Gerhard Tintner’s review essay assesses Henry Schultz’s 1938 treatise as both a landmark in empirical demand analysis and an occasion to reconsider its methodological foundations. Writing especially for agricultural economists, Tintner places Schultz at the culmination of the econometric tradition initiated by Henry Ludwell Moore: deriving Marshallian demand curves from economic time series. His admiration is substantial, extending to consumer theory, agricultural commodity studies, statistical technique, and policy relevance. Yet the review’s central argument is that demand measurement cannot adequately connect theory and evidence while treating an evolving economic process as essentially static.
It is the most important of all books and articles which deal with the derivation of statistical demand curves from time series.
This opening judgment establishes the significance of the work under review, rather than announcing Tintner’s agreement with every method it employs. The essay first surveys Schultz’s book, then develops two connected assessments under economic and statistical methodology. Its movement from exposition to criticism makes the review a proposal for further research as well as an appraisal of an accomplished synthesis.
Schultz’s first part supplies the historical and theoretical foundations: demand and indifference curves, methods for estimating demand from time series, family-budget approaches, and Schultz’s own procedures. Tintner praises the historical treatment while identifying its limited attention to dynamic problems. The second part estimates demand for ten American and Canadian agricultural commodities, including sugar, corn, cotton, wheat, and potatoes, comparing trend-ratio and link-relative methods. The third, which Tintner considers potentially still more important, investigates related demands: interactions among meats, among sugar, tea, and coffee, and among feed crops. Its discussion of complementarity connects empirical estimation with the theory of consumer choice. Appendices supplying basic data and explaining curve fitting, correlation, and significance testing further strengthen the book’s usefulness to economists and statisticians.
Under economic methodology, Tintner values Schultz’s careful assembly of consumer theories but questions their accessibility to economists without mathematical training. He also finds the treatment insufficiently responsive to recent work by Hicks and Allen, particularly on complementarity. These reservations are secondary to the omission of dynamic utility and demand theory. Static assumptions may be defensible for family-budget evidence, but observations extending across time require a more explicit account of economic change.
Hicks has shown in his recent book⁸ that the time has arrived for a consistent dynamic treatment of all economic problems and there is no reason why the field of demand or consumer's choice should be excluded.
Tintner traces this theoretical omission into Schultz’s estimation procedures. Dividing the period 1875–1929 into three segments and fitting separate regressions produces successive quasi-static curves rather than a coherent account of their movement. Treating time as a catch-all similarly permits parallel shifts while retaining a constant slope. The objection is therefore not simply that additional variables would improve a regression: the underlying conception of demand restricts what historical change the statistical model can represent.
The statistical assessment follows the same pattern of strong praise and fundamental qualification. Tintner regards Schultz’s exposition of least squares and his mathematical appendix as outstanding, especially for incorporating modern significance testing.
Only tests of significance can help us to distinguish statistically what is valid from what is spurious.
Nevertheless, such tests cannot establish validity when the distinctive properties of time series remain untreated. Tintner distinguishes random variation, which can be handled like randomness in other data, from a specifically temporal component. He invokes variate differences and serial-correlation methods as approaches to this distinction. Ordinary statistical procedures cannot simply be transferred to time-series observations, and an undifferentiated time term does not solve the problem. The deeper failure, in his account, is insufficient integration between economic theory and statistical treatment.
Tintner’s constructive alternative begins with a theoretically plausible demand function whose parameters vary through time. For the linear example (Y=A+BX), with price represented by (Y) and quantity by (X), the intercept describes the curve’s location and the coefficient its slope. Both may change as population, technique, expectations, and tastes change. The resulting formulation, (Y_t=A_t+B_tX_t), permits movement in the curve’s shape as well as its position.
There is no need of chopping up data into small parts in order to obtain a quasi-static demand curve for every period.
This proposal is explicitly provisional. Making parameters functions of time is only a first approximation; a stronger explanation would relate their movements to economically meaningful variables. Tintner also questions whether per-capita quantities and price-index deflation adequately approximate static conditions. Least squares remains useful, but its application must reflect an evolving economic relationship rather than impose stability by segmentation.
But we should not be content with this result without investigating if the residuals are random.
The closing requirement binds theoretical specification to statistical diagnosis. Residual randomness must be investigated before standard errors and significance tests can support reliable conclusions, although Tintner acknowledges that the available tests are imperfect. The review consequently treats Schultz’s achievement as a foundation for dynamic econometrics: meaningful demand measurement requires both an economic explanation of changing parameters and scrutiny of the temporal structure left unexplained by the fitted model.
This work was divided into 3 sections when it entered the library's research corpus—an apparatus for search and citation, not necessarily the author's own table of contents. Each title opens its summary.
Put a question to this work; the Librarian answers from its 3 sections and cites the passage.
Ask the Librarian