A General Procedure for Constructing Mortality Models

Abstract

This is the accepted version of the paper.This version of the publication may differ from the final published version. Amended 20 January 2015 Permanent repository link AbstractRecently, a large number of new mortality models have been proposed to analyse historical mortality rates and project them into the future. Many of these suffer from being over-parametrised or have terms added in an ad hoc manner which cannot be justified in terms of demographic significance. In addition, poor specification of a model can lead to period effects in the data being wrongly attributed to cohort effects which results in the model making implausible projections. We present a general procedure for constructing mortality models using a combination of a toolkit of functions and expert judgement. By following the general procedure, it is possible to identify sequentially every significant demographic feature in the data and give it a parametric structural form. We demonstrate using UK mortality data that the general procedure produces a relatively parsimonious model that nevertheless has a good fit to the data. * We are grateful to participants at a seminar at Cass

StMoMo: An R Package for Stochastic Mortality Modelling

“…This latter case could be extended to include the more general case of a pre-specified functional form with a set of parameters that need to be estimated, that is, β (i) x = f (i) (x; θ i ) with θ i being some model parameters. This generalisation would allow the implementation of the family of models considered in Hunt and Blake (2014).…”

Modelling longevity bonds: Analysing the Swiss Re Kortis bond

“…The first stage in assessing the Kortis bond is to construct suitable mortality models for England & Wales and the US. To do this, we apply the "general procedure" described in Hunt and Blake (2014) to data from the Human Mortality Database for both countries from 1950 to 2008 and for ages 50 to 100. We use data to 2008, to ensure comparability with the results of RMS, which would only have data available to 2008 when valuing the Kortis bond at issue in 2010.…”

Phantoms Never Die: Living with Unreliable Population Data

“…The format of the data varies from country to country. For example, in England & Wales, death counts, D(t, x) represents the number of persons who died in year t, 1 For example, Booth et al, 2002, andHyndman andUllah, 2007. 2 For example, Brouhns et al, 2002, Czado et al, 2005, Li et al, 2009 For example, Blake and Burrows, 2001, Coughlan et al, 2007, Cairns et al, 2008, Dahl et al, 2008, Li and Hardy, 2011, Li and Luo, 2012, Cairns 2011, 2013, 2014 4 For example, Cairns et al, 2006, 2009, Hyndman and Ullah, 2007, Plat, 2009, Currie, 2011, Hunt and Blake, 2014, and Mavros et al, 2014 5 For example, Willets, 2004, Renshaw and Haberman, 2006, Cairns et al 2009, 2011a For example, Li and Lee, 2005, Cairns et al, 2011b, Dowd et al, 2011, Jarner and Kryger, 2011, Li and Hardy, 2011, and Börger et al, 2014 7 Cairns (2014) has charted the genealogy of these new models. He argues that the accompanying complexity might not deliver improved forecasts, and that there is a need to get back to simpler, more robust models (see, also, Mavros et al, 2014, Hunt andBlake, 2014).…”