Pricing parking for fairness — A simulation study based on an empirically calibrated model of parking behavior
通过仿真实验，研究了动态定价和机器学习定价策略对停车公平性的影响，发现动态定价可能加剧不公平，而机器学习有潜力平衡不同政策目标。
It has been widely recognized that public parking, if not managed correctly, can significantly decrease a city’s quality of life due to increased traffic and its impact on mobility and the environment. To avoid these negative effects, various parking policies have been proposed to reduce traffic while guaranteeing high accessibility, especially in city centers. This work investigates different pricing policies for public parking, including dynamic pricing and Machine Learning-based strategies that can directly optimize policy goals, such as improving mobility or accessibility. In doing so, we pay special attention to an aspect often ignored when implementing pricing policies for public parking: fairness with regard to equal outcomes for different social groups. Since the effects of pricing policies are very sensitive to financial inequality, we specifically investigate the impact of policies on different income groups. As a foundation for these experiments, we introduce a parking simulation featuring an empirically calibrated behavioral model of parking. We find that (1) dynamic pricing schemes may negatively impact fairness; (2) fair pricing for parking may require different fees for individual social groups; (3) focusing on single policy goals when devising pricing for parking results in unintended consequences; (4) Machine Learning shows potential for creating pricing strategies combining different policy goals. • Agent-Based Model for parking featuring an empirically calibrated behavioral model. • Parking preferences of agents are determined using a Discrete Choice Experiment. • Combining behavioral modeling and simulation with Machine Learning-based pricing. • Extensive experiments show fairness concerns regarding dynamic pricing for parking. • Machine Learning-based pricing shows potential to combine different policy goals.