RDP 2000-06: Inflation Targeting and Exchange Rate Fluctuations in Australia 1. Introduction

In 1993, the Reserve Bank of Australia (RBA) introduced an inflation target of 2–3 per cent per annum, on average over the cycle. The available evidence suggests that, despite initial criticism from some quarters, this inflation-targeting framework has served the country quite well, possibly better than inflation targeting has served some other countries (Brooks 1998). We hasten to add that it may have been possible to do even better, and what has worked well for Australia may not work as well for some other countries, but the fact remains that it has worked well for Australia to date.

As for whether even better outcomes were possible, a number of recent papers have focused on the implications of exchange rate fluctuations for inflation targeting and have explored the possibility that inflation-targeting central banks in small open economies pay too much attention to these fluctuations.^[1] The argument is that exchange rate fluctuations tend to have significant but transient effects on inflation and that monetary policy attempts to offset these effects could cause undue variability in output.

On first inspection, Australia would appear to be a good test case for this argument. The Australian dollar has ranged from around US$0.68 in 1993 to a high of around US$0.80 in mid 1996 and to its recent low of around US$0.55. Part of these movements reflect changes in Australia's commodity-driven terms of trade, which help dampen output variability, but there has also been considerable short-term volatility not related to such fundamentals, such as in mid 1998. Australia also fits the description of a small open economy: it exerts little or no influence on world prices and our exports and imports account for a large share, each around 20–25 per cent, of GDP.

On closer examination, it could be argued that Australia's monetary policy framework already deals with the problem adequately. The inflation target, for example, is not hard-edged: relatively small divergences from the 2–3 per cent target band are tolerated provided inflation is forecast to be back within 2–3 per cent in the medium term. The forward-looking nature of policy should also be sufficient to prevent the RBA from responding to exchange rate shocks which are only expected to have a temporary effect on inflation.

These counterarguments are clearly an empirical matter and should be assessed accordingly. Thus, while accepting that it is ultimately aggregate inflation (and output) that the central bank cares about, in this paper we compare the implications of allowing policy to respond to different measures of inflation: aggregate, non-tradeable and growth in unit labour costs in the non-tradeable sector. Such a comparison enables us to assess the validity of the argument that central banks, by reacting a lot to temporary exchange rate movements, can cause undue variability in output.

Figures 1 and 2 show four-quarter-ended and quarterly movements in underlying tradeable, non-tradeable and (for Figure 1) aggregate inflation. Non-tradeable inflation was particularly high in the early 1980s, when wages growth was high in some sectors of the economy. The large exchange rate depreciation in the mid 1980s pushed tradeable inflation higher than non-tradeable inflation. More recently, the Australian dollar price of imported items in the CPI fell in absolute terms, causing the prices of tradeables to be roughly flat.

Figures 1 and 2 may suggest that non-tradeable inflation is at least as variable as tradeable inflation and that, at least occasionally, the two measures are negatively correlated such that aggregate inflation is less variable than non-tradeable inflation. If true, it would seem to be a foregone conclusion that one should target aggregate, rather than non-tradeable inflation. The issue is, however, more complicated than that. First, simple statistical analysis indicates that the two components generally move together and that aggregate inflation is almost as variable as non-tradeable inflation. More importantly, policy-makers are forward looking and non-tradeable inflation may be more predictable than tradeable inflation because it is less dependent on forecasts for the exchange rate, a variable which is notoriously difficult to predict.

The structure of the remainder of the paper is as follows. The next section reviews some previous research on open economy inflation targeting, motivates our distinction between aggregate and non-tradeable inflation and describes the simple monetary policy feedback rules which we use later in the paper. Section 3 describes the model that we use to investigate the properties of various monetary policy rules. The results of simulating the model under the different policy rules are discussed in Section 4. Section 5 concludes. As a preview, our findings suggest that there are no clear gains to be had from changing the nature of Australia's inflation-targeting framework.

Footnote

See, for example, Ball (1998), Svensson (1998), Bharucha and Kent (1998) and Conway et al (1998). For an earlier reference to targeting a measure of inflation that abstracts from exchange rate effects, see Pitchford (1993). [1]