The Keynesian Revolution

The first student to complete an economics PhD at MIT was Lawrence Klein. As an undergraduate at the University of California, Berkeley, where he grad­uated in 1942, he had spent most of his time in mathematics and economics classes.¹ He was fascinated with university mathematics and was convinced that it could play a role in economics.

One of the problems that concerned Samuelson was that of trying to iden­tify separate saving and investment functions from a single set of data, for he was not convinced by the ways in which people were trying to do it. When Trygve Haavelmo circulated a paper that analyzed the identification problem—the problem of how to decide whether data on prices and quan­tities traced out a demand curve or a supply curve—Samuelson was very interested, and he set Klein the task of investigating the equivalence between the saving-investment problem and that of identifying demand and supply functions?

Samuelson believed that the two problems were formally identical, imply­ing that the methods used in one problem could be applied to the other.⁵ Klein used that idea to criticize a major study by Mordecai Ezekiel of the relationships between saving, investment, and income.⁶ Ezekiel had tried to identify an investment function separately from a saving function by divid­ing investment into four categories, estimating separately the relationship between each of them and national income.

The problem of estimating saving and investment functions fell squarely within the range of problems that Samuelson was tackling at this time in both his academic papers and at the NRPB. Samuelson had himself criti­cized Ezekiel's estimates of the consumption function in July 1942, just before Klein's arrival. Another paper Klein wrote while at MIT also related directly to Samuelson's work at the NRPB—“The Cost of a ‘Beveridge Plan' in the United States.”⁸ He took the eight categories of social spend­ing, from retirement and unemployment benefits to marriage and funeral grants, and estimated what they would cost to implement in the United States—a task that involved deciding on levels of benefit and working out how many people would be entitled to those benefits. He emphasized that his aim was simply to evaluate the cost of the scheme, not to comment on whether such benefits were appropriate, but he concluded that the cost was “not excessive in terms of expected level of postwar national income,” representing not more than 10 to 13 percent of a high level of income.⁹

a.

The most important topic to which Samuelson directed Klein was analyz­ing the Keynesian system, on which Klein wrote his thesis. In the preface to the thesis Klein wrote that “Oftentimes I feel that I have in many cases done nothing more than paraphrase what I have learned in classes and innu­merable discussions with Professor Samuelson.”¹⁰ Despite Klein’s penchant for using mathematics, and his econometric skills, the thesis comprised a historical analysis of Keynesian economics, tracing the ideas of Keynes, then still alive and frequently visiting the United States on wartime missions, from his early work to the General Theory. Klein painted a picture of Keynes as having started out adhering closely to “orthodox doctrines” and exhibiting marks of “extreme classicism,” but as having broken with that tradition.¹¹ There were continuities in Keynes’s thinking—the need to avoid deflation and unemployment and a critical attitude toward the rentier, as well as a belief that fluctuations in investment were the prime mover of a capitalist economy. Intuition came first and led to the development of a formal theory. Theoretical revolution came when Keynes realized that the saving—invest­ment process determined the level of effective demand, the new theory being conceived in the middle of 1933.

There seems no reason to doubt that Samuelson was the source of many of the ideas in Klein’s dissertation. His key historical point, that the Keynesian revolution involved seeing that the interaction of saving and investment could determine national income, and its dating to the middle of 1933, can be viewed as a direct response to a challenge Samuelson had posed in “The Modern Theory of Income”—to explain what had happened in Cambridge between articles Joan Robinson published in 1932 and in 1933.^b He wrote:

Professor Samuelson has pointed out to me a very interesting develop­ment in the economic literature of 1933.

b. See chapter 18 this volume.

On the next page he reiterated this view, writing, “The differences in theoret­ical structure between these two articles of Mrs. Robinson are quite amazing and should lead us to suspect the occurrence of the Revolution in Cambridge during 1933.”¹³

Klein's dissertation echoed many of the themes that Samuelson had been developing in his writings on the multiplier over the previous two years. These include the importance of arguing in terms of schedules, or virtual movements, and relationships between observables that are true by defini­tion and the claim that unemployment arises because of the shapes of certain schedules (investment and savings being unresponsive in relation to rates of interest), not because of rigidities.

In a system of real economics with no frictions, it is necessary only to assume that certain schedules have different shapes from those assumed classically. If the savings and investment schedules are both interest­inelastic, as we now believe, then it is easy to see why there is no perfect equilibrium of perfect competition possible.¹⁴

Like Samuelson, Klein attached less importance to Keynes's theory of money and the rate of interest than to his theory of income determination, and he used the distinction between full employment and the level of output at which bot­tlenecks developed. He emphasized that the theory of the inflationary gap he had presented was entirely operational.

The state of thinking about Keynesian economics when Klein was writ­ing his thesis is nicely shown by a letter he wrote to Hansen in March 1944. In a conversation the previous day, Klein had failed to explain to Hansen what he had meant when he said “that the level of income is determined in the Keynesian system by savings and investment.”¹⁶ In his letter he explained the difference between an equation that would be satisfied for only one value of a variable and an identity true for all values of the variable. He then turned to the “statical Keynesian system used by Hicks, Lange, Samuelson and many others,” which comprised two equations: supply equals demand for money, and savings equal investment, where everything except the money supply (constant) depends on both income and the rate of interest.^c The first

c. The equations are M = M (i, Ó) and S(i, Ó) = I (i, Y). equation could be solved for the rate of interest as a function of income and the money supply. This could then be substituted into the saving equals investment equation to obtain a function in which income was the only variable.^d If investment were assumed not to depend on output, the result was a diagram such as figure 18.1, earlier. Klein's conclusion was that shifts in investment determined fluctuations in income, a result that was central to the Keynesian system. If the interaction of saving and investment did not determine income, the Keynesian system would be indeterminate. These ideas are now the staple of introductory macroeconomics courses, but were then less well understood.

