Policy Analysis Under Contested Evidence: A Road-Safety Levy
A national ministry is considering a congestion and road-safety levy on private vehicles entering the capital during peak hours. The transport ministry's model predicts a 14% reduction in peak traffic and a 9% fall in collisions, based on data from two comparable Asian cities. The finance ministry counters that the levy will raise the cost of living for 180,000 low-income commuters who cannot afford to change modes, and that the revenue estimate (about 4.2 billion taka annually) assumes near-universal compliance that has never been achieved in similar schemes. A civil-society group argues the model ignores informal transport, which carries roughly 40% of peak trips. The minister wants a decision before the next budget, and has asked for a policy note that 'tells me what to do, not just what the models say'.
1. Evaluate the quality of the evidence base and identify the key analytical gaps.
Read the worked answer
The evidence base is borrowed, partial, and internally contested. Three gaps stand out:
- External validity. The 14% and 9% figures come from two comparable cities, but comparability is asserted, not demonstrated. Mode share, enforcement capacity, income distribution, and informal transport differ, and each of these moderates the effect.
- Distributional analysis is missing. The finance ministry's 180,000 figure is an aggregate; a proper analysis would identify who bears the burden, whether it is regressive, and what compensating measures would cost. A levy can be efficient and still unjust.
- Compliance and informality. The revenue estimate assumes near-universal compliance, which is the weakest link in any enforcement-dependent instrument. The 40% informal share means the model's behavioural assumptions may simply not apply to a large part of the target population.
A good analyst also flags what is not measured: air quality, accident severity (not just count), and the effect on delivery and emergency vehicles. The note should state confidence levels explicitly rather than presenting point estimates as certainties.
2. Recommend a policy option and justify it in terms the minister can act on.
Read the worked answer
The recommendation should be a phased, differentiated levy with a compensating package, not a single blunt instrument:
- Phase 1 (pilot, 6 months): Apply the levy to a defined cordon and to private cars only, with a low initial rate and a published evaluation plan. This generates local compliance data rather than importing assumptions.
- Differentiate by ability to pay: Exempt or discount low-income commuters, and pair the levy with a visible improvement in public transport on the same corridors. Compensation must be credible, not promised.
- Enforce with technology and transparency: Automatic number-plate recognition with an appeal mechanism; publish compliance rates quarterly.
- Set decision triggers in advance: If compliance falls below a stated threshold, or if the low-income burden exceeds a stated level, the rate is adjusted or the cordon narrowed.
This answers the minister's request because it converts contested models into testable commitments. It also protects the ministry politically: the policy can be corrected on evidence rather than defended to the end.