Interview with Fritz Machlup

The Austrian Economics Newsletter

Summer 1980
Volume 3, Number 1

An Interview with Fritz Machlup (1902-1983)

Fritz Machlup's contributions to economics span a period of over 45 years. Born in 1902 in Wiener Neusstadt, Austria, he published his first book Die Goldkernwahrung (The Gold Exchange Standard) in 1925, the outgrowth of a dissertation written under the supervision of Ludwig von Mises at the University of Vienna.

In the 1920's, he participated in the debates over the German transfer problem. His "Austrian" analysis of The Stock Market, Credit and Capital Formation appeared in 1931 and in a revised English translation in 1940. This was soon followed by Fuhrer durch die Kirsenpolitik (Guidebook through Crises Policies) in 1934. In 1935 Professor Machlup defended traditional Austrian capital theory in his classic article "Professor Knight and the 'Period of Production' " Journal of Political Economy (Oct. 1935).

Among his other important works are International Trade and the National Income Multiplier (1943), The Political Economy of Monopoly (1952), The Economics of Seller's Competition (1952), International Payments, Debts and Gold (1964) and A History of Thought on Economic Integration (1977). His essays on methodology have been collected in two volumes, Essays on Economic Semantics (1963) and Methodology of Economics and Other SociaI Sciences(1978).

Professor Machlup has taught at Harvard University, University of Buffalo, Cornell University, The John Hopkins University, and was Professor of economics at Princeton University and New York University when this interview took place. This interview was conducted by Joseph T. Salerno and Richard M. Ebeling. His papers are now archived here.

AEN: While you were at the University of Vienna, with whom did you study?

MACHLUP: My first major teacher was Friedrich von Wieser. Wieser was quite old. In 1920-21, my first year at the University, he gave the big lecture on economic theory. Well, he never knew me. I was one of the very few people who attended the lectures regularly. I found the lectures pretty hard to take. They were terribly dull, and it was hard to keep one's mind on the subject. There was nothing wrong with the contents of his lectures, just with the delivery, He read from a manuscript. At the end of the year I felt I hadn't learned enough and I came back as an auditor for the second year, because I took my studies very seriously. So I studied for two years with Wieser. In the second year, however, I went also to the seminar of Ludwig von Mises and that was more fun. There was more vitality; much more was going on. I participated actively in the seminar and I had to do a paper. Thus, it was there that I really started to become an active economist.

AEN: You associated closely with Professor Mises during those years.

MACHLUP: Well, Mises was never very close to people before they had their doctorate. He was very formal. So the so-called Mises Privatseminar--private seminar--was open only to people who had their doctorate. I received my doctorate in December of 1923 and was invited to join the privatseminar in January '24. Now that doesn't mean that I saw very little of him in the earlier years. Indeed, I wrote my doctoral dissertation under him and that meant that I visited him in his office at the Chamber of Commerce, where he had his job, and I visited him perhaps every few weeks to discuss my problems arising from my reading and research.

AEN: What were some of the topics that were discussed at Mises' privatseminar?

MACHLUP: There the topics were, of course, left to the members though there was usually a general subject for a whole semester. The terms that I remember best were those that dealt with business cycle theory or monetary theory in general. But the one that perhaps made the biggest impression was the seminar on methodology. That was the seminar where many people who were not economists but interested in economics took part. There were the philosophers, Felix Kaufmann and Alfred Schutz, Schutz as you know, was a sociologist and philosopher. There was Fritz Schreier, the philosopher of law; and there were occasional visitors. I remember especially Herbert Feigl who as you know, later became a professor in Minnesota. There was a connection with the philosophers' circle, especially through Felix Kaufmann, who was a member not only of the Mises circle but also of the "Vienna Circle," which in Vienna was called the Schlick Circle, after Moritz von Schlick.

AEN: In those discussions on methodology, Professor Mises must already have been presenting his views on the foundations of economic science. How did the other members of the seminar react to his advocacy of an a priori deductive approach to economic theory?

