Athens University Economics and Business (AUEB)
Essays on textual analysis in finance
Abstract
dc:descriptionThis dissertation consists of five chapters. Chapter one is the introduction and presents in a clear and concise way the main objectives of this thesis. Chapter two employs natural language processing techniques in an effort to quantify the qualitative information emanating from U.S. banks’ annual reports. In turn, the obtained textual features are combined with conventional financial variables and fed as inputs to several machine learning models, including deep learning ones, to predict distressed delisting events from the NYSE/AMEX or NASDAQ exchanges. According to the documented results, the combination of these heterogenous information sources improves the classification performance of both traditional statistical techniques (e.g., logistic regression) as well as of sophisticated algorithms. Chapter three also employs textual analysis techniques, but this time from a different perspective. Particularly, annual report tone ambiguity proxies are constructed based on the fraction of ambiguous words, according to the sentiment word lists of Loughran and McDonald (2011). Then, the associations between the constructed tone ambiguity proxies and future stock price crash incidents are explored. However, the main innovation of this study is that these associations are examined after controlling for the state of the economy. The reported results are in favor of this conditional association, since we report statistically significant associations only during recessions. Chapter four is focused again on U.S. banks’ stock price crash incidents. Particularly, it explores the associations between the appearance of such incidents with the competition of the market that the banks operate, as well as with loan portfolio diversification activities. According to the documented results, banks that operate in highly competitive environments are more likely to experience stock price crash incidents. However, loan portfolio diversification activities are beneficial for them since they moderate their future stock price crash risk. Finally, chapter five presents the drawing conclusions of this dissertation and outlines avenues for future research.
Degree
thesis:*- Grantor dc:publisher
- Athens University Economics and Business (AUEB)
- Year dc:date
- 2024
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Gkoumas, Nikolaos
- Γκούμας, Νικόλαος
Subjects
dc:subject × 12Rights
- Language dc:language
- eng
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/55931