ESTIMATION OF RANDOMLY DISTRIBUTED VALUE OF TIME

This paper outlines various models and empirical studies aiming to capture variations in value of time (VOT). Its motivation is to improve the estimation of the 'implied' VOT captured in travel choice models, allowing for substantial heterogeneity between travellers, which can now be handled by advances in data acquisition, estimation methods, and computing. Forms of heterogeneity that are being addressed include decision protocols, choice sets, perceptions, attitudes, and taste variations. The types of mathematical models considered are: (1) individual-specific models, which have repeated observations of each individual, and use a separate travel-choice model estimated for each individual; (2) a pooled model, which bases its estimates on pooled data, and assumes completely homogeneous tastes; (3) a prior segmentation model, which is a practical procedure for introducing heterogeneous tastes; (4) a random coefficients model, which attempts to capture taste variations that cannot be explained by observed individual characteristics; and (5) a latent class choice model, which includes a probabilistic model of class membership. The advantages and disadvantages of each type of model are listed, and examples of empirical studies are tabulated for each. Four conclusions are presented. For the covering abstract, see IRRD E100587.