Returns to foreign languagues of native workers in the EU

Most papers on returns to languages are concerned with immigrants. We use the European Community Household Panel Survey (ECHP) to infer returns on non-native languages by non-immigrants in nine countries of the European Union. We differ from the few other studies that deal with the same problem in three respects. First, we correct for time-dependent measurement errors in self-reporting as suggested by Dustmann and Van Soest and find that the resulting IV estimates are much larger than those obtained by OLS. We also suggest that there is little room for time-persistent errors and heterogeneity, and that therefore our estimates should not suffer from the other usual biases. Secondly, instead of using a dummy for each language, we use the ratio of the population that is not proficient in a language in each country considered. Finally, we estimate instrumental variable quantile regressions to illustrate how returns to languages vary at different points of the distribution of earnings.