آیا مداخله ارزی توسط بانک های مرکزی خبر بد برای بازارهای بدهی است : مورد مداخله بانک مرکزی استرالیا 1986-2003
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|23559||2006||22 صفحه PDF||سفارش دهید||محاسبه نشده|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Journal of International Financial Markets, Institutions and Money, Volume 16, Issue 5, December 2006, Pages 446–467
We investigate the effects of the Reserve Bank of Australia's foreign exchange interventions on the USD/AUD market and 90-day and 10-year interest rate futures markets for the period July 1986–December 2003. Using recently released revised and updated intervention data, we investigate contemporaneous and disaggregated intervention influences and find significant evidence for (i) intervention effectiveness in moderating the contemporaneous exchange rate movements especially if interventions were cumulative and large, (ii) exchange rate volatility reducing effect with a day's lag, (iii) undesirable interest rate movements following interventions in some periods compromising monetary policy effectiveness, and (iv) a volatility reducing effect of cumulative interventions in the 90-day rate, and a volatility increasing effect of large interventions in both the 90-day and 10-year rate futures. These findings are a unique and significant contribution to the prevailing literature as they demonstrate that the RBA's interventions matter not only for the foreign exchange market but also for the debt markets.
Central bank intervention in the foreign exchange market has been based on a belief that central banks’ participation in the market was indeed successful in achieving the twin aim of influencing the direction of exchange rate and calming volatility. The extant literature on interventions reports a common finding of a higher foreign exchange volatility and undesirable exchanges rate movements being associated with intervention on the days of intervention. Various researchers who examined the operations of major central banks (the Federal Reserve, the Bundesbank, and the Bank of Japan) to have arrived at this conclusion include: inter alia Beine (2004), Bonser-Neal and Tanner (1996), Baillie and Osterberg, 1997a and Baillie and Osterberg, 1997b, Chang and Taylor (1998), Dominguez (1998), Frenkel et al. (2005). Similarly, the Reserve Bank of Australia's (RBA) intervention operations have been reported to be either ineffective in influencing the trend of the exchange rate or at the very worst to exacerbate second moment disturbances (Edison et al., 2003, Rogers and Siklos, 2003 and McKenzie, 2004). The most important factor that is common in these studies is the issue of simultaneity between the interventions and the first and second moments of contemporaneous exchange rate changes. Much of the empirical work has often assumed that the decision to intervene is made exogenous of market conditions. However, such an assumption is a rather extreme one considering that central monetary authorities have explicitly declared they intervene to calm disorderly markets (Baillie and Osterberg, 1997a, Dominguez, 1998, Frenkel et al., 2005 and Kim and Sheen, 2002). As a consequence, models that do not take into account this endogeneity will likely indicate a positive correlation between conditional volatility and intervention and lead one to the erroneous conclusion that higher volatility is a consequence of intervention, rather than a rationale for it. There are essentially two approaches in handling the simultaneity issue, modeling approach and data approach. The former aims to address the simultaneity by devising econometric models that avoid the endogeneity of intervention variables when modeling the intervention effects on exchange rates. This might be accomplished by utilizing instrument variables that are highly correlated with interventions but not correlated with exchange rate movements. If such instruments can be found, one might carry out an instrumental variable estimation of the first and second moments of exchange rate changes on the days of intervention. Alternatively, exchange rate changes and interventions can be jointly modelled as a part of systems estimation. However, parametric representation of such a system is very difficult, if not impossible, because of the non-standard nature of the distribution of the intervention variable. The intervention variable has three distinct types of observations, positive and negative values and zeroes, and so it would be appropriate to treat the intervention variable to have been generated from a mixture of three distributions rather than one continuous distribution. This makes it very difficult to jointly model intervention with exchange rate changes, which have a continuous distribution.1Kearns and Rigobon (2005) makes a contribution in this regard. They estimated a simulated GMM model of the mean equations of intervention reaction function and exchange rate changes. The development of an equivalent variance equation framework would also be warranted. On the other hand, the data oriented approach aims to circumvent simultaneity mostly by examining the lagged effects of intervention, so that the decision to intervene is predetermined. However, lagged interventions usually lack explanatory power (Baillie and Osterberg, 1997a and Lewis, 1995). Another data approach is to use ultra-high frequency (UHF) data of 5 or 10-min quotes, particularly for the USD, the Yen and the Deutschemark. Identifying the precise time-stamp of intervention trades is vital in these studies, yet such data are not released by intervention authorities.2 The focus here is on employing a rough proxy of intervention, e.g. time stamps of Reuters newswire reports (Dominguez, 2003), requiring arbitrary assumptions regarding the time lag between intervention transaction and news reports. The assumed lag varies substantially, between 10 min and 2 h, introducing greater ambiguity into investigation results. Further, Fischer (2003) finds the Reuters news reports are in fact highly “deficient in capturing the timing of intervention rounds” (p.2) leading him to directly challenge the findings of the UHF literature. Thus, the UHF studies are found to be impotent to distil the ex post volatility effects of intervention as there is an over-disaggregation of returns. This fact is highlighted by the conclusions of the UHF studies themselves. Peiers (1997), Chang and Taylor (1998), and Dominguez (2003) all find that volatility precedes their intervention proxies, but still conclude that this volatility is caused by intervention, due to order flow learning. Kim et al. (2000), however, deal with the simultaneity problem by modelling various disaggregated characteristics of interventions, and find evidence to indicate that sustained and above-average size interventions by the RB A were successful in stabilising volatility in the USD/AUD market. The literature so far considers only the direct impact of interventions on their targets; the first and second moments of exchange rate movements. Beine (2004) is one notable exception. He reports not only an elevated volatility in the Yen/USD and the Euro/USD exchange rates on the days of US FED interventions, but also in their covariance. Another is Fatum and Hutchison (1999) who report insignificant response of the US federal fund futures rates but interventions