中文摘要：

讨论了四种不同MCMC抽样方案在SV模型贝叶斯估计中的适应性和稳健性问题。蒙特卡洛模拟结果显示,随机误差项的近似处理方式和波动变量抽样结构直接影响SV模型的贝叶斯估计效率。具体来说,波动变量的＂成块＂抽样比＂逐分量＂抽样的效率更高;随机误差项有限混合近似比正态近似的估计精度更优。四种抽样方案中,正态近似和FFBS算法的收敛速度和运算时间最快,有限混合近似和FFBS算法的估计精度最优。

英文摘要：

In this paper,comparison between the most common MCMC schemes is performed in terms of different efficiency criteria,including processing time,inefficiency ration and estimator accuracy.The results support that rand walk Metropolis sampling time is shortest,normal approximation and FFBS method shows a fastest decay in the autocorrelation,mixtures of normal and FFBS sampling is the most precise.