Title: Forecasting emerging equity market volatility using nonlinear GARCH models
Abstract: In this paper we examine the usefulness of nonlinear Generalized Autoregressive Conditionally Heteroskedastic (GARCH) models for forecasting daily volatility in a number of Asian and Latin American emerging equity markets. Two of the most popular nonlinear GARCH specifications, the GJR model and the Exponential GARCH model, are found to outperform a linear GARCH model in terms of one-day ahead out-of-sample volatility forecasts. This conclusion holds both when volatility forecasts are evaluated by means of traditional criteria that rely upon a proxy for unobserved volatility or by means of indirect probability forecasts.