We all, or nearly all, consent,

            when wages rise by ten percent,

            it leaves the choice before the nation,

            of unemployment or inflation.

Kenneth Boulding

            Our general Aggregate Demand/Aggregate Supply framework hints at a possible negative relationship between unemployment and inflation: Boosting Aggregate Demand raises real output and employment when substantial productive capacity is idle, but ever greater upward pressures on prices and wages emerge as excess capacity disappears. Are such trade-offs between unemployment and inflation (if they truly even exist) strictly short-run phenomena, or do they persist into the long-run? These issues are critical for sound macroeconomic policymaking.

 

            More than three decades ago, A. W. Phillips plotted unemployment data for Great Britain during the preceding century against corresponding data for wage inflation. Curves that seem to fit such data have become known as Phillips curves.

The Phillips curve portrays an inverse statistical relationship between the rate of inflation and the unemployment rate.

Phillips's work focused on wage inflation, but later researchers addressed apparent trade-offs between unemployment and price inflation. For example, reducing inflation to zero might require unemployment of 8 percent, while pressing unemployment below 5 percent might require 6 percent annual inflation.

 

Productivity and Inflation

 

A fairly tight relationship links nominal wage inflation, price inflation, and productivity:

p = w - l

where p is the percentage rate of inflation, w is the percentage change in wages, and l is the percentage change in productivity.

 

            For example, if workers produce 5 percent more this year than last, then wages can rise 5 percent before firms feel pressure to maintain profit margins by raising prices. This relationship is illustrated in Figure 1 where a Phillips curve is superimposed over U.S. data for the 1960s. An annual rate of productivity gain (l) of 2 percent (based on rough historical averages) is assumed and is reflected by shifting the scale for price inflation (p) down by 2 percent relative to wage inflation (w). Thus, 4 percent wage hikes combined with 2 percent productivity gains only impose pressures for prices to rise 2 percent. But if productivity gains are nil and workers successfully demand higher wages, the likely result is price inflation.

 

Figure 1  A Phillips Curve for the 1960s (U.S. Data)

Phillips curves suggest that if Aggregate Demand is high and rising steadily, inflation will be substantial but unemployment will be low. On the other hand, consistently low Aggregate Demand yields little inflation but more unemployment. Wages may increase faster than prices if productivity rises.  Overall, movements during 1959 to 1969 reflected the expansionary fiscal and monetary policies of this era.

Note: Productivity is assumed to have grown by 2 percent annually, so the two vertical axes are calibrated differently.

Source: Economic Report of the President, 1994.

 

The 1960s: An American Phillips Curve?

 

According to the Keynesian theory that dominated economic thought by 1960, Aggregate Supply should be relatively stable. Thus, a stable Phillips curve would allow Aggregate Demand to be "fine-tuned" to yield the least harmful mix of unemployment and inflation. The dilemma seemed clear: If high inflation is associated with low unemployment, and vice versa, then the Phillips curve is a "menu" of the trade-off facing  policymakers who try to minimize our economic woes.[1]

                          The Phillips curve in Figure 1 was thought roughly reflective of the choices facing U.S. policymakers. At a target of zero price inflation, workers would suffer roughly 6 to 8 percent unemployment. Unemployment rates as low as 3 to 4 percent were thought feasible only with inflation of 4 to 6 percent. Policymakers would have preferred less of each, but Phillips curve analysis suggested that this was impossible.  

Shifts in Aggregate Demand:  A Stable Trade-Off?

National Output and Income are closely tied to employment. In turn, a close negative relationship exists between employment and unemployment---higher employment generally means lower unemployment, and vice versa. But changes in the size or composition of the labor force may obscure this relationship. For example, rapidly rising labor force participation rates might precipitate higher frictional unemployment even if real output and employment grew.

            An Aggregate Supply curve is a positive relationship between national output and the price level, while a typical Phillips curve depicts an inverse correlation between unemployment and inflation. Thus, Phillips curves roughly mirror Aggregate Supply curves, as shown in Figure 2. Points a, b, . . . , g in Panel A roughly correspond to points a, b, . . . , g in Panel B. Expanding Aggregate Demand from ADa to ADb in Panel A moves the economy from point a to point b in Panel B. Keeping the economy at point b requires increasing Aggregate Demand by the same proportion in each subsequent period. Similarly, expanding Aggregate Demand from ADa to ADc moves the economy to point c in Panel B; staying at point c would require continuous similar expansions of Aggregate Demand.[2] And so on, for points d, e, f, and g.

Figure 2  Aggregate Supply and the Phillips Curve

Ever larger sustained increases in Aggregate Demand in Panel A along a stable Aggregate Supply curve would trace out a series of equilibria like a, b, . . . , g as employment and output increased. Unemployment rates would fall during these movements, but prices would rise, tracing a pattern roughly like a, b, . . . , g along the stable Phillips curve in Panel B. Notice that the price level would be stable at points c in both panels.

 

The 1970s: Stable Phillips Curves at Bay?

The idea that the Phillips curve was stable unraveled in the 1970s. Four Phillips curves are drawn through 1970s data in Figure 3. It seemed that ever greater inflation had to be endured to keep unemployment in the 5 to 6 percent range. Why did the Phillips curve worsen (shift rightwards)?

 

Figure 3  The Shifting Phillips Curve Relationship

Adverse shifts of Aggregate Supply (due to higher energy prices and diminished productivity growth) caused the Phillips trade-off to worsen between the 1960s and early 1970s, between the early and mid-1970s, and again between the late 1970s and early 1980s.

Source: Economic Report of the President, 1994.

Supply Shocks and Unstable Phillips Curves

Continually expanding the growth rate of Aggregate Demand while Aggregate Supply remains stable results in upward movements along a stable Phillips curve, as in Figure 2. The 1960s seemed to provide evidence of a roughly stable Phillips curve, but bouts of stagflation in the 1970s betrayed the naïveté of this view. Shocks to Aggregate Supply apparently worsened the short-run trade-off between unemployment and inflation, as shown in Figure 4. Suppose Aggregate Supply began to shrink, declining from AS0 to AS1 to AS2, and so on, in Panel A. Output and employment fall while the price level rises (stagflation), confronting policymakers with ever harsher short-run options---curves PC0, PC1, PC2, and so on in Panel B. As Phillips curves deteriorate, achieving a given low unemployment rate entails ever more rapid inflation.

Figure 4  Shifting the Phillips Curve

Shifts in Aggregate Supply, as shown in Panel A, cause shifts in Phillips curves (PCs), as shown in Panel B, and account for the stagflation of the 1970s and early 1980s.

 

            Since its discovery, the Phillips curve has been in the eye of a stormy controversy. Keynesians initially heralded Phillips's discovery of a negative statistical relationship between unemployment and inflation as presenting policymakers with a stable trade-off. But economists of a classical bent dismiss the Phillips curve as a short-run artifact, and hold that policymakers who try to swap inflation for lower unemployment will soon find the relationship illusory. Their natural rate theories suggest that, in the long run, macroeconomic policies affect only absolute prices--- not such "real" variables as unemployment, output, or interest rates. We need to explore these differing interpretations of why Phillips curves may shift so that an economy experiences stagflation, as the United States did in the 1970s.

---