On accelerating the EM-based algorithms for the VAR(1) models with multivariate generalized scaled t-distributed innovations

A vector autoregressive model of order one with multivariate generalized scaled t-distributed innovations is considered here. The object is to estimate the parameters of the proposed model by using the well-known maximum likelihood estimation method. The maximum likelihood estimation method is perfo...

Full description

Saved in:
Bibliographic Details
Published inCommunications in statistics. Theory and methods Vol. 52; no. 13; pp. 4414 - 4428
Main Authors Mirniam, A. S., Nematollahi, A. R.
Format Journal Article
LanguageEnglish
Published Philadelphia Taylor & Francis 03.07.2023
Taylor & Francis Ltd
Subjects
Algorithms
Autoregressive models
BHHH algorithm
ECM algorithm
ECME algorithms
Innovations
Maximum likelihood estimation
Multivariate analysis
multivariate generalized scaled t distribution
VAR models
Online AccessGet full text
ISSN0361-0926
1532-415X
DOI10.1080/03610926.2021.1994608

Cover

More Information
Summary:A vector autoregressive model of order one with multivariate generalized scaled t-distributed innovations is considered here. The object is to estimate the parameters of the proposed model by using the well-known maximum likelihood estimation method. The maximum likelihood estimation method is performed by using the expectation-conditional maximization algorithms (ECM and ECME) to accelerate the basic EM algorithm. The proposed methods are also compared to a quasi-Newton well-known method, named BHHH in terms of the rate of convergence as well as the precision and validity of the obtained estimates, by implementing some numerical simulations. At last, the outcomes are used to fit the model and predict for a real data set.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:0361-0926
1532-415X
DOI:10.1080/03610926.2021.1994608