Case Study 2 — London's Congestion Charge: Pricing a Negative Externality

In February 2003, London became the first major city to impose a congestion charge — a fee for driving into the central city during peak hours. The charge was a direct application of the externality framework: driving into a congested city center imposes costs on other road users (they have to wait longer), on pedestrians (they face more danger and noise), on public-transit users (buses are slowed by traffic), and on residents (air pollution, noise). These are negative externalities that the individual driver does not pay for. The congestion charge is a Pigouvian tax designed to make drivers internalize the external cost of their decision.

This case study walks through London's experience and evaluates whether the Pigouvian-tax approach to congestion works in practice.

The design

The original charge was £5 (about $8 at the time) for any vehicle entering the central London "charging zone" between 7 a.m. and 6:30 p.m. on weekdays. The zone covered about 8 square miles of the densest part of central London — roughly the area bounded by the "inner ring road," including major destinations like the West End, the City of London, Westminster, and Southwark.

Enforcement was by camera: automated number-plate recognition cameras at every entry point photographed every vehicle entering the zone. Drivers could pay online, by phone, or at retail outlets. Failure to pay resulted in a fine (originally £80, later raised).

Exemptions included: motorcycles, taxis, buses, emergency vehicles, residents of the zone (who received a 90% discount), electric vehicles (later tightened), and disabled-badge holders.

The revenue from the charge was earmarked for investment in public transport — primarily bus service improvements.

What happened

Traffic fell. In the first year, traffic entering the charging zone during charging hours fell by about 18%. This was a large and immediate effect. Congestion (measured as average delay per kilometer) fell by about 30% in the first year.

Bus ridership rose. The combination of fewer cars and better bus service (funded by the congestion charge revenue) led to a measurable increase in bus ridership in central London.

Air quality improved. CO₂ emissions from traffic in the zone fell by about 16% in the first year. NOₓ and particulate matter also fell, though less dramatically (partly because the remaining vehicles were on average heavier — trucks and buses — than the passenger cars that switched to transit).

Economic activity was not significantly affected. One of the main fears before the charge was that it would hurt central London businesses — shops, restaurants, offices. The empirical evidence showed no significant negative effect on retail sales or employment in the zone. Some businesses reported initial disruption, but the effect was short-lived. Central London continued to thrive economically.

The charge has risen over time. The original £5 charge is now £15 (as of 2024) and the zone has been expanded. The Ultra Low Emission Zone (ULEZ), introduced in 2019 and expanded in 2023, added an additional charge for vehicles that don't meet emissions standards.

What the externality framework predicts

The London congestion charge is almost a textbook Pigouvian tax.

The externality: driving into congested central London imposes external costs — congestion, pollution, noise — on other road users, pedestrians, transit users, and residents.

The tax: the congestion charge is a fee set (roughly) equal to the marginal external cost of one additional vehicle in the zone. The fee makes drivers internalize the cost they impose on others.

The predicted effect: the tax raises the effective price of driving into central London. The law of demand says quantity demanded (number of vehicles) falls. The reduction in vehicles reduces the externality (congestion, pollution, noise). Total welfare rises because the social cost of the eliminated trips (the trips that were only worth making when the driver didn't have to pay the external cost) was higher than their private benefit.

The predicted revenue: the government collects revenue that can be used for public benefit (in this case, bus service improvement). Unlike a regulation that simply bans vehicles, the charge generates revenue and lets individuals choose: if the trip is worth more than £15 to you, you drive and pay; if it's not, you take the bus or stay home.

The match between prediction and outcome is remarkably close. The charge reduced traffic, improved air quality, generated revenue, and didn't destroy the central London economy. It's one of the cleanest examples of a Pigouvian tax working as designed.

Where it's more complicated

1. Distributional effects. A flat charge of £15 is a much larger burden for a low-income driver (say, a plumber making a service call) than for a high-income driver (a banker commuting to the office). The charge is regressive in the same way any flat fee is. London addressed this partly through the exemptions (taxis, buses, residents) and partly by investing the revenue in bus service that disproportionately benefits lower-income commuters. But the distributional concern is real.

