Mostly harmless econometrics? Statistical paradigms in the ‘top five’ from 2000 to 2018

Abstract

We explore the connection between four major inferential paradigms in statistical science and inferential practice in current econometrics. We develop the argument that econometrics is still largely characterized by John Stuart Mill’s conception of statistical inference from data, who saw a distinction between ‘theorists’ and ‘practical men’. We follow up with a review of all empirical papers published in the Top 5 economics journals in the period 2000–2018 (N = 2,258). In spite of Rodrik’s [(2015). Economics rules: The rights and wrongs of the dismal science. W. W. Norton & Company] much-debated notion of economics that sees issues of model selection at the core of the discipline, the ‘theory f irst’ / ‘pre-eminence of theory’ approach vastly dominates in the sample (94.0%). When model selection and model uncertainty is accounted for, this largely happens under the frequentist statistical paradigm. This finding may be explained by frequentism’s special role as an ‘orientational paradigm’ (Hoyningen-Huene and Kincaid, [2023]. What makes economics special: Orientational paradigms. Journal of Economic Methodology, 30(2), 188–202) in economics.