Methodology

Where the figures come from

Every series is the headline consumer price index published by the country's own statistics office. Most sit on a 2015 = 100 base; a few keep the base their publisher uses, the United States on 1982–84 and Argentina on the base of its composite series. It makes no difference to anything here, because every figure on this site is a ratio between two readings of the same series and a base cancels out of that. UK figures come from the ONS, US figures from the Bureau of Labor Statistics, EU members from Eurostat and the OECD, and the rest from OECD and IMF republications of each national release.

Where a country publishes more than one measure, the one used is the measure that country headlines, so a figure here matches the one its own office quotes. The UK is the exception a reader is most likely to notice: both CPI and RPI are offered, because both are in common use there.

What each figure means

Current inflation is the change in the index over twelve months, comparing the latest reading with the same month a year earlier. It is not the change since the start of the year.

Inflation in a year, in the year-by-year tables, is that year's average price level against the previous year's average. This is the definition national offices and most reference works use, so the figures match the ones quoted elsewhere: it gives the USA 11.0% for 1974 and 13.5% for 1980. Measuring January to January or December to December gives noticeably different answers, and would put this site at odds with every other source.

What a sum is worth today takes the amount in January of the year named and scales it by the ratio of the latest index to that month's. It is the change in the general price level, not the price of any particular thing.

What is left out

A year needs a complete set of readings on both sides to have an average, so a series that opens partway through a year has no rate for that year or the one after it, and a year still running has none until it ends. Those years are omitted rather than estimated from the months that exist.

Where readings stop and resume — Belgium's records break for the occupation and pick up again in 1947 — the missing years are absent from the tables and the page says the list is not continuous.

Comparing one country with another

The comparison tables rank countries by their own latest published figure, which means the dates differ: offices publish on different schedules, and each row says which month or quarter it is on.

Baskets and methods differ between countries too. Two national measures are not a perfect like-for-like comparison, and the tables say so where they appear. What they are good for is the order of magnitude and the direction of travel, which is what the ranking shows.

Taking the data away

Every country page carries a link to its whole series as a CSV: one row per month or quarter, the index and the year-on-year change, with the country, the measure and the publishing office named in each row so the file still says where it came from after it has been renamed and mailed on. Months an office has declared it never measured are present as empty rows rather than left out.

Where a country publishes two measures the file holds both, one after the other, told apart by the series column — the UK's CPI and RPI, and the national index alongside Eurostat's harmonised one for the six EU members that have both. The two can start decades apart, and the link says how far each of them reaches.

The homepage and the country index offer the whole archive in one file: every reading of every series for every country, which is every one of those files stacked into one, with the same columns and the same rules. It runs to tens of thousands of rows and a few megabytes, so the single country's file is the better download unless the whole set is what is wanted.

The same archive is published on Kaggle for anyone who would rather work with it there, in a notebook or alongside other datasets. It is the identical file, uploaded again whenever a statistics office publishes something new, so the two do not need reconciling and neither is the stale one.

Using the figures

Every number on this site is an official statistic, published free of charge by a national statistics office for anyone to use. Nothing here is proprietary and nothing is behind a paywall: the ONS, the US Bureau of Labor Statistics, Eurostat, the OECD, the IMF and the national agencies named on each country's page all release their consumer price index publicly, and this site gathers those releases into one place.

So the figures are free to use, and what is normally asked in return is a credit to the office that produced them. That office is named at the foot of its country's page and in every row of every file, so it travels with the data.

What this site adds is the gathering, the rebasing that lets any two dates be compared, and the arithmetic on top. It claims nothing over the figures themselves and cannot grant permission on any office's behalf: each sets its own terms, and where a particular use needs clearing, the named office is the body to ask. A link back here is welcome but is not a condition of anything.

Accuracy

The build checks itself. Figures quoted in the written histories are cross-checked against the series they sit on, superlatives are verified against the whole series rather than the part before them, and the build fails loudly rather than publishing a claim it cannot support.

None of that makes the site infallible. If a figure here disagrees with the office that published it, the office is right and this is a bug worth reporting.