As Schumpeter (1978,p.464) notes, '[t]here is ... one sense in which economics is the most quantitative, not only of 'social' or 'moral' sciences, but of all sciences, physics not excluded.....some of the most fundamental economic facts, on the contrary [to those of natural science], already present themselves to our observation as quantities made numerical by life itself. They carry meaning only by virtue of their numerical character". It is the numerical basis of economics that yields the discipline its standing as a science and implies a role for econometrics as the tool for the process of those 'quantities made numerical by life itself.' This paper will seek to argue that, regardless of the behaviour of some of those who have chosen to practice the subject, econometrics is scientific and therefore adds to the status of economics as a quantitative science. It is not the purpose of this paper to argue that econometrics has the ability to contribute so much, that economics is elevated in popular opinion, to a scientific standing akin to the hard sciences, but merely that without econometrics, economics would be less scientific. One will offer a definition of science, a definition of econometrics and show that despite well documented limitations, the definition of econometrics fulfils the definition of science. One will note in conclusion that it is econometrics contribution to the scientific status of economics which underlined Schumpeter's initial view of econometrics as 'the colossal step forward on the road to the development of our discipline [economics]".
It would be inconsistent to argue that econometrics contributes to the scientific status of economics if one failed to note that indeed economics had a scientific nature pre econometrics. Thus as Schumpeter notes, 'we have no common credo beyond holding: first, that economics is a science'.
In defining what is scientific, one has chosen two definitions, each of which highlight what one feels are important characteristics of items scientific, important not only in the context of this papers argument, but also in any definition of science. At a broad level, the encyclopaedia Britannica defines science as 'any mode of investigation by which impartial and systematic knowledge is acquired' Mises (1956). However one should observe the following two narrower definitions concerning elements necessary for a process to be scientific. Firstly, Hendry (1980) notes that Keynes in a review on econometrics came close to asserting that 'no economic theory is ever testable' to which Hendry adds, 'in which case of course, economics itself ceases to be scientific' Hendry (1980). One takes from this that if one could provide a method for testing economic theory, that economics would 'begin' to be scientific. Mises comments that 'scientific method is supposed to mean statistical method or the demand that every proposition be 'verified' by numerous sets of statistical data relating to sufficiently comparable situations'. Secondly, Schumpeter notes the old dictum that 'correct prediction is the best or only test of whether a science has achieved its purposes' (a href=#schumpeter>Schumpeter(1978). Therefore, correct prediction within the bounds of what one can reasonably expect of an uncertain future, is a requisite for scientific status. In the following description of econometrics it will be seen that the two of the characteristics of a scientific discipline noted previously, are to be found in econometrics.
Koutsoyiannis notes that 'applied econometric research is concerned with the measurement of the parameters of economic relationships and with the prediction (by means of these parameters) of the values of economic variables' (Koutsoyiannis, 1978) and that there are five desirable properties of any econometric model:
It is noticeable that these desirable qualities of any econometric model reflect the characteristics noted earlier, as requisites for a scientific discipline; the ability to verify results and to produce accurate forecasts. Therefore, by showing that econometrics methodology is essentially scientific, one hopes to argue that this methodology then adds to the scientific status of economics.
Hendry (1980) notes that 'the three golden rules in econometrics are test, test and test' yet one must take account of Machlup when he states that 'it is said that verification is not easy to come by in the social sciences, while it is the chief business of the investigator in the natural sciences' (Machlup, 1978). Therefore, if economics had access to a procedure of verification, then this social science could then engage in the chief business of the 'hard'sciences. As described earlier in the definition of econometrics, a process of verification is allowed for, for instance it is possible for one to construct a measure of closeness of ones estimate to the true parameter. One concludes that if economics could not refer to econometrics methodology for examination of its theories, then economics would be less scientific. As Hendry (1980) concludes about the ability to consistently verify results, 'rigorously tested models, which adequately described the available data, encompassed previous findings and were derived from well based theories would greatly enhance any claim to be scientific'
As noted in the definition of what is scientific and of the desirable properties of any econometric model, one encountered a consensus between the two in the importance of any scientific methodology to provide 'satisfactory' forecasts. Econometrics models provide the ability to make predictions, the accuracy of which is constrained only by the practicing econometrician and the economic theory upon which the model is based. Remembering Schumpeter's view that a science cannot achieve its objectives without correct prediction , this would imply that without the forecasting abilities of econometrics, economics would be less scientific.
