On Robustness of Average Inflation Targeting

Abstract

This paper considers average inflation targeting (AIT) policy in a New Keynesian model with adaptive learning agents. There are stability concerns regarding AIT when agents have imperfect knowledge and the averaging window length is not public knowledge. These stability risks near the inflation target steady state would likely be avoided under inflation targeting (IT) or price level targeting (PLT). Near the zero interest rate steady state, AIT under-performs PLT and does not necessarily outperform IT. Communicating the averaging window length or adopting an asymmetric average inflation target that judges below-target average inflation more negatively avoids these pitfalls.