International income distribution: comparing new ICP and the existing data

Purpose - The purpose of this paper is to provide a comparison of data on real (purchasing‐power‐parity – PPP) gross domestic product (GDP) per capita recently released by International Comparison Program (ICP) with the numbers reported in World Development Indicators (WDI) and Penn World Tables (PWT) which have been used by hundreds, perhaps thousands, of researchers over many years. Design/methodology/approach - A descriptive comparison is first provided by listing cases of largest absolute and percentage differences between ICP and WDI and PWT numbers. Second, well‐recommended measures of cross‐country inequality in real GDP per capita are computed, and inequality in terms of ICP data are compared with that based on WDI and PWT data. Findings - First, there are huge differences for numerous countries between the ICP numbers and the WDI and PWT data. Second, many of these differences are much larger than the highly publicized cases of China and India. Third, since ICP numbers are more accurate, existing WDI and PWT data are noted to substantially understate intercountry income inequality. Fourth, comparison of ICP with WDI shows a pattern which is similar to that indicated by a comparison of ICP and PWT. Fifth, the huge discrepancies in data provided by highly reputed sources, and used by numerous researchers, in such a prime indicator of economic and social well‐being seem to reflect a notable phenomenon. Originality/value - First, this is apparently the only attempt to provide a comparison of the new ICP data on country‐level real GDP per capita with that reported in the highly reputed and widely used WDI and PWT. Second, the enormous differences for numerous countries should suggest much caution to researchers in using the existing WDI and PWT data series. Third, the substantial understatement of intercountry income inequality by WDI and PWT data should be noteworthy. Fourth, although authors of WDI and PWT will probably identify reasons for the differences and reformulate their PPP data series, the present study may suggest need for some reflections on the context in which such large‐scale discrepancies in a variable of primary economic and social significance have existed for many years.