In the Nation's Interest
Oops: Debt and Economic Growth
By Joseph Minarik
Every now and then in economics, as I suspect in many other fields, some piece of supposedly settled wisdom is thrown into question by a revelation of human error. Such was the case in recent days with respect to a book and several papers by Carmen M. Reinhart and Kenneth S. Rogoff. Reinhart and Rogoff’s research, initially published just before the financial crisis but reprised over several years (and henceforth referenced in the common academic style as “RR”), has been widely cited to make the case that excesses of public debt lead to reduced economic growth. This finding, in turn, undergirds arguments in Europe and the United States for government budget retrenchment.
But earlier this month, three faculty members of the University of Massachusetts-Amherst (Thomas Herndon, Michael Ash and Robert Pollin, hereinafter stuck with the moniker of “HAP”), working in part with a data spreadsheet provided to them by Reinhart and Rogoff, discovered errors in the original calculations.
Given the prominence of Reinhart and Rogoff personally and of their findings in the budget debate, what should we conclude? How, if at all, should our thinking change?
Without burrowing deep into the weeds, but for your background: HAP found two families of what they consider to be errors. First, there were mistakes in spreadsheet formulas, which had the effect of omitting some relevant data from crucial calculations. On that count, RR plead guilty as charged. Second, there are some complex questions with respect to the relative weight in the calculations that should be assigned to different high-debt episodes in different countries. This one is a judgment call. However, given the discovery of the other errors, it looks like RR will need to mount a full defense on this count as well. And in practical terms, HAP present competing findings with this second “error” corrected; those findings have received and will continue to get attention. (In addition, there were some few errors in data entry, but those apparently are not consequential in terms of the results.)
Correcting those errors, it turns out, significantly attenuates RR’s finding that debt is an important determinant of growth. The HAP paper potentially could change the tone of the budget debate. However, I personally would conclude that it should not change the substance. Why?
In my judgment, RR put far too much weight in their original presentations on a curiosity of their findings. Specifically, they found that countries that had public debt in excess of 90 percent of GDP had sharply less economic growth than those below that level. They went to great length to reason why that 90 percent level had such particular significance. Their discussion, and the public reactions to it, have identified this research closely with the notion of a “magic number.” Many advocates of the RR position specifically referenced the 90 percent figure. Do an Internet search for Reinhart and Rogoff, and even if you then insist that the entry include the number 90, you still will get a boatload of references dated before the release of the HAP paper.
All of this seems like false precision – especially given the inevitable imprecisions and conceptual inconsistencies in their monumental multi-country, multi-era data set. (As one case in point, typical knowledgeable economists break out in hives when they recall that RR use gross debt – for example, in the U.S. case, counting the holdings of the Social Security, highway, and other trust funds.) RR’s real achievement, which is far less spectacular than the 90 percent boundary, is to document excessive public debt’s cost in terms of economic growth – which should be obvious enough intuitively, but remains easy for some to ignore or even deny.
HAP, for all of the sturm und drang surrounding the release of their paper, have not changed anything that matters. What HAP have done, really, is to discredit the false precision of the magic 90 percent threshold while leaving the key general point standing. To get a sense of this, consider the following table, derived from figures 1 and 2 in the HAP paper:
(N.R. indicates not reported. Note that HAP presented a finer gradation of the debt-to-GDP ratio to explore the relationship in more detail.)
The conclusion I would draw from this table, accepting all of HAP’s criticisms of the RR data and methodology, is that higher debt and lower growth are at least closely associated. (HAP report that a statistical test fails to reject their null hypothesis that these average growth rates across debt-to-GDP ratio classes are statistically the same. However, beyond the internationally recognized are-you-going-to-believe-what-I-tell-you-or-your-own-two-eyes test, my limited experience with practical statistics suggests that one could reverse the null hypothesis and a statistical test would fail to disprove that the growth rates are different.)
