you legal?

Legal Status and Wage Disparities for Mexican Immigrants

Matthew Hall, University of Illinois-CbicagoEmily Greenman, Pennsylvania State UniversityGeorge Farkas, University of California-Irvine

This article employs a unique method of inferring the legal status of Mexican im-migrants in the Survey of Income and Program Participation to offer new evidenceof the role of legal authorization in the United States on workers' wages. We estimatewage trajectories for four groups: documented Mexican immigrants, undocumeneedMexican immigrants, U.S-born Mexican Americans and native non-Latino whites.Our estimates reveal a gross 17 percent wage disparity between documeneed and un-documented Mexican immigrant men, and a 9 percene documented-undocumentedwage disparity for Mexican immigrant women. When worker human capital andoccupation are held constant, these wage gaps reduce to 8 and 4 percent, respectively.We also find large differences in returns to human capital with undocumented Mexi-can immigrants having the lowest wage returns to human capital and having veryslow wage growth over time.

Over ehe pase 20 years, ewo apparenely coneradiceory changes have been eakingplace in ehe U.S. labor markee: ehe quaneiey and qualiey of jobs available eo low-skill workers have seeadily declined, and ehe number of low-skill immigranes eneer-ing ehe counery in search of employmene has seeadily increased (Waldinger andLicheer 2003). The increase in ehe number of immigranes from Mexico has beenespecially noeable. Given ehe large size of ehis seream and ehe face ehae economiccircumseances have become less favorable for low-skill workers, underseandingbow Mexican immigranes fare in ehe labor markee is of considerable ineerese.Research by sociologists and economises on labor markee ouecomes for low-skillimmigranes, including Mexicans, aeeeses eo ehe ineerese of social scieneiscs in chisqueseion. Yee one poeeneially crieical influence on ehe labor markee ouecomes ofehis group is rarely examined: legal seaeus. By considering ehe influence of legalseaeus, ehis research provides a more compleee underseanding of ehe faceors shap-ing Mexican immigranes' earnings in ehe U.S. labor markee.

As governmeneal effores eo prevene employers from hiring unauehorized immi-granes have increased, and social services and polieical privileges for undocumeneedimmigranes have been rolled back, ehe imporeance of legal scacus for immigranes'ouecomes is likely eo have increased. In order eo make clear comparisons beeweeneehnic groups and examine erends in immigranes' earnings over eime, ie is eherefore

This research acknowledges support Jrom the Population Research Institute at Pennsylvania StateUniversity, which receives core funding Jrom the National Institute of Child Health and HumanDevelopment (Grant R24-HD041025). We are grateful to the editor, three anonymous reviewers,TedMoutv and Deborah Graefefor their helpful comments. Direct correspondence to Matthetv Hall,1007 West Harrison St., MC312, Chicago, IL 60607. E-mail: [email protected]." Tha Univaisity ol North Caralina Social Forças 89121491..514. Decambar 2010

492 • Social Forces 89(2)

crucial to disaggregate immigrants by legal status (Massey and Bartley 2005). Failureto do so is to risk distorting the realities faced by Mexican immigrants in the labormarket or attributing the effects of legal status to other factors, such as cultural dif-ferences between ethnic groups or declines in immigrants' unobservable skills overtime. Further, other factors influencing the earnings of Mexican immigrants mayinteract with legal status in ways that are not yet understood. For example, returnsto human capital have been shown to be lower for low-skill Mexican immigrantsthan for other low-skill workers (Hall and Farkas 2008). It is possible that this find-ing is driven by the inability of unauthorized workers, who make up a much largerproportion of immigrants from Mexico than from most other sending countries, totranslate human capital into higher earnings. It is therefore necessary to consider therole of legal status when evaluating the effects of human capital and other workercharacteristics that may be correlated with legal status.

This research employs a unique method of imputing the legal status of Mexicanimmigrants included in the 1996-1999 and 2001-2003 panels of the Survey ofIncome and Program Participation. We then use this information to provide amore complete understanding of earnings disparities faced by Mexican immi-grants in the low-wage labor market. In particular, we focus on two broad ques-tions. First, does legal status matter for the wages of low-skill Mexican immigrants,and if so, does the high proportion of Mexican immigrants who are unauthorizedexplain the wage disadvantage that has been documented in previous studies?Second, are the returns to human capital similar for documented and undocu-mented Mexican immigrants?

BackgroundLegal Status and Earnings Disparities for Mexican ImmigrantsIt is well known that recently arrived immigrants have lower earnings, on average,than native-born workers with similar educational levels. Labor market disadvan-tages common among foreign-born workers include employers' lower valuationof both work experience and educational credentials obtained outside the UnitedStates, lack of English language skills and little knowledge about the U.S. labormarket. There is a substantial research literature on the earnings "assimilation"patterns of immigrant workers that debates the extent to which immigrants' earn-ings catch up with those of natives over time spent in the United States (Borjas1985; Chiswick 1978; Schoeni 1997). While it is expected that there will bedisparities in earnings between foreign- and native-born workers, Mexican im-migrants stand out for their particularly low earnings. The median earnings ofMexican and Central American immigrants who worked fulltime in 2003 wereonly $20,840, compared with $38,486 among European immigrants, $40,297among Asian immigrants, and $36,784 among native-born workers (U.S. CensusBureau 2004). Poverty rates are also disproportionately high among Mexican andCentral American immigrants, with nearly 25 percent living in poverty (compared

Legal Status and Mexican Immigrants' Wages • 49^1

with only 10 percent, 12 percent and 11 percent of European immigrants, Asianimmigrants and the U.S.-born population, respectively) (Census Bureau 2004).

The low average educational attainment of Mexican immigrants undoubtedlycontributes to their low earnings. More than 40 percent of Mexican immigrantsover age 25 have less than a 9''' grade education (Census Bureau 2004), a low levelof education that is extremely rare among the U.S.-born population (at only 3.7%).Schoeni (1997) estimated that this disparity in education is responsible for about athird ofthe wage gap between male native workers and Mexican immigrant work-ers, but found that a substantial wage differential remained unexplained even aftercontrolling education. For most other immigrant groups studied, Schoeni foundlittle or no unexplained earnings disadvantage after controlling for educationand age. He proposed the high proportion of undocumented immigrants in theMexican foreign-born population as one possible explanation for this difference.

This speculation appears to be well-founded. Both popular accounts and theacademic literature have portrayed undocumented immigrants as especially vul-nerable to discrimination and exploitation by employers (Massey 1987; Rivera-Batiz 1999). Employing undocumented immigrants makes it easier for employersto skirt regulations designed to protect workers, such as the minimum wage andovertime pay. Lack of documentation also weakens workers' bargaining positionrelative to employers, potentially resulting in lower wages. However, empiricalverification of the labor market disadvantages associated with undocumentedstatus has been limited. Donato and Massey (1993) demonstrated that the pas-sage ofthe Immigration Reform and Control Act in 1986, which introduced newpenalties for establishments employing undocumented immigrants, substantiallyincreased the wage penalties associated with undocumented status. Therefore, thefindings of studies based solely on data collected prior to 1986 can no longer berelied upon to indicate how undocumented immigrants are currently faring. Therelatively small number of post-IRCA studies have found evidence of substantialearnings penalties associated with undocumented status, on the order of 14-24percent for Mexican immigrants (Donato and Massey 1993; Kossoudji and Cobb-Clark 2002; Rivera-Batiz 1999).

