Prior research has used inaccurate classification rules to distinguish between stock splits and stock dividends. The CRSP classification of two-for-one stock distributions agrees with the actual accounting treatment only 23% of the time. In addition, the accounting treatment impacts the announcement period reaction—two-for-one distributions accounted for as stock dividends are associated with five-day announcement period returns of 2.70%, significantly greater that the 0.93% announcement returns for distributions accounted for as stock splits. Announcement returns are positively related to earnings growth in the two years following the distribution for stock dividend firms but not for stock split firms. The accounting choice appears to be used to confirm management's private information about future earnings revealed at the time of the distribution announcement.

We observe significant post-split excess returns of 7.93 percent in the first year and 12.15 percent in the first three years for a sample of 1,275 two-for-one stock splits. These excess returns follow an announcement return of 3.38 percent, indicating that the market underreacts to split announcements. The evidence suggests that splits realign prices to a lower trading range, but managers self-select by conditioning the decision to split on expected future performance. Presplit runup and post-split excess returns are inversely related, indi? cating that our results are not caused by momentum.