*3.1. Stationary Test and Multicollinearity Test of Variables*

Before regression, it is required to test the stationarity of the data to avoid pseudo regression. The unit root test is a common method to test the stationarity of data. For example, Levin, Lin and Chu [65] derived the unit root test (LLC test) from Levin and Lin [66], and it was proposed by IM, Pesaran and Shin [67] that the unit root test considers panel heterogeneity (IPS test). The test results are shown in Table 3. In the LLC test, all the variables deny the original hypothesis of the "existence of a unit root" at the level of 1%, while, in the test of IPS, the variables such as ln*NE*, ln*pgdp*, ln*r*&*d*, ln*is*, ln*ee* and ln*hc* were unstable. After the first-order difference, all the variables rejected the original hypothesis of the "existence of a unit root" at the level of 1%. At the same time, multicollinearity will distort the regression estimation model. The value of the variance inflation factor (VIF) was calculated to judge whether there was multicollinearity between the prediction variables or not. It was found that the VIF values were less than 10. Therefore, from this, it could be seen that there was no multicollinearity among the variables to ensure the effectiveness of the following regression estimation results.

**Table 3.** Unit roots and multicollinearity test.


Note: \*, \*\* and \*\*\* indicate significance at the levels of 10%, 5% and 1%.
