Association for Public Policy Analysis & Management

Despite the contentious debate surrounding undocumented immigrants in contemporary political discourse, evidence of how these workers affect labor market outcomes in the U.S. remains relatively sparse. Standard competitive models predict that labor market outcomes of legal low-skilled workers would deteriorate with an increase in undocumented workers, assuming undocumented workers serve as substitutes for legal low-skilled workers. However, Haltiwanger and Davis (2014) propose reasons for the secular decline in labor market fluidity in the U.S. since the 1980’s, many of which do not apply to undocumented immigrants. Furthermore, they document the importance of labor market fluidity, defined as worker hires and separations, to the health of labor markets. In this paper, I seek to understand the effect of undocumented immigrants on labor market fluidity outcomes in U.S., low-skilled, local labor markets.

I present empirical evidence that metropolitan areas and states with higher proportions of undocumented workers have increased fluidity and subsequent increased employment rates. These findings emerge from instrumental variable and dynamic panel models using a city-year panel constructed from the U.S. Census Bureau's American Community Survey (ACS) and Quarterly Workforce Indicators datasets. Specifically, I find that a one percent increase in the composition of the low-skilled labor force that is Hispanic, from Mexico or Central America, not a U.S. citizen, recently immigrated, and does not speak English well, increases separation and hiring rates by 0.94 and 0.92 percent in the low-skilled labor market, respectively. As predicted by Haltiwanger and Davis (2014), increased fluidity also generates increased employment rates for substitute, legal workers. These findings are especially salient in industries where undocumented workers make up a larger share of the workforce, such as construction, agriculture, and personal services. Furthermore, placebo tests indicate that low-skilled workers from the same origin countries who either have citizenship or who have been in the U.S. for more than 10 years do not have similar fluidity-generating effects. My estimates additionally correct for known, classical measurement error in the independent variables generated by sampling error in the ACS.

In order to place these empirical results in a tractable context, I develop a search model of the local labor market with heterogeneous agents, building upon the work of Chassamboulli and Palivos (2014) and Chassamboulli and Peri (2015). I augment their models by allowing for on-the-job search in order to accurately predict the response of separation and hiring rates. The model demonstrates that as the composition of the low-skilled labor force increasingly includes undocumented immigrants with lower outside options and higher productivity re-draw probabilities than their legal counterparts, labor market tightness, separations, hires, and employment rates all increase consistently with empirical evidence. Unlike a competitive model of the labor market, my model can deliver these predictions without assuming undocumented and legal workers are complements in production—in fact, it considers them to be perfect substitutes. Thus, this paper proposes and empirically validates a model that highlights the importance of undocumented immigrants in sustaining and improving the health of low-skilled labor markets.