Multimodality in GARCH regression models

M. Ooms, J.A. Doornik

    Research output: Contribution to JournalArticleAcademicpeer-review


    It is shown empirically that mixed autoregressive moving average regression models with generalized autoregressive conditional heteroskedasticity (Reg-ARMA-GARCH models) can have multimodality in the likelihood that is caused by a dummy variable in the conditional mean. Maximum likelihood estimates at the local and global modes are investigated and turn out to be qualitatively different, leading to different model-based forecast intervals. In the simpler GARCH(p,q) regression model, we derive analytical conditions for bimodality of the corresponding likelihood. In that case, the likelihood is symmetrical around a local minimum. We propose a solution to avoid this bimodality. © 2008 International Institute of Forecasters.
    Original languageUndefined/Unknown
    Pages (from-to)432-448
    JournalInternational Journal of Forecasting
    Issue number3
    Publication statusPublished - 2008

    Cite this