Approximation de la réserve d'une compagnie d'assurance par un processus de diffusion et étude de quelques indicateurs de risque
|Abstract:||Risk management is an area that continues to evolve each year. Indeed, several models are built to model the wealth of an insurance company and follow its behavior over time. One of the targets of this modeling is to provide risk indicators that give visibility about the company’s situation and help the company’s managers make the necessary decisions. The majority of models rely on the composed Poisson processes and consider the number and time of sinisters. We propose in this thesis a new stochastic model based on stochastic differential equation for risk management. It is a reserve approximation model obtained by a diffusion process. In this model we do not take into account the number or the instants of sinisters, we only take into account the total of losses and of incomes together with the growth of each business line. Some risk indicators are also defined and adjusted according to our model. We consider then a multidimensional risk process, where each component of the vector is the reserve process for one line of business for the company. We assume the independence between the different lines to facilitate the modelling. Finally, we propose a simulation study using an Euler-Maruyama scheme coupled to a Monte- Carlo method. Then, we explain the behavior of each line and we compute the approximation of some risk indicators. The findings of the numerical study support the conclusion that our method works and provide good results. With regard to the numerical results, it can be concluded that the initial capital has a great role and can in some cases save the company’s situation. Moreover, the threshold level that has been introduced into the model is also very important for the insurance company’s health.|
|Document Type:||Mémoire de maîtrise|
|Open Access Date:||12 July 2019|
|Collection:||Thèses et mémoires|
All documents in CorpusUL are protected by Copyright Act of Canada.