HAP do deserve real credit. The 90 percent figure could have become an excessive deterrent to necessary action in extremis – almost like the old shibboleth that if man flew faster than the speed of sound, his ears would fall off. There may be times when policymakers must grit their teeth and go to a forbidding place, to prevent an even worse outcome. But that having been said, and beyond eyeballing the RR data with the HAP corrections, we can apply some common sense. Ask yourself the simple question: Why would I be happier if my nation carried a smaller rather than a larger debt burden? I can think of four easy answers:
• I’d rather spend my scarce tax dollars on something other than debt service. A former boss of mine, the former Rep. John M. Spratt (D-SC), used to say that if you wanted to destroy popular esteem for government, the easiest way would be to run up an outsized debt burden. People often complain that they pay substantial amounts of taxes, but they don’t receive anything in return. Putting aside that people in their daily lives tend not to think about national defense, law enforcement, food inspection, roads, air traffic control, etc., etc., to the extent that their taxes go to pay the federal government’s debt service, people’s reflex criticism is correct: They don’t get anything back for the taxes they pay.
• I want flexibility to respond to problems like the financial crisis. Suppose that back in 2007-2008 the public debt already had been at the roughly 75 percent of GDP that we “enjoy” today – or an even higher burden. Or suppose that the nation in such circumstances were hit by another major natural disaster. Would you want to find yourself in a dilemma where you could not revive the economy or help your neighbors because the nation had no credibility in the credit markets? Joseph Heller was right: Catch-22 can be a hopeless situation.
• I don’t like to worry about a financial meltdown. If you had trouble sleeping during 2007-2009, enough said.
• There is a relationship – not a cliff, but a meaningful slope – between debt and growth. This reason is associated with the first one above. An extra dollar of debt service means one fewer dollar for investment in infrastructure, human capital, or scientific and technical knowledge; or one greater dollar of tax burden, which distorts economic incentives and the allocation of resources; or some amount of additional debt service next year (repeat ad infinitum). (Some will counter that the nation could cut wasteful government spending instead of true public investment. But once you have cut all the way down to the government you want – and by the way, lots of luck trying to get a majority of the rest of the country to agree to precisely the government you want – the point remains the same. Debt service then will continue to crowd out things you want government to do.)
There are all sorts of qualifications to the RR analysis. An important one is the ambiguous direction of causation. There is plenty of reason to fear that higher debt causes slower growth. But slower growth for some totally separate reason certainly would reduce revenues and increase spending, causing larger deficits and debt. Still, this is just one more reason to control debt in the first place – to avoid a debt crisis should bad news strike. Again, Catch-22 is real, and a real potential problem, and fiscal prudence can avoid it.
Another conceptual issue is the timing of the relationship between debt and growth. A nation with a still-moderate debt that is nonetheless setting off alarm bells (think Spain five years ago) could be hit by financial problems and forced to impose contractionary budget measures, and therefore might take a growth hit. Later, while debt is elevated, the economy might turn around and register very rapid growth in a recovery. Thus, a true causal relationship of high debt leading to slow growth still could populate the RR data set with years of comparatively low debt and slow growth, and other years of elevated debt and very rapid growth. Quantitative analysis of this sort always is perilous and difficult.
But all of that said, we can be confident following our common sense. All else equal, there is every reason – including, but not limited to, the prospects for economic growth – to seek lower debt. Of course, all else is never precisely equal. His one colossal economic misjudgment aside, the late President Lyndon B. Johnson showed sound instincts when he said that you can’t save the chicken farm by starving the chickens. The middle of a down economy is not the right time to squeeze the budget, any more than the morning after a heart attack is the right time to climb on the rowing machine.
But in fact, in an amicable and I think constructive one-on-one that I just had with Robert Pollin, the “P” of HAP (available here), he readily agreed to the broad, common-sense point about debt (though we disagreed significantly on some other issues). It is easy to argue over the precise shape of the relationship between debt and growth, but it is just as easy to agree with the fundamental point that debt matters – on the down side. You can safely follow your instincts – regardless of the graphic headlines you might have seen in recent days.