Carpooling with heterogeneous users in the bottleneck model

We investigate the effects of carpooling in a dynamic equilibrium model of congestion, which captures various dimensions of heterogeneity: heterogeneity in the preference for and cost of carpooling, heterogeneity in values of time and values of schedule delay. We investigate various policy scenarios: no-toll, first-best pricing, and subsidization of carpooling. The optimally differentiated subsidy equals each type of users’ marginal external benefit (MEB) of switching to m-person carpooling or carpooling, which turns out to be heterogeneous for “ratio heterogeneity”, where the ratios of the values of time and schedule delay vary, and homogeneous for “proportional heterogeneity”, where these values vary in fixed proportion over the population. If such differentiation over users is impossible, the subsidy is a weighted average of the MEB's, with the weights reflecting the relative sensitivity of the group size of carpoolers to the subsidy. Using a numerical example, we investigate the welfare effects and distributional effects of different policies. The relative efficiency of the differentiated subsidization first increases and then falls with the degree of ratio heterogeneity, and decreases over the entire parameter range and more with the degree of proportional heterogeneity.