Keynes, Einstein And Scientific Revolution

by James K. Galbraith

SUBJECT: KEYNES AND EINSTEIN: TWO GENERAL THEORIES. This article is forthcoming in the Winter 1994 issue of THE AMERICAN PROSPECT. It is Copyright 1994, by The American Prospect. It is transmitted over the Internet by permission, as a matter of possible interest. Permission is also granted to make copies for classroom and other non-commercial use, and to forward electronically to other users, provided that this notice accompanies all copies. .

One of the most intriguing and little-noted facts about John Maynard Keynes' masterwork, The General Theory of Employment Interest and Money, concerns the first three words of its title. These are evidently cribbed from Albert Einstein.* Alone that would be only a curiosum; but there is more. The parallels between Keynes' economics and relativity theory are deep enough, and evidently intentional enough, to provide a useful framework for thinking about what Keynes meant to do with his scientific revolution.

Keynes and Einstein had met. Keynes traveled to Berlin in 1926 to lecture; Einstein attended. Keynes' impressions were not published until 1972:

Wordsworth, who had not seen him, wrote of Newton's statue: `The marble index of a mind for ever Voyaging through strange seas of Thought, alone.' I, who have seen Einstein, have to record something apparently " perhaps not really different " that he is 'a naughty boy', a naughty Jew-boy, covered with ink, pulling a long nose as the world kicks his bottom; a sweet imp, pure and giggling. (Collected Writings, Vol. X, p. 382.)

A second reference appears in The New Statesman and Nation of 21 October 1933. For this issue, Keynes prepared a short commentary to accompany a sketch by the artist Low, of Albert Einstein. The playful imagery is now gone, Keynes was by this time becoming a champion of Jewish refugees. Now, to Keynes' eye, Low's drawing evokes an Einstein under attack. Keynes quotes Einstein in the German:

Assuredly you too, dear reader, made acquaintance as boy or girl with the proud edifice of Euclid's geometry " thus begins the 'Essay on the Special and General Theory of Relativity' " Assuredly by force of this bit of your past you would beat with contempt anyone who casts doubts on even the most out of the way fragment of any of its propositions." It is so indeed. The boys, who cannot grow up to adult human nature, are beating the prophets of the ancient race " Marx, Freud, Einstein..." (Collected Writings, Vol. XXVIII, p. 21)

The first extant complete table of contents of Keynes' next book, then titled simply The General Theory of Employment, was found in a bundle of papers dated December 1933. (Collected Writings, Vol. XIII, p. 421). In the first proofs of that book there is a sentence, deleted from later proofs, that occurs exactly at the point where Keynes declares that the classical theory cannot be applied to the problem of unemployment, and just before this passage:

The classical theorists resemble Euclidean geometers in a non-Euclidean world who, discovering that in experience straight lines apparently parallel often meet, rebuke the lines for not keeping straight " as the only remedy for the unfortunate collisions which are occurring. Yet, in truth, there is no remedy except to throw over the axiom of parallels and to work out a non-Euclidean geometry. Something similar is required in economics.

The deleted sentence reads, "We require, therefore, to work out a more general theory than the classical theory." (Collected Writings, Vol XIV, p. 366)

Mark Twain writes somewhere that "some circumstantial evidence is very strong, as when you find a trout in the milk."

But what, if anything, does it mean?


Albert Einstein came of age in a world where the classical physics of Sir Isaac Newton still reigned. Two features of Newton's worldview are pertinent to understanding the classical economics that Keynes meant to attack.

The first is that Newton's physics presupposes an absolute separation of space and time. Space is Euclidean: a three- dimensional void stretching infinitely in all directions. The position of any particle in space can be defined, by means of a system of coordinates, with respect to any observer or any fixed reference point. Motion is the displacement of the particle from one position to another. Velocity is motion, divided by the number of ticks on a clock that it takes for the motion to occur. The clock that is used to measure velocity lies, in a conceptual sense, outside the universe itself. In other words, all observers of an event, provided they were equipped with accurate timepieces, no matter where they might be, would always agree on the exact time that the event occurred. Newton imagined time as an absolutely regular phenomenon that could not depend on the location of the clock or be affected by its movement or any other physical force.

