Statisticians and the History of Economics

Over twenty years ago, I had the pleasure of speaking to the Royal Statistical Society about the life and work in statistics of Francis Ysidro Edgeworth. In a preamble to that talk, I commented upon some similarities between Edgeworth and an even earlier scientist, Daniel Bernoulli, who lived from 1700 to 1782. One similarity I mentioned would be so well-known to this audience that it scarcely bears repetition: Both Edgeworth and Bernoulli were important in the histories of both economics and statistics. Of Edgeworth's work in economics I need not comment—I expect that his work on taxation, price theory, international trade, and index numbers is familiar to many in this audience. Edgeworth's work in statistics was no less important in at least four ways: (1) He was an astute and original philosopher of the application of probability and quantitative thinking in social science; (2) he had sufficient mathematical genius to situate Galton's concept of correlation in the framework of the multivariate normal distribution, thereby pointing the way for Karl Pearson to develop correlational analysis—Edgeworth even has a claim on being the inventor of the product moment correlation coefficient; (3) he pioneered the exploration of many topics, such as variance component estimation, the theory of maximum likelihood, and series expansions for sampling distributions; and (4) he had an amazing ability to explore the limits of statistical concepts by constructing subtle and surprising counterexamples to widely accepted truths.