The Dog that Did Not Bark: A Defense of Return Predictability
Review of Financial Studies 21(4) 2008 1533-1575. Taken alone, returns may not look that predictable. However, price-dividend ratios vary, so either returns or dividend growth must be forecastable (or both). Implications for dividends, and long-run forecasts give strong statistical evidence against the null that returns are not forecatsable. I address the Goyal-Welch finding that forecasts do badly out of sample, and the long literature criticizing long-run forecasts. The most important practical takeaway: even if you assume that all variation in market p/d ratios comes from time-varying expected returns, and none corresponds to dividend growth forecasts, you will typically find that market-timing strategies based on fitting the regression don’t work. Corrected Table 6. Three numbers were wrong in the published version. Thanks to Camilla Pederson for catching it.