The significance of uncertainty in monetary policy.

• Author: Froyland, Espen
• Position: Example of Norway's Norges Bank - Statistical Data Included

Observations seem to indicate that central banks adjust their key rates gradually. Uncertainty regarding economic relationships and measurement errors in the data may point to a need for key rates to be adjusted gradually. Uncertainty relating to possible shocks in exogenous variables on the other hand does not indicate the need for a gradualist approach to setting interest rates. The decisive factor regarding how the uncertainty inherent in monetary policy should be taken into account is whether the setting of interest rates in itself affects the degree of uncertainty and thereby the variance in the outcome of monetary policy. If the setting of interest rates affects both the expected outcome and the uncertainty of the outcome, it will be optimal to take both factors into account. Conclusions in the theoretical literature must be interpreted with caution. The models are highly simplified. Nevertheless, theory makes an important contribution to shedding light on the problems facing central banks in the conduct of monetary policy.

Introduction

Interest rates are the most important monetary policy instrument available to central banks. Experience shows that in many countries there is a tendency for central banks to take a gradualist approach to setting interest rates. It has been suggested that uncertainty in monetary policy is one possible explanation for this. Central banks have always had to take this uncertainty into account in the basis for monetary policy decisions, but it is only in recent years that there has been increased interest in this topic in the academic literature.

This article will review various types of uncertainty of relevance to monetary policy and the consequences for the setting of interest rates. We will begin by illustrating how gradual actual monetary policy has been in a number of countries, including Norway. Then we will go on to review various types of uncertainty and how they affect optimal monetary policy. We start by investigating whether gradual adjustments of monetary policy can be attributed to sudden, unexpected economic events, or shocks to the economy. We then discuss whether uncertainty in the parameters of the economic models offers an explanation of observed interest rate movements. Next, we consider the effect of measurement error on some variables in the models. Finally, we note that there is uncertainty associated with the models used by the central bank in its analyses. A simple model is used to illustrate the effects.

This article uses the terms gradual and cautious to describe monetary policy. Goodhart (1996) defines gradual monetary policy as the central bank adjusting interest rates in several small stages instead of a single jump when an inflation impetus arises which leads inflation away from the inflation target. A cautious monetary policy can be defined as a policy where interest rates react less to deviation from the inflation target than that which is optimal without uncertainty.(2) Actual monetary policy tends to be both cautious and gradual, and we will discuss whether uncertainty is sufficient to explain the actual setting of interest rates. In many cases we will use the terms cautious and gradual monetary policy interchangeably.

Optimal monetary policy and actual monetary policy

In many countries, monetary policy is oriented towards low and stable inflation.(3) It is often emphasised that the objective is to be achieved without incurring excessive real economic costs in the form of high unemployment and lost production. In economic theory it is common to use a loss function for the central bank, comprising inflation gap, measured as the deviation of actual inflation from its target, and output gap, ie the difference between actual output and potential output. By minimising the loss given the economic relationships, the central bank can derive an optimal path for interest rates.(4)

It can be shown that in a static model, optimal monetary policy will mean that the probability that the next change in interest rates will be an increase is the same as the probability of a reduction, see Goodhart (1999). Assume, for instance, that random shocks continually occur in the economy, necessitating increases or decreases in interest rates. Over time it will be natural for these shocks to be evenly divided between those requiting tighter monetary policy and those requiting more expansionary monetary policy. Subsequent changes in interest rates will therefore be uncorrelated.

In a dynamic model, however, it may be optimal for an adjustment of interest rates to be carded out in several stages. This is because economic variables such as output and inflation change slowly, and because the effects of monetary policy are associated with relatively long lags. However, it is apparent that the dynamic structure of the economy does not provide an adequate explanation of the gradual setting of interest rates observed in many countries. Central banks adjust their key rates more gradually than indicated by optimal monetary policy. For this reason, some authors define the setting of interest rates as gradual if interest rates change less than can be explained by the dynamic structure of the economy, see for instance Sack (2000).

Table 1 shows actual interest rate setting in various countries. In the period August 1989 to March 1998, the US Federal Reserve lowered interest rates twice in succession on 22 occasions. During the same period, the Federal Reserve shifted from lowering interest rates to increasing them in only two cases. Interest rate changes have also been gradual in Norway.(5) Since December 1992, Norges Bank has on 28 occasions lowered interest rates twice in succession, whereas it has only twice raised interest rates and then lowered them.

Table 1. Number of interest rate adjustments in selected countries Start date/end date ++ +- -+ -- US 10 August 1989/ 31 March 1998 6 1 2 22 UK 1 January 1978/ 31 March 1998 28 17 18 84 Sweden 1 June 1994/ 31 March 1998 14 1 2 24 Germany 19 June 1979/ 31 March 1998 65 31 31 107 Norway 1 January 1992/ 15 June 2000 7 2 5 28 ++ = Two successive increases: +- = Increase followed by decrease: -+ = Decrease followed by increase: -- = Two successive decreases

Sources: BIS (1998) and Norges Bank's calculations for Norway

The following section discusses what economic theory says about how to take uncertainty in monetary policy into account, and whether various economic theories can explain why central banks adjust interest rates gradually and cautiously.

Additive uncertainty

We will illustrate various types of uncertainty using a simplified version of Svensson's model (1997).(6) This model uses a highly simple representation of economic relationships, but is nevertheless useful for illustrating the effect of various types of uncertainty on optimal monetary policy.

We assume that the inflation process can be modelled as a backward-looking, expectations-augmented Phillips curve:

(1) [[Pi].sub.t+1]=a[[Pi].sub.t]+[Beta][y.sub.t]+[[Epsilon].sub.t+1], a, [Beta] [is greater than] 0

where [[Pi].sub.t] is the inflation rate and [y.sub.t] is the output gap. The output gap provides an indication of pressures in the economy. [[Epsilon].sub.t] is a cost shock in the form of an additive residual, with zero expectation and constant variance, [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]. There are lags in inflation, which means that inflation from the previous period is part of the reason for inflation in the current period. These lags may be due to structural factors in price-setting and the fact that expectations regarding future inflation are backward-looking. A positive (negative) output gap will contribute to higher (lower) price inflation in the subsequent period.

The output gap depends solely on the nominal interest rate in the same period, [i.sub.t]:

(2) [y.sub.t]=-[Delta][i.sub.t] [Delta] [is greater than] 0

Potential production is normalised to zero, so that [y.sub.t] denotes the output gap. As can be seen, higher interest rates in isolation will contribute to less pressure in the economy in the form of a lower output gap in the same period. By substituting (2) into (1) we obtain:

(3) [[Pi].sub.t+1]=a[[Pi].sub.t]-b[i.sub.t]+[[Epsilon].sub.t+1],

where b=[Beta][Delta] From equation (3) we...