Gerhard Tintner · 1949
Gerhard Tintner’s article, based on a lecture delivered on 21 December 1948, defines econometrics through its relations to mathematical economics, statistical research, and the wider empirical sciences. Its central argument is that economic knowledge requires both explicit theoretical assumptions and quantitative verification. Mathematical deduction alone cannot establish empirical validity, while collecting observations without theory leaves the principles of selection and interpretation unstated. Tintner proceeds from these methodological distinctions to the difficulties of analysing nonexperimental data, illustrates structural estimation through agricultural demand, and concludes with the unresolved problem of assessing entire theories.
L'économétrie se propose l'atteinte de résultats numériques et la vérification de théories économiques par application des méthodes mathématiques et statistiques.
English translation: Econometrics aims to obtain numerical results and to test economic theories by applying mathematical and statistical methods.
This opening makes econometrics a conjunction of theory, measurement, and testing. Mathematical economics supplies deductions from assumptions: preferences and utility functions support analyses of consumer choice, while production functions and market structures support analyses of profit maximization. Tintner also notes the promise of von Neumann and Morgenstern’s strategic games, although he considers their ideas insufficiently developed for a usable theory of economic behaviour. His defence of mathematics is nevertheless not a claim that verbal economics addresses fundamentally different questions. Mathematical reformulations of Keynes and Böhm-Bawerk clarify assumptions, generalize results, and expose theoretical limits. Following Bertrand Russell’s connection between mathematics and logic, Tintner argues that economists who accept logical deduction cannot consistently reject mathematical reasoning.
The scope of such deductions remains conditional:
Toutes les lois économiques sont conditionnelles.
English translation: All economic laws are conditional.
This qualification becomes especially important in welfare economics. Economic analysis can investigate how resources should be used given specified social objectives, but cannot itself establish those objectives. Tintner’s tariff example distinguishes an empirically testable proposition—that reducing duties increases national output—from the judgment that duties ought to be reduced. Higher living standards may conflict with economic self-sufficiency and wartime national security. Econometrics can estimate consequences, not decide which ends deserve priority. Its usefulness for policy therefore does not confer authority to settle normative disputes.
Tintner next rejects the opposite reduction of economics to statistical description. His criticism of the National Bureau of Economic Research’s approach, discussed through the controversy over “measurement without theory,” is that observation never begins without assumptions. Selecting data and arranging material already presuppose a conception of the phenomenon, even when that conception remains implicit. A complete description of a concrete event is impossible in natural as well as social science. National-income estimates and production statistics are indispensable, but they acquire explanatory force through theoretical interpretation:
Mais les données statistiques ne sont pas suffisantes. Il faut les interpréter par la théorie et les analyser par les méthodes de la statistique moderne.
English translation: But statistical data are not sufficient. They must be interpreted through theory and analysed using the methods of modern statistics.
Econometrics thus occupies a mediating position between radical empiricism and pure theory resistant to concrete investigation. This position does not make it universally sufficient. Tintner explicitly preserves historical methods for studying the development of capitalism where suitable quantitative evidence is lacking, and institutional methods for understanding the legal and social arrangements governing banking. Economics studies the administration of scarce resources to satisfy individual needs; it is consequently empirical, like physics and biology, rather than a priori, like logic and mathematics. Econometrics is its principal method of verification, within a broader methodological repertoire.
The article’s technical discussion explains why importing statistics requires adaptation. Sampling and multivariate analysis are important, but methods developed through agricultural experimentation and industrial applications cannot simply be transferred unchanged. Tintner presents economics as unable to conduct experiments, comparing it with astronomy and meteorology. The comparison counters the suggestion that this limitation uniquely disqualifies social science. Haavelmo and the Cowles Commission matter because they develop statistical approaches to observations not generated by controlled experiments.
Identification is the decisive connection between statistical results and economic explanation:
Il est nécessaire de reconnaître, dans les résultats statistiques, les relations qui ont un sens économique bien défini : par exemple, fonctions de la demande, fonctions d'offre, fonctions de production, fonctions de consommation, etc.
English translation: It is necessary to recognize, in statistical results, relationships that have a well-defined economic meaning: for example, demand functions, supply functions, production functions, consumption functions, etc.
Observed data arise from interacting relationships; policy analysis therefore requires structural coefficients rather than unexplained associations. Tintner contrasts the Cowles approach, which allows equation errors from omitted variables, with his own work allowing errors in the variables. His illustration uses annual United States agricultural data for 1920–1943 and maximum-likelihood estimation. He reports a price elasticity of demand of −0.123, with a standard error of 0.0341901, and interprets demand as comparatively insensitive to price under the period’s conditions and a cæteris paribus assumption. He uses significance testing, confidence limits, and rejection of unit elasticity to show how numerical estimates can inform policy and discriminate among hypotheses. These are presented as provisional results; his probability-based explanations of significance and confidence belong to the article’s own statistical exposition.
The conclusion distinguishes testing a coefficient from evaluating a whole theoretical framework:
Il n'existe pas encore une méthode d'induction assez générale pour résoudre ce problème.
English translation: There does not yet exist a sufficiently general method of induction to solve this problem.
The problem is how evidence could establish the relative probability of Keynesian and classical theories. Tintner sees promise in Carnap’s work on probability and induction but acknowledges that a complete method remains unavailable. The article’s relevance lies in this combination of ambition and restraint: econometrics makes economic claims explicit, measurable, and testable, while leaving normative choice, institutional understanding, and comprehensive theoretical confirmation as distinct problems.
This work was divided into 2 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 2 sections and cites the passage.
Ask the Librarian