Financial crime risk assessment: machine learning insights into ownership structures in secrecy firms.

This study examines the relationship between corporate secrecy and financial crime, presenting an analytical framework to strengthen risk assessment efforts. We develop secrecy indicators using corporate ownership data from over 2.6 million firms across eight European countries. These indicators are validated using machine learning models built upon evidence of crime committed by firms and/or their owners. The results demonstrate robust predictive power: firms with complex structures and owners from high-risk jurisdictions show a higher likelihood of engaging in illicit activities. Incorporating macro-level information, such as geographic location and economic sector, enhances the understanding of this phenomenon. These findings advance empirical knowledge about the nexus between secrecy firms and crime, offering anti-money laundering authorities novel machine learning tools for effective risk assessment. [ABSTRACT FROM AUTHOR]