Inflation target regimes provide interesting new monetary policy experiments , but s o far the literature has been mainly descriptive and institutional. In Discussion Paper No . 1249, Research Fellow Lars Svensson provides a more analytical treatment of inflation target regimes, and in particular relates them to the literature on rules and discretion in monetary policy. An inflation target regime is interpreted as having both an explicit inflation target and an implicit output/unemployment target. Thus, inflation is not the only target for monetary policy, and a loss function for the central bank is hence interpreted as including both the loss from deviating from the explicit inflation target as well as the loss from deviating from the implicit output or unemployment target.

The paper considers both the standard static situation when there is no persistence in output and unemployment, as well as the realistic dynamic situation when out put and unemployment persists. Without output/unemployment persistence, delegation of monetary policy to a central bank acting under discretion with an optimal inflation target can eliminate the discretionary inflation bias, mimic the optimal linear inflation contract and achieve the equilibrium corresponding to an optimal rule with commitment. The optimal inflation target is equal to the socially-best inflation rate less the inflation bias. Thus an `inflation target-conservative' central bank dominates a `weight-conservative' central bank with increased weight on inflation stabilization, since the latter reduces the inflation bias but suboptimally increases output/unemployment variability. With output /unemployment persistence , a constant inflation target is equivalent to a constant linear inflation contract. They can both eliminate the average inflation bias but not the state-contingent part of the inflation bias. Inflation variability is too high, and output variability too low, compared to the equilibrium corresponding to an optimal rule. An optimal state-contingent inflation target can remove all inflation bias, but in contrast to an optimal state-contingent linear inflation contract , it still leaves inflation variability too high. Delegation with an optimal state-contingent inflation target to a `weight-conservative' central bank can then achieve the equilibrium corresponding to an optimal rule. Inflation targets may on average be exceeded, and they may have imperfect credibility. Nevertheless they may usefully reduce inflation, and they appear much easier to implement than linear inflation contracts.

Optimal Inflation Targets, `Conservative' Central Banks, and Linear Inflation Contracts
Lars E O Svensson

Discussion Paper No. 1249, October 1995 (IM)