³ÉÈË´óƬ

This paper develops a theory of the labor market in which both union status and the wages of union and nonunion firms are endogenously determined. This brings together two classes of research - on the wage effects of unions and on the incidence of unionism - that have been largely separate in previous work. The implications of this unified theory diverge from current conventional wisdom. If employers set wages taking into account their effect on workers' incentives to organize, union density and the union wage gap are negatively correlated. In turn, union organization takes place only when it is less costly than dissuading workers from organizing, which happens if unions increase total rents, or if there is asymmetric information. Against conventional wisdom, organization is more likely in firms with lower than average rents, and average industry rents, by themselves, are uninformative to predict union density or the wage gap. The theory reconciles the time series behavior of union density and the wage gap in the U.S., and it is the first to provide an explanation for the cross-sectional distribution of these variables.