Keynes on the methodology of econometrics
from Lars Syll
There is first of all the central question of methodology — the logic of applying the method of multiple correlation to unanalysed economic material, which we know to be non-homogeneous through time. If we are dealing with the action of numerically measurable, independent forces, adequately analysed so that we were dealing with independent atomic factors and between them completely comprehensive, acting with fluctuating relative strength on material constant and homogeneous through time, we might be able to use the method of multiple correlation with some confidence for disentangling the laws of their action … In fact we know that every one of these conditions is far from being satisfied by the economic material under investigation.
Letter from John Maynard Keynes to Royall Tyler (1938)
Leave a comment
—– look inside —– $5.94 / $20.00
—– look inside —– $4.90 / $8.00
—– look inside —– $15.99
—– look inside —– $5.99 / 12.99
—– look inside —– $5.93 / $12.99
—– look inside —– $4.97 / $9.90
—— Ugarteche, Puyana and Madi ——
Gerson Lima / Maria Alejandra Madi
Edward Fullbrook and Jamie Morgan
————— Michael Hudson ————–
Maria Alejandra Madi / Jack Reardon
————- Edward Fullbrook ————-
—————— Steve Keen —————–
————— Richard Smith —————
————– Gustavo Marques————
– Victor Beker and Beniamino Moro –
————– Lars Pålsson Syll ————-
—————– Stuart Birks —————-
Edward Fullbrook and Jamie Morgan
———— Armando Ochangco ———-
Shimshon Bichler / Jonathan Nitzan
————— Mauro Gallegati ————–
————— Herman Daly —————-
————— Asad Zaman —————
—————– C. T. Kurien —————
————— Robert Locke —————-
Real-World Economics Review Blog 
Keynes was writing before the computer was in use. Regression analysis had to be conducted laboriously on a mechanical calculator. That severely limited specifications that could be handled. Linearity and constancy of coefficients were unavoidable assumptions. Now, in contrast, the only limitation on a specification is the imagination of the theorist. And the only limitation on testing it is the quantity of data.. “atomistic factors” do not have to be independent; interaction, effects, multiplicative relations or other non-linearities can be dealt with. If we don’t think relations are “constant and homogenous through time”, there is a battery of tests to ascertain whether they are. If the nature of change can be specified, regressions can deal with time-varying coefficients or breaks in relationships. People even use the Dirac equation to model sudden jumps in a relationship. Keynes had no means of forseeing the potentialities of computing, of course, nor could he have guessed at the advances in statistical methods it would facilitate.
In what other discipline would someone quote remarks that are 82 years old about the methods of research? Which empirical disciplines have not been revolutionised by the development of the computer and the extensive accretion of usable data? .None of the points Keynes made in the quote above have any current validity.
It used to be a problem in economics that if relationships of any complexity were postulated there was no means to test them empirically. Now that is much less the case, the current problem is that people don’t welcome empirical testing of their favoured theoretical propositions and cling to them regardless of evidence.
Gerald, you are missing the premise of Keynes’ comment: that he is discussing the logic of the methods of multiple correlation being used, and saying – much as he said in the General Theory – that the correlations are producing wrong answers by using the wrong logic. They are still correlating quantities in apparently complete ignorance of Cartesian coordinates being needed to define direction of action and geodesic context, i.e. complex number and non-Euclidian geometry. Even worse, they are still using logic without defining their terms, either deliberately – for rhetorical or criminal reasons – or because they don’t understand what has become much more obvious from writing computer programs at system level: that terms have to be defined before one can calculate reliably with them, and that even minor miscalculations in the course of a calculation can have cumulative effects when carried forward. Cf. the proverbial story of storms being caused by the flap of a butterfly’s wings.
I recommend to you the distinction made by library scientist S R Ranganathan between a generalisation and an abstraction: the one leaving out apparently irrelevant data and the other subtracting the dimension in which the data appeared, like showing the plan of a house but not the elevations needed to see how it had been worked out in practice. The plan (like the logical foundations) constrains what is possible in the resultant elevations.