 percet o the workers i mauacturig ad miig were oreig-bor ad evetually this idig helped to itroduce literacy requiremets i A quota system based o atioal origi took eect i 1929 i a eort to ease public tesio over the alleged depressig impact o the ilux o immigrats o earigs o atives. Followig the 1965 Immigratio ad Natioality Act, the Uited States abolished the atioal origi based quota system ad replaced it with kiship-based immigratio as the primary criterio used to ratio residecy visas. Fially, the Immigratio Act o 1990 expaded the legal immigratio chaels by establishig diversity ad employmetbased immigratio categories with the itetio o attractig skilled ad taleted workers. The Act also implemeted the H1-B specialty occupatio visa or temporary workers employed i certai jobs which require theoretical ad practical applicatio o a body o highly specialized kowledge alog with at least a bachelor's degree or its equivalet. Specialties such as architecture, egieerig, mathematics, 89

2 Southwester Ecoomic Review physical scieces, medicie ad health, educatio, ad busiess have qualiied uder the H1-B provisio. The reality o a cotiuous low o immigrats as a result o legal ad illegal chaels has caused a o-goig debate o whether immigrats are depressig the wages o atives. To deal with growig cocer about the icreasig amout o legal migratio as a result o the 1990 act, Cogress added various other provisios regulatig the employer-based immigratio ad H1-B temporary work visa i 1991 such as the requiremet o a Labor Coditio Applicatio (LCA). The LCA is a attestatio by a employer seekig to hire a oreig worker that the employer is payig the worker at least the higher o the actual wage paid by the employer to others i the same occupatio with similar experiece ad qualiicatios or the prevailig wage 1 or the occupatio i the geographical area o the work site ad that the employmet will ot adversely aect the workig coditios o similarly employed ative workers. The Immigratio Acts o 1965 ad 1990 altered the traditioal compositio o ew immigrats to the U.S. Due to atioal origi based quotas prior to the 1965 Act, immigratio rom Europea coutries was ecouraged while idividuals rom third world coutries were discouraged. Betwee 1965 ad 1990, amily-sposored immigratio icreased the diversity amog the ew immigrats, the majority o whom beloged to lower socio-ecoomic classes o their societies. The ew system beeited the ormerly disadvataged Asia ad Lati America populatio i the U.S ad eabled these groups to icrease their share i total immigratio. The, the 1990 Act added aother elemet by allowig highly taleted ad skilled oreigers to immigrate to the U.S. through employmet. Today, it is well kow that the curret immigrats i the U.S. are more likely to be i either the lower or upper tail o prestige i their occupatio (Tog, 2006). The cotributio oered by this study is to provide a detailed aalysis o salary dieretial with respect to birth ativity ad citizeship by employig the most recet atioal level data available. This study ids that there is a +3.5% total salary dieretial i avor o oreig-bor aculty o which +7.1% accout or edowmet diereces ad 3.6 % accout or uexplaied actors, which are geerally attributed to discrimiatio (whe the decompositio is based o ative-bor aculty salary structure). This implies that oreig-bor aculty are ot paid as much as they are expected to be paid give their characteristics compared to atives. Further ivestigatio o the data reveals that this gap is maily due to relatively better characteristics o oreig-bor but aturalized aculty who represet over 55% o the overall oreig-bor aculty sample. LITERATURE REVIEW Extesive research ito the ative versus immigrat earigs dieretial i the U.S. labor market has bee coducted. However, scholars are ar rom a cosesus o the magitude o the gap i earigs, as well as the speed o covergece i earigs. I oe o the pioeerig work i this area, Chiswick (1978 ad 1980) ids that earigs o immigrats who are ot reugees ted to catch up to ad subsequetly surpass those o the ative-bor at about 10 to 20 years i the Uited States whe other variables, icludig race ad ethicity, are the same. Chiswick (1978) poits that immigrats have relatively high earig growth i this period also otes that the legal status o a oreig-bor perso whether a alie or 90

