Probability and Depreciation: A History of the Stochastic Approach to Index Numbers

Early in this century economists began using the new statistical techniques of correlation and regression developed in biometrics. Thus began a period of uninterrupted growth in what became known as econometrics. ’ There were precedents. Long before, Cournot and Jevons had proposed several “econometric” projects. The statistical methods underpinning these projects had originated in astronomy. One of the projects, the estimation of demand laws, was reinvigorated to become the paradigm econometric investigation of the twentieth century. Another never became part of the canon of econometrics although it had received the full higher statistics treatment in the nineteenth century. This project-which involved the use of probabilistic reasoning in the analysis of price data-is my subject. At the end of the nineteenth century, this project may well have seemed more promising than the estimation of demand laws. In the 1860s, Jevons failed to estimate “laws of variation of price” but succeeded in applying probabilistic reasoning to the depreciation of gold. In the 1880s, Edgeworth continued the work, combining research in statistical theory with applications to economics in a way that became