It’s Time to Bury the Phillips Curve

The stagflation of the 1970s, during which high and variable inflation coincided with high unemployment, should have put the final nail in its coffin. Yet, central bankers continue to incorporate the Phillips curve in their forecasting models, which failed to predict the Great Recession, and are scratching their heads over the question of why inflation has remained so low when the economy is near full employment.

The world's most important central banker, Janet Yellen, in her June 14 press conference, noted that the "Phillips curve appears to be quite flat," meaning "that inflation doesn't respond very much or very quickly to movements in unemployment." However, she still believes "that relationship remains at work."

Likewise, Mark Carney, governor of the Bank of England, said in a recent lecture at the IMF that "the global Phillips curve appears alive and well," even though "globalization has been accompanied by a weakening in the relationship between domestic slack and domestic inflation."

The Phillips curve tells us nothing about the monetary transmission process — that is, how newly created base money (currency plus reserves) affects the broader monetary aggregates and nominal income. Therefore, it does not inform us about why the quadrupling of the Fed's balance sheet since 2008, and the creation of trillions of dollars of base money, has not been inflationary.

Preoccupation with the Phillips curve has diverted attention from the fact that we have entered a new monetary regime in which interest on excess reserves (IOER), macroprudential regulation, and regime uncertainty have significantly weakened the impact of a given increase in base money on the broader money supply and, hence, on inflation.

The original Phillips curve was specified in terms of changes in nominal, not real, wages, and failed to take account of expectations. It also ignored the concept that later became known as the "natural rate of unemployment" — namely, the rate consistent with equilibrium real wages, full employment, and steady (nonaccelerating) inflation. Milton Friedman and Edmund Phelps, both recipients of the Nobel Prize in economics, predicted that if inflation is fully anticipated, it will be independent of the natural rate of unemployment and the long-run Phillips curve will be vertical.

Their research returned common sense to monetary theory by emphasizing that inflation is determined by the growth rate of the money supply, not by the state of labor markets. Nevertheless, with activist policy and a lack of transparency, it is conceivable that trade-offs between inflation and unemployment could occur in the short run. However, once people fully anticipate inflation, they will revert to their normal job search behavior and unemployment will return to normal (i.e., consistent with market forces).

Finally, when Robert Lucas and others formulated the theory of rational expectations — which assumes that individuals take account of all information, including on likely future policy responses, and that markets always clear — no workable or exploitable trade-off between inflation and unemployment can occur, even in the short run.
In reality, there is a knowledge problem and policymakers can still surprise markets, especially in the absence of robust rules.

Moreover, by the 1970s, the Phillips curve had become upward sloping, highlighting its instability. Indeed, stagflation sent an arrow into the heart of the conventional Phillips curve.

The experience since the global financial crisis has once again confirmed the limits of monetary policy to stimulate the real economy. Unemployment has come down but mostly because it always does so following a deep recession as market forces pick up.

Instead of continuing to hold onto the presumption of a trade-off between inflation and unemployment, policymakers should focus on the misallocative and distributive effects of unconventional monetary policy. The Fed needs to recognize that (1) the Phillips curve is not a reliable policy compass, (2) there are limits to monetary policy, and (3) an appropriate (nonactivist) rules-based regime would help reduce uncertainty and so lead to better investment decisions.

A monetary policy based on short-run activism, such as that implied by a "data-dependent" Fed, is not a good substitute for one based on a transparent, long-run strategy guided by a robust monetary rule.

Congress could help by providing the Fed with a rule, such as maintaining the growth path of nominal GDP, so as to avoid downturns such as the one in 2008. In turn, the Fed needs to normalize its balance sheet, end IOER, and use open-market operations to control the growth of base money, which will feed through to the broad money supply and nominal income.

Meanwhile, by eliminating regulatory and other impediments to free markets, especially high marginal tax rates on capital, Congress would lay the ground for higher future growth of real GDP.

And in the meantime, may the Phillips curve RIP.