Comparing Wage Gains from Small and Mass Scale Immigrant Legalization. Programs

Transcription

1 UNR Economics Working Paper Series Working Paper No Comparing Wage Gains from Small and Mass Scale Immigrant Legalization Programs Sankar Mukhopadhyay Department of Economics /0030 University of Nevada, Reno Reno, NV (775) Fax (775) May, 2016 Abstract: In this paper I estimate the difference in wage gains from legalization between immigrants who were legalized on the basis of family ties, or smaller scale legalization programs, and those who were legalized in a large scale amnesty program (the 1986 Immigration Reform and Control Act, IRCA). Estimates suggest that the increase in wage after legalization is 11.7% higher for immigrants who were legalized on the basis of family ties, or smaller scale legalization programs, compared to IRCA beneficiaries. Further analysis suggests that supply shock restricted the wage gains of male but not female IRCA beneficiaries. Results also show that previously illegal immigrants receive a lower return to U.S. education, and work experience acquired as illegal workers, than the return legal immigrants receive from similar U.S. experiences. JEL Classification: J3, J6 Keywords: Immigration, Illegal Immigration, Legalization, Wage gain. 1

2 Comparing Wage Gains from Small and Mass Scale Immigrant Legalization Programs Sankar Mukhopadhyay Associate Professor Department of Economics (MS 030) University of Nevada, Reno, NV, Phone: Fax: Abstract In this paper I estimate the difference in wage gains from legalization between immigrants who were legalized on the basis of family ties, or smaller scale legalization programs, and those who were legalized in a large scale amnesty program (the 1986 Immigration Reform and Control Act, IRCA). Estimates suggest that the increase in wage after legalization is 11.7% higher for immigrants who were legalized on the basis of family ties, or smaller scale legalization programs, compared to IRCA beneficiaries. Further analysis suggests that supply shock restricted the wage gains of male but not female IRCA beneficiaries. Results also show that previously illegal immigrants receive a lower return to U.S. education, and work experience acquired as illegal workers, than the return legal immigrants receive from similar U.S. experiences. JEL Codes: J3, J6 Keywords: Immigration, Illegal Immigration, Legalization, Wage gain. 2

3 1. Introduction Estimates suggest that there are about 10.8 million illegal immigrants in the U.S. (Hoefner, Rytina, and Baker, 2010). Policy debate on legalization of previously illegal immigrants is fierce. 1 One topic that receives particular attention from economists is the wage premium to legal status. Many previous studies show that legalization leads to modest increase in wages. Most of these studies use the variation brought on by the 1986 Immigration Reform and Control Act (IRCA) to estimate the wage returns to legalization. There are a few papers that have focused on smaller legalization programs. However, these papers do not compare the wage gains from largescale legalization programs, to small-scale legalization programs which is the primary focus of this paper. Large-scale legalization programs may bring labor supply shocks, which may restrict the gains from legalization, at least in the short run. Thus comparing the legalization premium for those who were legalized on the basis of family ties or small-scale legalization programs, to the legalization premium from a large-scale amnesty program (IRCA) may, under certain assumptions, provide an estimate about the importance of labor supply shocks brought on by large-scale legalization programs. It may also provide an estimate of the long-term wage gain (as the supply shock of any large-scale amnesty program dissipates) from large-scale legalization programs. I estimate the difference in wage gains among immigrants who were legalized through two very different avenues: (1) on the basis of family ties or smaller scale legalization programs, 2 and (2) by a large scale amnesty program (IRCA). Another objective of this paper is 1 The comprehensive immigration reform bill (The Border Security, Economic Opportunity, and Immigration Modernization Act of 2013, or BSEOIMA) provides a path to Legal Permanent Residency (LPR), and citizenship, for illegal immigrants who were in the U.S. before December 31, The small-scale legalization programs include: the Nicaraguan Adjustment and Central American Relief Act (NACARA), the Chinese Student Protection Act, Employees of Hong Kong Businesses, Denied Parolees from the 3

4 to explore how legal status affects returns to human capital. First, I use the 2003 New Immigrant Survey (NIS) to estimate wage gains from legalization, for immigrants who were legalized through family sponsorship, or through small-scale legalization programs. 3 The estimated wage gain from legalization, for a previously illegal NIS respondent, is 18.2%. Among immigrants with high school or less education the estimated wage gain from legalization is lower (13.6%). Estimates also suggest that illegal immigrants get a lower returns to human capital compared to legal immigrants. Then, I compare wage gains from legalization of NIS respondents (who were legalized through family ties or small-scale legalization programs) to the Legalized Population Survey (LPS) respondents who were legalized by the 1986 IRCA. My estimates suggest that the increase in wage after legalization is 11.7% higher for NIS respondents compared to LPS respondents. Further analysis suggests that supply shock restricted the wage gains of male IRCA beneficiaries. Many previous studies show that legalization brings economic opportunities to illegal immigrants. Most of these studies use the variation brought on by the 1986 IRCA, to estimate the wage returns to legalization. Cobb-Clark, Shiells, and Lowell (1995) found a small, but statistically significant, wage increase after legalization. Rivera-Batiz (1999) estimates that Mexican legal immigrants earn about 20% more than illegal immigrants, after controlling for observable differences. Kossoudji and Cobb-Clark (2002) (KCC from now on) report that the wage gains from legalization are around 6%. Amuedo-Dorantes, et al. (2007) estimate that legalization increased wages of immigrant men by 9.3%, and immigrant women by 21%. Amuedo-Dorantes and Bansak (2011) report similar results. Soviet Union., American Indians Born in Canada, and Employees of the Panama Canal Company or Government, etc. 3 See Jasso, Massey, Rosenzwig and Smith (2008) for details on avenues of legalization, and how previously illegal persons can be identified in the NIS data. In the current sample, about half (48.5%) were legalized through family sponsorship, about one third (33.3%) through legalization programs, and the rest through other means. 4

