A0817
Title: Improving the supervisors anti-money laundering risk rating approach using news event monitoring
Authors: Kris Boudt - UGent, VUB, VUA (Belgium) [presenting]
Olivier Delmarcelle - Ghent University (Belgium)
Pascal Ringoot - NBB (Belgium)
Abstract: A tool is developed to support the regulator's risk assessment process of financial institutions in the context of money laundering and financing terrorism. As part of their mandate to safeguard financial stability, supervisors rate each institution and provide them advice on how to improve their policy. This is most often done using a manual analysis and expert judgement. Due to the yearly frequency of this process, it is important not to overlook important developments between successive reportings. It is shown how the integration of a news event and monitoring system in the Belgian supervisors' AML process reduces the manual work and can preselect relevant news articles to ease the risk rating. We specify how natural language processing and fuzzy matching techniques improve the selection process and how an articles importance is calculated by combining the keywords and institutions' relevance. Finally, we demonstrate how well this compares to the current news monitoring.