How income satisfaction impacts driver engagement dynamics in ride-hailing services

Abstract

To keep up with the rapid growth in ride-hailing demand in a competitive market, it is critical to cultivating active engagement with drivers so that they would spend more time on the ride-hailing platform and serve more passengers. Although it has been recognized that income satisfaction plays an important role in managing driver engagement, little is done to quantify such an impact. To fill this research gap, we develop a hidden Markov model to uncover latent driver engagement dynamics and capture the effects of income satisfaction on them. Specifically, our model differentiates the impacts of incomes above and below expectations through piecewise linear regression utility functions, the breakpoints of which are the latent psychologically expected incomes of drivers. We conduct numerical experiments using real-world ride-hailing data. Results show that our model enhances the goodness of fit and predictive accuracy by more than 35% and identifies the transitions among five distinct driver engagement levels. We derive estimates of expected hourly incomes and sensitivities toward incomes above and below expectations when drivers consider different engagement transitions, and interpret their heterogeneity across drivers through regression coefficient estimates.