for time-separable isoelastic preferences. Equations (57) through (59) represent to first-
order accuracy the capital accumulation equation, the resource constraint, and the aggregate
production technology.
Turning now to monetary policy, I assume, for simplicity, that the central bank’s decision
problem is to choose {rt+ι}∞ to optimize the primal loss function
∞
eO ∑ βt (πt + M⅜2) , (60)
t=0
subject to equations (48) through (59) and the known initial conditions To, go, uo, and ko.
5.1.2 Results
As earlier, I now analyze policy performance on a parameter grid. Specifically, I condition on
the depreciation rate and on the parameters in the shock processes,13 since preliminary inves-
tigations indicated these “persistence” parameters were largely unimportant for the results,
and analyze the performance of discretion relative to timeless perspective policymaking on a
grid of values for a, ξ, σ, ε, χ, and μ. The results are shown in Figure 3, which displays, for
each parameter, the share of the parameter space for which discretion is the superior policy.14
13For this exercise, I set ρg = ρv = 0.3, ρu = 0.95, and δ = 0.025. Further, the initial state is described by
vo = go = uo = ko = 0, such that the economy initially resides at its nonstochastic steady state.
14Blake and Kirsanova (2008) have shown recently that this model can exhibit multiple discretionary equi-
libria. In my simulations, I addressed this possibility by using multiple starting points to search for the
worst-performing discretionary equilibrium.
20