Forecasting Volatility of RTS Stock Index Using GARCH Models

Stock markets are characterized by uncertainty associated with economic and political situation, as well as the behavior of market participants. The negative impact of the COVID-19 pandemic on business activity in the world and oil crisis in 2020 only increase the amplitude of fluctuations. The RTS composite stock index is an indicator of the stock market condition in the Russian Federation. Also, it is used as a separate financial instrument for trading. Thereby, the ability to predict the volatility of the stock index over time allows not only to determine the securities market conditions, but also to benefit in the correct prediction of trends. That is why this research is useful for both government bodies and private investors. The conditional heteroskedasticity models that create the GARCH fanily allow to model volatility through conditional variance. During the study symmetric (ARCH, GARCH) and asymmetric (EGARCH, TARCH, APARCH) models are used on the time series of RTS index returns from January 2012 to February 2020. The results of in-sample estimation and out-of-sample prediction demonstrated the same result: the EGARCH(1,1) model with the specification for the mean equation ARIMA(1,0,0) has the greatest predictive power. This research allowed identifying econometric features of the volatility of RTS stock index: clustering, leverage effect and significant GARCH effect, indicating the presence of a positive risk premium. For the stock market these properties mean the inability to eliminate the effects of shocks in the short-run. Volatility increases in response to negative news more than in response to positive shocks. To increase the market’s resistance to fluctuations, recommendations are proposed. Comparison of the obtained results with the existing studies on forecasting other indices revealed a partial similarity between the RTS index and the Shanghai Stock Exchange (SSE) index. Key words: stock market, RTS index, volatility, GARCH models, modeling, forecasting, econometric features.