with the central bank’s other policy objectives, this volatility in real marginal costs raises the
volatility of the commitments that characterize the timeless perspective policy, penalizing its
performance.
The remainder of this paper is organized as follows. Section 2 introduces the timeless
perspective approach to policy design, applying it to a simple New Keynesian model. Sec-
tion 3 shows how standard control methods for rational expectations models can be used to
construct and analyze the equilibrium of a timeless perspective policy. Section 4 turns to
methods for evaluating timeless perspective policies, showing why the treatment of the auxil-
iary states matters importantly for performance comparisons. Subsequently, Section 4 shows
how the auxiliary state variables can be conditionally integrated out to construct a measure of
conditional loss that is easy to compute and that is suitable for comparing the performance of
discretion and timeless perspective policies. Applying this measure of conditional loss to the
simple New Keynesian model described in Section 2, Section 4 shows that discretion can be
superior to timeless perspective policymaking. Section 5 extends the analysis to a small-scale
DSGE model and shows that factors that flatten the slope of the Phillips curve increase the
likelihood that discretion will be superior to timeless perspective policymaking. Section 6
concludes.
2 Timeless perspective policy design
Commitment policies that are optimal in the sense of Kydland and Prescott (1980) are not sta-
tionary. If a commitment mechanism is available, then the optimal policy is to exploit private
sector expectations in the initial period while promising never to do so in the future. Such
policies are not stationary because the initial period, the period in which the optimization just
happens to occur, holds a special significance. To some, this feature of optimal commitment
policies, although fundamental to their being optimal, is unattractive and undesirable. After
all, when conducting policy today, why should a policymaker implement a policy that, while
optimal from the perspective of some arbitrary date in the past, is suboptimal today?
To overcome this “initial-period problem” and obtain a stationary policy, it is now common
to assume that policymakers approach policy design from a timeless perspective (Woodford,
1999a). Broadly speaking, a timeless perspective policymaker promises not to exploit initial
conditions. Instead, the policymaker ties its hands and commits to behave in the initial period
as it does in all subsequent periods. Put differently, the timeless perspective policymaker
implements policy as if that policy had been formulated in the distant past, such that any