University of California Transportation Center

Traffic congestion is a substantial time cost for many urban commuters. This dissertation first studies the response of subjects in experimental settings in which subjects choose between a short direct route that becomes increasingly congested as more people travel on it and a more indirect route that does not become congested. More specifically, I investigate three different toll implementations. Consistent with previous experiments, my first toll design imposes monetized homogeneous time costs. Within this framework the implementation of a toll comes very close to achieving the efficient use of the travel network predicted by theory. Two other toll designs implement heterogeneity in the subject pool. In the first design, I implement a design that more closely simulates boring commutes by forcing subjects to sit and wait for a period of time after the experiment where the length of time they have to wait is an increasing function of experimental travel time. Paying the toll can reduce waiting times. There is substantial heterogeneity in outcomes between groups, which is likely due to the distribution of values of time by session. This may help explain why similar traffic networks have different commuting patterns when they serve different populations. The second toll design imposes monetized heterogeneous time costs. As the level of heterogeneity rises, the number of travelers on each route becomes more stable. This contrasts with other experimental work, which shows a substantial level of instability in a homogeneous framework. Finally, I also analyze various models to study behavior in a no-toll homogeneous framework. While no theory explains individual behavior well, I do find that some theories explain aggregate behavior quite well.