The role of site and job characteristics in hedonic wage models

Recent economic analyses of wage rate differentials have generally used some form of hedonic framework in explaining the observed wage structure. While these studies have adopted this common theoretical justification, the models have been used to serve two distinct objectives. The first of these is an empirical test of the theory of equalizing differences originally introduced by Adam Smith. In such applications (see Brown [7], Lucas [26], Olson [28], Thaler and Rosen [36], and Viscusi [40] as examples) specific job characteristics, measured with technical information or with worker perceptions, have been used to explain differences in wage rates across individuals and jobs. While there are a number of these studies,’ there has not been a consistent pattern of support for the theory. Indeed, with the exception of job risks, many characteristics that a priori would be expected to lead to equalizing differences have often exhibited incorrect and/or statistically insignificant measured effects on wages.* The second use for wage models has been as the basis for an index of the quality of life in urban areas. In this framework, households are assumed to select residential locations so as to maximize their individual welfare. Migration between cities provides the mechanism for an equilibrium household assignment. Under ideal conditions the pattern of equilibrium wage rates reflects the attributes of the cities involved (see Hoch [20], Izraeli [23], Cropper [lo], Cropper and Arriaga-Salinas [ 111, and Rosen [31] as examples). The purpose of this paper is to develop a model for wage rate differentials that includes both job and site characteristics, and to estimate the model with a large micro-data set. Our wage models provide estimates of a market clearing envelope which includes each of two types of past analyses as a