Macroeconomics and the Phillips Curve Myth.
Here is a story told hundreds of times--in this case by a retiring Australian central bank governor, in a radio lecture for the general public, on the history of monetary policy:
In 1958 Professor A.W.H. Phillips... published a paper with the uninspiring title of 'The relation between unemployment and the rate of change of money wage rates in the United Kingdom, 1861-1957'. This showed that there had been an inverse relationship between the rate of wage inflation and the rate of unemployment over that period; that is, when wage inflation was low, unemployment was high and vice versa. This relationship became known as the Phillips Curve, and it lent empirical support to the view that there was a choice between a low unemployment- -high inflation situation, or a high unemployment--low inflation situation, and the various combinations between these two extremes. Not surprisingly, given this choice, the vast majority would choose a point that had low unemployment even though it meant higher inflation. Unfortunately, there was a very serious dynamic problem with the Phillips Curve, of which its creator was well aware... [T]o get unemployment below the level that was originally consistent with low inflation will take a series of increases in inflation with no apparent equilibrium end-point in sight. Thus, you cannot go to a permanently lower unemployment rate by accepting inflation at a constant higher level; what is required is a constantly increasing rate of inflation... This critique of the overly ambitious use of Keynesian demand management policy was mainly the work of Milton Friedman. For a decade or more it was hotly debated, but was ultimately proved right and is accepted today by economists of virtually all political persuasions. (Macfarlane 2006, 21-22, 24)
It is a story everybody knows. According to James Forder, it is a myth. Each part of it is wrong: that Phillips convinced the profession of a stable inverse relationship between wage inflation and unemployment; that economists saw it as a menu of options; and that policymakers chose inflationist policy from the menu. Few will be ready to accept such strong claims, and Forder knows that. He is saying that the textbooks are misleading; that someone like Macfarlane--no dummy, who ran a central bank for years--can have a fundamental misconception about fairly recent history in his area of expertise; essentially, that the profession has developed a false memory. He better be bringing strong evidence.
He does. He softens the reader by pointing out that to believe the conventional wisdom is to believe something implausible: it implies that the economics profession of the 1960s was stupid: 'struggling to articulate the simplest ideas, and disputing the obvious even when it was stated'. The myth suggests 'an interlude in which either economics was bizarrely primitive or its practitioners were extraordinarily slow-witted' (pp. 1-2). But, ultimately, Forder's book convinces by sheer brute force of literature review. He acknowledges he has not read everything, and perhaps there is a hidden seam of Phillips curve inflationism somewhere to vindicate the textbook story. But he has clearly read a lot--he discusses more than a thousand papers and books--and he demonstrates in detail that the myth's standard supports do not say what we thought they say, and that in any case these papers are not representative of the literature of the 1950s and 1960s.
First, Forder tackles 'the curve of Phillips' itself, Phillips' 1958 Economica paper. Phillips' results were not novel in suggesting a negative relationship between wage or price change and unemployment--the idea goes back at least as far as Hume, and statistical estimations to Fisher and Tinbergen in the 1920s and 1930s. In 1958 'there is not the remotest possibility that such an idea would ever have surprised anybody' (p. 12). What was new was the suggestion that the nature of the relationship had been consistent for a very long time in the United Kingdom, since Phillips' data covered nearly a century, from 1861 to 1957.
It was widely accepted that an increase in demand could put upward pressure on the wage and price levels even before full employment was reached. This was the major point of Chapter 21 of Keynes' General Theory. Some of the reasons might be thinkable in terms of a smooth curve: diminishing returns to labour and other factors, the development of bottlenecks in some industries. But others would be discontinuous and somewhat unpredictable, and Keynes thought this was true of money-wage determination. Wages would not be unresponsive to demand, because entrepreneurs will be more likely to accede to pressure for higher wages when business is booming--but the relationship would not be smooth, depending as it did on 'the psychology of the workers and by the policies of employers and trade unions' (Keynes 1936, 301). Forder establishes that the main line of wage theory in the 1950s was very much along these lines--though more suspicious than Keynes himself of general explanations of wages in terms of marginal productivity or supply and demand--and focused on institutional frameworks, bargaining strategies and the evolution of norms.
