مداخله در بازار ارز و انتظارات: نرخ ارز ین / دلار

Using press reports and official statistics on central bank intervention in conjunction with options market data, we find that central bank intervention had no statistically significant systematic impact on the mean or higher moments of the exchange rate during 1993–2000. We also find that Japanese authorities appeared to intervene mainly in response to deviations of the exchange rate from some implicit target levels and to a rise in market uncertainty. U.S. monetary authorities intervened only in cooperation with the Japanese authorities.

This paper presents empirical evidence on the relationship between central bank intervention and market expectations of the daily yen/dollar exchange rate. Over the past years, a number of studies have analyzed the effect of intervention on the exchange rate level as well as on the instantaneous and expected volatility of the exchange rate of the yen/dollar, the mark/dollar and other exchange rates. The results are generally mixed and depend on both the sample period being investigated and the intervention strategies being used. This paper provides further evidence on this issue in four ways. First, it focuses not only on the expected level and variance of the exchange rate, but on the entire distribution of expected exchange rates, which is derived from option prices. The moments of this distribution allow a more complete characterization of market sentiment. Second, this paper uses two new data sets on intervention episodes. One data set covers actual intervention episodes and was recently provided by the Japanese Ministry of Finance through its website. The other data set contains interventions perceived by market participants and is based on Reuters reports. This second data set is more comprehensive than the data sets based on newspaper articles that have been used in the literature. Third, this paper uses a comprehensive data set of “news” about macroeconomic variables that helps to distinguish the effect of intervention from that of the arrival of other relevant information. Finally, the paper also looks at two directions of causality in the relationship between intervention and market expectations. We first examine the reaction function of monetary authorities and estimate how changes in the moments of the expected distribution of future exchange rates affect the likelihood of central bank intervention. We then investigate the effect of intervention on the moments, taking account of the fact that intervention and market expectations are determined simultaneously. Our analysis is conducted over the period from September 20, 1993 to April 30, 2000.1 Our results suggest that during the period of dollar weakness between 1993 and 1996, the Bank of Japan seems to have responded to deviations of the spot exchange rate from what traders perceived as implicit target ranges. It also appears to have intervened on average as a response to a rise in market uncertainty, consistent with statements made by central banks in the past that emphasized their attempt to “calm disorderly markets”. We find evidence that between 1997 and 2000, the Bank of Japan seems to have intervened mainly following a rise in uncertainty. During the whole sample period, the Federal Reserve intervened always in conjunction with the Bank of Japan. Consistent with some of the findings in the literature, our regression analysis suggests that, on average, intervention in the yen/dollar market had no statistically significant effect on the mean of the PDF during the period under review. We also find that, between September 1993 and April 2000, central bank intervention directed at the yen/dollar exchange rate was not associated with a significantly higher variance of expected future spot rates. This result indicates that, on average, central bank intervention was not followed by an increase in uncertainty in the market about future exchange rate movements. Consistent with our results for the mean, we find that, on average, intervention did not have a significant effect on the skewness. Hence, following intervention in support of the dollar, market participants did not change the weight that they put on a much stronger rather than a much weaker dollar with respect to the forward rate. Finally, we find that intervention did not influence kurtosis in a significant way. In terms of intervention strategies, results for the period 1993–1996, when all interventions involved dollar purchases, indicate that at 90% or 95% confidence levels, the impact of coordinated or officially announced interventions was not statistically different from that of unilateral interventions.2 Our results can be easily reconciled with the consensus view on the effects of intervention in the foreign exchange market. According to this view (see the literature review in the following section) sterilized intervention can be effective if it is announced publicly, coordinated across central banks, and most importantly, consistent with underlying fiscal and monetary policies. When central banks entered foreign exchange markets in a coordinated, public fashion in the 1980s—particularly around the Plaza or Louvre Agreements—these interventions did succeed in affecting traders' expectations of future exchange rate movements (see Galati and Melick, 2002). For the period under study in this paper, however, all of these conditions were not met, with the result that intervention is found to have little influence on market outcomes and expectations. The remainder of the paper is organised as follows. Section 2 reviews the main contributions to the literature on the effect of central bank intervention and on central banks' reaction functions. In Section 3, we describe our intervention data, the data on macroeconomic news and the option market data that was used to estimate the empirical PDFs. We also discuss the interpretation of PDFs. In Section 4, we first estimate reaction functions for the Federal Reserve and the Bank of Japan that relate intervention to past market conditions. We then analyze the impact of intervention on market expectations. Section 5 concludes.

This paper presents empirical evidence on the relationship between market expectations of the yen/dollar exchange rate and central bank intervention carried out by Japanese authorities and the Federal Reserve. We examine the period September 20, 1993 to April 30, 1996, during which all central bank intervention in the yen/dollar market was directed at supporting the dollar, and the period November 1997 to April 2000, in which central banks both bought and sold dollars for yen. We use official data on intervention provided by the Federal Reserve and, recently, by the Japanese Ministry of Finance, as well as a new data set on intervention episodes perceived by market participants, which is based on Reuters reports. The paper uses the moments of the expected distribution of future exchange rates as a way to describe market expectations. The four moments of the empirical density functions allow us to give a complete characterization of market expectations. The paper quantifies how market expectations affected the likelihood of central bank intervention and the impact of intervention on market expectations. We find that Japanese and U.S. authorities followed different intervention strategies during the periods 1993–1996 and 1997–2000. During the first sample period, Japanese authorities appeared to respond to deviations of the spot yen/dollar rate from some implicit target ranges. We also find evidence that the Bank of Japan intervened to reduce market uncertainty. For the sub-sample 1997–2000, only increases in market uncertainty influenced the likelihood of intervention by the Bank of Japan. Between 1993 and 2000, the Federal Reserve only intervened in support of interventions by the Japanese authorities. Consistent with the findings of some of the literature, the regression results for the impact of intervention on the moments of the PDF suggest that, on average, intervention in support of the dollar had no statistically significant effect on the mean of expected exchange rates. In particular, we do not find evidence of a contemporaneous impact of intervention on the forward rate, nor any deferred effect. Consistent with the results for the mean of the PDF, we do not find evidence that, on average, intervention influenced the skewness of the PDFs, i.e., traders' balance between a much stronger and a much weaker dollar around the forward rate. An important finding is that, between September 1993 and April 1996, and between November 1997 and April 2000, intervention directed at the yen/dollar exchange rate was not associated with a higher variance of expected future spot rates. This suggests that on average central bank intervention did not influence the uncertainty prevailing in the market regarding future movements in the yen/dollar rate. In summary, our results are in line with the consensus view that sterilized intervention can be effective if it is announced publicly, coordinated across central banks, and most importantly, consistent with underlying fiscal and monetary policies. As we are able to control for public announcement and central bank coordination, our findings suggest that intervention during our sample period was not consistent with underlying fiscal and monetary policy and therefore did not have a lasting influence on exchange rate moments.