DISAGGREGATE BEHAVIOR MODEL OF AUTOMOBILE OWNERSHIP

This paper describes a series of disaggregate behavioral models that forecast the probability that various combinations of automobile ownership and travel modes to work will be selected by households. The models assume that workplace and residential location are predetermined. The multinomial logit model is used in a joint structure that captures the complex interrelationship of automobile ownership and travel-to-work decisons. The paper describes the considerations in the choice of independent variables and the specification of the utility functions. The estimation results for each of 7 distinct socioeconomic groups, or market segments, with different behavioral characteristics are presented and analyzed. One of the models than is used to examine the shifts in automobile owernship for a suburban household resulting from alternative levels of transit service improvements.