Post-IRCA research on the effect of legal status on immigrants' earnings hasbeen limited, primarily due to data constraints. The sensitive nature of questionsabout legal status keeps them from being included in nationally representativesurveys that form the basis for most studies of labor market outcomes. Thus, mostexisting studies have relied on smaller, non-nationally representative samples.Donato and Massey (1993) utilized unique data, collected on both sides oftheborder, on migrants from 13 Mexican communities. They found substantial earn-ings penalties associated with being undocumented, but only after the passage ofIRCA. Donato et al. (2008), using a later, expanded version ofthe same dataset,found that legal status not only affected earnings but was also related to a greaterlikelihood of employment in the informal sector. The results also showed that


women's situation relative to men had deteriorated Following IRCA, increasingthe importance oF considering potential gender differences in the effects oF legalstatus. Most other previous studies have relied primarily on one data source-theLegalized Population Survey (Kossoudji and Cobb-Clark 2002; Powers and Seltzer1998; Powers, Seltzer and Shi 1998; Rivera-Batiz 1999). This survey was con-ducted among a sample oF immigrants who had applied For legalization underIRCA. The immigrants were surveyed twice, once prior to becoming legalized(in 1988) and once aFter legalization (in 1992). Researchers have taken advantageoF this unusual longitudinal data to attempt to estimate the causal effect oF thechange in legal status on earnings.

We believe that our use oF SIPP data presents significant advantages over theLPS. Although the LPS data are unique in that they directly identify undocu-mented immigrants, their usefulness is limited by the Fact that they do not containa comparison group of workers who did not experience a change in legal status.Rivera-Batiz (1999) addresses this problem by comparing Mexican immigrantworkers included in the LPS with similar Mexican immigrant workers from the1990 U.S. Census, who are assumed to be predominantly legal-a problematicassumption given the high proportion of Mexican immigrants who are undocu-mented. Kossoudji and Cobb-Clark (2002) use Latino immigrant and native menfrom the NLSY as a comparison group. While in this case the comparison sampleis not as potentially contaminated by containing a substantial share of workerswho are actually undocumented, it differs from the undocumented sample onmany characteristics other than legal status, such as nativity, length of residencein the United States and specific Hispanic ethnicity. By contrast, our study usesa single nationally representative data source to identify both documented andundocumented Mexican immigrants, as well as Mexican-American and whitenative comparison groups. We are thereFore able to more effectively estimate theunique influence oF legal status on workers' wages.

Returns to Human Capital

In addition to providing an improved estimate oF the effect oF undocumentedstatus on earnings, we explore the extent to which returns to human capital varyby legal status. Hall and Farkas (2008) analyzed differences by race and nativity inearnings returns to education among low-skill workers. They Found that oFall therace/nativity groups studied. Latino immigrants were the only group For whomeducation did not significantly raise earnings, and that returns to education wereparticularly low For Mexican and Central American immigrants. We suggest thatthe high proportion oF undocumented workers among low-skill Mexican immi-grants may contribute to their uniquely low returns to education.

Theories oF labor market segmentation, recently summarized by Hudson(2007), posit that the economy contains two distinct sets oFjobs: those at the

"core," which offer stability, good working conditions, higher pay and prospects For

Legal Status and Mexican Immigrants' Wages • 49!5

advancement; and those at the "periphery," which tend to be low-paying, "dead-end" jobs with few prospects for improvements in earnings, regardless of humancapital. Although the existence of such a clear demarcation has been challenged onempirical grounds, it is widely recognized that low-skill immigrant workers tendto be concentrated disproportionately in jobs witb many "peripberal" cbaracteris-tics. While Hudson (2007) found that many workers who start out in peripheraljobs eventually move on to better jobs, this may not be as true for undocumentedworkers, whose lack of legal authorization to work in the United States presentsa barrier to switching employers. The lack of internal opportunities for advance-ment in peripheral jobs, coupled with greater difficulty in moving to better jobs,may combine to depress wages among undocumented workers.

Even disregarding tbeories of labor market segmentation, any barrier to switch-ing employers should have the effect of depressing earnings growth: Topel andWard (1992) demonstrate that at least a third of early-career earnings growthis attributable to workers moving to better-paying jobs, rather than to internalraises or promotions. To the extent that their legal status restricts undocumentedworkers' ability to seek better jobs, their human capital is likely to be rewarded lesshighly in the labor market, even in the absence of discrimination by employers.

We test this possibility by exploring differences between undocumentedMexican immigrants and other groups (including botb documented Mexicanimmigrants and native white and Mexican-American workers) in wage returns toeducation and work experience.

Legal Status and GenderFinally, we recognize the possibility that the processes influencing the labor marketoutcomes of Mexican immigrants differ by gender. Mexican migration is knownto be a gendered process. Traditionally, men bave been more likely to be primarylabor migrants, while women have been more likely to accompany them or fol-low them several years later as dependents (Donato, Wagner and Patterson 2008).Women's relatively later arrival within the migration stream, their greater likeli-hood of joining family members ratber tban migrating alone, and their greaterresponsibility for cbildren are all factors that may make their earnings trajectoriesdifferent from those of men. Occupational and industrial segregation by gender,race and nativity is another factor that may play different roles for male and femalemigrants. While occupational segregation of Latino immigrants is a factor thataffects both men and women (see Catanzarite's (2000) work on "brown-collar" oc-cupations), jobs are further segregated by gender, witb immigrant Latina womensbuttled into a particularly narrow range of jobs (Myers and Cranford 1998).Because of these and other potential gender differences in the effect of migration-related factors on earnings, we model both the effect of legal status and the returnsto human capital variables separately for men and women.

496 • Social Forces 8m)

Data and Methods

Study SampleThis areicle analyzes monehly wage daea from ehe 1996-1999 and 2001-2003panels of ehe Survey of Income and Program Pareicipaeion. SIPP is a panel seudyfocusing on eracking U.S. workers' employmene and public program experiences.The SIPP design draws a naeionally represeneaeive sample of U.S. households andineerviews each household member every four monehs for ehree (2001-2003) eofour (1996-1999) years. Ae each ineerview, respondenes are asked wave-specificeopical queseions and a see of core queseions ehae cover ehe reference moneh andpreceding ehree monehs. Thus, in ehe 1996 panel, for a respondene who compleeedall 12 ineerviews, ehere were 48 monehs of observaeion. SIPP provides eranslaeorsfor respondenes wieh poor English skills.

SIPP is uniquely suieed for ehis seudy for several reasons. Firse, ehese daeaconeain a wealeh of informaeion on wages, work experience and educaeional ae-eainmene. Second, ehe combined samples of ehe 1996 and 2001 panels include alarge number of Mexican immigranes and Mexican American naeives. Third, un-like oeher large, naeionally represeneaeive daea sources, SIPP includes key variables,such as immigrane visa seaeus and pareicipaeion in public welfare programs, whichcan be used eo assess ehe legaliey of Mexican immigranes.

