How Accurate is the Kakwani Index in Predicting Whether a Tax or a Transfer is Equalizing?

Abstract

The Kakwani index of progressivity is commonly used to establish whetherthe effect of a specific tax or transfer is equalizing. In the presence ofreranking or Lambert’s conundrum, however, a progressive tax could beunequalizing. While it is mathematically possible for counter-intuitiveresults to occur, how common are they in actual fiscal systems? Using anovel dataset that includes fiscal-incidence results for 39 countries, we findthat the likelihood of the Kakwani index to be progressive (regressive),while the tax or transfer is unequalizing (equalizing), is minimal, except inthe case of indirect taxes: in roughly 25 per cent of our sample, regressiveindirect taxes are equalizing (sign-inconsistent cases). Additionally, thelikelihood that the Kakwani index ranks the magnitude of the impact of atax or transfer wrongly also exists but it too is small. Finally, usingregression analysis, we find that increasing the size or progressivity of aprogressive tax or transfer is equalizing and statistically robust forsign-consistent cases. For sign-inconsistent cases, the coefficient for theKakwani index is not statistically significant. In sum, although theKakwani index could yield interpretations that are inaccurate in actualfiscal systems, the risk seems small, except for indirect taxes.