By James D. Agresti
May 6, 2013
Advocates for higher government spending are abuzz over a new working paper that disputes a famous paper often trumpeted by conservatives. The famous paper found that high levels of national debt are associated with lower economic growth, a result that conservatives have repeatedly cited to argue that governments should stop accumulating debt.
This new working paper exposes calculation errors in the famous paper, critiques its methodology, and presents competing findings. Liberals have latched onto these findings to argue that nations should be less concerned with government debt and should increase government spending to “stimulate” their economies.
While the authors of the working paper make significant contributions to this debate, they and numerous commentators who are citing their work have used their findings to mislead rather than inform. They have done this by leveling false accusations, ignoring an important follow-up paper written by the same authors, and failing to reveal that the new findings are similar to that of the famous paper: high levels of national debt are associated with slower economic growth.
For a 2010 paper published in the American Economic Review, Carmen M. Reinhart of the University of Maryland and Kenneth S. Rogoff of Harvard University researched and tabulated the national debt and economic growth in 20 advanced economies (such as the United States, France, and Japan). Using 2,000+ data points from over 200 years, the authors found that “high debt/GDP levels (90 percent and above) are associated with notably lower growth outcomes.” Relevantly, U.S. federal debt surpassed 90% of GDP in 2010 and has now reached 105% of GDP.
However, Reinhart and Rogoff’s paper has come under withering criticism in a working paper written by Thomas Herndon, Michael Ash, and Robert Pollin of the Political Economy Research Institute at the University of Massachusetts, Amherst. Herndon, Ash, and Pollin [HAP] assert that the Reinhart and Rogoff [RR] paper suffers from “coding errors, selective exclusion of available data, and unconventional weighting of summary statistics,” which “lead to serious errors that inaccurately represent the relationship between public debt and GDP growth….”
The existence of at least one coding error is a reality that RR admit is “a significant mistake in one of our figures….” Furthermore, this coding error appears to pervade the entire paper, a point that RR have yet to formally acknowledge. Beyond this, the other issues raised by HAP boil down to subjective interpretations of how data should be averaged and the use of data that was not verified until after RR’s paper was published.
Most importantly, even if one uncritically accepts all of HAP’s methods, their primary results are basically similar to RR’s: countries with debt/GDP ratios higher than 90% have notably lower economic growth. HAP’s results are graphed here:
Despite the association between debt and economic growth found by HAP, reporters and commentators have been leading their audiences to believe no such relationship exists. For example, Ben White and Tarini Parti of Politico reported that RR’s paper underwent a “very public demolition” at the hands of HAP, who found that economic growth “in countries with debt over 90 percent of GDP was around 2.2 percent, not much different from lower debt countries.”
In fact, HAP found that advanced countries with national debts over 90% of GDP had 31% less economic growth than when their debts were 60-90% of GDP, 29% less growth than when their debts were 30-60% of GDP, and 48% less growth than when their debts were 0-30% of GDP. As explained further below, these are significant differences with important implications, and Politico is not alone in masking these realities.
The Washington Post‘s editorial board wrote that HAP’s paper “debunks” RR’s “famous 2010 finding that a national debt-to-gross domestic product ratio above 90 percent may substantially retard economic growth.” A headline in the American Prospect has declared that “Reinhart and Rogoff’s Theory of Government Debt is Dead,” and Mike Konczal of the Roosevelt Institute has claimed that “one of the core empirical points providing the intellectual foundation for the global move to austerity in the early 2010s was based on someone accidentally not updating a row formula in Excel.” Countless other individuals and organizations have made similar claims.
These misrepresentations are somewhat understandable given the manner in which HAP present their findings. Their abstract denies any association between debt and economic growth, claiming that “average GDP growth at public debt/GDP ratios over 90 percent is not dramatically different than when debt/GDP ratios are lower.” What do they mean by “dramatically different?” One has to read ten pages into HAP’s paper before they provide a side-by-side comparison of the figures they arrived at for economic growth under different levels of debt: “The actual growth gap between the highest and next highest debt/GDP categories is 1.0 percentage point (i.e., 3.2 percent less 2.2 percent).”
To those unfamiliar with this issue, “1.0 percentage point” may not sound like much, but in this context, it amounts to 31% less economic growth per year. Compounded over time, this can cause genuine harm to people. For example, if economic growth in the U.S. were reduced by 1.0 percentage point per year over the past 20 years, GDP would have been $13.1 trillion in 2012 instead of the $15.7 trillion that it was. This portends far-reaching negative consequences, such as more poverty and reduced life expectancy. As explained in the textbook Microeconomics for Today:
GDP per capita provides a general index of a country’s standard of living. Countries with low GDP per capita and slow growth in GDP per capita are less able to satisfy basic needs for food, shelter, clothing, education, and health.
