Dependent risk modeling and management

Dependent Risk Modeling and Management

As insurance industry evolves, dependence starts to emerge. Risks are likely to be dependent particularly in the emerging insurance markets, such as cyber insurance and autonomous vehicle insurance. The existence of dependence brings challenges in insurance risk modeling and risk management.  

A conventional approach to model dependence is through copula. It requires high-dimensional data, which is not always available. We aim to develop and explore general dependence notions beyond copula. Compared to the copula approach, these notions should be more flexible and requires lower resolution data for statistical modeling. Meanwhile, these notions should still provide adequate decision for the purpose of decision-making.

Risk management in traditional insurance business crucially replies on the assumption of independence. When dependence comes into play, new methodologies are needed for pricing and risk management. We aim to design novel premium principles that can accommodate both insurer and insured’s interest in the presence of dependence, and develop risk management strategies accordingly. We shall also explore the potential in combing these pricing and risk management techniques with other approaches, such as peer-to-peer insurance and catastrophe bond design.