28-year old PhD student debunks the most influential austerity study
New study refutes Reinhart and Rogoff analysis that underpins austerity policy around the world; shows no relation between debt and lack of growth
PAUL JAY, SENIOR EDITOR, TRNN: Welcome to The Real News Network. I’m Paul Jay in Baltimore.
In capitals, both political and economic, across Europe, across North America, really, across the world, there’s been an assumption based on a study done by two eminent Harvard professors, economists Carmen Reinhart and Kenneth Rogoff, which presented in 2010 their conclusions that 90 percent debt-to-GDP ratio means a collapse in growth. It’s that conclusion that leads to policies like austerity, which says even in times of recession, debt’s more dangerous than high unemployment.
Now joining us to talk about their conclusions, because they’ve reached quite different conclusions about the same data, is, from the PERI institute, first of all, Thomas Herndon. He’s a doctoral student in economics at the University of Massachusetts Amherst. His research includes political economy and finance. And he coauthored a paper: Does High Public Debt Consistently Stifle Economic Growth?, which is a critique of Reinhart and Rogoff.
And with him is Michael Ash. Michael is a professor of economics and public policy at the University of Massachusetts Amherst. His research focuses on inequality and well-being in the United States. And he teaches graduate economics at UMass Amherst.
Thank you both for joining us.
THOMAS HERNDON, DOCTORAL CANDIDATE, ECONOMICS, UMASS AMHERST: Thanks, Paul.
MICHAEL ASH, PROF. ECONOMICS AND PUBLIC POLICY, UMASS AMHERST: Thanks for having us, Paul.
JAY: So, Michael, give us a little bit of the background. First of all, what is the importance of this study? How much has it affected public policy around the world? And then what did your study, your counterstudy find?
ASH: The Reinhart and Rogoff study, which came out in 2010, argued that when public debt reaches a level equal to about 90 percent of GDP, there’s a very substantial effect on growth, that economic growth measured by GDP essentially stops.
This finding, or the Reinhart and Rogoff result, had a profound effect on public policy. One of the authors testified before the Senate Budget Committee. The finding is one of fairly few academic studies cited in Paul Ryan’s “Path to Prosperity”, the House Republican budget resolution for fiscal year 2013, cited as giving evidence that debt is–that public debt is at the moment an overwhelming problem for the United States. So I think it’s also been cited by Democratic Senator Kent Conrad of South Dakota, who I think felt chastened by this study, that, yes, debt is a very serious problem. So it’s had a profound effect on public policy debate both in the United States, and even perhaps more so in Europe.
JAY: Michael just [snip] why this affected public policy. So is the theory, based on this study, that when you get such high debt-to-GDP levels, people don’t want to invest, they lack confidence, I mean, all these kinds of arguments? Is that what comes out of this?
ASH: Part of the influence of the paper–I think one of the reasons that it’s had such wide reach is that it was a very straightforward empirical exercise. So Reinhart and Rogoff proposed some mechanisms. They think that there may be a pathway through inflation and through interest rates that take hold at this 90 percent threshold or 90 percent cliff and that this has disastrous consequences for GDP growth. I hope that later in the interview we can discuss some of the reasons.
We think that they have the order backwards, because we pretty much believe that low GDP growth has a very straightforward effect on causing high debt. There is low GDP growth. Tax collection falls, and there has to be larger public outlay on things like unemployment compensation. So there’s a pretty straightforward path the other way.
But that’s not what they were arguing in their paper. They pretty straightforwardly–they straightforwardly argue that high levels of debt, of public debt, operate through this mechanism of interest rates and inflation and cause GDP growth to slow to a halt.
JAY: Okay. So what brought you to question their conclusions? This has been sort of the golden rule in a lot of economics.
ASH: Our study emerged from a paper [snip] class, a term paper for a class that was written by my coauthor, Thomas Herndon. And in this class we asked students to replicate an important result, important published result from the economics literature. Thomas selected the Reinhart and Rogoff paper, both because it had interesting empirical methods and because it had had a profound effect on debate in the discipline and on public policy debate. So he began the replication in fall 2012 for a term paper for the course, and that has since evolved into the study that we released on Monday.
JAY: Alright. So, Thomas, what did you find when you went to–when you looked at their conclusions and their data?
