The drivers of reputational risk are still far from explicit, making proactive risk management and quantitative research rather difficult. The Basel Committee on Banking Supervision encourages financial institutions to systematically identify reputational risk drivers; however, such drivers still represent an unsolved problem. Therefore, the objective of this paper is to systemically identify reputational risk drivers from textual risk disclosures in financial reports. We find that textual risk disclosures in financial reports contain abundant information about the causes of reputational risk, thus indicating the possibility of systematically identifying the reputational risk drivers. To accurately extract reputational risk drivers from massive and unstructured textual risk disclosure data, we modify a text mining method to make it more suitable for this type of textual data with noise words. Based on 352,326 risk headings extracted from 11,921 annual reports released by 1570 U.S. financial institutions from 2006 to 2019, a total of 13 reputational risk drivers are identified to extend upon existing studies. The importance of reputational risk drivers and their dynamic evolutions are also quantified to discover the drivers of greatest concern. This paper can clarify the sources of reputational risk to help companies realize proactive reputational risk management and provide a theoretical basis for further quantitative studies, especially the measurement of reputational risk.
Read full abstract