Forecasting Financial Volatility: Evidence from Chinese Stock Market

Volatility models and their forecasts are of interest to many types of economic agents, especially for financial risk management. Since 1982 when Engle proposed the Autoregressive Conditionally Heteroscedastic (ARCH) model, there have emerged numerous models for forecasting volatility. Given the vast number of models available, agents must decide which one to use. This paper explores a number of linear and GARCH-type models for predicting the daily volatility of two equity indices in the Chinese stock market. Under the framework of three distributional assumptions, the forecasts are evaluated using traditional metrics and by how they perform in a modern risk management setting-Value at Risk. We find that the relative accuracies of various methods are sensitive to the measure used to evaluate them. However, the worst performing method for forecasting the one-day-ahead volatility in the Shanghai and Shenzhen index is the random walk model.