raised their volatility. They concluded that interventions added to the noise in the market. In a similar vein, recent evidence suggests that interventions may be used to achieve monetary policy objectives (see Kim and Sheen, 2005 and Vitale, 2003) providing a natural information linkage between foreign exchange and short-term debt markets. In general, there is a potential for interventions to have significant impacts on other segments of the financial markets due to their system wide information releasing effects. For instance, both the first and second moment influences on the exchange rate would have important implications for international portfolio investors leading to a possible significant impact on the stock market. In addition, the equilibrium relationships between foreign exchange and short-term debt markets via interest rate parity conditions suggest potential information spillover to debt markets. The literature is rather silent on this issue of indirect effects of interventions on other financial market segments. In this paper, we have a dual aim of (i) investigating the effectiveness of the Reserve Bank of Australia (RBA)'s interventions over 1986–2003 using recently released revised data on intervention activities that are free from various errors plagued the previously released data set, and (ii) to investigate the nature of indirect effects of interventions on short- and long-term futures interest rates and see if there is a potential conflict between sterilized interventions and monetary policy effectiveness.3 To address the simultaneity issue we improve on the modelling strategy of Kim et al. (2000) and investigate both the contemporaneous (overall and disaggregated) and lagged influences of the RBA interventions in these markets. The important results of our investigations are as follows. Firstly, there are significant simultaneity effects of the interventions in both the first and second moments of the contemporaneous daily exchange rate changes which is consistent with the literature. Secondly, there are a number of offsetting influences which represent intervention effectiveness. Lagged interventions partially reversed this contemporaneous effect in both the first and second moments. In addition, large interventions and those that were a part of a continued campaign over a number of days had the desired effect of reducing the current exchange rate trends on intervention days. Without these interventions, the trends would have been more pronounced. Thus, the RBA's interventions were mostly successful in moderating the first and second moment trends in the currency. Thirdly, we show significant indirect effects of interventions in the interest rate futures markets. The contemporaneous effect of the RBA purchasing the Australian Dollars (AUD) was to raise both the short-and long-term interest rate futures and also their volatilities. Considering the stated aim of the interventions over the relevant periods was to relive downward pressures on the AUD in the midst of monetary easing, the interest rate raising effect of the interventions undermined the then current monetary policy stance. However, lagged and continued interventions had mild corrective influences, whereas large interventions tended to cause more volatility in the 90-day futures. In the case of the 10-year rate, the volatility was significantly higher on the days of continued and large intervention. We provide significant contributions to the intervention literature by offering important detailed insights into the direct and indirect influences of the RBA's intervention activities. The results of the paper have important implications for the policy makers (the RBA), foreign exchange market participants, and foreign investors in the Australian financial markets at large. The remainder of this paper is structured as follows. We discuss the data employed in this paper in the next section. Section 3 provides detailed discussions on the empirical methodologies employed, and the estimation results for the foreign exchange market and the 90-day and 10-year interest rate future markets are discussed in sections 4 and 5, respectively. Conclusions are presented in Section 6.
نتیجه گیری انگلیسی
Whether foreign exchange market interventions have been effective in moderating exchange rate trends and volatility is an empirical question that continues to promote substantial and ongoing controversy. Utilizing recently release revised intervention dataset that is free from various transcription errors, we provide new evidence on the effects of the RBA's interventions in the USD/AUD market and the interest rate futures markets in Australia. We make a number of important contributions to the intervention literature at large. We found that large interventions were particularly effective in moderating excessive trends in the exchange rate. Cumulative interventions were also found to break bandwagon surges, though the effect was not as strong as large interventions. However, we found that in order to effectively calm volatility, intervention operations should be carried out over a number of days. These results demonstrate that the homogenisation of information, and alleviation of uncertainty, is more adequately facilitated if the same message is repeatedly released to the market. Another important contribution we make is the investigation into the effect of the RBA's foreign exchange interventions on the first and second moments of the 90-day and the 10-year interest rate futures traded on the Sydney Futures Exchange. We found that there were both price and volatility effects of the intervention operations. In particular, cumulative interventions tended to calm volatility in the 90-day rate while large interventions exacerbated it. These results are consistent with the foreign exchange estimations where cumulative interventions had a calming effect while there was some mild evidence since 1995 to suggest that large interventions caused second moment disturbances in the exchange rate. Thus, the report the existence of common fundamentals between the two types of markets via interest rate parity conditions and/or monetary policy signalling approach. The results have strong policy implications for the RBA. Large interventions were found to be the most effective means for the RBA to break speculative surges in the exchange rate. However, this was at the cost of higher short term interest rate volatility and potential reduction in monetary policy potency in some periods. Consequently, the RBA may have to decide whether it is willing to promote a greater amount of volatility in the debt market, and threaten overall financial stability, in order to influence the trend of the exchange rate in the desired direction. That is, there is a trade-off between the two markets in terms of intervention effectiveness. In addition, the RBA should consider spreading its interventions over a number of days to maximize the exposure. Cumulative interventions have shown to be effective in moderating the exchange rate trends and partially offsetting a contemporaneous rise in volatility in both the foreign exchange and the 90-day future interest rate markets. In addition, the results we report could provide information that is useful for international investors for asset allocation and overall risk management purposes.