Methodology

Where the numbers come from

Every figure on this site is derived from published consumer price index data. Nothing is estimated by us, and no figure is produced by a model. Three sources are layered, with a later one winning for any year it covers:

• World Bank World Development Indicators — the consumer price index series, from 1960 onwards. This is the backbone of the dataset and the reason every country is on the same basis.
• IMF World Economic Outlook — the current year only. The World Bank has not published it yet, and without the IMF the site would stop a year short.
• Jordà-Schularick-Taylor Macrohistory Database — years before 1960, for the 18 countries it covers. This is what lets the British pound reach 1751 and the US dollar 1776.

All three ultimately draw on national statistics agencies and central banks. Using them rather than scraping 188 agencies separately is what keeps every country comparable.

How a conversion is calculated

The index is published on a 2015 = 100 basis, the World Bank's own base, so any value shown here can be checked against the source without rebasing. A conversion is the ratio of two index values:

amount × (index in the later year ÷ index in the earlier year)

Cumulative inflation is that ratio minus one. The average annual rate is the geometric mean — the constant rate that compounds to the same total over the same span — not the arithmetic average of the yearly rates, which would overstate it.

Purchasing power is reported as its own figure rather than as the mirror of inflation, because the two are not symmetrical. Prices falling 20% leaves money buying 25% more, not 20% more.

What the figures are, precisely

• Annual averages, not monthly. A year's index is the average of that year's months. A figure for "1990" is not January 1990 or December 1990.
• The current year is a projection. 2026 has not finished and no statistics agency has published it, so the 2026 value is the IMF's forecast. Every figure described as "today" on this site rests on it. Earlier years are published statistics.
• Redenominations are handled explicitly. Where a currency was replaced at a fixed rate — old francs, pre-1993 Mexican pesos, Turkey's 2005 lira — the series is continuous in index terms and the page says which currency a historical amount was actually denominated in.
• CPI is not everyone's inflation. It is a national basket. Housing, healthcare and education have moved very differently from the average almost everywhere, so a personal experience of inflation can differ substantially from these figures.

What is checked before anything is published

Every CPI value on this site ends up stated as fact in a page headline and in structured data, so the build gates on data quality: validate_cpi_data.py flags interpolated runs, outliers and gaps, and the generator refuses to run if it finds errors. A separate check validates the structured data on every generated page.

The US dollar series has been verified against the Bureau of Labor Statistics CPI-U and agrees to within 0.16% over 111 years. Other series are checked for internal consistency and plausibility rather than certified against each national agency individually.

Coverage

188 countries and currencies, 11,354 country-years of data in total, spanning 1751 to 2026. Coverage varies: most series begin in 1960 where the World Bank's does, and only the 18 countries in the Macrohistory Database reach the nineteenth century or earlier.

Use the data directly

The underlying series are published as JSON, one file per country, and may be fetched from any origin. The combined file is /data/cpi/_all.json; a single country is at /data/cpi/{ISO3}.json. These are the same files the calculators on this site read.

If you republish the data, cite the original sources above and check their licence terms, which govern it — not ours.

Corrections

Data is rebuilt from source when the World Bank and IMF publish their updates, which is normally once a year per country, and the date each page's content last changed is published in the sitemap. If a figure here disagrees with a national statistics agency, the agency is right and we want to know.