Why do central banks in the US, UK and EZ appear to focus on inflation, and ignore unemployment or the output gap? More specifically, why are they apparently not prepared to allow inflation to be above target by, say, 1% for a few years in order to generate a more rapid recovery? There are many potential answers (see, for example, Joe Gagnon), but one that is often mentioned is that excess (above target) inflation would destroy anti-inflation credibility. Here is Bernanke in April (via Felix Salmon, HT Brad DeLong):
“We, the Federal Reserve, have spent 30 years building up credibility for low and stable inflation, which has proved extremely valuable in that we’ve been able to take strong accommodative actions in the last four or five years to support the economy without leading to an unanchoring of inflation expectations or a destabilization of inflation. To risk that asset for what I think would be quite tentative and perhaps doubtful gains on the real side would be, I think, an unwise thing to do.”
What is exactly meant by this? In terms of a Phillips curve this statement seems odd. Over the last 4 years the recession has kept inflation and inflation expectations low. As long as unemployment remains above the natural rate, rising core inflation is not a danger. Is it plausible that QE might cause inflation without any impact on the real side?
The model that might make sense of such statements is the inflation bias literature. Those familiar with this can skip the next two paragraphs. This literature assumes that monetary policy makers are not content with unemployment at its natural rate. Although unemployment at this rate stabilises inflation, involuntary unemployment still exists at this rate because of market imperfections. As a result, with inflation and inflation expectations at target, policy makers will be tempted to raise inflation in order to reduce unemployment below the natural rate. For brevity, let’s call this the ‘overheating policy’.
However, the private sector knows this when setting inflation expectations. If expectations are set at the inflation target, they will be too low, because policy makers will raise inflation to get lower unemployment. The only inflation guess that is rational (by which we mean that it will on average be correct) is an inflation level that is so high that policy makers are no longer tempted to pursue the overheating policy. We therefore get a rational expectations equilibrium where inflation is above target, but unemployment is at its natural rate. This excess inflation is called inflation bias.
This idea was highly influential in persuading academics that independent central banks were a good idea. There are many versions of this argument, but one is that central bankers are less likely than politicians to be tempted to pursue the overheating policy (to reduce unemployment below the natural rate). However, ‘less likely’ is not the same as ‘never will’. So credibility in this context can simply mean the extent to which the private sector believes the monetary authority will not try the overheating policy.
But what, you may ask, has this got to do with our current situation? Today unemployment is clearly above the natural rate. So a central bank that today traded a bit of excess inflation in order to get unemployment a bit closer to the natural rate need not be showing any signs of wanting to pursue overheating. They would simply be optimising the kind of social welfare function that macroeconomists assume all the time. There are no necessary implications for credibility as defined above.
However, perhaps there is a more subtle point. Suppose there are two types of central bank: one that would like to try the overheating policy and one that would not. Credibility is about demonstrating that you are the second and not the first type. The private sector does not know which type the central bank is. In addition, the size of the output gap is uncertain, so evidence is messy. If the private sector is trying to guess which type its own central bank is, it might be more suspicious of a central bank that today raised inflation above target. As we noted above, this behaviour proves nothing, but it does not disprove anything either. However, a central bank that stuck with its inflation target even when unemployment is high is unlikely to also want to try the overheating policy, so this does prove something. Being indifferent to unemployment and strict on inflation demonstrates your credibility.
If this is the argument, I have a suggestion. An even more effective way of establishing credibility, in my view, would be for monetary policymakers to agree to donate all their annual income to charity if inflation turned out to go above target at the same time that unemployment fell below the natural rate. (I am, of course, borrowing an idea from Carl Walsh.) Such a pledge would give a pretty clear indication that they would not intentionally pursue overheating.
Now one might object that the members of the committee could turn out to be unlucky. We might get the combination of excess inflation and below natural rate unemployment because forecasts went wrong. So this is the potential cost of this proposal. But what is the certain cost of the current policy of using unemployment to demonstrate credibility? Happiness research (see, for example, the references here) suggests that something like the salary of a monetary policy maker is required to compensate just one unemployed person for the unhappiness caused by becoming unemployed. So if the policy of using unemployment to reinforce credibility reduces social welfare by the equivalent of just 20 or more extra unemployed, it is more costly than my suggestion. In reality, the current policy stance is reducing social welfare by the equivalent of thousands of additional unemployed.
You may think this suggestion is frivolous, and it is certainly unrealistic. But it is meant to illustrate an important point. Credibility should not be immune from a cost benefit analysis. Economists appreciate that few things have infinite worth, and credibility is not one of them.
Nor do I think credibility is as fragile as central bankers appear to imagine. The idea that a central bank that allowed above target inflation in a situation of high unemployment would suddenly loose all the credibility they had established over the previous 30 years makes no sense.
 By the equivalent of, I mean after netting off the value of any consequent change in inflation. Happiness research also suggests (e.g. Di Tella, MacCulloch and Oswald, AER 2001) that, even if we ignore the unemployed themselves, a 1% fall in unemployment is worth at least as much as a 1% fall in inflation in terms of individual happiness. Add in the convexity we normally assume, and the current policy of (at best) hitting the inflation target but keeping unemployment high is clearly suboptimal.