Telling True from False: Do Markets React Efficiently to Takeover Rumors?

Abstract: This thesis examines how U.S. markets react to 345 different takeover rumors, and how efficiently the market is capable of distinguishing between true and false rumors. The markets do appear to react differently to true rumors, with a statistically significant difference in the stock price reaction on the day of the rumor, giving weight to the hypothesis that the market can discern true rumors from false. Using this finding, we attempt to create a trading strategy by applying methods previously used for estimating the likelihood of success for takeovers to rumors. With the use of logit classification models and holdout samples we seem to be able to distinguish between true and false rumors with some success, but we cannot profit from trading on the rumors. Using an out-of-sample dataset also on the U.S. market as well as a sample of U.K. rumors, we manage to classify rumors with some degree of precision using the model constructed from estimates from the primary sample. The stock market seems to be reacting efficiently to rumors as we cannot find any evidence to the contrary. We also find a rumor run-up before the rumor publication, which we attribute to the rumor diffusing through the market rather than to insider trading.