آیا همه مداخلات بانک مرکزی برابر خلق شده اند؟ تحقیقات تجربی

This study investigates the relationship between Central Bank interventions and technical trading rule profitability in the spot foreign exchange market. Because interventions are not necessarily exogenous events, we analyze the relationships between interventions by the G-3 Central Banks, financial market conditions, changes in monetary policy and technical trading profitability. By considering announced, unannounced, unilateral and coordinated interventions separately, we provide more insight into the interrelationships between these factors than previous studies. We find that the level of technical trading profits and market uncertainty increase preceding and remain high during interventions, especially announced and coordinated, but decrease afterward. A preliminary investigation of the possible role of a time-varying risk premium around interventions cannot be rejected.

Because the foreign exchange market is the largest and arguably most important financial market in the world, it is believed that if any financial market should be efficient it should be this market. Unfortunately tests of even the weakest form of market efficiency are rejected in the foreign exchange market – technical analysis is consistently profitable. This apparent inefficiency has persisted from the first studies (Poole, 1967; Dooley and Shafer, 1976 and Dooley and Shafer, 1983) to the most recent studies (Neely et al., 1997; Gençay, 1999; LeBaron, 1999; Neely, 2002). Researchers have proposed that Central Banks may play a role in this apparent inefficiency because they can influence the supply and demand for currencies at any time. Consequently exchange rates are not always determined by the laws of supply and demand required for market efficiency (Friedman, 1953; Dooley and Shafer, 1983; Corrado and Taylor, 1986; Sweeney, 1986 among others). Consistent with the hypothesis that Central Banks may be related to this apparent inefficiency, Szakmary and Mathur (1997), Neely (1998) and LeBaron (1999) find that technical trading profits were correlated with periods of Federal Reserve intervention activity during the 1980s and early 1990s. To better understand this relationship Neely (2002) uses higher frequency data and finds that the profitability of technical analysis actually begins before the start of intervention activities. We build on these studies by investigating differences across types of interventions (e.g. announced versus unannounced) and whether interventions and these periods of apparent inefficiency may be related to some other economic factor(s). We address these issues using a dataset that is longer and more comprehensive than those used in previous studies. We have both a period of extensive intervention activity (the 1980s and early 1990s) as well as a period with little intervention activity (the mid- to late-1990s). This permits us to compare different types of interventions by the Federal Reserve, the Bank of Japan and the Deutsche Bundesbank. We compare announced and unannounced interventions as well as coordinated and unilateral interventions for these Central Banks.1 The existing empirical literature tends to concentrate on Fed interventions with little consideration for differences between these types of intervention. To understand the relationships between these different types of interventions, technical trading returns and other factors theory suggests may instigate and/or influence the effectiveness of interventions we use a vector autoregression (VAR) technique. We investigate, for example, the relationships between factors such as the volatility of exchange rates and Central Bank interventions – Fed policy states it intervenes “to calm disorderly markets” and “signal” the desired level of the exchange rate to the market (Cross, 1998). We also investigate several relationships between financial markets, interventions and exchange rate movements proposed by theories of exchange rate determination (for a survey see Frankel and Rose, 1995). All of these relationships are analyzed in the context of technical trading profitability to see if they can help explain this apparent inefficiency. We start our analysis by verifying that technical analysis can generate statistically and economically significant returns in the Deutsche Mark-$ and Japanese Yen-$ markets in our sample. We find an average annualized excess return of about 10% in the DM-$ market, for example, which is statistically significant. Because the set of rules we consider were profitable over both our sample period and an out-of-sample test period, it is unlikely they are the result of an ex-post bias. The economic significance of the returns is suggested by their robustness to market frictions such as transaction costs and their Sharpe Ratio being significantly better than for the S&P500 (despite the exceptional performance of the stock markets over this period). It is noteworthy that the profitability was concentrated in the 1980–1995 period – the period of active intervention. In our VAR analysis we find that information on Central Bank interventions (especially announced and coordinated interventions by the Fed and Bundesbank), some changes in monetary policy and changes in market uncertainty are related to technical trading profitability. The technical trading returns and measures of foreign exchange market uncertainty increase preceding interventions, peak on the day(s) of intervention activity and decrease on the last day and afterward. These results suggest that interventions change foreign exchange market expectations and end once they have “calmed disorderly markets”. The differences we find across types of interventions provide some insight into the apparently contradictory findings of many previous studies which treated all interventions in the same fashion (see Edison, 1993 or Frankel and Rose, 1995 for a discussion). The relationship between the level of technical trading returns, market uncertainty and interventions suggests that these profits may be the result of a risk premium at these times and not market inefficiency (increasing market uncertainty is frequently, but not necessarily, associated with the presence of a risk premium). Using an international CAPM, we are unable to reject the possible presence of a time-varying risk premium in the technical trading returns correlated with Central Bank intervention activity, especially announced and coordinated. The paper is organized as follows. A discussion of the data makes up Section 2. In Section 3, we measure and characterize the economic and statistical significance of the technical trading returns. Section 4 discusses the results from the VAR tests. Section 5 characterizes the behavior of foreign exchange and technical trading returns around interventions. A summary of the main results and areas for future research concludes.