Models for population-wide and portfolio-specific mortality

The life expectancy is a statistic that depends on forecasted human mortality rates. The last two centuries it has increased substantially, and it is expected that it will increase even further in the future. Pension funds and insurance companies have made obligations to participants respectively policyholders to make (pension) payments that often occur far in the future and which depend on survivorship or death. It is therefore crucial for these institutions that they accurately estimate how mortality rates will develop in the future to ensure that they hold sufficient capital to meet all future obligations. For good risk management, it is also important to understand the uncertainty associated with the developments in mortality rates. Moreover, people with a different socioeconomic background and different life style experience different levels of mortality, and this may result in large differences in life expectancy. Such differences can be found in the whole population but they also exist within a pension fund or insurance portfolio, and this should be taken into account when valuing liabilities. This thesis develops models to obtain better estimates of mortality developments and the uncertainty associated with these developments, and a framework in which adjustments to population mortality rates can be estimated on an individual level. This helps pension funds and insurance companies to value their liabilities more accurately.