Because we are ineereseed in ehe wage erajeceories of low-skill workers, we limie oursample eo workers ages 18 eo 60 whose highese level of educaeion is no more ehan ahigh school degree (or ies equivalene). We reserice ehe sample eo men and women whoself-ideneify as eieher naeive-born non-Laeino whiee or of Mexican descene (naeiveor foreign-born). Using ehe ancesery daea in SIPP, naeive Mexican Americans (sec-ond or higher generaeion) refer eo U.S.-born individuals who ideneify as "Mexican,"

"Mexican-American" or "Ghicano." Mexican immigranes, by conerase, are individualswho were born in Mexico, bue migraeed and are currenely living in ebe Unieed Seaees.Naeiviey daea comes from a see of eopical queseions asked during wave 2 (monehsfive eo eighe); and inclusion in our sample is ehus resericeed eo ehose who compleeedehe "migraeion" module. We segmene individual longicudinal records inco a series ofperson monehs, one person moneh for each moneh of daea coUeceed.

Study VariablesThe dependene variable in ehis analysis is respondenes' currene wage raee. Ourmeasure of wage raee corresponds eo ehe raee of pay of ehe respondene's primaryjob. For workers who are employed on an hourly basis, wage raees are reporeedby respondenes. Among workers who repore a pay unie oeher ehan "per hour," weeseimaee wage raees based on monehly personal earned income from ehe respon-dene's primary job and ehe number of hours worked in a given moneh ae ehae job.Because ehis procedure of eseimaeing wage raees occasionally resules in exeremevalues, we see lower and upper bounds of $1 and $250 per hour, respeceively.'

Legal Status and Mexican Immigrants' Wages • 497

We include two measure of individual human capital in our models: educa-tional attainment and work experience. Educational attainment is measured asyears of schooling completed (up to 12 years). Full employment histories arenot available in SIPP; therefore, we use "potential" work experience, defined asage (in years) minus educational attainment plus 6 [age-(education + 6)], as ourmeasure of time in the labor market. Unfortunately, language ability is unavailablein the 1996 SIPP panel, and combining the 1996 and 2001 panels is necessaryfor suflScient sample sizes.̂ However, research has shown that English skills havelittle effect on wages after controlling for educational attainment and time in theUnited States (Bleakley and Chin 2004). Thus, for models specific to immigrants,we include a term for age upon arrival in the United States, which in conjunc-tion with work experience controls for length of time in the country. To accountfor differences in occupation, we include terms indicating employment in oneof six Bureau of Labor Statistics-defined aggregated occupational sectors (basedon the 1990 classification scheme): managerial and professional; technical, salesand administrative; service; farming, forestry and fishing; production, craft andrepairers; and operatives and laborers.

Inferring Legal StatusThe key independent variable in our study is legal status. As with most surveys,undocumented status is not measured directly in SIPP. However, with knowledgeof Mexican migration patterns derived from earlier research and data from theDepartment of Homeland Security, it is possible to use information available inSIPP to infer legal status for Mexican immigrants. (See Appendix Figure 1 for aschematic presentation of the imputation strategy.) SIPP gathers information onwhether sampled immigrants are naturalized citizens or legal permanent residents.(See Appendix 2 for survey questions). We classify such immigrants as legal. If animmigrant "personally," as opposed to "dependently" (i.e., via a child's eligibility),receives federal welfare benefits (e.g.. Food Stamps, Medicaid, SSI, TANF), forwhich undocumented immigrants are not eligible, he or she is also classified aslegal. Remaining immigrants are either undocumented or have visas falling intoone of several categories: refugees and asylees, students and exchange visitors, tour-ist and business travelers, temporary workers, and diplomats and other politicalrepresentatives (U.S. Department of Homeland Security 2007). Tourists and othershort-term visitors are not sampled by SIPP and, historically, very few Mexicanshave been granted asylum in the United States. Those enrolled in school areexcluded from the sample, but to be certain that we are not mis-categorizing thelegal spouses of immigrants on student visas, we consider those with an immigrantspouse enrolled in college fulltime (student visas are contingent on full-time en-rollment) as legal. To account for those admitted as diplomats, we deem Mexicanforeigners that report being or married to "high-ranking public officials" to belegal. The only group of temporary migrants that we are unable to directly infer is

498 • Social Forces 89[2)

temporary workers. However, authorized temporary workers form a small minor-ity of Mexicans in the United States (see U.S. Department of Homeland Security2007). Nonetheless, readers should keep in mind that the group we refer to as

"undocumented" may include a small number of authorized temporary workers.

MethodsTo best utilize the longitudinal nature ofthe data and to model both initial wagesand wage growth aspects of workers' wage trajectories, we employ hierarchicallinear modeling techniques (Raudenbush and Bryk 2002), where both the in-tercepts and the slopes are a function of time-invariant, person-specific charac-teristics. More specifically, we estimate random-coefficient growth curve models(Greene 2000) for a pooled sample of Mexican immigrants, and separately forMexican documented immigrants, Mexican undocumented immigrants, and na-tive Mexican Americans and non-Latino whites. Our pooled Mexican immigrantgrowth curve models are specified as follows:

ßo = ïoo + ïoh^l-i + /oW^C'îfto«. + ïg^experience. + y^/irrival. + y^^. + 7gJ>anel. +ftg.ßi = 7,0 + / ; / % í + ï,2«ducation. + y^^experience. + y,/trrival. + y^^. + y,J>anel. + fi,.

where t refers to chronological month and j represents individuals. Each value oft corresponds to the number of months elapsed since first interview. There are,depending on the panel, 48 (1996 panel) or 36 (2001 panel) possible values oft.Ln(wages)^. represents logged hourly wages of respondent j at month t. ßg\s the in-tercept and ßj is the wage growth rate, which are both a function of person-specificcharacteristics legal, education, experience, arrival, Z znd panel. Legal is a dummyvariable representing the legal status of Mexican immigrant j (documented = 1 ;undocumented = 0); education, refers to years of schooling for person j ; experience.represents potential labor market experience for person j ; arrival, indicates age atarrival in the United States for person j ; Z. is a vector of occupational dummyvariables (with "managerial and professional" serving as the referent); panel, is adummy variable indicating the SIPP panel (1996 or 2001); ji. are unobserveddifferences that affect wages; and r. is a stochastic error term.

These models allow both the intercepts and slopes to be correlated (T), and forauto regressive error terms to adjust for serial correlation and heterogeneity in thevariance within individuals (p). To account for the non-random sample loss ofpoor households (Bavier 2002), both the descriptive and inferential parts of thisanalysis are adjusted using the person weights provided by SIPP.

ResultsTable 1 shows average wage rates in six month intervals, between the 1" and 36*month, for the pooled SIPP panels (1996 and 2001), separately for men and wom-


Table 1: Mean Wages (in $ per hour) by Observation Month

MenDocumented Mexican immigrantsUndocumented Mexican immigrantsMexican-Americans (natives)Non-Latino white nativesWomenDocumented Mexican immigrantsUndocumented Mexican immigrantsMexican-Americans (natives)Non-Latino white natives

1

9.828.27

10.7412.98

7.637.088.659.69

6

9.638.09

10.6212.90

7.807.268.659.81

12

10.248.57

10.5913.25

7.807.058.85

10.16

Month18

10.249.00

11.1813.66

7.916.689.14

10.44

24

10.329.42

11.0914.07

7.786.989.26

10.66

30

10.949.57

11.3/14.28

8.027.309.78

10.87

36

10.889.36

11.6214.49

8.187.079.80

11.01

en workers belonging to each oFthe Four low-skill analysis groups—documentedMexican immigrants, undocumented Mexican immigrants, Mexican-Americansand non-Latino white natives. Although there are some irregular movements inthese mean wages, almost all groups show an upward trend in wage rates overtime. Thus among low-skill males, the hourly wages oF documented Mexicanimmigrants rise From $9.82 to $10.88 over the 36-month span; those oF undocu-mented Mexican immigrants rise From $8.27 to $9.36 over this period; those oFMexican Americans rise From $10.74 to $11.62; and those oF non-Latino whitenatives rise From $12.98 to $14.49.