3 Salary Dieretials betwee Native-bor ad Foreig-bor Faculty aturalized citize, has o eect o earigs ater stayig i the coutry or a certai umber o years. He cocludes that oreig-bor workers with greater ability, work motivatio, ad ivestmets i huma capital through educatio ad traiig oset whatever earigs disadvatages they started with. The iitial disadvatage could have resulted rom discrimiatio agaist them or rom havig less kowledge ad skills relevat i U.S. labor markets. Duleep ad Dowha (2002) coirm Chiswick s idigs ad id that immigrat cohorts geerally show higher earigs growth tha ative cohorts or a major subset o immigrat ad ative populatio usig logitudial data coverig the period betwee rom 1960 through O the other had, ollowig the Immigratio o Act o 1990, Daeshvary (1993) examies the relative earigs o certai immigrats with at least a college degree ad ids approximately 7 percet uexplaied earigs dieretial i avor o atives ater cotrollig or the diereces i huma capital characteristics. Borjas (1985) also challeges Chiswick s idigs o a empirical ad theoretical basis. He argues that Chiswick s estimates might be ureliable sice variatio i immigrat quality caot be detected by cross-sectioal data, as oradom emigratio by immigrats will bias cohort-based aalyses. Usig time series data, Borjas ids that immigrats iitial wages have declied over time. 2 Usig logitudial data, Hu (2000) suggests that the rate o growth o immigrat earigs was overstated i Cesus-based studies ad the worseig o immigrat earigs or more recet arrival cohorts is steeper tha previously suggested. 3 Aother potetial problem with the Chiswick (1978 ad 1980) studies is that they use the data rom 1970 cesus i which most o the oreig-bor idividuals immigrated to U.S. based o atioal origi quotas. It is ot surprisig to id that these immigrats who were mostly white ad have Europea origis could catch up with atives over time. Moreover Chiswick, Le ad Miller (2006) report that adult male immigrats had hourly earigs i 2000, o average, aroud 17 percet less tha the ative-bor males. Ater cotrollig or the diereces betwee the birthplace groups i these immigrats productivity-related characteristics, this gap i mea hourly earigs arrowed, but still remaied at aroud 8 percetage poits amog recet arrivals. Chiswick, Le ad Miller (2006) also report that, i the U.S. labor market, whe other actors are held costat, immigrats rom Eglish-speakig coutries had mea hourly earigs aroud 12 percet greater tha the ative bor. I compariso, immigrats rom o-eglish-speakig coutries had mea hourly earigs aroud 12 percet less tha that o ative-bor workers. This idig is i lie with the immigratio literature that suggests that laguage is the key actor i determiig the assimilatio o immigrats ito the host coutry both socially ad ecoomically. Chiswick ad Miller (1995) show that immigrats i the U.S. who are proiciet i Eglish have earigs about 17 percet higher tha immigrats with limited Eglish skills, keepig other measured variables the same. Bleakley ad Chi (2004) report a sigiicat positive eect o Eglishlaguage skills o wages amog idividuals who had immigrated to the Uited States as childre. Bleakley ad Chi id that the estimated eect usig a istrumetal variable approach is greater i magitude tha that suggested by regressio strategies that do ot address edogeeity ad measuremet error. 4 They coclude that there is evidece o substatial dowward bias i the OLS estimate due to measuremet error ad somewhat smaller upward bias due to edogeeity. As summarized above, scholars have icreasigly become iterested i explorig ad uderstadig the earigs gap betwee atives ad immigrats or the 91

4 Southwester Ecoomic Review last three decades. However, the literature cotais very ew studies that examie the earigs gap i sigle labor market by employig disaggregated data rather tha cesus data. This study examies the compesatio structure ad potetial earig dieretials betwee the ative-bor ad oreig-bor aculty as well as citize ad ocitize aculty i the U.S. academic labor market. There is a cosesus i the literature that widespread earigs discrimiatio (uexplaied dieretials) exists betwee male ad emale aculty. 5 Some studies also claim that similar dieretials ca be oud betwee white ad o-white aculty as well as uioized ad o-uioized aculty. 6 However, to the author s kowledge oly two studies, Moks ad Robiso (2000), ad Lopez ad Mora (2006), examie the issue o earigs dieretial betwee ative-bor ad oreig-bor aculty. Moks ad Robiso (2000) ids a sigiicat earigs pealty or beig a aturalized citize or ocitize versus ative bor citize. Their study primarily ocused o geder ad racial diereces. They use the data rom the 1999 NSOPF survey while this study uses the most recet atioal level aculty data rom 2004 NSOPF survey. Moks ad Robiso s idigs, i terms o magitude o earig dieretials o citizes by birth, aturalized citizes ad ocitizes are geerally i lie with this study though there are some major diereces i terms o directios o the gap. Lopez ad Mora (2006) also employs the 1999 NSOPF survey ad id that ative-bor Hispaics ear sigiicatly more o average tha their o-hispaic white couterparts, but oreig-bor Hispaics ear the same as o-hispaics. However, the Lopez ad Mora study has some major methodological issues i terms o variables choices such as icludig part-time aculty i the sample. This caused a overestimate o the earig gap betwee Hispaics ad other aculty as well as withi the Hispaic aculty with respect to birth place. DATA AND METHODOLOGY This study employs the NSOPF-04 data collected by Natioal Ceter or Educatio Statistics (NCES) rom a atioally represetative sample o 35,000 aculty ad istructioal sta at 1,080 istitutios. About 26,100 aculty ad academic support sta completed the survey, either usig a sel-admiistered olie questioaire or via telephoe iterviews durig the all semester o NSOPF-04 is the ourth cycle o data collectios o postsecodary aculty coducted by the NCES. Previous collectios were coducted i 1987, 1992, ad As NCES has improved its data collectio methods, NSOPF-04 is superior to the earlier sets i terms o variable coverage ad quality or a detailed ecoomic aalysis. NSOPF-04 provides detailed iormatio o compesatio, research productivity, educatioal attaimet, geder, ethicity, atioal origi ad aculty status. For purposes o this study, the sample size is restricted to exclude may o the variables as well as some o the observatios. The deletios iclude aculty without rak iormatio (393 observatios) ad survey participats with o aculty status (1,676 observatios). Part-time aculty was also deleted (7,284 observatios) as part-time aculty pay is ot comparable i most istitutios. I additio, aculty with o degree (53 observatios) idicated were also deleted as well as the aculty who respoded egatively to the citizeship status questio (27 observatios) but reported that they were ative-bor. These 27 idividuals were deleted due to icosistecy i their resposes sice every ative-bor is automatically a citize. The ial sample 92

