Identification of a Company’s Business Model with Text Mining of Annual Reports

The herein paper is aimed to identify Business model types through text mining of companies' annual reports to fill the gap in the literature that lacks data-driven classifications and provide entrepreneurs with a map of existing Business models and the current trends. The study can be useful to businesses who want to know about the most effective ways of creating value and be aware of innovations in Business models. Using data on the S&P 500 companies' Item 1 part of 10-K reports in 2019 the 24 Business model patterns but not strict Business model elements or types were revealed by applying topic modeling. A company may contain several allocated patterns at the same time. The results are supported by the findings of linear regression models built on 2019 and 2020 data that some Business models perform better and show higher ROE and ROA, while some suffer a negative effect of belonging to a particular pattern.