MACHLUP: Well, deductive is fine, a priori is something else. Needless to say, you can deduce things from statements whether they are a priori or a posteriori. Now, Mises gave us his views on his a priori ideas and they were criticized by Kaufmann, Schutz and others, but you see it isn't really necessary to criticize these terms, because you could always construe, even in an entirely empirical science, you could construe an abstract, internally consistent system of propositions. You may call any model a priori because you can 'build' the model, according to your own specifications. A model can never be shown to be a posteriori; the builder can be influenced by what he has experienced but these experiences cannot contradict or falsify the model. A model can serve all sorts of purposes; you can say you understand the world better if you have a model, and the model will be a priori in that sense. It may contradict practically every experience that you have and yet it may be helpful in explaining observed phenomena. So one does not have to take the aprioristic position of Mises so seriously as he himself has done. You find statements of similar sorts by neopositivists. A statement that does not go back to experience is not, for that matter, worthless. Take Margenau of Yale, the physicist, who distinguishes the domain of experience, the epistemic domain, from the domain of construction. Construction is always a priori, even if you construe with some experience in mind. The domain of construction needs constructs and postulated relationships between constructs, but it is itself not the result of observation; it is a priori. So you don't have to take these distinctions so seriously as Mises himself did and as some of his followers do today.

This domain of construction is of immense use, of enormous use, but it will never allow you to predict a particular price or a particular quantity produced or a particular way of acting. It only shows us a system that is inherently consistent where all parts hang together and it is self-evident because it is the pure logic of choice which you analyze here. But it is only a skeleton without flesh and to say that this logic of choice will tell you that a steel price of twelve dollars fifty is the right price, that makes no sense.

AEN: Then the implications of a priori theory for prediction are just nil. What would you say about the attempts by economic science to predict, that is, the positivist methodology introduced by Friedman?

MACHLUP: Well, when we say "predict," we have to … maybe you want to say that all predictions are iffy, that means, if this and this, then that and that, in contradistinction to forecasts. If you are able to state all ifs, all conditions, that are necessary and sufficient for a certain result then you are not really making a prediction, because that is a tautology, in that one of these conditions would be "And there is nothing else that effects the result." In this case you have merely drawn an inference from your statement of conditions, and this is not a prediction. A prediction really should refer to the real world, where you can point to particular events or records or data.

But there is a kind of prediction, namely the prediction that something cannot take place, that means the conditions that you list--they may be quite insufficient in number, there may be many other things--but you can say the presence of these conditions is sufficient to make it impossible for that and that to follow. Now that is a negative prediction and I would say economics is most fertile in making negative predictions. For example, I can say if we print another so-and-so many billion dollars etc., etc., then it is practically impossible that we will not get a rise in prices. So negative predictions are the things that really are important.

Now let us come to forecasting. Forecasting really means that the forecaster believes that he knows all the things that are really likely to affect the outcome and that he can therefore dare to say the price level will rise by twelve per cent next year, and the gross national product, as measured in the ways we know, will not rise by more than one per cent or will fall by one per cent. I consider these forecasts absolutely foolish because these people who make the forecasts do not have the required knowledge. On this point I may refer to Hayek's "Pretense of Knowledge"--that was the title of his Nobel Prize lecture.

Now there are some forecasts that you can make. If you set upper and lower limits, you can say, from the things that you know, it is most unlikely that we will get next year a rate of price increase greater than twenty per cent and less than five per cent. Well, within that broad band, you can probably be right. But to make a forecast that something will rise by 3.6 percent--this is absolutely foolish.

AEN: On this point of predictability there has been more and more of a tendency, particularly in macroeconomics but even in microeconomics, to deal with economics in probability terms, to study economic activity as if the world was stochastic or purely probabilistic. How do you view this development?

MACHLUP: If you deal with phenomena or events that are very frequent like rainfall or something of that sort of which you have, in other words, a great number of readings or a great number of instances, what does the probability mean? When I say it will rain tomorrow with a probability of thirty per cent that doesn't mean anything for the rainfall tomorrow because if it rains you were right and, if it doesn't rain, you were also right, because you have said it will not rain with a probability of seventy per cent, so both the raining and the not raining will be correct.

Nevertheless, there is a meaning to it, namely, if you take a thousand cases where you have said it will rain tomorrow with a probability of thirty per cent and if you were right, then it should have rained, on about three hundred days and not an seven hundred days.

