Calculating welfare weights: practical steps

A3.19 To appraise the impact of policies using distributional weights, the equivalised income for two groups is estimated:

taxpayers as funders of policies (group t) - who are assumed to have an average income (using median equivalised income)

programme participants who benefit from the policies (group p) - who are assumed to be in the lowest equivalised income group, given DWP's overall objectives. For other departments applying this approach, interventions may be targeted at groups with higher incomes. If that is the case a higher income estimate should be used.

A3.20 Using the "taxpayer" and "programme participant" approach welfare weights can be estimated as follows:

divide median equivalised income of average taxpayers (proxied by median of all households) by the median equivalised income of programme participants (proxied by the quintile that matches the target for distributive effects)

raise this number by the power of 1.3 (the estimate of elasticity of marginal utility of income as set out above)

the result is the redistributive effect for an individual member of the group being affected by a policy change

Fujiwara (2010) uses this methodology to estimate a welfare weight of 2.5, based on income figures from the Office for National Statistics. Using more recent 2015 data37 yields a slightly lower welfare weight of 2.4.

A3.21 The weighted impact resulting from any redistribution is as follows:

impact on society = change in income, group p * welfare + change in income, group t

A3.22 There is uncertainty in both weighting and equivalisation methods. Presenting unweighted and weighted costs and benefits side-by-side shows the impact of the weightings. Testing the estimated weights through sensitivity analysis, including switching values where appropriate, is recommended. Switching values estimate the value that a key input variable (in this case the income weights) would need to take for a proposed intervention to switch from a recommended option to another option (see Chapter 5).

A3.23 Not all distributional effects can be or need to be quantified or monetised. To decide whether the intervention's distributional effects are of a scale necessary to monetise or quantify them, the underlying rationale or potential consequences of the intervention should be considered.

A3.24 When considering how to apply the guidance above consider the following steps:

does the policy have different financial impacts across income groups, or is redistribution a consequence? If yes, then consider applying welfare weights.

is the analysis targeted at individuals or a mixture of households of different size and composition? If the latter then equivalisation may be required, prior to applying weights.

is the income of the intervention group known? If known it should be used to calculate the welfare weight. If not, then the HBAI income groups can be used.

_________________________________________________________________________________

37 With median equivalised income per week for the bottom and middle quintile, respectively £244 and £481.