I want to test for the joint significance of two parameters (dcca1 and dccb1) estimated from a multivariate DCC GARCH model. does anyone know how to do it using R? any help is greatly appreciated.
Asked
Active
Viewed 274 times
1
-
thank you very much I have upvoted. – guest1 Jul 16 '19 at 03:19
-
If the answer is satisfactory, you may also accept it by clicking on the tick mark to the left. – Richard Hardy Sep 01 '19 at 12:16
1 Answers
2
Given that the process described by the DCC-GARCH model does not exist (McAleer, 2019), there is no point in testing hypotheses about the parameters of such a nonexistant process (such as the hypothesis that both parameters are equal to zero). From Section 4 of the paper:
<...> there is no likelihood function, and hence no derivatives that would enable the derivation of asymptotic properties for the Quasi-Maximum Likelihood Estimates (QMLE) of the estimated parameters. Therefore, any statements regarding the purported “statistical significance” of the estimated parameters are meaningless and lack statistical validity. It follows that any empirical results based on the DCC estimates are fatally flawed and lack statistical validity.
(emphasis is mine).
References
- McAleer, M. (2019). What They Did Not Tell You About Algebraic (Non-) Existence, Mathematical (IR-) Regularity and (Non-) Asymptotic Properties of the Dynamic Conditional Correlation (DCC) Model. Journal of Risk and Financial Management, 12(2), 61.
Richard Hardy
- 54,375
- 10
- 95
- 219
-
It is interesting that this "deficiency" of the DCC-GARCH model is not acknowledge even among many academics. Do you know what academics are doing when applying the DCC-GARCH. Do they try bootstrapping to get a finite sample distribution? Or something else? – Johan Stax Jakobsen Jun 05 '19 at 12:40
-
@JohanStaxJakobsen, No idea. I am also struggling with the notion of existence of a stochastic process and its need, as can be seen in [this](https://stats.stackexchange.com/questions/407725/) and [this](https://stats.stackexchange.com/questions/407295/) question. – Richard Hardy Jun 05 '19 at 12:44