Second, we conduct regression analyses to investigate whether CEO (CFO) LS can predict CFO (CEO) LS. In our first regression, we use CFO LS at the end of CEO-CFO tenure overlap as the dependent variable. The predictors include CEO LS, CFO LS, firm size (natural logarithm of total assets), firm performance (ROA), R&D intensity, CFO gender, and CFO age, and two-digit SIC industry dummy variables. All the predictors are measured at the beginning of CEO-CFO tenure overlap. We also include CEO-CFO tenure overlap as a predictor. In unreported results, the coefficient estimate of CEO LS (beginning) is positive and statistically significant (β=0.046, p