Abstract
The article proposes a new approach to the Reinhart–Rogoff affair. In contrast to the two explanations of the controversy put forward in existing literature, I argue that Growth in a Time of Debt and its criticism published by Herndon, Ash and Pollin (Does high public debt constantly stifle economic growth? A critique of Reinhart and Rogoff) exemplify the ‘emerging contrary result’ phenomenon (emerging recalcitrant result, ERR). Three arguments support this hypothesis. First, the infamous spreadsheet error did not cause the findings to differ. On the contrary, the results differed mostly due to employing alternative averaging schemes. Second, the cliometric techniques employed by both research teams are justified to a similar degree. Third, the pattern in the cliometric literature focused on the 90%-threshold hypothesis suggests that the Reinhart–Rogoff controversy exemplifies an ERR phenomenon caused by submission/publication bias.
Acknowledgement
The author would like to thankfully acknowledge Professor Robert Goldfarb for his help in understanding the ‘emerging contrary result’ phenomenon and Gabriela Staron, MMed for explaining how new evidence is corroborated by medical science. Last but not least, I also express gratitude for the comments received from the reviewers of the Journal of Economic Methodology which helped me in improving the line of argument.
Notes
1. Those opinions can be exemplified by the power words ‘Excel Error That Changed History’ from the title of an article on the controversy published by Bloomberg (Coy, Citation2013).
2. RR, henceforth.
3. HAP, henceforth.
4. Among others, the list includes Manuel Barroso (President of the European Commission), Olli Rehn (Vice-President of the European Commission), Angela Merkel (German Chancellor), Wolfgang Schäuble (German Finance Minister), George Osborn (British Chancellor of the Exchequer) and Paul Ryan (American Representative, the author of the report Path to Prosperity) (Botsch Citation2013; Smith Citation2007).
5. However, due to employing a different method, the article should rather be called a robustness check (Clemens, Citation2015).
6. HAP focused their criticism on the post-war developed countries analysis. In response, Reinhart and Rogoff (Citation2013a, 2013b) accused them of being biased since the other two studies are free from some of the raised issues. Similarly, a majority of econometric literature devoted to the 90%-threshold hypothesis also concentrates on the post-war developed countries.
7. I certainly do not state that the spreadsheet error may be defended on methodological grounds. Such mistakes should not happen in analyses of high academic and practical influence. However, the miscoded Excel formula biased the results insignificantly and did not undermine RR’s conclusion or created the difference between RR and HAP.
8. In connection with the Reinhart–Rogoff affair.
9. Their (2014) Cambridge Journal of Economics article was originally published as a working paper (Herndon, Ash, & Pollin, Citation2013).