In Macroeconomics and the Phillips curve myth, I presented lots of data on a variety of points of the history of economics, aiming at establishing various parts of my account of the history of the Phillips curve. While I was writing it, I did not find all the evidence that was there. So here is what I have found since.
I presented a collection of very clear early statements of the idea that as later terminology puts it ‘the Phillips curve shifts with changes in expectations of inflation’, and that was following up those I had already pointed to in The historical place of the ‘Friedman-Phelps’ expectations critique in the European Journal of the History of Economic Thought. The point was to show that the attribution of that argument to Friedman and Phelps, and its dating to the late 1960s is entirely erroneous since it was very well known by the time they stated it. And there are still more examples of its being stated.
First of all, here is one from Harry Johnson that no one seems to have noticed. He clearly doubts that as of 1958, the problem of accelerating inflation was manifesting itself, but the point is that his remarks make it impossible to sustain the idea that no one understood the character of the issue before Friedman talked about it at the end of the 1960s. Then one from Coombes – he was a central bank governor. His remarks clearly refute the view that the idea that the adaptation of expectations to ongoing inflation could generate accelerating inflation was invented by Friedman late in the 1960s. Thirdly, there is a harder-to-interpret one from Albert Hart in 1946. The ‘leakage’ is the dissipation of nominal demand in higher prices rather than in calling forth greater output. Their ‘institutionalization’ is the process by which the ordinary responses of the system come to incorporate the inflation as it comes to be recognized as normal. And he clearly appreciates that deceiving wage bargainers is essential to securing any on-going benefit from inflation.
‘But I would emphasise more strongly that a prolonged wage inflation sets up its own momentum through the establishment of mechanisms of automatic wage increase, analysed by both Mr. Robertson and Mr. Flanders, which will keep the inflationary process going for a considerable time after demand-restriction policies have been applied – especially if they have not been applied with convincing determination.’ … ‘though people and institutions have been adapting their behaviour to the assumption of continuing inflation, there is no evidence so far of a threatening acceleration in the rate of price increase.’
Harry G Johnson, (1958) ‘Two schools of thought on wage inflation’. Scottish Journal of Political Economy 5(2) pp. 149-153, cited parts from p. 149 and 151.
‘The longer prices continue to rise – however slowly – and the more confident become the expectations of further rises, the more people will seek protection and the more expensive it will become. Indeed, it is difficult to conceive an upward trend of prices remaining slow and gradual in a world where everybody is seeking to protect himself against its effects.’
Herbert Coombes, (1959) ‘A matter of prices’. Economic Record 35(72) pp. 337-348, cited part from p. 339.
‘The dilemma hinges chiefly on the danger of inflation with driving power from the cost side. Under existing and widely accepted practices, unions are likely to drive up wages whenever employment is reasonably good …
Inflation always indicates some leakage of the forces working to raise employment; and if the pace of inflation is fast enough to attract general attention these leakages can well become institutionalized (via farm parity adjustments, sliding-scale wage settlements, corrections to cost accounting, etc.). Economists who put faith in rising prices to solve the dilemma seem to rest their hopes on keeping inflation unnoticed – which means either having very little of it, or else deceiving organized labor, farm groups, etc., as to what is happening.’
Albert Hart, (1946) ‘The problem of full employment: Facts, issues and policies’. American Economic Review 36(2) pp. 280-290, cited part from p. 288.
The 'Phillips curve' before Phillips
Another daft thing sometimes said is that before Phillips’ paper no one had the idea that inflation and low unemployment might be associated. Well, it has been recognised that there were plenty of people had that idea and there is quite a fashion for showing one’s erudition by citing David Hume’s discussion in his essay Of money from 1752. The point though is not that someone got there before Phillips. It is that there was nothing at all surprising about that idea when Phillips wrote. The thought that broadly speaking, periods of inflation were periods of low unemployment and periods of high unemployment were periods of falling or stable prices was in the first place, just a bit of common sense, and in the second, an everyday observation of any consideration of the business cycle. The ordinariness of the idea is again the important thing. And again I found plenty of discussions of that idea, and plenty of econometric estimations of the relationship too, from the 1950s and earlier. Here are some more on that and closely related matters which I have discovered since.
Keith Isles wrote concerning Australian policy. He presented data showing the rate of unemployment (on an inverted scale) and the rate of change of wholesale prices. Using a one-quarter lag for unemployment he said ‘The two curves move in considerable sympathy’. They do too.
