'How can anyone express a view when he is not only ignorant, but also admits his ignorance? ....No, it's you who ought to speak, really, since you do claim to have knowledge and to be able to express it.......If I think somebody has a good idea, I'm quick to applaud it...'
'To the extent that 'con' connotes 'trickery', let us banish it from econometrics. ....; to the extent that 'con' denotes learning, let it flourish in an econometrics progressing by learning from data.'
The various valuable, but at the same time often technical, definitions of economics and econometrics, and their scientific status, have been extensively discussed, especially since the interaction of the revolution in 'macroeconomic measurement', and the 'theoretical revolution' induced by Keynes during the 1930s. Indeed, in this publication itself the topic has been covered by a number of authors. This essay hopes to contribute to the discussion in a simple way. An overview of some of the conventional arguments of economists is considered, but this essay then draws heavily on the valuable method of Socrates. It is a method that any scientist would do well to acknowledge. We shall then be able to consider, through slightly unconventional reasoning whether economics and its 'offspring', econometrics, are of scientific status.
'Science'
John Hicks supplies us with a definition of a science that he considers more widely applicable than merely to the natural sciences. Three distinguishing characteristics of the body of propositions constituting a science are laid out:
The third characteristic is named as the most important. Hicks stresses two issues: the nature of predictions and the methodology for formulating theories that are used in prediction. The only truly attainable prediction, unless the boundaries of a science have been reached (and who can tell whether they have or not?), is a weak conditional one, for example that an event will follow if there are no 'disturbances'. The methodology in question is that of 'induction', for example generalising from past observations supplying us with manifold propositions, which are then linked together to give us quite a strong prediction, or theory. This may then be tested and questioned at a later date, possibly considering further important variables.
Hicks considers a second approach to be of relevance in the field of science: the stochastic approach. Here, one concentrates on a finite sequence of experiments, drawing inferences about longer series using probability theory. If we can then identify some systematic disturbances, we may eliminate them.
Economics
How applicable is the above to economics? Hicks states that 'it is better to have some knowledge of what to expect than none at all', but this shows quite clearly how weak predictions in economics are, based on the dubious concept of ceteris paribus. Economics has always used the tying together of weak propositions, but due to a 'temporal' problem, predictions must end up weaker than in the natural sciences. Data are abundant and of a quantitative nature, but are of the past and the terms in which they are couched are those of another person, often made in a different context. Therefore, economics hangs on the edge of science:
'If a scientific theory is good, it is good now, and it would have been good a thousand years ago, if it had been available; but aspects of economic life which we need to select in order to make useful theories can be different at different times.'
The stochastic approach to economics is thus econometrics.
Economics and Econometrics
To complete Hicks' argument, econometrics faces the same 'temporal challenges' as economics, but cannot circumvent them. It is difficult to apply the stochastic methodology of a natural science to economics, as we cannot utilise our data in the same way. In the final instance, Hicks argues that we cannot apply the notion of randomness over a long period of time to economic data, as the circumstances change over time. He believes that we must turn to history and new arising events (still noting, however, that econometrics does have some role to play!).
Other views are not dissimilar to the above. Worswick (1972) paints a more depressing picture initially however, claiming that the 'gloomy science' has not led to a better, healthier economy. Interaction between fact and theory is challenged by the 'detached' nature of the facts. He is disillusioned by the econometric models, the manipulation of variables and the fudging of 'negative' results to make them 'positive': 'too much of what goes on in economic and econometric theory is of little or no relevance to serious economic science'. Econometricians use 'pretend-tools' to arrange and measure facts. Worswick is then also pessimistic, although again not to the extreme. He answers his primary question by noting that progress in economic science is possible, if very slow. Coefficients of econometric equations are however not durable, and predictions are hazy. 'Progress' must lie in gathering and processing ever more up-to-date data. There is little denying that econometrics is an undeveloped and contentious tool, as the above goes some way in showing. Hendry, an advocate of the methods and the scientific status of economics, notes a number of weighty critics, such as Brown or Leontief, who name 'glaring weakness of the data base' and 'deception by regression'.
