Patinkin and the Pigou effect: or how a Keynesian came to accept an anti-Keynesian argument*

Abstract

This paper intends to explain how was a supporter of Keynes like Don Patinkin led to integrate the Pigou effect, the arch anti-Keynesian effect, in his theory of involuntary unemployment. The reading of Patinkin's unpublished PhD thesis and the use of the Don Patinkin Papers from Duke University's archives shed new light on this key episode in the formation of the ‘neoclassical synthesis’. Using this material, we show that Patinkin changes his mind on this topic and that his incorporation of the real balance effect into the Keynesian apparatus is, paradoxically, an attempt at reinforcing it.

Keywords:

• Patinkin
• Pigou effect
• real balance effect
• Keynesian theory
• neoclassical synthesis

Notes

 * The author is grateful to Alain Béraud, Carlo Benetti, Mauro Boianovsky, Michel De Vroey and an anonymous referee for their comments and criticisms and to Michal Brezezinski for his indications concerning Alexander Henderson.

Pigou accepted and justified a doubtful argument. One hardly understands how positive savings can coexist with a zero rate of return on capital, hence a rate inferior or equal to the rate of discounting future satisfaction (assumed to be positive). In Ramsey's model, obviously the reference of the two economists, such an outcome is impossible. Pigou justified this possibility invoking ‘the desire for possession as such, conformity to tradition or custom and so on’ (Pigou 1943: 346) or ‘a desire actually to hold wealth for the amenity, so to speak, derived from holding it’ (Pigou 1947: 184). But these arguments are ad hoc. Furthermore, the problem considered by Keynes existed only because the expected rate of inflation was supposed to be nil. In these conditions, the real rate of interest and the money rate were identical, so the equilibrium rate could not be reached if it was negative. But a negative real rate of interest could be reached for a positive money rate if the rate of inflation was sufficiently high. Therefore, the obvious solution to Keynes' problem consisted in increasing the rate of growth of money in circulation. The fact that Pigou did not mention this solution is odd since, in his previous writings, he insisted on the fact that the economy's behaviour depended on the nature of the monetary policy: ‘If nothing is known about banking policy, anything may happen’ (Pigou 1937: 408). The care with which Pigou treated Keynes' argument in 1943 is mysterious since in 1936, in his account of the General Theory for Economica, he made fun of it (on this hostile first reaction see Laidler, 1999: 292 – 4).

In fact, Patinkin did not find his inspiration directly in Keynes, not even in Hansen, but in Klein (1947) (cf. Rubin 2002a)

We know that Friedman and Patinkin had the opportunity to meet at Chicago in 1947. Patinkin prepared his PhD thesis at the Cowles Commission during the academic year 1946 – 7 when this institution was located in this town. Friedman had just been recruited by the University of Chicago and sometimes attended the staff meetings of the Cowles Commission. Alexander Henderson stayed at the Rockefeller Foundation in 1947 and probably participated to some Cowles meetings. Hicks, who praised Henderson as the ‘true discoverer of [compensating variation and equivalent variation]’ for his paper entitled ‘Consumer's Surplus and the Compensating Variation’ (1941), indicated that ‘Henderson, who studied at Cambridge 1933 – 6, had been one of the most brilliant students of economics in that university in those critical days. He taught at Edinburgh before the war and at Manchester after it, but his career as a teacher was interrupted by a long period of service in the British army. In 1950 he left for the USA, where he died in 1954 in Pittsburgh, at the early age of 39’ (Hicks 1981: 114, n.3).

These microeconomic assumptions are documented by Patinkin's PhD thesis (Patinkin 1947a) and by ‘Price Flexibility and Full Employment’ (Patinkin 1948: 553 – 4).

• It should be noted that whether or not the Pigou effect existed was not part of the discussion between Friedman, Henderson and Patinkin. The Pigou effect exists only insofar as money, or any other financial asset, is a net wealth for the economy as a whole. As Kalecki (1944) pointed out, to the extent that it was backed by bank loans, hence private debts, (private) bank money could not give rise to the Pigou effect, for if a price fall enriched the holder of such bank money, it impoverished the debtors by the same amount. Assuming no distribution effects, the variation of global expenses was nil. Of course, this reasoning applied to all forms of private bonds.

• At this stage, the only possible basis for the Pigou effect was government money and government bonds. But this was the case only if the Ricardian equivalence was discarded. The Ricardian equivalence implies that the mode of financing public spending is neutral as long as the State repays its debts by tax levy. Therefore, public debt is not a wealth for the economic system as a whole. If a price fall increases the real value of the amount of money issued by the government or the central bank, the public debt increases to the same extent. The State must then increase its taxes. If private agents anticipate this behaviour correctly, they know that price deflation does not increase their real wealth. The Pigou effect disappears.

