This research is concerned with the econometric analysis of choices made by individuals or firms. In the case of an individual choosing among a discrete set of alternatives (such as what mode of transportation to take to work, whether to buy a house, etc.), a well- established approach is the model of stochastic utility maximization. In that approach, the utility or other objective function associated with each alternative is modelled as a parametric function of observed variables, plus a stochastic component which represents the effects of unobserved quantities, such as unmeasured attributes of the alternative and taste of the individual. Specifying a joint probability distribution of these stochastic components then leads to a choice probability model. For discrete choice, the standard models of this type include the probit, logit, and generalized extreme-value models. The principal focus of this project is the extension of these methods from discrete choice to continuous choice. Examples of continuous choice include time of day decisions (such as when to start a trip from home to work, when to make a phone call, or when to send out a shipment) and decisions involving geographic coordinates (such as where to locate a new facility within a metropolitan area). This extension involves considerable theoretical and practical difficulties, and has received much less attention in the research literature than has discrete choice modelling. The project is important because, first, it should result in a better understanding of the econometric theory associated with the maximization of stochastic processes. Secondly the study should lead to an improved understanding of the tradeoffs and other factors involved when individuals, firms, or institutions make decisions about geographic location and about the timing of economic activities.
