1. The Labor Supply of Undocumented Immigrants: Towards an Assessment of the Impact of Status Regularization George J. Borjas Harvard University August 6, 2015

2. 1. Regularizing the status of undocumented workers DHS estimates that 11.4 million undocumented persons reside in the United States (as of January 2012). Congress is considering proposals to regularize the status of the undocumented population and provide a “path to citizenship,” while President Obama has issued executive orders that grant some form of amnesty to about half of this population.

3. 2. Evaluating the impact of a regularization Predicting the impact of the regularization on the inflows and outflows of funds into any government program immediately runs into an impenetrable roadblock: We know almost nothing about the economic status of the 11.5 million undocumented persons in the United States. We know little about their employment history (so, for example, we do not know how many would potentially meet the 40-quarter eligibility requirement for Social Security). We know little about their earnings history, about their financial contributions to various government programs, or about how those earnings and contributions would change if their status were regularized.

4. 3. What this paper does This paper represents a first attempt at providing some of the requisite background information involved in conducting any such future evaluation. In particular, the paper provides a comprehensive empirical study of the labor supply behavior of undocumented immigrants in the United States. The analysis is based on data drawn from CPS files that attempt to identify the “likely undocumented” population at the individual level. This identification is an extension of the methodology employed by DHS to estimate the size of the undocumented population. I have the 2012-2013 CPS files created by the Pew Research Center, and have “reverse engineered” the method to all 1994-2014 March CPS files.

5. 4. Main findings Labor supply, men: Undocumented men > Legal immigrant men > Native men.’ Labor supply, women: Native women > Legal immigrant women > Undocumented Women. The undocumented are always at the “extreme” in labor supply. Labor supply of undocumented men rose dramatically between 1994 and 2014. The relative employment rates of undocumented women also rose, but at a much slower rate. The labor supply of undocumented men and women is much less responsive to wage changes than that of legal immigrants, which in turn is less responsive that that of native persons

6. 5. Undocumented immigration (DHS estimates) Jan. 2000: 8.5 million. Jan. 2005: 10.5 million. Jan. 2007: 11.8 million. Jan. 2008: 11.6 million. Jan. 2010: 10.8 million Jan. 2011: 11.5 million Jan. 2012: 11.4 million 25% live in California; 16% in Texas; 59% come from Mexico.

7. 6. Estimating size of undocumented population Residual Method. Originated in 1987 in work by Jeffrey Passell (now at PEW Research Center) and Robert Warren (Chief Statistician of INS at the time). We know how many “green cards” have been given out. We can calculate expected size of legal immigrant population by using mortality rates and age at migration, and accounting for out-migration. We have enumerations of number of foreign-born in country (Census, ACS, CPS). Adjust the number of foreign-born for persons in US with student visas, H-1Bs, etc. Difference between the adjusted number enumerated and the expected number of legal immigrants is the DHS estimate of the number of undocumented immigrants.

8. 7. Key assumption used in estimating size of illegal population by DHS

9. 8. Pew method to identify undocumented at micro level in CPS Created by Jeff Passell. Immigrants entering the U.S. before 1980 are legal. Immigrants entering as refugees are identified as such based on country of origin and year of entry. Immigrants with temporary visas (e.g., students, diplomats, high- tech workers) are legal. Immigrants in some occupations are legal (working for the government, licensed occupations, veterans). Immigrants receiving some types of public assistance are legal. Some family relationships extend legal status to relatives. Residual number of undocumented immigrants is larger than DHS estimates. So use a “probabilistic assignment process.” Passell then creates a new weight so that aggregates match DHS numbers. I have the micro files for March 2012 and March 2013 CPS.

10. 9. Pew: fraction of legal and undocumented immigrants in population

11. 10. Reverse engineering the process Legal if citizen. Legal if arrived before 1980. Legal if receives Social Security, SSI, Medicaid, Medicare, or Military Insurance. Legal if works for government or is a veteran. Legal if lives in public housing. Legal if works in occupation that requires some type of licensing (Examples: Legislators, Accountants, Architects, RNs, Teachers, Inspectors of Agricultural Products, Lawyers, Air Traffic Controllers). Legal if certain family members are legal immigrants. The residual group is undocumented. I have added the undocumented identifier to all March CPS files between 1994 and 2014.

12. 11. Fraction of undocumented immigrants in population

13. 12. Means of key variables, pooled 2012-2013

14. 13. Predicted age-employment profile, Pew, men

15. 14. Predicted age-employment profile, Pew, men, adjusted for education

16. 15. Predicted age-employment profile, Pew, women

17. 16. Predicted age-employment profile, Pew, women, adjusted for education

18. 17. Differences in labor supply (relative to natives)

19. 18. Labor supply trends, men, 1994-2014

20. 19. Labor supply trends, men, taking out year effects

21. 20. Labor supply trends, women, taking out year effects

22. 21. Labor supply elasticities, men Uses data aggregated at skill-year level. Log wage on r.h.s. is mean log wage of skill group in a particular year. Last column includes state-year fixed effects (so aggregated at skill-state-year level).

23. 22. Hours of work and wages, native men

24. 23. Hours of work and wages, legal immigrant men

25. 24. Hours of work and wages, undocumented men

26. 25. Labor supply elasticities, women Uses data aggregated at skill-year level. Log wage on r.h.s. is mean log wage of skill group in a particular year.

27. 26. Towards an evaluation of regularization programs This is just a first step. A “stylized fact” implied by the research is that undocumented men “come to the United States to work.” But that stylized fact does not apply to undocumented women. Because undocumented men work and their labor supply is insensitive to changes in the wage, it is almost surely the case that many will have worked 40 quarters by the time they reach retirement age. It is unclear, however, whether their employers collected FICA taxes. The implications for the Social Security system are yet to be determined. A similar analysis of earnings determination in the undocumented sample needs to be conducted, although preliminary data indicate that undocumented workers earn far less than both legal immigrant and natives, even after adjusting for observable skills.