Forecasting Volatility in Financial Markets Using a Bivariate Stochastic Volatility Model with Surprising Information

Most asset returns exhibit high volatility and its persistence. Heuristically, this paper focuses on the role of surprising information in high volatility processes and indicates that dismissing surprising information may lead to considerable loss in forecast accuracy. In response, this paper considers the corresponding extension of the modified MDH to surprising information, and proposes a bivariate stochastic volatility model incorporating surprising information in the volatility equations (BSV-SI), which is also designed to capture the dynamics of returns and trading volume. Using the South Korea stock index and trading volume series, it turns out that performance of the onestep- ahead forecasts of the BSV-SI model is apparently superior to those of other competitive models.