The eminent Yale monetary and capital theorist Irving Fisher is best known in the economics profession for the equation of exchange, the distinction between real and nominal interest rates, and an early analysis of intertemporal allocation. In fact, his contributions as an early macroeconomist were extensive, and his debt-deflation theory of depression both deserves and implicitly gets consideration currently. It is Fisher's misfortune, as well as that of the profession, that his analysis of the debt-deflation process, one of his most insightful contributions to macroeconomics, received little notice from his contemporaries.
Fisher has acquired lasting and unenviable fame for his predictions in September 1929 that "stock prices have reached what looks like a permanently high plateau" (Barber, 1985, p. 77) and for the consequences to Fisher of his predictive error. As John Kenneth Galbraith (1977, p. 192) remarked, "In the late nineteen-twenties Fisher went heavily into the stock market and in the Crash lost between eight and ten million dollars. This was a sizable sum, even for an economics professor." Fisher was known for this even to those, such as Robert Sobel (1968, pp. 97, 132), whose direct knowledge of Fisher and his work was vague enough to identify him as "Irving Fisher of Harvard." Kathryn Dominguez, Ray Fair, and Matthew Shapiro (1988) have now shown that even using modern statistical techniques and adding some retrospectively compiled time series to those available to Fisher and to the Harvard Economic Society, it would not have been possible to predict the onset, length or depth of the Great Depression by time-series analysis.
Dominguez, Fair, and Shapiro have done much to redeem Fisher's reputation as a forecaster relative to that of, for example, Roger Babson, whose successful prediction of a break in stock prices in the fall of 1929 must be balanced against his prediction of a stock price boom in 1930. If one views the bull market of the 1920s as a speculative bubble, all that could be predicted is that the bubble would eventually burst, not when.
Indeed, Fisher's formal statistical forecasting method, as distinct from his more subjective statements about future stock prices, held up quite well. In a series of journal articles, Fisher (1923, 1925, 1926) sought empirical verification of his monetary theory of economic fluctuations by correlating output and unemployment with a distributed lag of past changes in the price level (see Dimand, 1992). He was an innovator in the use of correlation analysis and distributed lags, and constructed his own price indices. His 1926 article was republished in 1973 as "I Discovered the Phillips Curve." Fisher (1925) found a correlation of .941 between a trend-adjusted measure of the volume of trade and a distributed lag of monthly inflation rates for the 114 months ending January 1923. Scott Sumner (1990) has found that, using the lag weights from Fisher's 1925 paper, Fisher's equation yielded a correlation of .851 for the period from January 1923 to July 1933, a close out-of-sample fit. The stable relationship between output and inflation collapsed only with the economic policy regime change after Franklin Roosevelt's inauguration when the United States left the fixed gold value of the dollar for what Maynard Keynes termed "a gold standard on the booze" and U.S. output recovered sharply. Thus "updating Fisher's model to the 1923-35 period," Sumner found that "The correlation between the predicted and the actual output series was only .256" because of the structural change in 1933 (1990, p. 721).
Apart from the regrettable impact on his personal finances, the stock market crash and subsequent depression had two important consequences for Fisher as a theorist of economic fluctuations. Fisher's attention was focused on a gap in his analysis in the 1920s of the business cycle as "a dance of the dollar": the need to explain why sometimes a deep and lasting depression occurred. …