5 Barcellos (2010) uses a regression discontinuity approach, to estimate the effect of IRCA legalization on a number of outcomes. She finds only a small, but positive, effect of legalization on wages. Pan (2010) estimates that the 1986 IRCA increased wages of legalized immigrants by about 5%. Lozano and Sorensen (2011) aim to identify the long-term effects of legalization, estimating that legalization increased wages by 20%. Lofstrom, Hill, and Hayes (2010) (LHH from now on) use NIS data to estimate gains from legalization through family sponsorship, and small-scale legalization programs. They do not find an illegal status penalty in the starting wage, and do not find any effect of legalization on wages. A few papers have focused on specific small-scale legalization programs. Kaushal (2006) uses the Current Population Survey (CPS), to estimate a 4% legalization premium, for individuals who were legalized through the 1997 Nicaraguan Adjustment and Central American Relief Act (NACARA). However, she notes (pp-639) that her treatment group includes legal permanent residents (LPRs), and therefore she estimates an intent-to-treat effect, and not per-se a treatment effect. Orrenius, Kerr, and Zavodny (2012) find that the 1992 Chinese Student Protection Act (CSPA) increased the wages of Chinese immigrants by almost 24%. In a recent paper Orrenius, and Zavodny (2015) found that temporary protected status (TPS) from deportation increases the wage of non-college educated men by 13.1%. In this paper, I start by estimating the wage gains from legalization of previously illegal immigrants, who were legalized through family ties, or small-scale legalization programs, using NIS data. The NIS uses a nationally representative sample of immigrants who achieved legal permanent resident (LPR) status in 2002 or Some of the new LPRs were previously illegal. Jasso, Massey, Rosenzweig, and Smith (2008), show that administrative record, and survey questions, can be used to identify the previous legal status of NIS respondents. I distinguish 5

6 between two types of illegality: those who entered the U.S. without documents (entry without inspection, or EWI) and those who overstayed their visa (over-stayers, or OS). 4 NIS data includes the wages of these previously illegal immigrants at their first U.S. job (when they were staying and working illegally) and after they received LPR status. The NIS data also includes wages of legal immigrants (in the rest of the paper I refer to this group as the always legal, or NIS-AL immigrants) when they started working in the U.S. on a legal temporary worker visa (such as H1B), and subsequent to receiving their LPR status. In the empirical implementation, NIS-AL immigrants work as a control for the two treatment groups: NIS-EWI, and NIS-OS immigrants. Therefore, a difference in difference estimator can be implemented. Estimates suggest that NIS-EWI immigrants had an 18.2% increase in wages after legalization, compared to NIS-AL immigrants. On the other hand, NIS-OS immigrants did not experience any wage gain after legalization. The difference between the results of LHH (who also use the NIS data), and this paper, is due to a difference in assumptions about returns to human capital. LHH assume that human capital (education and work experience) acquired in the U.S. as an illegal immigrant, provides the same return as human capital acquired in the U.S. as a legal immigrant. Some previous papers in this area (Calavita 1992; Borjas and Tienda 1993; KCC 2002; Caponi 2014) have postulated that illegal immigrants may get a lower rate of return to human capital, compared to legal immigrants. Therefore, I allow the return to human capital acquired in the U.S. to differ based on legal status at the time of acquiring the human capital. I find that previously illegal immigrants receive a lower return to U.S. education, and work experience acquired as illegal workers, than the return legal immigrants receive from similar U.S. experiences. This remains 4 In the rest of the paper, the term NIS-EWI immigrants is used to refer to the previously illegal NIS respondents who entered the U.S. without documentation, and the term NIS-OS immigrants is used to refer to the previously illegal NIS respondents who overstayed their visa period. 6

7 true even after legalization, i.e. human capital acquired when immigrants were staying in the U.S. illegally continues to get a lower return (compared to returns to human capital for NIS-AL immigrants) even after the individual immigrants have received LPR status. Next, I explore whether the gains from legalization are different for immigrants legalized through family ties, or small scale legalization programs, from those who are legalized through the 1986 IRCA. I compare the difference between starting wages (i.e. in their first U.S. job) and post-lpr wages, of illegal immigrants from NIS and LPS samples. My estimates show that the gain from legalization, of NIS-EWI immigrants, is 11.7% more than that of LPS respondents. This difference persists even after controlling for a plethora of individual level characteristics, and macroeconomic conditions. However, there are important differences across genders. Legalization increased the wages of male NIS-EWI immigrants by 18.3% more than male LPS respondents. However, there is no difference between the legalization premiums of female NIS- EWI immigrants and female LPS respondents. Then I explore the reasons behind the differential in wage gains: I compare the starting wages (i.e. in their first U.S. job) of illegal immigrants from NIS, and LPS samples. The estimates show that there are no differences in starting wages of male NIS-EWI, and male LPS respondents. However, starting wages of female NIS-EWI respondents are significantly higher than starting wages of female LPS respondents. The higher starting wages of NIS-EWI females may be due to changes in the illegal status penalty after the 1986 IRCA, or it may be due to unobserved differences between these two groups. These two effects cannot be separately identified from the data. However, it is not absolutely necessary for the purpose of this paper, since the objective here is to decipher whether there are systematic differences in starting U.S. wages, between these two groups of illegal immigrants. 7

