Micro Reserving in Non-Life Insurance – A Challenge for Risk Management
Recent surveys carried out among various market players have shown that aggregated reserving methodologies remain preferred in the calculations of non-life insurance reserves. However, the recent development of machine learning in actuarial sciences allows the emergence of new visions of claims and show a growing interest in new ways to estimate reserve based on individual assessment, especially for the atypical and long term risk.
This web session presents a new way to integrate these methods by the actuarial reserving team to produce a complete and innovated process.
As a good data quality is require for this type of studies, the use of data analysis and machine learning could help to check the accuracy of the database and complete it if necessary. Therefore, it allows the company to better understand the impact of each description variables on the claims experience, and thus to define finer homogeneous cohorts and large claims thresholds better adapted to the underlying risk. These preliminary analyses allow to better anticipate future deterioration of specific cases using a segmentation model for large and attritional claims, estimate reserves and prudence margin at a finest level, or challenge the reserving guidelines.
Moreover, the use of individual reserving approach helps to prevent the limitation of aggregated methodologies, especially when the development of claims strongly differs. The use of an innovative micro reserving method like ASICR will help to consider each case’s specificities to estimate the reserves. ASICR provides an Automatic Segmentation of the claims database with the help of data science and projects the future claims development and an Individual Claims Reserving distribution within the helps of actuarial tools.
Reserving can therefore be approached in a different way, with a global and innovative process starting with a precise data analysis and ending with an individual estimate of its reserves. In particular, this enhances the use of data available to the insurer in order to help him define an appropriate risk management policy, in view of the risk profiles that could affect its profitability.