Measurement of the Progressivity of Public Expenditures and Net Fiscal Incidence: Comment

Abstract

Donald Kiefer [1] raises two objections to my paper [2] which proposed methods of measuring the progressivity (regressivity) of public expenditures and net fiscal incidence. His first comment is concerned with the interpretation of points along the expenditure benefit curve, and his second comment is concerned with the ultimate usefulness of my summary statistic, K, which is designed to measure the progressivity of net fiscal incidence. In my judgment, the first comment is not useful. The second comment provides some useful qualifications to my index but is generally overstated. Regarding the first comment, it must be stressed at the outset that Suits's [4] index of tax progressivity and my [2; 3] indexes of expenditure and net fiscal incidence progressivity measure progressivity across the entire range of incomes under consideration. Appropriately, Kiefer [1, 578] reminds the reader that Both Suits and Kienzle emphasized the need for caution in interpreting the progressivity measures when the concentration curve crosses the proportionality line, that is when the distribution is progressive in one income range and regressive in another. The purpose of my Figure 2 and the related discussion [2, 199-200] was to show that it is possible for an expenditure benefit curve to intersect, and thus crossover, the diagonal in a box diagram with the accumulated percent of total public expenditures measured on the vertical axis and the accumulated percent of total income measured on the horizontal axis. It was also my purpose to indicate the implications of such a crossover. For example, suppose the expenditure benefit curve initially shows the accumulated percent of expenditure benefit to be greater than the accumulated percent of total income but then switches to a situation where the accumulated percent of expenditure benefit is less than the accumulated percent of total income. It is possible that the impact of the latter could outweigh the impact of the former and vice-versa. I attached significance to the crossover in the sense that it marks the point beyond which the data tend to change the index from progressive toward regressive. This seems natural given the construction of the index. In any event, whether the particular expenditure involved is progressive or not can not be known until we have accumulated across all income.'