A note on the Gordon growth model with nonstationary dividend growth

Researchers have sometimes argued that the recent ascent in stock prices could be explained in some measure by changes in expectations about long-run future dividend growth. For example, Barsky and De Long (1993) argue that a small random walk component in the growth rate of dividends, when extrapolated into the future, is capable of reproducing the large swings in US stock prices over the period 1880-1990. I show that the hypothesis of a nonstationary permanent growth rate of dividends is inconsistent with the Gordon growth model.