Dividend Yield

Dividend yield(D/P)equals current dividend divided by current price.We exam how does the dividend yield act as the state variable to affect the portfolio returns in the time and space dimension in the Chinese stock market.In the time-series dimension,we find that:(1)the Chinese stock market return can be predicted by the dividend yield in the short term,and can not be predicted in the long term,and vice versa for the dividend growth;(2)the dividend yield has more predict power for the tradable value weight market portfolio return,and has more predict power for the total value weight market portfolio dividend growth;(3)the predict power of dividend yield shows the inverted U-trend with the tradable-total value ratio,that is,when the tradable-total value ratio of a portfolio is extremely high or low,the predict power of dividend yield is weak,and vice versa.In the cross-section dimension,we find the dividend yield pattern in return,which means the average return is increasing with the dividend yield.Based on this,we use 6(2×3)size-dividend yield portfolios to construct the dividend yield factor GMS(Generous minus Stingy).Adding the factor GMS in Fama-French three-factor model can improve the description on the underling portfolios.For 5 groups of underling portfolios,the four-factor model reduces the average absolute intercepts of 0.1,3.2,1.7,3.0 and 2.8 basis points from Fama-French three-factor model.In addition,eliminating the value factor HML can further reduce the 25 size-BM,25 size-D/P and 32 size-B/M-D/P portfolios' average absolute intercepts of 1.0,2.0 and 4.5 basis points.Based on the full time sample,we divide it into 7 subsamples,and find the same results.The construction of GMS has no link with B/M,however,the factor GMS can explain the anomaly on value,which means that GMS is not a tautology factor.