An Evaluation of Alternative Indicator Regimes for Monetary Policy

Various indicators have been used to guide monetary policy. In many OECD countries, movements in the money supply itself are now a signal to the monetary authorities to adjust interest rates so as to correct any deviation of the money supply from target range. If targeting is adopted wholeheartedly, the money supply represents an intermediate target of policy, standing between the final objectives of policy (such as output and prices) and the instruments of policy (such as interest rates and other monetary control variables). The monetary authorities may be unwilling, however to accept the degree of variability in interest rates and other instruments required for strict short- term control of the money supply, and may prefer to trade off variability in the money supply against variations in interest rates. In this case, responses to deviations of the money supply from target will be more relaxed, and the money supply acts as an indicator of policy.

Other variables may serve as intermediate targets or indicators. The exchange rate was an intermediate target throughout the Bretton Woods era, as it is now within the EMS and in many third world countries; for many other countries, it constitutes an important indicator for monetary policy. The use of nominal income as an intermediate target has been advocated for the US (Tobin) and for the UK (Meade). A price index or an index of nominal wages might similarly be used as an indicator for policy.

The evaluation of indicator regimes requires an analysis under uncertainty, in the context of a fully specified macromodel. Currie and Levine uses an analytical model of this kind to compare how well each of the indicator regimes discussed above stabilises the small open economy. Their model includes dynamics arising from lags in the expenditure and money demand functions, from the wage/price sector, from asset accumulation via the government budget and the balance of payments, from exchange rate behaviour, and from persistence in some of the exogenous stochastic disturbances. They choose parameters to be plausible and consistent with available econometric evidence. The indicator regimes are evaluated on the assumption that price and output variability are equally undesirable, but these weights and other parameter values are varied extensively to test robustness. The evaluation is carried out for five types of stochastic disturbance: an unforeseen, exogenous shock to aggregate demand; a money demand shock; an aggregate supply shock; a foreign price level shock; and a foreign interest rate shock.

The way in which private sector forms its expectations of future variables is unlikely to be independent of the policy regime in force. This problem is best handled by assuming consistent or rational expectations, so that all model simulations assume that the private sector makes no systematic errors in prediction. A policy which performs badly under consistent expectations could perform well only by virtue of systematic forecasting errors by the private sector, which would provide an ill-founded and inherently unstable basis for policy. The comparison of alternative indicator regimes is best carried out in an optimal control framework, so that the rule linking interest rates to the indicator can be chosen optimally. Rational expectations pose problems for the theory of stochastic optimal control, and the first part of the paper deals with these. The theory is then applied to the model described above.

Currie and Levine find for monetary policy generates rather high variability in output and prices, and is therefore unsatisfactory. The money supply performs moderately well if the nature of the shock to the system is known, but rather badly if it is uncertain. Nominal income typically performs better than both these regimes and is more robust in the face of uncertainty about the source of shock. But the use of the price level as an indicator dominates all other regimes in stabilising the system in the face of both known and unknown disturbances. Moreover, the price rule is only marginally inferior to the fully optimal rule, where policy responds to all variables in the system. These results continue to hold when the parameters of the model are varied widely, so the price level may provide a rather robust indicator for monetary policy. Further work will pursue this line of inquiry for more developed models, including interdependent economies, and will also examine the use of fiscal policy in conjunction with monetary policy.


An Evaluation of Alternative Indicator Regimes for Monetary Policy
David Currie and Paul Levine

Discussion Paper no. 4, February 1984 (IM)