We use panel data to analyze the determinants of speaking fluency and wages
of immigrants. Our model takes account of two problems that may bias OLS e
stimates of the impact of speaking fluency on earnings. First, subjective v
ariables on an ordinal discrete scale, such as self-reported language abili
ty, can suffer from misclassification errors. The model decomposes misclass
ification errors into a time-persistent and a time-varying component. Secon
d, the model accounts for correlated unobserved heterogeneity in language a
nd earnings equation, The main finding is that these two generalizations of
the standard model both lead to substantial changes in the estimated effec
t of speaking fluency on earnings.