Sensitivity Analysis of Equivalence Scale in Income Inequality Studies

Coulter et al. (1992) argues that the use of different equivalence scales produces the estimations of the inequality indexes with U-shaped pattern where ϴ ≈ 0:5 marks the lowest estimate, and that the McClements Scale used in UK would produce lowest estimates of inequality indexes similar to the case where ϴ ≈ 0:6. Banks and Johnson (1994) counter-argues that such property of McClements Scale depends upon the data at hand and equivalence scale specification in terms of the way children are counted against the number of adults within a single household, rather than on the underlying property of the McClements Scale. This study will, first of all, illustrate that the main source of such dispute about the McClements Scale may come from the proportion of large-size family as well as the single parent households with two young children in the sample data. Secondly, it is argued that such differences in inequality measures from using any values of equivalence scale may not matter, contrary to the common belief, once we employ statistical inferences where the differences in estimated values are not statistically significant.