SPREADSHEET ERRORS (COMMENT). In a comment on this post, Henry Nejako pointed out an article in New York magazine about how a graduate student had found “a major flaw in an influential 2010 economic paper by Carmen Reinhart and Kenneth Rogoff”. This article by Carl Bialik in the Wall Street Journal (April 20-21) explains what happened. The paper had calculated that when a country’s ratio of debt to GDP gets above 90%, its rate of growth becomes negative (minus 0.1%). The error in the paper was in leaving out 5 countries in the calculations. Putting them back in gives a growth rate of positive 0.2 % rather than minus 0.1%. The paper which identified the error disagrees with other arguments in the Reinhart-Rogoff paper and concludes that the growth rate for countries with debt ratios over 90% is 2.2%.
Bialik points out some important lessons: 1. Peer review does not usually catch calculation errors. (This blog post by Bialik sheds light on what is actually done in peer reviews in economics.) 2. It is difficult to collect a large sample of data for many macroeconomic issues, and to do so may require using data from decades ago. (My post pointed out problems of constructing large enough data samples in neuroscience.) 3. Causation is difficult to establish (because, as I have pointed out, nature doesn’t do very good experiments). Does a high debt ratio lead to lower growth or does lower growth lead to a high debt ratio?