The Market Reaction to Stock Splits {Evidence from Germany," Schmalenbach Business Review, 54(3):280-290 A Monthly Double-Blind Peer Reviewed Refereed Open

Although stock splits seem to be a purely cosmetic event, there exists ample empirical evidence from the United States that stock splits are associated with abnormal returns on both the announcement and the execution day, and additionally with an increase in variance following the ex-day. This paper investigates the market reaction to stock splits using a set of German firms. Consistent with the U.S. findings, similar effects are observed for the sample of German stock splits. Institutional differences between Germany and the U.S. allow to disentangle the three main hypotheses on the announcement effect- signalling, liquidity, and neglected firm hypothesis – to gain further insights into their relative explanation power. This paper argues that legal restrictions strongly limit the ability of German companies to use a stock split for signaling. Consistently, abnormal returns around the announcement day are much lower in Germany than in the U.S. Although a significant increase in liquidity can be found after the split cross-sectional tests do not lend any support to the hypothesis that price changes are positively related to liquidity changes. This is in contrast to the results of Muscarella/Vetsuypens (1996) and Amihud/Mendelson/Lauterbach (1997). The paper shows that the announcement effect to German stock splits is best explained by a neglected firm effect. On the methodological side the effect of thin trading on event study results is examined. Using trade-to-trade returns increases the significance of abnormal returns but the difference between alternative return measurement methods is relatively small in short event periods. Thus, the observed market reaction cannot be attributed to measurement problems caused by thin trading. JEL classification: G 14 * I would like to thank Ekkehart Boehmer, Jörg Breitung and Richard Stehle for helpful comments. Support from the Deutsche Forschungsgemeinschaft (SFB 373, Quantifikation und Simulation ökonomischerProzesse) is gratefully acknowledged