Past studies have documented an ex-dividend day price drop which is less than the dividend per share and positively correlated with the corresponding dividend yield. In contrast to prior work, we show that, without additional information, the marginal tax rates cannot be inferred from this phenomenon which is, therefore, not necessarily the result of a tax induced clientele effect. Despite adjustments for potential biases in earlier work, however, the correlation between the ex-dividend relative price drop and the dividend yield is still positive which is consistent with a tax effect and a tax induced clientele effect.
THE EFFECT OF DIVIDEND policy on stock prices is an issue of growing interestand controversy in the financial literature. As Miller and Modigliani [19] show, if capital markets are perfect the dividend policy of the firm, for a given investment policy, does not affect its market value. However, in a world in which dividends are taxed more heavily than capital gains, investors may demand higher before-tax returns to hold securities with high dividend yield.' Furthermore, in such a world, investors could form "clienteles" each preferring a particular dividend yield.2 In particular, investors in high income tax brackets might find it advantageous to hold low dividend yield stocks, while those in lower income tax brackets concentrate their holdings in high dividend yield stocks.3Although the notion of a tax-induced clientele effect has intuitive appeal, there are serious questions as to its existence. Long [18] pointed out that the portfolio dividend yield choice cannot be made independently of the risk expected return trade off, since the dividend yield of all mean variance efficient portfolios is a linear function of their nondiversifiable risk. If, for example, dividend yield is positively correlated with risk, and wealthy investors have high tolerance to risk, they may hold high dividend yield portfolios even though they pay a higher tax on dividend income than on capital gains. Furthermore, as Miller and Scholes * Graduate School of Business Administration, New York University. This paper is a revised chapter of my Ph.D. dissertation at the University of Rochester, N.Y. I wish to thank my dissertation Committee-John Long (Chairman), Jerold Warner, and Ross Watts-for their support and encouragement. I would also like to thank the participants in the finance workshops of the University of Rochester, the University of Chicago, New York University, Northwestern University, and the University of Pittsburgh. Finally, I would like to thank Robert H. Litzenberger and Michael Brennan for their valuable comments. Unfortunately, the remaining errors are all mine.'Brennan [6] was first to extend the single period Capital Asset Pricing Model to include the preferential treatment of capital gains. He shows that, for the same level of risk, the before-tax expected rate of return of a stock is an increasing function of its dividend yield.
1059
