In a context of widespread instability, the actuarial analysis of emerging risks takes on particular importance to facilitate the adaptation of insurance to these profound changes. Whether it be old risks with poorly anticipated consequences (such as the epidemiological risk, as demonstrated by the recent Covid-19 crisis, whose consequences can significantly exceed purely healthcare boundaries), risks undergoing strong and worrying evolution (such as climate risk), or more recent risks (such as cyber risk), the task of actuaries is particularly delicate. The lack of data is notably a hindrance to a rigorous approach to these important questions. This session aims to lay the groundwork for a scientific approach to this problem and to introduce some tools for discussing the insurability of these new risks.

Actuaries with basic knowledge on statistics. Being familiar with Bayesian analysis is preferable but not mandatory.

Technical Requirements
Please check with your IT department if your firewall and computer settings support web session participation (the programme Zoom will be used for this online training). Please also make sure to join the web session with a stable internet connection.

This session will notably have two main angles. Firstly, it will raise the question of data, their quality, and the necessary reliance on expert judgment, crucial for risks where experience is limited. Bayesian analysis notably allows for this integration, but does not exempt from a critical examination of the quality of expert data, and we will thus present some methods to attempt to evaluate it.

Secondly, we will examine the shift from a "historical data" approach to a "scenario-based" approach. Because while the notion of scenario allows for anticipating situations never encountered before, their design must adhere to scientific standards to avoid being purely speculative.

A simplified illustration of these issues will be provided during the training. An R notebook will allow participants to follow and adapt the implementation without needing an in-depth knowledge of the software.

The language of the web session will be English.

OLIVIER LOPEZ
Olivier Lopez is Professor at Ensae Institut Polytechnique de Paris, co-supervisor of the master degree in Statistics, Finance and Actuarial Science. He formerly directed the Institute of Statistics of Sorbonne Universite from 2016 to 2023. He is a fully qualified member of Institut des actuaires (French actuarial association), and member of its Scientific board since 2013. His research is mainly related to the use of statistics and machine learning to actuarial sciences, with a particular focus on the insurability of emerging risks (epidemiology, natural disasters, cyber).