For low-skill women, the hourly wages oF documented Mexican immigrantsrise From $7.63 to $8.18, those oFMexican Americans rise From $8.65 to $9.80,and those oF non-Latino white natives rise From $9.69 to $11.01. The exceptionis Female undocumented Mexican immigrants-their hourly wages average $7.08in the first month oFthe panel, and $7.07 in the 36* month. This group has thelowest starting wages, and no wage growth, over the observed duration.

Table 2 shows the means For the exogenous variables used in the wage trajectoryregressions, separately For men and women in each oF the Four ethnic/legal groups.Not surprisingly, we find that low-skill native Mexican Americans and whites havehigher educational levels than Mexican immigrants. Among men, natives are muchmore likely than immigrants to hold managerial, proFessional, technical, sales andadministrative jobs, whereas immigrants are more likely to be employed in service.Farming, Forestry and fishing. For women, natives are also more likely than immi-grants to be employed in managerial, proFessional, technical, sales and administra-tive jobs while immigrants have higher representation in service. Farming, Forestry,fishing, production, craft, repair, operative and laborer occupations. Despite thesedifferences between native and immigrant workers, documented and undocumentedMexican immigrants appear quite alike: both groups have similar levels oFeducationand ages at arrivals, and are similarly distributed across the occupational categories;legal Mexican immigrants do, however, have about six more years oF experience inthe labor market than do undocumented Mexican immigrants.


io'Ser

UJ

I/)oa>O)oXm£inu

u

il3X

E'x.Z3 eu

csi •«-

CO COen ez>

l O CM <Z>t co irj

co csiCM

^ CD i nLO ei3 CJ)

CM CO

CSI •>—

_ _

SS

3 X .o eu

x .20}

LO r-

CD CO••- -^r

e-)CM

CM

•<- CO COCD o •<-

T— CO COCD CD -^

a> co co coCM ez) CD co

C35 co co tco CD CD CM

c o t~~.CZ) T -

co co co i nT - CD CM c o

= T CD co2 f-: C3Í CD CM l Oez) CD •

LO • ^ 1T - CM c o^ q

csi 1^ CDC M •<—

-= .en o ) eu

.Q E

leu ce .£

^ are?c ^ o"= -"'S

en . tgj oS c en

O ^>

JjiS<'ö'

o .— ^ etj£ E -a äj<u eo 2 5CO LL Q. O

UnCM CJ5

CMCOh - '

•>— C OI—

CM CDCO

r-... ooO) I•^'ij)

CSI

CO CSJCSI c oco en

CMCM

co cqcd

tn oco E1 o- Í2E Q

Legal Status and Wage TrajectoriesMenTable 3 shows growth curve regres-sions for the logged wages of low-skillMexican immigrant men. Our pre-sentation ofthe growth curve modelsshows the intercept or starting wageeffects in the upper half of the tableand the slope or wage growth rateeffects in the lower half Thus, forexample, the legal status effect in the

"intercepts" panel refers to the impactof being documented on wages at thestart of each panel period (one begin-ning in 1996 and the other in 2001),and legal status in the "slopes" panelshows the monthly wage return tobeing documented (i.e., the interac-tion between legal status and time).Wages have been log-transformed inour regression models, which allowsfor marginal effects to be interpretedas percent change in wages resultingfrom a one-unit change in predictorvariables. To ease the presentation ofthe results, all coefficients in Table3 and subsequent regression analy-ses bave been multiplied by a factorof 100. The first column of Table 3sbows a model in which legal statusis the only predictor; the second col-umn adds labor market experience,education and age at arrival in theUnited States to the regression, andthe third column adds occupation.

The first column shows thatdocumented male workers averagesignificantly higher starting wagesthan those who are undocumented.The gross wage premium associatedwith being a legal immigrant is 17.3


percenc. In our sample of Mexican immigrane men, wages grow ae an average raeeof .5 percene per moneh. However ehere is no significane effece of legal seaeus onehe raee of wage groweh during ehese panel periods, indicaeing ehae documeneedand undocumeneed immigranes' wages increase ae aboue ehe same pace.

In ehe second column of ehis eable, we eese wheeher ehe wage dispariey beeweendocumeneed and undocumeneed Mexican immigranes is due eo differences inhuman capieal. The resules show ehae, as expeceed, seareing wages are affeceed posi-eively by labor markee experience and educaeion, and negaeively by age ae arrival.The negaeive slope effece for experience indicaees ehae ehe posieive reeurn eo experi-ence (in ehe incercepc equaeion) is less posieive for workers wieh more experience.Educaeion and age ae arrival do noe significanely influence wage groweh-ae leasewiehin our relaeively shore period of observaeion. Wieh ehese variables conerolled,ehe wage advaneage for being legal is reduced by more ehan half (57%), bue ehelegal seaeus effece on seareing wages is seill large, posieive and highly significane;

Q. O

reo

(/>a>atre

o

2Q.

'Ü

ginin2a>

O

li

CQ

- •—•

en

*

oo

CMCM

176.

•K•i t

. î î

.^ ID IDOO 1 ^ OOCM C3 OO

eZ5 OO CI3I D I D COt^T-co

tt •¥ *

o IO •»- Tj-^ i ^̂ - O5 ^^

•k

COI DI D

COOO

**

77)

CM CM o en IDCO """^^^ '*~*'^ ' ^

, ^CO.,—

tt«

o "COI D

CO

"QTC3Îro5CD

[star

Inte

rce

Con

sta

" " " " ^

CNJ CO CO CN- ^ CO ' ^r^ x - CO

*CD"

cd

CO

D

C OOí C/3

<D * O9" c/]^ 1

i l.i

Lega

l sLa

bor r

Educ

at

ej>

CO

in th

e

"cn

ro

.Ifl"

year

Age

(in

oo

*

CM 1 ^- ^

CJ>

« + «m es m.<— o o C35

CO CM ^ ^ T -

CJ5 CM

'

eu>

rat

.Í2

adm

ir

ies

, sal

Occ

upé

Tech

i-1

2,

8

Serv

i

coco CM•t— r— esl-plDCD

en0 3 ^

j fis

hii

repa

ier

s

fn C f^

^ ^ —'fñ ro *-j

fore

n, C

lsa

r

ling,

uctio

ativ

e

ro 2 Q.Li. Q. O

ï03CJ-

i>.

inth

l

4c4c4c

• ^ —

COco

§£2-

CT>

en+

1 ^ C33co oo i D cz>

1

UD

^—

CO ^SC3 C3 C3 C3

1

I DOO CD

4<4c4c

o "C35

¿,.̂_LO

f

Slop

esTi

me

(il

1

CZ3

CM• < - ;

Leqa

l s

4 i

CO

¿C3>

1

803

Labo

r r

1

o enm -r-

(-1

CM C3es es

1

'S)"ö¿COo :en 3(/3 ^ ^ro .—

ion(

:ar

rivEd

ucat

Age

at••> -K «

* , -k 4t ** + a tt *

^^S'^^ío'íoO5 co ^ l O • ^CSJ "*— ^ - LO co

lo co r- co to• ^ CNJ t o í ^ " ^

1 1 1 1 •

CD

!n CD 03

adm

ir

i fis

hii

repa

ier

s

T3 ë "P oc ro í« oro >, roen i ' ^ -_eu 1ñ 2 ^

3 , sal

fore

n, C

lis

ar

ation

'ni

cal

ice ling,

uctio

ative

Occ

up«

Tech

Serv

iFa

rn"

Prod

Ope

r

CO>iesCM'

crioo^—

CMcoco

ikpiil"

c

roó

oesVo.

