CBDX: A Workhorse Mortality Model from the Cairns-Blake-Dowd Family

Kevin Dowd, Andrew John George Cairns, David Blake

Research output: Contribution to journalArticle

Abstract

The purpose of this paper is to identify a workhorse mortality model for the adult age range (i.e., excluding the accident hump and younger ages). It applies the “general procedure” (GP) of Hunt and Blake (2014) to identify an age-period model that fits the data well before adding in a cohort effect that captures the residual year-of-birth effects arising in the original age-period model. The resulting model is intended to be suitable for a variety of populations, but economises on the number of period effects in comparison with a full implementation of the GP. We estimate the model using two different iterative Maximum Likelihood (ML) approaches – one Partial ML and the other Full ML – that avoid the need to specify identifiability constraints.
Original languageEnglish
JournalAnnals of Actuarial Science
Publication statusAccepted/In press - 21 Jan 2020

Fingerprint

Mortality
Maximum likelihood
Identifiability
Accidents
Cohort effect

Keywords

  • mortality rates
  • Cairns-Blake-Dowd mortality model
  • CBDX mortality model

Cite this

@article{407f04cd597e49b899da18dba2a6d8ab,
title = "CBDX: A Workhorse Mortality Model from the Cairns-Blake-Dowd Family",
abstract = "The purpose of this paper is to identify a workhorse mortality model for the adult age range (i.e., excluding the accident hump and younger ages). It applies the “general procedure” (GP) of Hunt and Blake (2014) to identify an age-period model that fits the data well before adding in a cohort effect that captures the residual year-of-birth effects arising in the original age-period model. The resulting model is intended to be suitable for a variety of populations, but economises on the number of period effects in comparison with a full implementation of the GP. We estimate the model using two different iterative Maximum Likelihood (ML) approaches – one Partial ML and the other Full ML – that avoid the need to specify identifiability constraints.",
keywords = "mortality rates, Cairns-Blake-Dowd mortality model, CBDX mortality model",
author = "Kevin Dowd and Cairns, {Andrew John George} and David Blake",
year = "2020",
month = "1",
day = "21",
language = "English",
journal = "Annals of Actuarial Science",
issn = "1748-4995",
publisher = "Cambridge University Press",

}

CBDX: A Workhorse Mortality Model from the Cairns-Blake-Dowd Family. / Dowd, Kevin; Cairns, Andrew John George; Blake, David.

In: Annals of Actuarial Science, 21.01.2020.

Research output: Contribution to journalArticle

TY - JOUR

T1 - CBDX: A Workhorse Mortality Model from the Cairns-Blake-Dowd Family

AU - Dowd, Kevin

AU - Cairns, Andrew John George

AU - Blake, David

PY - 2020/1/21

Y1 - 2020/1/21

N2 - The purpose of this paper is to identify a workhorse mortality model for the adult age range (i.e., excluding the accident hump and younger ages). It applies the “general procedure” (GP) of Hunt and Blake (2014) to identify an age-period model that fits the data well before adding in a cohort effect that captures the residual year-of-birth effects arising in the original age-period model. The resulting model is intended to be suitable for a variety of populations, but economises on the number of period effects in comparison with a full implementation of the GP. We estimate the model using two different iterative Maximum Likelihood (ML) approaches – one Partial ML and the other Full ML – that avoid the need to specify identifiability constraints.

AB - The purpose of this paper is to identify a workhorse mortality model for the adult age range (i.e., excluding the accident hump and younger ages). It applies the “general procedure” (GP) of Hunt and Blake (2014) to identify an age-period model that fits the data well before adding in a cohort effect that captures the residual year-of-birth effects arising in the original age-period model. The resulting model is intended to be suitable for a variety of populations, but economises on the number of period effects in comparison with a full implementation of the GP. We estimate the model using two different iterative Maximum Likelihood (ML) approaches – one Partial ML and the other Full ML – that avoid the need to specify identifiability constraints.

KW - mortality rates

KW - Cairns-Blake-Dowd mortality model

KW - CBDX mortality model

M3 - Article

JO - Annals of Actuarial Science

JF - Annals of Actuarial Science

SN - 1748-4995

ER -