Four Essays on the Empirical Testing of the Efficient Market Hypothesis

In the finance field, hardly any topic is discussed as controversially as the hypothesis of efficient financial markets. For more than 50 years, economists have been tried to test this hypothesis both theoretically and empirically, without reaching a consensus. While partisans of the neoclassical school of thought propagate a perfectly rational investor and thus a fully efficient financial market, adherents of behavioral finance believe in bounded rational financial market players and market inefficiencies. Therefore, this dissertation aims to examine different areas of the capital market with respect to market efficiency and thus gain a deeper understanding of the price formation processes, as well as the investor’s information processing. More specifically, this thesis examines (1) price overreactions and underreactions in the cryptocurrency market and subsequent price patterns, (2) mergers and acquisitions of target and acquiring companies in the European Union with respect to price movements and their influencing factors, (3) takeover competitions in the European Union and their implications for stock prices and investor expectations, and (4) the causal relationship between investors' online sentiment and US stock prices in conjunction with the social reach of the users and their posted tweets. The first research article includes a methodologically comprehensive investigation of patterns following abnormal price reactions in the crypto market. This work contributes to the literature by first observing overreactions and underreactions, then detecting a so-called magnitude effect using specific thresholds, and finally examining the overreaction effect for several periods. While the results are mixed on the days after positive abnormal price reactions, we demonstrate a clear overreaction effect for negative abnormal price reactions followed by a price reversal on the following days. Furthermore, the market capitalization of the respective coins plays a significant role, implying that smaller cryptocurrencies experience stronger overreactions and market inefficiencies. The results are robust across different methodological approaches. The second research article offers new insights into the market for mergers and acquisitions in the European Union. Using an extensive dataset consisting of more than two thousand takeover announcements, the stock price movements of the target and the acquirer are examined before, during, and after the announcement. Important contributions are the consideration of multiple takeover forms and the linkage of the target’s and acquirer’s returns, suggesting synergy effects. This paper finds several indications of market inefficiencies, such as the abnormal price increase of the target before the takeover bid. The third research article proposes a method to study takeover competitions and their implications for stock price changes and investor expectations. The separate investigation depending on the bid order, as well as the examination of the anticipation of a subsequent bid, represent contributions to the literature. We show that abnormal returns are significantly high for target firms, but decrease with each subsequent bid. In contrast, there is no reaction of acquirer stocks, nor is there a discernible pattern regarding the bid order. The fourth research article looks at the online sentiment of users on the investor platform StockTwits in connection with price reactions in the U.S. stock market. We contribute to the literature with extensive insight into user activity, an examination of factors influencing social reach, and the incorporation of this user and tweet information into the stock price analysis. Using our deep learning model, we show that a sentiment classification accuracy far above established models is feasible. We hardly detect any effect of sentiment on stock price movements. The inclusion of social reach does not contribute to the explanation of stock price changes. In contrast, stock price changes appear to have a strong causal influence on online sentiment, particularly during intraday periods