Introduction

Most central banks nowadays gear monetary policy directly towards maintaining a low and stable rate of inflation (IMF 2005: chap. 4). There are, of course, important differences between central banks. While some are explicit inflation targeters, others, such as the European Central Bank, have a numerical definition of price stability as the overriding objective of monetary policy (Gerlach and Schnabel 2000). In either case, however, an understanding of how the public forms inflation expectations is of crucial importance for policymakers.

From the 1970s onwards the idea that expectations are rational has dominated much of the literature. Lately a new view on expectations has emerged, which views economic agents as econometricians when forecasting (an extensive overview of this literature is provided by Evans and Honkapohja 2001). This approach, referred to as adaptive learning, assumes that economic agents are boundedly rational but employ statistical forecasting techniques, which allows for the possibility of a rational expectations equilibrium to be learnt in the long run. One important insight from the adaptive learning literature is that policies that may be optimal under rational expectations are not optimal when individuals use a learning process. Orphanides and Williams (2005) show that the optimal monetary policy under a learning process should respond more aggressively to inflation and become more focused on inflation stability than if expectations were rational, since tight inflation control can facilitate learning and provide better guidance for the formation of inflation expectations.