8 Thus, the results reported above suggest that starting wages of male NIS-EWI immigrants: who were legalized through family ties, or through small-scale legalization programs, are not different from male illegal immigrants legalized through the 1986 IRCA. But, the former group received a significantly higher increase in wages after legalization, compared to the latter group. On the other hand, starting wages of NIS-EWI women are higher, and the legalization premium is not significantly different, from LPS women. Thus, it must be the case that female IRCA beneficiaries have had a larger legalization premium (i.e. wage gain from legalization) than male IRCA beneficiaries. There is evidence in the literature that, indeed, this is the case. Amuedo-Dorantes et al. (2007) estimate that IRCA increased wages of male immigrants by 9.3%, and female immigrants by 21%. Results in Amuedo-Dorantes et. al. (2007) and in this paper, suggest that increased supply of newly legalized immigrants restricted wage gains of males who were legalized through IRCA, but not females who were legalized through IRCA. In the results section, I discuss the possible reasons in detail. I also discuss the assumptions required for this inference. 2. Data This paper uses data from two different surveys: the New Immigrant Survey, or NIS (2003) and the Legalized Population Survey, or LPS (1992). NIS (2003) is a nationally representative sample of 8,573 new LPRs. They were interviewed between June 2003, and June 2004, after getting LPR status in the previous year. 4,402 (51.4%) of them were adjustees (those who were already in the U.S. on a non-immigrant visa, and changed their status to LPR) and 4,171 (48.6%) were new arrivees (those who arrived with a green card). Some of the adjustees who received green cards were previously illegal, and were legalized through family ties, or small-scale legalization programs. As explained earlier, in this paper the identification strategy is 8

9 to compare the change in wages of these previously illegal immigrants, to those who were always legal (NIS-AL immigrants). Therefore, the sample is restricted to adjustees (4402 immigrants). Out of all adjustees, 1579 immigrants have a valid wage observation for before and after receiving LPR status. I restrict the sample to individuals who are between 18 and 60 years old. After imposing this restriction, all relevant information is available for 1515 immigrants. The second data source used in this paper is the Legalized Population Survey. LPS (1992) surveyed only previously illegal immigrants, who were legalized by the 1986 IRCA. After imposing the same sample restrictions as above to the LPS data, I end up with a sample of 2600 previously illegal LPS immigrants. Table A1 in appendix presents the detailed sample selection criteria and how that changes the sample size for both NIS and LPS data. To identify previously illegal workers in the NIS (2003) I follow the algorithm suggested by Jasso, Massey, Rosenzwig and Smith (2008). I distinguish between two types of illegality: those who entered the country without documents (NIS-EWI immigrants) and those who overstayed their visa (NIS-OS immigrants). To identify EWI immigrants, I start with a survey question asking the respondents whether they entered the U.S. without inspection. Then, I use administrative data to identify NIS-OS immigrants. I use two sources of information in the administrative data: temporary visa status, and class of admission. Temporary visa status is not available for all immigrants. Jasso et al. (2008) suggest that this could be due either to the USCIS not knowing full histories, or statuses not being recorded in their computer system. Some of these immigrants have an unknown (UU) code, which Jasso et al. (2008) calls an euphemism for illegal status. Class of admission is another variable in the administrative data that can be used to identify previously illegal immigrants. Class of admission Z refers to immigrants who were admitted through legalization. Following Jasso et. al (2008) I assume an immigrant to be 9

10 previously illegal if either the temporary visa status, or class of admission, suggests they were previously illegal. Among those that were identified as previously illegal from administrative sources, some are NIS-EWI immigrants, as indicated by the survey question described above. Those illegal immigrants, who are not NIS-EWI immigrants, are classified as NIS-OS immigrants. Using this definition, 411(26.9% of all previously illegal immigrants) are NIS-EWI immigrants, and 97 (6.3% of all previously illegal immigrants) are NIS-OS immigrants. Taken together (33.2%) of the sample consists of previously illegal immigrants. This estimate is consistent with Jasso et al. (2008) who estimated, using NIS (Pilot) data, that between 31.7% and 34.8% of new LPRs were previously illegal Table 1 presents summary statistics. Columns one through three present: summary statistics for always-legal immigrants (NIS-AL), over-stayers (NIS-OS), and those who entered the U.S. without inspection (NIS-EWI). Column four presents summary statistics of LPS respondents. I start by comparing the average real hourly wages for different groups, both at their first U.S. job, and after receiving LPR status. All wages are in 2003 prices, and winsorized at 5% to reduce the effect of outliers. 5 At their first U.S. job, the average hourly wage for NIS-AL immigrants was $15.40, NIS-OS averaged $11.33, NIS-EWI averaged $8.25, and LPS respondents averaged $8.22. In their post-lpr job, NIS-AL respondents earned $20.87, NIS-OS earned $15.41, NIS-EWI earned $11.65, and LPS respondents earned $ Thus, based on average hourly wage, NIS-EWI immigrants, and LPS respondents, are similar, while the NIS- AL, and NIS-OS, groups seem to differ from all other groups. We observe a similar pattern in other economic, and demographic, dimensions. 5 If I do not winsorize the wages, the qualitative results do not change. I report the results based on un-winsorized wages in the robustness section. 10