Phillips' result went against the grain of this literature. As Forder puts it, it suggested that 'from the invention of the bicycle to the flight of the Sputnik, wage bargaining remained unchanged' (p. 22), in spite of the rise of the labour movement, the depression, the welfare state and decades of technological and industrial change. The conclusion was not generally accepted. The statistical techniques Phillips used were idiosyncratic, rough and easy for the many contemporary critics to find fault with. Almost nobody accepted the conclusion of an invariant relationship between unemployment and money wage growth.
This did not mean the rejection of any relationship between unemployment and wage growth. Phillips' paper was part of a wave of econometric studies of wage determination that began to surge around 1958 and continued through the 1960s. This has been retrospectively interpreted as a 'Phillips curve' or 'inflation-unemployment trade-off literature, for example in Santomero and Seater's (1978) survey of around 200 papers. If Forder's point were merely that this literature owed little to Phillips (1958) and so had been misnamed, the notion of a 'Phillips curve myth' would not be very interesting. But, he has bigger fish to fry. He aims to show that before 1968 (the year of Friedman's famous critique of the Phillips curve as trade-off) there was (1) no general neglect of feedback from price change to wage change, and (2) little interest in treating the relationship as an exploitable trade-off.
Forder goes through paper after paper. In most of the studies, unemployment (or something related) is one independent variable among others, including profits, bargained wages in key industries, the quit rate, productivity, the rate of change of unemployment, union militancy and so on. Unemployment was regularly found to be an important determinant, but rarely an exclusive one. Critically, price change was 'almost always' included among the explanatory variables in the literature of the early 1960s (p. 60).
The ubiquity of price change in the models is vital to Forder's argument, because it undermines the idea that macroeconomists had neglected the possibility of inflationary momentum before Friedman and Phelps. Price change was often not only included, but emphasised. Forder quotes one study after another from the late 1950s and 1960s to the effect that price inflation was the major determinant of wage inflation, that wage-bargainers were not subject to 'money illusion', that sustained price inflation may have an increased effect on wage inflation over time. Such statements can be found not only in marginal, forgotten papers, but in those widely recognised as the heart of the 'Phillips curve literature'. Lipsey (1960), for example, suggests that labour market supply and demand variables predict wage growth adequately for the nineteenth century, but that 'more of the twentieth-century variations can be explained in terms of wages "chasing" prices or of prices chasing wages' (Lipsey 1960, 26). To the extent that money wages in turn affect prices, it is difficult to untangle the causality here--and this is the note on which Lipsey ends his paper, far from endorsing a simple relationship between unemployment and wage or price inflation.
Forder also finds many writers suggesting long before Phelps (1967) and Friedman (1968) that workers and firms alike would adapt to persistent inflation over time and anticipate it in their bargains. Often the point was put in terms of catch-up from previous price rises, rather than anticipation of future inflation, but the inertial effect is the same. Phelps (1968) explicitly cited a number of forerunners. Neither Friedman (1968) nor those who reacted at the time claimed it to be a new idea.
How can this be squared with the idea of an exploitable tradeoff between inflation and unemployment? According to Forder, there is no mystery, because few made such an argument. The case everybody knows is Samuelson and Solow (1960), who do present an unemployment-price inflation Phillips curve as a 'menu of choice' (Samuelson and Solow 1960, 192)--which does seem clear cut, at least out of context. However, they present their analysis as 'phrased in short-run terms' so that 'it would be wrong... to think that our... menu that relates obtainable price and unemployment behaviour will maintain its same shape in the longer run' (Samuelson and Solow 1960, 192). Policy and, yes, adaptive expectations could shift the curve. Their estimate of the curve for the United States in fact represented a pessimistic conclusion about the compatibility of full employment and price stability: the curve seemed to have shifted upwards since the war. The policy conclusions were not about selecting the optimal point on the curve, but rather about the possibility of shifting the curve back down. In the 1960s, the paper was sometimes seen as making a case for a stable trade-off, but usually not, and those who did see it that way disagreed with the idea.
