This paper adapts Gouriéroux and Monfort's (2021) model risk framework to property and casualty insurance, focusing on policy-level data. It addresses model risk at two levels: the impact on predictions and out-of-sample uncertainty, and the need to account for risk during model selection.
The paper develops a machine learning algorithm using financial data to identify Italian private firms linked to organized crime. By analyzing firms with Mafia connections, it achieves a 74.9% AUC and 91.4% precision. This method serves as a risk management tool and supports legal enforcement actions.
“… the present study developed the AI Categorization and Classification (AI-CC) Method as a central artifact to provide guidance on the use of AI within the profession. The target users of the AI-CC Method are regulators, standard setters, the strategic management of the Big Four, and individual auditors.”
The paper explores how advanced technologies like AI pose both potential and complexity in risk and safety applications. It delves into explainability and interpretability within risk science, emphasizing their role in enhancing assessment, management, and communication of risks, illustrated with autonomous vehicles examples. Aimed at stakeholders navigating tech's impact on risk.
The era of big data revolutionizes operational management in enterprises, amplifying the challenges for auditors in managing vast corporate information and escalating fraud risks. This study explores machine learning's role in identifying financial fraud, constructing models based on fraud triangle theory and empirical data. The model, particularly LightGBM, achieves a 73.21% accuracy, showcasing its effectiveness in predicting fraud risks in publicly traded companies.