***

oV

03

;lud

d

•odro

C|3

ro

0303

03

co

§•o"m

. ei> oV =Q eñ

9m- E ^

è -g o

Q0


this suggests that net of human capital, documented Mexican immigrants hold a7.4 percent wage premium over their undocumented peers.

Labor segmentation theories posit that one reason unauthorized immigrant work-ers' wages are lower than legal immigrants is that they are permanently consignedto jobs in the "periphery" of the labor market. While the restilts in the final columnloosely conform to the segmentation prediction that peripheral occupations havelower starting wages and wage growth than professionals and managers do, control-ling for occupation does not attenuate the wage advantage of being documented.

WomenTable 4 repeats these calculations for low-skill female Mexican immigrants. Thefirst column shows a significant wage differential between documented and un-documented female workers of about 8.6 percent. Tbe coefficient of legal statusin the slope equation indicates that documented Mexican women's wages growslightly faster than undocumented workers, but not significantly so.

Xa>

So>

a>

3

SQ .(O"ca>'o

11ouco'in

0£

o•íí

05 (2

276

(1

T - CO

* * ^1

co co 1—t ^ •<— CD

SStñ^— 1

* *• ^ 1 ^ CD•51- - t - CM

o CO

^ C M CM -^-^

CO CO CM CO mi n t ^ CD •>- m

•<- CD

CM CO

g

01 (

-1

1

+CD CMCO i n

• ^ CD

* *•<- co co1 ^ co CMco CM T^

s ̂ ̂1 1 1

*

m CD• ^ i nT— CM

• 1

co CM t ^ co •«-CD t--. CD en CDcô -^-«a^ co coCM "—••— \ l .co co CD T— coCD h-; CD CD CDi r i CM ' .r^ >'t

•<s- mCM coir i CMm —'

CM CC3CD

CM co co CD> CM T— CO t ^

T— T— T— CO X—CD CM CD CI> C3

1 ^ COCM I—

CD COCM • . -

CDCDro

•Oro

g r o coro ro

o _ ;c/) 3*~ 0)

ro .^

c ' Eo ro

• o

roco75tn

ta

.°.o

I

till•-. i o o o « <̂

ro.•«

-aroT3crro

.3?roco

Í2O) d)

S-ii-^ëÊ E 'S 0) « ̂ —J2 ._Q'roc.S-Q.curog</> U- Q. O _o E g'-9

1ÇJ 03


Human capital characteristics of Mexican immigrant women are added to theequation in the second column. For Mexican women, starting wages are higher forthose with more labor market experience and schooling, and lower for immigrantsarriving at later ages. Over our relatively short observation period, wage growthamong this group of women is not differentiated by human capital characteristics.Most importantly, when these traits of workers are held constant, the legal statuseffect is reduced to a statistically insignificant 2.7 percent. This suggests that,for Mexican women, the wage premium associated with legal authorization islargely attributable to relatively benign labor market processes that reward bettereducated and more experienced workers.^

Lastly, the third column in Table 4 includes controls for Mexican immigrantwomen's occupations. Low-skill Mexican women working in service, farming,forestry and fishing report significandy lower wages than their counterparts in pro-fessional and managerial roles (the reference group). These occupational variablesdo little to further explain the (non-significant) legal status coefficient, and in factappear to slightly suppress it. Additional analyses (available upon request) showthat occupation alone (in models not including human capital characteristics)does not explain the initial effect of legal status. Thus, as was true for men, oc-cupational sorting does not appear to play a noticeable role in wage differentiationbetween documented and undocumented Mexican immigrant women.

Legal Status and Returns to Human CapitalMenTable 5 estimates separate growth curve models for males in each of the fourlow-skill ethnic/legal groups. This allows a full set of interactions between, onthe one hand, immigrant and legal status, and on the other, wage predictorssuch as labor market experience and education."* Significant differences in theseeffects between (1. documented and undocumented Mexican immigrants arenoted in the table with a "d;" (2. Mexican immigrants and Mexican Americanswith an "m;" and (3. Mexican immigrants and white natives with a "w." Thesesets of models thereby test whether the lower returns to human capital reportedfor Mexican immigrant men (Hall and Farkas 2008) are accounted for by theundocumented status of some of these workers.

The results clearly show that undocumented male workers receive lower startingwage returns to experience and schooling than the other groups, and that theiraverage wage growth is lower than that of the other groups. For undocumentedMexican immigrants, the coefficient of labor market experience is .87. This meansthat for each additional year of labor market experience, starting wages for thisgroup increase by .9 percent. There is a comparable effect for documented Mexicanimmigrants, about 1.1 percent. By contrast, Mexican American and non-Latinowhite natives gain significantly more from work experience, with each additionalyear associated with an increase in starting wages of about 1.5 percent.


Effects For years of schooling show even more striking inter-group differences. Thus,each additional year of schooling increases the starting wage rates of undocumentedMexican immigrants by 1.5 percent, whereas the effect for documented Mexicanimmigrants is 3.5 percent, that for Mexican American natives is 6 percent, and thatfor non-Latino white natives is 9.3 percent. This shows that a significant portion (butnot all) of the lower returns to schooling received by Mexican immigrants (Hall andFarkas 2008) is due to the very low returns experienced by undocumented workers.

c

SOl

i• o03OlOloOl

¡ctir

redi

o.•£Oí

Ici1Oot/lU)£S"oc.

lirve

o

Soo

Spec

i

¡dno.

oOla>

' c

EthTa

bie 5:

ihit

e

o

.atii

cozcreu• c

E<¿fi

exi(

.2

umer

uo•oc

sc

ii

ves

1

<n.5"re

.2c

1EC

s'Sa>S

1o'gEcreo

Me)

CQ

CQ

CQ

CQ

CD CO r.-in cp r̂..-d có -r^CNJ c o CNI

in co -^T- íci coco - ^ CDcz>

* * *

o~c^o"cq •<a; CNJ• ^ CD c d

CNI - ^ CI>co i n tz>CD •<-: coco

I I

1 ^ CO COCO CO i n

CNJ CO CD • ^

^Scöh...- •>«; CDCO CO CNJ

CO I ^ COCD CO i nCO r-̂CO

T— CO COCO CO t or~- cc) COCNJ ^ ^ - •

O Î CO COCNI T - i nCD - ^ COCO

O ) •<- Ti-co ^ CDCNJ CO

CO CNI CNICO t o CD

T - i n CDCNI CO ^

CD CNC O T -

1 ^ •<— CNICD CD CO

CDCJ)

iî l

I ? "̂: o <DI tn DJ: M- ro< O g

ro .CD

: CD CD: . . . - CD

: o o.

