Topic 5 of 64· Public value foundations

Distributional Weighting

Distributional weighting adjusts the monetary value of a cost or benefit according to who receives it, on the principle that an extra pound is worth more to a poor household than to a rich one. HM Treasury's Green Book supplies an explicit method for applying this weighting, built on the diminishing marginal utility of income, so that appraisals do not silently treat a pound gained by the wealthiest decile as equal in value to a pound gained by the poorest.

Why it matters

Standard cost-benefit analysis sums pounds without asking whose pounds they are, which implicitly assumes that a pound is worth the same to everyone — an assumption economists have long known to be false. A household earning £15,000/year experiences a £1,000 gain very differently from a household earning £150,000/year, because the marginal utility of income falls as income rises. Left unweighted, standard appraisal systematically favours interventions that benefit wealthier, already better-off groups, because their higher spending power inflates the monetary valuation of benefits reaching them (a park upgrade near expensive housing "shows" a larger property-value benefit than the same upgrade near cheap housing, purely because prices are higher, not because the welfare gain is greater).

The Green Book's supplementary guidance on distributional analysis, reinforced after the Treasury's 2020 review responded to criticism that appraisal methodology systematically favoured London and the South East, sets out a formal weighting approach based on an assumed elasticity of marginal utility of income of around 1.3 — meaning a doubling of income roughly halves (specifically, 2^-1.3 ≈ 0.41 times) the marginal value of an additional pound. This is not a rounding adjustment: applying it can change which of two competing programmes shows the higher net present value, particularly when comparing an intervention concentrated in a deprived area against one spread across the general population.

The maths

The Green Book's distributional weight for a pound of benefit accruing to a household at income level y, relative to a pound at the national average income level ȳ:

Weight(y) = (ȳ / y)^e

where:
  y  = household income (or income of the affected group)
  ȳ  = average (reference) household income
  e  = elasticity of marginal utility of income (Green Book: approximately 1.3)

Applying weights to net benefits:

Weighted benefit = Σ [unweighted benefit to group i × Weight(y_i)]

A group earning half the national average (y = 0.5ȳ) gets a weight of (1/0.5)^1.3 = 2^1.3 ≈ 2.46 — each pound of benefit to that group counts as worth roughly 2.46 times a pound to an average-income household.

Worked example

Two competing local programmes, each with an unweighted net benefit of £2 million/year, competing for the same regional growth fund:

  • Programme A: a business-support scheme in a prosperous town, average household income £45,000 (roughly 1.3× the assumed national average of £35,000).
  • Programme B: a skills programme in a deprived ward, average household income £18,000 (roughly 0.51× the national average).
Weight(A) = (35,000 / 45,000)^1.3 = (0.778)^1.3 ≈ 0.72
Weight(B) = (35,000 / 18,000)^1.3 = (1.944)^1.3 ≈ 2.53

Weighted benefit A = £2,000,000 × 0.72 = £1.44 million
Weighted benefit B = £2,000,000 × 2.53 = £5.06 million

Unweighted, the two programmes are tied. Weighted for distributional impact, Programme B's benefit is more than three times larger — a result that reverses the funding recommendation and reflects the Green Book's explicit purpose in requiring the weighting to be shown, not just the unweighted benefit-cost ratio.

Charity grant allocation: a funder comparing a £500,000 grant reaching 1,000 low-income households (weight ≈ 2.0, weighted value £1 million-equivalent) against the same £500,000 reaching 1,000 middle-income households (weight ≈ 1.0, weighted value £500,000-equivalent) should show the distributional case explicitly in its board paper, not leave it to be inferred.

Software engineering connection

Distributional weighting rarely appears directly in software delivery metrics, but it should shape how engineering and data teams design measurement and targeting:

  • When building an impact dashboard or benefits calculator, expose the income or deprivation profile of who is affected, not just an aggregate benefit total — aggregate figures without distributional breakdown hide exactly the reversal shown above.
  • Link targeting logic in service design to the same deprivation data the Green Book uses — see Index of Multiple Deprivation — so that a digital service's reach can be assessed for equity, not just efficiency (the contested fourth E in value for money).
  • When an algorithm allocates a scarce resource (appointment slots, caseworker time, a subsidy), an unweighted "maximize total benefit" objective function will, by construction, reproduce the same bias the Green Book's weighting exists to correct — flag this explicitly to policy owners before optimizing.

Pitfalls

  • Applying distributional weights inconsistently across a portfolio. Weighting one programme's benefits but not its comparator's produces a biased, not a fairer, comparison; the Green Book requires like-for-like treatment.
  • Using property or market values as a proxy for welfare without adjustment. Market prices are themselves distorted by existing income inequality, which is exactly what distributional weighting is meant to correct for — using unadjusted market values can double-count the bias.
  • Ignoring within-group variation. Weighting by area-average income (e.g. an Index of Multiple Deprivation decile) can misrepresent individuals who do not match their area's average; use the finest-grained income data reasonably available.
  • Treating the 1.3 elasticity as a universal constant. The Green Book itself notes this is an estimate with a plausible range; sensitivity-test major decisions against alternative elasticities rather than treating 1.3 as exact.

Sources