A Note on Choice of Residential Location in Travel Cost Demand Models

Abstract

where x is number of visits to a site, p is price, y is a (k x 1) vector of observed individual characteristics, and e is a random component. 1' is a (1 x k) parameter vector and p, is a scalar parameter. Hundreds of models with this basic demand relationship as a basis have been estimated (see Smith's 1989 review). My concern is that the price used in this model may be endogenous because people base decisions about where to live, in part, on proximity to recreation sites. If so, unobserved taste characteristics in the error term e will be