Now such things make very little sense in economics, because if you say that labor productivity will with a probability of 30 per cent, rise next year by 12.5 per cent, there are no thousand cases to test the reliability of your prediction. Thus, even it you say it will rise with such and such per cent probability by so-and-so much--it is meaningless: the stochastic element really serves another purpose, The stochastic element is usually used for working with something like regression analysis. There you try to regress a de-pendent variable on various variables for which you have to have information, good information, to make a regression.

Of course, then you usually have a constant and the various regression coefficients and you have a stochastic variable, an error term, that should catch not only the things that you have left out but also the things you have overspecified and so on, and this is a very different kettle of fish. The whole thing suffers then from the fact that what you put into your regression equation is a matter of your theory. You can have an indefinite number of theories and then you try out your regression coefficients for each of these specifications and you get plenty of different results and you select one. But if your theory had given you another variable, then the whole thing may have come out different again. So my trust in this kind of thing is next to zero.

AEN: During the twenties and early thirties, what did the Austrians see as distinguishing themselves from other schools of thought?

MACHLUP: Well this is very simple. As a matter of fact, I wrote about this recently and I can refer to my manuscript. Wieser himself had written an article in English in which he stated the Austrian position upon invitation of one of the journals. In that article he answered your question. He enumerated four tenets. One, marginalism. Second, diminishing marginal utility. Third, what he called the "cost theory," namely, that cost is foregone utility, or what was later called opportunity cost. And fourth, the imputation of value to complementary factors, the problem of attributing the value of a final good to complementary factors. He wrote that article in 1890.

But you are asking me how did the Austrians distinguish their own economics from others in the 1920s and here is my list. One, methodological individualism. Two, methodological subjectivism. Three, marginalism. Four, individual tastes and preferences, that is, subjective valuations of goods and services determine the demand for these goods and services. Five, opportunity costs (Wieser's law). Six--and this is something that Menger and Wieser did not accept, but in the twenties we did--the time structure of consumption and production (Böhm-Bawerk's time preference). Now these, I would say, were the things we regarded as the most important. Mises himself had a few extra tenets, at least two; one was consumer sovereignty and the other was political individualism, but these two tenets were not accepted by the majority of the Austrians, certainly not by Menger or Wieser of the older generation, but also not by some others who had achieved reputations as leading Austrian economists.

AEN: To what extent do you think these points still distinguish the Austrians from other economists today?

MACHLUP: I would say methodological individualism and subjectivism are the most important of the whole lot. Marginalism is internationally accepted. No one is an Austrian just because he is a marginalist. That people act according to their tastes and preferences and that these are basic to utility and to demand is clear. Hence, this would not be enough to make you an Austrian. Opportunity cost is so widely accepted, even in benefits-and-costs analysis. The time structure of production is perhaps not so widely accepted, and may be a mark of Austrian economics. But the one thing that really distinguishes the Austrians now is their methodological position: individualism and subjectivism. Here you can actually see differences, let's say, from the Chicago school, which does not sufficiently stress this approach, or, at least, they do not know the extent to which they are subjectivists and methodological individualists.

AEN: To what do you attribute the diminishing of interest in the Austrian approach following the 1930s?

MACHLUP: The Austrian approach has very little to say about macroeconomic problems, if you like that term (I don't like it, but it is so generally used that I can't help using it.). The Austrian approach is definitely a microeconomic approach--it goes back to individual decision-making in every respect. Of course, it does not use concrete decisions as building stones, or as empirical data of objective observation, but rather it constructs abstract models of individual decision-making. The Austrian school has little to say about total national income, total saving, total production. Of course, in Bohm-Bawerk you do find the total subsistence fund and the total of producers' goods, but not much was ever made of it, certainly not in quantitative analysis.

The problems of the 1930s were the spectacular deflations of price levels, of national income, and of foreign trade. At that time people were no longer interested--were definitely not interested--in the most efficient use of resources, since so many of the resources were not used at all. To explain opportunity cost when most foregone opportunities would be zero does not sound very meaningful. There was the phenomenon of widespread unemployment--which most of us Austrians explained by disparity between labor cost and commodity prices--but this was not enough, especially not at a time when the quantity of money had fallen so badly. So this was why all the Austrian attempts, even Hayek's Austrian business-cycle theory, did not satisfy people.

