Chapter 13: Technical Notes
the imposition of value judgments about the weight given to departures from proportionality at different points in the distribution (Lambert 1989). ADePT Health Financing applies the Kakwani index (Kakwani 1977), which is the most widely used summary measure of progressivity in both the tax and the health finance literatures (Wagstaff and others 1992, 1999; OΓ’€™Donnell and others 2005).3 The Kakwani index is defined as twice the area between a payment concentration curve and the Lorenz curve and is calculated as pK C G, where C is the concentration index for health payments and G is the Gini coefficient of the ability-to-pay variable.4 The value of pK ranges from 2 to 1. A negative number indicates regressivity; LH(p) lies inside L(p). A positive number indicates progressivity; LH(p) lies outside L(p). In the case of proportionality, the concentration curve lies on top of the Lorenz curve and the index is 0. The index could also be 0 if the curves were to cross and positive and negative differences between them cancel one another. Given this, it is important to use the Kakwani index, or any summary measure of progressivity, as a supplement to, and not a replacement for, the more general graphical analysis. Note 20: Progressivity of Overall Health Financing The progressivity of health financing in total can be measured by a weighted average of the Kakwani indexes for the sources of finance, where weights are equal to the proportion of total payments accounted for by each source. Thus, overall progressivity depends both on the progressivity of the different sources of finance and on the proportion of revenue collected from each of these sources. Ideally, the macro weights should come from the National Health Account (NHA). It is unlikely, however, that all sources of finance that are identified at the aggregate level can be allocated down to the household level from the survey data. Assumptions must be made about the distribution of sources of finance that cannot be estimated. Their distributional burden may be assumed to resemble that of some other source of payment. For example, corporate taxes may be assumed to be distributed as income taxes. In this case, we say that the missing payment distribution has been allocated. Alternatively, we may simply assume that the missing payment is distributed as the weighted average of all the revenues that have been identified. We refer to this as ventilation. Best practice is to make such assumptions explicit and to conduct extensive sensitivity analysis.
155