Surge Pricing Solves the Wild Goose Chase

Abstract

Ride-hailing applications (apps) like Uber and Lyft introduced a matching technology and market design that recent research has found is more efficient than traditional taxi systems [2]. However, unlike traditional street-hailing taxi systems, they are prone to a failure mode first anticipated by [1]. In this paper we model and empirically establish the existence of these dynamics. We then show how surge pricing and, to a lesser extent, other market design interventions can prevent this problem from crippling a ride-hailing market. An over-burdened dispatch system results in available idle drivers being too thinly spread throughout a city, forcing matches between drivers and passengers that are far away from each other. Cars are thus sent on a wild goose chase (WGC) to pick up distant customers, wasting drivers' time and reducing earnings. This effectively removes cars from the road both directly (as the cars are busy making pick-ups) and indirectly (as cars exit in the face of reduced earnings), exacerbating the problem. This harmful feedback cycle results in a dramatic fall in welfare, hurting both drivers and passengers. A ride-hailing market that falls into WGCs frequently might therefore perform worse than traditional street-hailing taxi systems, so it is essential to understand WGCs in order to design markets in a way that avoids WGCs and exploits the potential welfare gains from the new technology. [1] dismissed WGCs as Pareto-dominated equilibria and thus just a theoretical curiosity. However, we show that when prices are too low relative to demand all equilibria of the market are WGCs when using a first-dispatch protocol, in which an idle driver is immediately dispatched every time a rider requests a trip (as many ride-hailing services have committed to). This suggests two ways in which pricing can avoid WGCs. First, one might set a single high price all the time, sufficiently high to avoid WGCs even at peak-demand periods. Of course this design has the drawback that prices will be unnecessarily high, and thus demand inefficiently suppressed, at times of low demand. A more elaborate mechanism is to use dynamic ``surge pricing'' that responds to market conditions. Such a system was introduced by Uber early in its development. Prices are set high during peak-loads, but can fall when demand is more normal. Thus, against the common perception, surge pricing allows ride-hailing apps to reduce prices from the static baseline instead of increasing them.