In the midst of the greatest contraction of money circulation, of incomes, of employment and so on, our recommendations were to let the economy be shaken down and have everything that's wrong liquidated, and let all the structural distortions be repaired. Well, at that moment, in 1933, this was not the right recipe. At that moment, in the United States, with unemployment of 25 per cent it was more important to put a few, more people to work and, in these circumstances, every microeconomic argument was lost. Capital theory was no longer mentioned. This sort of economic theory is quite difficult--and why study and grapple with anything that wasn't directly applicable to the immediate problem, the problem of putting people to work? Pure theory was rejected, as irrelevant.

AEN: Of course, today's problem is not depression or deflation but inflation. How do you view now the Austrian theory of money and cyclical phenomena with its emphasis on relative price distortions and misallocation of resources?

MACHLUP: Well, I think that the problem now is totally different from that of the thirties. Today it is a question of distortions, today we can see an almost general rejection of the Keynesian recipe that you can cure things by just spending more. Most economists realize now that the more you spend, the more you distort, and they have observed that every increase in spending has led sooner or later to an increase in unemployment. So I believe the return to the Austrian tenets, both in capital theory and in monetary theory, and also in business-cycle theory, is absolutely needed.

I have not a speck of doubt that unemployment is chiefly a question of the cost of labor and the price of the goods that you can sell.

AEN: The major problem in this economy, both in this country and abroad, is not just inflation per se but rather the peculiar combination of inflation with unemployment that is often called "stagflation." There seem to be many explanations of why it's occurring and what can be done about it. How do you explain the phenomenon of stagflation?

MACHLUP: This is very simple. Incidentally, I wrote an article about it. But, before I start talking about it, I want to say that I don't use the word "stagflation" because it is linguistically unsound. The root of "stagnation" needs the "n;" if you leave out the n in "stagn" and say only "stag," you refer to an animal in the woods--a male deer--and not to stagnation. And "flation" alone can be "deflation" and "inflation" and hence it again leaves out the most important thing, the "in." Hence, "Stagflation" stands neither for stagnation nor for inflation, and is thus a corrupt word coinage. But it is by no means difficult to explain the phenomenon. An inflation of the money supply can lead, and is likely to lead, to an inflation of prices and an inflation of wages. As long as the inflation of price goes ahead of the inflation of wages, the inflation of the money supply can lead to an increase in employment. But when the inflation of wage rates, wage costs, and labor cost catches up with the inflation of prices, then you don't have any creation of employment. And, when the inflation of cost goes further and faster than the inflation of prices, you will have more unemployment. Now price inflation is associated with increasing unemployment. In the last years we haven't been creating more and more unemployment, but we just haven't created enough employment. This has been due to the fact that at our rate of money and credit inflation, wage rates have been rising too fast in relation to the prices. You find my article on that in the Bance Nazionale del Lavoro Quarterly Review. It is in English and bears the title "Different Kinds of Inflation Have Different Effects on Employment."

AEN: One of Professor Hayek's main arguments in the late twenties early thirties, was that even an attempt to keep the price level stable--because you still have to increase the money supply and bring about a disparity between the "natural rate" and the "money rate" of interest--could still generate a business cycle.

MACHLUP: Yes, you would still have industrial fluctuations. No doubt, Hayek was absolutely right. An attempt to keep the price level stable with a rising population, rising labor force, and rising productivity, and consequently increasing quantities of goods implies that, to keep the price level from failing, you have to create additional money; in creating money, you create distortions. The kind of distortion depends on to whom you give the new money. The essential question is: how do you distribute that money? Hayek's business-cycle theory is based on the fact that, in distributing the new money, in our customary way of creating money, you reduce the rate of interest; or in any case, the rate of interest will be lower than the natural rate would be.

