RDP 2000-07: The Effect of Uncertainty on Monetary Policy: How Good are the Brakes? 1. Introduction

In most industrial countries, policy interest rate changes tend to be ‘smooth’. That is, rates are adjusted relatively infrequently and in small steps. In contrast, the path of interest rates that emerges as optimal from macroeconomic models is, in general, considerably more volatile. This is also the case for the paths of interest rates that are implied by simple Taylor-type rules, unless a sufficiently large weight is put on an interest rate ‘smoothing’ term that penalises large movements in the policy interest rate. Do these contrasting outcomes imply that policy-makers are adopting sub-optimal monetary policy strategies, or are there factors that are not captured in the models that justify the strategies that are pursued in practice?

One possible explanation for the divergence between the models and observed practice is that the former fail to adequately capture the uncertainty that impinges on the monetary policy decision. Most notably, Brainard (1967) highlighted the fact that uncertainty about model parameters can induce less ‘aggressive’ actions on the part of the policy-maker than those that result when uncertainty is ignored. Consequently, this paper investigates the extent to which different forms of uncertainty affect the optimal path of interest rates. It does so by incorporating uncertainty in a simple model of the Australian economy, and examining the impact of various forms of uncertainty on the variability of the instrument of monetary policy.

This paper complements the analysis of Rudebusch (1999) who conducts a similar analysis for the US economy. One difference between his analysis and that in this paper is the inclusion of another transmission channel of monetary policy, namely monetary-policy-induced changes in the exchange rate on output and inflation. It also extends Shuetrim and Thompson's (1999) analysis of uncertainty in the Australian context.

Before investigating smoothness in Australia, in Section 2, the practice of interest rate smoothing by industrial-country central banks is documented, and some possible explanations that have been advanced for this behaviour are reviewed. Section 3 focuses explicitly on uncertainty as an explanation for smoothing and summarises the growing literature that examines the impact of various types of uncertainty on the monetary policy process. Section 4 describes briefly the simple model of the Australian economy and the methodology that will be used to examine the effect of uncertainty. Section 5 presents the empirical results, and Section 6 concludes.

The main findings of the paper are that the introduction of uncertainty does not explain the divergence between model-derived optimal policy and observed outcomes. Different types of uncertainty have differing effects on the degree of smoothness in the path of policy interest rates implied by the model of the Australian economy used here. General parameter uncertainty does not have much impact on smoothness. Shuetrim and Thompson (1999) find that the incidence and persistence of the uncertainty determines whether it results in more or less smoothness. Söderström (1999) obtains similar results to those in this paper in an analytical model. These findings, however, are in contrast to recent results for the US (Sack 2000) or the UK (Martin and Salmon 1999), and some possible explanations for the difference are discussed below.

Uncertainty about the average interest sensitivity of the economy – that is, how good the policy brakes are – is shown to have a significant impact on the degree of smoothness. Increasing the estimated interest sensitivity of the economy by one standard deviation in the model results in a smoother path of optimal policy interest rates. Nevertheless, the path of interest rates generated is still much more variable than that observed in practice.