The Vardham Congestion Charge and the Falling Accident Count
City councillors in Vardham introduced a £6 weekday congestion charge across a 12 km² central zone in January. In the twelve months before the charge the zone recorded 500 injury accidents; in the twelve months after, it recorded 410, a fall of 18%. Traffic entering the zone, measured by automatic counters, fell by 20% over the same period. In the local paper the council's transport spokesperson wrote: 'The charge cut accidents by nearly a fifth. Had we not acted, about ninety people would have been injured who were not. Other cities should copy Vardham.' Two other changes took effect in the same January: a bypass opened, diverting through-traffic around the zone, and the default speed limit inside the zone was cut from 50 km/h to 30 km/h. Average vehicle occupancy was unchanged. The council holds no data on walking and cycling volumes inside the zone, which residents report have increased.
1. The spokesperson claims the charge 'cut accidents by nearly a fifth'. Using only the figures in the setup, assess how far the figures support a safety improvement rather than simply a reduction in traffic.
Read the worked answer
Normalise by exposure. A count of accidents is not a measure of risk; it has to be divided by the amount of driving.
- Before: 500 accidents per (say) T vehicle-kilometres.
- After: 410 accidents per 0.80T vehicle-kilometres = 512.5 per T.
Interpretation. The number of accidents fell by 18%, but the distance driven fell by 20%. Per vehicle-kilometre the accident rate therefore rose by about 2.5%, from 500/T to about 512/T. Taken at face value, the roads inside the zone did not become safer per unit of travel; there was simply less travel.
Qualification. The conclusion depends on the denominator. If most of the traffic reduction came from through-traffic that was also the highest-risk traffic, the per-km comparison could be unfair. But on the data given, the honest statement is that the charge reduced exposure, not demonstrated risk. The spokesperson's headline conflates fewer accidents with safer driving.
2. Identify the two most serious threats to the inference that the charge itself caused the fall in accidents, and describe a comparison that would test the claim more convincingly.
Read the worked answer
Threat 1 - co-occurring changes (confounding). A bypass and a 50 → 30 km/h limit were introduced in the same January. Speed reduction is a plausible direct cause of fewer and less severe injuries, and the bypass physically removes traffic, and therefore crashes, from the zone. Either could produce an 18% fall with no contribution from the charge. The bypass also raises the possibility of displacement: collisions may simply have moved to the ring road outside the counted area.
Threat 2 - regression to the mean and secular trend. Central zones are usually chosen because their accident counts were unusually high, and unusually high counts tend to fall anyway. City-wide safety, vehicle technology and fuel prices were also changing.
A better test. A difference-in-differences design: compare the change inside the zone with the change in a comparable control area over the same period, and also count casualties in a band just outside the zone boundary to detect displacement. A simple before–after count in one area cannot separate the policy from its companions.
3. The spokesperson concludes 'other cities should copy Vardham'. State what would have to be true for that recommendation to follow, even if the charge did cause the fall in Vardham.
Read the worked answer
Even granting causation, the recommendation needs three separate things to hold:
- External validity of the mechanism. The effect must depend on features Vardham shares with the other city - dense centre, adequate public transport, an enforceable boundary. If Vardham's gain came mostly from the bypass absorbing displaced traffic, a city with no ring road may get displacement without benefit.
- Transferability of the outcome. The target must be the outcome measured. If the true gain was fewer casualties because of slower speeds, a city that adopts a charge but keeps 50 km/h limits should not expect the same result.
- Net effect, not gross. The recommendation compares the zone with itself, ignoring costs borne elsewhere (congestion and crashes shifted onto surrounding roads, compliance costs, effects on traders). A policy can be good for the charged zone and neutral or harmful for the city.
Without evidence on at least the first two, 'copy Vardham' is an over-generalisation from a single uncontrolled case.