RDP 1977-06: Interest Rates and Exchange Rate Expectations in the RBA76 Model 6. Simulation Analysis

This section of the paper illustrates the effects of policy impulses in the models discussed above. With Model B, a monetary/credit impulse of the type discussed in Jonson and But 1 in (1977) is simulated; an initial 1 per cent reduction in the bond rate is accompanied by a sustained increase in bank lending, the size of the joint impulse being designed to produce an effect on the monetary growth rate equal to that produced in the first quarter by a sustained 10 per cent increase in government current expenditure. With all four models a simple interest rate shock is also simulated; this comprises an initial 1 per cent reduction in the bond rate, and the exogenous shock which produces this impact effect is sustained during the simulation; the bond rate thus tends to fall relative to the control solution values throughout the period, although the endogenous reactions of the bond rate in the model mean that the deviations from the control solution levels are not a constant 1 per cent and in fact first fall by more than 1 per cent, and then the rate of change of the bond rate rises above the control solution value.^[16] In all experiments a managed float of the exchange rate is assumed, with the variation of the rate determined by the estimated policy reaction function in each model. The numerical results of the simulations should be regarded as illustrative only; the simulation analysis is subject to the usual objection that the impulse may be unrealistic and that private sector responses may have been different if alternative economic policies had been followed. The simulations do however provide some basis for comparison of the properties of four models considered in this paper.

Model B is the model in which the bond rate has the highest weight^[17] in determining the equilibrium debenture rate. The net effect of the bond rate and the debenture rate in the asset demand functions is similar to that of the bond rate in Model A, which is a version of the standard RBA76 model. It is therefore not surprising that the results of a monetary credit impulse are similar in Model B to the responses of the mid-1977 version of the model. The responses of Model B are presented, in terms of deviations of growth rates from the relevant control solution values, in the panels of figure 3. These responses include a monetary cycle which is roughly opposite to the cycle in the growth of international reserves, a cycle in the growth of real product, an initial increase followed by a decrease in inflation, and a devaluation of the exchange rate.

As the comparison of the paths in figure 3 depicting the monetary/credit impulse with those representing the effects of reducing the bond rate indicates, the quantity part of the joint impulse largely reinforces the effect of reducing the bond rate, and the remainder of the analysis will examine the effects of interest rate changes alone.

The responses of Models A, B and C to a reduction in the bond rate are broadly similar.^[18] In each case there is a tendency for the monetary growth rate to fall below the control solution values after an initial increase. Given the increased real income^[19] and reduced interest rates, both responses which increase the demand for money, the resulting excess demand for money puts mutually reinforcing downward pressure on prices and money wages. The initial increase in the monetary growth rate occurs because decreased bond sales are sufficient to more than offset the reduction in the growth of international reserves, but eventually the growth of bonds picks up because the effect of higher real income and expectations of revaluation of the exchange rate outweighs the effect of reduced interest rates.^[20]

The comparison of the responses of Models C and D, presented in figure 4, is of considerable interest. To recapitulate, Model D reverses the direction of causation between the bond rate and the debenture rate which is implied by the structure of Model C, and of Models A and B. For Model D, the responses of real product and the balance of payments are broadly similar to those in Model C, although there is a greater loss of international reserves,^[21] and hence a larger devaluation. Despite the greater loss of international reserves in Model D, the monetary growth rate is above the control solution value for much longer than it is in Model C, mainly because of a larger reduction in bond sales in Model D. As illustrated in Table 3, the response of the demand for bonds to changes in interest rates is similar in Models C and D, but a crucial difference between the two models is that in Model D the debenture rate does not fall in response to the falling bond rates, as it does in Model C. In Model D there is a fall in the bond rate relative to the debenture rate, as well as relative to interest rates in the rest of the world, and hence there is a greater reduction in bond sales in this model. This tendency is reinforced by the lower parameter on the relative prime term in the expression for exchange rate expectations in Model D. With higher monetary growth and similar growth of real product in Model D compared with Model C, inflation is higher in Model D.