*4.5. Robustness Test*

To ensure that the empirical findings are accurate, this paper instead measures *SG* using ROE (Cao and Wang, 2017; Song and Wang, 2018) [58,59]. The association between *ER*, *RE*, and *SG* is then measured again and the test results are shown in Table 7. Models 1, 2, and 3 measured the relationship between *ERA*, *ERB*, *ERC*, and *SG*, respectively, and the results showed that the coefficients of the *ERA*, *ERB*, and *ERC* terms were 0.216, 0.224, and 0.219, respectively, at the 5% level of significance. The relationship between the interaction terms of the *ER* series variables and *RE* variables on *SG* was measured in Models 4, 5, and 6, respectively, and the results showed that at the 5% level of significance the coefficients of the *ERA* × *RE*, *ER* × *RE*, and *ERC* × *RE* terms are −0.298, −0.344, and −0.316, respectively. In summary, it can be seen that the results of the robustness tests are basically consistent with the previous empirical results and the empirical findings are stable and valid.



Note: \*\*\* *p* < 0.01, \*\* *p* < 0.05, \* *p* < 0.1.