11 Age, and gender, composition are similar across groups. At the time of interview (2003 for the NIS respondents, and 1992 for LPS respondents) the average ages of: NIS-AL, NIS-OS, NIS-EWI, and LPS immigrants were years, years, years, and years respectively. About 39% of NIS-AL, 46% of NIS-OS, 39% of NIS-EWI, and 36% of LPS respondents are female. At the time of interview, 78% of NIS-AL, 66% of NIS-OS, 60% of NIS- EWI, and 61% of LPS respondents were married. Not surprisingly, NIS-AL immigrants have substantially more education than the other groups: NIS-AL, NIS-OS, NIS-EWI, and LPS immigrants have years, years, 9.39 years, and 8.77 years of total education respectively. It is possible to determine the number of years of both source country education, and U.S. education, in the NIS data, but not in the LPS data. Therefore, I control for whether the immigrants had any U.S. education, which can be determined in both data sets. 35% of NIS-AL, 44% of NIS-OS, 29% of NIS-EWI, and 24% of LPS respondents have some U.S. education. There are substantial differences across groups in U.S. work experience as well. NIS-AL immigrants have an average of 4.87 years work experience in the U.S., while NIS-OS have 7.5 years. On the other hand, NIS-EWI immigrants have years of work experience, while LPS have years. Exact U.S. work experience cannot be determined for about 31% of respondents in the NIS sample, because they were not surveyed about gaps in employment. In these cases, I use potential U.S. work experience. I define potential U.S. experience: as the number of years between the first job in the U.S. and the current job. The assumption here is that they were working continuously since they first started working. The results discussed in this paper are not sensitive to this assumption, which I discuss further in the robustness section. 11

12 English proficiency was evaluated in NIS using a five-point scale (very good, good, fair, poor, and no English ability) and in LPS using a four-point scale. In regressions involving both NIS and LPS respondents, I combined the bottom two categories (poor English, and no English ability) in NIS into one. 84% of NIS-AL, and 62% of NIS-OS immigrants, have very good or good English fluency. On the other hand, only 32% of NIS-EWI, and 36% of LPS respondents, have very good or good English fluency. There are differences between data sets in the respondents countries of origin. The top three source countries for previously illegal NIS immigrants are Mexico (27.5%), El Salvador (27.7%), and Guatemala (9%). The top three source countries for LPS respondents are also Mexico (53%), El Salvador (18.4%), and Guatemala (7%). On the other hand, the top three source regions for NIS always-legal immigrants are India (19.3%), Europe and Central Asia (15.4%), and East Asia South Asia and the Pacific (8.3%). 3. Methods and Results I start with estimating the gains from legalization of previously illegal immigrants who were legalized on the basis of family ties or small scale legalization programs. Next, I estimate the difference in gains between immigrants legalized under small and large scale programs. 3.1 Estimating gains from legalization for NIS respondents The NIS collected information on the wages of immigrants: when they started working in the U.S., and after they received LPR status. I compare the wage gains experienced by previously illegal (NIS-EWI, and NIS-OS) immigrants, and always legal (NIS-AL) immigrants. This strategy is similar to LHH, with some important distinctions elaborated below. The starting wages for NIS-EWI, and NIS-OS immigrants, refer to wages they earned when they were staying and working in the U.S. illegally. It is important to note, that while this 12

13 assumption is clearly valid for NIS-EWI immigrants, it may, or may not, be valid for NIS-OS immigrants. For NIS-OS immigrants, it is not possible to determine when their visa overstay period started. In other words, NIS-OS immigrants entered the U.S. legally, and then at some point during their stay, lost their legal status, when their temporary visas expired. At some time after that, they were legalized, and received LPR status. However, it is not possible to determine if they were staying in the U.S. legally at the time of their first U.S. job. The before wage for NIS-AL immigrants, refers to the wage they earned when they started working in the U.S., with temporary worker visas. At the time of first U.S. job, the legal status can be accurately determined for the NIS-AL group (i.e. control group), and NIS-EWI group (treatment group 1). But for the NIS-OS group (treatment group 2) legal status cannot be accurately determined. Since the second treatment group (NIS-OS immigrants) may be contaminated, it is important to distinguish between the two treatment groups. The difference-in-difference estimate, of gains from legalization, is given below (ignoring the individual and country subscripts for simplicity). Here, the subscripts refer to NIS legal status of groups: NIS-EWI ( ), NIS-OS (, and NIS-AL (AL). Gaining LPR status may change labor market opportunities, of a previously illegal immigrant, for two reasons: moving from illegal immigrant to legal temporary immigrant status, and moving from legal temporary immigrant to LPR status. The NIS-AL immigrants experience only the second effect. Previous research suggests that even the second reason may have substantial labor market implications. Legal immigrants on temporary work visas (such as H1B) face significant frictions in the labor market. Achieving LPR status may result in significant wage gains, even for workers who were legal but working on a temporary work 13

