Theory predicts that within a metropolitan area, employers located where th
ere is difficult commuting will have to compensate their workers with appro
priately higher wages. This should hold whether one is comparing central ci
ty work locations with those at the fringe. or simply differences between s
uburban "edge cities." In the longer run, if employers are concerned primar
ily with labor costs, firm mobility should equalize wages or else agglomera
tion economies must exist to sustain the differences. Using microdata from
the 1990 census for 2 large metropolitan areas in the U.S., wage equations
are estimated for urban workers allowing for different wage levels dependin
g on zone of employment. The results show that observationally equivalent w
orkers have wages that vary substantially across employment zones within a
metro area, and that this variation is strongly and significantly correlate
d with the average commute time of the workers employed in that zone. These
results hold across several different econometric specifications. Wages an
d average travel times also are found to be highly correlated with the aggr
egate number of workers in each zone. but are not affected by zone employme
nt specialization. Thus there is inconclusive evidence as to whether wage/c
ommuting cost differences result from equilibrium agglomeration effects or
from a disequilibrium distribution of employment. (C) 2001 Academic Press.