5 Salary Dieretials betwee Native-bor ad Foreig-bor Faculty used i this study cotais 16,207 aculty o which 13,158 are ative-bor citizes, 1,721 are aturalized citizes ad 1,328 are ocitizes. Secod, i order to cotrol or the istitutio quality per studet expediture ad studet-aculty ratio variables were created. However, due to high correlatio betwee these variables the studet-aculty ratio was dropped as a result o stepwise regressio results which idicated that per studet expediture had a higher explaatory power. Third, ew variables were created to deal with productivity measures. NSOPF surveys provide the iormatio about various measures o productivity such as reereed joural articles, o-reereed joural articles, books, chapters, art exhibitios, presetatios, patets ad copyrights. The value or importace o each o these cotributios, however, depeds o the disciplie. As the data set icludes heterogeeous disciplies, it might be icorrect to emphasize oe or some o these measures. Thereore, productivity idexes were created which are liear combiatios o all these various productivity measures by employig pricipal compoet aalysis. As NSOPF data provide both recet research output ad lietime research output, it was easible to create a idex or overall productivity ad recet productivity (umber o academic cotributios i last two years). Table 1 provides the descriptive statistics or some selected variables. It is worth otig that most o the variables are sigiicatly dieret across the groups suggestig that a sigle equatio is ot suitable to measure the salary dieretials because, usig a sigle regressio equatio or the whole aculty sample imposes a restrictio o coeiciets ad assumes that the impact o each characteristic o earigs should be the same. However, as it is well established i the literature, sigle equatio i earig dieretial studies is iappropriate whe the characteristics are sigiicatly dieret betwee the groups. Thereore the Blider-Oaxaca decompositio method was applied by employig separate regressios or each subset. Table 1 idicates that aturalized citizes ear o average $78,781, sigiicatly exceedig both atives ($70,685) ad ocitizes ($68,136), raisig the average earig or the overall oreig-bor aculty cohort. The majority o aturalized citizes ad ocitizes obtaied their highest degree i the Uited States (83 % versus 17% amog aturalized citizes ad 63% versus 37% amog the ocitizes). I additio, the majority o aturalized citizes received the graduate degree i the Uited States ad obtaied residecy through employmet-based immigratio chaels ad evetually became citizes 6 to 10 years ater their employmet. It is well kow that most ocitize aculty ollow the same path, as U.S. academic market cotiues to be the world leader i scietiic research ad iovatios. However, it is worth otig that the ocitize aculty set cotais three dieret types o idividuals; (1) U.S. educated graduates who stay i the Uited States ater the graduatio ad participate i the academic market uder the Optioal Practical Traiig (OPT) program, uder H1-B visa program or uder a employerbased gree card program, (2) oreig degree holders who come to the U.S uder the H1-B visa program or employer-based gree card program, (3) oreig aculty who come to U.S educatioal istitutios as visitig scholars uder H1-B or J exchage visa program. There are sigiicat diereces i the idividual ad istitutioal characteristics across the aculty sets which may be drivig the earig diereces oted above. Naturalized citizes have sigiicatly higher total research productivity 93

6 Southwester Ecoomic Review compared to atives ad ocitizes. However ocitizes catch ad eve exceed the aturalized citizes i recet productivity (withi the last two years) ad the dierece i total productivity ca be explaied by their lack o experiece as they are geerally youger ad ew to academia. Naturalized citizes have more experiece ad more likely to be teured ad to be ull proessors whereas ocitizes hold lower raks, have less experiece ad higher rate o doctoral degree while they cocetrate i uded research. Natives are more likely to be i teachig ad admiistratio, uioized, more balaced as to geder ad workig or istitutios with sigiicatly lower per studet expediture. Moks ad Robiso (2000) otes that NSOPF-99 data set reveals a clear geder ad racial segregatio by ield. This study ids a similar patter i NSOPF-04 data set i terms o ativity. Foreig-bor aculty are more likely to be i egieerig, atural scieces ad lie scieces while ative-bor aculty are more likely to be i educatio, health, art ad humaities disciplies. Table 2 presets i detail the distributio o the aculty sample used i this study with respect to disciplie, geder, race ad istitutio type. Oe iterestig poit is that oreig-bor proessors represet almost 50 percet o the Hispaic aculty ad a eve greater percetage o Asia aculty. The methodology ad variables used to coduct the aalysis o this study are rather stadard. A semi-logarithmic model is speciied i which the log o the sum o aual basic salary ad other icome rom the primary istitutio employed is the depedet variable. The earig equatio used i the computatios o the earigs gap is: Logicome i = 1 k = α + X i + Z (1) j = 1 j where the X i represet various persoal ad huma capital characteristics o aculty such as geder (male=1), marital status (sigle=1), race (white=1), educatioal attaimet measure (bachelors degree is the omitted category), productivity idex, rakig (ull proessor is the omitted rak), experiece (deied as the umber o years sice each respodet completed his/her highest degree), experiece-squared (iteded to capture the cocavity o the experiece-earigs proile), primary iterest area (teachig, research, cliical or admiistratio; ad research is the omitted area) ad teure status (teured=1). The Z j are dieret locatioal ad istitutioal characteristics such as per studet expediture, geographical locatio o the istitutio (west is the omitted regio ad big-city is the omitted populatio cotrol variable), type o degree grated by the istitutio (doctoral gratig istitutio is the omitted type) ad disciplies (busiess is the omitted disciplie). The Blider-Oaxaca (1973) model ca be expressed as: lw lw = β ( X X ) + ( β β ) X (2) where X ad X are vectors cotaiig meas o the variables or ative-bor ad oreig-bor aculty, while β ad β are vectors with the OLS coeiciet estimates or ative-bor ad oreig-bor aculty, respectively. The irst term o the right had side captures the wage dieretial due to dieret edowmets or characteristics o 94

