Reinhart And Rogoff Make More Mistakes While Admitting To Research Flaws
Sometimes mistakes are like potato chips, or sex partners: It's hard to stop at just one.
Carmen Reinhart and Kenneth Rogoff of Harvard have spent the past week defending themselves for mistakes they made in their influential research on government debt and economic growth. In the process, they have apparently made more mistakes.
Last week a graduate student named Thomas Herndon, along with a couple of his professors, published a paper pointing out big errors in Reinhart and Rogoff's 2010 paper, "Growth In A Time Of Debt," which has been cited repeatedly by policy makers pushing austerity in the U.S. and Europe.
In a detailed response to Herndon, Reinhart and Rogoff copped to some mistakes, including an error in creating a formula in Excel. But they stood by the general gist of their paper, that high debt is associated with slower economic growth. They suggested that Herndon's paper backed up that view, in fact.
And yet the mistakes just keep coming: In their response, Reinhart and Rogoff said that small differences in economic growth add up over time, saying a country growing one percent too slow for 23 years would end up with an economy 25 percent smaller than it should have been.
But Phil Izzo of the Wall Street Journal points out that, actually, a growth deficit of one percent per year for 23 years would leave an economy 20 percent smaller, not 25 percent.
Reinhart acknowledged the growth-rate error to the WSJ's Izzo, but said the number didn't mean much; she was only trying to illustrate a broader point, that small changes in growth have big effects over time.
For example, Reinhart and Rogoff point to a table in their initial research that they say proves their point still stands, even accounting for their errors. Herndon points out that the table only supports them because it includes their initial Excel spreadsheet error, along with an additional "transcription error" that printed the wrong number for Spanish GDP.
Herndon also disagrees with Reinhart and Rogoff's claim that his paper backs up their paper's broader point. In fact, he writes, once all of the errors are fixed, the statistical significance of Reinhart and Rogoff's findings go up in smoke like a Snoop Lion blunt.
And that's not to mention the chicken-and-egg problem at the root of Reinhart and Rogoff's work -- their "biggest mistake," as the Atlantic's Matthew O'Brien puts it: They imply strongly that high debt causes slow growth, when there is no evidence for that. In fact, one of Herndon's professors writes, there is evidence that the relationship could work the other way around.
It is doubtful that these errors will convince Reinhart and Rogoff to exactly abandon their research or its findings. And certainly the problems with their paper have not yet changed the minds of their fans, austerity pushers like Erskine Bowles, and Germany.
But with every one of these tiny errors the austerity argument weakens just a little bit more.