ABSTRACT
Most previous studies that have confirmed size anomalies in U.S. bank stock returns rely on a simple OLS regression. However, empirical evidence shows that all the financial variables used in their analyses are not normally distributed, which could cause inconsistent coefficient estimates. As a remedy for this possible issue, we introduce panel Gaussian copula marginal regression as an alternative approach because it does not require independent, identically distributed random variables in empirical analysis. In particular, we find that three Fama-French stock-risk factors have positive impacts on excess risk-adjusted bank returns, but no effects of bond risk factors. We also confirm that there exists the size effect on excess commercial bank stock returns in the U.S.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
1 In our analysis, we have and
.
2 Note that the model selection results with the different error correlation structures are available upon request.