Monitoring in target contracts : theory and experiment in Kenyan public transit

Speci cally, we introduce monitoring devices into commuter minibuses in Nairobi, Kenya, that track real-time vehicle driving behavior and daily productivity. We randomize which minibus owners have access to these monitoring data using a novel mobile app that we designed for the industry. In line with model predictions, we nd that treated vehicle owners modify the terms of the contract by decreasing the transfer they demand in exchange for higher e ort and lower risk-taking. Drivers respond accordingly by working more hours and decreasing risky driving behavior associated with higher repair costs. As a result, rm costs fall and pro ts increase. Structural estimation via simulated method of moments demonstrates a close match of the data to the contract model and suggests overall welfare increases stemming from lower rm costs