In “How Do Tougher Immigration Measures Affect Unauthorized Immigrants?” Amuedo-Dorantes et al. make an important contribution to the growing literature on the effects of state-level policies targeting unauthorized immigrants. Using a new data set on unauthorized migrants who have returned to Mexico, the study measures the effect of E-Verify mandates on outcomes, including access to legal, government, and health services, cross-state mobility, fear of deportation, and intent to return to the United States in the future. There are two main findings. Although living in a state with E-Verify does not affect migrants’ access to services, it appears to heighten deportation fears and to reduce cross-state mobility among migrants returning voluntarily. Among deportees, E-Verify also appears to result in significantly lower intent to return to the United States in the future. In this comment, I note the limitations of these data in addressing the effects of employer mandates and argue for a less-restrictive identification scheme.
E-Verify mandates, such as the one stipulated in Arizona’s 2007 Legal Arizona Workers Act (LAWA), oblige employers to check the work authorization of their new hires. Presumably, most employers follow up by firing a worker who is flagged as ineligible. The effect of E-Verify is therefore to reduce the demand for unauthorized immigrant workers, depressing their employment rates and wages (Amuedo-Dorantes and Bansak 2011).1 These direct effects may have further consequences for migrants, including increased self-employment, off-the-books employment, or out-migration to states with friendlier policies (Lofstrom et al. 2011). It may also increase the incidence of document and identity fraud as unauthorized workers try to pass themselves off as legal residents (Rosenblum 2010).
Holding other conditions constant, E-Verify programs make unauthorized immigrants worse off, lowering household income and increasing their families’ eligibility for assistance, which should increase their access to services, not decrease it as the study argues.2 In fact, an important drawback of employment verification programs is that they may throw more people onto welfare rolls. Although unauthorized immigrants are not eligible for most assistance programs, there are exceptions, such as pregnant women or mixed-status families with U.S.-born children.
E-Verify programs should also increase cross-state mobility, not decrease it as Amuedo-Dorantes et al. find, because affected migrants will face relatively better employment opportunities elsewhere.
And E-Verify, on its own, should not affect the probability of being deported because the federal government does not send out agents to look for migrants who have been flagged by E-Verify as unauthorized to work.3 State and local authorities cannot deport unauthorized immigrants, although they can turn them over to federal authorities.
Given these theoretical predictions, what can explain the empirical results? A number of factors could be confounding identification in the analysis. First, as the authors note in passing, other state and local enforcement programs and policies were coming into effect during this time, such as Secure Communities and 287(g), especially in Arizona. Both involve the cooperation of local law enforcement with federal authorities, which hence likely boosted deportation fears. It is important not to attribute the effect of these programs to the employment authorization mandate, which has very different theoretical implications.4 When migrants decide whether to access government services, for example, the effects of the two types of programs could cancel each other out.
The implementation of other policies during the study period also violates the assumption of the difference-in-difference identification scheme, which requires that outcomes of treatment and control groups follow a similar trend in the absence of treatment.5 In general, difference-in-differences analysis makes high demands on the data being used, which may be a concern with this particular data set in which a number of issues arise. For example, there is no information on the timing of when services were sought or a cross-state move was made. When a treated migration spell spans the pre- and post-period, it cannot be determined if these outcomes occurred before or after E-Verify was implemented.
It is also not clear how the treatment and control groups compare in the pre- and post-periods. The incidence of work in this overwhelmingly male sample is less than 30 %. Is this the result of E-Verify, or is it true of both groups? Difference-in-differences requires that the composition of the treatment and control groups remain intact over time. Using a single survey at a point in time to reconstruct pre- and post-period treatment and control groups can be problematic, particularly if the incidence and selection of return migration varies across the two groups and over time. EMIF data suggest that migrants to Arizona have significantly shorter durations of stay than migrants to California and are much less likely to work. And what about migrants who are headed to E-Verify states after implementation of the law? Are they truly similar to the migrants who came before?
Immigration researchers are no strangers to tricky data. Notwithstanding imperfect data and ambitious methodology, Amuedo-Dorantes et al. have authored a pioneering study with unique data that offer information not available elsewhere and a peek into the lives of the undocumented during a period of unprecedented change. The findings raise a host of interesting issues that demand further attention for their social implications and scientific content.
The opinions expressed herein are those of the author and do not necessarily reflect the views of the Federal Reserve Bank of Dallas or the Federal Reserve System.
The new regulation increases labor costs, which adversely affects overall employment (see Stark and Jakubek 2012).
Although chilling effects were found among immigrants following the passage of welfare reform in 1996 (Watson 2010), it is not clear why an employment authorization program should have such an effect on seeking public services.
Moreover, states cannot enforce the employment provisions of the 1986 Immigration Reform and Control Act, a role that law explicitly reserves for the federal government.
Detailed information on other state and local laws and policies implemented during this time is needed.
To examine the implications of state and local immigration policies more broadly would require changing the timing of treatment to correspond not simply to E-Verify, but to the earliest intervention in a state. See, for example, Orrenius and Zavodny (2009).