Comparable Indicators of Inequality Across Countries
narrow definition of the recipient unit. It is also worth noting that analysis of income inequality in certain countries
- notably the USA - tends to focus on the narrower nuclear family rather than the household.
Taking Household Size and Composition into Account
In addition to the within-household allocation, using the household as the income recipient unit raises complex
issues on how to compare incomes across households of different household composition - since a particular level
of income will have quite different implications for a person living alone versus a couple with three children. One
can calculate the income per capita of household members, but this ignores differences between adults and chil-
dren and the fact that larger households can benefit from economies of scale - the additional cost of heating and
light associated with an extra person in a large family may be negligible. Equivalence scales are the standard way
of taking these factors into account, deflating household income by a household-specific factor that is less than
one for each extra household member and may differentiate between adults and children. There is a large litera-
ture concerning the appropriate choice of equivalence scale relativities, derived from a wide variety of methods,
but with no consensus on the most satisfactory.6 This means that the scales used in practice are generally ad hoc,
most often the ‘modified OECD’ equivalence scale of [1 + (0.5 × number of additional adults) + (0.3 × number
of dependent children)] or the ‘square root’ scale which is simply the square root of the number of people in the
household, that is household size. Research investigating the sensitivity of distributional results to the choice of
equivalence scale (see, inter alia Buhmann et al., 1988, Coulter et al., 1992, Jenkins and Cowell, 1994) shows that
this can substantially affect estimates of the relative income position of single versus multi-adult households and
thus for example the position of elderly people relative to couples with children; cross-national comparisons can
also be affected, although within-country trends over time appear less sensitive.
6 Derivation methods are surveyed by Coulter et al. (1992) who emphasize the essentially normative aspects of equivalence scale specifi-
cation.
Page • 17