GARCH模型的貝葉斯推斷
頁(yè)數(shù) 6 字?jǐn)?shù)2566
摘要：本文闡述了如何用Griddy-Gibbs抽樣對(duì)GARCH模型進(jìn)行貝葉斯推斷。Gibbs抽樣方法是基于完全條件后驗(yàn)密度的，但在GARCH模型中，參數(shù)的密度比較復(fù)雜，從而無法用共軛法來構(gòu)造它的后驗(yàn)密度，對(duì)其直接抽樣不可行。因此，本文把Gibbs抽樣和數(shù)值積分法相結(jié)合,來從后驗(yàn)分布中進(jìn)行抽樣，并用CUMSUM統(tǒng)計(jì)量來判斷抽樣的收斂性。利用模擬產(chǎn)生的數(shù)據(jù)，本文分別用Griddy-Gibbs抽樣和MLE兩種方法來估計(jì)模型，并比較其結(jié)果，得出結(jié)論：這個(gè)Griddy-Gibbs抽樣是可行的，且結(jié)果與MLE互有優(yōu)劣
參考文獻(xiàn)：
[1] Luc Bauwens and Michel Lubrano (1998), Bayesian inference on GARCH models using the Gibbs sampler, Econometrics Journal, volume 1, pp. c23-c46
[2] Bauwens, L. and P.Giot (1998), A Gibbs sampling approach to cointegration, Comput. Stat. Forthcoming.
[3] Bollerslev, T., R.Y.Chou and K. F. Kroner (1992), ARCH modeling in finance. J. Econometrics 52, 5-59
[4] Engle, R. F. and T. Bollerslev (1986). Modelling the persistence of conditional variances. Econometric Rev. 5, 1-50