7 Salary Dieretials betwee Native-bor ad Foreig-bor Faculty ative-bor ad oreig-bor aculty. The secod term captures the wage dieretial due to the dierece i coeiciets. The Blider-Oaxaca decompositio model i equatio (2) calculates the wage dieretial assumig that the irst group s wage structure, i this case atives, is the prevailig wage equatio. However, just by chagig the idex umbers i the equatio a dieret wage equatio is obtaied which uses the oreig-bor group as the reerece group i determiig the wage. The, the wage gap ca be decomposed i the ollowig way: lw lw = β ( X X ) + ( β β ) X. (3) I other words, equatio (2) implies that i the absece o ay type o discrimiatio, the ative-bor wage structure would prevail i the market, while equatio (3) assumes that it is the oreig-bor wage structure that would prevail i a odiscrimiatory eviromet. Neumark (1988) suggests that this idexig problem ca be solved by usig a weighted average o coeiciets (β*) istead o β ad β, where β* is the coeiciet vector expected to prevail uder a o-discrimiatory wage structure. This coeiciet vector ca be calculated usig the sample sizes o ative-bor ad oreig-bor aculty sets as weights. As explaied above, the Blider-Oaxaca decompositio method decomposes earigs dieretials ito two compoets; Explaied ad Uexplaied compoets. Explaied compoet is a portio that arises whe two compariso groups, o average, have dieret qualiicatios or characteristics (e.g., age or years o schoolig) whe both groups are treated the same way. Uexplaied compoet is a portio that arises whe oe group is treated more avorably tha the other group while both groups have the same qualiicatios or characteristics. Uexplaied compoet eect is requetly iterpreted as a measure o labor market discrimiatio. Some researchers thik that the Blider-Oaxaca decompositio method is ot suitable to measure the discrimiatio as it overestimates the degree o discrimiatio whe the earigs equatio ails to iclude all relevat variables measurig skills ad qualiicatios. Thereore, idig a sigiicat uexplaied compoet eect does ot ecessarily idicate discrimiatio but it may be due to uobserved diereces i productivity or other characteristics. O the other had, some other researchers thik that the Blider-Oaxaca decompositio method may uderestimate the degree o discrimiatio eect as they suggest that eve the explaied compoet o earigs dieretial may be the result o iheret discrimiatio i the labor market. For example, idig that wome or blacks (disadvataged groups) have less years o schoolig may itsel be a result o discrimiatio. Thereore, the impact o years o schoolig o earigs dieretials might have bee higher tha what Blider-Oaxaca decompositio method reveals. I spite o the drawbacks o the Blider-Oaxaca decompositio method, this method has bee requetly used i all types o earigs discrimiatio studies sice 1973 because o its simplicity ad lexibility. 95

8 Southwester Ecoomic Review 96 TABLE 1 DESCRIPTIVE STATISTICS OF FACULTY SUBSETS Variables Native-bor Foreig-bor Naturalized No-citize Aual Salary 70,685 74,144 78,781 68,136 (38,497) (40,051) (42,008) (36,512) Productivity Idex (29.8) (33.6) (36.5) (29.0) Recet Productivity (3.6) (4.3) (4.6) (4.0) Proessor (0.45) (0.44) (0.48) (0.35) Associate Proessor (0.42) (0.43) (0.45) (0.40) Assistat Proessor (0.43) (0.47) (0.42) (0.50) Istructor (0.34) (0.28) (0.28) (0.28) Teachig (0.47) (0.49) (0.49) (0.50) Research (0.31) (0.43) (0.38) (0.47) Admiistratio (0.32) (0.25) (0.28) (0.21) Teured (0.50) (0.49) (0.50) (0.45) Uio (0.41) (0.41) (0.43) (0.38) Doctoral Degree (0.49) (0.42) (0.45) (0.38) Fuded Research (0.46) (0.50) (0.49) (0.50) Male (0.49) (0.47) (0.48) (0.46) White (0.33) (0.50) (0.49) (0.49) Experiece (11.2) (10.6) (10.8) (8.7) Per Studet Expediture (70.8) (92.1) (91.1) (93.3) Egieerig (0.25) (0.35) (0.35) (0.35) Educatio (0.30) (0.21) (0.23) (0.19) Natural Scieces (0.28) (0.36) (0.35) (0.37) Sample Size 13,158 3,049 1,721 1,328 Table 1 Note: Mea values are highlighted ad Stadard Deviatios are i parethesis. Most o the variables are sigiicatly dieret across the subsets at 1% sigiicace level. For example, all the variables are sigiicat betwee atives ad oreig bor aculty except Uio, Sigle ad Associate Proessor variables ad Per Studet Expediture, New Productivity, Istructor ad Egieerig variables betwee aturalized citizes ad ocitizes.

