Joseph A. Schumpeter: Henry L. Moore: Laws of Wages. An Essay in Statistical Economics (1913)

Schumpeter’s German book review presents Moore’s study of wages as a pioneering attempt to bring economic theory into precise contact with statistical evidence. Its central claim concerns the future of economics as much as the merits of the reviewed book: modern statistical methods, chiefly inherited from the Galton school, make possible a closer empirical engagement than earlier techniques allowed. They can introduce new facts into theory, establish foundations for analysis where ordinary methods fail, and test theoretical results.

Die Zukunft der Oekonomie hängt zu einem großen Teil vom Erfolge solcher Bestrebungen ab: Nur auf diesem Weg kann sie sich zu einer unmittelbar auf die Tatsachen anwendbaren, in ihrer Autorität durch zahlenmäßige Resultate gestützten Disziplin entwickeln.

English translation: The future of economics depends in large part upon the success of such endeavours: only along this path can it develop into a discipline immediately applicable to the facts and supported in its authority by numerical results.

Schumpeter does not suggest that theory previously lacked any relationship to facts. The advance lies in extracting precise answers from numerical series beyond the simplest cases. He judges Moore’s contribution against this methodological ambition, defending exploratory work against criticism that treats immature methods and imperfect materials as evidence of an enterprise’s futility. The book’s errors matter, but they do not cancel its achievement in demonstrating how advanced statistics might serve economic inquiry.

The review moves from this programmatic defense to an account of Moore’s investigations. Schumpeter praises the introduction and concluding chapter as accessible explanations of the undertaking, then follows the discussion of methodological foundations into correlations between wages, food prices, living standards, and skilled and unskilled pay. Here his endorsement becomes epistemologically exacting: Moore’s low correlation between wages and food prices does not establish the absence of a causal relationship.

Ueberhaupt liegt der Wert einer Bestätigung der Theorie durch entsprechend hohe Korrelation zweier Phänomenreihen nur darin, daß wir uns insoweit bei der betreffenden Theorie beruhigen können, während im Fall niedriger Korrelation ein neues Problem auftaucht: Bewiesen ist die Theorie im ersten Fall so wenig als sie im zweiten widerlegt ist.

English translation: In general, the value of a confirmation of the theory by a correspondingly high correlation between two series of phenomena lies solely in this, that to that extent we may rest content with the theory in question, whereas in the case of a low correlation a new problem arises: the theory is as little proved in the first case as it is refuted in the second.

Statistical confirmation thus gives grounds for retaining a theory without proving it; a disappointing correlation opens a problem rather than conclusively refuting the proposed explanation. This qualification governs Schumpeter’s reading of Moore’s finding that wages correlate closely with product value. The result supports the productivity theory’s empirical adequacy, not its definitive demonstration. Studies of strikes and industrial concentration likewise show the method’s promise, including its relevance to whether workers’ associations can influence wages. Schumpeter particularly values Moore’s choice of suitable indices where direct measurement is unavailable.

The review reserves its most extended scrutiny for the fourth chapter, “Wages and Ability,” which argues that wages are broadly distributed according to workers’ ability. Schumpeter distinguishes a theoretically intelligible association from a merely formal resemblance between frequency curves. If an association is already plausible, correspondence between distributions can provide the limited, negative confirmation available to statistics; without that intelligibility, the resemblance alone has little significance.

Moore’s further use of Galton’s difference problem, in Pearson’s formulation, asks whether differences in individual wages are proportional to differences in ability. Schumpeter reports the affirmative result but makes his approval conditional on the validity of the underlying problem and its solution. He does not settle that foundational issue within the review. The conceptual distinction is important: an ingenious application cannot itself establish the soundness of the framework it applies.

Zweifellos wird die Arbeit an den Problemen, die Moores Methoden zugrunde liegen, diese Methoden bald verändern. Das ändert nichts an dem Verdienst, schon heute getan zu haben, was eben heute zu tun möglich ist.

English translation: Doubtless the work upon the problems that underlie Moore's methods will soon alter those methods. That takes nothing away from the merit of having already done today what it is precisely today possible to do.

The closing judgment crystallizes the review’s balance of advocacy and restraint. Moore’s lasting merit lies in making a productive research program concrete, even where its methods remain provisional. Schumpeter’s review is consequently significant both as an endorsement of statistical economics and as a warning against equating correlation with causal proof or theoretical verification.