Gerhard Tintner, “Discussion: Estimation of Economic Relationships” (1956)

Gerhard Tintner’s journal discussion contribution briefly assesses papers by Nerlove, Suits and Koizuma, and Ladd. Its central concern is the uneven development of econometric research: promising advances in estimating economic relationships coexist with neglected subjects and unresolved empirical difficulties. The discussion moves from the economic functions requiring investigation to computational technique, then to observational error and multicollinearity.

Tintner welcomes the papers by Nerlove and by Suits and Koizuma against a specific research gap:

Supply functions have been very much neglected in econometric research.

His praise becomes an agenda for further work: he wishes that competent econometricians would also study cost functions. The point is not simply that these papers improve existing methods, but that they direct attention toward relationships insufficiently examined by econometric research.

Ladd’s paper receives a different assessment. Tintner calls it an important contribution to resolving difficult technical problems and emphasizes its computational significance:

It also demonstrates the value of large-scale digital computers for econometrics.

Yet technical progress does not remove questions about the adequacy of empirical assumptions. Invoking Morgenstern’s Accuracy of Economic Observations, Tintner introduces a cautiously phrased reservation:

A reader of Morgenstern's Accuracy of Economic Observations may wonder if the extent of the errors of observations in Ladd's model is not too small.

This is a question about the assumed magnitude of observational errors, not a demonstrated refutation of Ladd’s model. Tintner closes by identifying multicollinearity as another frequently encountered but neglected problem. The contribution’s relevance lies in this compact research agenda: broaden the economic relationships studied, exploit computational advances, and remain attentive to the difficulties posed by empirical data.