Title: State space models for stochastic claims reserving
Authors: Radek Hendrych - Charles University (Czech Republic) [presenting]
Abstract: The actuarial loss (claims) reserves represent estimates of insurers liabilities from all future claims obligations arising from insurance contracts currently in force and written in the past. There exist various actuarial (statistical) methods for calculation of outstanding claims reserves. The aim is to introduce and compare various methodological approaches to IBNR (incurred but not yet reported) claims reserving based on state space models and Kalman filter algorithms. In particular, if one transforms available claims data from the form of run-off triangles to the form of (multivariate) time series with missing observations, various state space models might be employed to project and/or interpolate IBNR claims reserves. Additionally, useful extensions of the loss reserving problem (e.g. dependent run-off triangles for correlated business lines or outliers in claims data) might be implemented when applying such a methodological framework. Results of numerical study for various claims data (univariate and multivariate ones) are presented.