g.tD^ S

•>a- r-.. COm -r- t—CNj cD i n

coCNJ c o CNJc o CNJ c oC D CO C D

CNj'CNJ

CDCD ÍC3CO CO CO

co'

^ CD coco - ^ CD^ •«-_ i n

" J

= S

_toCD

E

tb" =

Ig'roIs-.« "̂

ro^ • §

oCD

- ië Et s ^0̂ " ro

"2 92 .^

c ro<" co 5

T3 tD £ro S 'S"c ^ o5gj ~ ifc'o "O "Oig : . . >,

".S' — * CNJB V-to oÄ co

liro"A °̂o —^ ro

SiV «O.Q.

co

V o

om tDco -a

VCL !

totD

ro

S

m

jn"'Sro ,_:̂D)OI"-§ 5-'x totD g

TD ^tD •='S Sst

IIilE "?o .>"

Çî f i tDtD S .ytD

OJ ^ aj• ^

• i &O)'to


Finally, Table 5 shows average wage growth rates (the coefficient of time in theslope equation) of .1 percent per month for undocumented Mexican immigrants,1 percent per month for documented Mexican immigrants, .9 percent per monthfor Mexican Americans, and .8 percent per month for non-Latino white natives.Clearly, undocumented immigrants have, by far, the lowest rate of wage growth.

WomenTable 6 shows corresponding growth curve model estimates for low-skill femaleworkers, separately by ethnicity and legal status. As was true for men, documentedMexican immigrant women receive significantly higher wage returns to educationthan undocumented women. The second column shows that undocumentedMexican immigrant women receive a starting-wage increase of only .7 percent foreach additional year of schooling. In contrast, documented Mexican immigrantsreceive a starting wage return to schooling more than twice as high (2 percentfor each year of schooling). Importantly, the comparable coefficient for Mexican-American women is even higher (5.1 percent), but still far below that of non-Latino white natives (9.3 percent). Thus, while all groups receive positive startingwage returns to schooling, those for undocumented women are by far the lowest.While this supports the idea that Mexican immigrants' low returns to educationare partially due to large shares of undocumented immigrants among the Mexicanimmigrant population, the fact that documented Mexican immigrants and native-born Mexican Americans receive much lower returns than white natives suggeststhat other underlying processes may be at work.

The slope coefficients for women in Table 6 are generally nonsignificant.However, the large, positive and significant growth coefficient for white nativesindicates that the wages of foreign- and native-born Mexican immigrant womengrow significantly slower tban those for white natives. In addition, while the termsthemselves are nonsignificant, the interaction terms indicate that undocumentedMexican immigrant women's wage growth is lower than all the other groups.

DiscussionWe provide new evidence on the wage benefits of having legal authorization to be inthe United States for Mexican immigrants. In contrast to other research, this studyuses recent, nationally-representative data to assess wage differentials betweendocumented and undocumented Mexican immigrants, and the differential returnsto human capital between these groups and native-born Mexican Americans andnon-Latino wbites. Using two panels (1996-1999 and 2001-2003) of longitudinalSIPP data, and employing a unique strategy of inferring immigrants' legal status,we have shown that the wage premium of "being legal" for Mexican immigrants isapproximately 17 percent for men and 9 percent for women. Altbough a consider-able portion of tbe legal status wage premium is attributable to differences in labormarket experience, educational attainment, and age upon arrival in tbe United

506 • Social Forces 8m)

States, male documented workers retain a significant wage advantage relative totheir undocumented counterparts even when these factors are taken into account.Our analysis also demonstrated the very low returns undocumented Mexican im-migrant workers receive from human capital. For both male and female undocu-mented Mexican workers, the wage returns to education are half of those realizedby documented Mexican immigrants. Similarly, undocumented men receive lowerreturns to labor market experience than documented men. Also, in comparison

1a.

o

0£03

3U

Iou

a.2O)

03

* inII«o

1.1il

•a 2S.S

CO t ^ OOc o CD CM

ID CO •CO -«a-

co -«i- ooCD T- IDco ID co ^¿.-̂ l,-»—

.>—.<— ooTí- oq CD0Ó ' ID

co co .^h-, h- IDeji ^^^

• ^ CM ooO3 l~-- cocôr

CD co cooq T - C7>CM co coCM ~. - ' -—•

co T- . ÎD ID CDCM ' CM't

CO CD iCD CZ3 ez3

Tl- CI3 ^c o t ^ CM

1 ^ T— ICM CZ3 CD

CD

I D . ^ CMc o CZ3 CZ3

cqh T— coco t ^CM T -

co'

euen

i §-b 0303 en O3c M— roo o gX £».>'-03 rox:.1

8

03

03 ro.» eu

llllï'

ID c o - ÎD CM ~CM CO

• ^ CM ez>ID coO3 t^

co'

- a en

1103 • >

jn03

"8

03

°c roSS S" en

- a 03

SSa Sid d 039? -~ î£Ü "O T3

8g §2 ^ 1^ O3 "gSS ^ . -.•V 03 JÍ2§ S =S CD m

iilv• p - CL

V en c .•-.?5 -2-

S: s enCr> 03 03

ID 03 o ECD T3 O o

• -o E "°V ^ O ô. .c >fc •£+ .^î-g en85 i i l'S 2 îe 03z ro T3 en


to other low-skill workers (including documented Mexican immigrants) wagegrowth during our observation periods (1996-1999 and 2001-2003) was slowestfor undocumented Mexican immigrants of both sexes.

The 17 percent gross wage differential we find between undocumented anddocumented Mexican immigrant men is within the range of previous post-IRCAestimates (Donato and Massey 1993; Kossoudji and Cobb-Clark 2002; Rivera-Batiz 1999). Less is known about the legal status wage differential for Mexican im-migrant women, but our finding that documented women hold a wage premiumof about 9 percent indicates that, like men, undocumented women face employ-ment challenges in U.S. labor markets. After our estimates are adjusted for workercharacteristics (e.g., education, experience, age on arrival), legal Mexicans' wageadvantage drops to 8 percent for men, and a statistically insignificant 4 percentfor women. This net wage differential is also in line with other studies (Kaushal2006; Kossoudji and Cobb-Clark 2002; Rivera-Batiz 1999).

The gender differences in the effect of legal status and its mediating factorssuggest not only that legal authorization is more consequential for Mexican men'swage determination, but also that the underlying processes at work differ by gender.We initially suspected that occupational segregation played an even stronger rolefor undocumented women than men, given that previous research showed thatthe potent combination of gender, ethnicity and nativity leads to a narrow rangeof occupations available to newly arrived, low-skill Latina women (Myers andCranford 1998). We reasoned that moving into better-paid occupations would bemore difficult for workers without legal authorization, potentially contributing toearnings differences by legal status. However, this explanation was not borne outby the data: occupation explained little of the difference in the effect of legal statusamong women, either before or after taking into account human capital factors.

