Claim Reservation for Long Tailed Insurance Business

Specialeforsvar ved Arashk Jasemi

Titel: Claim Reservation for Long Tailed Insurance Business

Abstract:  In this thesis we wish to study the problem of claim reservation for long tailed insurance business. Here we focus on workers' compensation which is known for having a long settlement period and a very high development in accumulated claims. We work with three data sets from different real life insurance companies. We will study the classical Chain-ladder method and a Bayesian approach to the Chain-ladder method. For the Bayesian approach we will use the data from companies 2 and 3 to predict future claims for company 1. The main focus is to highlight the qualities and advantages of the Bayesian approach and study its parameters in order to see how they effect the model and hence the prediction of the future claims. We will also discuss the reason why it may be preferable to apply a priori development factors as an alternative to the Chain-ladder factors. Both models will be used to predict future claims and calculate the mean squared error of prediction in order to calculate the solvency margins for the company. Lastly, we will look at the solvency capital requirement for the company in both cases and discuss the difference. 

Vejleder:  Jostein Paulsen
Censor:    Mette M Havning