Asymmetric uncertainty of mortality and longevity in the Spanish population

Abstract

Using data of specific mortality rates, discriminating between males and females, we estimate mortality and longevity risks for Spain in a period spanning from 1950 to 2012. We employ Dynamic Factor Models, fitted over the differences of the log-mortality rates to forecast mortality rates and we model the short-run dependence relationship in the data set by means of pair-copula constructions. We also compare the forecasting performance of our model with other alternatives in the literature, such as the well-known Lee-Carter Model. Finally, we provide estimations of risk measures such as VaR and Conditional-VaR for different hypothetical populations, which could be of great importance to assess the uncertainty faced by firms such as pension funds or insurance companies, operating in Spain. Our results indicate that mortality and longevity risks are asymmetric, especially in aged populations of males.