Momentum, Book-to-Market Equity, and the 2003 Dividend Tax Cut

Abstract

We investigate market reactions and portfolio performance around the announcement of the January 2003 proposal to eliminate shareholder-level taxes on dividends. Although stock reactions during the period depend on dividend yield, we find momentum and book-to-market are also important in explaining cross-sectional stock returns. It appears that the dividend tax cut proposal triggered a market-wide re-valuation of stocks, which magnified the momentum and book-to-market effects, even for those stocks that were not paying dividends. Therefore, the period surrounding this announcement provides a unique opportunity to present new evidence on the causes for the momentum and book-to-market effects. We examine portfolios formed on dividend yield, momentum, and B/M during a 26-day window that includes a 9-day sub-window. Our most striking finding is that the momentum effect is extremely sensitive to the holding period. There is strong return continuation during the long window but reversal during the short window. The momentum and reversal effects are not driven by dividend yield, because these effects are among the strongest across zero-yield stocks. The coexistence of strong momentum and reversal in the same sample over a short period with reinforcing news announcements suggests that time-varying risk premia cannot explain the momentum and reversal effects, because the probability of the risk premium flipping signs in 26 days is remote, and it is equally improbable to expect this strong pattern of returns to be caused by chance, especially in view of the fact that we observe the effect in portfolios of stocks, not individual equities. We find that the B/M effect is not driven by dividend yield since, even for zero-yield stocks, high B/M portfolios outperform low B/M stocks in the 9-day window. After controlling for yield and momentum, we detect significant and positive B/M coefficients. While, in principle, the B/M effect could be driven by risk, the evidence does not appear to support such an explanation because (1) the B/M effect over the 26-day window is significant only when conditional on momentum, (2) the B/M effect is concentrated in the highest and lowest momentum stocks, (3) the reversal effect is strongest in high B/M stocks, and (4) we continue to find significantly positive B/M effects even after controlling for distress risk. Taken as a whole, our results are consistent with the notion that momentum and reversal are related to investor sentiment and the B/M effect a result of market's correcting its own misvaluation.