• It is not difficult to understand why this argument did not surface in the writings of Keynes' supporters. Indeed, the Ricardian equivalence would have weakened the Keynesian creed concerning the superiority of a policy of public spending financed by public borrowing. This can explain why Patinkin and all the authors of the neoclassical synthesis considered that government money and public debt were net wealth for the economy as a whole. But then they could not contest the existence of an aggregate wealth effect.

In the 1940s most Keynesians admitted that, assuming no distribution and expectation effects, the Pigou effect brought the economic system back to full employment. This was the result of Lange's influence, who stated, in Price Flexibility and Employment, that ‘Under these conditions, the orthodox theory is perfectly valid’ (Lange 1944: 65). This influence is visible when looking at the evolution of Hicks from the first to the second edition of Value and Capital (Hicks 1939, 1946). In 1939, Hicks was led to the conclusion that the temporary equilibrium with unit elasticity of price expectations and no distribution effects was generally unstable. But, in 1946, he modified his position ‘As a consequence of the work of Professor Lange and of Dr Mosak’ (Hicks 1946: 333) and recognized that ‘the stability of the system will be maintained by the income effect’ (ibid: 334). On these grounds, the Keynesian answer to Pigou consisted in stressing the importance of destabilizing distribution and expectation effects. The idea that (temporary) equilibrium was rendered unstable by elastic price expectations formed the main argument of Lange (1944). If price expectation elasticity was greater than one, a fall in the price level would induce negative inter-temporal substitution effects. Agents would expect an acceleration in the fall of prices, therefore postponing their spending. This argument was taken up by Klein (1947a), who claimed that relying on the Pigou effect would push the economic system towards a ‘deflationary spiral’: ‘prices will be depressed; and the economy will travel downward in a hopeless spiral’ (Klein, 1947a: 278). Hicks (1946) stressed that, in an economy with private debts, debtors could be more sensitive to price decrease than creditors. Where the negative effect that a fall in prices had on debtors' spending dominated, the aggregate demand could be an increasing function of the price level. This is what Tobin (1980) called the ‘Fisher effect’. According to Hicks, this would be a ‘perfectly general possibility’ (Hicks 1946: 335). Kalecki put forward the same argument in a more radical way. For him, the debt created by the deflation required for Pigou's mechanism to be efficient would ‘lead to wholesale bankruptcy and a confidence crisis’ (Kalecki 1944: 343).

In his 1947 manuscript, Patinkin claimed that significant deflation took a long time: ‘The very length of time necessary to complete the prolonged price fall militates against the success of the policy’ (Patinkin 1947c: 12). But this argument was used only insofar as it strengthened the danger of instability. In other words, stability was still the dividing line between a Classical and a Keynesian system.

Patinkin noted that whenever each equation of a system of K independent equations of K unknowns was homogeneous to a certain degree with regard to the same variables, this system was over-determinate. Given Walras law, a general equilibrium system of n equations contained n-1 independent equations n-1 unknowns (the money prices). If these equations were homogeneous with degree zero regarding the n-1 money prices, then the system was over-determinate.

Presley (1986) and Bridel (1987) brought as proof a letter that Keynes sent to Robertson in 1925 and in which he discussed the consequences of an increase in the supply of money and of the resulting inflation: ‘The real deposits of the public may fall to a highly inconvenient low proportion of their real income, so that they prefer to do new hoarding so as to raise them, rather than to maintain their current expenditure at its previous level’ (Keynes to Robertson, 31 May 1925).

Their common evidence is a letter from Keynes to Kalecki: ‘Assuming that interest is paid on this out of taxation, it cannot affect the wealth of the community one way or an other. This, it seems to me that Pigou is in reality depending entirely on the increase in the value of gold. The whole thing, however, is really too fantastic for words and scarcely worth discussing’ (Keynes to Kalecki, 8 March 1944).

Cf. Keynes definition of voluntary unemployment as a consequence of wage rigidity (The General Theory, 1936, chapter 2: 6), the first paragraph of chapter 19 (Keynes 1936: 257) and chapter 3 where one can read that ‘the essential character of the argument is precisely the same whether or not money-wages, etc., are liable to change’ (Keynes 1936: 27).

A pure Walrasian would accept the substance of Patinkin's argument but not Patinkin's justification for it. He would say that a general equilibrium model without solution is not acceptable because it is empty and not because it is ‘unrealistic’ (cf. Debreu 2001: 125). Patinkin's crude ‘realism’ may have been a reflection of ideas circulating inside the Cowles Commission under Marschak's directorship. This issue should be the subject of further research.

For a more detailed account see Rubin (2002a).

This argument plays an important role in chapter 14 of Money, Interest and Prices (Patinkin 1956: 234).

‘It looks an extremely plausible thing to take as one's standard assumption that elasticities of expectations are unity, that any change in current prices is expected to be a permanent change. It is also plausible that it has been simply taken for granted by the majority of economists, being assumed implicitly far more often than it is assumed explicitly’ (Hicks 1939: 251).

On this issue, see De Vroey (2002).