More Reasons to Call Off the Reinhart-Rogoff Witch Hunt

On Monday, I posted an article that one pundit introduced on Twitter with the words: “Ladies and gentlemen, we have a Reinhart-Rogoff defender.”

Needless to say, comment threads on the sites that published my article have been interesting.

Here are some thoughts on the second week of punditry about Carmen Reinhart and Kenneth Rogoff’s sovereign debt research, including those comment threads.  I’ll offer more reasons to see through the smear campaign and suggest a perspective on thresholds and austerity that seems absent from the public debate.

As in Monday’s article, I’ll use “RR” for Reinhart and Rogoff and “HAP” for Thomas Herndon, Michael Ash and Robert Pollin, authors of the now famous critique of the first of two papers by RR on the link between economic growth and debt.

RR’s influence, correlation versus causation, and thresholds

The debate over RR’s influence on policymakers continues to rage. The Washington Post weighed in last Sunday with an editorial titled, “Rogoff-Reinhart error is not the source of global austerity,” and went on to say that it was “preposterous” to think otherwise.

This was interesting because HAP and many other pundits had included the Washington Post’s editorial board among those who were allegedly over-influenced by RR. And was it?According to the editorial, this has been “rather overstated in some quarters.”

But this part of the dispute seems to have gone on long enough. It almost seems more constructive to pick your battles and say fine, let’s assume there was an RR present at every austerity discussion, swaying the outcome. And then ask what’s wrong with that.

I think I’ve heard three answers.

The first is that RR’s conclusions are invalid because their research was incorrect. But their conclusions are essentially unchanged, as I explained in Monday’s article, and the substantive differences between RR and HAP weren’t based on mistakes.

The second is that they never proved causation. This is true, but it doesn’t disqualify them from offering advice on how much of the correlation is growth-to-debt and how much is debt-to-growth.

To be clear, I’m not faulting anyone for stressing the difficulties in establishing causation. But if you then go on to say RR have no business sharing their opinions, then by the same logic the pro-stimulus academics should have no advisory role, either. Policies proposed by Krugman, Delong, Summers and others aren’t backed by any more evidence than RR’s recommendations. And some of their purported evidence is devoid of any sense of reality (see here).

The third is that RR put too much emphasis on the 90% threshold. As far as this goes, I’ll suggest asking yourself a question: Do you strive to keep your cholesterol below 200? If so, here’s the advice you may be heeding (from the American Heart Association’s website):

cholesterol levels

The AHA’s message is obviously based on extensive research. And just as obviously, research doesn’t show a “cliff” at the 200 level. Nor does it imply that risks are the same for every individual. In my case, I take cholesterol meds because of family history, not because I breached the “thresholds.”

Some people may assign more significance to the 200 level than warranted by the body of research, but does that reflect badly on the researchers?

Of course not.

People need markers to make sense of complicated risks. Researchers who help to provide those markers should be applauded, not condemned for the fact that the markers are never as precise as you would like them to be.

Pet peeves

This brings me to a couple of pet peeves about the RR witch hunt that I didn’t mention in my article on Monday.

First, I can understand laypeople misinterpreting the 90% threshold, but trained economists expressing shock and horror at the absence of a “cliff”?

These folks are either lying or incompetent. Either way, they’re frauds. If you have any experience working with economic data, the scatter plots in Figures 3 and 4 of HAP’s paper showed you exactly the type of relationship you would expect from the summary results in RR’s paper.

Trained economists should also know that when medians and arithmetic averages are far apart, there’s a high probability of outliers in the data. (I assume this is why RR led with their conclusions about the medians, not the averages that HAP critiqued.) It’s helpful to know more about the outliers, as I noted in my earlier article with regard to the 1951 New Zealand result, but they shouldn’t have been surprising, either.

Second, the question of how much debt is too much is an extremely important question. And RR’s many critics make no attempt whatsoever to answer it (if I’m wrong about this, please send me the research), even as they bash two researchers who’ve been leading the way to possible answers since well before their 2010 paper.

It seems to me that you can say two things about the many pundits on this issue:

  1. Commentators in the U.S. who’ve both looked under the surface of our fiscal challenges (see the problems with official numbers in articles such as this and this) and seriously considered the question “How much is too much?” have come away alarmed.
  2. Of the commentators on the other side – those now crowing over the misinformation that’s spread all the way from Rortybomb to the Colbert Report – I don’t know of any who’ve attempted to do the same due diligence and answer the all-important question.

If you’re not even trying to establish where the threshold is, you’re guaranteed to cruise right past it thinking “debt hasn’t harmed us yet, so why should it harm us now?” And when I say “cruise right past it,” I mean we won’t even realize what’s happened until years later. This is because the point at which a fiscal crisis becomes inevitable is likely to be much earlier for a country like the U.S. than the point at which the crisis actually occurs. It’s like the onset of “irreconcilable differences” in a bad marriage. The marriage may limp along for a long time before the legal battles begin, but the real mistakes were made much earlier.

