In 2010, two Harvard professors, Carmen Reinhart and Ken Rogoff, published a paper in the American Economic Review, one of the world’s most-respected economics journals, arguing that when the ratio of a nation’s federal debt to its GDP rises above a 90% tipping point, the nation is likely to slide into an economic recession.

This claim has potentially serious implications but there were serious problems with the paper (which are discussed more fully in my book Standard Deviations: Flawed Assumptions, Tortured Data, and Other Ways to Lie with Statistics). Thomas Herndon, a graduate student at the University of Massachusetts Amherst, tried to replicate the reported calculations and could not.

Reinhart/Rogoff had made a spreadsheet error that omitted five countries (Australia, Austria, Belgium, Canada, and Denmark). Three of these countries had experienced debt/GDP ratios above 90% and all three had positive growth rates during those years. In addition, some data for Australia (1946–50), Canada (1946–50), and New Zealand (1946–49) are available, but were inexplicably not included in the Reinhart/Rogoff calculations.

The New Zealand omission was particularly important because these were four of the five years when New Zealand’s debt/GDP ratio was above 90%. Looking at all five years, the average GDP growth rate was 2.6%. With four of the five years excluded, New Zealand’s growth rate during the remaining high-debt year was a calamitous -7.6%.

Unusual averaging

There was also unusual averaging. To illustrate, let’s just look at the above-90% years for the United Kingdom and New Zealand. There are 19 United Kingdom high-debt years, and the average GDP growth rate was 2.4% in those years. Because of the four omitted years, New Zealand had one high-debt year, with a GDP growth rate of -7.6%. The average growth rate for these two countries is 1.9% if we average all 20 years. But Reinhart and Rogoff instead averaged 2.6 and -7.6, and reported the average growth rate to be -2.5%.

The bottom line is that Reinhart and Rogoff reported that the overall average GDP growth rate in high-debt years was a recessionary -0.1% but if we fix the above problems, the average is 2.2%.

An even more fundamental problem is that correlation is not causation. Fiscal-austerity enthusiasts interpreted the Reinhart/Rogoff study as evidence that increased government debt reduces economic growth and that increasing the level of government debt past 90% of GDP is likely to cause an economic recession.

Perhaps the causation goes the other way. During recessions, government tax revenue falls and government spending on unemployment insurance, food stamps, and other safety nets increases, both of which increase government debt. If there is a statistical correlation between economic growth and the debt/GDP ratio, it may be mostly—or even entirely—due to the effects of the economy on the debt ratio, rather than the other way around.

An academic kerfuffle?

In a 2013 New York Times opinion piece, Reinhart and Rogoff dismissed the criticism of their study as “academic kerfuffle.” Unfortunately, this was much more than that. Millions of people lost their jobs, income, and self-respect; nations lost trillions of dollars of output.

When the Reinhart/Rogoff paper appeared in 2010, the US and many European countries were still using aggressive stimulus policies to keep the Great Recession of 2007–09 from turning into a second Great Depression. Fiscal hawks in many countries used the Reinhart/Rogoff study as proof that governments should reduce spending and increase taxes in order to balance their budgets or, even better, run a surplus so that government debt could be repaid.

On April 5, 2011, Reinhart and Rogoff testified before 40 US Senators and urged an “adjustment” to cut government debt. When Senator Johnny Isakson asked, “Do we need to act this year? Is it better to act quickly?,” Rogoff responded, “Absolutely. Not acting moves the risk closer.… You have very few levers at this point.”

Reinhart argued that the fact that there had been so few instances of debt/GDP ratios above 90% demonstrated that 90% was a tipping point: “If it is not risky to hit the 90 percent threshold, we would expect a higher incidence.” She warned ominously that, “The sooner our political leadership reconciles itself to accepting adjustment, the lower the risks of truly paralyzing debt problems down the road … Countries that have not laid the groundwork for adjustment will regret it.”

The US did not move quickly or aggressively but several European governments did cut spending and increase taxes and the average unemployment rate in Europe rose from 10% in 2011 to 11% in 2012 and 12% in 2013. The International Monetary Fund, a frequent advocate of austerity, later admitted that European austerity measures had been much more harmful than they expected.

The test of fresh data

As I have written in many places (for example, here), the replication crisis that is undermining the credibility of science and scientists involves the failure of reported research studies to replicate when they are tested with fresh data. Prominent examples are claims that hurricanes with female names are more deadly than are hurricanes with male names (here) and that power poses can increase a person’s testosterone and reduce cortisol (here and here).

One formidable obstacle to retesting many economic theories is that we may have to wait several years for new data to accumulate. The Reinhart/Rogoff study used data through 2009. We now have 14 years of additional data. In the US, the debt/GDP ratio went above 90% in 2010 and is currently 122. The figure below shows the overall ratio of central government debt to GDP for 38 OECD countries:

In 2009, the last year analyzed by Reinhart/Rogoff, the aggregate debt/GDP ratio was 84%, dangerously close to their 90% tipping point. The ratio crossed 90% in 2010 and kept going, never again falling below 90%. According to the tipping-point argument, we should have had a worldwide economic recession since 2010. We haven’t, because the tipping-point argument is bogus.

There was a COVID-19 recession that began in February 2020 and was over by May 2020 in the United States and somewhat later in many other countries. The surge in the debt/GDP ratio certainly did not cause the COVID recession. On the contrary, the Great Lockdown coupled with government efforts to keep their economies afloat caused a sharp increase in the debt/GDP ratio—which then fell somewhat after the lockdown ended and government support programs waned.

The post-2009 data clearly disprove the Reinhart/Rogoff assertion. First, the worldwide GDP ratio has been above 90% for many years without causing a worldwide recession. Second, the large spike in the debt/GDP ratio in 2020 and 2021 demonstrates that causality often runs in the other direction: fluctuations in the debt/GDP ratio are not the cause of business cycles but the result of them.

Reinhart/Rogoff’s dismissal of criticism of their deeply flawed study as “academic kerfuffle” is unconscionably cavalier. Their paper is a prime example of the harm that can be caused by sloppy science.