．單位首頁
．系所簡介
．系主任
．強力師資
．研究所
．課程概況
．各類學程
．期刊論文
．招生資訊
．演講專區
．榮譽榜
．系友天地
．就業訊息
97年5月23日(五) 2:00 ~3:00 P.M.
演講者姓名： 林孟樺 博士候選人
演講者服務單位： 逢甲大學應用統計研究所
Bayesian Analysis of Asymmetric Smooth Transition dynamic range models
Abstract
In this article, Bayesian estimation, volatility forecasting and model comparison for an asymmetric nonlinear smooth transition conditional autoregressive range (CARR) model are derived. An adaptive Markov chain Monte Carlo scheme is developed for these purposes. The logistic transition function is considered and its main properties are investigated and its’ characteristics discussed in detail. We suggest a proper prior distribution, thus ensuring the posterior is proper, for the smoothing parameter via a suitable, weakly informative prior. The proposed models can capture aspects such as sign asymmetry and heteroskedasticity, which are commonly observed in financial markets. The methods are illustrated using simulated data. Finally, deviance information criterion (DIC) is employed to compare the proposed model with its limiting class, the threshold CARR (TARR) models.
Keywords: smooth transition; volatility model; conditional autoregressive range (CARR) model; threshold variable; Bayes inference; MCMC methods.