9 Salary Dieretials betwee Native-bor ad Foreig-bor Faculty Variables TABLE 2 DISTRIBUTION OF FACULTY SUBSETS All Naturalized Male 9,677 7,643 2,034 1, Nativebor Foreigbor Nocitize Female 6,530 5,515 1, White 12,936 11,524 1, Black 1, Hispaic Asia 1, Other race Busiess ad Ecoomics Health 2,211 1, Educatio 1,450 1, Egieerig 1, Lie scieces 1,703 1, Natural scieces 1,608 1, Social Sciece 2,380 2, Humaities 2,292 1, Art 1, Other disciplies 1,206 1, Ph.D. gratig istitutio 8,359 6,446 1, Master s gratig istitutio 3,095 2, Bachelor s gratig 2,396 2, iassociate i i degree gratig 2,357 2, i i i Total Sample Size 16,207 13,158 3,049 1,721 1,328 The Blider-Oaxaca model has bee used i umerous pay equity studies ad decompositio results have bee reported repeatedly to prove or disprove the discrimiatio i various settigs. However, researchers usually ail to check the reliability o these idigs ad this practice raises a questio about the robustess o the idigs eve though the techique is available or a log time. Oaxaca ad Rasom (1994 ad 1998) show the appropriate methods to calculate the stadard errors or the decompositio compoets. Surprisigly, almost oe o the studies examiig geder or racial discrimiatio reported the stadard errors o their decompositio idigs despite the widespread use o the Blider-Oaxaca decompositio method. This study provides the stadard errors or all decompositio results calculated by the Oaxaca module available or Stata ad shows that most o the decompositio results are sigiicat at 1% level or better. EMPIRICAL RESULTS Table 3 presets the OLS regressio results o selected variables or the ull sample ad all subsets or compariso. Most o the reported ad ureported variables are statistically sigiicat at 1% ad all o the sigiicat oes have the expected sig. Ureported variables i Table 3 are istructor ad other titles such as lecturer, disability, sigle, cliical ad other duty, proessioal ad masters degree, type o 97

10 Southwester Ecoomic Review istitutio, degree type o the istitutio, size o the city ad most o the regios where the istitutio is located ad some o the disciplies. A sigle regressio or the whole aculty sample, assumig that the salary structure would have bee the same or all aculty, ids that the oreig-bor dummy is sigiicat ad egative, idicatig that the oreig-bor aculty are uderpaid. A careul examiatio o the separate regressios reveals that ew variables such as rak, uio membership ad size o the city where the istitutio located aect the ative aculty salaries more sigiicatly tha oreig-bor aculty salaries. Also, the pay reductio as a result o traditioally low-payig disciplies such as educatio seems much lower or the atives compared to oreigers i the same disciplies. O the other had, rak ad uio status do ot seem to aect the salaries o oreig-bor aculty sigiicatly. The retur rom uded projects, teure status ad admiistrative positios (or pealty or beig i teachig) is higher or oreig-bor aculty tha atives. The results also idicate that the retur o productivity seems to be the same across the groups. As Oaxaca (1973) ad Blider (1973) poited out these dieretials i the coeiciets are iterpreted as the source o uexplaied earigs dieretial or more commoly earigs discrimiatio. However discrimiatio itsel has a egative cootatio ad may be iterpreted as a result o the market participats itetios or behaviors. That is ot the positio o this study as this study maily iteds to id the dieretial without urther iterpretatio or explaatio as to why it might be happeig. It is worth otig that ew scholars suggested that immigrats, particularly Asias i the high-tech idustries, may be aced with a glass ceilig. 7 As poited i the data sectio, the distributio o Asia aculty betwee ative-bor ad oreig-bor aculty is ot proportioal as most o them are oreig-bor. I order to test the glass ceilig hypothesis a separate regressio was ru or Asia aculty. Iterestigly, there is o sigiicat dierece betwee ative-bor Asias ad oreig-bor Asias ater cotrollig or all other variables, but there is a substatial ad highly sigiicat dierece betwee aturalized citizes ad ocitizes amog Asias (as high as 30.8% total salary dieretial o which 18.5% is uexplaied). Table 4 presets the decompositio results or the regressio betwee ativebor ad oreig bor aculty. Table 5 presets the similar results or the regressio amog the oreig-bor aculty oly based o citizeship status. Table 4 suggests that there is 3.59% salary dieretial i avor o oreig-bor aculty. However, usig the ative equatio as the reerece equatio, explaied earigs dieretial which is the expected dierece i earigs as a result o diereces i edowmet is 7.20%. The uexplaied portio o the dierece could be explaied by laguage barriers 8, cultural ad social diereces which may adversely eect the adjustmet ad etworkig o oreig-bor aculty, or could eve be a orm o discrimiatio based o race hidde uder the atioality. This result suggests that the oreig-bor aculty actually are uderpaid compared to their ative-bor couterparts. There seems to be a 3.37% earig pealty or beig oreig-bor. As is show i Table 4 ad 5, the total dieretial stays same regardless o the reerece equatio but the distributio o the dieretial may slightly chage. 98

