The value of Shared Information for allocation of drivers in ride-hailing: a proof-of-concept study

TL;DR

This study explores whether ride-hailing companies should share demand and allocation information with drivers, using game theory and simulations to assess potential benefits of information sharing.

Contribution

It introduces the Expected Value of Shared Information as a new indicator and provides a simplified model with explicit reformulations for analyzing information sharing benefits.

Findings

Sharing information can be beneficial for drivers and companies

The model's simulations indicate potential improvements in driver allocation

Preliminary results support further investigation into information sharing benefits

Abstract

For drivers in ride-hailing companies, allocation within the city is paramount to get matched with rides. This decision depends on many factors, where some of them (such as demand and allocation of others) are unknown for the drivers, but are available for the company. In this work, we investigate whether it is beneficial or not for the ride-hailing company to share this information with their drivers. To do so, we study the problem through the lens of Stackelberg games, and we propose a new indicator called the Expected Value of Shared Information. We present a simplified model to conduct a proof-of-concept study: we provide explicit single-level reformulations of the bilevel programming problems derived from the model, and perform several simulations with randomly generated data. Our preliminary results suggest that sharing information could be beneficial and deserves to be further studied.