Deininger and Squire World Bank Inequality Database

BIBLIOGRAPHY

Measurements of income distribution are widely used in studies of social and economic inequality. Although the distribution of income is not the only factor that determines social stratification in a particular country or region, it is one of the most important factors. Economists have been especially interested in the presumed link between income distribution and economic growth. Some researchers have used income distribution as a dependent variable (i.e., a variable influenced by economic growth), whereas others have treated income inequality as an independent variable that affects the prospects for economic growth. Researchers who posit that income distribution and economic growth are reciprocally connected have tested this hypothesis by conducting least-squares regressions of simultaneous equations models.

Despite the frequent use of income distribution as a variable in econometric studies, researchers have long encountered problems in finding reliable statistics. The database entries for a time series of income distribution are fully valid only if they are derived from nationally representative household surveys. Moreover, the entries must be uniform in three respects: First, they must refer to the same type of income (wages, wages plus non-wage income, etc.); second, they must measure income at the same stage (before or after taxes); and third, they must designate the same unit as a recipient of income (households, families, individuals, wage-earners, etc.). A time series that does not include uniform data will not yield meaningful results when analyzed. The collection of uniform data, however, is difficult and expensive, especially for international comparisons of income distribution over many years. Statistical agencies in different countries vary in the way they measure income distribution (if they measure it at all), and many of them fail to provide adequate information about their measurement techniques. In some cases they have altered their methods without announcing it and without adjusting earlier data, thus causing temporal inconsistencies. Some international organizations that compile income distribution data for countries in specific regions, such as the European Bank for Reconstruction and Development and the Asian Development Bank, also have not disclosed enough information about their means of tabulation. Hence, panel datasets with long time series of income distribution covering dozens of countries around the world are rare.

The greatest success in producing comprehensive panel datasets of income distribution has been achieved by the World Bank and by the United Nations World Institute for Development Economics Research (WIDER). The first major advance along these lines was the effort in the mid-1990s by two World Bank economists, Klaus Deininger and Lyn Squire, to put together what they described in a 1996 article as a “new data set on inequality” that would “represent a significant expansion in coverage and a substantial improvement in quality” as “compared with earlier data sets” (Deininger and Squire 1996, p. 566). They noted that “although a large number of earlier studies on inequality have amassed substantial data on inequality, the information included is often of dubious quality,” and they indicated that their own compilation was “based on household surveys, on comprehensive coverage of the population, and on comprehensive coverage of income sources” (p. 567). Deininger and Squire explained why they used two types of distributional indicators—the Gini coefficient (an aggregate measure covering the full population) and the population-quintile shares (a disaggregated index permitting comparisons of population groups)—and highlighted the advantages and limitations of the new dataset. In particular, they explained how to distinguish “high-quality data” from entries that are less reliable and emphasized the advantages of using the new dataset for time-series analysis. The full Deininger-Squire dataset was made available to other researchers and became a standard resource for social scientists.

The Deininger-Squire dataset covers 138 countries in total, though the density of the coverage varies a good deal. For some countries dozens of observations are available, whereas for other countries the number of observations is much smaller. The coverage extends from 1890 to 1996, though the bulk of the information is from the 1960s to the early 1990s, after which the coverage is spotty despite periodic updates. As a result, researchers interested in studying patterns of income distribution since the early 1990s, particularly those who want to examine how income distribution has been affected by the economic and political transformations of the former communist countries in East-Central Europe, have relied on other sources of data that in some cases are of lower quality.

After Deininger and Squire presented their initial dataset, a few other large research institutes sought to compile new panel datasets that would fill in some of the gaps and bolster the quality of the data. Most of these undertakings were designed to build on the Deininger-Squire dataset, though a few were intended to supplant the Deininger-Squire project by relying on different distributional indicators and alternative sources of information. The most successful effort to put together a new dataset— one that built on but went well beyond the Deininger-Squire project—was WIDER’s compilation of its World Income Inequality Database (WIID), which increased the coverage to 152 countries, extended the timeframe to 2003, augmented the number of distributional indicators (including decile as well as quintile shares and the use of an adjusted Gini coefficient to supplement the reported Gini numbers), and expanded the number of data observations to provide for higher quality. The first version of WIID was compiled from 1997 to 1999 and made available in September 2000, and a second, substantially improved version was released in June 2005. WIDER has continued to update the WIID since 2005, allowing further improvements in the quality of data, the extent of international coverage, and the length of the time series.

Despite these advances, the Deininger-Squire project will remain important as the first systematic attempt to compile high-quality data for comparisons of income distribution in dozens of countries over several decades. The Deininger-Squire effort provided a crucial foundation for subsequent compilations of high-quality data on income distribution.

SEE ALSO Gini Coefficient; Income Distribution; Inequality, ncome; Inequality, Wealth; Kuznets Hypothesis; Theil Index; University of Texas Inequality Project

BIBLIOGRAPHY

Atkinson, Anthony B., and François Bourguignon, eds. 2000. Handbook of Income Distribution. Amsterdam and New York: Elsevier.

Babones, Salvatore J., and María José Alvarez-Rivadulla. 2007. Standardized Income Inequality Data for Use in Cross-National Research. Sociological Inquiry 77 (1): 3–22.

Deininger, Klaus, and Lyn Squire. 1996. A New Data Set Measuring Income Inequality. The World Bank Economic Review 10 (3): 565–591.

Deininger, Klaus, and Lyn Squire. 1998. New Ways of Looking at Old Issues: Inequality and Growth. Journal of Development Economics 57 (2): 259–287.

United Nations University, World Institute for Development Economics Research. 2005. World Income Inequality Database: User Guide and Data Sources. Vol. 2.0a (June). Helsinki: UNU/WIDER.

Mark Kramer