Consumers’ dishonesty and coarse thinking : empirical evidence from bike-sharing in China

Abstract

The rapid evolution of the sharing economy in transportation sector has transformed urban mobility and provided rich data for academic exploration. In the dissertation, I bring two classic topics into the context of dockless bike-sharing in China and conduct both theoretical analysis and empirical research with proprietary data from ofo, which used to be one of the leading bike-sharing companies in China. The first study investigates the impact of firm pricing strategy on consumer dishonesty. Using more than one million trip records from ofo and the shock from a free-riding campaign, I create a novel measure of cheating and employ the Regression Discontinuity Design (RDD) to find that bike-sharing users would submit more false reports of defective bikes to escape payments when facing a price hike. Heterogeneity analysis reveals that the increases in cheating behaviors after a price surge are more conspicuous for male and non-student users and concerns about their social images during the daytime hours could mitigate motives to cheat. I also find suggestive evidence that the free-riding campaign triggers reciprocity from consumers to the firm. The second study navigates the relationship between air quality information and bike-sharing usage patterns using both theoretical model and RDD. Air Quality Index (AQI) is a continuous score of air quality, which also serves as the basis for a coarser classification (e.g., AQI ≤ 50 classified to be “Excellent”). Applying the RDD method, I explore how people modify their bike-sharing usage as a response to the jump of air quality category at specific AQI values. Notably, I observe abrupt jumps in bike-sharing usage at these thresholds, and the direction of these shifts varies. I construct a coarse thinking model to elucidate these reactions, which holds that deteriorating air quality reduces people’s inclination to go outdoors but increases their demand for bike-sharing to minimize the time exposed to pollution. Those dual effects explain the different directions of the observed over-reactions. I further derive a set of testable implications, which are consistent with empirical evidence from additional analysis.