Joseph A. Schumpeter, Historical Approach to the Analysis of Business Cycles (1951)

Schumpeter’s conference contribution develops a research program in which historical investigation, economic theory, and statistical analysis jointly explain business cycles. Published as the unrevised draft prepared for the conference, it proceeds through five sections: a methodological clarification, a distinction between cyclical mechanisms and avoidable disasters, an interpretation of the American depression beginning in 1929, an examination of dynamic models, and a proposal for coordinated industrial histories. Its central claim is that understanding fluctuations requires reconstructing particular economic transformations, not merely describing aggregate movements.

Schumpeter begins by rejecting any competition between historical research and theoretical or statistical work. Theory supplies analytical tools as well as explanatory hypotheses; equilibrium analysis remains necessary alongside dynamic models. Historical inquiry matters because those tools must address an economy undergoing irreversible change:

Economic life is a unique process that goes on in historical time and in a disturbed environment.

This premise supports two distinct arguments. The first concerns depression policy: the most destructive features of crises need not belong to the cyclical mechanism itself. The second, which Schumpeter considers scientifically more important, concerns the mechanism’s explanation: aggregate fluctuations conceal changes in industries, firms, and economic structures. In both cases, cycles must initially be investigated as historical individuals before general conclusions can be established.

The depression of 1929 illustrates the policy argument. Schumpeter believes an unusually intense and prolonged depression was already due, but distinguishes this underlying movement from banking failures, abnormal mortgage indebtedness, and speculative excess. These aggravated the contraction and generated demoralization with lasting political consequences. His claim is neither that the depression could have been abolished nor that its institutional setting was a matter of chance:

Now the point is that those three factors were logically separable from the underlying process from which they arose and that they were practically avoidable.

“Practically avoidable” is conditional on an intelligent, sufficiently powerful government and a properly organized banking system, not a description of the capacities of the actual American authorities. Schumpeter attributes banking vulnerability to numerous inefficient small banks and mismanagement in larger institutions; he argues that resistance to concentration prevented a more resilient branch-banking structure. Agricultural catastrophe arose, in his account, from farmers’ debts, while urban real-estate credit compounded the trouble. Speculative excess required controls other than interest-rate changes and credit rationing. These are forceful historical judgments offered illustratively rather than fully demonstrated in this short contribution. Their purpose is to identify reforms that could reduce suffering without suppressing the cyclical process itself.

Section IV moves from avoidable aggravations to the underlying process. Dynamic models can reproduce investment fluctuations and illuminate oscillations that unaided historical description might miss. Yet mathematical fit does not establish what generated the movement. Investment expenditure is itself a surface phenomenon; autonomous investment requires explanation through the industrial processes that transform existing structures. Treating “expectations” as that explanation merely substitutes one unexplained category for another.

Schumpeter consequently qualifies the division between impulse and propagation problems. Historical analysis does not exclusively identify impulses while models exclusively describe their transmission. Models explain situations that help generate impulses, and historical inquiry can clarify propagation. Their relationship is reciprocal. He tentatively proposes that major prosperities and depressions since the late eighteenth century might be explained without endogenous oscillations of the elastic type, while explicitly declining to deny such oscillations’ existence. Multiple movements of different kinds may coexist; their relative importance must be established case by case.

Section V defines the evidence this program requires. Long time series are indispensable, but measurement alone does not supply an account of structural change:

Apart from the measurements it yields, a set of time series does not so much solve any problem as state in quantitative terms what problem there is to solve.

Chronological annals likewise aid interpretation and check theoretical claims without adequately explaining industrial transformation. Schumpeter therefore calls for a large collection of industrial and locational monographs prepared according to a common plan. These should investigate changes in production and consumption functions alongside the quality and conduct of leading personnel. His use of “consumption function” is expressly distinct from the Keynesian concept.

The proposal rests on a qualified counterfactual: if production and consumption functions remained unchanged, society would not exhibit the kinds of fluctuations ordinarily identified as business cycles. Schumpeter exempts wars, political disturbances, and harvest variation, and reserves judgment about short inventory cycles and comparable commodity fluctuations. Whether much remains after these exclusions is itself an empirical question. Industrial histories would explain the remaining mechanisms while testing and revising theoretical constructions.

Theoretical and statistical analysis is in this task as necessary as is historical work. In fact they are inseparable because there is an incessant give and take between them.

The contribution’s lasting relevance lies in this demand to connect aggregate behavior with the rise and decline of particular firms and industries. Such knowledge is necessary both for explaining cycles and for assessing questions such as monopoly’s effects on prosperity and depression. Historical inquiry supplies neither decorative examples nor an alternative to formal analysis: it identifies what models must explain and what policy might actually change.