Earlier this week, the Congressional Budget Office released its budget projections for the next decade. Its finding, that both the budget deficit and the debt-to-GDP ratio are recovering nicely from their recessionary spikes, is unsurprising. But its timing is impeccable. The CBO projections served as the closing rimshot to a surprisingly entertaining routine featuring two Harvard economists, a UMass graduate student, and a global cast of fiscal conservatives.
For the past few years, the work of Kenneth Rogoff and Carmen Reinhart has provided the intellectual foundation for the turn to austerity in the United States and elsewhere. The argument, most directly expressed in their 2010 paper “Growth in a Time of Debt,” is that the rate of economic growth slows dramatically when the level of public debt surpasses 90 percent of gross domestic product. Rogoff and Reinhart imply a clear causal relationship, concluding that high debt-to-GDP ratios are “associated with notably lower growth outcomes.” And they leave little doubt that the consequences are not confined to slow growth, warning that as economies approach the “debt tolerance ceiling,” interest rates will spike, “forcing painful adjustment.”
The political implications are clear. Deficit hawks invoked Reinhart and Rogoff’s “tipping point” to drag federal fiscal policy from stimulus to sequester in a few short months. Despite persistent unemployment (and absolutely no hint of upward pressure on interest rates), congressional Republicans leaned on the brakes. This was a strategy most economists (even the staid voices at the Federal Reserve) viewed as premature, counterproductive, and destructive.
And then things got interesting. In mid-April, economists at the University of Massachusetts-Amherst discovered elemental coding and calculation errors in Rogoff and Reinhart’s spreadsheets. In a devastating critique, Robert Pollin, Michael Ash, and Thomas Herndon called into question the basic math (and motives) of the high-debt-equals-low-growth argument. Rogoff and Reinhart tried to dismiss the dispute as an arcane academic kerfuffle and doubled down on the core argument. This tack served only to reinforce the perception that this was a political stance—a “damn-the-numbers” defense of austerity against all critics.
In the end, the stakes might be quite small: the satisfaction of being right is not worth much if fiscal policy remains unchanged. But the stakes could also be pretty high. Puncturing the intellectual legitimacy of the whole austerity agenda may not change Paul Ryan’s mind, but it could (or should) give pause to his timid enablers among moderates in both parties. For this reason, Rogoff and Reinhart’s “Excel for Dummies” error has become an opportunity to raise much broader questions about both the scholarship in question and its political uses. Why are Rogoff and Reinhart so wrong?
The spreadsheet error is a small element of this story but, like the pratfall of a mediocre figure skater, it is the one people will remember. In configuring an Excel formula (the part where you specify, or drag a box across, a range of cells), Rogoff and Reinhart mistakenly excluded five countries (Australia, Austria, Belgium, Canada, and Denmark) from a key part of their analysis. And then, in bringing growth rates for New Zealand from one part of the worksheet to another, Rogoff and Reinhart made a transcription error that dropped four years of data. Correcting these errors alone erases much of apparent gap between rates of growth under conditions of low debt and high debt.
2. The Omissions
Rogoff and Reinhart also omitted data for three countries (Australia, Canada, and New Zealand) for the late 1940s—an era of demobilization from war that featured particularly volatile rates of growth and debt. In these years, each of these countries managed respectable rates of growth despite high levels of public debt. At one point, Rogoff and Reinhart imply that the immediate postwar years in these countries are excluded as outliers—but they do not exclude the same years for the United States, where growth slowed much more dramatically with demobilization. When these omissions are corrected the gap narrows even further.
3. The Weighting Problem
A further mathematical curiosity in Rogoff and Reinhart’s work is their method for weighing and averaging rates of growth. For each year a country spends in a debt category (such as the above 90 percent threshold), there is a corresponding rate of growth in GDP. Rogoff and Reinhart add these up by country and then average them: if Country “A” spends ten years in the high debt category, its rate of growth is the average across those ten years; if Country “B” spends five years in the high debt category, its average is across those five years. What makes this an issue is the fact that Rogoff and Reinhart add these up and then average them again in order to establish a rate of growth for all countries in each debt category. The conventional method would be to weigh these averages by country and by year, so that a country spending twice as long in the high debt category carries double the weight in the final, across-country, average.
So what’s the net result? Rogoff and Reinhart, with much drumbeating, found that countries above the 90 percent debt threshold actually saw their economies shrink. But, once the coding errors and omissions and weighting issues are resolved, those countries show an average growth rate of 2.2 percent—a little slow, but nothing resembling the doomsday of stagnation and “painful adjustment.”
4. Thresholds, Ceilings, and Tipping Points
Once all the numbers are cleaned up, we are still left with the curious conviction that there is a magic threshold—debt at 90 percent of GDP—we dare not cross. This is a little like saying that nausea will abruptly kick in after you eat thirty-six cookies. Well maybe—but the results would presumably depend on what else you have eaten that day, your pre-cookie physical condition, the kind of cookie you are eating, and a range of other factors.
National, historical, and economic context matters a great deal in all of this. Cringing at the 90 percent threshold is just another version of the apocalyptic silliness that has the United States as “the next Greece.” As Brad Delong and others have pointed out, the level of debt may not matter much as long as interest rates remain low and inflation is not an issue. As Dean Baker has argued, this fixation on debt ignores the fact that countries also have assets: two countries at the 90 percent threshold may be as dissimilar as two homeowners with identical mortgages and incomes—but with very different equity positions and future prospects. And, as Josh Bivens and John Irons underscore, the American side of this story is almost entirely driven by the peculiar conditions of defense demobilization in the 1940s.
5. Chicken and Eggs
In a half-hearted mea culpa in late April, Rogoff and Reinhart acknowledged their errors but stuck to their view that there was “a vicious feedback loop between debt and growth.” Even the corrected numbers, they claimed, show slow growth (just not as slow) following from high debt. This reasoning, as Arin Dube and others pointed out, is their most profound and troubling error. Rogoff and Reinhart marshal no evidence to make the case that debt slows growth; indeed, the causality more plausibly (and demonstrably) runs in the other direction. When growth slows, public revenues shrink and demands on the public purse (especially from “automatic stabilizers” like unemployment insurance or food stamps) grow. This is not an accident of fiscal excess; it is what governments do (and should do) in hard times.
From a policy perspective, of course, this is no idle distinction. It could mean the difference between mindless budget slashing sure to prolong the recession, and public spending and public investments sufficient to allow growth, employment, and public revenues to recover.
Colin Gordon is a professor of history at the University of Iowa.