"...it might be thought that the proper mode of constructing a positive Social Science must be by deducing it from the general laws of human nature, using the facts of history merely for verification. Such, accordingly, has been the conception of social science by many of those who have endeavoured to render it positive, particularly by the school of Bentham. M. Comte considers this as an error. We may, he says, draw from the universal laws of human nature some conclusions (though even these, we think, rather precarious) concerning the very earliest stages of human progress, of which there are either no, or very imperfect, historical records. But as society proceeds in its development, its phaenomena are determined, more and more, not by the simple tendencies of universal human nature, but by the accumulated influence of past generations over the present. The human beings themselves, on the laws of whose nature the facts of history depend, are not abstract or universal but historical human beings, already shape and made what they are, by human society. This being the case, no powers of deduction could enable any one, starting from the mere conception of the Being Man, placed in a world such as the earth may have been before the commencement of human agency, to predict and calculate the phaenomena of his development such as they have in fact proved. If the facts of history, empirically considered, had not given rise to any generalizations, a deductive study of history could never have reached higher than more or less plausible conjecture" (J.S. Mill)
Mill on Comte concerning the distinction between the social statics approach and the social dynamics approach to economic modelling. Mill rightly points out Comte's strength along the social dynamics dimension. Think of the way that 1-variable calculus can be applied to all sorts of modelling. If there is a second Or third, or nth) variable which is less variable and less significant, then perhaps it isn't too bad a move to just assume them to be constant in the model. This allows all the 1-variable results from calculus to be used right away; this can often be a useful simplification. You can later loosen this condition once the basics of a model have been established. When you think about human economic behaviour (and the model of human behaviour which underlies your model), then there's certainly an aspect which would change on an evolutionary scale - regardless of the institutions of economics surrounding them, These surely are good candidates for 'holding constant'. And in this category might be the 'wealth maximising'/pecuniary benefit' motive. Even the economic institutions which a human economic actor finds himself in could reasonably be assumed to be constant - when you think of the institution of modern markets, then there are certain elements of it which are now several scores of generations old. Clearly some economic institutional structures can change dramatically and right now (e.g. after an economic crisis, when policy responses can construct a new world order which can last a generation, and which can cause fundamental readjustments in the behaviour of economic actors). It depends on your focus of interest - for example, compare and contrast Kaletsky and El-Erian. I get the impression Kaletsky's time horizon is longer than El-Erian's - hence they can focus on different parts of financial/economic behaviour and even though they can seem contradictory, they might both be valid. El-Erian's focus is markets-based and so can afford to be more sensitive to the short term markets view.
Also, going back to Comte for a moment, given his grander time scale - charting the development of a science through multiple generations (millennia, even), then it is perhaps understandable why he prefers to focus on 'Social Dynamics'. This also chimes with his more progressivist, revolutionary political agenda. Whereas the mathematicians just then beginning to analyse corporations mathematically (Cournot) and individuals logico-mathematically (Jevons, Walras) were also justified in a concentration on 'Social Statics' - it is a decent approximation, after all, and certainly has proved productive. At all stages, the performativity of the subject should make modellers and theorists even more sensitive to the possibility of radical critique of the simplifying assumption. The quotation from Mill at the beginning echoes this possibility of performativity.
This issue of assumptions being potentially 'too simple' versus 'usefully simple' has been central to the arguments between the rational expectations/efficient markets economists and the Keynesians in recent times. John Neville Keynes gives a fair assessment of how this debate stood at the end of the nineteenth century, and one of his main historical points is that many of the great economists of the classical school were already aware of this tension and were less doctrinaire than subsequent followers made them out to be. He's also sympathetic to the belief that the simplifying assumption of economic self-interest as the best single simplification you could possibly make, given the usually rich set of real considerations real people bring in when they make real economic decisions. This seems eminently fair, to me, and doesn't invalidate the increasingly mathematical models the classical (then neo-classical) school developed in economics. Similarly, the macro-economic modeller would seem reasonable if he decided to bring certain simplifications to the range of multiple-variable macro phenomena. It is of course, in both the neoclassical and in the Keynesian (neo-Keynesian) schools, a matter precisely of which simplifying assumptions to make which can result in productive and useful models versus intellectual dead ends. In criticising which specific simplifying assumptions have been made, we should always remember that this ought not to be a criticism of the very method of simplifying a model so as to allow the application of mathematical/logical thinking. Mill and Comte can both be right on this account.
No comments:
Post a Comment