THE RELEVANCE OF CBOE VOLATILITY INDEX TO STOCK MARKETS IN EMERGING ECONOMIES

We examine the capability of CBOE S&P500 Volatility index (VIX) to determine returns of emerging stock market indices as compared to local stock markets volatility indicators. Our study considers CBOE S&P500 VIX, local BRIC stock market volatility indices and BRIC stock market MSCI indices daily returns in the period !om January 1, 2009 to September 30, 2014. Research is conducted in two steps. First, we perform Spearman correlation analysis between daily changes in CBOE S&P500 VIX, local BRIC stock market VIX and MSCI BRIC stock market indices returns. Second, we perform multiple regression analysis with ARCH e"ects to estimate the relevance of CBOE S&P500 VIX and local VIX in determining BRIC stock market returns. Research reports weak correlation between CBOE S&P500 VIX and local VIX (except for Brazil). Furthermore, results challenge the assumption of CBOE S&P500 VIX being an indicator of global risk aversion. We conclude that commonly documented trends of rising globalization and stock markets co-integration are not yet present in emerging economies, therefore the usage of CBOE S&P500 VIX alone in determining BRIC stock market returns should be considered cautiously, and local volatility indices should be accounted for in analysis. Furthermore, the data con#rms the presence of safe haven properties in Chinese stock market index.


Introduction
Apparently, increase in globalization, countries liberalization and openness has rendered a strong interdependence between nancial market dynamics across the globe.Additionally, recently undersigned treaties like General Agreement on Tari s and Trade agreements (GA ), European Community (EC), North American Free Trade Agreement (NAFTA), and Association of Southeast Asian Nations (ASAN) signi cantly supported worldwide economic integration (Cavaglia et al., 2000).
Recent nancial crisis and its a ermath of volatility spillover e ects evoked growing interest in market sentiment indicators measured by stock market volatility indices (VIX).As outlined by Gemmil et al. (1997), shocks in stock markets implied VIX across nancial markets are interrelated and may be employed as indicators of the rise in volatility in other markets.As summarized by Hu (2006), "markets are more likely to crash together than to boom together".(Hu, 2006, p. 729).e la er summary of market turmoil contagion e ect encourages investors to be more sensitive to market distortions and highly look up for market sentiment parameters and their transmission when considering portfolio diversi cation alternatives (Shiller, 2013).e extent to which stock price indices in developed and emerging countries are a ected by volatility indicators "is important to the individual investor, the policy maker and forecaster, the researcher and more recently the investment banks that are specializing in new nancial innovations to minimize risk" (Natarajan et al., 2013, p. 56).
e context of globalization has fostered surging discussions about the choice of market sentiment indicators while constructing investment strategies.As outlined by Whaley (2009), CBOE S&P 500 VIX (therea er VIX S&P ) index has proven its applicability in regime-switching, threshold and transition models due to its property of forward-looking indication of market risk perception.
A stream of academic research has proven that implied volatility outperforms realized volatility in terms of forecasting power (for more details, see Christensen & Prabhala (1998), Szakmary et al. (2003), Corrado & Miller (2005) and Carr & Wu (2006).e plausibility of VIX S&P to serve as a proxy for market turmoil, risk aversion and benchmark of future volatility measure (Fassas & Siriopoulos, 2012) can be summarized by Carr & Wu (2009), who outlined two major components combined within a single index, speci cally, the quantity and the price of perceived risk.
Some scholars like Fassas & Siriopoulos (2012) refer VIX S&P to a global volatility indicator and argue that it alone may re ect market shocks worldwide in the context of recent markets co-integration trends, which diminishes signi cance of local stock market volatility indices (therea er VIX local ).e rationale behind such statements stems from arguments similar to those expressed by Nelson & Mossavar-Rahmani (2014), who claims that volatility of S&P 500 index is highly relative to the US economic cycle, capable to re ect the global cycle and is in line with volatility observed during global recession periods.
Nonetheless, evident asymmetries in volatility transmission foster contradicting stream of research which aims at challenging the "one size ts all" approach to volatility indicators.Aggarwal et al. (1999) report shi s in emerging stock markets volatility being explained solely by local market shocks (Mexican Peso crisis, Latin America hyperin ation, and Marcos-Aquino con ict in the Philippines) in the period of 1985-1995. Furthermore, Bailey & Chung (1995) report that sudden changes in volatility of emerging stock markets is highly related to contemporaneous local political events.
Recent studies by Chulia et al. (2009) estimate volatility transmission from the US to Euro zone (EZ) stock markets a er the September 11 a ack, but not from EZ to the US stock market a er terrorist a acks in Madrid and London on March 11 and July 7, 2009 respectively.As con rmed by studies of Bekaert & Harvey (1997) and Susmel & ompson (1998), the explanatory power of global events is weak when considering emerging markets volatility shocks.
e importance of emerging markets is signi cantly increasing as they have become integrated part of the global equity portfolio allocation with market capitalization of emerging countries hiking from only 1% in 19881% in to 11% in 20141% in (MSCI, 2014)).Currently, BRIC countries account for 41% of the world's population, hold USD 4.4 trillion of foreign reserves and create one-h of global domestic products at relatively low costs (MSCI, 2014).Establishment of joint BRIC and South Africa Bank in the mid-2014 aimed at providing money for infrastructure and development projects, higher BRIC co-integration a er the recent Crimean crisis and economic sanctions targeted at Russian Federation, and the established BRICS exchanges alliance in order to "expose international investors to their dynamic economies" (BRICS Exchanges Alliance, 2014) with the future perspectives to incline towards the development of Energy Association of BRICS altogether signify growing importance of the block.
In this paper we aim at estimating the capability of VIX S&P in determining emerging stock market returns as compared to that of local volatility indices.Our results document the empirical signi cance of BRIC stock market local volatility indices in determining stock market returns.We argue that even in the context of recent nancial markets globalization, openness and subsequent co-integration, stock markets in emerging economies do not fully re ect and absorb the e ects of global market turmoil measured by VIX S&P , but rather are sensitive to local events.Furthermore, we argue that the level of relevance of volatility indices in determining BRIC stock market returns is idiosyncratic and should be further individually explored.erefore, investors should not solely rely on VIX S&P , but also account for VIX local dynamics when considering emerging stock markets in portfolio allocation strategies.
e paper continues as follows.Section 2 discusses data used in the research.Section 3 elaborates on methodology.Section 4 presents results, and the last Section concludes.