Two lines of thought that one is subjected to in arguments that the contribution of Econometrics to the scientific nature of economics is constrained and possibly non-existent, is firstly that the unsatisfactory nature of the data generating process and secondly the standing of economics itself as a social science, questioning the ability to have a scientific study in the field of human behaviour.
In response to this first approach, the faults of the data generating process and hence of econometric models have been well documented; omission of variables, faulty measurement, misspecification and inadequate sampling. However one feels that these do not undermine what one has shown to be the scientific basis of econometrics, the provisions in its methodology for verification and prediction.
Hendry (1980) concludes that 'the ease with which spurious results could be created suggested alchemy, but the scientific status of econometrics was illustrated by showing such deceptions are testable.' Knowing that such errors exist allows one to account for them. As Machlup (1980) reports, 'everyone of us could name dozens of propositions that have been disconfirmed, and this means that the verification process has done what it is supposed to do.'
With regard to the standard of forecasting held against econometrics, one feels that this is not attributable to the methodology of econometrics but to what Machlup (1978) notes as the fundamental difference between the soft and hard sciences, 'experts in the natural sciences usually do not try to do what they know they cannot do and nobody expects them to do it' whereas 'social scientists, for some strange reason, are expected to foretell the future and they feel badly if they fail.'
With regard to the second line of reasoning noted earlier, questioning the ability to have a scientific study in the field of human behaviour, one refers to what Mises (1956) calls the inferiority complex of the social sciences. It has been offered that economics can never produce consistently accurate results. In essence, the argument is that a scientific status akin to that of the natural sciences can never be achieved. Mises (1956) notes that such individuals 'are apparently ashamed of the one thing that really distinguishes social sciences from natural sciences, namely, that fact that the student of human action is himself an acting human being and therefore has at his command a source of knowledge unavailable to the student of the phenomena of nature.' Similarly for hard sciences it must be noted that 'there exists no method-oriented definition of science under which all parts and sections of physics, chemistry, biology, geology and other generally recognised natural sciences could qualify as sciences' (Mises,1953).
In conclusion one must note that when comparing the hard and the social sciences, it is common practice to refer to the controlled laboratory experiments of the natural sciences in which predictions have proved so eminently successful and then to look at the social scientists inability to apply this methodology with the resultant level of accuracy. however, one would support Machlup (1978) when he notes that 'if a comparison is made it must be between predictions of events in the real natural world and in the real social world.'
As stated in the introduction, one has sought to argue not that econometrics is capable of raising economics to a scientific status akin to that of physics, but that without econometrics, economics would be less scientific. Given ones definitions of what constitutes a science and of what constitutes econometrics, one concludes that even taking account of its limitations, econometrics contributes to economics scientific standing. One wholeheartedly agrees with Schumpeter's (1978) conclusion about this union of economic theory and statistical science, 'I expect much from it; even the most modest result, no matter how much it will be ridiculed and criticised - as it surely will be, since nothing is easier than to criticise first attempts - will be a colossal step forward on the road to the development of our discipline [economics]'.
Hendry, T. 'Econometrics-Alchemy or Science?' in Economica November 1980
Koutsoyiannis, A. 'Theory of Econometrics-An introductory Exposition of Econometric Methods.' 2nd ed. 1977 Macmillan London
Machlup, F.'Methodology of Economics And Other Social Sciences' 1978 Academic Press
Machlup, F.'Are the Social Sciences really inferior?' as edited by Machlup, F. 'Methodology of Economics And Other Social Sciences' 1978 Academic Press
Mises, L. Von'The inferiority complex of the social sciences' 1956 as edited by Machlup, F. 'Methodology of Economics And Other Social Sciences' 1978 Academic Press
Nagle, E. 'If matter could Talk' 1969 as edited by Machlup, F. 'Methodology of Economics And Other Social Sciences' 1978 Academic Press
Schumpeter, J. 'Economic Methodology' as edited by Machlup, F. 'Methodology of Economics And Other Social Sciences' 1978 Academic Press