We propose, and empirically evaluate, a theory of the labor market in which
differences in pay between otherwise identical workers arise as a consequence of
(endogenous) differences in bargaining conditions. We use the model to help explain
trends in the gender wage gap, and explore the plausibility of the theory by
employing a city-level analysis using U.S. census data for the 1970-2007 period.
We address potential endogeneity by exploiting the time dimension of the data,
as well via instrumental variables methods. We find considerable support for the
model: bargaining effects are correct-signed and significant in virtually all speci-
fications, and our preferred estimates indicate that bargaining effects are roughly
of the same magnitude as mechanical composition effects.