Data
In this study, we use daily log-changes in VIX S&P and VIX local of Brazil, Russia, India and China stock markets, and daily log-returns of Brazil, Russia, India and China MSCI stock market indices 01.01.2009 to 30.09.2014.Our sample consists of 1412 daily observations a er excluding weekends and bank holidays.We selected this period as it is not contaminated by recent global nancial crisis dynamics, and is representative in terms of market shocks (summarized in Table 1 below) and resurgence periods amid.In our study, we set VIX S&P as a proxy for global market uncertainty and risk aversion.Our choice of implied volatility measure is supported by its capability to serve as a forward-looking indicator as opposed to alternative historical volatility (HV) measure.
e main aw of HV is its consideration of close-to-close prices of the stock without capturing the magnitude of intraday price movements, while VIX S&P measures implied volatility of 30-day period options on the S&P 500 index.In other words, it provides estimates of expected future realized volatility for 30 calendar days ahead suggesting that higher VIX indicates higher implied volatility in the S&P 500 index (Whaley, 2000).Figure 1 exhibits dynamics of VIX S&P and S&P 500 index.Market shocks, illustrated by plummeting stock market index returns, are captured by corresponding spikes in VIX S&P index.Furthermore, diverging trends a er 2012 signify diminishing market "fear gauge" and increasing economic resurgence of S&P 500 companies.
We set MSCI local (VIX local ) implied volatility indices of BRIC countries and corresponding stock market indices as summarized in Table 2     Growing global interest in emerging markets is exhibited in Table 3, which illustrates the structure of investment portfolio held by overseas investors in corresponding BRIC markets.In Brazil, India and China, most part of investment portfolio is comprised of equities, while in Russia, 93% of foreign investment is concentrated in debt securities.
e table highlights percentage of US investment in the corresponding country economy, which implies that VIX S&P may be more relevant in determining Brazilian stock market returns (42% of US share in total Brazil portfolio investment assets), as contrasted to Russia, where US investments account for less than 5%.Presence of relatively strong correlation between VIX S&P and VIX local would imply that VIX S&P alone might be relevant for representing BRIC stock markets risk aversion, given strong correlation between volatility and corresponding stock market indices.Similarly, correlation analysis between VIX S&P and BRIC stock market returns is performed.
Multiple regression analysis with ARCH e ects is performed by R i,t -i stock market log return at time t; c i,0 -constant; c i,n -coe cients of volatility indicators; e t -error term, V m,i -interaction term which is strictly positive and de ned by: Dummy variables are de ned by: We include dummy variables in regression analysis in order to capture the e ect of VIX S&P and VIX local on intercept (terms c i , 4 D1and c i , 5 D2, respectively) and on slope (terms c i , 6 (D1*VIX S&P ) t-1 and c i , 7 (D2*VIX i , local ) t-1 ).Interaction term V m,i is included in regression analysis in order to test for possible indirect (moderation) e ect of VIX S&P on relation between VIX i, local and the corresponding stock market indices, R i,t .

