Is it time to ditch the natural rate hypothesis?

from Lars Syll

Fifty years ago Milton Friedman wrote an (in)famous article arguing that (1) the natural rate of unemployment was independent of monetary policy, and (2) trying to keep the unemployment rate below the natural rate would only give rise to higher and higher inflation.

The hypothesis has always been controversial, and much theoretical and empirical work has questioned the real-world relevance of the ideas that unemployment really is independent of monetary policy and that there is no long-run trade-off between inflation and unemployment.

Although Olivier Blanchard also has his doubts — after having played around with a ‘toy model’ and looked at the data — he lands on the following advice:

Failure of either of the hypothesis leads to a more attractive trade-off​ between output and inflation, and, in the presence of shocks, suggests a stronger role for stabilization policy. If the independence hypothesis fails, adverse shocks are more costly, and stabilization policy more powerful. If the accelerationist hypothesis fails, there is more room for stabilization policy​ to be used at little inflation cost.

Where does this leave us? It would be good to have a sense of … the specific channels at work. The empirical part of this paper has shown that we are still far from it. Thus, the general advice must be that central banks should keep the natural rate hypothesis (extended to mean positive but low values of b and a) as their baseline, but keep an open mind and put some weight on the alternatives. For example, given the evidence on labor force participation and on the stickiness of inflation expectations presented earlier, I believe that there is a strong case, although not an overwhelming case, to allow U.S. output to exceed potential for some time, so as to reintegrate some of the workers who left the labor force during the last ten years.

My own view on the subject is that the natural rate hypothesis does not hold water simply because the relations it describes have never actually existed. 

The only thing that amazes yours truly is that although this is pretty ‘common knowledge,’  so-called ‘New Keynesian’ macroeconomists still today use it — and its cousin the Phillips curve — as a fundamental building block in their models. Why? Because without it ‘New Keynesians’ have to give up their (again and again empirically falsified) neoclassical view of the long-run neutrality of money and the simplistic idea of inflation as an excess-demand phenomenon.

The natural rate hypothesis approach (NRH) is not only of theoretical interest. Far from it.

The real damage done is that policymakers that take decisions based on NRH models systematically implement austerity measures and kill off economic expansion. The unnecessary and costly unemployment that this self-inflicted and flawed illusion eventuates, is something its New Classical and ‘New Keynesian’ advocates should always be kept accountable for.

According to the  [NRH], unemployment differs from its natural rate only if expected inflation differs from actual inflation. If expectations are rational, we should see as many quarters when inflation is above expected inflation as quarters when it is below expected inflation. That suggests the following test of the [NRH]:

Because a decade contains 40 quarters, the probability that average expected inflation over a decade will be different from average​ actual inflation should be small. If the [NRH] and rational expectations are both true simultaneously, a plot of decade averages of inflation against unemployment should reveal a vertical line at the natural rate of unemployment … This prediction fails dramatically.

There is no tendency for the points to lie around a vertical line and, if anything, the long-run Phillips is upward sloping, and closer to being horizontal than vertical. Since it is unlikely that expectations are systematically biased over decades, I conclude that the  [NRH] is false.

Defenders of the [NRH] might choose to respond to these empirical findings by arguing that the natural rate of unemployment is time-varying​. But I am unaware of any theory which provides us, in advance, with an explanation of how the natural rate of unemployment varies over time. In the absence of such a theor, ​ the [NRH] has no predictive content. A theory like this, which cannot be falsified by any set of observations, is closer to religion than science.

Roger Farmer

So, yes, it is definitely time to ditch the natural rate hypothesis!