Refining value-at-risk estimates using a Bayesian markov-switching gjr-garch copula-evt model

Marius Galabe Sampid, Haslifah M. Hasim, Hongsheng Dai

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

In this paper, we propose a model for forecasting Value-at-Risk (VaR) using a Bayesian Markov-switching GJR-GARCH(1,1) model with skewed Student’s-t innovation, copula functions and extreme value theory. A Bayesian Markov-switching GJR-GARCH(1,1) model that identifies non-constant volatility over time and allows the GARCH parameters to vary over time following a Markov process, is combined with copula functions and EVT to formulate the Bayesian Markov-switching GJR-GARCH(1,1) copula-EVT VaR model, which is then used to forecast the level of risk on financial asset returns. We further propose a new method for threshold selection in EVT analysis, which we term the hybrid method. Empirical and back-testing results show that the proposed VaR models capture VaR reasonably well in periods of calm and in periods of crisis.

Original languageEnglish
Article numbere0198753
JournalPLoS ONE
Volume13
Issue number6
DOIs
Publication statusPublished - 22 Jun 2018

ASJC Scopus subject areas

  • Biochemistry, Genetics and Molecular Biology(all)
  • Agricultural and Biological Sciences(all)
  • General

Fingerprint Dive into the research topics of 'Refining value-at-risk estimates using a Bayesian markov-switching gjr-garch copula-evt model'. Together they form a unique fingerprint.

  • Cite this