I show that firms can manipulate their stock price by restricting the ability of investors to sell. In Japan, new shares created from a stock split are not distributed to investors until the pay-date, several weeks after the ex-date. During this time, investors can trade their old shares but not their forward claims on the new ones. In a simple model, I show that when investors disagree about the economic implications of a split, a high split ratio tightens short sale constraints, pushing up prices. When the shares are distributed, the constraint is relieved and prices fall. The theory explains why (a) the average abnormal return associated with a split in Japan has grown to over 30%, (b) split ratios have risen from an average of 1.15-for-1 to over 10-for-1 between 1995 and 2005, (c) significantly negative (positive) ex-date returns (pay-date returns) are positively (negatively) related to the split ratio, a measure of the float reduction. Taken together, the results suggest that firms may actively attempt to restrict the float when differences of opinion are high.