In this paper we simulate a central bank subject to the misperception that prices are indexed to past inflation in periods when firms are unable to re-optimise. It thinks, in other words, that inflation is intrinsically persistent. The central bank sets monetary policy optimally subject to this belief. The central bank updates its beliefs about in¬dexation using a constant gain learning scheme. The data generated by such policy lead to beliefs about inflation persistence being effectively self-confirming in a wide variety of setttings. These results offer a tentative answer to why it appears that inflation is persistent at some times and in some countries, and at others not. The answer is that policymakers sometimes believe inflation to be persistent, and sometimes do not.
