Misunderstanding the “Phillips Curve”

Steve Keen  Debunking Economics

Part 3 : Complexities: issues omitted from standard courses that should be part of an education in economics.

(9) Let’s do the Time Wrap again.

Addendum Misunderstanding Bill Phillips, wages and “the Phillips Curve”  


“Bill Phillips the man was undoubtedly one of the most dynamic human beings of all time. Compared to that of Phillips , the lives of most economists – even non – neoclassical ones – are as pale as the theories that neoclassical economists have concocted about the world. He left school at fifteen , worked as a crocodile hunter and gold miner in Australia , learnt engineering by correspondence , was awarded an MBE for his role in the defence of Singapore in 1942 , and , as a prisoner of war , made a miniaturized radio from components he stole from the camp commander’s radiogram . Despite the effects of malnutrition and abuse in the camp , within five years of the war finishing – and while still an undergraduate student of economics – he had his first paper published in a leading journal ( Phillips 1950 ) . The paper described an analog computer dynamic simulation model of the economy ( MONIAC ) that he constructed at a cost of £ 400 , just three years after the first digital computer ( ENIAC ) had been constructed at a cost of US $ 500,000 ( Leeson 1994 , 2000 ) .

MONIAC put into mechanical – hydraulic form the principles of dynamics that Phillips had learnt as an engineer , and it was this approach which he tried to communicate to economists , on the sound basis that their preferred methodology of comparative statics was inappropriate for economic modeling :

S Keen PC 9.3.jpg

9.3 Phillips’s functional flow block diagram model of the economy

RECOMMENDATIONS for stabilizing aggregate production and employment have usually been derived from the analysis of multiplier models , using the method of comparative statics . This type of analysis does not provide a very firm basis for policy recommendations , for two reasons .
First , the time path of income , production and employment during the process of adjustment is not revealed . It is quite possible that certain types of policy may give rise to undesired fluctuations , or even cause a previously stable system to become unstable , although the final equilibrium position as shown by a static analysis appears to be quite satisfactory .
Second , the effects of variations in prices and interest rates cannot be dealt with adequately with the simple multiplier models which usually form the basis of the analysis .” ( Phillips 1954 : 290 )

Phillips instead proposed that economists should build dynamic models of the economy – models in which time was embraced rather than ignored via the device of comparative statics – and his underlying method here was the functional flow block diagram . This had been devised by engineers in the 1920s as a way to visually represent dynamic processes , which previously had been shown as either differential equations , or transformations of these equations into other mathematical forms . 10 Phillips drew such a diagrammatic representation of a simple dynamic economic model ( ibid . : Fig . 10 , p . 306 ; see Figure 9.3 ) , with symbols to indicate operations like time lags , differentiation and integration with respect to time , addition and subtraction , etc . The model recast the standard comparative – static , multiplier – accelerator models of the time into dynamic form .

This model was only the starting point of a project to develop a complete dynamic model of the economy , in which the feedback effects and disequilibrium dynamics that were ignored by the conventional ‘ Keynesian ’ models of the time could be fully accounted for.

In particular , Phillips extended his model to consider the impact of expectations upon prices . Given how much his work has been falsely denigrated by neoclassical economists for ignoring the role of expectations in economics , this aspect of his model deserves attention prior to considering the Phillips Curve itself :

“Demand is also likely to be influenced by the rate at which prices are changing [ … ] this influence on demand being greater , the greater the rate of change of prices [ … ] The direction of this change in demand will depend on expectations about future price changes . If changing prices induce expectations of further changes in the same direction , as will probably be the case after fairly rapid and prolonged movements , demand will change in the same direction as the changing prices [ … ]

If , on the other hand , there is confidence that any movement of prices away from the level ruling in the recent past will soon be reversed , demand is likely to change in the opposite direction to the changing prices [ … ] .” ( Ibid . : 311 ; emphases added )

S Keen PC 9.49.4 The component of Phillips’s Figure 12 including the role of expectations in price setting

Phillips didn’t merely talk about expectations : he extended his model to incorporate them – see Figure 9.4.  As part of this project , Phillips also hypothesized that there was a nonlinear relationship between ‘ the level of production and the rate of change of factor prices [ labor and capital ] ’ ( ibid . : 308 ) , and he sketched a hypothetical curve for this relationship – see Figure 9.5 .