I’ve called this threshold (for debt, not marriages) the “point-of-no-return” and described it as follows:

It’s based on the mostly mathematical but partly political impediments to lowering debt once it’s risen beyond a certain amount … You can think of it as a Catch-22. Once you reach the point-of-no-return, then the only way to get back to safe debt levels is through austerity. But austerity can have the opposite effect because it throws the economy into reverse and fiscal measures that are normally deficit-reducing can turn out to be deficit-increasing. Not to mention the fact that austerity kills political careers. The key is to stay clear of the point-of-no-return in the first place, because once you’ve reached it you’re toast … I propose that there’s no such thing as immunity from this Catch-22. My reasoning is that the mathematics and politics of austerity aren’t that much different from one country to the next. Therefore, America’s point-of-no-return isn’t that much different than it is for other large, developed nations.

And I offered an initial estimate here.

The Coburn book

Bloggers have been all over a passage from The Debt Bomb by Senator Tom Coburn that purportedly shows RR behaving inappropriately in their discussions with Congress. Here’s the passage:

Johnny Isakson, a Republican from Georgia and always a gentleman, stood up to ask his question: “Do we need to act this year? Is it better to act quickly?”

“Absolutely,” Rogoff said. “Not acting moves the risk closer,” he explained, because every year of not acting adds another year of debt accumulation. “You have very few levers at this point,” he warned us.

Reinhart echoed Conrad’s point and explained that countries rarely pass the 90 percent debt-to-GDP tipping point precisely because it is dangerous to let that much debt accumulate. She said, “If it is not risky to hit the 90 percent threshold, we would expect a higher incidence.”

One reader seemed to capture the anti-RR hysteria best by suggesting they should have told Senator Isakson this:

Certainly, you can read it that way, Senator–but to be fair, the 90% is not an actual threshold that we should see as a hard and fast rule. More importantly, it is absolutely possible that causation goes exactly the other way–economies that are doing poorly accumulate debt. The other side of the debate could be right about that. Our collective gut feeling is that debt is bad, and we need to keep it under 90%. That is our personal opinion.

I don’t get it – do people really believe this is how advisors should talk? I’ve offered a heck of a lot of research-based advice in my career, and I can guarantee that my colleagues would have locked me in my office if I’d spoken like that. It sounds fine in theory, and maybe you can do it if you’re sitting in a room full of Phds., but most of your audiences don’t have the patience for this kind of wiffle-waffle. They really just want to know what you think.

Apart from the facts that the passage in the book doesn’t relate to the means and medians of the 2010 paper, and summarizes what was obviously a longer conversation, it merely shows that RR were asked what they thought and then answered the question.

Big whoop.


Some anti-austerity pundits like to point out that austerity has weakened Europe’s peripheral economies and conclude that this means austerity’s bad. In my opinion, this perspective isn’t helpful at all. It’s just dumb, really. It’s like saying that the months you spent in rehab were miserable, and therefore, rehab is bad and heroin is good. We already know that fiscal restraint slows growth and drives up unemployment in the short-term, and sometimes longer. Of course it does.

The crux of the discussion should be whether those actions prevent an even more severe and painful crisis in the future.  Or not.

Although in the Eurozone, it’s more complicated because the alternatives aren’t so much in the future – the peripheral nations wouldn’t have been able to sell their bonds without help. Without German support, they would have been forced to completely balance their budgets and in the short-term there would have been far more austerity, not less.

Setting aside the Eurozone’s unique politics, another way to look at recent events is this: RR’s association between high debt and low growth is exactly what we’ve been witnessing.

The only reason for austerity is because debt was too high to begin with. At some point, something needs to be done about high debt, and that something leads to lower growth. (High Debt–>Low Growth.)

And if you wait too long before acting, such that deficits are way too high to be managed down quickly, then low growth feeds back into higher debt. (Low Growth–>High Debt.)

We’ve clearly seen both directions of the much debated causation at work.

Finally, causation from lower growth to higher debt reinforces the importance of not exceeding the type of threshold I discussed above – the point-of-no-return. Once you’ve exceeded this threshold, then neither austerity nor “not austerity” will restore sustainable debt levels, which is exactly what we’ve seen in Greece.

So, what have we learned from Europe?

We’ve learned the importance of trying to figure out the types of thresholds that RR’s critics are endeavoring to do away with, as part of the smear campaign that also seeks to do away with RR.  In other words, these critics are guiding their followers in the worst possible direction for all of us.

