Methodology
Inequality: sources, units and contests
Every figure on the inequality page is computed at build time from four maintained public series. This page states what each one measures, the license under which it is used, the unit conventions that move levels, and the two places where the field's headline numbers are genuinely contested. Nothing here is a forecast, and no exhibit endorses a policy.
What the numbers measure, and the units
Top shares (income and wealth concentration) come from WID.world, the Piketty-Saez-Zucman distributional national accounts. They are measured on EQUAL-SPLIT ADULTS: household income or wealth is divided equally among the adults in the household, and every adult is one observation. This is not the same as a household ranking or a per-capita ranking, and it typically shows lower top shares than tax-unit series. The unit is stated on every WID exhibit.
Gini coefficients come from household surveys: the OECD Income Distribution Database (disposable, i.e. post-tax post-transfer, income), the World Bank PIP (income OR consumption, depending on the country's survey), and the UNU-WIDER WIID (a compilation across concepts). A Gini summarises the whole distribution in one number between 0 and 1; the OECD and PIP report it on a 0-1 scale, WIID on 0-100.
Fiscal-data top shares and survey Ginis are different objects
Top shares built from tax and national-accounts data (WID) and Ginis built from household surveys answer different questions and have different blind spots, so they are never combined into a single ranking. Surveys undercount the very top (nonresponse and top-coding); tax data miss income that never appears on a return; and both differ from the equal-split-adult national-income concept WID uses. A country can also pair a low income Gini with high wealth concentration, because wealth is a stock and income is a flow. That is why income and wealth are shown as separate objects.
The US top-1% income share: Piketty-Saez-Zucman vs Auten-Splinter
The most-cited inequality statistic is the least settled. The maintained WID series follows Piketty, Saez and Zucman. Using the same tax data but allocating untaxed and retained income differently, Gerald Auten and David Splinter find top shares markedly lower and roughly flat since the 1960s. For 2014, the top 1% share of fiscal income is 21.5% under Piketty-Saez, 16.7% under the Congressional Budget Office, and 13.1% under Auten-Splinter. The gap is driven by the treatment of retained corporate earnings, underreported income, and imputed pension and government income: PSZ distribute more of these to the top, Auten-Splinter less. Neither has been refuted, and the disagreement itself is the current state of knowledge, so both are shown wherever a US top share appears.
Cite: Piketty, Saez and Zucman (Quarterly Journal of Economics 133(2), 2018), maintained in the World Inequality Database. Auten and Splinter, "Income Inequality in the United States: Using Tax Data to Measure Long-Term Trends", Journal of Political Economy 132(7), 2024, pp. 2179-2227; the 2014 comparison is from Auten and Splinter, AEA Papers and Proceedings 109, 2019, pp. 307-311.
Survey, administrative and capitalization methods for wealth
Wealth concentration is harder to measure than income and more sensitive to method. Surveys undercount top wealth; estate-multiplier and tax-capitalization methods (Saez-Zucman) infer wealth from capital-income flows and are sensitive to the assumed rate of return, with Smith, Zidar and Zwick's heterogeneous-returns estimates putting US top wealth materially lower. The WID wealth series used here is one maintained estimate, not the last word; wealth series also carry wider extrapolation than income, so the newest years are treated as estimates.
Poverty lines: pinned to 2021 PPP
Global poverty is counted at the World Bank's international poverty lines, which were rebased to 2021 purchasing power parities in June 2025: the extreme-poverty line rose from $2.15 (2017 PPP) to $3.00 (2021 PPP), with $4.20 and $8.30 companions replacing $3.65 and $6.85. A PPP vintage moves the global count by tens of millions, so every poverty figure states the line and the PPP year explicitly, and the platform pins the release (PIP 20260324, 2021 PPP). Nowcasts (years after the latest survey-based estimate) are labelled as such.
WID extrapolation and quality flags
WID extends every country's series to the most recent years, filling gaps by interpolation and extrapolation, so a ranked cross-section at the latest year mixes measured and estimated values. WID's own row-level data-quality flag is carried into the estate, and the long-run line charts, which rest on the measured historical record, carry more information than any single recent cross-section.
Bangladesh, and coverage gaps stated not hidden
Bangladesh appears wherever a source covers it: PIP (survey Gini and poverty), WIID (with quality flags), and WID (with extrapolation flags). It is absent from the OECD, which covers members and a few key partners only. Where a source does not cover an economy, the table says "not covered" rather than dropping the row, so the gap is visible.
Sources and licenses
WID.world (World Inequality Lab, Paris School of Economics). The project describes its data as "open and convenient access to the most extensive available database on the historical evolution of the world distribution of income and wealth"; it publishes no formal redistribution license. Classified display-with-attribution: FinObservatory shows computed statistics with citation and links wid.world, and never re-serves WID files or bulk data.
World Bank Poverty and Inequality Platform (PIP): Creative Commons Attribution 4.0 (World Bank Data Catalog, "Poverty and Equity Database"). Cite: World Bank (2026), Poverty and Inequality Platform, version 20260324_2021_01_02_PROD. pip.worldbank.org.
OECD Income Distribution Database: under the OECD Terms and Conditions, users may "extract from, download, copy, adapt, print, distribute, share and embed Data for any purpose, even for commercial use", with appropriate credit to the OECD.
UNU-WIDER WIID, Version 29 April 2025: the database "can be downloaded for free"; cite as UNU-WIDER, World Income Inequality Database (WIID), Version 29 April 2025, https://doi.org/10.35188/UNU-WIDER/WIID-290425.
In every case, the owner's no-redistribution ruling applies on top: the platform displays computed statistics with attribution and never re-serves source files, rows, bulk data, or an API.