The Magazine

It Just Gets More and More Dismal

Caution: economists at work.

May 6, 2013, Vol. 18, No. 32 • By ANDREW FERGUSON
Widget tooltip
Audio version Single Page Print Larger Text Smaller Text Alerts

Liberal economics writers agreed. “This is huge,” wrote one in the online magazine Slate. Another, called Dean Baker, suggested that RR’s hands were red with the blood—or at least the pink slips—of unemployed Europeans. Baker began his article in the Guardian with the question “How much unemployment did Reinhart and Rogoff’s arithmetic mistake cause?” His causal chain went like this: RR caused policymakers to worry about debt, the worry led them to impose austerity measures like spending cuts, and spending cuts caused unemployment to rise above 20 percent in Greece and Spain. “This is a mistake that has had enormous consequences.”

RR, Baker summed up, had tried to prove “that high ratios of debt to GDP lead to long periods of slow growth.” In fact, he continued, “the correct numbers tell a very different story.”

In fact, though, they don’t. HAP’s recalculations of RR’s data eliminate the dramatic drop-off in growth rates at the 90 percent ratio. But they also show an unmistakable drift toward slower growth as the debt ratio rises. Countries with up to 30 percent debt-to-GDP ratio, according to HAP’s paper, average 4.2 percent growth; growth falls to 3.1 percent at a 60 percent ratio. Growth increases to 3.2 percent as the ratio reaches 90 percent. After 90 percent, growth averages 2.2 percent.

In other words, by the time a debt ratio rises above 90 percent of GDP, growth will be cut roughly in half—according to HAP’s own calculations. High debt-GDP ratios, once they take hold in a country, tend to last for 15 years or more. An annual loss of 2 percentage points in growth over such a sustained period really starts to add up: A country will be significantly poorer than if it had kept debt under control.

This is why RR, acknowledging the spreadsheet and data errors with the appropriate amount of self-flagellation, seemed relatively undisturbed by HAP’s attempted debunking. “We do not,” they said in a statement, “believe this regrettable slip affects in any significant way the central message of [our] paper.”

By now, though, laymen who follow the RR-HAP controversy will have begun to have doubts about the whole enterprise. One thing that economists do agree on is the power of their elaborate, science-like calculations to describe and predict reality. It is this conceit, clung to by economists right and left, that is undercut by what liberal economists have come to call “the RR scandal.”

The scandal, if that’s the word, goes beyond simple negligence or incompetence or the injection of ideology into science. Problems with RR’s conclusions were plain from the start—so plain indeed that no highly trained economist could see them. For all their mounds of data—“data on forty-four countries spanning about two hundred years”!—only a handful of countries in RR’s set of numbers crossed the 90 percent threshold for any significant length of time; these provide a tiny sample from which to draw sweeping conclusions about how any given country will react under a similar debt load. And as for those mounds: It seems highly unlikely that statistics collected from dozens of sources about scores of countries over hundreds of years will yield reliable data that can be usefully compared across borders and epochs, especially when what’s being measured are such squishy concepts as “public debt” and “gross domestic product.” 

And RR made no distinctions between kinds of debt or their varying effects on growth. Not all debt works the same way. Money borrowed to build roads and bridges has enduring benefits that help an economy grow; money borrowed to invest in Solyndra .  .  . doesn’t. RR can’t tell us which countries had which kind of debt during which periods. 

Finally, there are the graphs themselves. The HAP graph shows a long slow decline in GDP under heavier and heavier debt loads. RR shows a dramatic drop at the threshold. The HAP graph is inherently more plausible precisely because RR’s is so dramatic. Cataclysms are rare in life; it’s why they’re so cataclysmic. Predictable cataclysms that happen on a regular schedule are even rarer. RR’s graph shows countries meandering along until .  .  . WHOMP! The HAP graph shows countries meandering along until .  .  . they keep meandering along, en route to more meandering along. Which is a more accurate picture of life? 

Recent Blog Posts