In a recent New York Times op-ed, Pollin and Ash (two thirds of HAP) write that a “coding error and partial exclusion of data” by RR altered one of their results for economic growth by 0.6 percentage points. They perceptively note that this difference “is quite substantial when we’re talking about national economic growth.” Yet, in their paper, HAP characterize a much larger 1.0 percentage point difference in national economic growth as “not dramatically different.”
Cause and Effect
One of the most important critiques of RR and those who have favorably cited their research concerns the issue of causality. In basic terms, HAP and company argue that slow economic growth causes high debt and not vice-versa. In the words of Mark Gongloff of the Huffington Post, RR “imply strongly that high debt causes slow growth, when there is no evidence for that.”
In truth, there is prominent evidence for this, but HAP and many others have ignored it. In 2012, the Journal of Economic Perspectives published a paper by RR and Reinhart’s husband, Vincent R. Reinhart, the chief U.S. economist at Morgan Stanley. In this paper, these scholars (hereafter referred to as RRR) specifically addressed the issue of cause and effect. Yet, from reading HAP’s paper and many news reports and commentaries about this issue, one would never even know that this paper existed.
RRR took a straightforward approach to the matter of cause and effect by limiting their analysis to “prolonged periods of exceptionally high public debt, defined as episodes where public debt to GDP exceeded 90 percent for at least five years.” They found that these countries averaged 1.2 percentage points or 34% less economic growth than when debt was below 90% of GDP. Note that this figure is very close to the 31% difference found by HAP. RRR explain the significance of this with regard to cause and effect:
Following Reinhart and Rogoff (2010), we select stretches where gross public debt exceeds 90 percent of nominal GDP on a sustained basis. Such public debt overhang episodes are associated with lower growth than during other periods. Even more striking, among the 26 episodes we identify, 20 lasted more than a decade. The long duration belies the view that the correlation is caused mainly by debt buildups during business cycle recessions.
RRR emphasize that the cause-and-effect issue has not been “definitively addressed,” but they assert that “the balance of the existing evidence” from their study and other recent studies “certainly suggests that public debt above a certain threshold leads to a rate of economic growth that is perhaps 1 percentage point slower per year.” This is precisely the figure arrived at by HAP.
This does not mean that cause and effect can’t run in both directions. No one disputes that economic recessions can increase government debt, and constructive debate over this matter will surely be ongoing. Nonetheless, there is a clear association between high debt and slow growth, and substantial evidence that the former can cause the latter.
A Universal Rule?
According to Thomas Herndon in an op-ed for Business Insider, he and his coauthors (HAP) “show that, contrary to R&R, there is no definitive threshold for the public debt/GDP ratio, beyond which countries will invariably suffer a major decline in GDP growth.” Likewise Mike Konczal of the Roosevelt Institute has claimed that “Reinhart-Rogoff was supposed to establish a universal rule that there was a speed limit where debt above 90 percent of GDP became dangerous. Now I think that’s out the door.”
Those statements border on defamatory. RR explicitly state in their original 2010 paper that “there is considerable variation across the countries, with some countries such as Australia and New Zealand experiencing no growth deterioration at very high debt levels.” Furthermore, Table 1 of HAP’s paper details the average economic growth rates under differing debt levels for each of the 20 advanced economies they study, and the values range from -1.8% to 4.6%. On top of this, in a 2011 Bloomberg op-ed, RR wrote:
We aren’t suggesting there is a bright red line at 90 percent; our results don’t imply that 89 percent is a safe debt level, or that 91 percent is necessarily catastrophic. Anyone familiar with doing empirical research understands that vulnerability to crises and anemic growth seldom depends on a single factor such as public debt.
Yet, in their New York Times op-ed, Pollin and Ash claim that RR in their 2012 paper “partly backed away from” the claim that “countries will consistently experience a sharp decline in economic growth once public debt levels exceed 90 percent of G.D.P.” How can RR possibly back away from a claim that they repudiated from the start? Moreover, anyone even vaguely familiar with economics knows that numerous factors affect economies, and thus, no single factor can possibly produce a consistent outcome for all economies. To claim that RR said or implied otherwise is a patent falsehood.
How Robust are the Results?
Other elements of the debate between RR and HAP concern the use of different mathematical methods, the inclusion/exclusion of certain data, and the significance of RR’s coding errors. RR and HAP have been battling over these issues in various venues, and links to their respective commentaries can be found here and here.
The views of other economists about the competing mathematical approaches of the scholars vary greatly. Nobel Prize-winning Princeton professor Paul Krugman has referred to RR’s statistical methods as “very odd” and “dubious.” Conversely, University of California professor James D. Hamilton, who is the author of a prominent graduate textbook about the types of mathematical issues involved here, has written that HAP’s method of handling such data is “less widely chosen” and “in my opinion less to be recommended.” Hamilton also stated that yet another approach is preferable, and it would produce results that fall in between RR’s and HAP’s.
