An Austrian rehabilitation of the Phillips curve.

• Author: Mulligan, Robert F.
• Position: Essay

William Niskanen (2002) estimated a Phillips curve for the United States using annual 1960-2000 data. By adding one-year-lagged terms in unemployment and inflation, he was able to show that this familiar equation is misspecified. In his improved specification, Niskanen found that the immediate impact of inflation is to reduce unemployment, confirming the traditional understanding of the Phillips-curve relationship, but also finding that after an interval as short as one year inflation has generally been followed by increased unemployment. Though Niskanen was perhaps unaware of it, his results lend strong support to the Austrian model of the business cycle. In that model, credit expansion results in a temporary but unsustainable expansion. Unemployment is lowered in the short run, but once the policy-induced malinvestment is recognized, total output and income will be permanently reduced, and unemployment will increase.

Beyond reinterpreting Niskanen's results, this article estimates a similar model using an expanded monthly 1948-2009 dataset. In particular, the article presents an improved specification with a statistically better-motivated and more-encompassing lag structure. Insights and cautions suggested by Reichel (2004), who estimated vector-error-correction (VEC) models for a variety of different countries, and by Moghaddam and Jenson (2008), who estimated a respecified error-correction-model (ECM), are also addressed.

Background

The Phillips curve is the purported relationship between inflation and unemployment. In his study of U.K. wage inflation and unemployment, A.W.H. Phillips (1958) found a consistent inverse relationship between unemployment and wage inflation--when unemployment was high, wages increased slowly; when unemployment was low, wages rose rapidly. Phillips conjectured that the lower the unemployment rate, the more firms needed to raise wages to attract scarce labor.

At the height of the Phillips curve's popularity as a guide to policy, Edmund Phelps (1967) and Milton Friedman (1968) independently challenged its theoretical foundations. They argued that nominal wages were largely irrelevant, and that worker behavior responded only to real--that is, inflation-adjusted--wages. In their view, real wages would adjust to make the quantity supplied of labor equal to the quantity demanded, and the unemployment rate would then stand at a level uniquely associated with that real wage--the "natural rate" of unemployment, often also called the "non-accelerating inflation rate of unemployment" or NAIRU.

In the expectations-augmented Phillips curve proposed by Friedman and Phelps, unanticipated inflation results in a temporary depression of the real wage, making labor a relatively cheap factor of production, and facilitating lowered unemployment. This short-run tradeoff between inflation and unemployment disappears as soon as workers learn to expect the prevailing rate of price inflation and start demanding higher nominal wages. When workers thus restore the real wage to its pre-inflation level, labor ceases to be an especially cheap resource, and unemployment rises back to its natural rate.

The Friedman-Phelps critiques of the Phillips curve failed to consider the impact of "Cantillon effects" of expansionary policy. Monetary expansion, or expansionary fiscal policy such as public works programmes, both increase demand for output, and therefore demand for labor, in particular sectors at the expense of others. The higher real wage in the initially favored sectors accompanies a reallocation of resources, including labor, to those sectors. The higher real wage is spread out throughout the economy as those workers boost demand in consumption sectors, often geographically near the location of the industries that first benefit from the Cantillon effect. At each successive wave of spending, the increase in the real wage is dissipated, until it is overcome by the general increase in prices, which rise to meet it throughout the economy, and eventually rise beyond the average increase in nominal wages introduced by the expansionary policy.

During the 1970s, the Phillips curve became badly discredited as a policy guide, as the experience of protracted stagflation unambiguously gave the lie to government attempts to exploit this supposed trade-off between inflation and unemployment, leading to more of both. It is thus somewhat curious that that under the guise of the so-called Keynesian resurgence, the Phillips curve is again being invoked to justify expansionary monetary and fiscal policy.

Niskanen's Auto Regression Distributed Lag (ARDL) Specification

Niskanen (2002) suggests equilibrium unemployment ([U.sup.*]) should be rendered as a positive function of lagged inflation:

(1) [U.sup.*] = a - bI + [cI.sub.-1] + u.

This becomes the natural rate of unemployment or NAIRU...