ABSTRACT
In this study we focus on extracting qualitative information from the management discussion and analysis (MD&A) section of an annual report and compare whether there are textually evident differences in textual expressions used between bankrupt and non-bankrupt companies. We extract high-frequency words, related concept links, and topics from MD&As and find that some high-frequency words appear to suggest differences between bankrupt and non-bankrupt companies regarding their financial position and ongoing status. However, the usefulness of concept links is mixed. Some concept links for high-frequency words do not seem to center around a theme or a key word, yet others provide some contextual information supporting our conjectures about the ongoing business status of non-bankrupt companies. Finally, we perform topic extraction based on a latent semantic analysis algorithm in order to investigate whether issues and themes discussed differ between non-bankrupt and bankrupt companies. We find that most of the top topics extracted merely recapture the characteristics of industries in which companies operate and do not provide information in differentiating between bankrupt and non-bankrupt companies. The reasons are discussed in the paper.
