Insurance risk: Realistic models for dependence

Research Project, 2010 – 2014

This industry PhD project is a cooperation between Folksam, Chalmers, the Royal Institute of Technology, and representative(s) from the Swedish Financial Supervisory Authority. An insurance company has to set aside buffer capital to meet financial risks, risks of cost increases, operational risks, political risks, and insurance risks. International accords such as Solvency 2 (and Basel II for banks) are the basis for national laws. If companies want to reduce their required buffer capital have to prove that they have sophisticated and realistic risk modeling systems which can estimate probabilities of extreme losses. A central challenge is dependence between risks - that problems can spread like a wildfire from one company to the next, and from one part of the economy to another, as vividly illustrated by the present financial crisis. The goal is understanding, measurement and management of risk dependence. The project ranges from development of theory to improvement of risk management at Folksam. The aim is also to influence risk management at other Swedish insurance companies. We will build on recent models for dependent extremes, and develop them into practical statistical methodology. Central questions are the relation between extremes of components of risk and extreme total risk, and time development of dependence between risks. The project will give unique access to data, revolve around data and practical needs, and will be used for practical risk management.