hayek: Scientific vs Pseudo-Scientific Approachs to Macroeconomics


“If the market really achieves a utilization of more information than any participant in this market process possesses, the outcome must depend on more particular facts than the scientific observer can insert into his tentative hypothesis which is intended to explain the whole process.

There are two possible ways in which economists have endeavored at least partly to overcome this difficulty.

The first, represented by what today we call micro-economics, resignedly accepts the fact that because of this difficulty we can never achieve a full explanation, or an exact prediction of the particular outcome of a given situation, but must instead be content with what I have occasionally called a “pattern prediction” or, earlier a “prediction of the principle.” All we can achieve is to say what kinds of things will not happen and what sort of pattern the resulting situation will show, without being able to predict a particular outcome.

This kind of microeconomics attempts, by the construction of simplified models in which all the kinds of attitudes and circumstances we meet in the real life are represented, to simulate the kind of movements and changes which we observe in the real world.

Such a theory can tell us what sort of changes we can expect in the real world, the general character of which our model indicates, which reduces (not so much in scale as in the number of distinct elements), the facts with which we have to deal, to make its workings still comprehensible or surveyable.

I still believe that this is the only approach which is entitled to regard itself as scientific. Being scientific involves in this connection a frank admission of how limited our powers of prediction really are. It still does lead to some falsifiable predictions, namely what sorts of events are possible in a given situation and which are not.

It is, in this sense, an empirical theory even though it consists largely, but not entirely, of propositions which are self-evident once they are stated. Indeed, I doubt whether microeconomic theory has ever discovered any new facts. Decreasing returns, decreasing marginal productivity or marginal utility, decreasing marginal rates of substitution were of course all phenomena familiar to ordinary people even if these did not call them by that name. In fact, it is only because ordinary people knew these facts, long before economists discovered their importance, that they have always been among the determinants of how the market actually functions. What the economic theorists found out was merely the relevance of these particular facts for the decision of individuals in their interactions with other persons .. ”


” .. economists ambitious to make [microeconomics] more precise have often spoiled microeconomics by a tendency which we shall encounter .. when I pass on to the second type of approach, macro-economics. They tried to deal with our inescapable ignorance of the data required for a full explanation, the macroeconomic one, by trying measurements ..

I will at this stage make only two further comments on this. The first is that it is an erroneous belief, characteristic of bad mathematicians, that mathematics is essentially quantitative and that, therefore, to build on the great achievements of the founders of mathematical economics, men like Jevons, Walras and Pareto, one has to introduce quantitative data obtained by measurements. That was certainly not the intention of the founders of mathematical economics. They understood much better than their successors that algebraic mathematical formulae are the pre-eminent method for describing abstract patterns without assuming or possessing particular information about the specific magnitudes involved. One great mathematician has indeed described a mathematician as a maker of patterns. In this sense mathematics can be very helpful to us.

The second point which I want to make is that a particular reason which in the physical sciences make measurements of concrete magnitudes the hallmark of scientific procedure for a very definite reason do not apply to the explanation of human action. The true reason why the physical sciences must rely on measurements is that it has been recognized that things which appear alike to our senses frequently do not behave in the same manner, and that sometimes things which appear alike to us behave very differently if examined.

The physicist, to arrive at valid theories, was often compelled to substitute for the classification of different objects which our senses provide to us a different classification which was based solely on the relations of objective things toward each other.

Now this is really what measurement amounts to: a classification of objects according to the manner in which they act on other objects. But to explain human action all that is relevant is how the things appear to human beings, to acting men. This depends on whether men regard two things as the same or different kinds of things, not what they really are, unknown to them. For our purposes the results of measurements (at least so far as these are not performed by the people whose actions we want to explain) are wholly uninteresting.

The belief derived from physics that measurement is an essential foundation of all sciences is very old. There was more than 300 years ago a German philosopher named Erhard Weigel who strove to construct a universal science which he proposed to call Pantometria, based as the name says on measuring everything. Much of economics, and if I may add in parenthesis much of contemporary psychology, has indeed become Pantometria in a sense in the principle that if you don’t know what measurements mean, measure anyhow because that is what science does. The social sciences building at the University of Chicago indeed still bears since it was built 40 years ago on its outside an inscription taken from the famous physicist Lord Kelvin: “When you cannot measure, your knowledge is meager and unsatisfactory.” I will admit that that may be true, but it is certainly not scientific to insist on measurement where you don’t know what your measurements mean. There are cases where measurements are not relevant. What has done much damage to microeconomics is striving for a pseudo-exactness by imitating methods of the physical sciences which have to deal with what are fundamentally much more simple phenomena. And the assumption that it is possible to ascertain all the relevant particular facts still completely dominates the alternative methods of dealing with our constitutional ignorance, which economists have tried to overcome. This of course, is what has come to be called macro-economics as distinct from microeconomics.

