Departure Times in Y-Shaped Traffic Networks with Multiple Bottlenecks

We study the departure time decisions of commuters traversing a traffic network with the goal of arriving at a common destination at a specified time. There are costs associated with arriving either too early or too late, and with delays experienced at bottlenecks. Our main hypothesis, based on the Nash equilibrium distribution of departure times, implies that, for certain parameter values, expanding the capacity of an upstream bottleneck can increase the total travel costs in the network. We report the results of a large-group laboratory experiment, which are strongly supportive of this counterintuitive hypothesis, and we discuss the implications. (JEL D85, R41)