Summary Statistics and Statistical Testing
Summary statistics is outlined in Table 4. e lowest daily return of 0.01% was generated by Brazilian stock market index (EWZSO) which was also the least resilient.e returns of Russian stock market (RTSI) were the most volatile with standard deviation of 1.97% a ributing it to the least a ractive of all BRIC for risk averse investors.Additionally, Russian stock market returns were negatively biased due to negative skewness in the period of consideration, as opposed to the rest of indices.Furthermore, Brazil and China generate the highest excess kurtosis (16.9 and 25.9 respectively), which may indicate possible overestimation of mean returns probability.As illustrated in Appendix 1, Jarque-Bera test results violate normality assumption for all data time series, which implies the relevance of Spearman test for correlation analysis.ADF test rejected the null hypothesis of unit root, therefore no further data transformation is needed.Due to the presence of autocorrelation in error terms (Appendix 2), regression with ARCH e ects is estimated.

Correlation Analysis Results
Spearman correlation analysis results summarized in Table 5 report weak correlation estimates between local volatility indices of BRIC stock markets, which implies weak (or absent) volatility transmission e ects between emerging economies under study, despite their recent growing interdependence discussed in Section 1.Similarly, correlation estimates between VIX S&P and Russian, Indian and Chinese stock markets appear to be low, entailing the absence of contagion e ect in risk aversion stemming from US stock market.However, relatively high correlation estimate between VIX S&P and Brazilian VIX local (0.71) is anticipated as it is in line with relatively high share of US investments (42%) in Brazil economy, as depicted in Table 3 and is suggested by geographic proximity.Spearman correlation estimates between volatility indices and BRIC stock market returns are summarised in Table 6.As anticipated, all correlation estimates are negative implying diverging dynamics between stock market daily returns and corresponding changes in volatility indices.VIX S&P correlation with BRIC stock market returns is reported to be very weak.Slightly higher, but still weak correlation is observed between VIX S&P and Russian stock market index.Furthermore, local volatility indices report weak/moderate correlation estimates with the corresponding local stock market indices.

Regression Analysis Results
Regression analysis with ARCH e ects is performed following methodology discussed in Section 2. Regression coe cient estimates are summarized in Table 7, while explicit regression results are presented in Appendix 3. Regression results illustrate evident idiosyncrasies in relationship between VIX S&P , VIX local and the corresponding BRIC stock market returns dynamics.In the case of Brazil, insigni cant coe cient estimates suggest that neither VIX S&P nor VIX local (VXEWZ) explain EWZSO returns.However, it is an anticipated result due to relatively low Spearman correlation between EWZSO and VIX S&P (Table 5) and between EWZSO and VXEWZ (Table 6): -0.01 and -0.06, respectively.Regression coe cients of Russian stock market index returns are insigni cant except for D1 dummy variable coe cient c 4 , which implies that daily returns of RTSE are negatively a ected by (and only in the case of) increase in VIX S&P , and are not explained by regression variables otherwise.Indian stock market returns are explained by both VIX S&P and VIX local .Furthermore, signi cant interaction term c 3 suggests the presence of moderation e ect of VIX S&P on the relationship between NIFTY and local volatility indicator, INVIXN.Regression analysis of Chinese stock market HIS index reports unanticipated results.VIX S&P , VIX local and interaction term coe cients are signi cant at 0.05 signi cance level and are of positive sign (except for interaction term).Positive c 1 and c 2 coe cients illustrate converging dynamics between Chinese stock market index and indicators of global and local risk aversion.In other words, in periods of rising VIX S&P and VIX local , HIS index exhibits safe haven properties by rising in value.
ARCH e ects, i.e. residual serial correlation, are present in all time series and are controlled in our study.ARCH e ects suggest volatility clustering together with possibility of omi ed variables in estimated regressions.e la er, together with relatively poor explanatory power of VIX S&P and VIX local in Brazilian and Russian stock market cases indicate the need of further study of the given phenomenon together with consideration of additional factors.

