مداخله در بازار لحظه ای ین ـ دلار : داستان قیمت، نوسانات و حجم

We test the effectiveness of Bank of Japan (BOJ)’s foreign exchange interventions on conditional first and second moments of exchange rate returns and traded volumes, using a bivariate EGARCH model of the Yen/USD market from 5-13-1991 to 3-16-2004. We also estimate a friction model of BOJ’s intervention reaction function based on reducing short-term market disorderliness and supplementing domestic monetary policy. Important finding of this study are that: (i) we find ineffectiveness of BOJ interventions in influencing exchange rate trends pre-1995, in general, but effectiveness post-1995; (ii) FED intervention amplified the effectiveness of the BOJ transactions; (iii) interventions amplified market volatility and volumes through a ‘learning by trading’ process; (iv) BOJ’s interventions were based on ‘leaning against the wind’ motivations on the exchange rate trend and volumes; and (v) BOJ interventions were vigorously used in support of domestic monetary policy objectives post-1995. Though some of our findings confirm recent studies, our analysis goes deeper to provide new findings with important implications for central banks and foreign exchange market participants.

We examine the relations between exchange rate changes, traded volumes and central bank interventions in the Japanese Yen/United States dollar (Yen/USD) market. The Yen/USD was the second most traded currency pair (17% of all global transactions valued at USD1.88 trillion per day in 2004, the €-USD having 28%) in 2004 (BIS, 2004). Further, 83% of transactions occurred between foreign exchange dealers and brokers. This suggests that the huge daily volumes traded are largely driven by the need for resolution of asymmetric information issues amongst dealers and brokers, often prompted or halted by the interventions of the Bank of Japan (BOJ) and the United States Federal Reserve (FED). The BOJ has been one of the most active central banks in the foreign exchange markets in the last 15 years. It has the distinction of having made the biggest ever intervention on a day – USD20.3 billion sales on April 10 1998, when all East Asian currencies were depreciating, and Japan, on the brink of a full scale financial crisis, was proposing a massive fiscal expansion package. At the end of 2003, the BOJ had accumulated USD653 billion of reserves, which represented nearly 22% of all official reserves held globally. Since the release of the historical intervention data by the BOJ in July 2001, a long list of researchers have investigated various aspects of the BOJ interventions. This literature has tested intervention effectiveness by determining whether the level and volatility of exchange rate returns are affected in the desired direction, or if prominent trends are reduced (Chaboud and Humpage, 2003, Fatum and Hutchison, 2003, Beine, 2004, Nagayasu, 2004 and Frenkel et al., 2005). Possible responses of the BOJ’s interventions to exchange rate returns have been examined (Ito, 2002, Beine and Bernal, 2004 and Ito and Yabu, 2004). This literature has focused on higher frequency issues in a market environment (Dominguez, 2003). However this focus needs extending in the case of the BOJ, which has used foreign exchange intervention as an alternative instrument to achieve its domestic monetary policy objectives (for example, see Fatum and Hutchison, 1999, Sellon, 2003 and Vitale, 2003). With short-term interest rates close to zero from 1995, intervention to stop the Yen appreciating (and perhaps to engineer a depreciation) was used to prevent further deflationary pressures. However, there is a growing agnosticism about the ability of a central bank to successfully influence exchange rates. For example, Schwartz (2000) claimed that US and European monetary authorities no longer believe that intervention works. She challenges the BOJ for its obduracy in pursuing sterilized intervention2 as an instrument targeted to domestic objectives. The basis for this challenge is essentially that there is no hard and robust evidence from around the world that interventions successfully affect exchange rates. What evidence exists suggests that interventions sometimes work because of the ‘presence’ effect of a central bank rather than the size of its operations and if it surprises the market. However, if there is going to be good evidence of successful size effects of transparent interventions, the chances are that it will be found from the BOJ’s activities post-1995 when it very actively intervened as a non-traditional alternative to monetary policy implementation. In fact, we report such findings. Microstructural modelling of financial markets has stressed the relation between asset price returns and volumes. In foreign exchange markets, this suggests that intervention analysis should be extended to incorporate volume processes: interventions ought to matter for volumes, and volumes ought to matter for interventions. Interventions should affect foreign exchange trading volumes because they change underlying market fundamentals – or beliefs about them – and should be reflected in trading activity. Further, central banks are likely to value the signal or information content of transaction volumes in making their intervention decisions. An examination of central bank interventions in the context of both exchange rate returns and volumes has never been done before because data on volumes is not readily available. We address this issue using a measure of trading volume – the daily reported trades by Tokyo foreign exchange brokers to the BOJ. Our joint estimation of exchange rate returns and volumes using a bivariate EGARCH model should not only yield more efficient results than separate estimations, but provide insights into the dynamics of the return-volatility-volume relationships. The main questions we address are: Was the BOJ able to achieve desirable effects on the trend and volatility of returns and volumes with its interventions? What were the main factors driving the BOJ’s intervention reaction function? Did the BOJ use intervention as a monetary policy instrument, and was it successful in doing so? The main findings of this paper are: (i) From a model that jointly estimates exchange rate returns and volumes, large and infrequent BOJ interventions after 1995 successfully corrected undesired exchange rate movements. However this came at the price of higher market volatility and volumes. (ii) ‘Coordinated’ interventions with the FED were successful only after 1995. (iii) From estimates of the intervention reaction function, the BOJ did intervene to correct deviations from an exchange rate trend, but this was moderated when there was high unexpected volume (but not exchange rate volatility). (iv) Lower interest rates had no effect on exchange rate returns before 1995, as the BOJ sold US dollars to stop depreciation of the Yen, but they did cause depreciation (or reduce appreciation) after 1995. Japan found itself in a liquidity trap situation with deflation and a growth slump, and so the BOJ bought USD in unsterilized interventions to encourage depreciation of the Yen. The results in (i) and (ii) are largely consistent with the recent literature (Ito, 2002, Chaboud and Humpage, 2003 and Nagayasu, 2004), however, we provide further significant insights into the nature of the effectiveness, putting to rest many of the concerns raised in Schwartz (2000). Our reaction function estimations for the BOJ intervention determinants substantially extend those considered by Ito and Yabu (2004), thus providing a more complete analysis of intervention in the Yen/USD market. Accordingly, the results in (iii) and (iv) represent new and important findings. The results discussed in the paper have important implications for central banks, and for foreign exchange market participants alike.