I learned about this case from Coleman, Cornish, and Hagger (2006 p. 52). They comment that Isles thereby ‘stumbled on’ the Phillips curve and leave it at that. That’s fair enough I suppose as Isles too just remarked on the point, and left it. But that sort of reaction is what shows that even two and half decades before Phillips, the relationship did not strike economists as a surprise. That is certainly true of those writing about it in the few years before Phillips.
Coleman, William, Selwyn Cornish, and Alf Hagger, Giblin’s Platoons: The trials and triumphs of the economist in Australian public life. Canberra: ANU ePress.
Isles, Keith (1932) Australian monetary policy reconsidered, The Economic Record, vol 8(2) pp.191-205
Peter Kenen’s British monetary policy and the balance of payments, 1951-1957, Harvard Economic Studies vol CXVI, published by Harvard University Press, Cambridge MA, in 1960 was originally his doctoral thesis submitted to Harvard in 1957. Appendix B is an econometric study of ‘Wages. costs, and prices’. He was mainly addressing the question of cost push inflation, but produced an equation in which the change in unemployment had a role in explaining changes in the price level. That is not quite the same thing as the level of unemployment doing the work, but in a period in which both prices and employment were (by later standards) extremely stable, and with a very short data series, the two would be hard to distinguish. The more interesting point, perhaps, is that it is yet another econometric treatment of the issue preceding Phillips.
Reading Samuelson and Solow
A third thing is the question of how Samuelson and Solow’s paper in the 1960 conference volume of the American Economic Review was actually interpreted. Here, the story often told is that it was their interpretation of Phillips’ work that led everyone else to think of it as offering a policy ‘menu’. Well, that is not what they were doing, but they are a little vague. What impressed me more was that in the relevant period – the 1960s – I could find no one who interpreted them as saying that, and who accepted that interpretation. Everyone who reads them as presenting a ‘menu’ interpretation of Phillips rejects that interpretation. It is only in the 1970s that people start to rewrite the history by saying that Samuelson and Solow had been influential in promoting inflationism. I tried to find everyone citing them from the 1960s, and I have no idea how I missed their appearance in a debate in the Economic Journal between Wilfred Beckerman and Béla Belassa. Anyway, I did not discuss that one. But here we go.
Balassa (1963) attributed to Beckerman (1962) the view that a high rate of growth of manufactured exports tended to be accompanied by improvements in productivity and hence reduced costs. That, obviously, might lead to further export growth and hence a virtuous circle. A question that arose was that of what happened to wages in the meantime. Balassa said that Beckerman had treated them as substantially determined by changes in productivity and questioning that, drew attention to the idea that wage change was related to unemployment – citing Phillips, and Samuelson and Solow. This, said Balassa suggested that Beckerman’s virtuous cycle would be attenuated as export growth reduced unemployment, thereby leading to a source of price increase which would offset the gains in productivity Beckerman considered. Hence his virtuous circle would be attenuated, and at some point very probably ended.
Beckerman (1963) responded by doubting the relevance of the evidence in Phillips and Samuelson and Solow. He suggested their results were explained by variations in the mean wage increase in the countries in question caused by variations in the level of demand. An inspection of international evidence showed that the mean wage increases were much better explained by productivity change than differences in unemployment levels.
The debate on the merits of the issue about export led growth went on between the authors, but as to the attitude to Samuelson and Solow, these responses are typical of the literature as a whole. One saw them as pointing to a danger of inflation as unemployment fell – thereby raising a problem, not suggesting a way of reducing unemployment. The other thought their work pretty uninteresting.
Balassa, B. (1963) ‘Some observations on Mr Beckerman’s ‘export propelled’ growth model’. Economic Journal 73(292) pp. 781-785.
Beckerman, W. (1962) ‘Projecting Europe’s growth’. Economic Journal 72(288) pp. 912-925.
Beckerman, W. (1963) ‘Some observations on Mr Beckerman’s ‘export-propelled’ growth model: A reply’. Economic Journal 73(292) pp. 785-787.
Who else discusses these things?
What have I missed? Other ‘Phillips curves’ before Phillips? Other statements of the expectations argument before 1967? Other citations of Samuelson and Solow from the 1960s? If it is not in Macroeconomics and the Phillips curve myth, and it isn’t here, let me know – particularly if you think it gives any support to the conventional story.