-- Posted by Neil H. Buchanan
Anyone who has been paying attention to the debates in the U.S. and Europe about austerity and debt has heard (at least indirectly) about a 2010 paper by two Harvard economists, Carmen Reinhart and Kenneth Rogoff, that purports to show that countries with government debt levels exceeding 90% of national income experience severe economic damage, in the form of much slower growth (and sometimes shrinkage). The paper drew a great deal of attention, mostly because it confirmed what the political class had already decided was the correct path: inflicting pain on citizens during the deepest economic downturn since the Great Depression -- actually, in some European countries that have taken the austerity medicine (including the U.K.), the current situation is even worse than the Great Depression -- in the name of growing more quickly in the future.
I had been aware of the R-R paper ever since it made news, but I do not recall writing about it, because it struck me as an extraordinarily weak paper (for reasons that I will explain in a moment). There has been a bit of news this week, however, that has thrown a harsh new light on the paper. Three economists at UMass-Amherst (Thomas Herndon, Michael Ash, and Robert Pollin) have written a paper that claims that Reinhart and Rogoff's statistical analysis was flawed, with the results biased by the exclusion of certain data points and/or coding errors in the computational algorithm. Correcting those errors, Herndon et al. say, severely undercuts the Reinhart-Rogoff result.
So, it turns out that the go-to academic article that every deficit scold considers the end of the argument on fiscal policy is potentially invalid even on its own terms. What is more important, however, is the paper's problems that were obvious all along. Several of my recent posts have discussed the problem of economists "committing politics" (e.g., here and here). My argument has been that economists are not trained to understand politics, and their professional socialization makes it an actual point of pride not to know anything about politics, which results in their making either avoidably silly statements or thinking that they can get away with the most crude political misanalysis, because "politics is easy."
There is a bit of that going on in the Reinhart-Rogoff affair. Yesterday's New York Times ran a news article that summarized the debate at this stage, which included the following gem from reliably Republican economist Douglas Holtz-Eakin: "There’s nothing about this that will change my view of the universe. The sun still rises in the east. It sets in the west. And a lot of debt is still bad." In other words, he's a hack. But Reinhart and Rogoff are not hacks, or they have not obviously been so up to this point.
Even so, this is another good example of how shoddy economics can be, when it is performed in the service of a right-wing political agenda (especially a right-wing political agenda that a Democratic president and much of his party endorses). Following are two of the most important things that we knew were wrong with the R-R analysis all along (that is, even before the current discussion of potential data manipulation):
-- The paper (which was not peer-reviewed) claims that something important happens when the debt-to-GDP ratio hits 90%. The action in the plus-90% range, however, is largely driven by two countries, Italy and Japan, which currently have debt levels of 126% and 219%, respectively. And, as some economists have noted (especially Paul Krugman, who has been on top of this since the R-R paper originally made news), those two countries perfectly illustrate why the paper's conclusion is so weak. R-R claim that debt causes slow growth, whereas both Italy and Japan are clear examples of slow growth causing debt to increase. This is a classic correlation-is-not-causation problem, in other words. Reinhart was recently quoted acknowledging the direction of causation problem, but defending her conclusion by challenging anyone to find a high growth country that has high debt. This, of course, is not the point the she and Rogoff were making at all.
-- Another example of the odd analytical strategy in the R-R paper is that they include the U.S. in the post-WWII period as a "high debt causes low growth" case. Everyone knows that the debt in the U.S. was very high coming out of the war, and that the economy experienced a brief recession after the wartime spending ended (and as the returning veterans were slowly reintegrated into the new economy). Saying that the debt had anything to do with the recession is not reverse causation, it is spurious correlation.
You get the idea. Apparently, R & R have issued a preliminary defense against the Herndon et al. paper, claiming that even their critics find that growth slows at higher debt levels. If that is the best they can do, however, this is pretty weak tea. Their paper's fame was based on that 90% breaking point -- a number conveniently set at about the level that the U.S. debt would have reached, if we had followed Keynesian policies to fight the Great Recession. If, instead, there is simply a slow tailing off of growth as debt rises, then that is a very different story -- and involves very different policy considerations -- than the one that they have been telling. And again, even that continues to confuse cause with effect.
A couple of years ago, the big pro-austerity argument was based on a paper co-authored by another Harvard economist, Alberto Alesina (which I discussed at various points, e.g. here), that purported to show that countries that engage in fiscal austerity are rewarded with higher rates of growth. That argument, too, ended up being based on extremely shoddy statistical analysis. What Alesina and R & R had going for them is that they gave intellectual backing to what the political classes in both the U.S. and Europe wanted to hear. I find it impossible to imagine that the authors of those papers were not aware of the weakness of their arguments. Even if they were acting in good faith, however, these episodes show that economists are sorta-kinda right: Politics is easy, if you tell powerful people what they already believe (especially if you dress it up with a bit of math).