Price level versus inflation targeting under heterogeneous expectations: A laboratory experiment

Since the 2007 crisis, macroeconomists have been interested in monetary policies that could help with stabilizing inflation and output (Honkapohja, 2015). Two ideas gained particular attention: (i) that inflation should be replaced by the nominal price level (PLT) as the target for the central bank; and (ii) that the central bank should provide explicit guidance about its interest rate rule. We conduct a laboratory experiment to test the validity of these two hypotheses. Our experiment uses a Learning-to-Forecast design based on a simple DSGE economy. Subjects are given a qualitative description of the economy and are asked to predict inflation and output gap two-periods ahead for 50 periods. There are five treatments. Baseline treatment (1) incorporates a standard inflation targeting rule. The other four treatments utilize a PLT Taylor rule and are based on a two-by-two design: a ‘weak’ PLT rule (2) with guidance and (3) without guidance; and a ‘strong’ PLT rule (4) with guidance and (5) without guidance. By guidance we mean that the central bank informs subjects about the price level deviation from its target. We find that subjects within each treatment coordinate on similar behavior, but large differences between the treatments prevail. Guidance has a negligible effect, whereas a weak or strong Taylor rule specification turns out to be crucial for stability. PLT can be a robust monetary policy, but only if it is sufficiently responsive to the deviations of output and prices