In this paper we examine the ex-dividend day returns of several taxable and non-taxable distributions.The ex-dividend day returns for the taxable common stocks are consistent with the hypothesis that dividends are taxed more heavily than capital gains. However, the cx-dividend day returns of preferred stocks suggest that preferred dividends are taxed at a lower rate than capital gains; non-taxable stock dividends and splits are priced on ex-dividend days as if they are fully taxable; and non-taxable cash distributions are priced as if investors recetve a tax rebate with them. We also find that each of these distributions exhibits abnormal return behavior for several days surrounding the ex-dividend day. We investigate several possible explanations for this anomaly, but none is capable of explaining the phenomenon.

We study the pricing effects of dividend and earnings announcements by taking advantage of the unique setting in Japan where managers simultaneously announce the current year's dividends and earnings as well as forecasts of next year's dividends and earnings. Defining surprises as deviations from analysts' forecasts, we find that share price reactions are significantly affected by earnings surprises, especially management forecasts of next year's earnings. The information content of dividends is marginal and is restricted to announcements of next year's dividends. Consistent with Modigliani and Miller's dividend irrelevance proposition, current dividend surprises have no material impact on stock prices in Japan. SINCE MODIGLIANI AND MILLER INITIATED MODERN corporate finance theory~seeModigliani and Miller~1958! and Miller and Modigliani~1961!!, academic wisdom has been that because dividends do not change a firm's underlying investment policy and cash f lows, dividend policy is irrelevant to firm value. It is earnings and earnings potential~the investment opportunity set! that are fundamental determinants of firm value: An extra dollar of dividends does not create value because it comes at the cost of a dollar of capital gains. Thus, it is of little surprise to find that numerous empirical studies show earnings announcements convey information that prompts investors to revise stock prices. In light of Modigliani and Miller's irrelevance theory, however, it is surprising to find that dividend research also reports a strong empirical tie between dividend announcements and stock price changes. To explain the observed pricing effects of dividends, a host of theories have been developed that model dividend announcements as a means of convey-* Darden Graduate School of Business, University of Virginia. We thank Citibank for support provided by its Global Scholars program, the Darden Foundation and TVA for financial support, and Morgan Stanley for supplying data. We also thank Sarah Garwood and Shinobu Suzuki for research assistance. Earlier versions of this paper were presented at the Office of Economic Analysis at the Securities and Exchange Commission, the 1998 FMA European meetings, and the University of Otago. We appreciate the comments of the participants at these presentations.

We investigate stock market rationality by examining the timeliness and unbiasedness of the market's response to dividend announcements.Our initial findings for market timeliness show a sluggish market reaction to dividend announcements; however, when the ex-dividend effect is controlled for, we find no evidence of a sluggish market reaction. We examine the unbiasedness of the market's response by testing whether the net announcement effect across a sample that is devoid of ex-post selection bias sums to zero. We observe a significant positive net announcement effect and examine several plausible conjectures for this puzzling phenomenon, but none provides a satisfactory explanation.

No abstract

No abstract