Stanford University


Do Digital Platforms Reduce Moral Hazard? The Case of Uber and Taxis

Erik Brynjolfsson
Meng Liu
Jason Dowlatabadi

Management Science
August 27, 2021

Show buttons

Traditional marketplaces are prone to market inefficiencies such as moral hazard due to information asymmetries between market participants. Digital platforms often build transparency and dispute resolution mechanisms into their marketplaces. This may reduce moral hazard. Examples of such tools include real-time monitoring, buyer and seller ratings, and low-cost complaint channels. In a newly published paper in Management Science, Meng Liu, Erik Brynjolfsson, and Jason Dowlatabadi study this research question through the lens of the vehicle for hire market. They find that Uber may mitigate some of the moral hazard problems existing among taxis.

Specifically, the authors focus on driver detour behavior, where detour is defined as the extra distance a driver adds to the fastest route. To compare driver routing choices between taxis and Uber, the authors leverage trip-level data from yellow taxis and UberX in New York City. These taxi and Uber trips are matched at a granular origin-destination-time level with the assumption that such strictly matched trips are subject to the same underlying optimal route and thus are comparable. The authors then exploit variations in pricing schedules (and thus driver incentives) to facilitate identification.

The findings are consistent with Uber’s market design reducing moral hazard: (1) taxi drivers drive longer distances (on average 8%) than Uber drivers on matched metered airport trips; (2) this difference disappears when airport taxi trips have a flat fare and on short trips in dense markets (e.g. within Manhattan). The authors explore several explanations besides moral hazard and find that they are not compatible with the evidence. For example, it is found that taxi driver detours are associated with both longer trip distance and duration, thus rejecting the explanation that taxi drivers are making time-distance tradeoffs desired by their customers. In addition, the evidence is not consistent with the hypothesis that Uber-taxi routing differences are an outcome of GPS use, because taxi and Uber trips are similarly efficient on flat-fare airport trips. Lastly, driver selection is rejected as a viable explanation by a driver fixed effects model, and the significant routing efficiency improvement after taxi drivers became Uber drivers. 

While Uber appears to achieve and overall improvement in market transparency, Uber may also create new scenarios of driver moral hazard. For example, it is shown that Uber drivers tend to take longer routes in times of high surge pricing.

This paper provides a direct comparison between taxis and Uber, finding that moral hazard problems can be mitigated by innovative market designs that improve transparency. These findings help provide a better understanding of online-offline competition and welfare in the digital economy.

Stanford University