The second feature is reductionism: Newton's universe was neither more nor less than the sum of its component particles. Gravity in Newton's system is the basic force exerted by one massive body on any other. Gravity produces the acceleration of a particle in space, according to the position and mass of all other particles in the universe that exert gravitational force. And, in Newton's view, this interaction of each particle on every other is all there is. Once you knew the position, mass, direction and velocity of every particle in the universe, you would not need to know anything else. Every future event would be fully determined by the laws of motion.

Without going into great detail, it is possible to trace out the role of each of the above features in the Classical economics of Keynes' time and -- in modern neo-Classical economics. The analog of space is the market. Look at any supply and demand diagram. The graph itself is a two-dimensional space. Every point on the graph is a position defined uniquely with respect to the origin. The relationships between variables are presented as forces in this space: in the labor market, demand aligns wages and employment in a downward sloping relation; supply aligns them along an upward slope. If two curves cross in that space, their point of intersection is an equilibrium position, where the forces balance and the market clears.

The analog of Newtonian time, in the classical economics, is money. Just as time is absolutely separated from space, money is absolutely separated from the market. Prices and wages may be measured in money terms, but this is only a convenience. The prices that count are relative prices -- prices in relation to the prices of other goods. The wages measured in a proper labor market are real wages-- an hour's work in terms of the commodities that an hour of work can purchase. Like time, money is an invariable standard. And just as it does not matter whether one measures time in seconds or in hours, or from Andromeda or Cassiopeia, it does not matter whether one measures prices in dollars or dimes, in pesos or yen, or in dollars of 1958 or dollars of 1993. The quantity of money has no effect on the equilibrium of the market; nothing real depends on money in any important way.

The reductionism of Newton's system is equally fundamental to the classical economics -- and remains so today. Economists are taught that societies, like Newton's universe, are nothing more than the sum of their individual components. Macroeconomic expressions, though they purport to describe the behavior of society as a whole, are only a shorthand for the mass of individual human actions. In principle, therefore, the best macroeconomics would be built strictly and rigidly from the theory of individual behavior, or micro-foundations. If there are operational difficulties with this, they must lie mainly in the difficulty of acquiring all the information that is necessary about all of the individuals whose preferences and behavior must be considered. Fundamental difficulties of theory do not arise.


By the time Keynes came along, the Newtonian view of the physical universe had crumbled. Einstein's theories of relativity had done it in.

The absolute separation of time and space collapsed with Einstein's introduction of a new universal constant, the speed of light. If light traveled through empty space, everywhere and always and irrespective of the direction and velocity of the observer (as Einstein argued and experiments have confirmed) at the same identical speed (186,000 miles per second), then the absolute simultaneity of two or more very distant events could no longer be defined. Clocks in different places will record these events at different times, and none is more correct than any other. Moreover, Einstein showed that space and time were interrelated " time itself advances more slowly near massive bodies than it does in empty space.

Furthermore, this newly unified concept, space-time, also destroyed the Euclidean concept of emptiness extending forever in all directions. Space-time is curved, and Einstein's relativity is the extension of the Riemannian geometry of curved spaces to the physical universe. Near any massive body the shortest distance between two points curves around it (as does the path of a ray of light, a point verified by experiment). For this reason, parallel lines may meet if extended far enough. (Keynes' reference to overthrowing Euclid's axiom of parallels is an unmistakeable allusion to this feature of Einstein's theory.)

But if space-time is curved by the presence of matter, then the shortest distance between two points is no longer defined independently of the distribution of matter in space. And then the system is no longer reducible to its elements; you can no longer get to the whole merely by adding up the parts. The universe is, instead, more easily and more correctly understood by looking at the whole and placing the parts within it. The whole can impose rules on the parts: in a famous phrase "space tells matter how to move; matter tells space how to curve."


