Original title:
Bayesian vector auto-regression model with Laplace errors applied to financial market data
Authors:
Šindelář, Jan Document type: Papers Conference/Event: Mathematical Methods in Economics 2010, České Budějovice (CZ), 2010-09-08 / 2010-09-10
Year:
2010
Language:
eng Abstract:
The article presents alternative version of Bayesian vector auto-regression model with Laplace distributed innovations. Bayesian estimation in such model is more computationally demanding than estimation in a model with normally distributed innovations, but because of the heavier tails of Laplace distribution, it is more robust. In the article I try to present the way of proceeding with the estimation, obtaining a full posterior distribution of the parameters as a result. At the end an efficient algorithm is sketched, but this part of the research is still unfinished.
Keywords:
auto-regression; parameter estimation; robust Project no.: CEZ:AV0Z10750506 (CEP), 1M0572 (CEP), GA102/08/0567 (CEP) Funding provider: GA MŠk, GA ČR Host item entry: Proceedings of Mathematical Methods in Economics 2010, ISBN 978-80-7394-218-2