14 visa (Mukhopadhyay and Oxborrow 2012, Gass-Kandilov 2007). 6 The difference-in-difference structure will difference out the second effect, to identify the first effect. Thus the parameter of interest here is, strictly speaking, the wage return to a change from illegal immigrant to temporary but legal immigrant. Conceptually this is somewhere in between the return to temporary protected status discussed in Orrenius, and Zavodny (2015) and return to an LPR status discussed in the broader literature. The regression equation is given by Here, is the change in (log of) real hourly wage, of individual i. represents a vector of control variables: age, gender, marital status, years of source country education, U.S. specific human capital (such as U.S. education, and U.S. work experience), and English ability. 7 is an indicator for change in legal status (from NIS-EWI to LPR); is also an indicator for change in legal status (from NIS-OS to LPR). denotes cohort effect (the year of first entry into the U.S.) and is a source country/region fixed effect. I allow return to human capital acquired in the U.S. to depend on legal status. For example, the return to human capital accumulated as a NIS-EWI (NIS-OS) immigrant, may be different from return to human capital accumulated as a NIS-AL immigrant. LHH do not allow for this possibility. I do not include year of first U.S. job. If I include U.S. work experience, then, even with longitudinal data, the time effect is not separately identified. KCC (2002) use dummies for the year in which an immigrant first started working in the U.S., but they exclude work experience from their specification. Borjas (1989) on the other hand, includes years since migration (he 6 It should be noted that both of these papers focus on individuals who gained LPR status through employment based green cards, and who may have more to gain from less of labor market friction. While neither of these papers control for previously illegal status of temporary immigrants, the share of previously illegal immigrants, who gained LPR status through employment based green cards, is likely to be negligible. 7 No English is the omitted category in the regression. 14

15 did not have a measure for U.S. work experience in his data) but no year dummies. KCC (2002) found that while the magnitude of entry-into-the-u.s.-job-market-year effect declines with time spent in the U.S., there may still be some residual effects. Since returns to experience and if they differ by legal status is one of my questions of interest, it is my preferred specification to include years of U.S. work experience, and not the year of first U.S. job. Kahn (2010), reports that U.S. natives who graduate during recessions, earn less over their lifetimes than those who graduate in normal times. This effect is significant and persistent. To account for economic conditions at the time of first U.S. job, I include a dummy variable if the respondent first started working in the U.S. in a recessionary year. I use source country dummies, to control for unobserved heterogeneity across countries. To control for entry year effect, or cohort effect, I include dummies for the year of U.S. entry. Hanson and Spilimbergo (1999) show that apprehension of potential illegal immigrants at the U.S. borders, rise with a rise in the U.S.-Mexico wage differential. Therefore, it is plausible that those immigrating to the U.S., during a U.S. recession, may have different abilities than individuals who immigrate during a boom. Also, it may be noted that year of entry dummies are included in the regressions; these are often close of year of first U.S. job. This specification requires an identifying restriction: that the aggregate changes in the labor market are same for always legal, and previously illegal workers, who started working in the same year. Most papers using difference-in-difference framework impose a similar restriction. Table 2 shows the results from estimating equation (1). In the first column, I only include change in legal status dummies, without any controls. Results show that the wage gain is 10.6% more for NIS-EWI immigrants, compared to NIS-AL immigrants. The wage gain for NIS-OS immigrants is not statistically different from NIS-AL immigrants. Next (column two) I include 15

16 the full set of controls: gender, marital status, years of source country education, and U.S. specific human capital (such as U.S. education, and U.S. work experience), English Ability, year of entry fixed effects, and source country fixed effects. 8 The resulting coefficient estimates, shows that legalization did not increase the wages of either NIS-EWI, or NIS-OS immigrants. In fact, the point estimates do not have the expected (positive) sign, and are insignificant. This is consistent with LHH. In the third column, returns to human capital acquired in the U.S. are allowed to differ across legal status. LHH do not consider this possibility. Estimates show that returns to human capital accumulated in the U.S. by illegal immigrants, are lower than returns to human capital accumulated in the U.S. by always legal immigrants. For example, always legal immigrants with any U.S. education earn 14.8% more than always legal immigrants without any U.S. education. However, NIS-EWI immigrants with any U.S. education earn 7.0% less than (marginally significant with a p-value of 0.09) NIS-EWI immigrants without any U.S. education. Accetturo and Infante (2010) found a similar result for illegal immigrants in Italy. For an NIS-AL immigrant, 10 years of U.S. work experience increases wage by 61.4%. On the other hand, for a NIS-EWI immigrant, 10 years of U.S. work experience increases wage by 40.5%. The difference (20.9%) is statistically significant (p-value 0.007). This result is consistent with KCC (2002) who found that returns to experience of previously illegal immigrants in LPS was lower than that of NLSY respondents. For NIS-OS immigrants there is no difference in returns to education but the returns to work experience are significantly lower than for NIS-AL immigrants. 8 The effects of time constant controls should be differenced out in a first difference regression. I nonetheless include them, since the intercepts may be different for different groups. For example, immigrants from different countries may have different baseline wage growth rates. Also this specification is consistent with LHH. Qualitative results are robust to the exclusion of time constant regressors. 16

