آمریکا و اخبار اقتصاد کلان ژاپن و نوسانات بازار سهام در آسیا و اقیانوس آرام
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|19562||2009||17 صفحه PDF||سفارش دهید||10387 کلمه|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Pacific-Basin Finance Journal, Volume 17, Issue 5, November 2009, Pages 611–627
I use a new comprehensive dataset to analyze the impact of ten U.S. and six Japanese macroeconomic announcements on stock market volatility in Japan, Hong Kong, South-Korea and Australia. A GARCH model that allows for multiplicative announcement effects and asymmetries is employed. Overnight conditional variances are significantly higher on announcement days and significantly lower on days before and after announcements, especially for U.S. news. The impact of announcements on implied volatilities, in contrast, is much weaker. Out-of-sample trading strategies that systematically buy delta-neutral straddles on announcement days generate statistically significant profits, but these disappear after transaction costs are taken into account.
Stock market volatility plays a key role in portfolio construction, risk management and the pricing of derivative securities. Furthermore, changes in market risk affect expected returns of all individual securities in asset pricing models. Understanding the determinants of stock market volatility is therefore important to both academics and practitioners. Since the contributions of Engle (1982) and Bollerslev (1986), substantial progress has been made in the empirical modeling of time-varying volatilities in financial markets. However, most volatility models only include past (squared) returns and lagged volatilities and remain silent on the fundamental determinants of volatility. The studies that do link macroeconomic fundamentals to asset return volatility focus primarily on U.S. financial markets and conditional volatility. In contrast, this paper analyzes stock markets in Asia-Pacific and studies the impact of macroeconomic news on both conditional and implied volatility. The first contribution of this paper is to show that macroeconomic announcements from the U.S. and Japan have a different impact on implied and conditional stock market volatility. International macroeconomic news strongly affects conditional stock market volatility. The impact on implied volatility, on the other hand, is very weak. This discrepancy suggests an interesting difference in how primary and derivatives markets incorporate economic fundamentals in Asia-Pacific. If underlying stock markets significantly move in response to macroeconomic announcements, but this is not reflected in ex-ante options prices, profitable trading strategies can possibly be formed. This would constitute important information for the (in)efficiency of international options markets. Testing whether this difference can be exploited economically forms the second contribution of this paper. The studies of Connolly and Wang (2003) and Kim (2003) are most closely related to the current paper. Connolly and Wang (2003) study the causes of contemporaneous return comovement in U.S., U.K., and Japanese equity markets. The authors analyze the impact of news announcements made in these three countries during the period 1985–1996. Using various GARCH-specifications, macroeconomic announcements cannot explain comovements between international markets. By focusing on volatility rather mean return dynamics, I take a different perspective than Connolly and Wang (2003). Kim (2003) investigates the nature of macroeconomic announcement spillovers from the U.S. and Japan to conditional stock market volatility in Australia, Hong Kong and Singapore. The analysis includes six macroeconomic surprises from the U.S. and Japan from Money Market Services International (MMS) over the period 1991 to mid-1999. Using an exponential GARCH (or EGARCH) model, Kim (2003) shows that announcements in the U.S. and Japan have a significant impact on both returns and conditional volatilities in Asia-Pacific. Furthermore, markets respond differently to good and bad news, with bad news announcements in general leading to increases in volatilities in the other markets. Apart from using a different data source with many macroeconomic announcements and a long and recent sample period, I extend the work of Kim (2003) in two dimensions. First, in addition to analyzing conditional volatility with GARCH models, I study the impact of macroeconomic announcements on implied volatilities. This has, to the best of my knowledge, not been done before. Interestingly, I document a lack of importance of news about fundamentals for implied volatilities. Second, I test whether the difference in how primary and derivatives markets incorporate information can be exploited economically by systematic option trading strategies. Most prior studies take a U.S. perspective. Pearce and Roley (1985), for example, show that monetary policy surprises significantly affect stock prices in the U.S. The evidence for inflation surprises, however, is weaker and there is there is no noticeable impact of real activity surprises. McQueen and Roley (1993) show that the response of U.S. stock prices to news is much stronger after allowing for different stages of the business cycle. When economic conditions are good, the stock market reacts negatively to news about higher real economic activity. Flannery and Protopapadakis (2002) estimate a GARCH model and document that macroeconomic variables affect both conditional returns and volatilities in the U.S. Whereas most studies typically examine only a single market, Kim et al. (2004) consider the impact of economic news on U.S. stock, bond and foreign exchange markets. With a GARCH model, the authors show that markets respond primarily to macroeconomic surprises, rather than the act of releasing information about fundamentals itself. Balance of trade news is the most important release for the foreign exchange market. For the bond market, news about the internal economy is most important and for the stock market consumer and producer price information. Finally, volatility increases for news announcements of some variables, but falls with releases for others. Andersen et al. (2007) study the response of international financial markets to U.S. news releases using high-frequency data. Macroeconomic announcements cause conditional mean returns to jump. Once the stage of the business cycle is taken into account, the response of the foreign exchange and equity markets is as strong as the response of bond markets. Another strand of the literature shows that there are significant