Jakob Wilhelmus lores
Jakob Wilhelmus
Associate Director, International Finance and Macroeconomics Research
Capital Flows and Finance and Regulation and Systemic Risk
Jakob Wilhelmus is an associate director in international finance and macroeconomics at the Milken Institute. He studies topics relating to systemic risk, capital flows, and investment. Concentrating on market-level information, his work focuses on identifying and analyzing financial data to produce a better understanding of the behavior and underlying structure...
read bio

ECON 101: Revisiting the Phillips Curve

By: Jakob Wilhelmus
February 25, 2016

In 1958 A.W. Phillips showed a stable, negative correlation between unemployment rates and nominal wages.[4] A typical Phillips curve follows a relation such as depicted in Figure 1. A high (low) unemployment rate was associated with low (high) nominal wages.

FIGURE 1: 1950—1970

This relationship held up exceptionally well for quite some time, but as the U.S. economy entered a period of stagflation—high unemployment and high inflation—in the 1970s, the relation seemed off track (see Figure 2).

Was the Phillips Curve, even though persistent, just an accidental pattern in the data rather than a fundamental economic structure? According to research during the late 1960s, the answer is both yes and no. Both Milton Friedman and Edmund Phelps had argued that the general idea of the Phillips Curve was correct, but it was concentrating on the wrong determinant, as workers are not concerned about their nominal wage but instead their real wage—what is left after taking into account future inflation. Friedman introduced what he called the natural rate of unemployment, an equilibrium level around which the short-term unemployment rate would deviate. He argued that any level below (above) this natural level indicates an excess demand (supply) of labor that will lead to rising (falling) real wages, thereby moving the unemployment rate towards its equilibrium.[5] This led to an adjustment of the curve, the so-called expectation-augmented Phillips Curve, where the curve moves inward (outward) depending on the inflation expectations of households and businesses. The new curve allows for a simple explanation of the ‘70s hiccup: the abolishment of the gold standard in 1971 had caused workers to anticipate higher inflation, leading to the joint movement of inflation and unemployment.


FIGURE 2: 1970—1980

The expectation-augmented Phillips curve does explain the homogenous rise (fall) of inflation and unemployment based on the changing expectations of future inflation. However, the relation has not been as stable as it was in its earlier days, as can be shown by the below graph covering the period 2000—2015. The Phillips curve is almost equivalent to a horizontal line, making the interpretation of today’s inflation-unemployment relation much more complicated.

Figure 3: 2000—2015

One explanation of this horizontal Phillips curve is the increased uncertainty of households and businesses about expected inflation. The years 2001 to 2008 were accompanied by consistent economic growth, and the fitted curve over this time span (green line) therefore observes the classical inverse relation of unemployment and inflation. However, with the financial crisis and economic turmoil that followed, inflation expectations became highly volatile, and the trend from 2009 to 2015 shows a positive relation (blue line) similar to Figure 2.

At the same time, one has to keep in mind the difficulty of measuring the unemployment rate, as this depends on many other economic developments, such as the labor participation rate. This point based on the current movement of the two indicators: can we expect inflation to increase due to excess labor demand, or is the dropping unemployment rate actually representative of the declining labor participation rate? The most recent data don’t provide a clear answer.

This does not mean, however, that the Phillips curve and models that depend on its assumption, e.g. most dynamic stochastic general equilibrium (DSGE) models (those used by the Fed and other central banks worldwide), are not useful theoretical tools. The dynamics of the inflation-unemployment tradeoff, under the right conditions, hold in both the long and short term. Nevertheless, the problem of the Phillips Curve is one that is very common: it works quite well until it doesn’t. The Phillips curve as a forecasting tool should be viewed cautiously, as demonstrated by the consistency of forecasting errors by officials.


[1] The question on the Phillips model starts around minute 43.


[3] The following is limited to the simple idea behind the concepts of the Phillips curve for the sake of argument. For a more detailed look at the Phillips curve and its use in the New Keynesian Model, see the influential paper by Clarida, et al., “The Science of Monetary Policy: A New Keynesian Perspective”.

[4] Phillips based his observation on data from the UK, but the relationship was soon found to hold for other economies, in general replacing wages with price inflation, and is nowadays accepted as a sort of short-term fundamental economic relation.

[5] The short-term deviation is mostly driven by what is called a money illusion, in that workers do not adjust right away for the fact that prices are increasing at a higher rate than their wages.

PHOTO CREDIT: HOUSEKEEPING AT THE RESERVE BANK... "Keep still dammit!" Inflation. Sunday News, 5 August 2000. Walker, Malcolm, 1950-: [Digital cartoons published from 2000 in Sunday News]. Ref: DCDL-0009203. Alexander Turnbull Library, Wellington, New Zealand.