A Worst--Case Approach to Inflation Zone Targeting

In this paper we present an algorithm for continuous minimax problem where a quasi--Newton direction conditional on appropriate maximizers is used. The direction involves a quadratic subproblem to compute the minimum norm subgradient. An application of the algorithm to a monetary policy design is given. A simple model, due to Orphanides and Wieland, is used for practice of inflation zone targeting. In this paper, however, the approach is different as we minimize the worst--case with respect to inflation and economic activity. Also a comparison with the $H^infinity approach is included.