A hstract-Estimates of the demand for the characteristics contained in a differentiated product such as housing can be based on hedonic regressions. However, previous studies have not dealt with several potential problems in such estimation. A more appropriate framework for the estimation is provided in this paper. Also data from a number of cities are used to eliminate the identification problem, and the endogeneity of the marginal prices derived from non-linear hedonic equations is considered in the estimation. The results on own and cross price elasticities, expenditure elasticities, and the effects of socio-economic vafiables accord well with expectations.

T HE rapid rate of increase in the prices of houses in recent years has resulted in renewed interest in trends in residential real estate. A widely quoted study by the Harvard-MIT Joint Center for Urban Studies (Solomon, et al., 1977) showed that in 1976 only 27% of families could afford to buy the median priced new home, a significant change from 1970 when 46% of families were able to make such a purchase. Such rapid changes could result in significant income redistribution effects, and policy proposals have been numerous. To assess these trends and proposals, it is necessary to have accurate methods of developing residential real estate price indexes. Most real estate indexes have been based on the average or median selling price in each year for new or used houses. Yet there may exist substantial differences in the houses on the market at different times, so such indexes contain quality changes as well as pure price changes. Because housing is a highly heterogeneous commodity, the measurement of quality-adjusted price changes has proven difficult. Such measurement is desirable not simply because of the recent significant price changes in the real estate market, but also because economists have found property values to be one of the best sources of information on goods for which markets do not exist. Although in the past cross-sectional studies have generally been used for this purpose, questions remain concerning the timing of the impacts. Comparisons of real estate price indexes can provide insights into the dynamics of such effects. Among the techniques suggested for developing quality-adjusted price indexes, approaches using hedonic regressions and repeat-sale regressions appear to be the most promising. In this paper, alternative local price indexes are developed using modifications of the hedonic and repeat-sale techniques. For the case considered, the two independent techniques provide statistically identical indexes of the real price of housing. The estimated rate of increase in house prices using the hedonic and repeat-sale techniques is substantially lower than the non-quality adjusted rate implied by the change in average selling price.

This paper proposes a method for measuring the effects of substitutions in the timing of recreational use on people's willingness to pay for nonmarketed resources. Using the three markets (peak, pre-peak, and post-peak) for weekly rentals of vacation properties along the Outer Banks of North Carolina, we are able to control for changes in the mix of site characteristics selected at different times and estimate the effects of temporal substitution on tradeoffs between other characteristics. Proximity to the ocean was found to be a significant determinant of temporal substitution between the peak and pre-peak seasons with ocean front properties having 1.9 percent to 4.7 percent smaller discounts for preseason rentals relative to other properties. Copyright 1994 by MIT Press.

V. Kerry Smith, Raymond B. Palmquist, Paul Jakus, Combining Farrell Frontier and Hedonic Travel Cost Models for Valuing Estuarine Quality, The Review of Economics and Statistics, Vol. 73, No. 4 (Nov., 1991), pp. 694-699