Abstract
We propose an alternative method for identifying undocumented immigrants in public use household data. Our approach departs from the commonly used likely unauthorized (LU) method in that we do not resort to classifying individuals on the basis of their specific geographic origin. We instead consider health insurance status of different family members coupled with the Affordable Care Act insurance mandates, referring to this strategy for sorting individuals by documentation status as the Affordable Care Mandate (ACM) approach. The LU method is generally limited to identifying individuals originating from Latin America. By contrast, the ACM approach enables us to identify individuals who are apparently undocumented and originate from all regions of the world. We also compare our results to the Reconstructed Pew Algorithm (RPA). The RPA is not limited in terms of which adults it can classify, but it does involve making a larger number of judgment calls with respect to the final design of the algorithm. The ACM has its limits in that it can only be applied to individuals who have US-born children who currently reside in the household. Using the 2015–2017 March Supplements to the CPS, we test the ACM against the LU and RPA approaches by estimating simple models of labor force participation and fertility by presumed legal status. Using the alternative methods, we obtain similar, though not identical, results and argue that the ACM approach holds promise for studying the behavior of undocumented immigrants that does not resort to ethnic profiling and uses fewer judgment calls.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
The estimated foreign born population living in the USA in 2012 was 34 million people. Approximately 22.6 million of these people were legal residents, leaving the remaining 11.4 million people as the estimated population of undocumented immigrants (Baker and Rytina 2013). Newer estimates that combine the residual method with additional information (e.g., Warren 2018) estimate the unauthorized population to have declined to 10.8 million by 2016.
One might also question the ability to truly assign documentation status even if accessing the restricted version of the SIPP. A number of assumptions/judgments still need to be made to convert, for example, visa status upon entry to the USA into documentation status today.
Note that DACA-status individuals are not eligible to participate in the health insurance exchanges either (www.healthcare.gov/immigrants/immigration-status). While DACA and non-DACA undocumented can purchase private insurance that is not linked to the exchanges, the mandate does not require either group to be insured. Furthermore, private non-exchange linked health insurance tends to be relatively expensive.
We do not report on these results as they are not the focus of this study. However, they are available from the authors upon request.
The occupations we chose that require licensing include physicians and surgeons, registered nurses, dentists, pharmacists, air traffic controllers, lawyers and judges, accountants and auditors, personal financial advisors, architects and civil engineers, counselors and social workers, and K-12 and special education teachers.
One might wonder why we chose to work with the CPS rather than the American Community Survey (ACS). The ACS has a larger sample and contains the variables needed to implement our algorithm. We argue that the algorithm is likely to improve the higher the penalty for noncompliance with the mandate (penalty is higher in later years) and with the greater the passage of time since the mandate was put in place given that information dissemination has improved with time since passage. At the time of the writing of this paper, the 2017 CPS was available while the 2017 ACS was not yet released, making the case for using the CPS over the ACS.
Information on the current penalty for not having insurance can be found on the healthcare.gov website (https://www.healthcare.gov/fees/fee-for-not-being-covered/).
To check for a chilling effect on insurance coverage, we obtained a measure, at the state level, of internal immigration enforcement measured by the Immigrant Legal Resource Center (ILRC) at https://www.ilrc.org/immigration-enforcement through the use of 287(g) agreements. These are contracts between local law enforcement agencies and Immigration Customs and Enforcement (ICE) that allow local law enforcement to perform ICE tasks. We regressed the share of category 2 parents (uninsured foreign born non-citizen parents with domestically born uninsured children/all foreign born non-citizen parents) on the intensity of enforcement as measured by the ILRC. The coefficient on enforcement at the state level is 0.02 with a probability value of 0.135. These results do not point to a chilling effect on the acquisition of insurance due to higher levels of internal enforcement, though we cannot rule it out.
The three groups are constrained in order to make them comparable. Since the LU method limits the age range to 16–45, for this comparison, we only include this age range in the three groups. Since the ACM method requires that the family unit contain US-born children, we constrain the three groups to be likewise for the purposes of this comparative exercise.
Appendix Table 9 produces the ACM sample showing finer cuts beginning with the first set of columns, which display the sample of uninsured foreign born parents with insured children under the age of 18. The middle columns strip away all observations where the children were born abroad. In the last column, we obtain the final sample used in our analysis, the sample that we term as undocumented, by imposing the final condition that the children in these family units carry health insurance.
