Disaggregate models of choice in a spatial context: paper presented at the IIASA-seminar on 'Urban Systems Modelling', Moscow, October 1980

INTRODUCTION The analysis of spatial allocation and choice patterns in urban and regional systems is a central issue in regional science and geography. In many traditional analyses, cost-minimizing or utility-maximizing principles have been used to explain and predict spatial behaviour of people; economie decision criteria were pivotal elements in analyzing regional and urban mobility patterns. Since, however, many western industrialized countries have reached a full maturity, the locational and mobility choices are guided by less purely economic-oriented criteria. Many qualitative aspects (such as environmental conditions and the quality of the housing stock) are increasingly influencing spatial choice behaviour. This evokes the need for a broader analysis of spatial choice mechanisms. Furthermore, the allocation and choice patterns demonstrate an increasing heterogeneity and diversity among spatial actors, so that also the question as to the scale of analysis (aggregate versus disaggregate) becomes crucial. This paper is devoted to a critical survey of modern choice models, which are being used nowadays, among others, in attempts to approach spatial allocation and choice patterns in a more realistic way, by means of taking into account the abovementioned qualitative aspects and elements of scale in addition to economie decision criteria. After a metho-dological introduction, several categories of such choice models are reviewed in a spatial context. Having evaluated these classes of choice models, we will pay more specific attention to two disaggregate choice models, viz. the multinomial logit and the multinomial probit model. The features of the latter pair of models will be illustrated by means of a numerical exercise. METHODOLOGICAL REMARKS The methodology of spatial choice analysis can be based on several theoretical frameworks. Two main categories can be distinguished, viz. the traditional (mainly neoclassical) and the behaoiourdl theories. The traditional approach to spatial choice analysis takes for granted the notions of utility and indifference. Usually, the choice criteria in micro-economie decision-making are assumed to be the same for all individuals , though the shape of the individual utility functions is not necessarily equal. Consequently, the same set of attributes of a commodity or of an alternative or the same set of commodities will normally not lead to the same utility for all subjects, while also interpersonal utility compar-2-isons are often not possible. It is clear that the application of this traditional theory to spatial groups at a more aggregate level (e.g., social classes, income groups) has until now assumed a uniform utility function for …