The Stratified Sampling Bootstrap for Measuring the Uncertainty in Mortality Forecasts |
| |
Authors: | Valeria D’Amato Steven Haberman Maria Russolillo |
| |
Institution: | (1) Institut de Statistique & Institut des Sciences Actuarielles, Universit? Catholique de Louvain, Louvain-la-Neuve, Belgium |
| |
Abstract: | In this paper, we propose a procedure for reducing the uncertainty in mortality projections, on the basis of a log bilinear
Poisson Lee Carter model (Renshaw and Haberman Appl Stat 52:119–137, 2003a). In the literature, because the non-linear nature of the quantities under consideration has prevented analytical solutions,
simulation techniques have been used in order to provide prediction intervals for forecasted quantities (for example, Brouhns
et al. Scand Actuar J 3:212–224, 2005, Renshaw and Haberman Insur Math Econ 42:797–816, 2008). In this respect, we adopt the bootstrap simulation approach in order to measure the uncertainty affecting mortality projections.
In particular, we propose making the bootstrap procedure more efficient by using a specific variance reducing technique, the
so-called Stratified Sampling technique. To this end, we propose a two stage simulation bootstrap procedure where variance
reducing techniques are combined with the simple bootstrap of the Poisson Lee Carter version. Numerical applications are shown
using the results for some datasets. |
| |
Keywords: | |
本文献已被 SpringerLink 等数据库收录! |
|