4.3. The relationship between operating efficiency and intellectual capital
As the data used in this study are drawn mainly from publicly available sources, we have to restrict our sample period to only 2009 and 2010. In other words, we indirectly place an additional restriction on our sample for regression analysis that include our intellectual capital proxies, since data on these proxies are only available from the annual reports of the Chinese life insurers.9 A strong correlation is expected between the input/output variables and explanatory variables in small samples, indicating that the regression assumption that error terms are independent of explanatory variables is violated [47]. Note, however, that the small sample is unlikely to introduce any bias since this study applies the truncated regression with bootstrapping approach, which could overcome the above problem.
Before conducting the regression analysis, we first perform a test of the possibility of multicollinearity. The results of the Pearson correlation coefficients are reported in Table 7. The correlation coefficients between OE and HCE, SCE, and CEE are positive, with SCE reaching the conventional significance level. Other correlation coefficients are generally lower than 0.5. In addition, the untabulated values on the diagnostics of variance inflation factors (VIF) are all less than 2.8. The findings suggest the non-existence of the potential multicollinearity problem.
4.3. The relationship between operating efficiency and intellectual capital
As the data used in this study are drawn mainly from publicly available sources, we have to restrict our sample period to only 2009 and 2010. In other words, we indirectly place an additional restriction on our sample for regression analysis that include our intellectual capital proxies, since data on these proxies are only available from the annual reports of the Chinese life insurers.9 A strong correlation is expected between the input/output variables and explanatory variables in small samples, indicating that the regression assumption that error terms are independent of explanatory variables is violated [47]. Note, however, that the small sample is unlikely to introduce any bias since this study applies the truncated regression with bootstrapping approach, which could overcome the above problem.
Before conducting the regression analysis, we first perform a test of the possibility of multicollinearity. The results of the Pearson correlation coefficients are reported in Table 7. The correlation coefficients between OE and HCE, SCE, and CEE are positive, with SCE reaching the conventional significance level. Other correlation coefficients are generally lower than 0.5. In addition, the untabulated values on the diagnostics of variance inflation factors (VIF) are all less than 2.8. The findings suggest the non-existence of the potential multicollinearity problem.
การแปล กรุณารอสักครู่..
4.3. The relationship between operating efficiency and intellectual capital
As the data used in this study are drawn mainly from publicly available sources, we have to restrict our sample period to only 2009 and 2010. In other words, we indirectly place an additional restriction on our sample for regression analysis that include our intellectual capital proxies, since data on these proxies are only available from the annual reports of the Chinese life insurers.9 A strong correlation is expected between the input/output variables and explanatory variables in small samples, indicating that the regression assumption that error terms are independent of explanatory variables is violated [47]. Note, however, that the small sample is unlikely to introduce any bias since this study applies the truncated regression with bootstrapping approach, which could overcome the above problem.
Before conducting the regression analysis, we first perform a test of the possibility of multicollinearity. The results of the Pearson correlation coefficients are reported in Table 7. The correlation coefficients between OE and HCE, SCE, and CEE are positive, with SCE reaching the conventional significance level. Other correlation coefficients are generally lower than 0.5. In addition, the untabulated values on the diagnostics of variance inflation factors (VIF) are all less than 2.8. The findings suggest the non-existence of the potential multicollinearity problem.
การแปล กรุณารอสักครู่..