The basic idea on which this approach proceeds is fairly simple and obvious. If we cannot know all the individual facts which determine individual action and thereby the economic process, we must start from the most comprehensive information which we can obtain about them, and that is the statistical figures about aggregates and averages.

Again, the model which is followed is provided by the physical sciences which, where they have to deal with true mass phenomena such as the movement of millions of molecules with which thermodynamics has to deal, where we admittedly know nothing about the movement of any individual molecule, the law of large numbers enables us to discover statistical regularities or probabilities which indeed, in this way, provide an adequate foundation for reliable predictions.

The trouble is, unfortunately, that in the disciplines which endeavor to explain the structure of society, we do not have to deal with true mass phenomena.

The events which we must take into account in any attempt to predict the outcome of particular social processes are never so numerous as to enable us to substitute ascertained probabilities for information about the individual events. As a distinguished thinker, the late Warren Weaver of the Rockefeller Foundation has pointed out, both in the biological and in the social sciences frequently we cannot rely on probabilities, or the law of large numbers, because unlike the positions which exist in the physical sciences, where statistical evidence of probabilities can be substituted for information on particular facts, we have to deal with what he calls organized complexity, where we cannot expect to find permanent constant relations between aggregates or averages.

Indeed, this intermediate field between the simple phenomena of the physical sciences, where everything can be explained by theoretical formulae which contain no more than two or three unknowns, and the instances where a large enough number of events to be able to deal with true mass phenomena to rely on probability is our subject. In the social sciences we have to deal with something which Warren Weaver called organized complexity, phenomena which are not made up of sufficiently large numbers of similar events to enable us to ascertain the probabilities for their occurrence.

In order to provide a full explanation we would have to have information about every single event which you can never possibly obtain. But while micro-theorists have resigned themselves to the consequent limitations of our powers and admit that we must be content with what I’ve called mere pattern predictions, many of the more ambitious and impatient students of these problems refuse to recognize these limitations to out possible knowledge, and possible power of prediction, and therefore also of our possible power of control.

What drives people to the pursuit of statistical research is usually the hope of discovering in this way new facts of general and not merely historical importance. But this hope is inevitably disappointed. I certainly do not wish to underrate the importance of historical information about the particular situation. I doubt, however, whether the observation and measurement of true mass phenomena has significantly improved our understanding of the market process. What we can find by this procedure, as by all observation of particular circumstances, may possibly be special relations, determined by the particular circumstances of the moment and the place, which indeed, perhaps for some time may enable us to make correct predictions. But with general laws which help to explain how at different places the course of economic affairs is determined, these quantitative relations between measurable magnitudes have precious little to do. Indeed, even the very moderate hopes which I myself had at one time concerning the usefulness of such economic forecasts based on observed statistical regularities has mostly been disappointed. The concrete course of the process of adaptation to unknown circumstances cannot be predicted. All we can predict is certain abstract features of the process, not its concrete manifestations.

It is now frequently assumed that at least the theory of money, in the nature of that subject, must be macro-theory. I can see no reason whatever for this. The cause for this belief is apparently the fact that the value of money is usually conceived as corresponding to an average of prices. But that is no more true than it is of the value of any other commodity. I do not see for instance, that our habitual use of index numbers of prices, although undoubtedly very convenient for many purposes, has in any way assisted our understanding of the effect of monetary changes, or to draw relevant conclusions, except, perhaps about the behavior of index numbers.

The interesting problems are those of the effect of monetary changes on particular prices, and about these index numbers or changes of general price levels, tell us nothing.

It seems to me more and more that the immense efforts which during the great popularity of macroeconomics over the last thirty or forty years have been devoted to it, were largely misspent, and that if we want to be useful in the future we shall have to be content to improve and spread the admittedly limited insights which micro-economics conveys.

I believe it is only microeconomics which enables us to understand the crucial functions of the market process: that it enables us to make effective use of information about thousands of facts of which nobody can have full knowledge.”

From F. A. Hayek, “Coping with Ignorance”, Imprimus, July, 1978.

This entry was posted in Keynes, Scientism. Bookmark the permalink.

5 Responses to hayek: Scientific vs Pseudo-Scientific Approachs to Macroeconomics

  1. Jonathan says:

    Fabulous. Thanks for posting this. Could someone forward it to the apparently scholarly Obama?

  2. Lynn says:

    In other words, respect facts. The devil is in the details. Glossy generalizations, while perfectly valid as generalizations, can easily mislead.

  3. Pingback: The failure of modern macroeconomics – how Hayek and McCloskey end up saying the same thing « FOO Law and Economics

  4. Pingback: Methodology of economics - Hayek and McCloskey vs. the Keynesians « FOO Law and Economics

  5. Pingback: Methodology of economics - Hayek/McCloskey vs. the Keynesians « FOO Law and Economics

Comments are closed.