RDP 9701: Inflation Regimes and Inflation Expectations 5. Interpretations and Extensions

The basic point of this paper is a stark one: monetary regimes with inflation targets that are quite different from the average inflation rate across previous regimes may never be fully credible to long-term financial markets. This lack of credibility is not necessarily due to slow learning by private agents or to a lack of resolve on the part of the central bank. Even when agents understand and believe in the central bank's target inflation rate, they must attach some probability to a change in the regime. For example, the central bank governor may die or resign, or the government may change the institutional framework of monetary policy. There is no way to guarantee that these things will not happen.

The key to credibility over the long term is the expected value of inflation under the next regime. This paper considers two approaches to modelling expectations of inflation under the next regime. In the first approach, it is assumed that agents know the constant mean inflation rate across regimes. If the current regime inflation target is equal to this long-run inflation mean, then policy is credible in the long run. If the current regime inflation target is far from the long-run inflation mean, then policy is not credible and policy will never become credible no matter how long the current regime lasts or how often similar regimes arise. In the second approach, agents update their expectations about inflation in the next regime based on inflation in the current and previous regimes. Under this approach, a sequence of low (or high) inflation regimes would change agents' expectations of inflation in the next regime. However, as demonstrated in a simple example, significant changes in long-run inflation expectations may still require a very long time.

One plausible extension of these models is to consider learning on the part of the central bank. For example, one may argue that the high inflation of the 1970s was a mistake, that central banks have learned their lessons, and that the public understands that this episode will not recur. Under this hypothesis, agents ought to place more weight on recent inflation rates when forming expectations about inflation in the next regime, and long-run credibility would be easier to obtain than in the basic models. Nevertheless, as long as agents place some positive weight on past inflation targets in forming expectations about future inflation targets, the credibility problem will remain.

Another extension of the model would be to consider variation over time in the probability of a regime shift. One way to increase the long-run credibility of the current inflation target is to take steps to reduce the probability of a regime shift. Recent attempts in many countries to increase the independence of the central bank may be interpreted as reducing the probability of a regime shift and thus strengthening credibility. Nevertheless, it is not possible to guarantee that any regime will last forever.

Although the hypothesis of central bank learning seems plausible and many central banks have achieved greater independence in recent years, I would like to conclude the paper by noting several caveats. First, the persistence of inflation in some countries that have a long history of inflation (particularly in less developed countries) argues for caution about the idea that a bad experience with inflation innoculates one against future inflation. At the very least, one should keep in mind that lessons learned may become lessons forgotten. Second, even if central banks have learned their lessons well and permanently, the public may be skeptical and the time needed to convince the public may be measured in decades rather than years. Third, even if one does discount the possibility of a return to double-digit inflation, it is harder to justify ignoring the possibility of a return to moderate inflation rates of around 5 per cent or so. In light of the fact that no one is recommending a regime with negative inflation rates, an inflation target that is very close to zero can never be credible in the long run as long as there is some possibility of a return to positive inflation.