Nonetheless, regardless of which mathematical techniques are used, once the coding errors are corrected and all available data are included, the results up to this point have been basically the same: high debt and slow economic growth go hand in hand. The exact figures that inform the strength of this relationship differ in material ways, and this needs to be sorted out in a comprehensive and methodical manner. The graph below, which compares RR’s and HAP’s results for advanced economies since World War II, is a modest start toward this end. Given that there is no objective “best way” to compute this data, all these results collectively serve to enlighten the issue.
In addition to the above, RR’s 2010 paper also contains average and median results for the period from 1790-2009 and for emerging economies (such as India, Brazil, and Nigeria). HAP’s paper only contains average results for advanced economies from 1946-2009, but Just Facts requested additional results from them, some of which they published in a New York Times op-ed and appendix. These results are incorporated above. Just Facts also requested corrected results from RR for all other results affected by their coding error, but a response is not yet forthcoming.
In their paper and subsequent commentaries, HAP present results for progressively shorter time periods to as little as a decade. In the New York Times, they assert that the “pattern for the most recent decade” is “especially significant” because it is “more informative and useful for assessing present-day policy concerns than data from the post-World War II era or, say, the Industrial Revolution.” For this period (2000-2009), they find “no evidence in these most recent years for any drop-off at all in economic growth when public debt exceeds 90 percent of G.D.P.”
Whether intended or not, restricting the observation period to 2000-2009 has the effect of reducing the sample to a relatively small dataset that happens to cut off near the end of one of the worst global recessions in modern history. This is precisely the type of scenario that would produce anomalous results. In contrast, the extensive dataset compiled by RR (2,000+ observations) has the advantage of being little affected by anomalies. This is necessary for robust results, although it is not an inherent guarantee of such.
In the closing statement of their paper, HAP assert that “RR’s findings have served as an intellectual bulwark in support of austerity politics,” and the “fact that their findings are wrong should therefore lead us to reassess the austerity agenda itself in both Europe and the United States.” Likewise, writing about HAP’s findings, Paul Krugman asserts that the “main reason our economic recovery has been so weak is that, spooked by fear-mongering over debt, we’ve been doing exactly what basic macroeconomics says you shouldn’t do — cutting government spending in the face of a depressed economy.”
Such claims are belied by actual data on government spending from the U.S. Bureau of Economic Analysis (BEA). The Great Recession lasted from December 2007 through June 2009, and from 2007 to 2009, the portion of the U.S. economy consumed by local, state, and federal governments increased by 17%. This higher level of spending was supposed to be a temporary response to the recession that would speed recovery, but in 2012, government was still consuming 9% more of the U.S. economy than in 2007, and the economy was still sputtering.
From a long-term perspective, since 2008, total government spending has consumed more of the nation’s economy than at any time since 1960, which is as far back as this BEA data goes. BEA also tracks a slightly less inclusive measure of government spending (called current expenditures) that dates back to 1929. By this measure, government spending consumed more of the nation’s economy during 2009-2011 than ever recorded in the history of the nation, including the peak of World War II. In 2012, current spending was just a hair below the peak of World War II.
Over the past several years, prominent journalists, commentators, and public policy organizations have misled their audiences into believing that government spending had fallen while it had actually risen. They have done this by making palpably false assertions, redefining government spending to exclude large portions of it, and cherry-picking misrepresentative baselines from which to make calculations.
Many of these same individuals have also blamed a host of ills on reduced government spending. As RR revealed in a recent New York Times op-ed responding to HAP’s criticisms, they have been receiving threatening emails blaming them “for layoffs of public employees, cutbacks in government services and tax increases.” This is despite the reality that the unemployment rate for government workers is 3.3%, as compared to 7.5% for the entire population.
Likewise, an October 2012 poll commissioned by Just Facts found that 23% of all voters (including 43% of Obama voters and 6% of Romney voters) did not know that government spending was consuming a larger portion of the economy than it was ten years ago, notwithstanding the fact that it was consuming 12% more.
On top of all this, politicians and reporters are now advancing the narrative that the national debt is going to be stable for the next ten years, when in fact, this claim is based upon an unrealistic scenario that requires major departures from current policy. All of this serves to mislead people into believing that the national debt does not pose a significant problem, even though the Congressional Budget Office recently reported that the U.S. government is on a path of “high and rising debt” that will have “serious negative consequences.”
Condensing the key points above, there is a clear relationship between high levels of debt and slow economic growth. This is a not a universal rule but a robust association based upon extensive observations and disparate mathematical methods. The precise strength of this relationship is debatable, but existing results center around the outcome that growth in countries with debt over 90% of GDP is about 30% lower than when debt is below this level. There is also considerable but not definitive evidence that high debt can cause slow growth, as well as vice versa.
The current U.S. debt is at 105% of GDP and is projected to keep growing under current polices. This elevated level of debt may be a factor in weak economic growth that the U.S. has been experiencing, but advocates for increased government spending are blaming this and other problems on reduced government spending. This is in spite of the fact that since 2008, government spending has been much higher than it has been for the vast majority of U.S. history.