A Longer Look at Dividend Yields

This article re-examines the evidence on the ability of dividend yields to predict long-horizon stock returns. We use two new series beginning in 1871, a monthly series for the United States, and an annual series for the United Kingdom. Conditional on survival over the entire 122 years, dividend yields display only marginal ability to predict stock market returns in either country. We also argue that tests over long periods may be affected by survivorship. Simulations show that regression statistics based on a sample drawn solely from surviving markets can be seriously biased towards finding predictability.