Estimating Network Effects on Migrant Labor Outcomes Using a Shift-share Instrument

This paper estimates the causal effects of ethnic networks on the labor outcomes of new migrants. Empirically, estimating this relationship is challenging as migrants endogenously sort into geographic regions with higher wages. Using the local concentration of migrants from the same country to proxy network effects, I construct a modified shift-share instrument to address endogeneity. I normalize the share component of the instrument by the total local immigrant population in order to reflect the predicted local coethnic concentration. I use microdata from the American Community Survey from 2005 to 2019, at the Public Use Microdata Area (PUMA) level. I construct the instrument for each origin country, year, and PUMA combination. The baseline estimates indicate inconclusive evidence for network effects on employment outcomes and wages. There is some evidence for heterogeneous effects for migrants based on origin country characteristics, occupation and industry. I also consider competition among coethnics and labor market equilibrium effects as alternative mechanisms through which local concentration of migrants affect new migrants' employment outcomes. I find that while the directional correlations are generally consistent with competition and labor market equilibrium effects, there is no conclusive evidence that any single mechanism is driving differences in new migrants' labor outcomes.