The Congestion Charge Model: How Much Should a €180m Welfare Gain Be Trusted?
A transport ministry is evaluating a congestion charge for a city of 1.4 million residents. Consultants built a spreadsheet model in which weekday car trips fall according to a single assumed elasticity of −0.35 with respect to the cost of a trip, and in which all charge revenue is recycled into bus services. The model reports a 12% fall in car trips, a 9-minute reduction in average peak journey time and a net welfare gain of €180m a year. The ministry's own analysts note three things: the elasticity was estimated from a 2019 fuel-price study in a different country; the model holds the road network, population and commuting patterns fixed; and it values journey-time savings using average wages. A residents' group argues the charge is unfair to low-income drivers, while a business group argues it will damage the local economy.
1. The €180m welfare gain is presented as a single point estimate. Identify the three assumptions that most drive that number and explain how the analysts should test them before advising the ministry.
Read the worked answer
1. The transferred elasticity (−0.35). This parameter generates the 12% trip reduction, which then generates the time savings, so it is first-order. It was estimated on fuel prices, a cost paid weekly and spread across all car use, whereas a charge is salient, paid per crossing, and easier to avoid by re-timing or re-routing. Short-run elasticities are also smaller than long-run ones. Test: re-run the model across a plausible range (say −0.15 to −0.60), report the range of outcomes rather than one number, and run a small pilot in one corridor.
2. Valuation of journey time at average wages. The theory is that time should be valued at the individual's marginal rate of substitution between time and money, which for non-work time is typically a fraction of the gross wage, and average wages overstate the value for retired, part-time and unemployed travellers. Test: sensitivity to a 30–60% wage fraction and to differentiated values by trip purpose.
3. Ceteris paribus on the network and population. A comparative-static model cannot capture induced demand, re-routing onto residential streets, diverted bus passengers, or land-use change over ten years. The assumed full recycling of revenue into buses also risks double counting: the transfer is not itself a resource gain, and bus benefits must be counted once as time savings, not again as revenue.
Best practice is threshold analysis: state how low the elasticity or the time value must fall for the €180m to become zero or negative. If the sign survives the whole plausible range, the policy case is robust; if it flips at −0.25, the ministry is being sold a forecast dressed as a measurement.
2. Classify each statement as positive or normative and justify your classification: (a) 'the charge will cut peak car trips by 12%'; (b) 'the charge is unfair to low-income drivers'; (c) 'revenue should be recycled into buses rather than used to cut fuel duty'.
Read the worked answer
(a) Positive. It is a conditional prediction that can in principle be falsified against data from a pilot or from the outturn. Note that 'will' makes it a forecast, not a certainty; a positive statement can still be wrong.
(b) Normative, but with a positive core. The word 'unfair' embeds a value judgement about which distribution of burdens is acceptable. To make it testable you must first supply the criterion: is the charge regressive relative to income, relative to car use, or relative to ability to pay? The distributional incidence of the charge - who pays relative to who benefits from the faster buses - is a factual question; whether that incidence is unfair is not.
(c) Normative. 'Should' signals a policy prescription and therefore a value judgement about objectives (equity versus efficiency versus emissions). The supporting arguments are usually positive - that bus investment has a higher benefit-cost ratio, or that fuel-duty cuts leak heavily to higher-income drivers - but the conclusion does not follow from the facts alone.
The practical point for a policy analyst is not to pretend to be value-free, but to separate the two layers so that disagreement about facts can be resolved by evidence and disagreement about values can be made explicit and debated politically.
3. The ministry proposes to ring-fence the charge revenue for bus improvements rather than paying it into general taxation. Explain the opportunity cost of that choice and why the model's €180m welfare gain cannot simply be added to the estimated benefits of the bus programme.
Read the worked answer
Opportunity cost. The relevant cost of ring-fencing is the best alternative use of the same money - either lower general taxation, or higher spending on health or on road maintenance, or debt repayment. Ring-fencing does not create resources; it changes who decides. If bus investment earns a return of, say, 25% per euro while the marginal use of general tax revenue earns 40%, ring-fencing destroys value even if the bus scheme is worthwhile in absolute terms. The correct comparison is always marginal: the bus scheme should be evaluated against the next-best marginal project, not against doing nothing.
Why the €180m cannot simply be added to the bus benefits. First, the charge revenue is a transfer, from drivers to the transport authority, not a resource cost or gain in itself. Counting the revenue as a benefit and then counting the bus improvements it buys as a further benefit double counts the same euro.
Second, the model already monetises the journey-time savings that faster buses deliver to remaining bus users and to car users who switch. Adding a separate bus-benefit figure risks counting those same minutes twice.
Third, there is a genuine welfare gain from the charge that has nothing to do with the buses: it prices the congestion externality, so trips that are worth less than their marginal social cost are deterred. That gain survives even if the revenue is burned.
The right presentation separates three items: the efficiency gain from corrected prices, the distributional transfer, and the net present value of the bus investment measured against its own counterfactual.