This study employs panel smooth transition regression (PSTR) models with different lagged variables of earnings components as regressor to evaluate earnings persistence effects. The models can resolve collinearity problems between predictors, reflect firms' volatile or irregular earnings streams...