Why Money Talks and Wealth Whispers: Monetary Uncertainty and Mystique: A Comment

Abstract

We demonstrate that in important cases Propositions 3 and 4 in Eijffinger, Hoeberichts, and Schaling (Journal of Money, Credit, and Banking, May 2000) may fail. Moreover, their monetary policy delegation arrangement, which advocates that central banker preference uncertainty may be desirable, is dominated by other arrangements without any such uncertainty. Finally, their way of modelling preference uncertainty leads to arbitrary effects on average monetary policy. Without these, preference uncertainty is never desirable.

In a recent article in this journal, Eijffinger, Hoeberichts, and Schaling (2000), henceforth EHS, argue that monetary policy uncertainty may be welfare enhancing. More specifically, uncertainty about the weight that the central bank attaches to inflation stabilization helps to reduce output variability, as the central bank on average reacts more vigorously to supply shocks. This gain may dominate the losses associated with weight uncertainty in their setup, which take the form of a higher inflation variability, and—if a Barro and Gordon (1983) type of credibility problem prevails—a higher inflation bias. Taking central bank preference uncertainty as a proxy for central bank secrecy, this result leads EHS to conclude that their paper "explains why high credibility institutions such as the former Bundesbank can afford to be relatively closed, and why low credibility institutions such as the Reserve Bank of New Zealand and the Bank of England need to be very open ..." (p. 231).

This paper comments on EHS. Firstly, we show that precisely when the central bank has high credibility, uncertainty about its preferences may actually be undesirable. This implies that EHS' central Proposition 4 may fail and, hence, that their claim that "high credibility institutions ... can afford to be relatively closed" is unwarranted. The main reason for this potential failure is that EHS' Proposition 3, which states that output variability is always minimized for a positive level of preference uncertainty, may also fail.

Secondly, we argue that the role for preference uncertainty is generally much weaker than EHS claim. In their setup, preference uncertainty can be beneficial if monetary policy is delegated to a central bank with a suboptimal degree of conservatism (in the sense of Rogoff 1985), which causes inefficiently high output variability. Introducing uncertainty about the preferences of this central bank may help to correct this inefficiency because it may induce the bank to act in a more "liberal" way on average. However, imposing the appropriate inflation contract (Walsh 1995) or the appropriate inflation target (Svensson 1997) takes the economy right to the socially optimal equilibrium, thereby obviating the need for preference uncertainty. Even if these arrangements are not feasible, one might do better by choosing the optimal degree of conservatism right away. This is confirmed by numerical results for a wide range of parameter combinations.

Lastly, we demonstrate that the particular direction in which preference uncertainty in EHS affects the central bank's average responses to, for example, supply shocks is arbitrary. We do this by illustrating that if uncertainty is modeled in just a slightly different (but analogous) manner, its effects on average monetary policy responses are exactly the opposite of those obtained by EHS. In order to isolate the implications of preference uncertainty on policy uncertainty, we therefore examine a preference specification adopted from Beetsma and Jensen (1998), from which the average effects of preference uncertainty are absent. With this specification, central bank preference uncertainty is unambiguously harmful for society. Hence, if one concurs with EHS that preference uncertainty is a proxy for secrecy in monetary policymaking, then one must be very careful in advocating such secrecy.

1. The EHS Model

We briefly present the EHS model using their notation. Output is given by a reduced-form Lucas supply function:

where y is the (log of) output, y* > 0 is the natural level of output, π is inflation, π^e is expected inflation, and ε is a white noise shock with zero mean and variance σ_2π. Society's loss function is

where desired output exceeds the natural level if k > 1. This implies the familiar inflation bias under discretionary monetary policy (cf. Barro and Gordon 1983). [End Page 130...