13. We log-linearized FRB and linearized TAYMCM around sample means. Because the two models are nearly linear anyway, this linearization has virtually no effect on the second moments studied in this paper. On a Sun Ultra Enterprise 3000 computer (SPECfp_base95= 8.45, about twice the speed of an Intel Pentium 233), it takes seconds to solve and compute moments for the Fuhrer-Moore and MSR models and about five minutes each for the FRB and Taylor models. The solution algorithm has been modified to use sparse matrix operations. The moment computations take advantage of the doubling algorithm described in Hansen and Sargent (1997).

14. Throughout the paper, inflation refers to the growth in the GDP price level for FM, MSR, and TAYMCM, and that of the personal consumption price index for FRB.

15. Throughout the analysis in this paper, we assume that the inflation target is high enough to avoid the distortions documented in Orphanides and Wieland (1997) for low inflation regimes.

16. As seen in Figure 3, along the vertical asymptote the level and first-difference rule frontiers overlap. For very low weights (0.0 – 0.05) on output variance in the policy objective function, the optimal value of p ranges from 0.5 to 0.85, but this coefficient quickly rises to nearly unity as the weight on output variance rises above 0.05.

17. Sources of mismeasurement relevant for policy rules include the noise in the data due to sampling and data imputation methods and imprecise estimation of potential output and the natural rate of unemployment. Orphanides (1997) studies the relevance of measurement problems for policy rules. Wieland (1997) examines optimal policy when the natural rate is unknown. guaranteed payday loans

13. We log-linearized FRB and linearized TAYMCM around sample means. Because the two models are nearly linear anyway, this linearization has virtually no effect on the second moments studied in this paper. On a Sun Ultra Enterprise 3000 computer (SPECfp_base95= 8.45, about twice the speed of an Intel Pentium 233), it takes seconds to solve and compute moments for the Fuhrer-Moore and MSR models and about five minutes each for the FRB and Taylor models. The solution algorithm has been modified to use sparse matrix operations. The moment computations take advantage of the doubling algorithm described in Hansen and Sargent (1997). 14. Throughout the paper, inflation refers to the growth in the GDP price level for FM, MSR, and TAYMCM,