THE ESTIMATION OF CONSUMER PREFERENCES IN HEDONIC MARKETS: A SIMULATION TEST

The hedonic markets model is widely used to estimate the demand for housing and neighborhood attributes, and to evaluate the welfare effects of changes in attributes resulting from public policy. This dissertation assesses the accuracy with which preferences for housing attributes can be inferred by estimating the parameters of the hedonic model using single market data. The results are based on simulations of housing market equilibria in which consumers with specified preferences for housing attributes bid for a fixed housing stock. Equilibrium housing prices, together with housing attributes and buyer characteristics, provide the data used to estimate the hedonic model.;A comparison of estimated coefficients of the marginal bid functions with true coefficients provides insights into problems caused by identifications and simultaneous equations bias. The marginal bid functions are then used to estimate willingness to pay for nonmarginal change in each in each attribute. The estimated value of each nonmarginal change is compared with the true value of the change to evaluate which specification of the hedonic price function produces the most accurate willingness-to-pay estimates.;Unlike theoretical results, simulation results pertain only to the set of specific assumptions used in the simulation. In order to strengthen the dissertation results, a variety of assumptions about utility functions, about the distribution of housing attributes, and the distribution of buyer characteristics is used to create six distinct scenarios. Because random components are included in each housing market model to allow for unobserved variation in tastes, a Monte Carlo approach is used; each scenario is repeated using twenty different draws of the vector of random parameters.;Marginal prices for individual attributes of the housing stock are estimated using a variety of specifications of the hedonic price function. These estimates are compared with true marginal attribute bids to evaluate which hedonic price function specification produces lowest errors in estimating marginal bids. Marginal prices, along with buyer characteristics are used to estimate the mean household's marginal bid function for each attribute.