A Note on Diagnosing Multivariate Conditional Heteroscedasticity Models (Q4956037)
From MaRDI portal
!
This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:
scientific article; zbMATH DE number 1452439
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | A Note on Diagnosing Multivariate Conditional Heteroscedasticity Models |
scientific article; zbMATH DE number 1452439 |
Statements
A Note on Diagnosing Multivariate Conditional Heteroscedasticity Models (English)
0 references
24 May 2000
0 references
GARCH model
0 references
diagnostic tests
0 references
Box-Pierce statistics
0 references
Ling-Li statistics
0 references
0.9277917
0 references
0.92651474
0 references
0.9232869
0 references
0.9115727
0 references
A multivariate generalized autoregressive conditional heterocsedasticity model (GARCH) is defined by \(X_t=\mu_t+\varepsilon_t\), \(\varepsilon_t=V_t^{1/2}\eta_t\), where \(X_t\) is the observed \(k\)-dimensional time series, \(\eta_t\) are i.i.d. zero mean vectors with identity variance matrix, \(\mu_t\) and \(V_t\) are, respectively, the conditional mean and variance of \(X_t\) based on the past. E.g., in the BEKK model: \[ V_t=V+\sum_{h=1}^pB_hV_{t-h}B'_h+\sum_{h=1}^p C_h\varepsilon_{t-h}\varepsilon'_{t-h}C'_h, \] where \(V\), \(B_h\), \(C_h\) are some parameter matrices. The authors consider three model diagnostics tests: Ling-Li [see \textit{S. Ling} and \textit{W.K. Li}, J. Time Ser. Anal. 18, No. 5, 447-464 (1997; Zbl 0882.62081)], Box-Pierce and the residual-based \(F\)-test. The tests are compared via simulation studies. The authors' conclusion is that the Ling-Li test has weak power against the alternatives considered, the F-test demonstrates under-rejection, but the Box-Pierce test has good empirical size as well as power.
0 references