When Keynes wrote his General Theory, he had in his gunsights -- I shall argue -- both Newton's reductionism and his space-time dichotomy, as both were reflected in the Classical economics. First, Keynes sought to disestablish the absolute space of classical markets, and to end the separation of markets from the world of money. Keynes characterized his theory as a monetary theory of production, giving lectures on this subject in the fall of 1933 as the General Theory of Employment (the preliminary title) was taking shape. Keynes contrasted monetary-production economics with what he called the real-exchange economics of the classical view. In so doing, he broke down the traditional non-monetary concepts of a labor market and a capital market, suffusing both of these subjects with ideas -- effective demand and liquidity preference -- that cannot be conceived of properly except in monetary terms.

Monetary-production is Keynes' space-time: the marriage of conceptual domains previously held to be distinct. In the classical theory of the labor market, for example, Keynes found a first postulatewhich held that the demand for labor would rise when real wages fell, and vice versa. This was a consequence of the principle of diminishing returns, and an idea that Keynes did not choose (at that time) to dispute. But the idea that demand for labor rises as wages fall cannot, by itself, establish either the actual level of employment or the value of the real wage.

The Classics had closed their model with a second postulate, which held that work-time offered would increase when real wages rose. This second postulate was precisely that part of the classical vision that reduced unemployment to a matter of individual decision. If a person was apparently unemployed, it should always be possible for him or her either to find work, or else to eliminate the desire to work and therefore the appearance of unemployment, by sufficiently cutting the wage.

For Keynes, this second postulate, the upward-sloping supply curve of labor, was akin to the axiom of parallels in Euclid's geometry. It should likewise be rejected. In doing so, Keynes threw over not only the supply curve of labor but also the whole idea of a self-contained labor market in the normal supply-and-demand sense, a construct in which real wages and employment could be modeled together as though one depended directly on the other. In its place, Keynes offered the now-familiar, but then revolutionary, idea that employment was determined by effective monetary demand for output. Since there was no reason why the total demand for output would necessarily correspond to high or full employment, involuntary unemployment in the strict sensewould now be possible in economics.

But what would determine effective demand? Such demand could be divided into two major elements: the consumption demand of households, and the investment demands of business. Here Keynes' reasoning led him to dismantle the second metaphorical classical, supply-and-demand market, namely the capital market. In the classical theory, the supply of and demand for capital jointly determined a quantity, namely the total volume of savings and investment, and a price, namely the rate of interest. Investment was demanded by firms, with more being demanded at low interest rates than at high. Savings was supplied by individuals, with more being supplied at high interest than at low. Thus a market for capital determined how much of current output would be consumed, and how much saved and invested. This market, it should be noted, operated wholly apart from the determination of output. Investment and savings did not affect employment and output, but only the division of output between current consumption and capital formation.

Here again, Keynes' attacked the supply curve. Savings, he proposed, had nothing to do with the interest rate. They were, instead, merely the leftovers after consumption out of income. Investment, he believed, did depend on the interest rate. But a curve of investment demand alone could not determine both the volume of investment and the rate of interest. Keynes now needed an independent theory of the interest rate.

To get an interest rate, Keynes brought in a new market, up to that point largely ignored in economics: the market for debt instruments and, in particular, for money. Interest, he proposed, was not a reward for saving, but the reward for giving up the liquidity, the easy access to immediate purchasing power, that could be had by holding money. As anyone who has bought a bond or a CD knows, the longer the term (the greater the liquidity foregone), the higher the rate of interest. Keynes argued that the interest rate thus reconciled the supply of liquidity (quantity of money) with the demand for it. And in Keynes' new sequence, the interest rate determined in the money market in turn determined the volume of investment.

To complete his theory, Keynes tied these elements together. The market for money determined interest. Interest (together with the state of business confidence) determined investment. Investment, alongside consumption, determined effective demand for output. Demand for output determined output and employment. Consumption out of incomes determined savings. Employment determined the real wage.

In this world, a change in monetary policy, such as a cut in interest rates leading to an increase in bank credit, now had fundamental real consequences. The classical dichotomy, in economics as in physics, had been broken. And with the deconstruction of labor and capital markets, the reductionist idea of microfoundations had also necessarily to be abandoned. Workers, Keynes pointed out, bargain for money wages, not real wages. The act of dropping money wages would generate feedbacks through previously unrecognized -- monetary -- channels in the system. In particular, prices would fall, and real wages (the ratio of wages to prices) would therefore not necessarily change. Falling prices might, however, depress business profit expectations and so cut into demand for investment. This would actually reduce the demand for workers and prevent total employment from rising. The system interacts with itself, and an equilibrium of full employment cannot be achieved within the labor market. Economic space-time is curved.


