son (1999)]. The small number of studies carried out on a narrowly defined
spatial scale may be justified by the lack of available data sets (although the
information available is growing). Also, the challenge posed by modeling
large spatial choice sets within the CLM may have constituted a signifi-
cant hurdle. When confronted with the large data set problem, researchers
have followed McFadden’s (1978)suggestion to work with a smaller sam-
ple of alternatives randomly drawn from the full choice set [Hansen (1987),
Woodward (1992), Friedman, Gerlowski & Silberman (1992)and Guimaraes,
Figueiredo & Woodward (2000)].4 A different approach (aggregation alter-
natives) was proposed by Bartik (1985)who justified the choice of U.S. states
as resulting from the aggregation of the true alternatives considered by firms.
However, these solutions to overcome the large data set problem are unsatis-
factory because they disregard useful information. The resulting estimators
are clearly less efficient.
An econometric problem posed by the CLM in the use of narrowly de-
fined spatial sets is that the Independence of Irrelevant Alternatives (IIA)
assumption is more likely to be violated. Conditional logit models rely on
the assumption that the error terms are independent across individuals and
choices. Typically, industrial location researchers have acknowledged the
potential problem caused by the existence of unobserved site characteristics
that may induce correlation across choices and therefore a violation of the
IIA assumption.5 When dealing with small geographical units, this problem
may be more important because site characteristics that are unaccounted
for can more easily extend their influence beyond the boundaries of the con-
sidered spatial units.6 Some researchers have attempted to control for the