This was no aberration, but basically representative of pre-1968 'Phillips curve' literature. Very few used the idea of a ceteris paribus relationship between unemployment and wage inflation to make a case for accepting higher inflation. Combing the papers for Phillips curve inflationism, Forder finds plausible candidates mostly among a few Canadians: 'Among the British work, there is nothing; amongst the American, there is a small amount' (p. 75). Those who used the language of 'trade-offs' tended to prioritise price stability rather than full employment.
There were economists arguing for the acceptance of gentle inflation, before and after 1968. Most commonly, they made some version of what Forder calls the 'lubrication argument': the idea that a modest rate of inflation--even if it is anticipated--makes room for relative price adjustment without the pain of local deflation in industries whose relative prices are too high. This could be interpreted as a 'trade-off, since it suggests that unemployment would be higher if inflation were too low. This explains, writes Forder, 'what those few Phillips curve inflationists had in mind' (p. 79), though the idea pre-dates Phillips (1958). But this was not generally an argument for higher inflation, simply for accepting the low but persistent inflation that already prevailed. It must be seen in context, where a sustained upward price trend was a novelty and with other voices urging absolute price stability. The lubrication argument is now utterly conventional, with, say, the European Central Bank defining price stability as 'inflation below, but close to,
2 per cent' for essentially lubricationist reasons (quoted by Forder, p. 202). The remarkable thing is not based that there was some mild inflationism in the 1960s, but that there was not more.
Forder does see changes in the literature after 1968. The term itself was more widely used--'the Phillips curve' was more prominent after Friedman (1968) than before. Previously it had more often than not referred to a relationship between unemployment and wage growth; now it more often described the unemployment-price inflation relationship. There was no single Phillips curve: it played a role in a number of distinct research programs. These were not dominated by reactions to Friedman's paper, positive or negative, though testing the hypothesis of a long-run vertical curve was one of the programs. Again, the inclusion of expectations was not an innovation: feedback from the price change of previous periods was already a familiar component of wage and price inflation models, and it was simple to reinterpret this in terms of adaptive expectations. This research did develop a general consensus that there was no long-run trade-off between unemployment and inflation.
So, Forder does not dispute the emergence of this consensus in the 1970s. What he does dispute is that this was a break from an earlier, naive consensus around the possibility of exploiting a stable trade-off. It is not the case that the kind of model that became standard--the 'expectations-augmented Phillips curve'--explains the stagflation of the 1970s, as if it were all the fault of a misguided attempt to exploit a supposed trade-off. Low unemployment was evidently compatible with a stable rate of inflation for two decades after World War II; it remains to be explained why the 'natural rate of unemployment' or 'non-accelerating inflation rate of unemployment' apparently jumped upwards in the 1970s.
In the penultimate chapter, Forder turns to explaining the emergence of the myth itself. He dates it precisely to 1975, with the first clear statements that policy was led astray by the Phillips curve idea from Friedman himself. But, the ground was cleared earlier in the decade. People read and engaged in the expanding inflation literature in the 1970s without needing to know much about the literature of the 1950s and 1960s. They got a sense of 'Friedman's position', 'Tobin's position' and so on, and took an understanding of 'the Phillips curve' from its uses at the time--now referring mainly to a relationship between unemployment and price rather than wage inflation. It hardly mattered for the contemporary debate that the ideas did not necessarily originate with the names they became associated with. But, it did feed misunderstandings of the past. Phillips (1958) came to be mistakenly seen as the origin of the idea of a negative relationship between unemployment and wage change; Phelps (1967) and Friedman (1968) were misidentified as originators of the expectations argument. Those reference points from the past that did surface often gave a misleading impression if they were taken as representative: for example, Samuelson and Solow (1960) were unusual for their time in referring to a 'menu of choice' and in not systematically incorporating past price change into their analysis (though they did, as noted, acknowledge it could be important).