Our estimations of a bivariate EGARCH model of exchange returns and trading volume in the Yen/USD market lend support to the ‘mixed distribution hypothesis’ and a ‘learning by trading story’. We found that the change of intervention philosophy by the BOJ around July 1995 resulted in successful market responses to its intervention transactions. Before 1995, interventions were small and frequent and the contemporaneous effects of interventions appeared to be counter-productive. The effects on trade volumes were similar – volume was successfully calmed only on unusual days of sustained intervention – though interventions of above-average size also helped. The interventions by the FED were of little consequence in this sample. After 1995, interventions were less frequent and substantially larger in size. In contrast to results widely obtained in the literature the contemporaneous effects of these large interventions were to move the Yen/USD rate in the desired direction. Rare, but large coordinated interventions by the US FED now added to the effectiveness of the BOJ interventions. Overall, while small interventions could be destabilizing, large, sustained and coordinated interventions were actually successful. However they came at a price of more market disorderliness in the form of higher volumes and volatility. Our result that ‘size matters’ is an important finding, since the literature suggests that it is the ‘presence’ of the central bank that matters, not size of the activity as such. Similarly the positive effects of sustained (and therefore expected) interventions are important because previous results tend to suggest that only ‘surprise interventions’ are effective. We argue that continued presence improves the signalling effect of intervention. The BOJ were motivated to intervene to correct short-term trend deviations and unexpectedly high volume in both samples. This is consistent with the ‘leaning against the wind’ hypothesis, but interestingly, exchange rate volatility had no impact. However volume mattered to the BOJ, not exchange rate volatility, as a measure of disorderliness. We argue that it preferred to use an observable rather than a statistically derived variable as a measure of disorderliness. Finally, BOJ interventions responded significantly but differently to changes in the Japanese interest rate before and after 1995. Before 1995, it sold US dollars when the Japanese discount rate fell, thus preventing the Yen from depreciating and creating inflation. After 1995, it bought US dollars as Japanese interest rates fell towards zero. Thus, as it fell into a liquidity trap, the BOJ used unsterilized foreign exchange intervention as an alternative instrument to achieve the objectives of its domestic monetary policy. It intervened to weaken the Yen, or at least to inhibit any appreciation in the face of deflation in its slumping economy. We are thus able to conclude that the BOJ’s intense interventions post-1995 had significant desirable effects on the foreign exchange market, and these successful actions were in part motivated by domestic monetary policy objectives.