In the long run, Keynes did not achieve what he hoped. His parallel to Einstein went virtually unnoticed. Lawrence Klein, writing an early interpretation in his 1947 work, The Keynesian Revolution, did emphasize Keynes' attacks on microeconomic supply curves. But in the United States the prevailing view became that of Paul Samuelson, who transposed Keynes' unemployment theory into the proposition that wages are "sticky." In this interpretation, unemployment occurs simply because labor markets, characterized by supply and demand curves just as in the good old days, do not clear. What Samuelson did -- and he is, I think, too good a student of physics not to have known it -- was to push the daemon of Keynesian relativity back into its box. And modern American Keynesians, even down to the New Keynesians presently in fashion around Harvard, MIT, Princeton and the Council of Economic Advisers, are Newtonian and Samuelsonian to the core (though with a touch of Von Neumann thrown in nowadays). As such, they have denied themselves the high ground of principle Keynes sought to claim, conceding an enormous advantage to classical free market conservatives on every important policy matter.

Too bad. For one cannot say, as one can with Newtonian physics, that Newtonian economics is good enough for practical situations. The scale of the whole, in the economic case, is not that of the universe or the solar system; it is merely that of the nation state or the global region. Interdependence afflicts us all. The global irrationality of wage cutting, American budget balancing, zero-inflation Federal Reserve targets and Third World austerity programs is an everyday occurrence. The failure of Keynesian macroeconomics to establish full theoretical independence from the classical labor market and the classical neutrality of money means that we are, in effect, now denied fair discussion of Keynesian solutions to policy problems. The end result is that we cannot cope now, any more than could the classics in their day, with stagnation and involuntary unemployment.

***************** FOOTNOTE TO PAGE ONE

* This point was first made to me in private conversation by Robert Skidelsky. The economists Ching-Yao Hsieh and Meng-Hua Ye devote a short chapter in their excellent book to relativity and economics, stating it would not be an exaggeration to assert that Keynes's theory of involuntaryunemployment was inspired by Einstein. They do not, however, explore the parallelism between space-time and monetary production. Nor does Skidelsky, who discusses the Einstein link in his second volume on Keynes. Philip Mirowski, whose 1989 book More Heat that Light is a fundamental treatment of the relationship between physics and economics, also leaves out the Keynes-Einstein tale.



Ching-Yao Hsieh and Meng-Hua Ye, Economics, Philosophy and Physics, Armonk, ME Sharpe 1991.

Lawrence Klein, The Keynesian Revolution, New York: MacMillan, 1947.

Axel Leijonhufvud, On Keynesian Economics and the Economics of Keynes, New York: Oxford University Press, 1968. Philip Mirowski, More Heat Than Light: Economics as Social Physics, Physics as Nature's Economics, Cambridge: Cambridge University Press, 1989.

Robert Skidelsky, John Maynard Keynes: The Economist as Saviour, 1920-1937, London: MacMillan, 1993.


James K. Galbraith is co-author with William Darity, jr. of Macroeconomics, a new textbook from Houghton-Mifflin. This article will be published in the Winter 1994 issue of The American Prospect. Comments are welcome, and may be sent directly to the author at GALBRAITH.JAMES@UTXVM.CC.UTEXAS.EDU


Date: Sat, 11 Dec 1993 13:45:29 EST



Subject: Sraffa and Long-Run Equilibrium

Message-ID: <>

This is from PKT's archives by Robert Vienneau 8.0 Theories of Distribution

At least three theories of distribution are consistent with long-run equilibrium:

1) Neo-Marxian. The wage is externally specified. One possibilty is that Malthusian population pressure reduces it to a physically specified subsistence level when it is above the equilibrium value, while it cannot fall below without workers dying. The assumption that the wage is socially determined is more in keeping with Marx. Class struggle may raise it for a while. The resulting pressure on profits will induce technological innovation eventually bringing it down. A complex story arises explaining the interaction of business cycles and growth. Marx was wrong in thinking that technological change logically must be capital-using and labor-saving. Increased productivity in the iron or tin industries can result in a more capitalistic technique being adopted by the economy as a whole. The issues brought up here are still a matter of debate.

