With mobile workers and competitive markets, equilibrium nominal wage rates
rise with the local cost of living but fall with the value of local amenit
ies. Earnings and wage regressions that ignore such effects may suffer from
omitted variable bias because observed education and demographic attribute
s affect both worker skill levels and location choice. Geographic fixed eff
ects can be used to control for unobserved locational attributes provided t
hat their scope is at least as narrow as the underlying labor markets, but
not so narrow as to introduce simultaneity problems arising from the endoge
nous choice of location on the basis of income. Estimates from the 1985-198
9 American Housing Survey suggest that SMSA-level fixed effects control for
unobserved locational attributes without introducing simultaneity problems
. In addition, failure to control for location leads to biased estimates of
the effect of important demographic characteristics. (C) 1999 Elsevier Sci
ence B.V. All rights reserved.