S Keen PC 9.59.5 Phillips’s hand drawing of the output – pricechange relationship

The role of this relationship in his dynamic model was to limit the rate at which prices would fall when unemployment was high , in line with ‘ the greater rigidity of factor prices in the downward than in the upward direction ’ ( ibid . : 308 ) . In a dynamic model itself , this does not lead to a stable trade – off between inflation and unemployment – which is the way his empirically derived curve was subsequently interpreted – but rather limits the volatility of the cycles that occur compared to what a linear relationship would yield .

This was hard for Phillips to convey in his day , because then functional flow block diagrams were merely means to describe a dynamic model – they didn’t let you simulate the model itself . But today , numerous computer programs enable these diagrams to be turned into active simulations . There is also an enormous analytic framework for analyzing stability and incomplete information supporting these programs : engineers have progressed dramatically in their capacity to model dynamic processes , while economics has if anything gone backwards .

S Keen PC 9.6.jpg9.6 A modern flow – chart simulation program generating cycles , not equilibrium

Figure 9.6 illustrates both these modern simulation tools , and this difference between a linear and a nonlinear ‘ Phillips Curve ’ in Goodwin’s growth cycle model . One of these programs ( Vissim ) turns the six – step verbal description of Marx’s cycle model directly into a numerical simulation , using a linear ‘ Phillips Curve . ’ This model cycles as Marx expected , but it has extreme , high – frequency cycles in both employment and wages share .

Embedded in the diagram is an otherwise identical model , which has a nonlinear Phillips Curve with the shape like that envisaged by Phillips . This has smaller , more realistic cycles and these have a lower frequency as well , closer to the actual frequency of the business cycle .

What this model doesn’t have – and this is a very important point – is an equilibrium ‘ trade – off ’ between inflation ( proxied here by the rate of change of wages ) and unemployment . Instead the model economy is inherently cyclical , and Phillips’s overall research agenda was to devise policy measures – inspired by engineering control theory – that might attenuate the severity of the cycles .

Had Phillips stuck with just a sketch of his hypothesized nonlinear relationship between the level of production and factor prices , it is possible that he would be known today only for these attempts to develop dynamic economic analysis – and possibly relatively unknown too , given how other pioneers of dynamics like Richard Goodwin ( Goodwin 1986 , 1990 ) and John Blatt ( Blatt 1983 ) have been treated . Instead , he made the fateful decision to see whether he could find such a relationship in the UK data on unemployment and the rate of change of money wages .

This decision led to him being immortalized for work that he later told a colleague ‘ was just done in a weekend ’ while ‘ his best work was largely ignored – his early control work ’ ( Leeson 1994 : 613 ) .

S Keen PC 9.7.jpg9.7 Phillips’s empirically derived unemployment – money – wage – change relation

To do his statistical analysis , Phillips assembled annual data for the UK from 1861 until 1957 from a range of sources . He then used the subset from 1861 till the outbreak of World War I to derive a nonlinear function that appeared to fit the data very tightly ( see Figure 9.7 ) . When he fitted the post – WWI data to this curve , the ‘ out of sample ’ data also had a relatively close fit to his equation ( except for some deviations which he explained as due to negotiated inflation – wage deals between unions and employers , and the impact of World War II on forcing up agricultural prices in Britain ) .

He then summarized his results in the following accurate but poorly considered statement :

Ignoring years in which import prices rise rapidly enough to initiate a wage – price spiral , which seem to occur very rarely except as a result of war , and assuming an increase in productivity of 2 per cent per year , it seems from the relation fitted to the data that if aggregate demand were kept at a value which would maintain a stable level of product prices the associated level of unemployment would be a little under 2 per cent . If , as is sometimes recommended , demand were kept at a value which would maintain stable wage rates the associated level of unemployment would be about 5 per cent .” ( Phillips 1954 : 299 ; emphases added )

To actually achieve the preconditions that Phillips set out here – keeping aggregate demand ‘ at a value which would maintain a stable level of product prices ’ or ‘ at a value which would maintain stable wage rates ’ – would have required a whole host of control mechanisms to be added , even to Phillips’s model of the economy , let alone the real economy itself . As the Goodwin model indicates , a dynamic model of the economy will have endogenous tendencies to cyclical behavior , and these in turn are merely a caricature of the cyclical nature of evolutionary change in a capitalist economy .
Developing these control mechanisms was , as noted , Phillips’s main research agenda , but the economics profession at large , and politicians as well , latched on to this statement as if it provided a simple menu by which the economy could be controlled . If you wanted stable prices ( in the UK ) , just set unemployment to 2 percent ; if you wanted stable money wages instead , set unemployment to 5 percent ; and pick off any other combination you like along the Phillips Curve as well .