2) Neo-Ricardian Keynesian. The rate of interest is externally determined by the monetary authority. A Keynesian macrotheory shows how the level of income varies with the resulting investment. Despite a famous aside by Piero Sraffa, I do not feel comfortable in ascribing this value to him.

3) Post-Keynesian. The rate of profit is determined by the interactions of investment, profit, and savings. Given fixed savings rates, the distribution of income is determined by the rate of growth, which, in turn, is determined by investment. The rate of growth that firms desire is a function of "animal spirits" and the expected rate of profit. Two equations determine the two unknown quantities the rate of growth and the rate of profit. Prices fall out in the wash. Investment is not constrained by savings, but savings adjusts to the investment decisions of capitalists. Consumption per head is known given the technique and the rate of growth.

8.1 Neoclassical Theory

Is there a consistent long-run Neoclassical theory of distribution? This is a debated point. Whatever it may be, it cannot be equated with marginalism. Marginal techniques are the common possession of all theoretical economists.

The theory of intertemporal equilibrium is commonly thought to provide the Neoclassical theory. Consider an economy in equilibrium at a single point of time in the sense that all markets clear. All markets are barter and all transactions planned for forever are contracted for at this single moment. In other words, all possible future markets exist. The initial endowment is given and individuals must make decisions trading leisure for goods at various times and consumption decisions across time. Since markets are assumed to clear, the entire future course of the economy is pre-coordinated.

Overlapping generation models are another variant. Instead of the future being precoordinated, it is assumed to be perfectly known. All markets are then assumed to be spot markets operated at various points of time. In either case, the ultimate determates of equilibrium are

1) Initial endowments 2) Tastes 3) Technology.

Now imagine that the long-run condition is imposed that the rate of interest be the same in all lines of production actually in use. Furthermore, relaitve prices are stationary. Then, in general the model is overdetermined. So the method of intertemporal equilibrium is, at best, a short run model.

Is this a sufficient basis for economic theory? I do not think so. A model of capitalism that does not show competition tending to level out the rate of return by the ebb and flow of finance just seems to be a contradiction in terms. On the other hand, a lot of economists have seemed content in the last two decades to articulate this model. Perhaps they are not aware of these limitations.

9.0 Conclusions and Open Questions

9.1 The Rebirth of Classical Theory

The implications of the Sraffian approach are far reaching. First, they have vindicated the approach of Classical economists and Marx. Here is a theory in which prices of production attract market prices. Inputs become parameters determined by the model, not externally given endowments. The ability of the economy to generate a surplus provides an alternate basis for economic theory. Economics is not about the allocation of scarce resources among competing ends.

The classicals had the insight that only objective conditions of production matter. Subjective tastes can be ignored. Of course, the classicals made lots of mistakes that need to be cleared up in developing this theory. For example, the labor theory of value goes by the wayside. Whether or not this is an historically accurate interpretation is very much a point of contention. Samuel Hollander says not, that the classicals were striving to develop Neoclassical theory. Sraffians, of course, disagree.

9.2 The Failure of the Neoclassical Revolution

Second, a whole body of Neoclassical theories developed between 1870 and 1930 are seen to be mistaken. These theorists has prices being determined by endowments, tastes, and technology. They had well-behaved supply and demand curves, a supposedly complete theory of marginal relationships, and substitution working in the expected directions. They also had the economy approaching a state of long run equilibrium whose values are determined by the same forces.

This theory now seems to have been mistaken. The stumbling block is the existence of goods that serve as inputs into production and are themselves produced. Some of the best Neoclassicals, for example, Marshall and Wicksell, were quite aware that they had not totally figured out the theory of capital.

(Although this note is not promarily directed at Libertarians, I will mention that their favorite theorist, von Mises, is revealed to have been wrong on a most basic level. In Human Action he has the usual nonsense about supply and demand. He calls a long run equilibrium stationary state an "Evenly Rotating Economy," and has the economy always tending there as the forces of competition level out differences in prices and profit. This theory is wrong, and he is not even aware of the problems or, for that matter, General Equilibrium theory.)

