Common Stochastic Trends in European Mortality Levels: Testing and Consequences for Modeling Longevity Risk in Insurance

This paper highlights the long-term trends of general mortality levels in the European countries, investigates the existence of some common stochastic trends determining the mortality of the elderly population, and proposes to forecast the mortality rates taking into account these common stochastic trends. As a first step, a method for regionalization is used in order to provide homogeneous contiguous clusters. The tests for cointegration detect the number of common stochastic trends determining the mortality indices, derived from the Lee-Carter model, for countries from a homogeneous cluster; forecasts are generated by the vector error correction model. Based on out-of- sample forecasts, this approach leads to estimates of life expectancy near those provided by the Lee Carter model. In addition, using the VEC approach, the common stochastic trends, developed by the long-term mortality experience for countries from a cluster are preserved in the long run.