Generalized Mixed GARCH-Jump Models for Financial Assets Returns

Abstract

In most financial asset returns, the most commonly used parametric specification for the return distribution is the standard normal distribution and discrete jumps in returns for observed data. The advantage of the GARCH specification can lead to a rich model with conditional variances and leptokurtosis for asset returns. We develop a GARJI-GED model which extends the GARJI models in Chan and Maheu (2002) to capture the fundamental characteristics of spot and index futures. We show that the GARJI-GED model specification provides a significant statistical improvement over the assumption of normal return distribution model.