Efficient MCMC estimation of some elliptical copula regression models through scale mixtures of normals. (24th September 2018)

• Record Type: Journal Article
• Authors: Wichitaksorn, Nuttanan; Gerlach, Richard; Choy, S.T. Boris
• Other Names: Pievatolo Antonio guestEditor.; Insua David Rios guestEditor.; Wilson Simon guestEditor.
• Is Part Of: Applied stochastic models in business and industry. Volume 35:Number 3(2019)
• Page Start: 808
• Page End: 822
• Publication Date: 2018-09-24
• Subjects: elliptical copulas -- Markov chain Monte Carlo -- multivariate statistics -- scale mixtures of normal -- seemingly unrelated regression; Stochastic analysis -- Periodicals; Stochastic processes -- Periodicals; Business mathematics -- Periodicals; Finance -- Mathematical models -- Periodicals; Industrial management -- Mathematical models -- Periodicals; 338.00151923
• DOI: 10.1002/asmb.2410
• Languages: English
• ISSNs: 1524-1904

Abstract

Abstract: This paper proposes an efficient estimation method for some elliptical copula regression models by expressing both copula density and marginal density functions as scale mixtures of normals (SMN). Implementing these models using the SMN is novel and allows efficient estimation via Bayesian methods. An innovative algorithm for the case of complex semicontinuous margins is also presented. We utilize the facts that copulas are invariant to the location and scale of the margins; all elliptical distributions have the same correlation structure; and some densities can be represented by the SMN. Two simulation studies, one on continuous margins and the other on semicontinuous margins, highlight the favorable performance of the proposed methods. Two empirical studies, one on the US excess returns and one on the Thai wage earnings, further illustrate the applicability of the proposals.