Dividend Yield, Risk, and Mispricing: A Bayesian Analysis

In the asset pricing literature, time-variation in market expected excess return tracked by financial ratios like dividend yield is typically attributed either to changing risk, related to the business cycle, or irrational mispricing. Extending the work on asset allocation and dividend yield by Kandel and Stambaugh (1996) to accommodate variation in risk as well as expected return, we develop Bayesian methods to examine the interaction between the data and an investor’s initial beliefs about the sources of return predictability. Our investor is also uncertain about the tradeoff between risk and the market risk premium and uses Merton’s (1980) proportionality condition as a prior reference point. We find that predictability in monthly market risk is statistically and economically important, but most of the yield related-variation in expected return is unrelated to risk. Differences in beliefs about mispricing and the risk-return tradeoff can have smaller but still “significant” effects on the utility of an investor allocating funds across a market index and riskless T-bills.