The Need for New Inequality Measures
If science consists in a search for patterns in data—and just as much, if it consists in applying formulae to facts—then the study of economic inequality suffers from an original sin. From the beginning, the job of measurement was badly done. In most countries, measures of economic inequality never became part of the official statistical routine, in the national income accounts or labor statistics. Among governments, the United States is one of just a handful that release an annual measure of income inequality based on a substantial household survey. Observations and measurements of inequality across countries and through time have for the most part relied on occasional and in many cases unofficial surveys, with results that are sparse, often conceptually inconsistent, affected by differences in top-coding practice and subject to the hazards of sampling.
The historical record of these efforts, once undertaken, is what it is. One cannot take a retrospective survey; there is no way to go back to a household and ask what its income was five, ten, or twenty years in the past. Thus the gaps can’t be filled; the methods with which the original data were created cannot be used to repair the archives. And yet interest in inequality persists, the need for information persists, so economists and applied statisticians make do with the data at hand. For much of the postwar period, data were sparse, so the few researchers who worked in the area concentrated on developments within single countries, such as the United States, the United Kingdom, or India, extrapolating common patterns of economic development from a small number of historical cases. Everyone knew this was not a very satisfactory way to proceed.
In 1996, Klaus Deininger and Lyn Squire of the World Bank (hereafter DS) published a collection of many disparate surveys of income and expenditure inequality and compiled those meeting certain criteria1 into a single