What explains the urban wage premium? Sorting, non-portable or portable agglomeration effects?

Using administrative data for West Germany, we study the relative importance of different determinants of the urban wage premium. More explicitly, we distinguish worker sorting, as well as portable and non-portable agglomeration effects. Our results indicate that worker sorting explains about two thirds of the urban-rural wage gap. We show that the estimated fraction of the urban wage premium attributed to worker sorting differs considerably depending on the selectivity of the sample used for identification and provide guidance how this selectivity can be reduced. Agglomeration effects explain about one third of the urban wage premium, with portable and non-portable agglomeration effects being of similar importance.