Oskar Morgenstern, “Foreword” (1964; reprinted 2015)

Oskar Morgenstern’s foreword introduces a book on spectral analysis undertaken by Clive W. J. Granger, with two chapters contributed by Michio Hatanaka. It combines a methodological critique, a personal history of research, and an assessment of the new methods’ significance for economics. Its central claim is that spectral analysis, supported by digital computation, can replace intuitive accounts of economic fluctuations with more rigorous and revealing analysis.

Only quite recently has the analysis of economic time series reached a level commensurate with the inherent difficulties.

The difficulty, Morgenstern argues, lies partly in economists’ reliance on apparently plausible components of time series and inadequate methods for separating them. He singles out A. Wald’s 1936 work on seasonal variations as an important but neglected advance. More generally, economists resisted Fourier analysis because irregular economic cycles seemed incompatible with periodicity, while unfamiliar computing demands and inconclusive early applications reinforced their skepticism. Subjective identification of peaks and troughs consequently persisted.

This procedure is contrary to the spirit of statistical analysis where every effort is made to extract the last amount of information from data that are difficult to obtain.

This criticism makes the extraction of information, rather than conformity to familiar economic categories, the standard of sound analysis. Morgenstern then recounts his plans with John von Neumann to apply Fourier analysis on a large scale, adapting it to economic series and more precise analytical objectives. Their project was delayed by the availability of computing machinery and competing atomic-physics work; von Neumann’s death in 1957 ended their collaboration. The narrative connects this unrealized project to advances in spectral analysis and prediction theory, John Tukey’s sustained interest, and Granger’s research.

The foreword’s final movement explains why the new approach demands conceptual as well as technical change. Economists must reconsider established ideas, while computers make both applications and exploratory research possible.

Even more, these modern facilities offer possibilities of experimental exploration which can serve to guide theoretical work.

Computation thus becomes an instrument of theoretical discovery, not merely a means of accelerating calculation. Morgenstern identifies the book’s treatment of non-stationarity as a particularly important example. He maintains that earlier objections to conventional Fourier techniques no longer invalidate spectral analysis and that its application can reshape the study of aggregate fluctuations and individual markets. Hatanaka’s chapters illustrate the prospect of uncovering more complex relationships and exposing widely accepted notions as spurious.

It will contribute toward moving economics from an often intuitively oriented and severely limited approach to a germane penetration by proven modern concepts and methods.

The foreword’s relevance lies in this programmatic link between mathematical innovation, computational experimentation, and the revision of economic understanding. Its concluding announcement of further studies sponsored by Princeton’s Econometric Research Program presents the book as an opening achievement in an expanding inquiry: applications will test the new methods while also prompting their further development.