An introduction to gevistic regression mortality models

Research output: Contribution to journalJournal articleResearchpeer-review

Abstract

Many common stochastic mortality models can be formulated as a generalized linear model (GLM). When these GLMs are used to model one year-death probabilities, qx, deaths are assumed to be binomially distributed, and the canonical logit link function has been used by default. In this work we present the quantile function of the Generalized Extreme Value distribution as an alternative link function to the standard canonical logit link and show that its theoretical advantages enable a better fit for mortality models in cases when data are highly imbalanced or sparse. We provide an example that shows that this link function also enables superior fits to mortality data at the very highest ages in the case of the Cairns Blake Dowd family of mortality models.
Original languageEnglish
JournalScandinavian Actuarial Journal
Volume2019
Issue number7
Pages (from-to)604-620
ISSN0346-1238
DOIs
Publication statusPublished - 9. Aug 2019

Fingerprint

Mortality
Link Function
Regression
Logit
Generalized Extreme Value Distribution
Quantile Function
Generalized Linear Model
Model
Alternatives

Keywords

  • GEV distribution
  • Stochastic mortality models
  • gevit link
  • high age mortality
  • link functions
  • logit link

Cite this

@article{8cb6a3d377924dc296a5c9a192c8f07b,
title = "An introduction to gevistic regression mortality models",
abstract = "Many common stochastic mortality models can be formulated as a generalized linear model (GLM). When these GLMs are used to model one year-death probabilities, qx, deaths are assumed to be binomially distributed, and the canonical logit link function has been used by default. In this work we present the quantile function of the Generalized Extreme Value distribution as an alternative link function to the standard canonical logit link and show that its theoretical advantages enable a better fit for mortality models in cases when data are highly imbalanced or sparse. We provide an example that shows that this link function also enables superior fits to mortality data at the very highest ages in the case of the Cairns Blake Dowd family of mortality models.",
keywords = "GEV distribution, Stochastic mortality models, gevit link, high age mortality, link functions, logit link",
author = "Anthony Medford and Vaupel, {James W.}",
year = "2019",
month = "8",
day = "9",
doi = "10.1080/03461238.2019.1586756",
language = "English",
volume = "2019",
pages = "604--620",
journal = "Scandinavian Actuarial Journal",
issn = "0346-1238",
publisher = "Taylor & Francis",
number = "7",

}

An introduction to gevistic regression mortality models. / Medford, Anthony; Vaupel, James W.

In: Scandinavian Actuarial Journal, Vol. 2019, No. 7, 09.08.2019, p. 604-620.

Research output: Contribution to journalJournal articleResearchpeer-review

TY - JOUR

T1 - An introduction to gevistic regression mortality models

AU - Medford, Anthony

AU - Vaupel, James W.

PY - 2019/8/9

Y1 - 2019/8/9

N2 - Many common stochastic mortality models can be formulated as a generalized linear model (GLM). When these GLMs are used to model one year-death probabilities, qx, deaths are assumed to be binomially distributed, and the canonical logit link function has been used by default. In this work we present the quantile function of the Generalized Extreme Value distribution as an alternative link function to the standard canonical logit link and show that its theoretical advantages enable a better fit for mortality models in cases when data are highly imbalanced or sparse. We provide an example that shows that this link function also enables superior fits to mortality data at the very highest ages in the case of the Cairns Blake Dowd family of mortality models.

AB - Many common stochastic mortality models can be formulated as a generalized linear model (GLM). When these GLMs are used to model one year-death probabilities, qx, deaths are assumed to be binomially distributed, and the canonical logit link function has been used by default. In this work we present the quantile function of the Generalized Extreme Value distribution as an alternative link function to the standard canonical logit link and show that its theoretical advantages enable a better fit for mortality models in cases when data are highly imbalanced or sparse. We provide an example that shows that this link function also enables superior fits to mortality data at the very highest ages in the case of the Cairns Blake Dowd family of mortality models.

KW - GEV distribution

KW - Stochastic mortality models

KW - gevit link

KW - high age mortality

KW - link functions

KW - logit link

U2 - 10.1080/03461238.2019.1586756

DO - 10.1080/03461238.2019.1586756

M3 - Journal article

VL - 2019

SP - 604

EP - 620

JO - Scandinavian Actuarial Journal

JF - Scandinavian Actuarial Journal

SN - 0346-1238

IS - 7

ER -