HERNDON: When I first started replicating the paper, I had to construct the data set out of the publicly available data, and I was unable to replicate their basic summary statistics and number of observations for the over-90 category. It took me a while. At first I thought that it was likely that I had just made the mistake, but as I started ruling out all the possible errors I could have made, I started more and more to start to believe that it was likely that they had made the mistake by constructing the negative average.
JAY: Just to be clear what you’re looking at, you’re looking at countries that actually did reach this 90 percent ratio, and then you looked at what happened to growth, and you were comparing these things. Is that right?
HERNDON: That’s correct. They examine a sample of 20 advanced countries from 1946 to 2009, and they look at all the cases of growth or when debt-to-GDP ratios were above 90 percent. They find a negative average. And this negative average was, I think, the strongest result that received the most media attention and attention in policy-making studies or circles. And that’s why I chose to replicate it.
JAY: So what did you find when you redid the data?
HERNDON: I was unable to replicate the negative average. I tried a number of ways, and I was unable to do it based on the publicly available data. Once they were kind enough to provide me with their spreadsheet, I was able to identify the error rather quickly, because I had already done a lot of the work of really getting to know the data well.
JAY: And what was the error? And then once you fix the error, then what do you find?
HERNDON: There was actually a few errors. The one that’s received a lot of the media attention has been the spreadsheet error where they average 15 out of the 20 countries instead of 20 out of the 20 countries. Additionally, we found that there was kind of an unconventional way to make the average. Instead of looking at all 96 observations for their over-90 category and then taking the average with those 96 observations, they first averaged by country and then averaged the country averages. This has the effect, though, of making countries with very little experience over 90 weighted higher than countries who have spent, for example, two decades over 90.
ASH: If I could make that concrete, to give you an–so, much of their result turns out to hinge on New Zealand’s experience in 1951. New Zealand had a serious recession in 1951 at a time when it had a high public debt ratio. And so that value for New Zealand in 1951 (happens to be -7.6 percent) ends up having as much importance in their paper as Great Britain’s 19 years in the high public debt category when its average growth was 2.6 percent per year. So we’re holding up -7.6 for New Zealand for one year against 20 years for Britain at 2.6 percent growth, and those have equal weight in their finding.
To give you a baseball analogy, someone who’s been up five times and is hitting .200 is going to get equal weight in deciding whether that person’s a good hitter or not [incompr.] look at them at the end of the season and [incompr.] hit, you know, .315 with, you know, several hundred at-bats.
JAY: Right. So you can’t take too narrow a sampling to come to a real conclusion. So then what do you conclude from all of this? Does this–does debt-to-GDP ratio have any effect on growth? I mean, I saw in one of the reports of your study you actually found there was growth in many countries with over 90 percent. So, in other words, are there just other factors involved here that there’s–it’s been too narrow interpretation of the effect of this debt-to-GDP ratio?
ASH: What we find is that average growth is modestly diminished when countries hit the 90 percent–as countries approach the 90 percent public debt to GDP ratio. There’s no cliff. I encourage your listeners, your viewers to take a look at the paper. We show all of the data, so you can see that there is no cliff.
But more importantly, we observe that there’s an enormous range of GDP growth experiences at every level of public debt to GDP. So there are countries that grow quickly and countries that grow slowly at low public debt to GDP levels and countries that grow slowly and countries that grow quickly at high public debt to GDP levels. So I think that it’s the–really that there is no threshold at 90, there is no precipitous drop in growth, I think, is the most important message of our replication.
JAY: So this narrow-minded focus on this one indicator is what’s wrong, because there’s many factors at play here.
ASH: Yes, there are many factors at play in determining whether or not countries grow quickly. And the public debt to GDP ratio does not have an obvious strong effect on GDP growth, again, at any level.
JAY: So I saw Reinhart and Rogoff have refuted your study, saying that their weighting was correct, they still stand by their conclusions, the data still supports what they say. Have you seen any more from them? How do they deal with your conclusions? ‘Cause it seems like you’ve found a more or less straightforward error.
HERNDON: Yes. Reinhart and Rogoff have said that our findings are somewhat consistent with theirs, and they’ve argued that the way they calculated the average was not unconventional. There are many ways to calculate the averages, and there might be good reasons to take country averages first. We mention serial correlation in the paper as one of them.
