| Understanding the forces that determine residential land use can be of immense value to households, builders, and urban policy makers. One method of modeling land use is to describe the "highest-and-best-use" phenomenon. To operationalize this use concept in economic terms, it is hypothesized that the marginal product of a site, as an input in the production of housing, is determined by its locational traits and will vary relative to each dwelling type alternative. The dwelling that maximizes a site's marginal product, as measured by its land residual value, is shown to be that site's "highest-and-best" use. Within Rosen's implicit market approach, two related forecasting models are developed based on this use theory. The models are formulated form the producer's perspective which describes how landowners maximize the residual value of their land through their choice of dwelling attributes supplied by builders. Each model will predict the dwelling characteristics that would be built optimally on a residential site given its unique locational characteristics and the builder's cost function.; Both models produce theoretically identical results, yet are estimated by different approaches. The first approach, the land residual maximization model, is a land residual function maximized with respect to its dwelling attribute variables. The second approach uses Rosen's widely accepted hedonic "two-step" model modified to reflect this dissertation's producer-oriented perspective. This model facilitates the estimation of the landowner's derived demand for dwelling attributes and the builder's dwelling attribute supply functions. Thus the impact that locational attributes have on the quantity of dwelling attributes demanded is made explicit along with the nature of builder supply. With the demand and supply functions estimated, market determined equilibrium output levels of dwelling attributes can be forecasted for various sites.; The forecasting ability of both models was tested. The predictions of the land residual maximization model were reasonable in meeting theoretical and intuitive expectations. The modified Rosen model was, however, less successful. Several statistical problems were encountered with the estimation of the house price and builder's supply functions. Many of the independent variables proved insignificant, possessed the wrong sign, and were highly correlated with one another, creating severe multicollinearity problems. |
