Methodology & honesty

These are careful best guesses, not audits. Comparing incomes across 145 countries honestly requires a few choices, and here are ours:

• PPP dollars, not exchange rates. We compare what money buys where you live, not what it trades for at the airport. All figures are in international dollars.
• Take-home income ÷ household size. The surveys behind these curves measure what households actually have to live on — disposable income in most higher-income countries, household consumption in most others — so take-home is the closest honest input to both. If you enter a before-tax figure instead, we approximate your country's deductions from published single-worker rates (~35 countries; we always show you the take-home we derived, and your own number beats our estimate). Incomes support households, so we divide by the people they support — the standard convention in this field. It also means a single person ranks higher than the same salary feeding a family of four; that's the convention working, not a bug.
• Past the top 0.1%, we stop counting. Inside the top 1% our global distribution still resolves tenths ("the top 0.4%"), so we show them. But every survey thins to a handful of respondents at the very top; "the top 0.1%" is as fine a claim as the data can honestly carry, so that's where our claims end. Surveys also tend to miss or under-report the richest households: estimates built from tax records, such as the World Inequality Database's, put the top of the curve higher than any survey shows. We don't splice those in, because the machine stays on one consistent source, and it is one more reason the top is where our claims end.
• Income, not wealth. Wealth data is patchy and self-reported net worth is noise; every serious comparison tool makes the same call.
• The machine has a worldview, and we'd rather name it. The numbers show a world that keeps getting better across the decades, even through years that set it back. Our reading — and it is a reading — is that the improvement came mainly through ordinary work and exchange: people making something for someone and being paid for it. If the data ever stops showing that improvement, this page will change.
• "Extreme poverty" means the World Bank's international line — $3.00 a day at 2021 PPP, the same dollars as everything else here, via Our World in Data. It is a deliberately austere line; crossing it is a floor, not a finish line. The trend's last few years are the Bank's nowcast, not a survey: the line reaches today because the Bank projects it there, and we show the projection as a round number only.
• Country curves come from the World Bank's Poverty and Inequality Platform — survey-based percentile distributions, the best public source there is. They are still estimates, and at the very top of every curve surveys run out of respondents: if you're in the top fraction of a percent, we say "about" and mean it.
• Personas are illustrations, not statistics. The occupations we name sit roughly where typical earnings for that work sit on each country's curve. They are there to make a percentile feel like a person — treat them as portraits, not payroll data.
• Four levels, for every country — and the only description where no vignette exists. For countries without hand-written personas the machine describes the income level instead — the objects that tend to arrive with it around the world, never the character of the people or of any one country. The cuts are the World Bank's: $3.00 a day (its international poverty line) and $8.30 (its upper-middle-income line), both at 2021 PPP. The top cut, $33, is our own — four times the $8.30 line. How much of each country and of the world lives on each level is read straight off the percentile curves.
• Your income never leaves this page. The math runs on your device. We keep no accounts, no histories, and no record of any income anyone ever typed. We do count page views — in aggregate, with no cookies and no identifiers — so we know which pages get read, never what anyone typed.

Snapshot: World Bank PIP percentile distributions, fetched September 2026 · 2021 PPP dollars · poverty trend via Our World in Data · world population 2025 (World Bank SP.POP.TOTL).

The sources, precisely

• Country income curves: World Bank Poverty and Inequality Platform, 100-percentile distributions per country-year, 2021 PPP; merged into a 1,001-point global distribution weighted by population (World Bank API).
• PPP and prices: World Bank consumption-PPP conversion factors (PA.NUS.PRVT.PP, 2021) and national CPI (FP.CPI.TOTL) to express today's money in 2021 prices.
• Extreme poverty over time: Our World in Data, share of people below the World Bank's $3.00/day (2021 PPP) international line.
• Population, age, births: UN World Population Prospects; the ever-born anchor from PRB, "How Many People Have Ever Lived on Earth?" (2022).
• Two scales, stated side by side: official development assistance runs about $220 billion a year (OECD DAC, 2023, current US dollars); the poorer half of the curve earns for itself about $6 trillion a year (integrated from this site's own curve, 2021 PPP dollars). Different units, so the machine never states them as a ratio.
• Take-home from before-tax: curated single-worker effective rates informed by OECD Taxing Wages, ~35 countries; the derived figure is always shown, and your own number wins.
• Land: Natural Earth.

Suggested citation: The Perspective Machine (theperspectivemachine.com), built on World Bank PIP data, 2021 PPP.