Answers are grounded in this paper and cite their sources.
To keep the conditional variances generated by the GARCH (p, q) model nonnegative, Bollerslev imposed nonnegativity constraints on the parameters of the process. We show that these constraints can be substantially weakened and so should not be imposed in estimation. We also provide empirical examples illustrating the importance of relaxing these constraints.
No takes yet. Share an insight, caveat, or question.
Nelson et al. (1992) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: