Measurement and mitigation of the “wild goose chase” phenomenon in taxi services

This paper examines the so-called wild goose chase phenomenon in taxi systems and proposes a new mitigation strategy based on dynamic vehicle swaps. An analytic queuing network model, including a system of differential equations, is derived to yield approximate formulas for the expected system performance in the steady-state equilibria under the vehicle swap strategy. This model is solved numerically to yield equilibrium points and the expected system performance metrics. To corroborate the theoretical predictions, a new data-processing method based on dynamic programming is proposed to identify and measure multiple distinct equilibrium states from observed data. Agent-based simulations are conducted to generate examples of such data, and to verify the formulas. Numerical results show that (i) the proposed measurement method works well in identifying the distinct equilibria, and (ii) the vehicle swap strategy works quite effectively in mitigating the wild goose chase phenomenon.