9.3 Fixed Capital and Land

A third point of debate is whether or not introducing heterogeneous labor, fixed capital, and land restores the Neoclassical theory. I think not. It is always the case that given the technical conditions actually used and all but one of the rate of profit and the wages of the various types of labor; the remaining distributive variable, prices of production, rents, and depreciation are determined. The Sraffian core remains unchallenged.

Of course, the theory becomes considerably more complicated. Suppose a number of different qualities of land exist. Each quality of land each year can provide a fixed level of services and no additional land can be produced. Given the rate of profit or the wage, land can be ordered from high rent to low rent. This order varies along with the wage or rate of profit. Ricardo was wrong in thinking the order to be technically determined, but he was right to think the marginal no-rent land to be of particular importance.

Now consider how the equilibrium position is altered by a change of tastes. What land is actually used depends on the total output of corn. But the total output does vary with changes in tastes. If different lands are employed, the solution to the whole system of prices changes. So consumer demand can affect prices and distribution. Even though the direction of influence is through the technique of production, this is quite a concession for a Sraffian to make.

But the Neoclassical theory is not restored. Substitution and prices do not move in the expected way. The paradoxes are not removed. In fact, new ones arise. For example, consider a reproducable good used in production that lasts several years (a fixed capital good). Even though its physical life is many years, its economic life can be much shorter. The case can arise in which a higher interest rate induces a firm to junk this good sooner.

9.4 Sraffians and Post-Keynesians

Another point of contention is the relationship to Sraffianism to Post-Keynesianism. Post-Keynesians have always emphasized "historical time." Prior to Sraffa's book, they have argued that Classical theory was set in historical time, while Neoclassical theory was not. Neoclassical theory abstracts from the fact that the future is not known.

Contrary to Peter Dorman, it was the leading Post-Keynesian Joan Robinson who interjected the idea of historical time into the Cambridge capital controversy, not the Neoclassicals. She had an independent criticism of Neoclassical capital theory.

She has also criticized the Sraffians for abstracting from historical time. Some Post-Keynesians seem to think Sraffian theory is just another mistaken method that relies on a known future. On the other hand, different methodologies have been assigned to Sraffa. It's all very complicated, but as long as neither is in the mainstream, Sraffians and Post-Keynesians have an interest in cooperating. In fact, some have argued their theories are consistent.

9.5 Extensions and Empirical Work

Where to go from here is also a matter of dispute. Developing Sraffianism beyond the state of things in 1960 is a difficult task. What short-run theories will lead to long-run equilibrium? How can theories of less than perfect competition be incorporated? What are the implications for empirical work? These are all areas of research.

9.6 A Scandal

Finally, I think that both the teaching and the practice of mainstream economic theory is scandalous. Nothing above is new, and much of the argument has been around for thirty years now. Yet most undergraduate microeconomic texts do not even give a hint of these disputes and how much of theory is a matter of contention. And they remain full of, at least, misleading theories. As far as I can tell, mainstream graduate eduction is not any better. Where the theories are correct (i.e. intertemporal equilibrium) the student is not told about the limitations and the problem domain the theory was developed to address, or rather, avoid. I suppose I should not be too hard on teachers for this last bit. Technical subjects are always taught with this limitation.

But the professional literature is equally awful. The Sraffian criticism has never been really addressed by mainstream theorists. The Cambridge Capital Controversy focused on some minor points and never reached a consensus. Based on sci.econ comments, it's already shrouded in myth.

Instead, theorists have blithely developed a short-run theory with hardly any discussion of why this is the proper way to go. You can see why I remain extemely uninterested in rational expectations, for example. What good is basing macro on micro when microeconomic theory is in complete disarray and the resulting macrotheory exhibits all sorts of characteristics that have been shown to only apply to one good models?

So does anybody want to offer a suggestion to why the Sraffian critique and alternative are ignored? Robert Vienneau

Modification date: March 07, 2004 © Copyright 2004 by Brad Cox
Served by John Companies