The Econometricians' Statisticians, 1895–1945

Abstract

"Modern statistical theory originated in England, and is today advancing faster there than in any other country." So Harold Hotelling (1930a, 186) informed the American Statistical Association after a visit in 1929. Hotelling traced the origins to Karl Pearson and attributed the current progress to R. A. Fisher. A few years later there were two new theorists to watch, Jerzy Neyman and Abraham Wald, Fisher's main rivals/successors in the theory of statistical inference. I will be considering these four and how they interacted with the econometricians-with H. L. Moore, Mordecai Ezekiel, Henry Schultz, Tjalling Koopmans, and Trygve Haavelmo especially. I also note some statisticians who might have been, but were not, the econometricians' statisticians.Research in the history of econometrics has concentrated-naturally enough-on the distinctive concepts developed by econometricians, "structure" and "identifi cation" being outstanding instances, but the history of econometrics is also a history of using ideas from statistics. These ideas and how they traveled will be my concern; more broadly, I will be considering the way statisticians and econometricians worked together, or,

Big Data and Audience Measurement: A Marriage of Convenience?

“…Among the earliest techniques are Hebb's cell assembly theory (Hebb, 1949) (which later gave birth to the "perceptron" in the 1950s, and then to neural networks), with Widrow and Hoff (1960) demonstrating, around fifteen years later, the links with least-squares methods, the SVM (support vector machine) and, more recently, boosting methods. While the two communities have developed in parallel, big data require links to be built between the two approaches by bridging the "two cultures" referred to by Breiman (2001a), contrasting mathematical statistics, which may be likened to traditional econometrics (Aldrich, 2010), with computational statistics and machine learning more generally.…”

The Role of the Cowles Commission and RAND Corporation in Transforming Mathematical Economics in the Mid-twentieth Century

“…The literature on the history of econometrics is rich and varied In addition to the classic studies by Epstein 1987 Morgan 1990 and Qin 1993 there are more recent ones by Aldrich 2010Bjerkholt 200520072015and Qin 2013 Modern statistics was established around the turn of the twentieth century by English biometrical researchers committed to Darwinism and eugenics Francis Galton Karl Pearson and others presented their notions of correlation coefficient and probability distributions American economist Henry Ludwell Moore regarded Pearson highly and started statistical studies by using their ideas and the least square method which had been developed in the early nineteenth century to smooth out measurement errors Henry Schultz Moore s pupil at Columbia University followed in his teacher s footsteps and worked on the estimation of the demand curves of agricultural products Norwegian economist Ragnar Frisch started his career by quantitatively…”

Multiple Regression, Longitudinal Data and Welfare in the 19th Century: Reflections on Yule (1899)

“…In fact, Yule () is part one of what Yule intended to be a two‐part investigation with a planned extension to cover Wales, to deal with ‘unaccounted’ changes and to examine the economic correlations of changes in outrelief. But part two never appeared in print and there is some doubt about how much influence part one had on developments in statistics: Hepple () argued for diffusion but Aldrich () is less convinced. Certainly, Pearson () is rather critical of Yule's general approach and Kendall (), in his warm appreciation of Yule and his contributions to statistics, barely mentions Yule ().…”