格兰杰因果检验中不同检验方法的功效比较

发布时间：2018-01-29 23:23

本文关键词： 格兰杰因果检验 ECM模型数据生成蒙特卡洛模拟　出处：《统计与决策》2017年23期 　论文类型：期刊论文

【摘要】：文章运用蒙特卡洛模拟方法,对F、LR、Wald、Mwald四种格兰杰因果检验统计量的检验效果进行了小样本模拟。研究结果表明:(1)极小样本条件下(n￡50),LR统计量检验效果最佳,MWald统计量检验效果较差。(2)随着样本的增大,Wald、Mwald统计量检验效果将会逐步改善。(3)较大样本条件下(n=200,400),F、LR、Wald、Mwald四个统计量检验效果差异变小;在二维系统中,Wald、Mwald统计量检验效果最佳,LR统计量检验效果最差。(4)F统计量检验效果比较稳定,不随样本的增加出现较大改变。因此,在极小样本情况下,适合采用LR统计量检验序列间的格兰杰因果关系;在较大样本情况下,适合采用Wald、Mwald统计量检验序列间的格兰杰因果关系;而F统计量既适用极小样本的检验,也适用较大样本的检验。
[Abstract]:In this paper, Monte Carlo simulation method is used to study the effect of LRX Wald. The test effect of four Granger causality test statistics of Mwald is simulated with small sample. The results show that the test effect is the best under the condition of very small sample (n50 / L). The effect of MWald statistic test is poor. 2) with the increase of sample size, the effect of MWald statistic test will be improved gradually. 3) under the condition of large sample, the effect of Mwald statistic test will be improved gradually. The difference of the four statistical test results of FRRLRLRDwald Mwald was smaller. In the two-dimensional system, the effect of Waldwald statistic test is the best and the effect of LLR statistic test is the worst. Therefore, in the case of very small sample, the Granger causality between the sequences is suitable for LR statistic test. In the case of large samples, Granger causality is suitable for Waldberg Mwald statistic test. The F statistic is suitable for both the test of very small sample and the test of large sample.
【作者单位】：西南财经大学中国西部经济研究中心;
【基金】：中央高校基本科研业务费专项资金青年教师成长资助项目(JBK160169)
【分类号】：F224.0
【正文快照】： 0引言Granger(1969)[1]以信息集为基础给出了因果关系的定义:如果序列At中含有预测序列Bt+j的信息(A有助于解释B未来的变化),则A是B的格兰杰原因。根据这一定义,学者们提出了多种检验格兰杰因果关系的思路。Sims(1972)[2]研究发现,如果A不是B的格兰杰原因,那么A关于B未来值的

【相似文献】