Cyber Risk: Statistical Approaches and Insights

Cyber risk is a major challenge in the digital economy, requiring appropriate statistical methodologies for accurate assessment and evaluation. This work illustrates different statistical approaches to model cyber risk, addressing multiple aspects of the problem. Our analysis is structured around various methodological frameworks. First, we employ ordered response models to evaluate the severity of cyber attacks, expressed on an ordinal scale, as a function of the characteristics of the attacks. Second, we employ network-based models to assess vulnerabilities among cyber attack victims. Finally, we introduce a distributional inference approach based on Wasserstein propagation to estimate cyber risk at the country level, particularly for regions with limited data availability. The proposed methodologies are applied to real-world data on cyber attacks that have occurred worldwide over the past decades. This research provides valuable information for policy makers and industry stakeholders to improve cybersecurity strategies.