This chapter analyzes several model specifications of the asymmetric relationship between the implied volatility index (VIX) and return series for the CESEE (Central-Eastern and South-Eastern European) stock markets. Several different country-origin VIX indices are examined (US, emerging markets, Russian, and EU) to analyze which one has the best forecasting ability of the return series. Based on daily data analysis and six different model specifications, resulting in 240 models in total, asymmetric and non-linear relationships were found between the selected VIX and return series. As the results differ over all stock market return series, international investors are advised to consider such results when making decisions about their portfolio selection.
TopIntroduction
Prediction of stock return and risk series has become a widespread tool in practice, and a topic of empirical analysis for quite some time now. The reward of bearing additional volatility risk is recognized in previous research (Bollerslev et al., 2009, 2011; Carr and Wu, 2009), indicating that alongside many different factors, volatility presents another important pricing factor in the pricing models. Sentiment and investor fear has become regular variables in the modeling process, ever since the contributions of De Long et al. (1990), Brown and Cliff (2004), Baker and Wurgler (2006), etc. The implied volatility index (VIX) is now commonly used as the investor fear gauge within the research, since initial considerations in Whaley (2000), Simon and Wiggins (2001), Giot (2005), and Whaley (2008). Today, VIX is known to be useful in return and realized volatility forecasting, alongside having diversification possibilities, and it does not include solely noise traders that are often modeled in web searching based indicators. This index includes the professional market participants' perceptions of short-run future realized volatility in stock markets (Hibbert et al., 2008). This means that it can be perceived and used to test whether contrarian strategies on stock markets can be based on the results from models that analyze the return and implied volatility relationship. Moreover, as VIX can be useful in modeling realized volatility or conditional variance (see Bekaert and Hoerova, 2014), it could provide good information about future diversification possibilities regarding individual risk series within the portfolio.
As volatility is a basis for hedging, portfolio selection, risk management in general, and pricing of derivatives (Bollerslev et al. 2018; Yang et al. 2015), alongside risk aversion and sentiment being important factors in the portfolio management process, there exists a lot of research on constructing such indicators (see Courdet and Gex, 2008), and trying to utilize them for forecasting purposes. In general, there are two hypotheses on the negative return-volatility relationship (Hibbert et al., 2008): the leverage and volatility feedback hypothesis. The first one is related to the firm value. If it falls, the value of the equity of that firm has a smaller share in the total value of the firm, which increases the volatility of the equity, as it bears the firm's risk. The second hypothesis states that increases in volatility lead to a decrease in expected stock returns, and consequently to a decrease in current stock prices (i.e. lower returns). Another popular term in the literature is asymmetric volatility (see Schwert, 1989, Engle and Ng, 1993, Bekaert and Wu, 2000), meaning that there exists a negative correlation between the return and variance of the future returns. This is most visible during bear markets. Practice recognizes that VIX-based strategies of investing are useful for hedging and speculation (Nagel, 2012).