RDP 2000-06: Inflation Targeting and Exchange Rate Fluctuations in Australia 5. Conclusion

Inflation targeting in a small open economy is complicated by the exchange rate channel of monetary policy. In this paper we have used a small econometrically estimated model of the Australian economy to examine a number of issues relevant to the pursuit of inflation targeting in an open economy.

Our results, some of which are already well established, are summarised here.

Policy rules in which the interest rate responds to model-consistent forecasts of inflation and the output gap are preferable to those which respond only to lagged outcomes. This will be particularly relevant for an open economy where inflation is likely to be influenced by more than just past inflation and output. Forward-looking policy takes account of all available information to forecast future inflation and is better able to account for lags in the monetary policy transmission mechanism. This result lends support to the current central bank practice of ‘pre-emptive’ monetary policy.

While forward-looking policy can certainly improve outcomes, our results indicate that there is a limit on how forward-looking policy can be before it begins to generate worse outcomes. This finding merely reflects the fact that forecasting future inflation is an imperfect science – the further the forecast horizon, the less chance that the forecast will eventually prove correct because there is more time for unforeseen circumstances to occur.

Many of the forward-looking policy rules which we consider are ‘efficient’ in the sense that they stabilise the economy better than most other rules, but they typically imply interest rate variability which is much larger than that seen in practice. One explanation is that our model ignores a number of sources of uncertainty which the policy-maker actually faces in practice. The implication of uncertainty for monetary policy is an issue which is the focus of a lot of ongoing research so we did not attempt to address the issue in this paper. Instead, we investigated a number of ways of trying to solve the problem. Smoothing the response of the interest rate by means of a partial adjustment mechanism in the policy rule, for example, did not sufficiently reduce interest rate variability. The approach which we ultimately took was to ignore those policy rules which generated excessive interest rate variability. For the remaining feasible rules, the results of the paper were unchanged.

In this paper we also examined the implications of following policy rules in which the interest rate responded to a measure of inflation which was less susceptible to temporary exchange rate shocks. We also noted whether these alternative rules helped alleviate the problem of excessive interest rate variability.

Our results suggest that in the highly simplified world of a one-off shock to the exchange rate – the sort of world that one might think would most favour an alternative interest rate rule – an aggregate inflation rule actually seemed preferable.

Accepting that one cannot draw firm conclusions based on these results, we proceeded to stochastic simulations of the model. Non-tradeable inflation (and unit labour cost growth) rules appeared to generate at least as much variability in output and aggregate inflation – it is aggregate inflation that we care about – without any significant reduction in interest rate variability.

Our results appear to be robust to a number of simple modifications to the model. For example, we considered a more theoretically consistent version of the model in which policy had an immediate (rather than lagged) impact on the exchange rate. Another version allowed for inflation expectations in the non-tradeable sector to be determined by past non-tradeable rather than aggregate inflation. Neither of these variations, nor the other modifications which we tried, altered the main result of the paper.

Notwithstanding this, we do not rule out the possibility that our results are model dependent. However, to the extent that we have adhered to data-consistency wherever possible, we believe that our findings should be relevant to Australia.