Our approach to portraying policy frontiers differs slightly from that commonly found in the literature, where interest rate volatility is included in the objective function and each frontier is drawn using a different weight on interest rate volatility. The standard approach combines information on model-imposed constraints on policy with policymakers’ preferences regarding funds rate volatility. We prefer to maintain the strict distinction between policy constraints— given by the model—and preferences. In this paper, each frontier represents the best obtainable combinations of output and inflation variability for a specified level of funds rate volatility.

We use the estimated rule given in equation (1) as a benchmark for funds rate volatility. For each model, we impose the estimated rule and compute the standard deviation of the one-quarter change in the funds rate. The outcomes from this experiment provide reference points for comparisons across the different models. We refer to frontiers constructed under the restriction on funds volatility that results from the estimated rule as E-frontiers. This normalization is needed because each model is estimated over different sample periods and generates quantitatively different distributions for the endogenous variables. For example, the moments for the FRB and TAYMCM models depend in part on shocks from the 1970’s-a period of relative economic upheaval—while the shocks for the FM and MSR models are from the relatively tranquil 1980’s and early 1990fs. This difference in samples implies significant differences in interest rate volatility across the four models for a given policy rule.

Our approach to portraying policy frontiers differs slightly from that commonly found in the literature, where interest rate volatility is included in the objective function and each frontier is drawn using a different weight on interest rate volatility. The standard approach combines information on model-imposed constraints on policy with policymakers’ preferences regarding funds rate volatility. We prefer to maintain the strict distinction between policy constraints— given by the model—and preferences. In this paper, each frontier represents the best obtainable combinations of output and inflation variability for a specified level of funds rate volatility. We use the estimated rule given in equation (1) as a benchmark for funds rate volatility. For each model, we impose the estimated rule and compute the