Assessing the impacts of tradable credit schemes through agent-based simulation

Abstract

Tradable credit schemes (TCS) are an alternative to congestion pricing, offering revenue neutrality and the potential to address equity concerns through the credit allocation. Past research on the performance of TCS has largely relied on simplified network and market equilibrium models that may fail to capture the complexities of transportation demand, supply, and credit market interactions. Agent- and activity-based simulation provides a more comprehensive approach by explicitly modeling individual traveler behaviors and market dynamics. This study proposes an integrated simulation framework for TCS implementation within the open-source urban simulation platform SimMobility, featuring: (a) a flexible TCS design that accounts for multiple trips and individual trading behaviors; (b) a simulation framework that models interactions between travelers, the TCS regulator, and the market; (c) TCS optimized using Gaussian Processes and Bayesian Optimization, and (d) simulation experiments on a large-scale mesoscopic multimodal network. Results show that network and market performance stabilize over time, aligning with theoretical TCS properties from network equilibrium models. We confirm the efficiency of TCS in reducing congestion and explore its varied impacts on users, travel behavior, and market dynamics. Our framework allows for designing different TCS configurations and testing their effect in mitigating potentially undesirable trading and market behavior, ultimately contributing to a closer-to-practice design and assessment.