"Previous research showed that the dividend price ratio process changed remarkably during the 1980's and 1990's, but that the total payout ratio (dividends plus repurchases over price) changed very little. We investigate implications of this difference for asset pricing models. In particular, the widely documented decline in the predictive power of dividends for excess stock returns in time series regressions in recent data is vastly overstated. Statistically and economically significant predictability is found at both short and long horizons when total payout yield is used instead of dividend yield. We also provide evidence that total payout yield has information in the cross-section for expected stock returns exceeding that of dividend yield and that the high minus low payout yield portfolio is a priced factor. The evidence throughout is shown to be robust to the method of measuring total payouts"--National Bureau of Economic Research web site.