17 Allowing for differential returns to human capital, changes the estimated legalization premium. Estimates show that the wage growth of NIS-EWI immigrants is 18.2% more than that of NIS-AL immigrants. I do not find any evidence of a legalization premium for over-stayers. Columns four, and five, present the results for male, and female, subsample respectively. Results for each reported gender are similar to the full sample. Results in Table 2, show that previously illegal immigrants (especially NIS-EWI immigrants) get a lower return to the human capital they accumulated in illegal status, even after getting LPR status. There could be a number of explanations behind this result. First, it may be that employer sanctions make it difficult to document work experience (to a prospective new employer) accumulated as an NIS-EWI immigrant. Since the current employer knows that the outside opportunities for a previously NIS-EWI immigrant are limited, they do not increase his wage. Second, it may be that illegal immigrants work in occupations that are bad matches, and/or where returns to experience are low. Third, it may be that NIS-EWI immigrants may be acquiring a different level of education (high school degree for NIS-EWI immigrants, vs. professional degree for NIS-AL immigrants) compared to NIS-AL immigrants. As shown in Table 1, on an average, NIS-EWI immigrants have 8.03 years of source country education, compared to years of source country education for NIS-AL immigrants. One concern behind the second, and third explanations, is that maybe the relatively low education level (and not per se the illegal status) is the reason behind the lower returns to human capital. It has been previously documented, that returns to experience are higher for more educated individuals (Farber and Gibbons, 1996). To ascertain whether the differential in returns to experience (and therefore gain from legalization) depends on the level of source country education, I re-estimate equation (1) with only those immigrants who have 12 or less years of 17

18 source country education. Sample size reduces to 726, from 1529, as a significant part of the always legal population has more than 12 years of education. This also makes the always legal, and previously illegal, groups more comparable. Results from this regression, are presented in Table 2, column six. Compared to the full sample, the returns to both education, and experience, are lower in this sample, but the differential in returns persists. For a NIS-AL immigrant (with 12 or less years of education) 10 years of U.S. work experience increases wage by 44.6% ; but for a NIS-EWI (with 12 or less years of education) 10 years of U.S. work experience increases wage by 31.6%. The estimated gain from legalization, for NIS-EWI immigrants, in this sample is 13.6%. This is smaller than the estimate for full sample (18.2%) but still substantial, and statistically significant. This estimate is similar to return from temporary protected status (13.1%) reported in Orrenius, and Zavodny (2015). I performed a number of additional robustness checks. The results are presented in appendix table A2. First, I discussed in the data section, that wages are winsorized at 5%. To check the sensitivity to this assumption, Table A2 column one presents the results for unwinsorized wages. The qualitative results are similar to the results described above, although the point estimates are higher. Column two presents the results for those immigrants whose exact experience can be determined from data (1050 immigrants, or about 69% of the original sample). Again, the qualitative results are similar to the results for the full sample. Next, I restrict my sample to full-time workers. In this case, I include only those who were working 40 hours or more per week, both in their first U.S. job, and in their post LPR job. Again, the qualitative results, in column three, are similar to the results for the full sample. Finally, I include only one treatment group, and the control group, at one time. Column four, presents the results comparing 18

19 NIS-EWI, and NIS-AL immigrants (i.e. NIS-OS immigrants are excluded from the sample). Column five, presents the results comparing NIS-OS, and NIS-AL immigrants (i.e. NIS-EWI immigrants are excluded from the sample). Results in columns four, and five, are similar to the pooled regression results. 3.2 Comparing the legalization premium for NIS and LPS respondents Next, I explore whether the gains from legalization are different: for immigrants legalized through family ties, or small-scale legalization programs (such as NIS-EWI immigrants) from those legalized in a large-scale amnesty program (such as the IRCA). There are a number of papers (discussed in the introduction section) that use IRCA variation to estimate gains from legalization. For identification, most of these papers rely on the change in wage between the time of application, and at the time of interview (which was up to three years after receiving LPR status). However, the NIS did not collect information on wage of respondents at the time of application for legalization. The NIS only contains information on wages when the immigrants first started working in the U.S., and in post LPR jobs. Therefore, to implement a difference-in-difference regression, I compare the change in wages (of previously illegal NIS respondents, between their first U.S. job, and post LPR job) to the corresponding change in wage of those who were legalized through the 1986 IRCA. I estimate the following regression equation (2) In the above equation, represents the change in (log of) real hourly wage, between the first U.S. job, and post LPR job, of previously illegal immigrants. represents changes in 19

20 the control variables. 9 The treatment groups are two previously illegal groups of NIS respondents: NIS-EWI, and NIS-OS immigrants. As discussed earlier, these groups were legalized through small-scale legalization programs, or through family ties. LPS respondents, who were legalized through the 1986 IRCA, are the control group in this regression. is an indicator of change in legal status: from NIS-EWI, to LPR. is also an indicator for change in legal status: from NIS-OS, to LPR. Since both the treatment, and the control group, consist of only previously illegal immigrants, the question of a differential return to human capital by legal status does not arise. The coefficient ( represents the difference in wage gains from legalization for NIS-EWI (NIS-OS) immigrants legalized through family ties, or small-scale programs, compared to those legalized through the 1986 IRCA. Before discussing the results, it is important to discuss two issues. First, the parameter of interest in this section is how the wage return to a change from illegal immigrant to LPR differs by the scale of legalization. It should be noted that, strictly speaking, this is not comparable to the parameter estimates from section 3.1 without further assumptions. However, if the wage returns to a movement from temporary legal immigrants to LPR is zero or negligible (we do know that that is not true for employment based (EB) immigrants from Mukhopadhyay and Oxborrow 2012, Gass-Kandilov 2007) then in equation 2 should be less than in equation 1. If the wage returns to a movement from temporary legal immigrants to LPR is non-zero then may or may not be less than depending on the size of wage return from legal immigrants to LPR. Second, it is important to discuss assumption required for identification. One potential problem is that post-lpr interviews were conducted in 1992 for LPS respondents, and 2003 for 9 The effects of time constant controls should be differenced out in a first difference regression. I nonetheless include them, since the intercepts may be different for different groups. For example, immigrants from different countries, may have different baseline wage growth rates. 20