This was not the work of Friedmanites alone. Some of Friedman's opponents took to defending the idea of a trade-off and even fed the idea that policymakers had sought to exploit one in the 1960s. Tobin, for example, not only defended policy based on a 'Phillips trade-off, but also seemed to suggest that the Kennedy administration had targeted a point on the Phillips curve. On the latter point, Forder argues that Tobin was simply mistaken, pointing to a lack of any corroborating evidence. As for his defence of a 'trade-off, this was a restatement of the 'lubrication' idea Forder traced through the 1950s and 1960s, and not based on any idea of perpetually mistaken expectations. Meanwhile, a sense of the tide turning against the idea of a short-run trade-off developed in the econometric literature. A number of results challenged the hypothesis that persistent inflation would eventually be fully anticipated and incorporated into bargains, but later in the 1970s the consensus came to accept it. It would be easy to jump to the conclusion that this was a victory of Friedman over post-war Keynesian convention, but really both sides of the argument developed after 1968.
The 'Phillips curve' as we know it was actually a construct of this period. 'It was not that the Phillips curve was established in thought, and the revolutionaries needed to show that money was neutral in the long run. It was that long-run neutrality was the established view, and what was sought was a way to incorporate the Phillips curve into that established thinking' (p. 135). Phelps was quite explicit about this: the problem for him was to explain why there seemed to be a curve in the short run. The short-run curve was born alongside the vertical long-run curve; only later was the former mistakenly attributed to Keynesian convention.
For the historian of economic thought, the book could be demoralising. The Phillips curve story is 'more or less the only piece of the history of economic thought' that economics students are exposed to. It is the origin story of modern macroeconomics, 'a story of how the errors that led to policy failure have been put behind us, and how the dissent for which economists were once so notorious was ended' (p. 217). And it is wrong.
But the book is also a call to arms. A bogus story about Keynesian fools and Friedmanite heroes has long lent rhetorical support to the counter-revolution in macroeconomics. As Robert Gordon (2009, 2) writes, the term 'Keynesian economics' 'has been forever tainted by association... with the discredited 1960s-era Phillips Curve and its implication of an exploitable inflation-unemployment tradeoff. Forder wants to set the record straight not only for historical interest, but also to suggest that there is wisdom to be recovered from pre-1968 macroeconomics. Not, obviously, an exploitable Phillips curve, but a rich literature on wage and price determination that was less prone to assume any kind of mechanical inflation-unemployment relationship than we are today.
The author reports no conflicts of interest. The author alone is responsible for the content and writing of this article.
Friedman, M. 1968. "The role of monetary policy." American Economic Review 58 (1): 1-17.
Gordon, R. J. 2009. "Is modern macro or 1978-era macro more relevant to the understanding of the current economic crisis?" Paper Presented at the International Colloquium on the History of Economic Thought, Sao Paulo, Brazil, August 3. http://faculty-web.at.northwestern.edu/economics/Gordon/GRU_ Combined_090909.pdf
Keynes, J. M. 1936. The General Theory of Employment Interest and Money. London: Macmillan.
Lipsey, R. G. 1960. "The relation between unemployment and the rate of change of money wage rates in the United Kingdom, 1862-1957: a further analysis." Economica 27 (105): 1-31.
Macfarlane, I. 2006. The Search for Stability. Sydney: ABC Books.
Phelps, E. S. 1967. "Phillips curves, expectations of inflation and optimal unemployment over time." Economica 34 (135): 254-281.
Phelps, E. S. 1968. "Money-wage dynamics and labour-market equilibrium." Journal of Political Economy 76 (4): 678-711.
Phillips, A. W. 1958. "The relation between unemployment and the rate of change of money wage rates in the United Kingdom, 1861-1957." Economica 25 (100): 283-299.
Samuelson, P. A., and Solow, R. M. 1960. "Analytical aspects of anti-inflation policy." American Economic Review 50 (2): 177-194.
Santomero, A. M., and Seater, J. J. 1978. "The inflation-unemployment trade-off: a critique of the literature." Journal of Economic Literature 16 (2): 499-544.
Department of Political Economy, University of Sydney, NSW, Australia
|Printer friendly Cite/link Email Feedback|
|Publication:||History of Economics Review|
|Article Type:||Book review|
|Date:||Jun 1, 2016|
|Previous Article:||Nature's Gifts. The Australian Lectures of Henry George on the Ownership of Land & other Natural Resources.|
|Next Article:||A History of Economic Science in Japan: The Internationalization of Economics in the Twentieth Century.|