Congestion and incentives in the age of driverless fleets

The diffusion of autonomous vehicles (AVs) will expand the tools to manage congestion. Differently than fleets of traditional vehicles, operators of fleets of AVs will be able to assign different travelers to different routes, potentially inducing different congestion levels (and speed). We look at the effects of the technological transition from traditional to autonomous vehicles. Our model exhibits a unit mass of heterogeneous individuals. Some of them use the services of a fleet, while others do not, and travel independently. With few fleet users, the fleet technology (traditional vs automated vehicles) is immaterial to welfare. On the contrary, when there are many fleet users, we show that, if fleets do not price any individuals out of the market, the differentiation in congestion across routes under the automated fleet is welfare-reducing. When, instead, fleets price some individuals out of the market, the welfare effects of the transition are ambiguous and depend on the interplay between the extent of rationing by both types of fleets and the extent of differentiation by the AVs fleet. Finally, we characterize the tax restoring the first best with AVs. It involves charging different taxes across lanes, starkly different between independent travelers and the fleet. While independent travelers should be charged lane-specific congestion charges, the fleet should be imposed a scheme involving a congestion-based tax and a subsidy.