21 NIS respondents. The macroeconomic conditions prevalent in 2003, may be different from those prevalent in For example, the national unemployment rate was 7.5% in 1992, but it was 6.0% in However, I am not comparing the levels of wages. I am comparing the changes in wages between first U.S. job, and post-lpr job. Even then, the change in economic condition between the first U.S. job, and post-lpr job, may be different for each immigrant. To account for the role of changes in the macroeconomic environment, on the change of wages, I control for change in the national unemployment rate: between first U.S. job year, and post-lpr interview year. This variable is defined as: (unemployment rate at time of post LPR interview) (unemployment rate at time of first U.S. job). In this sample, this variable (i.e. the difference in unemployment rate) varies between -6.0, and 2.7 percentage points. The mean of this variable is percentage points. The identifying assumption is that the change in the unemployment rate (along with other control variables, such as entry year dummies) is enough to absorb the effect of change in macroeconomic environment. Also, it is worth noting that this structure implicitly assumes that legalization programs do not affect the unemployment rate. If they do, for example, if mass legalization such as the IRCA increased the unemployment rate, by bringing out a large number of workers from the shadow labor market, then this would lead to a downward bias in the estimated difference between the two groups. Another Potential problem in comparing the wage gains of LPS, and NIS respondents, is that skill prices in the U.S. may have changed between 1980 and 2000 (Lubotsky, 2009). I do not explicitly allow change in skill prices in my empirical specifications. However, as long as the rates of change during the 1980s and 1990s were not different, their effects will be differenced out. Furthermore, entry year dummies, and checking robustness after restricting the sample to immigrants with high school or less education, should reduce this problem. 21

22 The regression results are presented in Table 3. Column one shows the results without any control variables. The increase in the wage after legalization is 7.1% higher for NIS-EWI immigrants, than IRCA beneficiaries (i.e. LPS respondents). On the other hand, the legalization premium for NIS-OS immigrants is not different from IRCA beneficiaries (i.e. LPS respondents). In column two, I add the full set of controls. They include change in national unemployment rate between first U.S. job year and post LPR interview year, and a plethora of individual level controls such as: age, gender, marital status, total education, whether the respondent had any U.S. education, U.S. work experience, if they started working in a recession year, and English ability. 10 Regressions also include entry year dummies, and source country dummies. The control variables have the expected signs: U.S. work experience increases wage, individuals with more education earn more, and individuals with better English skills earn more. However, U.S. education does not lead to a significant increase in wages, which is consistent with results on NIS-EWI immigrants in Table 2. An increase in the difference in unemployment rate between the first U.S. job year, and post LPR interview year, reduces wage growth. Estimates suggest that for a one percentage point increase in the change in unemployment rate (between first U.S. job year, and post LPR interview year) reduces wages by 1.61%. Adding in the controls does not change the qualitative result obtained using simple mean difference-in-difference. Estimates suggest that the increase in wages following legalization was 11.7% more for NIS-EWI, than LPS respondents. Again, there was no difference in the legalization premium of NIS-OS, and LPS respondents. Next, I estimate the difference in legalization premium separately for male, and female samples. Column three presents results for the male sample. It shows that the post-legalization wage increase is 18.3% more for male NIS-EWI immigrants, compared to male IRCA 10 Poor/No English ability is the omitted category in the regressions. 22

23 beneficiaries. However, column four shows, there is no difference in legalization premium between female NIS-EWI immigrants, and female IRCA beneficiaries. This difference between male and female samples may seem puzzling, especially given that there is no difference in legalization premiums of male, and female, NIS-EWI immigrants (Table 2, columns four and five). Thus, it must be the case that female IRCA beneficiaries have experienced a larger legalization premium (i.e. wage gain from legalization) that male IRCA beneficiaries. And there is evidence in the literature that this is indeed the case. Amuedo-Dorantes et. al. (2007) estimate that IRCA increased wages of male immigrants by 9.3%, and female immigrants by 21%. I further discuss this issue at the end of this section. Even though the total years of education is similar for NIS-EWI, and IRCA beneficiaries (9.39 years, vs years), the difference is statistically significant. Next, I restrict the sample to immigrants with 12 years or less of education, and estimate the difference in legalization premiums of previously illegal NIS respondents, and LPS respondents. Now the difference in average education between the two groups is not significant anymore. I again estimate regressions separately for males and females, as the results in columns three and four suggest substantial differences in the results for males and females. Column five shows that in the lowskill male sample the post-legalization wage increase was 12.0% more for NIS-EWI immigrants, compared to IRCA beneficiaries. Again, in column six, there is no difference in the legalization premium between female NIS-EWI immigrants, and female IRCA beneficiaries. Thus, the results suggest a difference in the legalization premium: between those are legalized through small-scale programs, or family ties (i.e. NIS-EWI immigrants) and IRCA beneficiaries. This is true in the male sample, but not in the female sample. Next, I explore the potential reasons behind this difference in legalization premium between NIS-EWI immigrants, 23