Differences in migratory patterns for men and women may be a more promisingexplanation. In the migration stream from particular communities or families, thetiming of men's migration tends to be earlier than that of women. Women often fol-low men as dependents or, if they themselves are labor migrants, migrate to join fem-ily or community members who have already established themselves in the UnitedStates (Donato, Wagner and Patterson 2008). Thus undocumented male workers aremore likely to consider themselves temporary migrants whose intention is to returnto families in Mexico, whereas women may see their migration as more permanent,regardless of legal status. Women are also more likely than men to be accompaniedby families. Bonacich's (1972) split labor market theory posits that workers whoview themselves as temporary migrants and who are unencumbered by dependentswill be more willing to accept low pay and substandard living and working condi-tions. Thus for men, more so than women, lacking papers may correlate factors thatmake them willing to accept lower wages While it would be a mistake to completelydismiss the potential influence of legal stattis on earnings for Mexican immigrantwomen, our findings are consistent with previous literature in suggesting that other

508 • Social Forces 8^(2]

factors-particularly ethnicity, gender and education-are more critical determinantsof earnings for this group (Baker 1999). Mexican documented and undocumentedwomen share ethnicity and gender and have similar levels of education. Given theextremely strong relationship between these factors and earnings, there may simplybe little room for legal status to have an additional effect.

If, for men, the cost of being undocumented is not accounted for by their hu-man capital or by the types of jobs they hold, what explains the wage disparity?Labor market processes that cheapen the value of undocumented immigrants' la-bor are one possible explanation. From the perspective of employers, a legal statuswage gap is justified if governmental sanctions levied against them make hiringundocumented workers costly. These costs are, in turn, passed off as an effectivetax on the wages of undocumented workers (Massey, Durand and Malone 2002;Phillips and Massey 2000).

Employers, of course, are foremost in the business of maximizing profits, andwill, assuming similarities in skill, hire workers with the lowest reservation wages.It is widely recognized that because of dual reference frames between origin andhost countries, the reservation wages of immigrant workers are lower than thoseof natives, but they may also differ between documented and undocumentedworkers if the motives for work vary (see Bonacich 1972; Waldinger and Lichter2003). Undocumented workers are, for example, more likely to intend their staysin the United States to be temporary or seasonal (Massey et al. 2002), and maythus be willing to accept lower wages to meet provisional or fixed goals. If thisis true, our finding that undocumented workers receive lower returns to humancapital and have wages that grow more slowly than other workers may indicatethat undocumented workers substitute immediate earnings over investments inpotential earnings. This line of reasoning parallels labor segmentation models, butthe underlying mechanism-motives vs. exclusionary practices-differs.

Institutional and legal constraints also undoubtedly play important roles inexplaining the legal status differentials in wages and returns to human capital.Undocumented immigrants are ineligible for many governmental assistanceprograms, and heightened animosity toward undocumented immigrants andharsh municipal policies about where they can live and seek employment (Singer,Hardwick and Brettell 2008) create an unstable and potentially hostile socialenvironment. As a result, undocumented immigrants often take the first job theyare offered, continue to work in jobs even if the pay is low, or accept exploitativeor illegal work conditions out of fear that they will be exposed.

We may actually be under-tsûramn^ the legal status wage premium if theseprocesses are imposed on legal Mexican-origin workers (documented Mexicansand native Mexican Americans). Lowell, Teachman and Jing (1995) demonstrateddiscrimination in hiring for all Latino-looking workers, not just the undocument-ed, as a result of IRCA-implemented employer sanctions. Bansak and Raphael(2001) and Orrenius and Zovodny (2009) show evidence of lower earnings for


Latino workers as a function of increased enforcement. Donato et al. (2008) docu-ment similar processes for Mexican-origin women. Massey et al. (2002:126) go astep furtber and argue that the employment prospects and wages of/íga/Mexicanimmigrants are more greatly affected by governmental sanctions and employerdiscrimination than those of undocumented Mexican immigrants: "Rather thanprotecting domestic workers, the criminalization of undocumented hiring endedup marginalizing [legal Mexican workers] by exacerbating income inequality, en-couraging subcontracting, and generally promoting the informalization of hiring."Thus, if the wages of documented Mexican immigrants are penalized as a result ofillegal immigration, out legal status wage differentials are likely biased downwards.

If in fact these processes tied to legal status detrimentally affect the economicwell-being of all Latino workers (regardless of legal status), they might explainwby legal Mexican immigrants and (to a lesser extent) Mexican-American workersare less able to convert additional schooling into higber wages than non-Latinowhite natives. Nonetheless, the fact that only part of Mexican-origin workers' lowoverall returns to human capital is explained by the abundance of undocumentedmigrants among tbeir population is troubling, and is consistent with Telles andOrtiz's (2008) finding that Mexican Americans struggle to reach the kind of edu-cational equality that other immigrants have achieved.

An important research question concerns what the future holds for undocu-mented workers. Nativism and hostility toward undocumented Mexican immi-grants has increased in recent years, and as political efforts further marginalize theseindividuals, it is plausible that the wage penalties for not holding legal authorizationmay increase. Likewise, in times of economic crisis, immigrants are ofi:en blamed. Insupplemental models available by request, we tested for the possibility that the effectof legal status changed over time.^ We found no evidence that it did, but this may beattributable to our observation window (1996-2003) in wbich no major legislationwas offered, public sentiment being less antagonistic, and economic conditions thatwere reasonably robust, at least in comparison to the post-2005 period. Continuingto explore these issues clearly needs to be on the agenda for future research.

Continuing to explore the multifaceted and complex mechanisms through whichlegal status affects workers' economic well-being is clearly needed. In addition, itis important for future research to examine tbe impact of legal status on other di-mensions of social life, particularly health, education, civic engagement, crime anddeviance, and health. The imputation strategy advanced here offers a new approachto understanding these issues, and we encourage researchers to employ it.

Notes1. We re-estimated our models using lower bounds of $.01 and $2.50 and upper bounds

of $100 and $500. The results were consistent regardless of wage bound placement.

2. Descriptive statistics for Mexican immigrants in the 2001 panel with valid Englishability information (N = 695 men and women) indicate that undocumented Mexican


workers are less likely than documented Mexican immigrants to speak English well(36.3 vs. 51.1 percent, respectively), but the correlation (.12) between Englishability and legal status is considerably smaller than those between English abilityand education (.30) and age at arrival (-.26).

3. Additional analyses reveal that legal status is attenuated only when the full set ofhuman capital characteristics is accounted for.

4. As the main goal of the group-specific models is to test differences in the returns tohuman capital, occupational controls are excluded from group-specific growth curveregressions.

5. These models were carried out by a series of interaction terms between legal statusand (1. linear observation year, (2. individual year dummy terms, and (3. pre- andpost-9/11. Intercept and slope coefficients for all of these terms were small and mostlynonsignificant.

ReferencesBaker, Susan Gonzalez. 1999. "Mexican-Origin Women in Southwestern Labor Markets."

Pp. 244-69. Latinas and African American Women at Work. Irene Brown, editor.Russell Sage Foundation.

Bansak, Cynthia, and Steven Raphael. 2001. "Immigration Reform and the Earnings ofLatino Worker: Do Employer Sanctions Cause Discrimination?" Industrial and LaborRelations Review 54(2):275-95.

Bavier, Richard. 2002. "Welfare Reform Impacts in the SIPP." Monthly Labor Review125(11):23-38.

