北大经院工作坊915场
ebicop: ensemble bivariate copulas for modeling multivariate cyber data breach risks
风险、保险与不确定性经济学工作坊
主讲人：徐茂超（伊利诺伊州立大学精算学教授）
主持人：
（人大财金）陈泽
（北大经院）贾若
（清华经管）冯润桓
参与老师：
（人大财金）魏丽
（北大经院）郑伟
时间：2024年6月18日（周二）10：00-11：30
线上形式：腾讯会议，会议号：777 394 864
线下地点：中国人民大学明德主楼515b
主讲人简介：
dr. maochao xu is a professor in the department of mathematics at illinois state university. he is also a co-founder of rankiteo ltd, a risk management company headquartered in london. dr. xu offers advisory and consultancy services in cyber insurance to various industry companies. his research specializes in cyber insurance, statistical modeling, and risk analysis, with his work published in prestigious journals and recognized with numerous awards, including the 2019 best paper award from the society of actuaries.
摘要：
cyber risk has emerged as a significant threat in recent years, with potential consequences including the exposure of sensitive information, identity fraud, and financial losses. modeling cyber risk is critical for effective risk management and underwriting in the insurance industry, presenting a formidable challenge due to intricate multivariate dependencies and limited data availability. in this talk, we introduce a novel ensemble learning approach that effectively captures both the temporal and cross-sectional dependencies inherent in cyber risks. unlike traditional methods that directly model multivariate dependence among risks, our approach leverages bivariate copulas to generate predictive members, thereby capturing multivariate dependence. the resulting predictive distribution is calibrated by minimizing the distribution score.
furthermore, we apply the proposed model to insurance pricing, demonstrating that it can lead to more profitable contracts. through extensive simulations and analysis of real-world data, our findings reveal that the proposed model has satisfactory fitting and predictive performance, outperforming existing models in the literature.
供稿：科研与博士后办公室
美编：初夏
责编：度量、雨禾、雨田