Abstract
SummaryForecasts of mortality provide vital information about future populations, with implications for pension and healthcare policy as well as for decisions made by private companies about life insurance and annuity pricing. The paper presents a Bayesian approach to the forecasting of mortality that jointly estimates a generalized additive model (GAM) for mortality for the majority of the age range and a parametric model for older ages where the data are sparser. The GAM allows smooth components to be estimated for age, cohort and age-specific improvement rates, together with a non-smoothed period effect. Forecasts for the UK are produced by using data from the human mortality database spanning the period 1961–2013. A metric that approximates predictive accuracy is used to estimate weights for the ‘stacking’ of forecasts from models with different points of transition between the GAM and parametric elements. Mortality for males and females is estimated separately at first, but a joint model allows the asymptotic limit of mortality at old ages to be shared between sexes and furthermore provides for forecasts accounting for correlations in period innovations.
Highlights
The future level of mortality is of vital interest to policy makers and private insurers alike, as lower mortality results in greater expenditure on pension payments and higher social care spending
The work of Dodd et al (2018a) in producing the 17th iteration of the English Life Tables provided a methodology for mortality estimation that combines smoothing based on generalized additive models (GAMs) (Wood, 2006) at the youngest ages with a parametric model at older ages
This paper extends this approach to a forecasting context and introduces period and cohort effects, producing fully probabilistic mortality projections within a Bayesian framework
Summary
The future level of mortality is of vital interest to policy makers and private insurers alike, as lower mortality results in greater expenditure on pension payments and higher social care spending. Individuals are living longer because of improved mortality conditions and will reach higher ages in greater number as the post-war baby boom cohort ages, and forecasts of mortality at the oldest ages are becoming more important. The work of Dodd et al (2018a) in producing the 17th iteration of the English Life Tables provided a methodology for mortality estimation that combines smoothing based on generalized additive models (GAMs) (Wood, 2006) at the youngest ages with a parametric model at older ages This paper extends this approach to a forecasting context and introduces period and cohort effects, producing fully probabilistic mortality projections within a Bayesian framework
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