J.A. Schumpeter wrote in the first issue of Econometrica that
'It was definitively established that economic theory involves quantity, and therefore requires the only language or method available to deal with quantitative argument as soon as it outstrips its most primitive stage.'
Both Maddala and especially Koutsoyiannis give good descriptions of what exactly is involved in econometric research, although both stress that their 'lists' are neither perfect nor final. They state that econometrics establishes the usefulness or insignificance of certain factors in economic relationships suggested by theory, and can thereby reshape theoretical economics through its analysis. Relationships are more precisely defined as regards coefficients, and explanations of parameters become more exact. A useful definition is then the following:
'Experience has shown that each of these three viewpoints, that of statistics, economic theory, and mathematics is a necessary, but not by itself sufficient condition for a real understanding of the quantitative relations in modern economic life. It is the unification of all three that is powerful, and it is this unification that constitutes econometrics.'
So, the above argument, seems to revolve around some form of 'line' between the 'exact' sciences, and the 'moral sciences', a definition made by philosophers. Both place the train of econometrics somewhere on this 'railway line', but those who advocate it suggest it is moving ever closer to making economics a science, whereas those who oppose it state that it has been diverted by some 'specification side-track', therefore making it irrelevant whether the train itself gets ever faster or not!
The Socratic Ignorance of a 'Progressive Research Strategy' Approach to Econometrics
Although Socrates proclaimed to be 'ignorant', and not able to put forward an opinion, he invariably did. Named by Apollo through an oracle as the wisest man alive, the Platonic Socrates sought to refute this by questioning anyone who purported to be wise, and find one wiser than himself. If they were not wiser, however, he questioned their ideas and proved their ignorance beyond a doubt, using a negative questioning process, the 'elenchus'. In a certain sense, he shows that, through his awareness of his ignorance, he remains wiser than those who are not. However, if this essentially negative result were the full extent of the ideal and aims of Socrates, it would not provide a fitting background for the current discussion. What is important is the following point:
'For Plato has designed Socrates' questions not only to lead Socrates' interlocutors to the recognition that they are as ignorant as he, but to guide first the conversation within the dialogue and secondly the reader to an improved understanding, and ultimately perhaps knowledge, of positive answers to the fundamental questions with which Socrates starts'
The goals of Socrates (and Plato) were often of a more fundamental nature, examining virtue, justice and courage amongst others, but there are similarities in the methodology. The topics discussed were difficult to grasp, and their nature often almost impossible to agree upon. However, Socrates did not give up, and continued to question and argue. Even if a recognition of ignorance was the only outcome, this was considered positive and a basis to perhaps finding the right way, the right formulation of the problem and the right answer to it. The science of economics may not be exact, nor may econometrics produce positive, well-founded results all the time, but it is potentially as virtuous as the Socratic method of producing some positive result from such intensive questioning.
In an abstract way, Hendry and Mizon promote an approach similar to this process. It is a process that should not be strange to econometricians.
In formulating a model, a number of steps are invaluable:
As modelling is a process, it is infinitely improbable that these criteria will produce a model that is everlasting - this is in fact an essential criticism of econometrics. However, the above approach does not claim to ascertain an 'ultimate truth', although at first this may seem to be the case. As do Hendry and Mizon, McCloskey mentions the 'context of discovery' and the 'context of justification'. These are vilified as postmodernistic tools to deal with the embarrassing fact that not all can be deduced from economic theory, and that economic scientists are convinced by 'metaphor, case study, upbringing, authority, introspection, simplicity, symmetry, fashion, theology, and politics....' McCloskey rejects the old style, modernist scientific method approach to economics. Economics as a science must go beyond this eternal and non-viable quest for prediction, control, observability. It promises knowledge free from 'doubt, metaphysics, and morals' (p.488) and utilises a methodology to test practice that is arrogant and pretentious, based on a 'dubious epistemological principle' (p. 491).