Klein had come to MIT as an enthusiast for using mathematics in eco­nomics, possessing considerable statistical and econometric skills that he augmented while studying with Samuelson, attending courses in the math­ematics department.

Keynes was quite unkind to the mathematical economists in the General Theory, but it is hoped that this book along with basic works of Lange, Smithies, Hicks, Samuelson, Kaldor will show that only by laying bare the mathematical skeleton models of the theory can all its implications be traced.¹⁷

However, the mathematical skeleton was not enough on its own, hence the need to go back into history to understand the systems they were being used to analyze. Thus, he wrote,

Mathematical models of the skeleton system of the General Theory are very useful in bringing out certain important structural aspects of Keynesian economics, in disproving certain false conceptions about the new theory, and in contrasting the Keynesian and classical sys­tems. The models show the building blocks on which the complete, interrelated system rests.¹⁸

There is a clear distinction here between “model” and “system.”

d. S(i(M₀,Y),Y) = I(i(M₀,Y), Y).

Klein provided a cogent discussion of the relationship of a model to the world, and to policymaking, using the analogy of a physical machine.

That is to say, the Keynesian economic system is essentially a machine which grinds out results according to where the several dials con­trolling the system are set. The dials are the functional relations and the setting of the dials is taken care of by the banking system, the government, the psychology of consumers, the attitudes of inves­tors, the achievements of the technologists, etc. Is it correct to blame the machine if the dials are consistently set at pessimistic levels? If the machine is a true model of the way the system of the real world behaves, then we are not justified in criticizing the machine because other factors set the dials at particular levels. It is just as easy to explain one phase of the business cycle as any other with the Keynesian analy­sis provided we take into account the correct structure of the rela­tionships involved during each phase. Our experience in the past has been that the relationships have been sufficiently alike over a period of years so that we can take the stationary system for an equilibrium solution. If the conditions of our time are such that this stationary solution is not one of full employment, then we must realize this fact and do something about it. On the other hand, the future may be such that we get a continuously changing structure for the economic model. Without serious modification, this can be incorporated into the theory. There is nothing to make us work with constant, unchang­ing functional relations. Shifting equilibria and dynamically changing relations can also be ground out of the machine. The principles of the Keynesian Revolution need not be discarded; rather they must receive elaboration and be extended to handle more complex situations.¹⁹

Samuelson did not express himself this way—in his publications up to this point he had not talked in terms of “models,” though in his unpublished paper “The Modern Theory of Income,” he had referred to his own simpli­fied equation systems and those of others as models. In contrast, Klein's dissertation is littered with the term, which appears on at least 43 of its 195 pages.²⁰ It contains some themes that can be found in Samuelson's work, such as the importance of invariant, structural relationships, but it can be read as expressing an optimistic strategy for empirical modeling that goes beyond what can be found in Samuelson's writings. It is a view that arose out of Klein's ongoing relationship with Samuelson and with which Samuelson must have engaged even if he did not himself go down this route.

Klein's views on the nature of the Keynesian revolution are very different from the views Samuelson had expressed in his publications up to this point, which had been colored by his having approached Keynes from the perspec­tive not of classical economics but from American business cycle theory as represented by Hansen. Samuelson's earliest work on business cycle theory followed Hansen closely, seeing Keynes as having added the concept of the multiplier to a body of literature that adopted a more dynamic perspective on business cycle theory than was to be found in the General Theory. There was no suggestion that he thought in terms of a dichotomy between Keynesian and classical theory. Even in “The Modern Theory of Income,” he wrote of the Keynesians from the perspective of an outsider: adopting a modern theory of income determination was not the same as being a Keynesian.^e For example, he criticized “the Keynesians” for having two different theories of the equality of saving and investment, between which they continually switched back and forth. This is very different from the picture painted by Klein of Keynes as the economist who had put everything together to create a new system. Moreover, when Klein compared Keynes's theory with the ideas of other economists, his list included no Americans, unless Schumpeter is counted as American.^f This raises the question of whether, despite Klein's claim that in many of his dis­cussions he did no more than paraphrase what he had learned from Samuelson, Klein had helped shift Samuelson to a different stance on Keynes.

Samuelson's lectures in early 1943, when Klein was working on his the­sis, focused on “the Keynesian system,” a significant change compared with earlier work in which he wrote in terms of incorporating Keynesian ideas into the more general Hansen theory.^g Whether or not it was Klein who per­suaded Samuelson to change his emphasis, it was while working with Klein that Samuelson began to talk of Keynes rather than Hansen as the key figure in the New Economics. Klein was also politically much more committed than Samuelson. He believed in socialism, and shortly after leaving MIT, he even became a member of the Communist Party.²¹ In contrast, Samuelson was never a socialist, but as he came closer to Hansen, the roots of whose political ideas lay in the New Deal, he grew to adopt a more intervention­ist position that was branded socialist by its conservative critics. The two had much in common, and it seems certain that their relationship was less unequal than Klein suggested.

e. This judgment is based on the way he uses the terms “Keynesian” (to refer to ideas of Keynes) and “Keynesians” (to refer to an identifiable group of economists).

f. His list comprised Hayek, Schumpeter, Myrdal, Pigou, Hawtrey, and Hobson.

g. See chapter 20 this volume.