However, the problem with this average is that it allows one observation, New Zealand in 1951, to move the central tendency of the sample very far. It drops it quite a bit. And so it’s overweighting it.
You know, there’s always many ways you can measure things in statistics, but I think the problem with the way this is goes back to the baseball analogy. If someone has 500 at-bats and is batting a 2.0, and then one person has one at-bat and gets a perfect hit, the average between them will be about a batting average of 600. And that’s not at all very–a representative characterization of the data.
JAY: Now, I thought on these sorts of studies you’re supposed to kind of get rid of the outliers on both sides and try–you know, if you have one that’s an extreme argument on one side, one the other, get rid of both of them and deal what’s in the middle. Did they not do that?
ASH: So, Paul, thank you for raising that issue, because another very important problem that we found with the Reinhart and Rogoff paper was the selective exclusion of years, particularly at the early years of the sample. So, remember, Reinhart and Rogoff are telling us about the advanced economies between 1946 and 2009. So those early years of the sample happen right in the wake of World War II.
The U.S., for example, had a massive demobilization in 1946 and very large negative GDP growth, completely explicable by that demobilization. I don’t think that has any lessons for us for the current economic crisis that we’re facing. But that is nonetheless in the Reinhart and Rogoff study.
Excluded from the Reinhart and Rogoff study, however, is data for Australia, Canada, and New Zealand for those early post-World War II years, when in fact those countries had reasonably high growth despite or with no relation to the very large public debt loads that they were carrying. So those data are available. They’re in the spreadsheet that Reinhart and Rogoff supplied to us, but they weren’t included in the study, while the early years to the U.S. were.
So one of the other very important problems that we see with the Reinhart and Rogoff study is this selective exclusion of data from the early years for some countries but not for others. There’s no explanation or discussion in the paper of that decision, and I think it at least requires some additional discussion.
JAY: So, Thomas, what are the public policy conclusions from this? ‘Cause it seems to me if you’re right, you’ve just unraveled the whole economic argument for austerity.
HERNDON: I think a lot of the public policy conclusions is that we need to think about the uses of debt in a more specific way. Rather than just saying in general that high debt is bad, I think we need to look at the historical moment, we need to see that there’s a lot of unemployment. And when you’re in a recession, the use of stimulus policies can have a very large effect at getting the economy out of the recession. So I think looking at this specifically, asking what positive effects stimulus can have under the historical moment is a much better way to do it than just a, you know, debt-to-GDP is always bad and there’s a nonlinear threshold [incompr.]
ASH: So Reinhart and Rogoff have proposed a mechanism by which public debt is bad for GDP growth. And they argue that operates through raising interest rates. If you take a look at the current economic situation, interest rates are at historic lows. So the mechanism through which public debt is supposed to have a negative effect on GDP growth doesn’t seem to be operating at all.
On the contrary, the burden of servicing the U.S. debt is at all-time lows even though the U.S. debt is at relatively high levels compared to history. So the Reinhart and Rogoff mechanism by which public debt is supposed to have this negative effect doesn’t seem to be in play at the moment.
And so the combination of the collapse of the empirical result that high public debt is inevitably associated with greatly reduced GDP growth and the weakness of the theoretical mechanism under current conditions, I think, render the Reinhart and Rogoff point close to irrelevent for current public policy debate.
JAY: So we’re going to invite both of the economists, Reinhart and Rogoff, to come on to The Real News and debate this with you or perhaps your coauthor, Bob Pollin, and I hope they do.
Thank you, gentlemen, for joining us.
ASH: Thank you, Paul.
HERNDON: Thank you, Paul.
JAY: And thank you for joining us on The Real News Network.
And don’t forget we’re in our spring fundraising campaign. We have a matching grant of $50,000. So if you donate $1, we get another $1. If you donate $10, we get another $10. And you don’t need to be an economist to figure out that if you donate more, we’ll get more. And if you would like to see more economic coverage from The Real News Network like this story and many others that we do, we need your support.
Thanks for joining us on The Real News Network.
DISCLAIMER: Please note that transcripts for The Real News Network are typed from a recording of the program. TRNN cannot guarantee their complete accuracy.