With the changes in economic structure and social development, the paths of most economic variables show the characteristics of structural changes. Traditional unit root tests, such as ADF and PP tests, often lead to wrong conclusions about data stationarity characteristics when considering such structural variation factors. This paper considers the test power of ADF statistic with one structural break using Monte Carlo simulation when the data generating process has double breaks in both intercept and slope. The results show that when there is a structural break in the data generating process if the conventional unit root test is performing without considering this break, the ADF test is still efficient when the degree of the structural break is relatively low. Furthermore, this paper discusses the influence of the size of the sample, break position and break degree on the power of the ADF statistic test under the premise of a low break degree.
Authors： xinchen chen
Journal of New Economics and Finance, Volume 2, Issue 1