Skip to Main content Skip to Navigation

Sur l’utilisation des modèles multi-états pour la mesure et la gestion des risques d’un contrat d’assurance

Abstract : With the implementation of the Solvency II framework, actuaries should examine the good adequacy between models and data. This thesis aims to study several statistical approaches, often ignored by practitioners, enabling the use of multi-state methods to model and manage individual risks in insurance. Chapter 1 presents the general context of this thesis and positions its main contributions. The basic tools to use multi-state models in insurance are introduced and classical inference techniques, adapted to insurance data with and without the Markov assumption, are presented. Finally, a development of these models for credit risk is outlined. Chapter 2 focuses on using nonparametric inference methods to build incidence tables for long term care insurance contracts. Since there are several entry-causes in disability states which are useful for actuaries, an inference method for competing risks data, seen as a Markov multi-state model in continuous time, is used. In a second step, I compare these estimators to those conventionally used by practitioners, based on survival analysis methods. This second approach may involve significant bias because the interaction between entry-causes cannot be appropriately captured. In particular, these approaches assume that latent failure times are independent, while this hypothesis cannot be tested for competing risks data. Our approach allows to measure the error done by practitioners when they build incidence tables. Finally, a numerical application is considered on a long term care insurance dataset
Document type :
Complete list of metadatas

Cited literature [291 references]  Display  Hide  Download
Contributor : Abes Star :  Contact
Submitted on : Monday, March 14, 2016 - 5:15:06 PM
Last modification on : Friday, October 23, 2020 - 4:44:28 PM
Long-term archiving on: : Wednesday, June 15, 2016 - 2:50:19 PM


Version validated by the jury (STAR)


  • HAL Id : tel-01288199, version 1



Quentin Guibert. Sur l’utilisation des modèles multi-états pour la mesure et la gestion des risques d’un contrat d’assurance. Gestion et management. Université Claude Bernard - Lyon I, 2015. Français. ⟨NNT : 2015LYO10256⟩. ⟨tel-01288199⟩



Record views


Files downloads