The Effects of Road Pricing on Driver Behavior and Air Pollution (Gibson & Carnovale)

Summary

Matthew Gibson and Maria Carnovale's "The effects of road pricing on driver behavior and air pollution" (Journal of Urban Economics 89, 2015, pp. 62–73) is the tightest identified causal study of a cordon charge. Evaluating road pricing is normally confounded because cities bundle it with other changes — Stockholm expanded (older, dirtier) bus service at the same time, muting measured emission effects. Gibson and Carnovale exploit a rare clean shock: in 2012 an Italian court unexpectedly and without press coverage suspended Milan's "Area C" congestion charge, which the city then reinstated eight weeks later (on 17 September 2012). That unanticipated on/off/on sequence is a natural experiment, and — combining high-resolution vehicle counts with monitor-level pollution data — lets the authors recover unconfounded estimates of what pricing does, net of drivers' evasive behaviour. It is the causal capstone to the Stockholm and London before/after evaluations.

Key Findings

• Pricing sharply cut entries. Suspending the charge produced roughly 27,000 additional vehicle entries per day into the priced area; equivalently, the Area C charge reduces vehicle entries by 14.5%.[1]
• And it cut pollution. "Net of these behavioral responses, we find the Area C policy reduces vehicle entries into the priced area by 14.5 percent and air pollution by 6 to 17 percent" — measured across carbon monoxide, PM10 and PM2.5 monitors.[1]
• The welfare gain is large. Scaling a well-identified US willingness-to-pay estimate to Milan incomes, the authors calculate the pollution reduction increases welfare by approximately $3 billion annually — "particularly given that the priced region is just five percent of Milan's land area and the city has an unusually clean vehicle fleet."[1]
• Drivers substitute, but the net effect survives. They document intertemporal substitution toward unpriced hours and spatial substitution toward unpriced roads, and find the traffic effect varies with public-transit availability — routes without good transit alternatives show smaller reductions. This is the direct empirical answer to the standard worry (from license-plate bans and other non-price policies) that evasive behaviour can drive net benefits to zero or below.[1]
• A credible price elasticity. Using Milan's long-run pricing changes (2008–2011 vs. 2012), city-center entries by charged vehicles fall about 0.3% per 1% price increase (passenger vehicles ≈ −0.17), giving a clean elasticity estimate for cordon demand.[1]
• It confirms the prior literature causally. The authors note that earlier cordon studies — Olszewski & Xie on Singapore, Santos on London, Eliasson et al. on Stockholm — "find cordon charges do reduce traffic within the priced area"; their contribution is to establish this with an unconfounded discontinuity.[1]

What It Supports

• Congestion pricing reduces traffic and congestion — supplies the causal keystone: where London and Stockholm are before/after evaluations, Milan is a clean natural experiment (an unanticipated court suspension) showing pricing cuts entries ~14.5% and pollution 6–17%, with large net welfare gains after accounting for driver substitution.
• Congestion pricing — strengthens the concept page's evidence base with a fourth city and the strongest identification strategy in the file.

What It Cuts Against / Honest Limits

• Substitution is real and must be netted out. The paper's own finding is that drivers shift time and route; the 14.5% and 6–17% figures are the effects after those responses. Where transit substitutes are poor, the traffic reduction is smaller — a genuine limit on cordon schemes' reach.
• Local, not global, pollution. The clean-fleet, small-zone context means the welfare figure is Milan-specific; the authors are explicit that elasticities "are necessarily local."[1]
• Externality first. Like the other cases, the paper's frame is externality correction (pollution and congestion), not commons-rent capture — the Geoist reading is this wiki's.

Bears On

• Outcome: Congestion pricing reduces traffic and congestion
• Concept: Congestion Pricing
• Research: The Stockholm congestion-charging trial 2006 (Eliasson et al.) — the before/after case Milan's discontinuity de-confounds
• Research: The London Congestion Charge (Leape 2006)

See Also

• Congestion pricing reduces traffic and congestion
• Congestion Pricing
• The Stockholm congestion-charging trial 2006 (Eliasson et al.)
• The London Congestion Charge (Leape 2006)

Sources

1. Matthew Gibson & Maria Carnovale (2015), "The effects of road pricing on driver behavior and air pollution," Journal of Urban Economics 89, 62–73. DOI 10.1016/j.jue.2015.06.005 — used for (full text fetched and read this session, author-hosted copy, Williams College) the 2012 court injunction and eight-week suspension, the +27,000 entries/day (14.5%) effect, the 6–17% air-pollution reduction across CO/PM10/PM2.5, the ~$3 billion/year welfare estimate, the intertemporal/spatial substitution and transit-dependence findings, and the ~−0.3 (passenger ≈ −0.17) entry elasticity. Williams PDF · RePEc