Tarduno, Matthew “For Whom the Bridge Tolls: Congestion, Air Pollution, and Second-Best Road Pricing” (November 2021) | WP-323 | Blog

Abstract:
Cities are increasingly adopting road pricing policies to address the congestion and air pollution externalities associated with urban driving. A first-best road pricing scheme would charge drivers according to the social damages associated with each trip. In practice, road pricing often takes the form of cordon zones – regions in the center of a city where road users are charged for entry. These pricing schemes deviate from the first-best policy in two key ways: First, feasible cordon systems cannot account for all of the heterogeneity in trip-level externalities. Second, cordon zones leave nearby roads unpriced, allowing for externality leakage. As a result, it is generally unclear how to optimally set cordon prices. In this paper, I adapt models from public finance to demonstrate how to optimally set cordon prices in the face of these policy imperfections. Calculating optimal prices requires information about (i) the heterogeneity in marginal trip-level externalities, (ii) the relationship between these externalities and individual price-responsiveness, and (iii) the elasticity of substitution between priced and unpriced trips. I then use administrative microdata from bridge tolls in the San Francisco Bay Area to back out each of these parameters. Armed with this model of urban driving demand, I calculate optimal prices for planned cordon zones in three cities – San Francisco, Los Angeles, and New York. In each city, I find that leakage drives optimal peak-hour prices ($2-7) well below average social damages ($4-12). Due to the blunt nature of cordon pricing, these policies are relatively ineffective at internalizing congestion and pollution externalities. In these three cities, I estimate that second-best optimal cordon prices recover 15 to 40% of the welfare gains that would be achieved under an ideal Pigouvian policy. To conclude, I discuss the prospects for improving the performance of congestion pricing through expanding spatial coverage or allowing for granular time-of-day pricing.