11 Salary Dieretials betwee Native-bor ad Foreig-bor Faculty TABLE 3 OLS REGRESSION RESULTS OF FACULTY SUBSETS Depedet Variable: Log (Aual Salary) Variables All Native-bor Foreig-bor Naturalized No-citize Foreig-bor 0.031*** (0.010) Productivity 0.002*** 0.002*** 0.002*** 0.002*** 0.003*** (0.0001) (0.0001) (0.0003) (0.0004) (0.0005) Associate Proessor 0.114*** 0.124*** 0.055** 0.056* (0.010) (0.011) (0.027) (0.031) (0.055) Assistat Proessor 0.128*** 0.138*** 0.071* (0.013) (0.014) (0.037) (0.044) (0.071) Teachig 0.101*** 0.115*** 0.070*** 0.107** (0.012) (0.013) (0.026) (0.036) (0.039) Admiistratio 0.117*** 0.095*** 0.191*** 0.131*** 0.272*** (0.015) (0.016) (0.040) (0.049) (0.072) Teured 0.105*** 0.095*** 0.157*** 0.154*** 0.126** (0.010) (0.011) (0.028) (0.033) (0.051) Uio 0.022** 0.030*** (0.010) (0.010) (0.024) (0.031) (0.039) Doctoral Degree 0.180*** 0.182*** 0.191** 0.264*** (0.019) (0.020) (0.077) (0.098) (0.130) Fuded Research 0.084*** 0.078*** 0.100*** 0.073*** 0.134*** (0.008) (0.009) (0.019) (0.025) (0.030) Male 0.085*** 0.084*** 0.092*** 0.078*** 0.123*** (0.007) (0.008) (0.020) (0.025) (0.032) White ** (0.009) (0.011) (0.018) (0.023) (0.029) Experiece 0.012*** 0.014*** (0.001) (0.001) (0.003) (0.004) (0.005) Experiece Squared *** *** ( ) ( ) ( ) ( ) (0.0002) Southwest 0.088*** 0.080*** 0.125*** 0.125** 0.141** (0.014) (0.016) (0.038) (0.048) (0.061) Midsize city 0.022*** 0.025*** (0.008) (0.008) (0.020) (0.025) (0.032) Large tow 0.070*** 0.071*** (0.016) (0.017) (0.044) (0.063) (0.063) Per Studet *** *** *** *** ( ) ( ) ( ) ( ) ( ) Egieerig 0.092*** 0.078*** 0.184*** *** (0.018) (0.020) (0.042) (0.052) (0.069) Educatio 0.204*** 0.172*** 0.366*** 0.282** 0.518*** (0.017) (0.018) (0.053) (0.064) (0.093) Natural Scieces 0.220*** 0.191*** 0.333*** 0.225** 0.495*** (0.017) (0.019) (0.041) (0.052) (0.068) Itercept *** *** *** ** *** (0.032) (0.034) (0.101) (0.129) (0.166) Adjusted R-squared Sample Size 16,207 13,158 3,049 1,721 1,328 Note:Stadard errors are i parethesis. *** idicates 1% level o sigiicace, ** idicates 5% level o sigiicace ad * idicates 10% level o sigiicace. 99

12 Southwester Ecoomic Review TABLE 4 DECOMPOSITION RESULTS Foreig-bor Faculty versus Native-bor Faculty Oaxaca 1 Oaxaca 2 Neumark Explaied Salary Dieretial (%) 7.20*** 6.69*** 6.78*** (1.34) (0.92) (0.88) Uexplaied Salary Dieretial (%) 3.37** 2.90*** 2.99*** (1.33) (1.05) (1.01) Total Salary Dieretial (%) 3.59*** 3.59*** 3.59*** (1.19) (1.19) (1.19) 100 Note: Stadard errors are i parethesis. *** idicates 1% level o sigiicace, ** idicates 5% level o sigiicace ad * idicates 10% level o sigiicace. Blider-Oaxaca1 decompositio estimates are based o the average salaries calculated rom oreig-bor aculty salary equatio coeiciets. Blider-Oaxaca2 decompositio estimates are based o the average salaries calculated rom ativebor aculty salary equatio coeiciets. Neumark decompositio estimates are based o the average salaries calculated rom weighted coeiciets rom both oreig-bor ad ative-bor aculty salary equatios. Table 5 suggests a 16.69% earigs dieretial betwee aturalized citize aculty ad o-citize aculty. Similar to Table 4, usig the aturalized aculty salary equatio as the reerece, 9.27% o this dierece is due to the diereces i huma capital ad the rest, 6.74% is due to ukow reasos. This dierece could be due to the lack o a chace or ull participatio i the market as people with visa or eve gree card caot be eligible or all the jobs, or istitutios may avor America citizes or various reasos. 9 A separate regressio usig total icome, which icludes cosultig ad other icome eared by aculty i additio to the basic ad supplemetal icome rom the mai istitutio as the depedet variable, produced somewhat dieret results i terms o overall salary dieretial. This model oud virtually o overall earigs dieretial but the decompositio results were i lie with the previous model. The ew model suggested that there was a +5.0% (p-value < 0.001) characteristics dierece i avor o oreig-bor aculty which could ot be relected i the earigs due to 5.0% uexplaied dierece (p-value < 0.001). I additio, separate regressios by rak, geder ad istitutio type were ru to test the sigiicace o uexplaied dierece. The overall salary dieretial is sigiicat at 1% level or oreig-bor associate ad assistat proessors (+4.6% ad +8.1% respectively) but ot sigiicat or ull proessors (3.9% ad p-value = 0.102). The Blider-Oaxaca decompositio results or raks also vary. Explaied or characteristic dierece is sigiicat or all raks ragig rom +4.9% or oreigbor associate proessors to +12.7% or assistat proessors while it is +7.5% or ull proessors. However, uexplaied dierece is oly sigiicat or assistat proessors at 1% level ad or ull proessors at 10% level.