Conclusion
In this paper we challenge the commonly accepted notion of VIX S&P as a global "fear gauge" indicator and argue that stock markets of emerging economies are idiosyncratic in their pa erns of risk perception.We claim that BRIC stock markets are still too slow to absorb global risk aversion and therefore contagion e ect caused by nancial markets globalization, co-integration and speed of information transmission does not (yet) have signi cant e ect on BRIC stock market returns.Due to this fact, allocating investments in BRIC stock markets can be a viable diversi cation tool because of BRIC peculiar response to shocks in global and local volatility indicators.Furthermore, our study contradicts prevailing a itude to emerging markets as volatile and solely risky.Data indicates the presence of safe haven properties in Chinese stock market index supported by statistically signi cant regression coe cients.
As implied by our research results, VIX S&P does not explain changes in Brazil and Russian stock market returns and is relevant in determining Indian and Chinese stock market returns only when considered together with VIX local .erefore, considering VIX S&P as an indicator of global stock market risk aversion should be treated cautiously when evaluating investment opportunities in emerging stock markets.Furthermore, relatively poor explanatory power of volatility indices in Indian and Russian cases, and no explanatory power in the Brazil case call for further research in the eld.
Our empirical study carries certain limitations which should be kept in mind when assessing results.First, we have considered only four stock market indices in our analysis, therefore implications regarding emerging economies risk aversion should be solely considered within the context of BRIC stock markets.Second, we assume that each stock market's returns are explained only by two factors, VIX S&P and VIX local .We ruled out macroeconomic, social and political factors as well as other market sentiment indicators, such as ZEW economic indicator, Consumer Con dence Index (CCI), and others which may have impact on stock market returns, albeit these are not the focus of this particular study.ird, we employed parsimonious multiple regression model with ARCH e ects.Alternatively, more complex threshold and regime-switching models should be considered to capture shi s in stock market returns dynamics.Finally, we looked at a relatively short time period covering past 5 years and used daily return data as we aimed at studying most recent trends in volatility transmission to emerging markets.Considering longer time span and less frequent data might report less noisy results.
To conclude, our results imply that investors should revisit their a itude towards emerging economies and consider them as a potential source of a di erent type of risk to be added to their investment portfolio.

Figure 2
Figure 2 exhibits dynamics of implied volatility indices of VIX S&P , VIX local of corresponding BRIC and developed (Australia, Germany, Japan and United Kingdom) countries, which we include for graphical comparison.Apparent though lagging comovement between market uncertainty indices in emerging and developed economies illustrates interdependence and transmission of volatility.

TABLE 1 .
Summary of the major market shocks during 2009-2014 Date Event May, 2010 "Flash Crash" led by Greek sovereign debt concerns 21 July 2010 Dodd Frank Wall Street Reform and Consumer Protection Act (DFA) January 2011 Civil uprising in Syria and Lybia March 2011 e earthquake and tsunami in Japan 6 August 2011 USA credit rating downgrade from AAA to AA+ (by S&P) August 2011 Possibility of transmission of European sovereign debt crisis to Italy and Spain (I) January 2012 France, Austria, Spain, Italy and Portugal credit ratings downgrade (S&P) Mid of 2012 Possibility of transmission of European sovereign debt crisis to Italy and Spain (II) February 2014 Ukrainian Revolution March 2014 Bazil credit rating downgrade, Crimean Crisis and the rst sanctions to Russian Federation Note.Prepared by the authors, 2015

TABLE 2 .
MSCI Emerging Stock Market Indices and VIX local

TABLE 3 .
Investment portfolio composition held by overseas investors in BRIC, 2013 Note.International Monetary Fund, 2013.Retrieved from: h p://cpis.imf.org/3.MethodologyFirst, we obtain and discuss summary statistics.A er that, Jarque-Bera test for data normality, Augmented Dickey-Fuller (ADF) test for stationarity and tests for ARCH e ects are implemented prior to analysis.Subsequently, correlation analysis between VIX S&P and VIX local of the corresponding BRIC stock markets is performed according to the classi cation of relative correlation strength provided below:

TABLE 4 .
Summary statistics of BRIC stock market indices

TABLE 5 .
Spearman Correlation Estimates between VIXS&P and VIXlocal of BRIC

TABLE 6 .
Spearman Correlation Estimates between VIXS&P, VIXlocal and BRIC Stock Market Indices