Twelve global databases publish Gini coefficients. They cover 222 countries from 1867 to 2024 — over 122,000 observations. The Gini coefficient is a single number between 0 and 1 measuring income inequality. Simple, comparable, universal.
Except it isn't. For the same country in the same year, published Gini estimates diverge by up to 50 percentage points. Not 5. Fifty. A country could be described as moderately egalitarian or severely unequal depending on which database you consult.
The divergence comes from a chain of measurement decisions, each individually reasonable, that compound into incomparability. The largest source is welfare metric selection: income versus consumption. Measuring what people earn and measuring what they spend produce systematically different pictures of inequality. After that, sub-metric definitions (gross versus net income, which transfers to include), equivalence scale choices (how to compare households of different sizes), and post-survey adjustments each contribute additional variance.
Any single Gini figure is the product of this chain. The decisions are rarely fully disclosed. Researchers comparing countries across databases are often comparing measurement methodologies, not societies.
The structural point is not that measurement is hard. It is that a single scalar cannot absorb its own provenance. The Gini coefficient looks like a fact — a number attached to a country and a year. It is actually a function of a long sequence of methodological choices, and the sensitivity to those choices exceeds the variation between countries. The measurement frame is more variable than the thing being measured.