Econometrics is included in the problem, as it is inevitably drawn into the 'official rhetoric' as McCloskey puts it. But, it is also part of the 'workaday rhetoric'. There are elements involved, which are not readily explainable. 'Unreasonable' results are then 'looped' and changed regarding the hypothesis or specification, wreaking havoc with the 'objectivity' of modernists. The official theory is quickly left behind, and here we return to Hendry and Mizon, and to Socrates. The former do not align with McCloskey's views, and this is not their intention, but implicitly their analysis is complementary:
'Intrinsically, therefore, models embody the notions of design and a corresponding focus of interest, but are inherently approximations and inevitably 'false'. .....Nevertheless, models can differ radically in their usefulness, (or otherwise!) relative to the objectives for which they were constructed. The extent to which they are useful depends on their actual design, not on the origins or methods whereby that design was achieved'
The design must be dependent on the designer, and this is where subjective notions become important. Objectivity in economics is overstated, according to McCloskey, and in a certain sense is also claimed by the above authors. The former asks why such a firm line must be drawn between scientific and unscientific study. The latter seem to comply to this by giving us a way forward that relies on both feeling and introspection as well as quasi-objective checks and methods in the shape of computer programs run by 'honest' econometricians. Econometrics then plays a valuable part. Although it seems not to be able to be a pure method of science in economics, if we accept economics as different from science learnt in school then it remains a most useful tool. We are then returning to Socratic ignorance. This is of course not ignorance as such, but rather a notion that whatever we produce, we are analysing a system that is far too complex as to be fully understood through any methods. Socrates is also quick to disarm any attempt to impose false theories to further ones' own ends. A less formal analysis built on 'workaday rhetoric' and some progressive, critical system of empirical testing allow us to know how little we really do know, and also uncover 'frauds'. We must remain critical of econometric results, but nonetheless utilise them:
'If econometric argument does not persuade, it is because the field of argument is too narrow, not because the impulse towards thoughtfulness and explicitness which it embodies is wrong. The arguments need to be broadened, not merely dismissed.'
Even if the 'train' considered above has been diverted, who is to say that any particular track might not be the right way to go? The constructive humility and wisdom of the Socratic ethos are proven by their continuing salience. The authors discussed above have implicitly applied it to economics and econometrics. They have shown that economics escapes from a rigid determination of science, but fulfils criteria much wider and more important to the topics at hand. It seems that a certain amount of humility and attempting to test other people's theories as well as one's own, constantly learning from and questioning others, and one's surroundings, are a most excellent basis for any science to build on. Even if we must continually confess collective ignorance as to the possible causations of complex models, we may draw from such results and look for new ways forward. Econometrics allows honest quantification of these ideas, as beset with problems as it may be. It is as useful an exercise as Socrates questioning Thrasymachus in Polymarchus' garden.
Bibliography
Gower, Barry S. and Stokes, Michael (eds.) (1992) Socratic Questions. New Essays on the philosophy of Socrates and its significance. Routledge: London.
Hendry, David F (November 1980) 'Economics - Alchemy or Science ?' in Econometrica vol. 47. London School of Economics and Political Science: London.
Hendry, David F. and Mizon, Grayham E. (1985) Procrustean Econometrics: Or Stretching and Squeezing Data. Centre for Economic Policy Research: London.
Hicks, John (1986) 'Is Economics a Science?' in Foundations of Economics - Structures of Inquiry and Economic Theory. Baranzini, Mauro and Scazzieri, Roberto (eds) Blackwell: Oxford.
Koutsoyiannis, A. (1977) Theory of Econometrics.
McCloskey, Donald M. (June 1983) 'The Rhetoric of Economics' in Journal of Economic Literature, vol. 21.
Plato (1993) The Republic. Translated by Robin Waterfield, Oxford University Press: Oxford.
Schumpeter, Josef. (1933) 'The Common Sense of Economics'. Reprinted in The Foundations of Econometric Analysis. Hendry, David F. And Morgan, Mary S. (eds.) (1995). (Originally printed in Econometrica Vol. I, 1933) Cambridge University Press: Cambridge.
Worswick, G.D.N. (1972). 'Is Progress in Economic Science Possible?' in Economic Journal ( 82), p. 73-86, Macmillan (Journals): London.