Figure 2 - available via license: CC BY
Content may be subject to copyright.
Three-year-ahead forecast from 2014-2016 for GB males aged (a) 50, (b) 60, (c) 70 and (d) 80 based on mortality data since 1950.
Source publication
Extrapolative methods are one of the most commonly-adopted forecasting approaches in the literature on projecting future mortality rates. It can be argued that there are two types of mortality models using this approach. The first extracts patterns in age, time and cohort dimensions either in a deterministic fashion or a stochastic fashion. The sec...
Context in source publication
Context 1
... comparison of the 3-, 5-and 10-year-ahead mortality forecasts for the four models against the real mortality experience for GB males aged 50, 60, 70 and 80 is illustrated in Figures 2-4. It can be seen from these plots that the 2D KS model outperformed the other three models in the majority of circumstances. ...
Citations
... This procedure is also known as an "evaluation on a rolling forecasting origin one-step ahead" [61]. In a particular application of LOOCV, Reference [62] compared the forecasting ability of parametric and nonparametric mortality models. ...
... This procedure is also known as an "evaluation on a rolling forecasting origin one-step ahead" [61]. In a particular application of LOOCV, Reference [62] compared the forecasting ability of parametric and nonparametric mortality models. ...
The accuracy of the predictions of age-specific probabilities of death is an essential objective for the insurance industry since it dramatically affects the proper valuation of their products. Currently, it is crucial to be able to accurately calculate the age-specific probabilities of death over time since insurance companies' profits and the social security of citizens depend on human survival; therefore, forecasting dynamic life tables could have significant economic and social implications. Quantitative tools such as resampling methods are required to assess the current and future states of mortality behavior. The insurance companies that manage these life tables are attempting to establish models for evaluating the risk of insurance products to develop a proactive approach instead of using traditional reactive schemes. The main objective of this paper is to compare three mortality models to predict dynamic life tables. By using the real data of European countries from the Human Mortality Database, this study has identified the best model in terms of the prediction ability for each sex and each European country. A comparison that uses cobweb graphs leads us to the conclusion that the best model is, in general, the Lee-Carter model. Additionally, we propose a procedure that can be applied to a life table database that allows us to choose the most appropriate model for any geographical area.
Currently, most academic research involving the mortality modeling of multiple populations mainly focuses on factor‐based approaches. Increasingly, these models are enriched with socio‐economic determinants. Yet these emerging mortality models come with little attention to interpretable spatial model features. Such features could be highly valuable to demographers and old‐age benefit providers in need of a comprehensive understanding of the impact of economic growth on mortality across space. To address this, we propose and investigate a family of models that extend the seminal Li‐Lee factor‐based stochastic mortality modeling framework to include both economic growth, as measured by the real gross domestic product (GDP), and spatial patterns of the contiguous United States mortality. Model selection performed on the introduced new class of spatial models shows that based on the AIC criteria, the introduced spatial lag of GDP with GDP (SLGG) model had the best fit. The out‐of‐sample forecast performance of SLGG model is shown to be more accurate than the well‐known Li–Lee model. When it comes to model implications, a comparison of annuity pricing across space revealed that the SLGG model admits more regional pricing differences compared to the Li‐Lee model.
