论文部分内容阅读
在无漂移平稳过程回归中,最近的研究表明传统t检验和经Newey-West修正的t检验具有较高的拒绝率,因此,标准误的不适当估计是产生伪回归的主要原因.本文利用修正的HAC法来估计标准误,且基于该HAC估计所构造的参数t^*检验统计量具有不依赖于冗余参数的极限分布.我们通过一系列的蒙特卡罗模拟研究发现,新的t^*统计量大大减少了拒绝率;同时发现随着数据过程的持久性增强,传统t检验的拒绝率增加的速度很快,而新的t^*检验的拒绝率增加速度很慢,即新统计量对数据过程持久性不敏感,这大大增加了计量经济分析的可靠性.“,”The latest research shows that a spurious regression occurs in the regression between stationary processes without drifts. Therefore, the standard errors estimated inappropriately have an important impact on spurious regression among stationary variables. In the paper, we use a modified HAC method to estimate standard errors, and our test statistic constructed based on our HAC method has a limiting distribution that does not depend on nuisance parameters. A set of MC simulations show that the traditional t statistic and t statistic modified by Newey-West method have a larger rejection rate of null hypothesis, and our new statistic can greatly reduce rejection rate. At the same time, simulation results show that the rejection rate of traditional tests increased very rapidly with a strong persistence, and the rejection rate of our statistic increased very slowly, that is to say, our statistics is not sensitive to the persistence of the data process, which helps to increase the reliability of econometric analysis greatly.