13 Salary Dieretials betwee Native-bor ad Foreig-bor Faculty TABLE 5 DECOMPOSITION RESULTS Naturalized Faculty versus No-citize Faculty Oaxaca 1 Oaxaca 2 Neumark Explaied Salary Dieretial (%) Uexplaied Salary Dieretial (%) Total Salary Dieretial (%) 9.27*** 12.25*** 10.93*** (1.85) (2.35) (1.81) 6.74*** 3.90* 5.14** (2.31) (2.39) (2.13) 16.63*** 16.63*** 16.63*** (2.52) (2.52) (2.52) Notes: Stadard errors are i parethesis. *** idicates 1% level o sigiicace, ** idicates 5% level o sigiicace ad * idicates 10% level o sigiicace. Blider-Oaxaca1 decompositio estimates are based o the average salaries calculated rom aturalized aculty salary equatio coeiciets. Blider-Oaxaca2 decompositio estimates are based o the average salaries calculated rom ocitize aculty salary equatio coeiciets. Neumark decompositio estimates are based o the average salaries calculated rom weighted coeiciets rom both aturalized aculty ad o-citize aculty salary equatios. CONCLUSIONS The results o this study suggests that there are sigiicat salary diereces betwee oreig-bor ad ative-bor aculty as well as betwee aturalized oreigbor aculty ad ocitize aculty. Further aalysis reveals that the higher earigs o aturalized citizes are the mai source o discrepacy betwee the earigs o oreig-bor ad ative-bor aculty. For example the overall dieretial oud by this study is aroud +3.5% betwee oreig-bor ad ative-bor aculty but it is +10.8% betwee aturalized citizes ad ative-bor citizes ad -5.3% betwee ative-bor ad ocitize aculty. It is worth otig that most o the dieretial i irst two cases above is due to edowmet diereces (uexplaied dieretial is small ad ot statistically sigiicat with p-value > 0.25) while the situatio is opposite i the third case (this time explaied dieretial is statistically isigiicat with p-value > 0.80). Numerous aculty have sued public ad private uiversities ad colleges over geder based discrimiatio practices with modest results. Very ew lawsuits are iled each year regardig discrimiatio i earigs based o atioal origi i higher educatio. I these lawsuits typically courts rule i avor o the istitutio because aculty ail to establish the lik betwee salary ad promotio decisios ad their atioal origi. Eve though this study does ot ecessarily claim the existece o such discrimiatio i U.S higher educatio system, the results o this study are ot icosistet with this possibility. However, it should be oted that this study may suer rom missig variable bias sice the laguage skill is ot cotrolled i the data. Empirical results reveal that the extet o the earigs dieretial is substatially greater i teachig istitutios 101

14 Southwester Ecoomic Review compared to research istitutios. Laguage skill may explai this dieretial, as it is more relevat i teachig aculty s salary ad promotio decisios. This study also ids that sigiicat uexplaied salary dieretials exist withi academic raks ad by geder. The causes o those dieretials caot be idetiied rom this data set ad requires urther ivestigatio. REFERENCES Ashra, J The Iluece o Geder o Faculty Salaries i the Uited States, Applied Ecoomics 28 (7): Ashra, J., Aydi R., & Shabbir, T Geder Earig Dieretials i Academia: A Update ad Some New Fidigs. Joural o Busiess ad Ecoomic Perspectives 32 (2): Barbezat D. A History o Pay Equity Studies. New Directios or Istitutioal Research 115: Barbezat D. A Airmative Actio i Higher Educatio: Have Two Decades Altered Salary Dieretials by Sex ad Race? Research i Labor Ecoomics 10: Bleakley, H., & Chi, A Laguage Skills ad Earigs: Evidece rom Childhood Immigrats. Review o Ecoomics ad Statistics 86 (2): Blider A Wage Discrimiatio: Reduced Form ad Structural Estimates. The Joural o Huma Resources 8 (4): Borjas, G. J Assimilatio, Chages i Cohort Quality ad the Earigs o Immigrats. Joural o Labor Ecoomics 3 (4): Borjas, G The Ecoomics o Immigratio. Joural o Ecoomic Literature 32 (4): Borjas, G The Ecoomic Aalysis o Immigratio i O. Asheelter ad D. Card (Eds.), Hadbook o Labor Ecoomics, Amsterdam, Elsevier Sciece Publishig. Borjas, G The Labor Market Impact o High-Skill Immigratio. The America Ecoomic Review 95 (2): Byog-Shik, R Faculty Salary Revisited: Discrimiatio by Sex ad Race. Paper preseted at the Aual Meetig o the Associatio or the Study o Higher Educatio, Albuquerque, NM November 6-8, Chiswick, B. R The Eect o Americaizatio o the Earigs o Foreigbor Me. Joural o Political Ecoomy 86 (5): Chiswick, B. R Immigrats Earigs Patters by Sex, Race, ad ethic Groupigs. Mothly Labor Review 103 (10): Chiswick, B. R Speakig, Readig ad Earigs amog Low-Skilled Immigrats. Joural o Labor Ecoomics 9 (2): Chiswick, B.R., & Miller, P.W The Edogeeity betwee Laguage ad Earigs: Iteratioal Aalysis. Joural o Labor Ecoomics 13 (2): Chiswick B. R., & Miller, P. W Laguage ad Earigs amog Immigrats i the Uited States. Research i Labor Ecoomics 17: Chiswick, B. R., Le, A. T., & Miller, P. W How Immigrats Fare Across the Earigs Distributio: Iteratioal Aalyses. IZA Discussio Paper No Available at SSRN: 102