24 and IRCA beneficiaries. One reason behind the increased legalization premium may be an increased illegal status penalty when they started their first U.S. job. For example, if the employer sanction provisions in the 1986 IRCA increased the illegal status penalty, then the NIS respondents will have a lower wage in their first U.S. job. This in turn may explain why they have a bigger legalization premium. Thus, I compare the starting wages (i.e. wages on their first U.S. job) of illegal immigrants from NIS and LPS samples. The regression equation is given by Where is the real hourly wage, in the first U.S. job. represents a vector of control variables: age at the time of first job, total education (in years), gender, marital status, and English ability. I also include source country dummies, and year of first U.S. job dummies. Table 4 presents the results from estimating equation (3). Again, I start by comparing the (log of) real hourly wages, in first U.S. jobs, without any control variables. The estimates in column one, show that in their first U.S. job, NIS-EWI immigrants earn 5.1% less than LPS respondents, while NIS-OS immigrants earn 22.5% more than the LPS respondents. In column two, the full set of controls is added. The control variables have expected signs. Wage increases with age, education, and English ability. Females earn less than men. Once the full set of controls are added, there are no differences between the NIS-EWI immigrants, and LPS respondents. The difference between NIS-OS immigrants, and LPS respondents, become statistically insignificant as well. Although in this case, the t-stat is 1.60, approaching significance at the 10% level. Next, I estimate equation (3) separately for males and females. Estimates in column three show that there is no difference in the starting wages of male NIS-EWI, and male LPS 24

25 respondents. In column four, starting wages of female NIS-EWI immigrants are significantly higher than starting wages of female LPS respondents. It is worth nothing, that the results in Tables 2 and 3 are about comparing the change in wage between first U.S. wage, and post-lpr wage. Thus, individual level unobserved heterogeneity was differenced out. In Table 4, the comparison is at the level of wages, thus individual specific unobserved heterogeneity may be a concern. In other words, represent effects of changes in the illegal status penalty after the 1986 IRCA, and unobserved difference between NIS, and LPS respondents. In this case, these two effects cannot be separately identified. For example, it may have happened that IRCA did increase the illegal status penalty, but NIS-EWI immigrants are better than the LPS respondents in unobservable ways. In that case, these two effects may cancel each other, rendering them insignificant. However, the pattern of results across gender, suggests that it is an unlikely scenario. The female results suggest that an increase in illegal status penalty in the post-irca period is unlikely. LHH also find a similar result. While separate identification of changes in the illegal status penalty, and unobservable characteristics is ideal, it is not absolutely necessary for the purpose of this paper. The objective here is only to check whether there are systematic differences in the starting U.S. wages of NIS-EWI, and LPS respondents. The results suggest that there is no difference in male sample, but there is in female sample (most likely because the female NIS-EWI immigrants are different from female LPS respondents in unobservable ways). Thus, these results suggest that starting wages of male NIS-EWI immigrants (legalized through family ties, or small-scale legalization programs) are not different from male LPS immigrants (legalized through the 1986 IRCA). But, the former group has a significantly higher increase in wages after legalization, compared to the latter group. On the other hand, starting 25

26 wages of female NIS-EWI immigrants are higher, and the legalization premium is not significantly different from female LPS respondents. Taken with the results reported in Amuedo- Dorantes et al. (2007) this suggests that increased labor supply of newly legalized immigrants, restricted the wage gains of males, but not females, legalized through IRCA. Next, I discuss why this might be the case. KCC (2002) postulated that a mass legalization increases the relative supply of legal workers (compared to illegal workers) in occupations frequented by illegal workers. A number of other studies have documented that certain occupations have a high density of illegal immigrants (Gill and Long 1989; Taylor 1992; Massey et al. 1987; Kossoudji and Cobb-Clark 1996) possibly due to network effects (Patel and Vella 2013). Passel (2009) estimates that even though illegal immigrants constituted only about 4.9% of the total civilian labor force, 40% of all brick-masons, block-masons, and stone-masons are illegal immigrants. Similar concentrations can be found in other occupations, such as drywall installers, ceiling tile installers and tapers (37%), roofers (31%), construction helpers (28%), and construction laborers (27%) (Passel, 2009). These occupations are male dominated. In general, more than 10% of all workers employed in farming, cleaning, construction, and food preparation industries are illegal immigrants. Furthermore, Passel (2006) estimates that the labor force participation rate of male illegal immigrants is 94%, but the labor force participation rate of female illegal immigrants is only 54%. Previous papers also suggest that legalization does not lead to significant changes in employment probabilities. While some papers find a small decline in employment rate (Amuedo-Dorantes, Bansak, and Raphael 2007; Amuedo-Dorantes and Bansak 2011), others find a small increase in employment rate especially among women (Pan, 2010). Thus there is no reason to think that the labor force participation rates reported in Passel (2006) would change significantly after legalization. 26