Joseph A. Schumpeter, Mitchell’s Business Cycles (1930)

Joseph A. Schumpeter’s journal review article examines Wesley C. Mitchell’s Business Cycles: The Problem and Its Setting as both a major contribution to cycle research and a statement about economic science. Its four sections move from methodology and the relation to Marshallian economics through institutional analysis, statistics, and business annals. Schumpeter’s central argument is that Mitchell’s empirical program requires analytical theory—not merely hypotheses awaiting statistical verification, but instruments that make economic facts intelligible. His criticism proceeds within admiration for Mitchell’s achievement and seeks cooperation rather than a division into opposing schools.

The opening section explains why a conventional introduction or assessment would be inadequate: Mitchell’s standing is already assured, while detailed criticism would exceed a review’s limits. Schumpeter instead considers the book’s “general message,” treating business-cycle inquiry as an entry into the fundamental problems of capitalist society. He praises Mitchell’s combination of observation, statistical technique, and rational hypotheses, and his rejection of fruitless methodological controversy. Mitchell’s engagement with previous scholarship also affirms scientific continuity:

His teaching on this point may be summed up, I believe, saying that any “New Economics” can come about not by program but only by achievement.

Yet Schumpeter finds Mitchell insufficiently appreciative of theoretical construction and of theoretical proof or disproof. The distinction becomes clearer through comparison with Marshall. Where Marshall built an analytical engine, Mitchell approaches theory chiefly as a stock of hypotheses, doctrines, or empirical generalizations. Schumpeter argues that this difference in intellectual temperament need not become an epistemological opposition:

This being the case there is a great difference in emphasis, and yet no epistemological gulf between Mitchell and Marshall; not even between Mitchell and Marshall’s fifth book (appendix included).

He considers four possible grounds of disagreement. Analytical tools need not depend on an objectionable psychology; disputes over particular tools should be settled by theoretical argument rather than methodological allegiance; existing theory often anticipates numerical investigation; and theory necessarily changes as factual knowledge develops. This last concession does not subordinate theory to data collection. Factual study and analytical construction transform one another, and improving theoretical instruments remains an independent scientific task.

Section II examines Mitchell’s institutional account of economic organization. Schumpeter welcomes its positive contribution and its freedom from the tendency, associated with aspects of Schmoller’s and Veblen’s reasoning, to substitute institutional investigation for economic analysis. His reservations concern particular explanatory choices. Money rationalizes conduct and has profound civilizational importance, but this does not establish that monetary and nonmonetary economies require fundamentally different principles. He therefore prefers the division of labor to making and spending money as the foundation of the business economy.

Similarly, Mitchell’s account of interconnected prices is compatible with Walrasian theory but loses precision by avoiding its technical apparatus. Elasticity, cost curves, and quasi-rent would help empirical investigators distinguish relations that observation alone cannot disentangle. Schumpeter’s discussion of equilibrium makes the same point: processes tending toward balance should be distinguished from processes producing disequilibrium. His objection is less that Mitchell contradicts theory than that he leaves useful distinctions underdeveloped.

Section III gives statistics the largest methodological role. The decisive advance is not simply acquiring more observations but learning to think in statistical quantities and time series. Cooperation must run both ways: economics must learn from statistics, while statistical procedures must be adapted to the economic phenomena they measure. Averages, correlations, and trends do not acquire substantive meaning merely through correct calculation. Schumpeter questions a distribution of 106 cycle durations because changing historical norms and inconsistent criteria for identifying cycles undermine its interpretation as a statistical sample. Price indices likewise require a prior account of what a general price level means, while correlations may obscure real relations or suggest nonexistent ones.

Trend analysis brings these difficulties into sharpest focus:

Failing this, a trend is no more than a descriptive device summing up past history with which nothing can be done. It lacks economic connotation. It is, in fact, merely formal.

Curve fitting cannot decide whether trends generate cycles, cycles generate trends, or a trend represents a distinct economic process at all. Schumpeter proposes distinguishing gradual growth, industrial and commercial innovation, and influences originating outside the economic system, including wars and harvests. Theoretical schemata for these different processes would give empirical research more definite questions and results.

Section IV values business annals as historical evidence capable of revealing what time-series analysis misses, especially when supplemented by studies of industries and firms. But Schumpeter finds Mitchell excessively cautious in drawing explanatory conclusions. Without theory, differences between countries remain insufficiently interpreted, and the classification of recessions has no firm stopping point: ever smaller fluctuations can become additional cycles.

The final criticism concerns Mitchell’s attempt to establish cycles as a distinct phenomenon through recurrent depression, revival, prosperity, and recession. Recurrence alone cannot preserve the phenomenon’s explanatory contours if movements caused or altered by wars and other disturbances are indiscriminately included. Schumpeter also suggests that long waves may prove more significant than the cycles currently foregrounded. These objections culminate in the review’s governing methodological claim:

Such difference in outlook as remains between us can be summarized by saying the theory of the cycle is not the last but the first step on the road to our goal.

The review thus locates Mitchell’s lasting importance within a reciprocal relationship between empirical inquiry and theory. Its relevance lies in showing that measurement, historical description, and even the identification of the object called a business cycle depend on explicit analytical decisions.