This simplistic , static ‘ trade – off ’ interpretation of Phillips’s empirically derived curve rapidly came to be seen as the embodiment of Keynesian economics , and since the 1960s data also fitted the curve very well , initially this appeared to strengthen ‘ Keynesian ’ economics .

But in the late 1960s , the apparent ‘ trade – off ’ began to break down , with higher and higher levels of both inflation and unemployment . Since the belief that there was a trade – off had become equivalent in the public debate to Keynesian economics , the apparent breakdown of this relationship led to a loss of confidence in ‘ Keynesian ’ economics – and this was egged on by Milton Friedman as he campaigned to restore neoclassical economics to the position of primacy it had occupied prior to the Great Depression .

Phillips’s empirical research recurs throughout the development of macroeconomics , as I am about to recount in the next chapter – as Robert Leeson observed : ‘ For over a third of a century , applied macroeconomics has , to a large extent , proceeded from the starting point of the trade – off interpretation of the work of A . W . H . “ Bill ” Phillips . It is hardly an exaggeration to say that any student destitute of the geometry of the Phillips curve would have difficulty passing an undergraduate macroeconomics examination ’ ( Leeson 1997 : 155 ) .

However , even his empirical research has been distorted , since it has focused on just one of the factors that Phillips surmised would affect the rate of change of money wages – the level of employment . Phillips in fact put forward three causal factors :

When the demand for a commodity or service is high relatively to the supply of it we expect the price to rise , the rate of rise being greater the greater the excess demand . Conversely when the demand is low relatively to the supply we expect the price to fall , the rate of fall being greater the greater the deficiency of demand . It seems plausible that this principle should operate as one of the factors determining the rate of change of money wage rates , which are the price of labor services .

When the demand for labor is high and there are very few unemployed we should expect employers to bid wage rates up quite rapidly , each firm and each industry being continually tempted to offer a little above the prevailing rates to attract the most suitable labor from other firms and industries . On the other hand it appears that workers are reluctant to offer their services at less than the prevailing rates when the demand for labor is low and unemployment is high so that wage rates fall only very slowly . The relation between unemployment and the rate of change of wage rates is therefore likely to be highly non – linear .

Phillips then added that the rate of change of employment would affect the rate of change of money wages :

It seems possible that a second factor influencing the rate of change of money wage rates might be the rate of change of the demand for labor , and so of unemployment . Thus in a year of rising business activity , with the demand for labor increasing and the percentage unemployment decreasing , employers will be bidding more vigorously for the services of labor than they would be in a year during which the average percentage unemployment was the same but the demand for labor was not increasing . Conversely in a year of falling business activity , with the demand for labor decreasing and the percentage unemployment increasing , employers will be less inclined to grant wage increases , and workers will be in a weaker position to press for them , than they would be in a year during which the average percentage unemployment was the same but the demand for labor was not decreasing .

Thirdly , he considered that there could be a feedback between the rate of inflation and the rate of change of money wages – though he tended to discount this except in times of war : ‘ A third factor which may affect the rate of change of money wage rates is the rate of change of retail prices , operating through cost of living adjustments in wage rates ’ ( Phillips 1954 : 283 ) .

In subsequent work , Phillips went farther still , and considered that attempts to control the economy that relied upon the historically observed relationship could change the relationship itself : ‘ In my view it cannot be too strongly stated that in attempting to control economic fluctuations we do not have two separate problems of estimating the system and controlling it , we have a single problem of jointly controlling and learning about the system , that is , a problem of learning control or adaptive control ’ ( Phillips 1968 : 164 ; Leeson 1994 : 612 , n . 13 ) .

Phillips didn’t consider the other two causal relationships in his empirical work because , at the time he did it , and with the computing resources available to him ( a hand – operated electronic desk calculator ) , quite simply , it was impossible to do so . But today it is quite feasible to model all three causal factors , and adaptive learning as well , in a modern dynamic model of the kind that Phillips had hoped to develop .

Unfortunately , Phillips’s noble intentions resulted in a backfire : far from helping wean economists off their dependency on static methods , the misinterpretation of his simple empirical research allowed the rebirth of neoclassical economics and its equilibrium methodology – and ultimately , the reduction of macroeconomics to applied microeconomics .”


Steve Keen  Debunking Economics  Zed Books  2011  Part 3 : Complexities: issues omitted from standard courses that should be part of an education in economics. (9) Let’s do the Time Wrap again.  Addendum Misunderstanding Bill Phillips, wages and “the Phillips Curve”   kindle loc 4675 ff , pages 195 -202