Sent back home: How the forced removal and return of Mexican migrants affect local Mexican workers: Guest post by Thomas Pearson

This is the sixth in this year’s series of posts by PhD students on the job market.

Today, nearly 10% of Mexican-born individuals—11 million people—live in the United States. Such large migrant populations have increased Mexico’s exposure to shocks across the border. Important among these has been the growing intensity of interior immigration enforcement and deportations, which have increased return flows to Mexico (Figure 1).

In my job market paper, I study how these deportations impact Mexican labor markets using variation in migrant networks and Secure Communities (SC), a policy which expanded local immigration enforcement. SC began sending the fingerprints of everyone arrested directly to Immigration and Customs Enforcement (ICE), who could then quickly identify, detain, and remove deportable individuals.

Deportation Shock

To identify the effects of deportations, I construct a shift-share instrument which predicts the number of deportees for 2,300 Mexican municipalities in two periods (2009-2010 and 2011-2015). This measure sums up the product of the number of Mexicans deported through SC in each U.S. state and the share of undocumented Mexicans living in that state. Data on deportations comes from the Transactional Records Access Clearinghouse. I measure the shares using data from the Mexican government’s Matrícula Consular de Alta Seguridad program. This program provides IDs to undocumented Mexicans living in the U.S and also publishes the annual number of IDs issued for each origin Mexican municipality-U.S. state of residence combination.

I divide the predicted number of deportees by the municipality’s lagged population, providing an intuitive measure, the deportee population share (Figure 2).

 

Three Channels

Perhaps the most salient channel through which the return of hundreds of thousands of Mexican migrants might impact Mexican labor markets is through increased return migration. However, there are two alternative channels worth highlighting. First, exposure to U.S. deportations could alter the emigration decisions of prospective migrants. Second, deportations could decrease remittances by decreasing the stock of migrants living in the U.S. and/or by affecting their employment outcomes.

Given these two alternative channels, I first identify the reduced form, causal effects of deportations on return migration, earnings, out-migration, and remittances to determine whether deportations impact local Mexicans’ labor market outcomes primarily through return migration. To understand the dynamic adjustment process, I include the lagged period’s deportee shock in the reduced form specification, following the advice in Jaeger et al. (2018) who show that shift-share instruments can conflate the short run effect of migration with the positive response to previous inflows. I then estimate the short and long run causal effects of return migration on local Mexicans’ earnings using the deportation instruments to benchmark my results to the migration literature.

Identification

The identification assumption is that the deportee shocks are uncorrelated with changes in unobserved economic conditions conditional on the controls. These controls include Mexican state fixed effects, initial municipality demographics, emigration rates in 2000 (using the Mexican Census), recent emigration flows to the U.S. (proxied by the number of matrículas issued from 2006 to 2010), and pre-trends in the outcome variables. Variation in exposure to deportations thus hinges on different municipalities having migrant networks in different U.S. states.

Example of a threat to identification: Suppose poor labor market outcomes and/or crime/violence at origin lead migrants to locate in places that will eventually have higher rates of SC enforcement. This could be because these migrants are more likely to compete with the U.S.-born for jobs or be criminals, which could lead to more deportations and more exposure. If conditions continue to deteriorate in these municipalities, then I would mistakenly infer that deportations negatively affect labor market outcomes.

There are no differential pre-trends in return migration and homicide rates, and eventual exposure to deportations is uncorrelated with initial earnings. These results rule out the possibility that migrants sort across U.S. states based on origin-municipality economic conditions and crime/violence, which helps validate the identifying assumption. I conduct these same checks using the shares for the U.S. states driving most of the variation (Arizona, California, and Texas) as instruments as recommended by Goldsmith-Pinkham et al. (2020) who show that in certain shift-share designs, exogeneity should be interpreted in terms of the shares. These diagnostics help provide further validity to the identifying assumption and are consistent with the findings in Munshi (2003) who shows that similar municipalities often send their migrants to different destinations in the U.S.

Main Results

Using microdata from the 2010 Mexican Census and the 2015 Intercensal Survey, I show that in both the lagged (2005-2010) and contemporaneous (2010-2015) periods, deportations increased return migration. The increased inflows are driven mostly by return migrant men with less than a high school degree.

• For the local Mexicans most similar to the return migrants in terms of gender and education, deportations decreased earnings in the short run. In the average municipality, deportations decreased earnings in 2015 for men and women with less than a high school degree by 5.5 and 2.1 percent respectively. These negative short run effects are similar in magnitude to the positive effects of the flip side of this migration. Mishra (2007) shows that out-migration to the U.S. from 1970 to 2000 increased the wage of an average Mexican worker by 8.0 percent

• Increased exposure to deportations does not deter emigration. Rather, the decline in earnings increases undocumented migration to the U.S. Using the issuance of matrículas as a proxy for emigration, the estimates suggest that 2 additional predicted deportees result in 1 new undocumented immigrant.

The negative short run effects on earnings are not due to changes in labor demand resulting from falls in remittance income as deportations increase both the share of households receiving remittances and the total amount received.  They are also not the result of an income effect from increased remittances as the effects are similar when I restrict the sample to individuals in households that do not receive remittances. Overall, the reduced form results instead point to return migration as the primary channel through which deportations impact the earnings of local Mexicans.

To ensure I am identifying the causal effects of deportations and not picking up on other factors correlated with SC enforcement, I control for several other shocks for robustness. These include U.S. unemployment rate shocks following Caballero et al. (2021), exposure to border enforcement as measured by networks in border states, and exposure to industry shocks as measured by initial industry labor shares. The results when including these controls are nearly identical to the baseline results.

Consequences of Underdevelopment

Frictions in credit and capital markets in Mexico exacerbate the negative effects on earnings from the deportee-induced labor supply shock. For a municipality with one predicted deportee arrival per 100 people, having a bank branch mitigates the drop in men’s earnings by 5.6 percent. Meanwhile, a one standard deviation increase in road density mitigates the fall for such a municipality by 3.7 percent. I show that banks allow locals to work less when wages fall, as they can borrow to smooth consumption. I also show that increased road density leads to larger expansions of capital in response to deportations, which dampens the wage effects and results in smaller drops in earnings.

These results help explain the large effects on locals’ earnings as many migrants return to regions where these frictions are prevalent. They also provide new insights into the effects of immigration in underdeveloped economies.

Policy Implications

1.       U.S. interior immigration enforcement can impact economic conditions in migrant-sending countries.

2.       These economic consequences can increase undocumented immigration to the U.S.

3.       Resources that help forced return migrants’ social, cultural, and economic re-integration are important.

Thomas Pearson is a PhD Student at Boston University.