Determinants of Dividend Payouts: The Relative Contribution of Dividend Predictors in KOSPI and KOSDAQ Firms

Abstract

We decompose the traditional dividend model of Fama and French to investigate the relative contribution of major dividend predictors and the evolution of the model’s explanatory power over the economic crisis periods using KOSPI and KOSDAQ sampled firms. The relative weight analysis suggests that the predictive power of the traditional dividend model mainly comes from the earnings related firm characteristics, such as profitability and firm lifecycle, while the simple firm size, investment opportunity or risks can only play the supplementary role. The rank and the magnitude of this relative contribution remain very stable over time and the model as a whole explain both KOSPI and KOSDAQ firms’ dividend behaviors fairly well during the noncrisis period. However, the model’s dividend predictors all lose explanatory power during the crisis period. Particularly, KOSDAQ firms are more severely damaged by macroeconomic shocks and the changes in their firm characteristics can no longer explain majority of the variations in their dividend policy. Rather, the credit rating information that could only marginally contribute to the model accuracy, now emerges as a top tier influencer during the crisis period. Therefore, caution should be taken because the naive universal application of the model may bear noisy results.