One might ask why we are conditioning the labor force participation regressions on being a parent already. This is because the ACM technique for assessing whether an immigrant is undocumented relies on the individual being the parent of at least one child less than 18 and born in the USA To not introduce a bias with respect to the definition of one or another group, we condition the estimation model on being a parent when comparing the techniques.
We choose to use a linear probability model over probit since linear probability models are less sensitive to specification errors. Standard errors are clustered at the state level.
While we cautioned that our sample is selected (as it is limited to adults with an own child in the family less than 18), the labor force participation estimates by Borjas (2017b) using the RPA method are not limited to individuals with children, yet he finds the same overall results for undocumented men and women relative to men and women who are documented.
Lack of insurance and the undocumented indicator (Ui) are positively correlated by virtue of the construction of the undocumented immigrant indicator. One would expect that fertility would be negatively associated with lack of insurance. This suggests that lack of explicitly incorporating an insurance status variable for all parents may be causing the coefficient on Ui to be positively biased in the fertility model results in Table 5.
References
Amuedo-Dorantes C, Lozano F. On the effectiveness of SB1070 in Arizona. Econ Inq. 2015;53(1):335–51.
Baker, B. and Rytina, N.. “Estimates of the unauthorized immigrant population residing in the United States: January 2012”, Population estimates, Office of Immigration Statistics, Department of Homeland Security;2013
Bohn S, Lofstrom M, Raphael S. Did the legal Arizona workers act reduce the state’s unauthorized immigrant population? Rev Econ Stat. 2014;96(2):258–69.
Borjas, G.. “The earnings of undocumented immigrants,” NBER working paper No 23236;2017a
Borjas G. The labor supply of undocumented immigrants. Labour Econ. 2017b;46:1–13.
Capps R, Bachmeier JD, Van Hook J. Estimating the characteristics of unauthorized immigrants using U.S. Census data: combined sample multiple imputation. AAPSS. 2018;677(1):165–79.
Flores SM. State dream acts: the effect of in-state resident tuition policies and undocumented Latino students. Rev High Educ. 2010;33:239–83.
Kaushal N. Amnesty programs and the labor market outcomes of undocumented workers. J Hum Resour. 2006;(41):631–47.
NILC (National Immigration Law Center). (2014). “Immigrants and the Affordable Care Act (ACA)”, Last revised January 2014, URL: http://www.nilc.orgimmigrantshcr.html.
Passel, J.. “The size and characteristics of the unauthorized migrant population in the US: estimates based on the March 2005 current population survey”, Research report, Pew Hispanic Center; 2006
Passel, J. and Cohn, D.. “A portrait of unauthorized immigrants in the United States”, Research Report, Pew Hispanic Center; 2009
Passel J, Cohn D. Unauthorized immigrant totals rise in 7 states, fall in. States: Decline in Those from Mexico Fuels Most State Decreases”, Research Report, Pew Research Center; 2014. p. 14.
Ross J. The myth of the ‘anchor baby’ deportation defense. Washington Post, August. 2015;20:2015.
Warren, R. (2018), “The US undocumented population fell sharply during the Obama era: estimates for 2016.” URL: http://cmsny.org/publications/warren-undocumented-2016/
Warren R. Democratizing data about unauthorized residents in the United States: estimates and public-use data, 2010 to 2013. J Migr Hum Secur. 2014;2(4):305–28.
Warren J, Warren R. Unauthorized immigration to the United States: annual estimates and components of change, by state, 1990 to 2010. Int Migr Rev. 2013;47(2):296–329.
Watson T. Inside the refrigerator: immigration enforcement and chilling effects in Medicaid participation. Am Econ J Econ Pol. 2014;6(3):313–38.
Acknowledgements
We are very grateful to three referees and to participants at the 2018 Western Economics Association International meetings for their comments and suggestions.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of Interest
On behalf of all authors, the corresponding author states that there is no conflict of interest.
Appendix
Appendix
Rights and permissions
About this article
Cite this article
Mengistu, D., Pozo, S. & Sarnstrom, T. An Alternative Approach for Identifying a Hidden Immigrant Population. J Econ Race Policy 2, 121–135 (2019). https://doi.org/10.1007/s41996-018-0020-x
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s41996-018-0020-x