The advisors who should be quieted are those who haven’t credibly attempted to answer the obvious question:  How much debt can we bear before we get locked into a long-term path that leads to even more hardship than people are suffering today.  And we should pay attention to those who’ve taken on the unpopular task of trying to figure out the answer.

The balance of power in Europe

It’s occurred to me that many people seem to misunderstand the politics behind austerity decisions in the Eurozone. The reality is that these decisions rest largely with the German public, not a bunch of pundits in America and not even Angela Merkel, who’ll be tossed out if she gives away too much.

Another reason this is a witch hunt

Getting back to the comment threads that followed my article and others, it’s clear that the public’s understanding of economic research has taken a giant step backward in the past few weeks. And that’s thanks to a handful of intellectually dishonest pundits who are whipping up a whole stew of misconceptions.

Believe it or not, economists discuss the reasons for causation in papers that don’t “prove” causation all the time. And they do it using the same type of language that RR uses, where they highlight an “association” or “relationship,” while offering an interpretation of its causes. The second part necessarily introduces directionality.

And not only was RR’s paper devoid of the supposedly grave sins that people claim, but they actually avoided common pitfalls of empirical research. One of the appeals of their work is that it’s not full of statistical inferences. In economics, these inferences are always inherently flawed. Whenever an economist makes a purely statistical inference (using the parametric statistics that dominate research), she’s assuming that some characteristic of the economy doesn’t change over time. In the real world, you won’t find such a thing. And yet, people are judging right and wrong as if economists have figured out how to glean the truth from historical data.

In my opinion, one of the clearest “truths” of empirical research is that overreliance on flawed statistical methods is one of the reasons for the profession’s lousy track record (as Nassim Nicholas Taleb has explained in great detail).  And RR’s approach is unusual only in the fact that they’ve done what they can to strip away these pretenses.


Here’s a post by Greg Mankiw (h/t Marginal Revolution) that I recommend for anyone who sees RR’s Excel error as a game-breaker, regardless of it not having a meaningful effect.

And don’t miss the posts from James Hamilton on Econbrowser, such as this one, or RR’s N.Y. Times Op-ed on Friday.

Thank yous

Thanks to the reader who let me know that we’ve been blogging for two and a half months without noticing that our site header said cynicoMomics. And also to the reader who noticed the stray axis label on my second chart in Monday’s article. It’s a shame I’m not world famous – Rortybomb could have gotten a post out of these errors.

And thanks to the sites that published the article but changed the title. I suck at titles and I’ll change the first one soon. Any volunteers to do our titles for bloggers’ wages let us know.

Bookmark and Share
This entry was posted in Uncategorized. Bookmark the permalink.

One Response to More Reasons to Call Off the Reinhart-Rogoff Witch Hunt

  1. Unanimous says:

    I’m sure that many people don’t offer a limit of government debt to GDP because they think other things are limiting factors rather than government debt to GDP, and debt itself isn’t a limit. It’s like asking how light a chair can be before it breaks. There might well be a statistical correlation between chair lightness and how often chairs break, but when you are not sure about a particular chair and trying to decide if it is safe to use, it is better to focus on the more direct factors, such as what load it can hold rather than its weight.

    Imagine someone has done a study on the weight of 200 broken and unbroken chairs and finds that half of the chairs under 5 kg broke and this figure became well publicised. Also imagine that the raw chair data contained unbroken chairs as light as 2kg, broken chairs as heavy as 20 kg, and that chairs in the 4-5 kg range didn’t break any more than chairs in the 5-6 kg range but the categories used in the statistical analysis didn’t pick this up because they spanned 5 kg ranges. Next, a bunch of chair pundits and politicians go around saying that all chairs that weigh less than 5 kg should be banned. Some people try to point out that the 5kg limit isn’t justified, and that really weight itself is a bad thing to use as a limit. Are those people on a witch hunt because they won’t offer an alternative weight limit?

    Factors that spring to mind and that make an enormous difference to the debt carrying capacity of a government include the currency that the debt is denominated in, how much of it is owed overseas, other government assets, demographics, and the strength of the legal system.

    In theory, a government could have any amount of debt provided that it had sufficient counter balancing assets generating an adequate income, and that the variations in income are matched to the variations in the debt repayments. In practice, it will be difficult to have things matched up well at very high debt levels, but it wouldn’t surprise me to see Japan for example manage 300%, and still have real GDP per working age person continuing on it’s steady rise. If Japan had an active immigration policy, it could probably handle more. Third world nations with externally denominated debt, widespread corruption, and a poorly performing tax office could run into trouble with 20% debt. Given a factor of ten or more variation in reasonable outcomes, why would I want to put forward or accept a general debt limit?

    RR aren’t evil, and most opponents aren’t on a witch hunt. They did a feeble analysis (not unusual in economic circles as you point out), and parts of that were taken out of context by people pushing a view. The end result was a deception.

Comments are closed.