'Economics is in time, and therefore in history in a way that science is not.'
- J.R. Hicks
'All too often in economics the choice is between being roughly accurate or precisely wrong.'
- Kamarck
The social laboratory that is the world and the difficulty of its analysis has called into question the position of economics as a science or not. This essay will attempt to deal with this question, and will show that economics perhaps tends towards the epistemic dimension of science i.e. the collection of knowledge for its intrinsic sake, but that econometrics has a role in drawing economics into a more non-epistemic or applied role. This will be done by first considering the question of what a science is and then whether economics fulfils the aims of science i.e. what scientific status does it have? Then, econometrics will be considered in the same light, and the analogy of meteorological practices will be drawn in. Finally it will be concluded that economics is firmly a moral science and also that econometrics can only reduce the historical and epistemic emphasis of economics, and thus push it closer to applied science.
What is a Science?
Hicks defined a science to be a body of propositions which have three characteristics: these propositions are about real phenomena; they are general propositions about classes of phenomena; and they are propositions on which it is possible to base predictions - which command some degree of belief.
Hicks goes on to consider predictions. There are two types: unconditional predictions - but scientists don't claim to predict the future - and conditional predictions. Conditional predictions are the normal type of scientific prediction of which there is a strong form, in which given the stated conditions the event will follow, and a weak form. The weak form states that the event will follow if there is no disturbance. Thus the theory cannot be categorically refuted by what happens. We are thus in the land of 'ceteris paribus' and the area of economics.
Economics: A Science?
The science of economics has a twofold aim, namely to accurately describe the economic events in both our social world and past social orders and second, to furnish scientific explanations of these events. Hempel distinguished between the pragmatic usefulness of science and its use in satisfying our intellectual curiosity. The latter he termed 'the epistemic dimension of science.' Applied science such as medicine falls within the non-epistemic dimension. According to Boylan and O'Gorman:
'Description, theory and observable causal webs are domiciled in the epistemic domain of science, and explanation is housed in the non-epistemic domain.'
Thus one can picture a 'scientific' continuum, and at one end of the scale pure epistemic science, and at the other end is the non-epistemic or applied sciences. One might now ask where economics lies on this continuum?
The relation of economics to time and its historical dimension is one reason why economics is, in the opinion of the author, to be nearer the epistemic end of the spectrum. Economic 'experiments' are merely past events and so while it can satisfy our intellectual curiosity, the pragmatic usefulness of theories derived are questionable. Does econometrics then push economics nearer the applied end, that is out of the purely historical dimension into the practical application and ability to test theory?
A role for Econometrics?
'A scientific theory is a body of reasoning with an empirical basis.' J.R. Hicks
Econometric analysis must add to the empirical basis of economics and therefore can only add to its scientific status. The British philosopher of science Popper argued that all scientific laws are inductions i.e. generalisations from observations. This basis on observations implies that no induction can ever provide a complete proof. An econometric test provides no proof of a theory as it is an induction. It does, though, provide the tools with which to formulate a body of propositions.
Hendry, Leamer and others have seriously questioned the basis of econometrics. The presence of 'priors' in the construction of econometric models is one such criticism. If one has prior expectations about the results of a study, these will be reflected in the way one specifies one's model. Econometrics may provide a false sense of certainty, purely because it is numerically based. Morganstern exposes sources of errors in economic observations:
(i) economic data not usually secured from planned experiments: they are by-products.
(ii) usually gathered not by highly trained economic observers but by clerical personnel or people employed ad hoc.
(iii) the data collected are often defined by legal rather than economic categories.
Keynes sees economics as a moral science: '...it deals with introspection and with values.' McCloskey brings this further in saying that economist's criticisms are in the form of literary criticism. That is, they criticise without passing judgement on how good or bad their work is. He says the use of mathematical models does not make this any less so - they are 'non-ornamental metaphors.'
Yet in any science inaccuracies can be shrouded, the fact that there is morals in the subject does not mean that it cannot be scientific. Indeed econometrics, for all its problems, must expose some of this rhetoric by nature of its hypothesis testing framework. For Friedman, what matters in an economic hypothesis is only successful prediction, but science demands a logical reasoning as to why. Econometrics can help test if these theories are correct and can also help in prediction. To return to Popper, a critical attitude is the key. Falsification is the prime objective of the scientific economic theorist. Testability is a criterion for distinguishing science and non-science. From this we may conclude that econometrics can only push economics along the continuum towards the non-epistemic domain.
The econometrician has much in common with the meteorologist. In meteorology, the system to be coped with is in constant flux. Even if one could measure accurately the pressure patterns and frontal surface patterns over the earth at any one point in time, change can be significant within a short period of time. There is also a near to infinite set of endogenous and exogenous factors - in meteorology the ultimate force is the sun, while in economics it is the elusive 'invisible hand' of millions of decision makers. The variability of probable outcomes grows exponentially with time, thus forecasts can only be in the short term. Marshak notes that the historical nature of the data also ties the two disciplines. According to Kamarck, though, compared to the record of meteorology in predicting weather and climactic change, the accomplishments of economics are not unfavourable. The major difference between the two is probably that measurement techniques are more accurate and unbiased in meteorology, but the variables are much more volatile in most cases than economic data. Would we then place meteorology nearer the epistemic domain than economics? Perhaps not, as its practical emphasis would make it more non-epistemic than economics.
Conclusion
Econometrics is not about hiding behind a veil of numerical complexity. Yes it is a moral science, but econometrics combined with the Popperian critical view can help see through the unsubstantiated rhetoric of publication-loving economists. Economic predictions are weak predictions, and more often than not are short-term, and though predictions relate to the future, the records of the past are our phenomena. Thus economics soaks in a bath of history and this led Hicks to comment '[economics] is on the edge of science and on the edge of history.' To be confined to history is to be confined to the epistemic, but theory must use the pattern of the past to push into the applied, and econometrics can only help validate or refute these theories. Thus, while retaining the epistemic to 'furnish accurate theory-laden descriptions of the observable economic world', the applied aspects of economics are enhanced by econometrics and so the Hicksian 'body of propositions' can have predictions based on them. Thus econometrics pushes economics from the history books and out into the scientific world.
Bibliography
Boylan, T.A. & O'Gorman, P.F. (1995): Beyond Rhetoric and Realism in Economics Routledge, London and New York
Hicks, J.R. (1986): Is Economics A Science? in Baranzini and Scazzieri Foundations of Economics, Blackwell
Kamarck, A.M. (1983): Economics and the Real World, Basil Blackwell
Marschak, J. (1994) On Econometric Tools in Hausman, D.M. The Philosophy of Economics, Cambridge University Press
McCloskey, D. (1985) The Loss Function has been Mislaid: The Rhetoric of Significance Tests, American Economic Review, 75 (May) 201-205