Thewissen, JamesSekulic, SashaSashaSekulic2025-05-142025-05-142025-05-142023https://hdl.handle.net/2078.2/35955This master thesis aims to determine whether the content of press releases disclosed by listed U.S. companies can be analysed using topic modelling to detect firms with aggressive attitudes towards taxation. Three objectives were undertaken in order to answer this research question. It includes the identification topics discussed in the firm’s press releases. Then, the assessment of the informativeness of thematic content in detecting tax-aggressive firms, and finally, determine which topics are associated with tax aggressiveness. The methodology involved the use of Latent Dirichlet Allocation (LDA) through the topic modelling process to determine and quantify in an empirical manner different topics from the 8-K narrative covering a period from 2004 to 2016. The research methodology employed Tobit regressions with financial tax aggressiveness indicators (ETR, CETR, MPBT), and control variables which account for firm characteristics. Additionally, an ANOVA test was employed to examine disparities in thematic content between tax-aggressive and non-aggressive firms. It has been found that topic modelling enables to generate significant lists of relevant topics. Moreover, the use of topic modelling in press releases improves the identification of tax aggressiveness, revealing topics such as finance, accounting, risk, legal disclosures, and development to be associated with tax aggressiveness. By incorporating these topics into the Tobit regression models, it has been observed that it significantly enhanced their performance. Overall, the results suggest that topic modelling is a useful tool to evaluate the content present in press releases. It offers important insights into the detection of tax aggressiveness in U.S. listed companies. This method could assist regulators, investors and other interested parties in the identification of firm’s tax aggressivity. It could facilitate informed decision-making and contribute to developing strategies to monitor and address corporate tax aggressiveness.Topic modellingCorporate disclosurestext::thesis::master thesisthesis:40308