ABSTRACT
In criticizing Samuelson (1962), Garegnani repeated what was essentially the argument of his own 1960 book in Italian (Garegnani 1960),1 namely that the amount of capital cannot be determined independently of the rate of profit. But this time, drawing upon Sraffa (1960), he put his criticism in terms of the wpf, which he had not done in his book, and he mentioned reswitching, without, however, assigning any central role to it in his criticism. I suggest that the subsequent course of events was as follows: Garegnani’s
criticism referred to Sraffa (1960) and mentioned reswitching. This prompted Samuelson to read Sraffa’s book, or study it more closely,2 and it was not until then that he realized reswitching was a counterexample to some of the predictions of neoclassical models which he had believed to be true. One of these was ‘Samuelson’s theorem’.3 So it was only via Garegnani’s criticism and through Sraffa’s book that at least one ‘mainstream economist’, namely Samuelson, realized that his ‘orthodoxy’ was at stake, as Robinson relates.4 I think that we have to take Robinson’s observation quite literally: the only other economist who realized the importance of the reswitching result was mainstream, orthodox Samuelson. One more proviso seems in order. Solow in his 1955 article had written his
own criticism of the aggregate production function, which had nothing to do with reswitching. He stated the conditions under which it is legitimate to use an aggregate production function, and concluded that these are so restrictive as to make the aggregate production function of very limited usefulness. It is he who expressly coined the term ‘pseudo production function’5 in order to emphasize that this instrument of static analysis was not up to the task of describing the process of production and accumulation in time.6 But neither he nor anyone else pursued this line of criticism at the time. Solow even continued to use the production function in his own subsequent empirical work. Solow the econometrician was not inhibited by the results of Solow the theoretician.7 But with the exception of these authors it was apparently generally believed that Samuelson (1962) provided a satisfactory account of the neoclassical, idealizing approach to distribution theory: ‘For
several years everyone (except Garegnani) was somewhat baffled by the surrogate production function.’ (Robinson 1975c: 36).8 In the meantime, Samuelson had become convinced that the possibility of
reswitching was a serious criticism, and he tried to find a way of saving his 1962 model. But he did not reply to Garegnani’s main criticism, which, as we have seen, did not make use of reswitching. Instead, Samuelson focused on the different question of how reswitching might be avoided. He thought that the reswitching counterexample, which he had encountered in Sraffa’s book, was not so strong as to render his surrogate production function model inapplicable in general. So, Samuelson changed the problem. And he sought the solution in the conjecture that indecomposable models are not affected by reswitching. Then ‘sometime in 1964’ (Solow 1983: 184) Samuelson told David Levhari, who was writing his Ph.D. thesis under Samuelson’s supervision, about his conjecture. In 1965 Levhari published his proof of Samuelson’s conjecture that reswitching cannot occur in an indecomposable production system.9 Pasinetti was the first to come up with what he claimedwas a counterexample to this result. During the 1965Econometric Society Conference in Rome a symposium was organized to discuss Levhari’s proof. The symposium papers were published in Quarterly Journal of Economics, 1966. They demonstrated in various ways the falsity of Levhari’s proof: ‘Counterexamples were soon produced and reswitching became an issue instead of a curiosum.’ (ibid.). The ‘Cambridge debate’ had gained momentum. How this came about will be shown in more detail with the aid of the following case studies.
