where the consumer has already optimized his choice of lot size
and other consumption for each location, given the rent and the
disposable income at that location. Put differently, the utility
maximization process is two-stage: in stage one, the consumer
maximizes utility for each location, allocating his disposable
income between the composite good and lot size were he to
choose that location; in stage-two the consumer chooses between
the two locations by considering the value of the indirect utility
plus her idiosyncratic preference n i for each location.
The idiosyncratic utilities n 1 and n 2 represent particular taste
constants for each location that do not depend on the character-
istics with respect to which utility has been maximized. These
idiosyncratic tastes are distributed randomly among the
consumers for each location as independent draws from the
double exponential distribution with heterogeneity parameter
b ¼ p=s
ffiffiffi
6
p
(with s 2 the distribution’s variance). Thus, for exam-
ple, if for a particular consumer n 14n 2 , then that means that the
consumer would prefer location in the core to location in the
suburb if the core and suburb were identical in all other respects.
Under the double exponential distribution of the idiosyncratic
utilities, the choice of location (core versus periphery) would be a
binary logit model. The strictly positive probability that the
randomly selected consumer most-prefers the core is