A stochastic unit root test under the existence of correlation between disturbing term and errors

• Main Authors: Mao-Ning Li, 李茂寧
• Other Authors: Lee, Chingnun
• Format: Others
• Language: zh-TW
• Published: 2019
• Online Access: http://ndltd.ncl.edu.tw/handle/62bq4s

碩士 === 國立中山大學 === 經濟學研究所 === 107 === In this paper, we attempted to find out a new stochastic unit root test under the existence of correlation between disturbing term and errors in the model. From the past literatures, we observe that these test statistics of STUR lack to consider the correlation between disturbing term and errors in the regression due to the null hypothesis. Therefore, in this paper, we use the regression which contains the correlation between disturbing term and errors. After finding out the test statistic, we use the Monte Carlo method to support our theorem. And, we notice that the power of statistic only affected by the variance of disturbing term. Compare to the other stochastic unit root test, the performance of power doesn''t as good as expected. The reason of result may owing to single variable test which is the variance of disturbing term that we focus on. We didn''t consider the variable which is the combination of correlation and variance of disturbing term in the regression as well. If we change to multiple regression analysis, the performance of power may better than this one.