Bleakley, Hoyt, and Aimee Chin. 2004. "Language Skills and Earnings: Evidence fromChildhood Immigrants." The Review of Economics and Statistics 86(2):481-96.

Bonacich, Edna. 1972. "A Theory of Ethnic Antagonism: The Split Labor Market."American Sociological Review 37(5):547-59.

Borjas, George J. 1985. "Assimilation, Changes in Cohort Quality, and the Earnings ofimmi^ïAnts!' Journal of Labor Economics ?>{A):AG'5-?>9.

Catanzarite, Lisa. 2000. "Brown Collar Jobs: Occupational Segregation and Earnings ofRecent-Immigrant Latinos." Sociolo^calPerspectives 43(l):45-75.

Chiswick, Barry. 1978. "The Effect of Americanization on the Earnings of Foreign-BornMen." Journal of Political Economy 86(5) :897-921.

Donato, Katherine M., Brandon Wagner and Evelyn Patterson. 2008. "The Cat andMouse Game at the Mexico-U.S. Border: Gendered Patterns and Recent Shifts." TheInternational Migration Review 42(2):330-59.

Donato, Katharine M., Chizuko Wakabayashi, Shirin Hakimzadeh and Amada Armenta.2008. "Shifts in the Employment Conditions of Mexican Migrant Men and Women:The Effect of U.S. Immigration Policy." Work and Occupations 35(4):462-95.

Donato, Katharine M., and Douglas S. Massey. 1993. "Effects ofthe Immigration Reform andControl Act on the Wages of Mexican Migrants." Social Science Quarterly 74(3):523-4l.

Greene, William H. 2000. Econometric Analysis. Prentice Hall.Hall, Matthew, and George Farkas. 2008. "Does Human Capital Raise Earnings for

Immigrants in the Low-Skill Labor Market?" Demography 45{3):6l9-59.

Legal Status and Mexican Immigrants' Wages »

Hoefer, Michael, Nancy Rytina and Christopher Campbell. 2007. "Estimates of the UnauthorizedImmigrant Population Residing in the United States: January 2006." Washington, DC: U.S.Department of Homeland Security, Office of Immigration Statistics.

Kaushal, Neeraj. 2006. "Amnesty Programs and the Labor Market Outcomes ofUndocumented Workers." fournal of Human Resources 41 (3):631-47.

Kossoudji, Sherrie A., and Deborah A. Cobb-Clark. 2002. "Coming out of the Shadows:Learning about Legal Status and Wages from the Legalized Population." fournal ofLabor Economics 20(3):598-628.

Lowell, B. Lindsay, Jay D. Teachman and Zhongren Jing. 1995. "UnintendedConsequences of Immigration Reform: Discrimination and Hispanic Employment"Demography 32(4):6l7-28.

Massey, Douglas S. 1987. "Do Undocumented Migrants Earn Lower Wages than LegalImmigrants?" International Migration Review 21(2):236-74.

Massey, Douglas S., and Katherine Bartley. 2005. "The Changing Legal Status Distributionof Immigrants: A Caution." International Migration Review 39(2):469-84.

Massey, Douglas S., Jorge Durand and Nolan J. Malone. 2002. Beyond Smoke and Mirrors:Mexican Immigration in an Era of Economic Integration. Russell Sage Eoundation.

Myers, Dowell, and Cynthia J. Cranford. 1998. "Temporal Differentiation in theOccupation Mobility of Immigrant and Native-Born Latina Workers." AmericanSociological Review 63(l):68-93.

Orrenius, Pia M., and Madeline Zavodny. 2005. "Self-Selection among UndocumentedImmigrants from Mexico." fournal of Development Economics 78(l):215-40.

. 2009. "The Effects of Tougher Enforcement on the Job Prospects of Recent LatinAmerican Immigrants." fournal of Policy Analysis and Management 2S{2):239-57.

Phillips, Julie A., and Douglas S. Massey. 1999. "The New Labor Market: Immigrantsand Wages after IRCA." Demography 36(2):233-46.

Raudenbush, Stephen W, and Anthony S. Bryk. 2002. Hierarchical Linear Models. Sage.Rivera-Batiz, Francisco L. 1999. "Undocumented Workers in the Labor Market: An

Analysis of the Earnings of Legal and Illegal Mexican Immigrants in the UnitedStzKs." fournal of Population Economics 12(1):91-116.

Schoeni, Robert F. 1997. "New Evidence on the Economic Progress of Foreign-Born Menin the 1970s and \%Ç)s" fournal of Human Resources 32(4):683-740.

Singer, Audrey, Susan W. Hardwick and Caroline B. Brettell. Editors. 2008. Twenty-EirstCentury Gateways: Immigrant Incorporation in Suburban America. Washington, DC:Brookings Institution.

Telles, Edward, and Vilma Ortiz. 2008. Generations of Exclusion: Mexican Americans,Assimilation, and Race. Russell Sage.

Topel, Robert H., and Michael P Ward. 1992. "Job Mobility and the Careers of YoungMen." Quarterly fournal of Economics 107(2):436-79.

U.S. Census Bureau. 2004. Annual Social and Economic Supplement. Available at: http://www.census.gov/population/www/socdemo/foreign/ppl-176.htmlfla (earnings) andhttp://www.census.gov/ population/socdemo/foreign/p20-534/tab0111.pdf (poverty).

U.S. Department of Homeland Security. 2007. Yearbook of Immigration Statistics: 2006.Washington, DC: U.S. Department of Homeland Security, Office of Immigration Statistics.

Waldinger, Roger, and Michael Lichter. 2003. How the Other HalfWorks: Immigration andthe Social Organization of Labor. University of California Press.


Appendix 1. Imputation of Legal Status for Mexican Immigrants in SIPP

Universe: Foreign-born Mexican adults ages 18-60with less than a high school education who participated in theWave 2 migration module.

When arriving in the U.S. to live, did respondententer as a legal permanent resident?

Yes No

Has respondent's immigrant status beenchanged to permanent?

Yes No

Does respondent personally receiveassistance from any federal program?

Yes No

Is respondent or respondent's spousea high ranking political official?

Yes No

Is respondent or respondent'sspouse enrolled in coiiege?

Yes No

Documented Undocumented

Legal Status and Mexican Immigrants' Wages « 5 1 3

Appendix 2. SIPP Survey Questions on Citizenship and Immigration Status

Citizenship:"Is [R] a U.S. Citizen?"

(l.Yes(2. No

Initial Legal Permanent Resident Status:"When [R] moved to the United States to live, what was [R's] immigration status?"

(1. Immediate relative or family sponsored permanent resident(2. Employment-based permanent resident(3. Other permanent resident(4. Granted refugee status or granted asylum(5. Non-immigrant (e.g., diplomatic, student, business, or tourist visa)(6. Other

This full list of responses is restricted in the public access data to:(1. Permanent resident (which corresponds to response numbers 1, 2 and 3)(2. Other (response numbers 4, 5 and 6)

Current LPR Status:"Has [R's] status been changed to permanent resident?"

(l.Yes(2. No

Copyright of Social Forces is the property of University of North Carolina Press and its content may not becopied or emailed to multiple sites or posted to a listserv without the copyright holder's express writtenpermission. However, users may print, download, or email articles for individual use.