作 者: ;
机构地区: 广州大学
出 处: 《预测》 2010年第6期 71-75,共5页
摘 要: 本文应用FM-OLS来估计阈值协整参数,并利用Monte-Carlo模拟详细研究了其小样本性质,结果表明FM-OLS法能修正OLS估计的小样本偏差,且数据过程的持久性(均值回复速度)、随机干扰项与解释变量的相关程度以及样本容量是影响FM-OLS小样本性质的主要因素。模拟结果还表明不论是阈值协整还是线性协整,FM-OLS都比OLS估计具有明显优势,因此在宏观经济协整分析中,利用FM-OLS法能获得较准确的参数估计,同时还可以利用标准分布对协整参数进行W ald检验。 The paper examines small sample properties of parameters estimation with fully modified OLS(FM-OLS) method in threshold cointegration through performing a set of Monte-Carlo simulations,and the resulting evidence shows that FM-OLS method can modify small sample bias resulted from OLS,and some factors exert an influence on properties of FM-OLS method including persistence of the data process,the correlation between stochastic error and explanatory variables as well as sample size.As addition,regardless of threshold cointegration or linear cointegration,FM-OLS gains an obvious advantage over OLS.Therefore,when performed in the cointegration analysis,FM-OLS can obtain an exact estimator about cointegration parameter,and the wald statistics constructed through FM-OLS method have standard limiting distributions.
关 键 词: 阈值协整 完全修正的OLS 偏差和标准差 MC模拟
分 类 号: [F224.0]
领 域: []