In non-life insurance, modelling based on aggregate data is still common practice when estimating claims reserves. As a result, valuable information, such as individual claim features, is disregarded even though both the relevant data and the computational capacity for more complex modelling are often available to insurers. In this thesis, we present and compare five recently proposed models aimed at utilizing more than the classical aggregate data, expanding on conventional methods to varying degrees. To gain a deeper understanding of these models’ functionality and limitations, they are applied to a synthetic data set generated using the individual claims history simulation machine of Gabrielli and Wüthrich. Additionally, we examine the real-world applicability and challenges of individual claims modelling through a survey of several non-life insurance companies’ actuarial departments and interpret its results with regard to potentially relevant aspects for advancing this area of research.     «

In non-life insurance, modelling based on aggregate data is still common practice when estimating claims reserves. As a result, valuable information, such as individual claim features, is disregarded even though both the relevant data and the computational capacity for more complex modelling are often available to insurers. In this thesis, we present and compare five recently proposed models aimed at utilizing more than the classical aggregate data, expanding on conventional methods to varying...     »