Multimodality in the GARCH Regression Model
Jurgen Doornik and
Marius Ooms
No 2003-W20, Economics Papers from Economics Group, Nuffield College, University of Oxford
Abstract:
Several aspects of GARCH(p,q) models that are relevant for empirical applications are investigated. In particular, it is noted that the inclusion of dummy variables as regressors can lead to multimodality in the GARCH likelihood. This invalidates standard inference on the estimated coefficients. Next, the implementation of different restrictions on the GARCH parameter space is considered. A refinement to the Nelson and Cao (1992) conditions for a GARCH(2,q) model is presented, and it is shown how these can then be implemented by parameter transformations. It is argued that these conditions may also be too restrictive, and a simpler alternative is introduced which is formulated in terms of the unconditional variance. Finally, examples show that multimodality is a real concern for models of the £/$ exchange rate, especially when p>2.
Keywords: Dummy variable; EGARCH; GARCH; Multimodality. (search for similar items in EconPapers)
JEL-codes: C22 C51 (search for similar items in EconPapers)
Pages: 30 pages
Date: 2003-09-15
New Economics Papers: this item is included in nep-ecm, nep-ets, nep-fin and nep-ifn
References: View references in EconPapers View complete reference list from CitEc
Citations: View citations in EconPapers (17)
Downloads: (external link)
http://www.nuff.ox.ac.uk/economics/papers/2003/W20/GarchMode1.pdf (application/pdf)
Related works:
Journal Article: Multimodality in GARCH regression models (2008) 
This item may be available elsewhere in EconPapers: Search for items with the same title.
Export reference: BibTeX
RIS (EndNote, ProCite, RefMan)
HTML/Text
Persistent link: https://EconPapers.repec.org/RePEc:nuf:econwp:0320
Access Statistics for this paper
More papers in Economics Papers from Economics Group, Nuffield College, University of Oxford Contact information at EDIRC.
Bibliographic data for series maintained by Maxine Collett ().