How is inequality measured?

There is no single number for inequality. Which measure you pick, and which data it comes from, shapes the conclusion. Two approaches do most of the work in the research.

Summary indexes: the Gini coefficient

The Gini coefficient compresses a whole income distribution into one number between 0 (perfect equality) and 100 (one person has everything). It is the workhorse of cross-country comparison because almost every statistical agency produces one.

The catch is comparability. Agencies define income differently — before or after taxes and transfers, per household or per person, gross or net. Frederick Solt (2020) built the Standardized World Income Inequality Database (SWIID) to reconcile those definitions into one comparable panel, with uncertainty estimates attached. It is the dataset behind a large share of quantitative comparative research on inequality and democracy, precisely because it is the only source with wide enough country-and-year coverage. Our country comparison chart uses it.

Top-income shares from tax records

A Gini can hide a lot at the very top, because household surveys undercount the rich and cap ("top-code") the highest incomes. Thomas Piketty & Emmanuel Saez (2003) sidestepped that by building income series directly from individual tax returns — first for the United States back to 1913 — which is how we know the share of income going to the top 1 percent and top 10 percent with real precision over a century.

Anthony B. Atkinson et al. (2011) surveyed the method across more than twenty countries and laid out its hard parts: income definitions differ, tax avoidance and evasion bias the numbers, and comparability across tax systems takes care. Our U.S. top-income-shares chart uses data from the World Inequality Database, the successor project to that work.

Income is not wealth

Both measures above are about income — what people earn in a year. Wealth — what they own — is more concentrated and harder to measure. Thomas Piketty (2014) put wealth and inheritance back at the center of the debate; his data were also seriously challenged on exactly these measurement grounds, which is why the library flags that book as contested.

The practical rule

When you see an inequality statistic, ask two questions: which measure (Gini? top-1% share?), and which data source (survey or tax records — and does it capture the top properly?). Every chart on this site answers both, with a "data as of" stamp.