The Natural Rate

The case for this stable tradeoff was shattered in the late 1960s and early 1970s. On the theoretical side, the attack took the form of the natural-rate hypothesis of Friedman (1968) and Phelps (1968). Friedman and Phelps argued that the idea that nominal variables, such as the money supply or inflation, could permanently affect real variables, such as output or unemployment, was unreasonable; in the long run, they argued, the behavior of real variables is determined by real forces.

In the specific case of the output-inflation or unemployment-inflation tradeoff, Friedmans and Phelpss argument was that a shift by pohcymak-ers to permanently expansionary policy would, sooner or later, change the way that prices or wages are set. Consider again the example analyzed m Figure 5.16. When poUcymakers adopt permanently more expansionary polices, they permanently increase output and employment, and (with this

*See also Lipsey (1960) and Samuelson and Solow (1960).

wage exceeds the period-1 wage by a factor of Pi/Pq:

Wz APi Wl ~ APo

(5.34)

Tfiis imphes that if the price level in period 2 is the same as in period 1, the real wage is APi/Pi = A, which is the same as the real wage in period 0. Thus employment and output would be the same as they were in period 0. That is, ASz goes through the point (lo,fi); this is shown in the hgure. Thus if policymakers shift the aggregate demand curve out further to ADz, output remains at Yi and the price level rises further to Pz.

This process can continue indefinitely, with the price level continually rising and Y equal to Yi every period. And if policymakers pursue even more expansionary policies, they can keep output at an even higher level, at the cost of higher inflation. Thus the model implies a permanent output-inflation tradeoff. Since higher output is associated with lower unemployment, it also implies a permanent unemployment-inflation tradeoff.

In a famous paper, Phillips (1958) showed that there was in fact a strong and relatively stable negative relationship between unemployment and wage inflation in the United Kingdom over the previous century. Subsequent researchers found a similar relationship between unemployment and price inflation-a relationship that became known as the Phillips curve. Thus there appeared to be both theoretical and empirical support for a stable unemployment-inflation tradeoff.

version of the aggregate supply curve) they permanently reduce the real wage. Yet there is no reason for workers and firms to settle on different levels of employment and the real wage just because inflation is higher: if there are forces causing the employment and real wage that prevail in the absence of inflation to be an equilibrium, those same forces are present when there is inflation. Thus wages will not always be adjusted mechanically for the previous periods inflation. Sooner or later, they will be set to account for the expansionary pohcies that workers and firms know are going to be undertaken. Once this occurs, employment, output, and the real wage wiU return to the levels that prevailed in the absence of inflation.

In short, the natural-rate hypothesis states that there is some "normal" or "natural" rate of unemployment, and that monetary policy cannot keep unemployment below this level indefinitely. The precise determinants of the natural rate are unimportant. Friedmans and Phelpss argument was simpl> that it was determined by real rather than nominal forces. In Friedmans famous definition (1968, p. 8):

"The natural rate of unemployment".. .is the level that would be ground out by the Walrasian system of general equilibrium equations, provided there is embedded in them the actual structural characteristics of the labor and commodity markets, including market imperfections, stochastic variability in demands and supplies, the cost of gathering information about job vacancies and labor availabilities, the costs of mobility, and so on.

The empirical downfall of the stable unemployment-inflation tradeoff is illustrated by Figure 5.17, which shows the combinations of unemployment and inflation in the United States from 1961 to 1994. The points for the 1960s fall along a fairly stable downward-sloping curve. The points since then do not.

One source of the empirical failure of the Phillips curve is mundane: if there are disturbances to aggregate supply rather than aggregate demand, then even the models of the previous section imply that high inflation and high unemployment can occur together. And there certainly are plausible candidates for significant supply shocks in the 1970s. For example, there were tremendous increases in oil prices in 1973-74 and 1978-79; such increases are likely to cause firms to charge higher prices for a given level of wages. To give another example, there were large influxes of new workers into the labor force during this period; such influxes may increase unemployment for a given level of wages.

Yet these supply shocks cannot explain all of the failings of the Phillips curve in the 1970s and 1980s. In 1981 and 1982, for example, there were no identifiable large supply shocks, yet both inflation and unemployment were much higher than they were any time in the 1960s. The reason, if Friedman and Phelps are right, is that the high inflation of the 1970s changed how prices and wages were set.

Thus, the models of price and wage behavior that imply a stable relationship between inflation and unemployment do not provide even a moderately

4 5 6 7 8 9

Unemployment (percent)

FIGURE 5.17 Unemployment and inflation in the United States, 1961-1994 (data from Citibase)

accurate description of the dynamics of inflation and the choices facing policymakers. They must therefore be modified if they are to be used to address these issues.

The Expectations-Augmented Phillips Curve

[n analyzing the long run, it is easiest to state directly that prices and wages are fully flexible, so that changes in aggregate demand have no real effects. Thus the long-run aggregate supply (or LRAS) curve is vertical, and disturbances on the demand side of the economy do not affect output in the long run. The level of output at which the long-run aggregate supply curve is vertical is known as the natural rate of output, or potential or full-employment output, and is denoted Y. This is shown in Figure 5.18.

The conclusion that the long-run aggregate supply curve is vertical does not answer the question of how to model aggregate supply in the short run. Modern Keynesian formulations of short-run aggregate supply differ from the simple models in equations (5.31)-(5.33) and in Section 5.4 in three ways. First, neither wages nor prices are assumed to be completely unresponsive to the current state of the economy. Instead, higher output is assumed to be associated with higher wages and prices. One implication is that the short-run aggregate supply curve is upward-sloping even if it is prices rather than wages that do not adjust immediately to disturbances. Second, the possibility of supply shocks is allowed for. Third, and most important, adjustment