inter-market linkages and volatility spillovers between international financial markets.1Koutmos and Booth (1995) document strong volatility spillovers between stock markets in New York, Tokyo and London, especially for negative returns. Furthermore, links have increased substantially since the 1987 stock market crash. Ng (2000) also studies volatility spillovers from Japan and the U.S. to equity markets in Asia-Pacific. Although U.S. spillovers are most important, Japanese spillovers also play a role. Hamao et al. (1990) show strong evidence of spillovers from the U.S. and the U.K. to Japan, but much weaker evidence from Japan to the U.S. and the U.K. Bae and Karolyi (1994) consider overnight and daytime return and volatility dynamics in the U.S. and Japan. Volatility spillovers are substantially underestimated if the asymmetric impact of news is ignored. These studies focus on the international spillover effects of return innovations from one market to another. The focus of this paper is different in the sense that local conditional return volatility is explicitly modeled as a function of international macroeconomic news. The current study therefore gives content to foreign macroeconomic influences rather than leaving these as unspecified return innovations. My research design is as follows. I use a new and comprehensive dataset from Bloomberg with real-time macroeconomic expectations and announcements for six Japanese and ten U.S. variables. I start with an analysis of the impact of macroeconomic announcements on conditional overnight return variances in Japan, Hong Kong, South-Korea and Australia. The GARCH model that I employ allows for asymmetry and, apart from announcement effects, also models pre- and post-announcement effects. These effects have often been overlooked in previous work, but turn out to be important empirically. Conditional variance is abnormally low on days immediately prior and after announcement days and abnormally high on announcement days. The effects are much stronger for news emanating from the U.S. than from Japan. Interestingly, this is even the case in Japan. For U.S. news announcements, announcement-day variance is boosted by as much as 28 to 67%. Pre- and post-U.S.-announcement-day conditional variances are between 10 and 20% lower than normal. Several strict robustness tests do not alter these conclusions. Since macroeconomic news boosts conditional variance, a natural follow-on question is whether the importance of macroeconomic information is also reflected in implied volatilities. Interestingly, the impact of news on implied volatilities is much weaker. When I distinguish between good and bad news and small and big surprises, the results do not improve. Conditioning the response on the stage of the business cycle, working with levels of implied volatilities rather than first differences and leading or lagging the announcements does also not make any qualitative difference. This suggests that primary equity markets and derivatives markets in the Asia-Pacific region incorporate economic fundamentals differently. If this difference can be exploited by option trading strategies, this would violate efficiency of options markets. I test this hypothesis directly by designing option trading strategies that exploit this feature of the data. These involve buying delta-neutral combinations of puts and calls before announcements and selling them afterwards. The option strategies generate statistically significant mean returns over the sample period for all markets. The median daily return of the straddle strategies is substantially lower than the mean and the payoff profile is highly positively skewed. This means that the return is relatively low on typical days, but occasional extremely high returns on announcement days increase the mean substantially. When transaction costs are taken into account, both mean and median returns decrease rapidly and become negative. This leads to the conclusion that the efficiency of options markets in Asia-Pacific cannot be rejected based on the evidence presented in this paper. The organization of the remainder of this paper is as follows. Section 2 describes the announcement, financial futures, and implied volatility data. Section 3 discusses the econometric methodology and Section 4 presents the empirical results. Section 5 concludes.
نتیجه گیری انگلیسی
This article addresses the empirical question of how international macroeconomic news announcements affect conditional and implied stock market volatility in Japan, Hong Kong, South-Korea, and Australia. I use a comprehensive real-time dataset with ten U.S. and six Japanese announcements during the period 1996–2007. There are three main conclusions from this paper. First, international macroeconomic news announcements have a statistically significant and economically large impact on conditional (GARCH) stock market volatility. U.S. announcements are more important than Japanese announcements, even for Japan. This may be caused by the relative importance of the U.S. economy for the world and Asia-Pacific in particular. Another explanation is the more timely release of U.S. statistics. U.S. announcements boost conditional variances between 28% and 67%, whereas pre- and postannouncement days have significantly lower conditional variances (between 10% and 20% lower than normal). These effects survive several strict robustness tests. Second, the impact of international macroeconomic announcements on implied volatilities is very weak. Separating big from small surprises and good from bad news does not alter this conclusion. Analyzing implied volatility levels or first differences and conditioning the response on the stage of the business cycle does not make a difference either. These effects are puzzling and seem to suggest that option markets in Asia-Pacific do not rationally reflect the impact of announcements on volatility. Third, strategies designed to exploit this apparent discrepancy by buying delta-neutral straddles on announcement days generate statistically significant profits before transaction costs. Mean returns are substantially higher than median daily strategy returns. This indicates that large moves in volatility associated with macroeconomic news substantially benefit the returns of the strategy. When I take transaction costs into account, the results deteriorate rapidly and substantially. For even modest levels of transaction costs, mean returns are not significantly different from zero anymore. Hence, inefficiency of the options markets in Asia-Pacific cannot be rejected based on this study.