آیا آب و هوا بر نوسانات بازار سهام تاثیر می گذارد؟
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|10957||2010||10 صفحه PDF||سفارش دهید||5580 کلمه|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Finance Research Letters, Volume 7, Issue 4, December 2010, Pages 214–223
This paper investigates the empirical association between stock market volatility and investor mood-proxies related to the weather (cloudiness, temperature and precipitation) and the environment (nighttime length). Overall, our results suggest that cloudiness and length of nighttime are inversely related to historical, implied and realized measures of volatility. The strength of association seems to vary with the location of an exchange on Earth with respect to the equator. Weather deviations from seasonal norms and dummies representing extreme weather conditions do not offer additional explanatory power in our datasets.
Investment professionals appear to have been well aware of the behavioral effects of the weather for over a century now. Characteristically, Nelson (1902, p. 163) reports: ‘During normal markets, brokers have observed that the psychological factor is so strong that speculators are not disposed to trade as freely and confidently in wet and stormy weather as they are during the dry days when the sun is shining, and mankind is cheerful and optimistic’.2 More recently, several papers have investigated in depth the links between stock market returns and prevailing weather conditions. The main empirical finding in this literature is the so-called ‘sunshine effect’ according to which cloudiness, as measured by cloud cover, has a significant negative correlation with daily equity index returns (see Saunders, 1993, Hirshleifer and Shumway, 2003 and Chang et al., 2008 among others). This relationship has been explained using arguments from psychology on the basis of ‘mood misattribution’. Simply put, sunny weather is thought to influence the mood of some investors making them more optimistic and thus more willing to enter into long positions, which in turn leads to higher returns. Other weather and environmental variables which have been considered in the financial literature as mood-proxies include, among others, temperature (e.g., Cao and Wei, 2005), daylight savings time changes (see, e.g., Kamstra et al., 2000) and the ‘Seasonal Affective Disorder’ (SAD, see, e.g., Kamstra et al., 2003 and Garrett et al., 2005). Rather than concentrating on expected returns, a recent strand of research has examined the effect of weather and environmental factors on volatility. This is of great academic and practical interest since volatility underlies a variety of key financial decisions, problems and applications in asset valuation, portfolio theory, derivatives pricing, risk management, etc. The main obstacle in this research is that volatility is largely unobservable. In the present paper, we consider all three of the most widely used proxies: historical, implied and realized volatility (for a detailed description of these and relevant references see Poon and Granger, 2003 and Mills and Markellos, 2008). Specifically, we extend in four main directions the empirical literature which examines the impact on volatility of cloudiness, variation in nighttime hours, temperature and precipitation, respectively. First, in addition to the three deseasonalized weather variables, we consider also the effect of absolute deviations from seasonal norms and of dummies which reflect extreme weather conditions. This is because mood variations could be potentially better correlated with the magnitudes of deviations, or, with extreme deviations of weather, from seasonal norms, respectively. For example, we may feel particularly uncomfortable when the weather is (significantly) hotter or colder than expected during a particular season. In this manner, deviations of weather variables from seasonal averages may then lead to variations in mood states and to shifts in volatility. Since the strength of association between weather/environmental variables and stock market returns has been found to depend also on stock exchange location (see, e.g., Keef and Roush, 2007), we also consider the effect of latitude when looking at international data. Second, we analyze the effects on historical volatility using an ARCH-type model on the extensive dataset of Hirshleifer and Shumway (2003) which consists of stock market index returns for 26 stock exchanges internationally between 1982 and 1997. Third, we analyze four implied volatility indices for the CBOE (namely: VIX, VXO, VXN and VXD) along with the term structure of the VIX volatility index (seven volatility duration buckets). Implied volatility is derived from traded options and is a measure of expected volatility as this is perceived by investors in the derivatives market. The variety of indices used enhances the robustness of our results and allows us to see if the effect of weather and of environmental factors depends on the composition of the volatility index and the underlying option market investment horizon. Finally, we analyze realized volatility which is constructed on the basis of high-frequency returns for the S&P 500 index. Realized volatility offers a great advantage in that it is considered as the most accurate representation of the unobserved volatility process. Empirical evidence is mixed between the existing studies that have investigated the effects of weather and environmental conditions on volatility. Chang et al. (2008) show that New York City cloudiness has a significant positive effect on intraday volatility of NYSE firms over the entire trading day. Two volatility proxies are used by these authors: one based on the range of the intraday prices and the other on the basis of the standard deviation of the bid-ask mid-point returns. Both of these proxies are uncommon in the literature and their accuracy is unknown. Dowling and Lucey (2008) study the empirical effect of seven mood-proxies on both the returns and variances of 37 national equity market indices and 21 small capitalization indices. They employ GARCH-type processes to approximate and model the variations in the conditional variance of returns. Their results show that wind, precipitation, geomagnetic storms, daylight savings time changes and the SAD are all positively related to conditional volatility for most of the indices considered. Kaplanski and Levy (2009) consider the effect of SAD and temperature on the VIX option’s implied volatility index which is traded in the Chicago Board Options Exchange (CBOE). They