Searching for a Tide Table for Business

Abstract

The article investigates the methods and conceptions of statistical inference used in business forecasting in the United States and in Europe in the 1920s. After presenting the methods and arguments used by the members of the Harvard Committee on Economic Research in the first years after its establishment in 1919, the article explores the far-reaching changes in method and conviction from 1922 on. The members’ realization that the future evolved differently than predicted prompted them to give up their hope for mechanical means of forecasting and to revoke their calls for the employment of the mathematical theory of probability in economics. Instead, they established an extensive correspondence with economic and political decision-makers that allowed them to base their forecasts on “inside information.” Subsequently, the article traces European attempts to adopt the Harvard Index of General Business Conditions in the early 1920s. Impressed by the seemingly mechanical working of the Harvard index, European economists and statisticians sought to establish similar indices for their countries. However, numerous revisions of the Harvard index in the mid-1920s cast doubt on the universality of the index and the existence of stable patterns and led European researchers to pursue different paths of investigation. The article complicates the larger history of statistical inference in economics in two meaningful ways. First, it argues that statistical inference with probability was not the long-sought solution for the problem of objectivity but a long-contested, and repeatedly discarded, approach. Second, it shows that these contestations were often triggered by deviations between forecasts and the conditions actually observed and by this means argues for the importance of the historical context in the history of economics.