A simulation approach to studying the sensitivity of commuting-flow predictions with respect to specific changes in spatial structure

Economic evaluations of investments in transportation infrastructure in general call for traffic-flow predictions. In many studies such predictions are based on a doubly constrained modeling framework, with no explicit attempts to account for possible effects from changes in the location pattern of population and employment. In this paper we focus on the sensitivity of commuting-flow predictions to specific changes in the marginal totals of a trip-distribution model. We approach this problem through a series of simulation experiments where population and employment are systematically redistributed between different zones within the region. With this procedure we provide some quantitative estimates of prediction errors that follow from the inability to take into account long-term consequences of the location pattern of firms and households. We also carry through a simulation experiment that focuses on how benefits of different road investments are interrelated, and depend on spatial-structure characteristics.