Figure 6 provides further information on the benefits of interest rate smoothing in comparison with level rules. The dotted line indicates the E-frontier associated with level rules; i.e., the inflation-output volatility frontier for rules having the form of equation (4) withp = 0, under the constraint that the standard deviation of Ar does not exceed that generated by the estimated rule in equation (1). The solid and dashed lines show the E-frontiers for 3-parameter rules and first-difference rules, as previously depicted in Figure 2. As previously noted, the E-frontiers for 3-parameter rules and first-difference rules are virtually identical for three of the models, while the optimal 3-parameter rules associated with the FM model incorporate a high degree of interest rate smoothing but with values of p noticeably smaller than unity.

Figure 6 reveals very substantial gains from interest rate smoothing in all four models.

In fact, these are the largest gains we have found among all the permutations of simple policy rules that we have investigated. In the FRB model, for example, using a first-difference rule instead of a level rule can reduce the standard deviation of output by a full percentage point.

Figure 6 provides further information on the benefits of interest rate smoothing in comparison with level rules. The dotted line indicates the E-frontier associated with level rules; i.e., the inflation-output volatility frontier for rules having the form of equation (4) withp = 0, under the constraint that the standard deviation of Ar does not exceed that generated by the estimated rule in equation (1). The solid and dashed lines show the E-frontiers for 3-parameter rules and first-difference rules, as previously depicted in Figure 2. As previously noted, the E-frontiers for 3-parameter rules and first-difference rules are virtually identical for three of the models, while the optimal 3-parameter rules associated with the FM model incorporate a high degree of interest rate