Hyperinflation/Archive

Home / About

About, data
& copyright

How the figures are measured, where they come from, who owns what, and what this site does not claim.

How the data works

What counts as a hyperinflation

Phillip Cagan’s definition (1956), the one most economists use: a hyperinflation begins in the month prices rise 50% or more and ends when the monthly rate falls below 50% and stays there for at least a year. 50% a month is about 13,000% a year. The archive holds 20 hyperinflations and one contrast case, Turkey 1990–2005: decades of high inflation, never 50% in a month.

Worst month

The highest price rise in a single month, from the reference table of Hanke & Krus (World Hyperinflations, 2012) unless a case says otherwise. Never an annual rate: mixing the two makes cases impossible to compare. The exceptions are named on their pages: Venezuela (National Assembly estimate) and Lebanon (an implied rate, inferred by Steve Hanke from the black-market exchange rate).

Doubling time and dates

“For prices to double” is the number of days (or hours) prices took to double at the worst month’s pace (Hanke & Krus). The years in a case’s title cover the wider crisis; “above 50% a month” gives the months of the hyperinflation itself.

The curves

Banknotes and histories

Each banknote label is read off the note: denomination and the date printed on it. Each history is written from a knowledge base where every figure carries its source; anything that could not be verified is left out of the page.

Data for reuse, and for language models

Every case has a JSON file (figures, timeline, banknotes, series, sources) and a CSV of its series; all.json holds every case. llms.txt describes the site for AI assistants. Please cite the case page.

Found an error?

Historical figures differ between sources and we will have made mistakes. Open an issue on GitHub with the page, the figure and your source.

Main sources

Disclaimer