We have repeatedly flagged that standard CBA measures efficiency and is silent on — even biased against — distribution, because it aggregates willingness to pay, which is capped by ability to pay. This module brings distribution back in, through two approaches: the classic distributional weights, and the modern marginal value of public funds, which is reshaping applied welfare analysis.
The distribution problem, restated
Unweighted CBA counts a shilling of benefit equally regardless of who receives it, which (since the rich express their preferences in more shillings) tilts toward projects serving the well-off (module 2's clean-water example). The traditional Kaldor-Hicks defence — let CBA handle efficiency and let the tax-and-transfer system handle distribution separately — assumes that costless lump-sum redistribution is available to compensate losers and adjust the distribution. But real redistribution is costly and incomplete (it distorts behaviour, the optimal-tax problem; and compensation is often not paid, the hypothetical-compensation problem). When you cannot rely on a separate transfer system to fix distribution, the distributional consequences of each project matter in their own right — and the appraisal should reflect them.
Distributional weights
Weighting a shilling by who receives it
Distributional weighting multiplies each group's costs and benefits by a weight that reflects the social value of a shilling to that group. The standard basis is the diminishing marginal utility of income: a shilling is worth more to a poor person than to a rich one, by a factor that depends on the elasticity of marginal utility η (the same η as in the Ramsey discount formula): weight ∝ (reference income / group income)^η So with η = 1, a shilling to someone with half the average income is weighted twice as much; a shilling to someone with twice the average, half as much. Applying these weights, a project's benefits to the poor count for more and a project serving the rich must clear a higher bar. The World Bank's Squire-van der Tak framework formalised this for project appraisal. The objection: the weights depend on η, a value judgement, and choosing them can look like rigging the answer — which is why weighting is often shown alongside the unweighted result, making the distributional judgement explicit and transparent rather than hidden.
The marginal value of public funds
The most important recent development in applied welfare analysis is the marginal value of public funds (MVPF), developed by Nathaniel Hendren and Ben Sprung-Keyser. It reframes the evaluation of any government policy around a single ratio.
MVPF: benefit per net government dollar
MVPF = (beneficiaries' willingness to pay for the policy) / (net cost to the government) • The numerator is what the recipients value the policy at (their WTP). • The denominator is the net cost to the government — the upfront cost MINUS any fiscal externalities (e.g., if a child-health or education programme leads recipients to earn more and pay more tax later, that returned revenue lowers the net cost; if a programme induces behaviour that reduces tax revenue, it raises the net cost). The MVPF tells you how much benefit a policy delivers to its recipients per net dollar of public funds it ultimately costs. A policy with an MVPF of 4 delivers $4 of benefit per $1 of net government cost; one with an infinite MVPF actually pays for itself (the fiscal externality covers the cost). Comparing policies' MVPFs — and plotting their MVPF against who benefits (rich or poor) — lets you see both efficiency (bang per buck) and distribution (for whom) in one framework. Hendren and Sprung-Keyser's striking empirical finding: many policies aimed at children (health, education) have very high or infinite MVPFs, because the long-run earnings-and-tax gains largely pay for them — they are not 'costs' but investments. The MVPF has become the organising tool of modern policy evaluation precisely because it unifies efficiency, distribution, and fiscal feedback in one transparent number.
Why the MVPF matters here
For the appraiser, the MVPF reframes the whole exercise. It naturally incorporates the long-run fiscal externalities that conventional CBA often misses (the future tax revenue from a healthier, better-educated population — the human-capital returns of the Development and Health/Education economics courses), and it makes distribution explicit by always asking 'benefit to whom?'. It is especially powerful for comparing very different policies (a cash transfer vs a job-training programme vs a child-health intervention) on a common footing. It does not replace careful CBA of the WTP and the fiscal flows — those are exactly its inputs — but it organises them in a way that answers the question decision-makers actually have: among the things we could spend public money on, which delivers the most welfare per net shilling, and to whom?
Exercise
A government must choose between two policies of equal upfront cost: (A) a job-training programme for unemployed adults, and (B) a primary-school health-and-nutrition programme for poor children. Standard CBA (unweighted) gives them similar NPVs. (1) Explain why unweighted CBA might miss what matters in choosing between them. (2) Apply distributional weights — how might they change the comparison, and what value judgement is involved? (3) Define the MVPF of each and explain how fiscal externalities could make B's net cost much lower than its upfront cost. (4) Explain why B might have a very high or even infinite MVPF and what that implies for calling it a 'cost'.