Discretionary Policy, Potential Output Uncertainty, and Optimal Learning

We compare inflation targeting, price level targeting, and speed limit policies when a central bank sets monetary policy under discretion, and must learn about the level of potential output over time. We show that if the central bank learns optimally over time, a speed limit policy dominates [is dominated by] a price level target if society places a high [low] weight on inflation stability. Inefficient learning on the part of the central bank can radically change this conclusion. A speed limit policy is favoured if the central bank places too much weight on recent data when estimating potential output, while a price level target is favoured if the central bank places too much weight on historical data.