Conclusions: What Have We Learned?

Our case studies indicate that both the adoption of inflation targets and the design choices for that framework have made a difference in the operation of monetary policy. The design choices of the targeting countries have tended to converge over time with regard to the operational design questions posed in Part II, suggesting that a consensus is emerging on best practice in the operation of an inflation-targeting regime. Where the design choices have differed, however, the experiences in the countries examined provide some insight about what has resulted from the different choices. In general, the public announcement of numerical targets for inflation has been very effective in balancing the needs for transparency and flexibility in monetary policy. The areas of operational design that show a convergence of practice include the use of inflation as the target variable. Despite all the rhetoric associated with the pursuit of price stability, all the targeting countries examined here have chosen an inflation target--ranging from 0 to 4 percent annual inflation--rather than a price-level target. This choice reflects concerns that a price-level target may require deflation when prices overshoot the target, an outcome that could entail far higher costs in output losses than are acceptable. Reversals of past target misses, which would be required by a price-level target, do not appear to be necessary for the maintenance of low inflation. Relatedly, targeting countries that have chosen target values for inflation greater than zero make the possibility of deflations less likely. It is-important to emphasize that maintaining an inflation target at a level even somewhat greater than zero for an extended period, as the Bundesbank has done, does not appear to lead to instability in inflation expectations or diminished central bank credibility. Even with a positive inflation target, admission of occasional errors does not appear to be damaging. These design choices are also consistent with building a high degree of flexibility into the inflation-targeting regimes in all the countries studied here, in which central bankers do demonstrate concern about real output growth and fluctuations. This is seen particularly in the gradualism all targeting countries have exercised when disinflating, as well as in the treatment by some countries of the inflation target's (implicit or explicit) floor on price movements as a stabilizing factor. While the targeting countries differ in the degree to which they emphasize particular indicators of inflation in their decision making, all rely on an inclusive information framework untied to specific intermediate target variables. All of these design choices support the contention in Bernanke and Mishkin (1997) that inflation targeting should be seen as a framework rather than a rule. In addition, all of the targeting countries allow for deviations from their targets in response to supply shocks. Usually, the central bank will take action at its own discretion, when such a response is not already built into the target definition, and then explain its actions. Only in New Zealand has an explicit escape clause been invoked to justify such actions, although the Reserve Bank of New Zealand has also engaged in the more discretionary forms of response. Actual inflation targets have been moved over time by all targeting countries, whether up--as in the case of Germany after the 1979 oil shock or New Zealand after the 1996 election--or down--as in all countries considered as disinflations proceeded. As long as target movements are announced sufficiently far in advance, there is no sense that the target is being moved to meet the short-run outcome; target movements are perceived as adaptations to economic conditions. The key to the exercise of discretion in a disciplined manner has been the central banks' ability to convey to the public the distinction between movements in trend inflation and onetime events. …