15 Salary Dieretials betwee Native-bor ad Foreig-bor Faculty Daeshvary, N Earigs Dieretials betwee Natives ad Immigrats with College Degree. America Ecoomist 37 (2): Duleep, H.O., & Doowha, D. J Isights rom Logitudial Data i the Earigs Growth o the U.S. Foreig-Bor Me. Demography 39 (3): Fog, Timothy P The Cotemporary Asia America Experiece: Beyod the Model Miority. Upper Saddle River, N.J.: Pretice Hall. Hu, W Immigrat Earigs Assimilatio: Estimates rom Logitudial Data. America Ecoomic Review 90 (2): Kassoudji, S. A Eglish Laguage Ability ad the Labor Market Opportuities o Hispaic ad East Asia Immigrat Me. Joural o Labor Ecoomics 6 (2): Lee, S. M. & Edmosto, B The Socioecoomic Status ad Itegratio o Asia Immigrats. I Immigratio ad Ethicity: The Itegratio o America s Newest Arrivals, edited by Barry Edmosto ad Jerey S. Passel. Washigto, D.C.: Urba Istitute Press. Lee, S. M Do Asia America Faculty Face a Glass Ceilig i Higher Educatio? America Educatioal Research Joural 39 (3): Lopez, M. H., & Mora, M. T The Earigs o U.S. ad Foreig-bor Hispaic Faculty, Marvasti, A U.S. Academic Istitutios ad Perceived Eectiveess o Foreig-bor Faculty. Joural o Ecoomic Issues 39 (1): McMaus, W., Gould W., & Welch, F Earigs o Hispaic Me: The Role o Eglish Laguage Proiciecy. Joural o Labor Ecoomics 1 (2): Moks, J Uioizatio ad Faculty Salaries: New Evidece rom 1990s. Joural o Labor Research 21 (2): Moks, J., & Robiso, M Geder ad Racial Earigs Dieretials i academic Labor Markets. Ecoomic Iquiry 38 (4): Oaxaca, R Male-Female Wage Dieretials i Urba Labor Markets. Iteratioal Ecoomic Review 14 (3): Oaxaca, R., & Rasom, M O Discrimiatio ad the Decompositio o Wage Dieretials. Joural o Ecoometrics 61 (1): Oaxaca, R., & Rasom, M Calculatio o Approximate Variaces or Wage Decompositio Dieretials. Joural o Ecoomic ad Social Measuremet 24 (1): Taier, E Eglish Laguage Proiciecy ad Earigs amog Foreig-bor Me. Joural o Huma Resources 23 (1): Tog, Y Birth Nativity, Citizeship, ad Geder Dierece o Immigrat Scietists/Egieers Earigs. Upublished. Dowloadable rom Toutkoushia, R Usig Citatios to Measure Sex Discrimiatio i Faculty Salaries. The Review o Higher Educatio 18 (1): Toutkoushia, R Racial ad Marital Status Diereces i Faculty Pay. The Joural o Higher Educatio 69 (5):

16 Southwester Ecoomic Review ENDNOTES 1. Employers were allowed to pay as low as 95% o the prevailig wage util New regulatios ow require that the worker must be paid at least 100% o the prevailig wage. 2. Despite this cocer, this study employs cross-sectioal data as oe ca saely assume that the cohort quality o upper tail immigrats such as the subject o this study, higher educatio aculty, is ot expected to chage dramatically over the time. 3. It should be oted that most o these studies use the cesus data despite the pitalls as poited out by Hu (2000). 4. See McMaus, Gould ad Welch (1983), Kassoudji (1988), Taier (1988) ad Chiswick (1991). 5. See Ashra (1996), Ashra, Aydi, & Shabbir (2006), Toutkoushia (1994). Also see Barbezat (2002) ad Byug-Shik (1997) or a excellet review o earigs discrimiatio literature i academia. 6. See Barbezat (1989), Toutkoushia (1998) ad Moks (2000). 7. See Lee ad Edmosto (1994), Lee(2002) ad Fog (1998). 8. Marvasti (2005) suggests that the perceptios about the liguistic ability o oreigbor proessors may create a prejudice ad may reduce the demad or oreig-bor aculty i certai istitutios hece lower the bargaiig power o them. 9. H1-B visa requiremets are very light ad the process is easy. I additio, the higher educatio istitutios are ot subject to the aual quota. Still, may higher educatio istitutios are reluctat to hire o-citize or o-residet aculty or ew graduates or various reasos. This automatically limits the umber o jobs available especially or the graduates rom relatively lower raked schools. O the other had, legal residets do ot have most o these limitatios eve though they still caot apply to the govermetal jobs as citizeship have icreasigly become a strict coditio or employmet i govermet agecies ater 9/