Evolutionary Implementation of Optimal Traffic When Values of Time are Unknown

Abstract:     
We consider evolutionary implementation of optimal traffic via a price scheme as in Sandholm [Evolutionary Implementation and Congestion Pricing, Review of Economic Studies 69 (2002), 667-689] in which the planner makes traffic converge to an optimal one in the long run without knowledge of preferences. However, we analyze the opposite case to Sandholm's in which the planner knows the demand functions for travel on one link but he does not know the values of time of potential drivers. We show that the planner can achieve evolutionary implementation of an optimal traffic by adjusting his estimates of value of time so that optimality conditions hold at current traffic. If all potential drivers have the same value of time, it is possible to use properties of potential game as in Sandholm (2002). However, if the value of time is heterogeneous, it is no longer possible and we make several additional assumptions to make the implementation possible.

Number of Pages in PDF File: 23

Keywords: Congestion, Implementation, Value of time, Potential game, Stable game, Evolutionary game theory

JEL Classification: C72, D62, R41, R48

working papers series