Language and the Earnings of Immigrants

Abstract

Several studies, most of them employing straightforward regression analysis, have concluded that immigrants' proficiency in the language of their adopted country is correlated with their productivity, as measured by earnings. Two weaknesses of these studies are potential unobserved heterogeneity, which could result in over-estimated effects if overall ability is linked with language acquisition, and potential measurement error in the language proficiency measures, which would probably result in under-estimation of language effects. The present study, which uses panel data for a 10-year period in Germany, yields evidence that the latter bias tends to be much larger than the former, implying that language proficiency is far more important than suggested by the existing literature.