This paper provides evidence that firms signal their private information about future earnings by their choice of split factor. Split factors are increasing in earnings forecast errors, after controlling for differences in pre‐split price and firm size. Furthermore, price changes at stock dividend and split announcements are significantly correlated with split factors, holding other factors constant, and with earnings forecast errors. These correlations suggest that management's choice of split factor signals private information about future earnings and that investors revise their beliefs about firm value accordingly. The analysis also suggests, however, that announcement returns are significantly correlated with split factors after controlling for earnings forecast errors. This suggests that earnings forecast errors measure management's private information about future earnings with error, that split factors signal other valuation‐relevant attributes, or that a signaling explanation is incomplete.

No abstract

A RECENT PAPER BY Ohlson and Penman [7] (henceforth OP) documents that, for stock splits larger than two-for-one (one hundred percent), the volatility of stock returns after the ex-split date is significantly higher than the presplit volatility. The increase in the standard deviation of daily returns is of the order of twenty-eight to thirty-five percent and persists for as long as a full year after the ex-date.' Even more interesting is their finding that there is no (permanent) change in the variance of returns at the announcement date. They investigate several potential explanations for this "aberration" but are unable to find a satisfactory answer. Grinblatt, Masulis, and Titman [6] (henceforth GMT) examine the behavior of expected returns around announcement dates and exdates for stock distributions exceeding ten percent. Classifying stock distributions as either stock dividends (split factors of twenty-five percent or less) or stock splits, they conclude that return behavior differs across these two types of distributions, with stock dividends exhibiting larger abnormal returns than splits. This is interpreted by them as evidence in favor of the "retained earnings hypothesis": the dissimilar accounting treatment of these two types of stock distributions affects retained earnings in different ways, and hence splits are not simply large stock dividends. This paper extends the work of OP to all types of stock distributions. In Section I, postsplit variance is shown to be different from the presplit variance for small stock splits and stock dividends. While small splits display an increase in volatility like large splits, stock dividends are associated with a decrease in the volatility of returns. This evidence also supports GMT's conclusions that stock splits and stock dividends are "different" types of events. Section II investigates the volatility of returns for reverse splits and demonstrates that postsplit volatility decreases for these events. This evidence is complementary to Woolridge and Chambers' [9] findings that reverse splits are also associated with negative abnormal returns at their announcement and ex-dates. Section III concludes the paper. I. Small Splits and Stock DividendsOP present convincing evidence (in particular, see OP Figure 1, page 257) that the variance of returns after the ex-date for large stock splits (greater than or * Graduate School of Business, Stanford Univeniity. I am grateful to Allan Kleidon, Robert Litzenberger, Myron Scholes, Eric Terry, and especially Michael Gibbons for their helpful comments. I am responsible for all remaining errors.' This is in contrast to the temporary increases (related to trading activity) documented by Beaver [1] and others for events such as earnings announcements. 163

No abstract

No abstract