مداخله بانک مرکزی و نوسانات نرخ ارز - شواهد استرالیا

This paper examines the key characteristics of foreign exchange intervention by the Reserve Bank of Australia in the period 1983–1997, which can be broken into five distinct phases. We investigate the changing effectiveness of daily intervention on the $US/$A exchange rate by decomposing the exchange rate response to the intervention into various separate components. We find contemporaneous positive correlation between the direction of intervention and the conditional mean and variance of exchange rate returns. We show that sustained and large interventions have a stabilising influence in the foreign exchange market in terms of direction and volatility. Without these interventions, the market would have moved further and exhibited more volatility.

The effectiveness of foreign exchange intervention by central banks has been studied at length in the last 30 years. Much of the early literature focused only on the longer term implications and objectives of intervention, being constrained by the availability of only low frequency data on the central bank activities. Generally the evidence on the effectiveness hypothesis from this data was weak, and that is a generous interpretation. Yet central banks have continued to intervene, believing from their experience that intervention does indeed work. In the last 20 years, many central banks have made available information on their daily interventions, which have allowed researchers to study the inter-daily features of exchange rate intervention on the foreign exchange market. Although tick-by-tick data is really needed to fully evaluate the effectiveness on an intra- and inter-daily basis, daily data availability has meant that it is possible to test effectiveness arising from micro-structural features of the foreign exchange market. There is a popular view that central banks (or their treasuries, where they are dependent) have no special knowledge or ability in the busy foreign exchange market place, nor do they have adequate reserve resources to determine the direction of the exchange rate. Any intervention is believed to add further confusion in periods of turbulence exacerbating uncertainty and thus volatility. If the case for a fixed exchange rate, or an EMS bandwidth system, cannot be supported, then the central bank is advised to keep out of the market altogether. Looking at intervention data, it is immediately apparent that there are often long stretches of time when intervention has been dormant, even in periods when there has been considerable turbulence in the markets. So this popular view has certainly had some influence. In apparent support, almost all the empirical work with high frequency data has found that the intervention on any day is positively correlated to the conditional variance of exchange rate change for that day, or else uncorrelated. Dominguez (1998) finds some significant rises in conditional volatilities of US exchange rate returns on the days of secret intervention by the Federal Reserve Bank, the Bundesbank and the Bank of Japan (although there were some falls for reported interventions by the first two banks). Baillie and Osterberg's (Baillie and Osterberg, 1997) GARCH research shows that foreign exchange intervention by these central banks in 1985–1990 had no significant impact on the conditional mean and variance of changes in the spot US exchange rates. Others who report significant positive intervention effects on exchange rate volatilities include Almekinders and Eijffinger (1994) and Bonser-Neal and Tanner (1996). A similar phenomenon sometimes shows up with the conditional mean, which may suggest the central bank adds to destabilising speculation in the market. Baillie and Osterberg (1997) conclude that intervention generally had no effect on the change in the exchange rate, though purchases of the $US by the Fed was correlated with contemporaneous depreciations of the dollar. Dominguez (1998) estimates but does not report conditional mean effects, though Dominguez and Frankel (1993) do find that in 10 out of 11 episodes of clustered interventions involving more than one central bank, intervention moved the exchange rate in the appropriate direction (see Edison, 1993, for a survey). However these may be occurring because of the standard simultaneity problem — intervention occurs in response to turbulent foreign exchange market conditions on a day, and the exchange rate process on that day may be affected by that intervention. This sort of evidence is not a good test of that popular view. Despite the evidence for effectiveness being generally mixed, there is increasing support for the alternative belief that central banks can have some influence on the stochastic properties of exchange rate processes. And of course, this view has some support in central bank circles, because there are so many daily occurrences of strategic interventions, conducted secretly or accompanied by official statements. In an influential monograph, Frankel and Dominguez (1993) concluded that intervention has a maximum impact when it occurs unexpectedly, with as much publicity as possible on the day and when it is done in coordination with other central banks. In Australia, the Reserve Bank is responsible for conducting interventions on the $US/$A exchange rate, and for the post-floating periods (since December 1983) five distinct periods of intervention can be identified. The aim of this paper is to investigate the changing effectiveness of the RBA's intervention on the exchange rate (in conjunction with monetary policy) with particular emphasis on the effects on the conditional volatility of the daily exchange rate returns. The paper is organised in sections. In Section 2, we discuss some theoretical explanations of the possible effects of intervention, and link the standard explanations to recent developments in understanding the microstructure of financial markets. In Section 3, we develop the key characteristics of the Australian dollar foreign exchange market, and of the intervention patterns of the Reserve Bank of Australia (RBA). In Section 4, we develop the GARCH model that we use to estimate the effectiveness of the intervention. In Section 5, our results are discussed, and the paper ends with our key conclusions in Section 6.

We have found evidence that the Reserve Bank of Australia has had some success in its foreign exchange intervention policy. Its stated objectives and the resulting performances varied over five distinct phases in the period 1983–1997. By checking the various components of the exchange rate response to the intervention, we find evidence of a stabilising influence on the $US/$A exchange rate process. That is, purchases of $A on a day tended to strengthen the currency and reduce its conditional volatility. The Reserve Bank will have been prompted to intervene when the exchange rate moved sufficiently far (down) on a particular day, breaching its threshold of inactivity. On that day, we would expect to see (a defensive) intervention action positively correlated with volatility and the (depreciating) change in the currency. This obviously accounts for the observed contemporaneous effects of intervention documented in the paper. Having laid the groundwork, one direction of future research would be to explicitly model this policy endogeneity effect. In general, we find that official statements concerning intervention had little effect on the exchange rate process. In the second period (1986–1991), when there were some large swings in the exchange rate (e.g. a current account deficit blow out and a subsequent ‘Banana Republic’ statement by the Federal Treasurer in 1986, 1987 stock market crash, monetary policy tightening from 1988, the 1990 Gulf War) and the RBA aimed to ensure orderly conditions, its official statements had a very significant destabilising effect, which apparently undermined its intervention activity.