Do digital platforms reduce moral hazard? The case of uber and taxis

Digital platforms provide a variety of technology-enabled tools that enhance market transparency, such as real-time monitoring, ratings of buyers and sellers, and lowcost complaint channels. How do these innovations affect moral hazard and service quality? We investigate this problem by comparing driver routing choices and efficiency on a large digital platform, Uber, with traditional taxis. The identification is enabled by matching taxi and Uber trips at the origin-destination-time level so they are subject to the same underlying optimal route, by exploiting characteristics of the pricing schemes that differentially affect the incentives of taxi and Uber drivers in various circumstances, and by examining changes in behavior when drivers switch from taxis to Uber. We find that (1) taxi drivers route longer in distance than matched Uber drivers on metered airport routes by an average of 8%, with nonlocal passengers on airport routes experiencing even longer routing; (2) no such long routing is found for short trips in dense markets (e.g., within-Manhattan trips) or airport trips with a flat fare; and (3) long routing in general leads to longer travel time, instead of saving passengers time. These findings are consistent with digital platform designs reducing driver moral hazard, but not with competing explanations such as driver selection or differences in driver navigation technologies. We also find evidence of Uber drivers' long routing on airport trips in times of surge pricing, suggesting that the tech-enabled market designs may not be binding in our setting.