use also a measure of so-called ‘actual’ volatility based on the historical standard deviation of a monthly window of daily returns. The authors find that the number of daylight hours (temperature) is negatively (positively) related only to the ‘perceived’ volatility proxied by the VIX and not to the ‘actual’ historical volatility measure. Another study which indirectly shows a positive relationship between volatility and bad weather is Kliger and Levy (2003). These authors find using S&P 500 index options data that bad mood, as proxied by total cloud cover and precipitation, make investors place higher-than-usual probabilities on adverse events. At a theoretical level, our research effort is motivated by Mehra and Sah (2002) who show that even small fluctuations in investors’ attitudes towards risk, which could result from weather-related shifts in their mood states, can have a non-negligible impact on market volatility. Chang et al. (2008), mention two competing, but not mutually exclusive, explanations with contradictory empirical implications for the relationship between weather and volatility. On the one hand, since poorer social moods can be associated with more disagreement in valuation opinions among investors, bad weather can be expected to be inversely related to market volatility (see Harris and Raviv, 1993, Shalen, 1993, Baker and Stein, 2004 and Lucey and Dowling, 2005, among others for a thorough discussion). On the other hand, studies such as Brown, 1999, Gervais and Odean, 2001 and Statman et al., 2006, suggest that when investors are in a good mood, which can be associated with fair weather, then they tend to trade more, which in turn increases volatility. Α third explanation has been given by Kaplanski and Levy (2009), who argue that if SAD induces seasonality in returns, and returns are negatively correlated with volatility, then SAD can indirectly create seasonality in volatility in the opposite direction. We can assume that a similar indirect effect on volatility holds also for other weather and environmental conditions which may affect returns. Finally, another explanation of a positive association between bad weather and volatility could be based on psychological studies which link poor mood with an increase in the subjective probability of undesired outcomes (see Kliger and Levy (2003) and the references therein).
نتیجه گیری انگلیسی
The empirical results in this paper suggest that SAD and cloudiness are negatively associated with various measures of stock market volatility. In line with Dowling and Lucey (2008), we find that historical volatility according to a GJR-GARCH(1, 1) model is significantly inversely related to the mood-proxies associated with cloudiness and variation in nighttime hours for 26 stock exchanges and cities internationally between 1982 and 1997. Despite the fact that we use a latitude-corrected SAD-proxy, we find that the effect of this variable depends on the location of a city on Earth with respect to the equator. Our results concerning implied volatility and realized volatility offer some additional support. Specifically, implied volatility indices for the CBOE and realized S&P 500 index returns tend to be negatively related with cloudiness and variation in nighttime hours. However, the underlying coefficients are statistically significant only in a pooled sample of four implied volatility indices. The direction of association for the SAD-proxy and the VIX implied volatility index is consistent with that reported by Kaplanski and Levy (2009). Our disparity with respect to statistical significance is possibly due to the adoption of a different sample. In general, our analysis suggests that absolute deviations of weather variables from seasonal norms and dummies related to extreme weather conditions do not offer additional explanatory power in attempts to model volatility. Our results are consistent with the explanation that good mood is associated with increased trading and volatility, respectively. As mentioned, it could also be the case that we are simply observing the indirect result of the ‘leverage effect’. Our results are unlikely to be influenced by data-snooping since we use several different but comparable volatility datasets to evaluate our hypothesis and we validate our results, when possible, using subsamples of our original data. It would be useful to evaluate also the economic significance of our results, as in Hirshleifer and Shumway (2003). However, building volatility trading strategies is far from straightforward since it requires combined derivative positions. This note adds to the empirical literature but does not extend our theoretical understanding of the relationship between weather and financial markets. The psychological effects involved in weather are both interesting and complex and deserve further research. A potentially useful direction could consider the heterogeneity in trade responsiveness to weather and environmental-related changes in mood. For example, Levy and Galili (2008) show that males, low income, and young individuals tend to be net buyers on cloudy days. To the extent that these groups have differences in characteristics such as risk aversion, the variations in investor mix could affect intertemporal market returns and volatility. We believe that it would also be interesting to explore rational causes in addition to the behavioral explanations that have been discussed. For example, extending the arguments by Goetzmann and Zhu (2005), if market participants tend to leave early on rainy days in order to beat the rush, then we can expect a negative impact of cloudiness on liquidity and volatility, respectively. Indeed, as Loughran and Schultz (2004) demonstrate, trading volume is significantly lower during blizzards in a city, since investors may take longer to get to work as a result of, for example, the need to shovel snow or dig out cars. This leaves less time for trading. In general, during bad weather it can be expected that commuting times of investors will also be significantly longer. Alternative explanations could be based on the effect of weather on the cognitive behaviors of market participants (see Keller et al., 2005, inter alia). It could be that volatility increases due to weather-related shifts in information consumption by investors. It is well known that social interaction has a significant effect on stock prices (Hong et al., 2004). It could be that during sunny weather investors tend to socialize and communicate more which increases the amount of effective information and volatility.