Toll Credits and Discounts Help Low-Income Drivers Access Roads More Fairly

Researchers are increasingly focused on equitable implementations of congestion pricing to address societal concerns surrounding road usage costs. Chih-Yuan Chiu from the Georgia Institute of Technology, School of Electrical and Computer Engineering, Devansh Jalota from Columbia University, Data Science Institute, and Marco Pavone from Stanford University, Department of Aeronautics and Astronautics, present a comparative study of two prominent approaches: credit-based congestion pricing (CBCP) and discount-based congestion pricing (DBCP). This collaborative work, undertaken across these three institutions, investigates which method more effectively reduces costs for users and maximises toll revenues. By formulating a non-congestion game and analysing equilibrium flows, the team demonstrates conditions under which DBCP demonstrably outperforms CBCP in achieving optimal societal outcomes, validated through a case study of the 101 Express Lanes Project in the San Francisco Bay Area. This research offers crucial insights for policymakers considering the implementation of fairer and more efficient congestion pricing schemes.

Researchers are refining strategies to address traffic congestion and its disproportionate impact on lower-income commuters. Congestion pricing, a system of levying tolls on busy roads, offers a potential solution, but concerns about fairness have limited its widespread adoption. This work directly compares two approaches, credit-based congestion pricing (CBCP) and discount-based congestion pricing (DBCP), designed to mitigate these equity issues by subsidizing access for those with limited means; CBCP allots travel credits, while DBCP offers direct toll discounts. The study establishes a robust framework for analysing these policies, formulating a model of how drivers choose routes in response to tolls and subsidies. Within this framework, low-income users are granted either travel credits or toll discounts when accessing express lanes. Researchers demonstrate that a stable state, a Nash equilibrium, always exists and can be determined using convex programming, a powerful optimisation technique. The core of this investigation lies in a non-atomic congestion game, a mathematical representation of traffic flow where individual drivers make decisions without significantly impacting overall congestion, ensuring the feasibility and tractability of analysing both CBCP and DBCP policies. The primary result centres on identifying conditions where DBCP consistently achieves lower societal costs than CBCP. This societal cost incorporates both the expenses incurred by drivers and the revenue generated from tolls. Under specific conditions, the study proves that if maximising toll revenue is prioritized, DBCP will yield equal or lower societal costs compared to CBCP; furthermore, with a minor additional assumption, DBCP demonstrably minimizes societal cost, although CBCP can outperform DBCP if the societal cost prioritizes minimising travel costs for eligible users. Comparing the resulting equilibrium values, they assessed users’ travel costs and generated toll revenue. The experimental results align with the theoretical predictions, confirming that DBCP outperforms CBCP under the specified conditions, strengthening the policy implications of the work. Deploying discount-based congestion pricing (DBCP) demonstrably outperforms credit-based congestion pricing (CBCP) under specific network conditions, inducing equilibrium outcomes that minimise societal cost. Analysis reveals that DBCP policies consistently achieve lower equilibrium societal costs when network structure, user attributes, and cost parameters align with the research’s detailed Assumptions 1-4. Specifically, the study establishes conditions where DBCP provably minimizes a societal cost function encoding both user cost reduction and toll revenue maximisation. These experiments compared the equilibrium values of user travel costs and generated toll revenue realised by optimal CBCP and DBCP policies, indicating that DBCP policies consistently outperform CBCP policies in minimising the defined societal cost. The research team adapted zeroth-order gradient descent methods to compute optimal policies for consecutive segments of the US-101 highway, providing a practical validation framework. This computational approach allowed for a detailed comparison of the two pricing mechanisms across a realistic traffic network. The work establishes that Nash equilibrium flows always exist within the formulated non-congestion game and can be accurately approximated using convex programming techniques. This ensures the feasibility and tractability of analysing both CBCP and DBCP policies. Furthermore, the study’s formulation of CBCP allocates credits solely for express lane access, differentiating it from prior work focusing on monetary payouts, while the DBCP policies specifically target low-income users, addressing equity issues within existing tolling systems. The persistent challenge of funding infrastructure while ensuring equitable access has long vexed urban planners and policymakers. This research delves into credit-based and discount-based congestion pricing, moving beyond theoretical debate to rigorously compare their practical efficacy. What distinguishes this work is establishing clear conditions under which one demonstrably outperforms the other in balancing cost reduction with revenue generation. The findings suggest that, under certain circumstances, directly discounting tolls for low-income users is more efficient at minimising overall societal cost than allotting travel credits, with significant implications for cities considering such schemes. However, the modelling relies on specific assumptions about driver behaviour and cost structures; real-world implementation will inevitably introduce complexities not captured in the simulation, such as varying income distributions or unpredictable traffic patterns. Future research should explore the impact of these factors and investigate how these pricing models interact with other transport policies, like public transport investment, to move towards a truly integrated and equitable transport system.