This chapter illustrates how Machine learning techniques can be used to improve both fitting and forecasting of traditional stochastic mortality models to better understand processes that are not fully identifiable by standard models. we present a numerical application based on real mortality data of three European countries: we implement different mortality models in order to show how some ML techniques improve the fitting and modify the forecasts. Some of the forecasted mortality rates are then used to price two life insurance products whose payoffs depend on the future realized lifetime. The time profile of the actuarial reserves is shown to highlight the impact of longevity risk on such products.
Improving Longevity Risk Management through Machine Learning / Levantesi, Susanna; Nigri, Andrea; Piscopo, Gabriella. - I:(2021). [10.4324/9781003037903-3]
Improving Longevity Risk Management through Machine Learning
GABRIELLA PISCOPO
2021
Abstract
This chapter illustrates how Machine learning techniques can be used to improve both fitting and forecasting of traditional stochastic mortality models to better understand processes that are not fully identifiable by standard models. we present a numerical application based on real mortality data of three European countries: we implement different mortality models in order to show how some ML techniques improve the fitting and modify the forecasts. Some of the forecasted mortality rates are then used to price two life insurance products whose payoffs depend on the future realized lifetime. The time profile of the actuarial reserves is shown to highlight the impact of longevity risk on such products.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.