Uncertainty in forecasting inflation and monetary policy design: Evidence from the laboratory

This paper designs a laboratory experiment for studying subjects' uncertainty regarding inflation in different monetary policy environments. We find that the contemporaneous Taylor rule produces a lower uncertainty and higher accuracy of interval forecasts than the forward-looking Taylor rule. The latter also produces a lower uncertainty when the reaction coefficient is high, 4, than rules with lower reaction coefficients, 1.5 and 1.35. Subjects perceive the underlying inflation uncertainty correctly in only 60% of cases, and tend to report asymmetric confidence intervals, perceiving a higher level of uncertainty with respect to inflation increases.