A large volume of the literature documents the existence of large and persistent wage differentials among industries for workers of comparable skills. This work provides empirical evidence measuring the extent to which unobserved abilities explain the existence of highly persistent inter-industry wage differentials, based on Georgian quarterly household longitudinal data that allow us tracking workers over time. We find that inter-industry differentials that are measured on a cross-sectional basis reflect inter-industry variations in unmeasured labor characteristics. When estimated at longitudinal level the dispersion of inter-industry wage differentials measured by weighted standard deviation decreases roughly seven times and the actual differentials reduce 4-5 times, indicating that a substantial part of the differentials merely reflect differences in labor quality across industries. In addition, we check whether there is endogeneity of mobility within the longitudinal model by Instrumental Variables (IV) method, concluding that endogeneity of mobility is not a major source of bias in the model and the results of the test are fully consistent with our earlier findings.