Capital Market Efficiency and the Influence of Investor Sentiment on Stocks with Positively Skewed Returns

Jakob Lars Bertil Kostet & Stefan Meyer

Student thesis: Master thesis


According to the Efficient Market Hypothesis, investor psychology should not have consistent effects on asset pricing. We analyze whether a specific asset pricing anomaly, the underperformance of stocks with positively skewed return distributions, is influenced by investor psychology in the form of investor sentiment. In such a case, our results may present challenging evidence for the Efficient Market Hypothesis and its relevant implications for asset pricing, corporate finance and the economy. The study is furthermore motivated by prior research suggesting that the asset pricing anomaly is driven by investors’ misestimates of return probabilities and that these misestimates may be amplified during periods of high investor sentiment. We test this hypothesis by dividing historical return data into different states of investor sentiment to analyze whether the underperformance of positively skewed stocks is strongest following periods of high sentiment. To approximate investor sentiment, we use two existing sentiment measures of different kinds: one financial measure and one survey-based measure. Furthermore, we construct two text-based sentiment measures by analyzing two decades of news articles from the New York Times with a bag-of-words algorithm. To analyze the influence of sentiment on future stock performance, we perform a series of Fama and MacBeth (1973) regressions of next-month returns during low, medium and high periods of investor sentiment. While controlling for several stock risk factors, we find an underperformance of stocks with positively skewed return distributions. This underperformance, in line with prior research, is statistically significant and economically relevant. However, we do not find indications that the underperformance is driven by high investor sentiment. On the contrary, we find indications for a reverse effect in comparison to our predictions. Further tests reveal that the underperformance of positively skewed stocks may be influenced only marginally by investor sentiment or not at all. For example, we find that other factors, such as GDP growth, may influence the underperformance of positively skewed stocks to a greater extent. Therefore, our results do not challenge the validity of the EMH. As a research contribution to the measurement of investor sentiment, we find indications that the New York Timesbased sentiment measures approximate some fraction of true investor sentiment. Moreover, correlation analyses of multiple additional sentiment measures indicate that only few of the measures seem to react to changes in investor sentiment in a similar fashion

EducationsMSc in Accounting, Strategy and Control, (Graduate Programme) Final Thesis
Publication date2018
Number of pages152
SupervisorsMarcel Fischer