Survey_Orphanides_2003

Abstract

Given the noise in the real-time data from 1965 to 1993 exhibited in figures 3 and 4, one would expect at least some differences in the performance of activist policy rules once the imperfections in the data are accounted for. The next step is to investigate the quantitative magnitude of these differences. In parallel with the earlier simulations, the counterfactual simulations based on the real-time data employ the historical values of all variables up to 1965Q4 as initial conditions and the estimated residuals from equations (8) and (9) from 1966:1 to 1993:4. In addition, in determining the interest rate setting these simulations employ the noise in the historical data, xt, and zt. Thus, in every quarter, the policymaker is assumed to set the interest rate responding to the perceived simulated paths of inflation and output which equal the true simulated path minus the historical noise for that quarter.