Nicholas Buchholz, Princeton University

Rethinking Reference Dependence:  Wage Dynamics and Optimal Stopping Rules in Taxis

Abstract:  Workers with flexible hours, like taxi drivers, stadium vendors, and call center workers, have generated key puzzles in the empirical analysis of labor supply. The typical approach used in the literature is to pose a static model in which hours worked depends on wages or cumulative earnings. Many workers in these settings, however, face uncertain and variable earnings, suggesting that the choice of work time is a forward-looking decision. Using a large sample of shifts of New York City taxicab drivers, we estimate a dynamic optimal stopping model of drivers' work times and quitting decisions. We motivate our model by documenting two new stylized facts: drivers' overall and relative hourly wage trends downward with time worked, and trip-by-trip wages exhibit negative serial correlation. In a static analysis, both phenomena lead to downward-sloping labor supply curves, which would be consistent with income-targeting or reference-dependent preferences. In contrast, our results demonstrate that some of the apparent behavioral biases documented in the literature can be reproduced using entirely standard preferences. We use the model to provide new estimates of individual earnings elasticities. We then use data spanning a 2012 fare change to measure a market-wide elasticity, accounting for the equilibrium impact of prices on supply and demand.  We find market elasticities to be approximately 60% less than individual elasticities. Our analysis suggests that existing estimates of the benefits to proposed wage legislation in the taxi and ride-hail industries are overstated.