An Economist's Mea Culpa: I Relied on Reinhart and Rogoff

Here is a link to my 22d column on Quartz: “An economist’s mea culpa: I relied on Reinhart and Rogoff.”

Let me also reprint here from my update to “Noah Smith Joins My Debate with Paul Krugman: Debt, National Lines of Credit and Politics” in the light of recent events:

You can see what I have to say in the wake of Thomas Herndon, Michael Ash and Robert Pollin’s critique of Carmen Reinhart and Ken Rogoff’s work on national debt and growth in my column “An economists mea culpa: I relied on Reinhart and Rogoff.” (You can see my same-day reaction here.) Also, on the substance, see Owen Zidar’s nice graph in his post “Debt to GDP & Future Economic Growth.” I sent a query to Carmen Reinhart and Ken Rogoff about whether any adjustments are needed to the two figures from the paper with Vincent Reinhart that I display below, but it is too soon to have gotten a reply. I think that covers most of the issues that recent revelations raise. 


Note that I have revised “What Paul Krugman got wrong about Italy’s economy.” [My post “Noah Smith Joins My Debate with Paul Krugman: Debt, National Lines of Credit and Politics”] is now the go-to source for what I originally said there, relying on “Debt Overhangs, Past and Present” (which has Vincent Reinhart as a coauthor along with Carmen Reinhart and Ken Rogoff). 

One final thought. Given the spotlight they put on Reinhart and Rogoff, and the spotlight that is therefore on them as well, Thomas Herndon, Michael Ash and Robert Pollin may not be as careful as they should be. Am I mistaken in what I said in this tweet:

Herndon, Ash & Pollin report 11% p-value for 30-60% debt GDP vs. BOTH 60-90% and >90% bins, but don’t report p-value for 60-90% vs. >90% !

One way or the other, they should report the p-value for the 60-90% bin vs. the above 90% bin alongside the test they do report, which is less germane to the controversy about the 90% threshold. They should have made the results of the 60-90% vs. above 90% statistical test impossible to miss. How easy is it to find their report of that statistic in their paper?

Note on Comments on this Blog: I want to encourage more commentary on my blog. I need to approve each comment unless you are whitelisted. But if you send me a tweet to let me know you need a comment approved, I will get to it quicker, and normally approve it and whitelist you. I do try to enforce a certain level of civility and decorum (including a language filter), but on substance, I want a robust debate. For Quartz columns, the link I post on my blog (usually the day after the column appears) is a good place to make comments.