World income inequality databases: an assessment of WIID and SWIID

Abstract

This article assesses two secondary data compilations about income inequality -the World Income Inequality Database (WIIDv2c), and the Standardized World Income Inequality Database (SWIIDv4.0) which is based on WIID but with all observations multiply-imputed. WIID and SWIID are convenient and accessible sources for researchers seeking crossnational data with global coverage for relatively long time periods. Against these undoubted benefits must be set costs arising from lack of data comparability and quality and also, in the case of SWIID, questions about its imputation model. WIID and SWIID users need to recognize this benefit-cost trade-off and ensure their substantive conclusions are robust to potential data problems. I provide detailed description of the nature and contents of both sources plus illustrative regression analysis. From a data issues perspective, I recommend WIID over SWIID, though my support for use of WIID is conditional.JEL codes: C81, C82, D31

Determinants of governmental redistribution: Income distribution, development levels, and the role of perceptions

“…An intense discussion has arisen on whether to use the WIID or the SWIID for cross-country analyses on inequality. As Jenkins (2015) argues, any researcher employing cross-national income inequality data needs to acknowledge the benefit-cost trade-off and has to ensure that any analytical conclusions drawn are in accordance with the underlying data concept. In our case, there are some strong arguments advocating for the utilization of the SWIID.…”

“…Whereas earlier versions of the SWIID are entirely based on the WIID, version 5.0 utilizes over ten thousand Gini coefficients from national statistical offices, scholarly articles, Eurostat, the OECD, SEDLAC, Deininger and Squire (1996), as well as Milanovic (2014). Some concerns have been raised with regard to the multiple imputation procedure of version 4.0 of the SWIID (Jenkins, 2015). However, version 5.0 has addressed many of these issues.…”

“…The SWIID 5.0 also contains 100 multiply-imputed values for each of the inequality measures, allowing for multiple imputation (MI) estimation of the empirical models. Using MI yields larger standard errors on coefficients, as it takes into account the imputation variability (for a detailed discussion with a focus on the SWIID, see Jenkins, 2015). To estimate the impact of the uncertainty introduced by the MI procedure on estimation precision, Table (A1) in the appendix reports the results when estimating the models of Table (1) based on multiply-imputed market inequality and redistribution.…”

“…Finally, Table (A2) in the appendix provides a detailed robustness check concerning the influence of top income earners, as there has been some concern regarding the data quality of the top income shares in the SWIID 4.0 (Jenkins, 2015). The table reports the effect of the top-1%, top-0.5%, top-0.1%, top-0.05%, and the top-0.01% on redistribution based on the WID, the source data upon which the top income series of the SWIID 4.0 relies.…”

Finance and Income Inequality: A Review and New Evidence

“…SWIID is based on WIID, but is supplemented by other sources and has all of its observations multiply-imputed (Jenkins, 2015). 14 WIID often provides more than one Gini coefficient for the same country/year.…”