摘要
现有门限协整检验方法由于模型似然函数具有多峰、不连续特征,导致冗余参数识别存在困难,最优化计算相对复杂。本文提出基于非线性误差修正模型的贝叶斯门限协整分析,结合参数的后验条件分布设计MCMC抽样方案,进行贝叶斯门限协整检验;并利用Monte Carlo仿真研究了贝叶斯门限协整检验的有限样本性质,发现贝叶斯门限协整检验方法具有良好的有限样本性质。同时,利用不同期限的美国利率序列进行了实证研究,结果发现1个月与3个月利率之间、3个月与6个月利率之间以及3个月与1年利率之间均存在门限协整关系。研究结果表明:贝叶斯门限协整检验方法解决了冗余参数识别的难题,使计算变得相对简单,并提高了估计的精确度和检验的准确性。
In the existing threshold cointegration methods, the jagged and potentially muhimodal nature of the likelihood function of threshold model complicates optimization and also makes the identification of unknown nuisance parameters more difficult. This paper proposes nonlinear threshold ECM and conducts Bayesian inference. Based on the posterior conditional distributions of the parameters, MCMC samplers are designed. The finite sample property is studied via a combination of Monte Carlo simulation. Finally, through the empirical application of the US interest rates with various maturities, we find there are threshold cointegration relationships between interest rates with maturities of 1 month and 3 month, 3 month and 6 month, 3 month and 1 year, respectively. Therefore, the usefulness of this Bayesian method is demonstrated. It shows that Bayesian threshold eointegration solves the problem in complicated optimization and improves the precision of cointegration test.
出处
《统计研究》
CSSCI
北大核心
2013年第1期96-104,共9页
Statistical Research
基金
国家自然科学基金项目(NSFC71031004,71171075,70771038)
教育部长江学者和创新团队发展计划项目(IRT0916)
教育部留学回国人员科研启动基金项目(教外司留[2010]609)
湖南省研究生创新项目(CX2011B134)资助
