Even though we cannot reject the null hypothesis based on the
result of the Hausman test, we still have a concern about the testing
power of the test and the choice of instrument variables. To alleviate
the suspected endogeneity, we next try to use the lagged explanatory
variable to run the QR. The result is reported in Table 10.
The F tests show that differences across various quantiles are significant
at the 1 per cent level for most situations. Therefore, we
conclude that the previous quantile-varying result of the impact of
stock-based compensation on RoE is robust after considering the
endogeneity.