2. Boundary effects. The charging zone has a sharp boundary. Some drivers change their routes to avoid the zone, which shifts congestion to roads just outside the boundary. This is an imperfect outcome — the externality isn't eliminated, it's partly relocated. Expanding the zone (as London has done) helps but doesn't fully solve the problem.

3. The "right" price is hard to set. The external cost of congestion varies by time of day, by location, and by conditions. A fixed £15 charge is too high during quiet periods and too low during the busiest hours. Dynamic pricing (varying the charge with congestion levels in real time, as Singapore does) would be more efficient but is harder to implement and harder for drivers to predict.

4. Political sustainability. The congestion charge was initially unpopular. Ken Livingstone, the Mayor of London who introduced it, won reelection in 2004 partly on the charge's success — but his successor, Boris Johnson, expanded the zone reluctantly and was always politically uncomfortable with it. The charge has survived politically because it works — traffic and pollution are measurably better — but it remains an easier target for political opposition than a subsidy or a regulation.

5. Electric vehicles and the future. The original congestion charge exempted electric vehicles. As EVs become more common, the congestion benefit of the charge declines (EVs still cause congestion even though they don't pollute). London has begun to reduce the EV exemption, which is the right response from an externality-framework perspective (the relevant externality is congestion, not just pollution).

What other cities have done

London's success inspired similar programs elsewhere:

• Singapore has had road pricing since 1975 (originally manual, now electronic). It uses dynamic pricing that varies with congestion levels.
• Stockholm introduced a congestion charge in 2007 after a trial period. The charge was initially opposed but became popular after residents experienced the benefits. A referendum confirmed it.
• Milan introduced "Area C" in 2012. Traffic fell about 30%.
• New York City has been planning a congestion-pricing program for Manhattan south of 60th Street since 2019. Implementation was delayed by political and legal challenges but is expected to launch in the mid-2020s.
• Many cities have not adopted congestion pricing, despite the evidence, because of political resistance. Los Angeles, Chicago, and most other major U.S. cities have no plans for congestion charges.

The pattern: congestion pricing works well empirically but is politically difficult to introduce. Once introduced, it tends to become popular because the benefits are visible and immediate. The political challenge is the introduction, not the continuation.

What this case study illustrates

Lesson 1 — Pigouvian taxes can work in practice, not just in theory. The London congestion charge is one of the cleanest real-world examples of an externality tax producing the predicted outcome: less of the externality-producing activity, more investment in alternatives, and net welfare gains.

Lesson 2 — Distributional effects are real and need to be addressed. A flat charge is regressive. London's response — exemptions for taxis and buses, revenue invested in public transit — partially offsets this but doesn't eliminate it.

Lesson 3 — Political feasibility is the binding constraint, not economic design. The economic case for congestion pricing is very strong. The political case is much harder. Introducing a new cost that drivers have to pay, even when the benefits (less congestion, better air, better buses) exceed the costs, is politically risky. Cities that have overcome this risk have generally benefited; cities that haven't tried are still stuck with congestion.

Lesson 4 — The externality framework applies to many everyday situations. Congestion is not exotic — it's something millions of people experience every day. The framework from Chapter 11 (externality → tax/cap/regulation → closer to optimal outcome) applies to congestion, parking, noise, water use, and dozens of other everyday externalities. The London charge is just the most visible example.

Discussion questions

1. The London congestion charge is £15 per day ($20). Is this the "right" price? How would you determine the optimal congestion charge? What information would you need?

2. A flat charge is regressive. What alternative pricing structures could reduce the distributional burden while maintaining the congestion benefits?

3. Why has the U.S. been so much slower than European cities to adopt congestion pricing? Is it a political problem, a design problem, or something else?

4. As electric vehicles become more common, the pollution argument for the congestion charge becomes weaker but the congestion argument remains. Should the charge apply equally to EVs?

5. Imagine you are the mayor of a mid-sized U.S. city (say, Millbrook, but much bigger). Apply the externality framework. Would a congestion charge work? What obstacles would you face?