How government and platform policies influence delivery riders' behavior: A theoretical and empirical study

To uncover the mechanisms underlying the frequent traffic violations committed by food delivery riders during deliveries, this study constructs a “mechanism-behavior” evolutionary game model involving government, platform, and rider stakeholders. Centered on rider payoffs and incorporating order allocation, reward-punishment, and salary mechanisms, the model investigates the impact of policies on rider behavior. Empirical data collected from eight intersections in Wuwei City, Gansu Province during July–August 2023 were used as benchmark parameters, while homologous data from January 2024 served as a control group for comparative evolutionary analysis. The results validate the applicability and explanatory power of the proposed model. Riders' behavioral decisions are primarily driven by the trade-off between economic incentives and risk costs, shaped by the regulatory mechanisms implemented by both government and platforms. The findings indicate that the government and platforms can effectively regulate rider behavior only when the combined cost of joint government supervision and subsidies to platforms is less than the benefits obtained under this strategy, and when the sum of the platform's active supervision cost and subsidies to riders is less than the platform's revenue plus the government's subsidies to the platform. Moreover, riders' strategy will effectively converge toward traffic compliance only when the total fines imposed by the government and the platform exceed the additional income gained from traffic violations. Subsidies provided by the government to the platform or by the platform to riders alone cannot drive the three participants to converge toward positive strategies. Within a certain range, reducing regulatory costs for the government and platforms, while increasing penalty intensity and detection rates for rider violations, can promote the convergence of tripartite strategies toward proactive and cooperative outcomes. These findings provide a scientific basis for platforms to optimize reward–penalty algorithms and for governments to formulate technology-enabled regulatory policies. • From the perspective of participants' interests, a dynamic evolutionary game model is constructed to analyze in detail the relationship between decision-making payoffs and losses of the three participants. It reveals the influence of government and platform policies on riders' behavioral choices, analyzes the tripartite evolutionary paths and stability conditions, and proposes that optimizing policies from both government and platform sides can effectively constrain riders' violations, thereby providing theoretical support for subsequent applications by governments and platforms. • Based on empirical data, this study validates the applicability and reliability of the “mechanism–behavior” model using MATLAB. In the future, the model can serve as the foundation for constructing a “supervision–incentive” linkage algorithm to optimize delivery strategies in real time, promoting the sustainable development of the food delivery industry. • The study underscores the critical role of government and platform policies, technology, and interest alignment in addressing illegal riding behavior, providing practical guidance for local government–platform collaboration and offering useful insights for related sectors such as logistics and express delivery.