There are however other ways of spending the new money. I would be quite willing to consider devoting the modest annual increase in money to obtain things outside the sector in which the market mechanism operates, provided you are not thereby creating a structure in the economy that is unsustainable. You see, the Hayekian trade-cycle theory says you create an unsustainable structure in the economy. If you use, however, the one or two percent of new money--it won't be more than that if the labor force stops rising (as it will be soon) and if productivity rises only quite slowly (as it now does)--so, if the increase in the money supply is only one or two per cent per year, I think you probably can find such uses for it as will not produce unsustainable structures. I can imagine that reductions of income taxes or slight increases in social welfare services (which may be politically unavoidable in any way) need not distort the economic structure. So I think there is a possibility of spending that modest increase in money that is needed for keeping price levels from falling, in ways not resulting in any deepening of the capital structure and not creating industries that are unsustainable.

AEN: Given the subsequent problems that occurred once the banking system started collapsing in the United States in the early thirties, do you feel that Professor Hayek's analysis of what brought on the depression was correct?

MACHLUP: Yes, I do, absolutely. I have always accepted it. The stable price level during the early twenties was the beginning of our downfall; this was a period of particularly fast growth of total output and, to keep price levels from falling, one had to create a good deal of money which was all fed into investment and this high rate of investment turned out to be unsustainable in the long run.

AEN: I take it then that you feel that the Chicago approach to analyzing monetary phenomena--with its focus on anticipated changes in the price level--is missing an essential ingredient of monetary problems?

MACHLUP: Absolutely. I don't know why a man as intelligent as Milton Friedman doesn't see that point. It is quite true that his ideas on price level and expectations of rising price levels explain much -- but not everything. There are many things that are very important but cannot be explained by that approach alone. I don't understand why he doesn't give more emphasis to relative prices, relative costs, even in an inflationary period.

AEN: You mention that one of the subjects that was sort of put by the side during the 1930s was capital theory. You had participated in the capital controversies of the 1930s along with Professor Hayek and Professor Knight. How do you evaluate that debate and your role in it?

MACHLUP: I think it's very valuable and I think we should go back to it. It's too bad that Hayek's Pure Theory of Capital is no longer being read and that nobody tries now to work on it, after all, it is not the end. The relation of money capital to pure capital has still to be worked in. It's terribly important, but there is still not sufficient interest in it. There are so many things that the few people in the Austrian school have to do; they just can't do everything. Many years ago, I wrote a review of Schneider's three or four volumes of economics and found that this, the most complete, the most didactic textbook in the world in economics, did not have a chapter on capital.

AEN: Professor Machlup, what is your opinion of the newly emergent monetary approach to the balance of payments and exchange-rate determination, especially as propounded by Laffer and Mundell?

MACHLUP: I wouldn't have been that critical of the new approach if you hadn't mentioned Letter and Mundell. I think Laffer's version is so preposterous that it's almost ridiculous. But I concede that there is always a kernel of truth in everything that any economist says. Now the monetary element in explaining the balance of payments has always been present. No one in his right mind has neglected the monetary approach, changes in the quantity of money or in the availability of bank credit, etc., ...you could never explain an import surplus, a more than temporary import surplus, if you did not refer to a monetary policy of offsetting, or to a constant increase in the quantity of bank credit; so the monetary issue was always there, even if some people didn't stress it.

But what some people did with the monetary approach was to concoct some terribly--well, how should I say?--esoteric formulations, which are quite unhelpful. To give you an example, no explanation of the balance of payments is of much help if it does not immediately distinguish goods and services, or perhaps, separately from the personal services, capital services (because they are of a different type). We must distinguish unilateral payments, spontaneous and autonomous capital movements, long-term and short-term, and plain money movements. Now to put everything into one pocket and say, "I look only at financial items"--you don't explain anything.

It is clear that if you have only non-money items and money items, you have a tautology, that the money items determine the non-money items, or vice versa. If you assume that money is the independent variable, then obviously, the money variable determines the sum of the non-money variables. That's it. But this is not an explanation. First of all, it isn't an explanation for those who discover that there is a dependent part in the money variable; it is no explanation for those who want to have the trade balance and the capital balance separated.

Incidentally, the portfolio theory does a very similar bracketing of items which ought to be distinguished. If you don't distinguish, let's say, foreign and domestic goods and services, foreign and domestic securities, foreign and domestic direct investments, foreign and domestic liquid balances, and so on, you don't really explain anything. Now I would say, in its extreme versions the monetary approach to the balance of payments is not useful.

AEN: Thank you very much, Professor Machlup.