The academic two-step


Carmen Reinhart has written yet another defense of the discredited study that she and fellow Harvard economist Kenneth Rogoff wrote. This one takes the form of an open letter to Paul Krugman, one of her harshest critics in academia. She says that it was not their fault if policymakers misread their statements about the impact of debt reaching 90% of GDP and arrived at an alarming conclusion that resulted in them pursuing the debt-reduction austerity programs that have caused such hardship around the world.

What Reinhart and Rogoff are doing is the ‘academic two-step’, writing things in such a way that it seems to imply one thing while at the same time inserting caveats in case things turn out differently. Then as long as things are going your way, you don’t mention the caveats, but if you start to take a beating, shift to the other foot and haul them out as defenses.

Krugman counters with a short piece that essentially says that she is doing the academic two-step.

There is, as everyone in this debate has acknowledged, a negative correlation in the data between debt and growth. As a result, draw a line at any point — 80 percent, 90 percent, whatever — and countries with debt above that level will tend to have slower growth than countries with debt below that level.

There is, however, an enormous difference between the statement “countries with debt over 90 percent of GDP tend to have slower growth than countries with debt below 90 percent of GDP” and the statement “growth drops off sharply when debt exceeds 90 percent of GDP”. The former statement is true; the latter isn’t. Yet R&R have repeatedly blurred that distinction, and have continued to do so in recent writings.

Another economist Brad DeLong looks closely at the correlation between debt and growth rates and shows how there is no cliff at the 90% mark but how Reinhart and Rogoff encouraged policymakers to think there was.

The academic two-step is unfortunately not an uncommon practice. It was on clear display in The Bell Curve and more recently in the Jason Richwine affair when the authors encouraged readers to draw all manner of sweeping conclusions about the relationships between IQ and race and success in life, while burying the caveats as insurance policies in case they were taken to task, as indeed happened.

Comments

  1. Reginald Selkirk says

    His Royal Highness Pope Emeritus Benedict Palpatine Ratzinger was fond of a variant of this technique. He was in the habit of expressing his ideas as quotes from others. That way, when one backfires (like suggesting that a competing religion has spread only by the sword, or claiming that Galileo deserved his treatment at the hands of the church; references available upon request), then you just say, ‘Hey, I was just quoting this guy. It’s not my idea.’

  2. Corvus illustris says

    She says that it was not their fault if policymakers misread their statements about the impact of debt reaching 90% of GDP and arrived at an alarming conclusion that resulted in them pursuing the debt-reduction austerity programs that have caused such hardship around the world.

    If I recall their previous defense correctly, it was that their piece was a “working paper” and did not represent their final word on its subject. (This defense also excused the use of Excel instead of a standard statistical package, the weird weighting factors, etc.) This then permitted the passive-aggressive “it wasn’t our fault that our tentative results were taken as definitive,” etc., parallel to their now-current defense, the one you cite.

    But it is your fault–at least in the math biz–when these things happen. So when you find someone using your “preprint” results as definitive you caution them strongly and loudly; you set your own grad students to checking your seemingly airtight results (as in the case of Fermat’s Last Theorem); you make computational details available (as in the case of the Four-Color Theorem). If a result looks important, this is certainly what you do. If R&R had done the same, their reputations would be intact. It will be interesting to see whether their colleagues just circle the wagons or whether this level of academic malpractice will have consequences.

  3. sunny says

    It will be interesting to see whether their colleagues just circle the wagons or whether this level of academic malpractice will have consequences.
    —–

    Nothing will happen. In fact, they will be rewarded with plum postings in the next administration. At some point one realizes academia – at least in the social sciences – is a game without any notion of accountability. Does any one remember Paul Collier’s misguided policy advice – based on running mindless cross-country regressions – to support a military coup in Cote D’Ivoire? Would Collier have been held accountable had the policy been implemented? No, it is just a game but such work has the aura of rigour because of quantification. As Feynmann aptly put it: it is nothing but a cargo cult.

  4. Uri says

    For those who haven’t read it, or who have read it but not recently, Noam Chomsky’s “The Responsibility of Intellectuals” is an excellent essay that poses a simple moral axiom (“it is the responsibility of intellectuals to speak the truth and expose lies”) and shows how dramatically many of our public intellectuals fail to abide by it. The two-step just seems like a variant on this unfortunate trend.

  5. Jeffrey Johnson says

    According to a recent study by Wang and Kimball, R & R’s work is even worse than it has looked. Not only is the drop in GDP at 90% debt wrong, but the idea even that there is a causal relationship between debt and GDP is wrong. While R & R found correlation between hight debt and low growth, which they have tried to fall back on after their 90% threshhold was demolished, this latest work, if correct, demonstrates that the causality does not run from debt to low growth, but rather the other way around: slow growth leads to a high debt to GDP ratio.

    http://www.slate.com/blogs/moneybox/2013/05/30/reinhardt_rogoff_debunked_yet_again.html

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>