June 6, 2018 Reading Time: 2 minutes

In retrospect, it looks obvious why WWI had to generate high inflation: Germany, for example, had to pay back a ridiculous amount upon losing the war and then printed its way to hyperinflation. Looking across countries, however, it is not obvious why hyperinflation happened where it did. A new NBER working paper by Jose Lopez and Kris Mitchener argues a major driver of hyperinflation was general policy uncertainty.

Many countries saw high inflation: over 30 percent per year over 1914–18 for such countries as France, Holland, and Italy. However, only Germany, Austria, Hungary, and Poland went through hyperinflation, defined in the paper as at least a doubling of prices within one month. 

The common explanations for the hyperinflation do not seem sufficient. It was not just about who won the war. Bulgaria and Turkey (losers in the war) did not see hyperinflation, while Russia (a winner in the war) did see hyperinflation. High government debt was also not sufficient. Belgium, France, Italy, and the UK all had debt-to-GDP ratios of over 100 percent by 1920 but did not have hyperinflation.

War reparations, border uncertainty, high debt, communist parties’ taking power, and other circumstances all combine to generate policy uncertainty. After WWI, people did not have any idea what was going to happen politically. Who would fall under what political jurisdiction? What would the new governments formed in the losing countries have to pay back to the victors?

As the authors put it: “Unresolved political issues cast a pall over all of postwar Europe and proved counterproductive for generating credible fiscal policy. The most widely known of these was the intractable debate over reparations payments.” The authors claim this uncertainty was an important driver in tipping countries into the self-fulfilling process of hyperinflation. 

To make this claim, Lopez and Mitchener build on a recent branch of macroeconomics that indirectly measures policy uncertainty, which unfortunately can never be measured directly. The authors collected daily exchange rates to generate a measure of realized volatility as a proxy. The idea is that if a country is going through a time of uncertainty, such as after the assassination of Germany’s foreign minister, some people are going to flee the currency and others are going to try to profit on those fleeing. The combined forces move around the exchange rate.

For Germany, Austria, Hungary, and Poland, the authors show a strong correlation between volatility and inflation within a particular month. That is not surprising since once a hyperinflation starts, one should expect volatility to increase as both domestic citizens and foreigners adjust their holdings of the currency experiencing a hyperinflation. 

More interestingly, volatility precedes the hyperinflation. For Germany, Hungary, and Poland, inflation correlates with volatility in the few months before. In contrast, for those countries that did not experience a hyperinflation, there is almost no correlation between volatility and future inflation. 

This leads the authors to their main claim: Mild policy uncertainty does not drive mild inflation. However, large policy uncertainty does drive large inflation — that is, hyperinflation. The paper finds more evidence that hyperinflation, such as that in Venezuela today, is different in kind from the 1–3 percent inflation experienced in the United States, not just different in size.

Brian C. Albrecht


Brian C. Albrecht is an assistant professor of economics at Kennesaw State University. His research interests include political economy and monetary economics. He has published articles in scholarly journals, including the Journal of Economic Methodology and the NYU Journal of Law & Liberty.

Albrecht earned his M.A. in Economics from the University of Minnesota, his M.Sc. in Economics of Public Policy from the Barcelona Graduate School of Economics, and his B.A. in Physics and Political Science from St. Olaf College.

Get notified of new articles from Brian C. Albrecht and AIER.