On Fisher's information matrix of an ARMA process (Q4357569)
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 1066380
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | On Fisher's information matrix of an ARMA process |
scientific article; zbMATH DE number 1066380 |
Statements
25 September 1997
0 references
stationary ARMA process
0 references
state space form
0 references
white noise process
0 references
autoregressive moving average
0 references
Fisher's information matrix
0 references
Sylvester's resultant matrix
0 references
0.9008501768112184
0 references
0.82999587059021
0 references
0.8160709738731384
0 references
On Fisher's information matrix of an ARMA process (English)
0 references
The Cramér-Rao bound is of considerable importance for evaluating the performance of (stationary) autoregressive moving average (ARMA) models, where the focus is on the error covariance matrix of the estimated parameters. For computing the Cramér-Rao bound and the asymptotic distribution of a Wald test statistic the inverse of Fisher's information matrix is needed. The latter is singular in the presence of common roots of the AR and the MA polynomial and vice versa. This fact is considered to be well-known in time series analysis. We give an elementary proof of this equivalence by linking Fisher's information matrix to Sylvester's resultant matrix and an interpretation in terms of a state space realization. For the entire collection see [Zbl 0870.00040].
0 references
