(joint with Ingo Isphording and Sebastian Otten)
This paper uses data from the German Socio-Economic Panel (SOEP) to study the effect of language skills on labor market outcomes. To address problems related to endogeneity and measurement error, we construct an instrumental variable based on differences in language acquisition profiles of immigrants across the distribution of linguistic distance between German and the language spoken in the country of origin. Our findings indicate that the effect of language skills on employment probabilities is insignificant, while the positive effect of language skills on wages may be attributed to the sorting of immigrants across occupations. Our results further indicate that OLS regressions systematically underestimate the returns to language skills in the German labor market.