Stochastic mortality using non - life methods

Although longevity and mortality risks have been studied for a long time, there has been a remarkable increase in the number and scope of these studies due to recent theoretical and practical developments. Such developments paved the way for a more detailed analysis of the asset-liability balance of life insurance and pension companies. The impact of these risks is huge albeit long-term. In addition to life insurance and pension companies, governments are also considerably affected through health expenditures and pension payments. Further, environmental impact on food and water supplies and pollution should also be considered as well as financial ramifications. Hence, the accurate modelling and forecasting of longevity and mortality risks has become more important than ever as the accelerating population increase is taking its toll on mankind both financially and environmentally. This thesis focuses on the most widely used stochastic mortality models. The results pertaining to historical death probability, force of mortality, life expectancy, and rectangularization behavior are analysed in detail. These models are applied to the data from 20 different countries. In terms of the variety of stochastic mortality models and the number of different countries, this study is singled out as the most extensive study to date. The main contribution of this thesis is the introduction of a new approach to model mortality. This approach is based on IBNR calculations in non-life insurance. The comparison of this approach with two extensively used models in practice, namely Lee and Carter and Renshaw and Haberman models, shows that the new approach outperforms the aforementioned models with UK data.