Coherent modeling of male and female mortality using Lee–Carter in a complex number framework

Research output: Contribution to journalArticleResearchpeer-review

Abstract

Forecasts of female and male mortality that are conducted independently run the risk of projecting implausible sex differentials and fail to exploit correlations that are known to exist between the sexes. We present a new model for the simultaneous modeling of female and male mortality. The model casts mortality as a complex-valued process where the real and imaginary parts correspond to female and male mortalities, respectively. Calculations proceed similarly to the usual Lee–Carter model, via the singular value decomposition, albeit in complex form. Initial applications suggest that the complex Lee–Carter gives fits that are broadly comparable to independent real fits, while offering the advantage of explicit modeling of the relationship between the sexes. Furthermore, model parameters are informative and easily-interpretable.

LanguageEnglish
Pages130-137
Number of pages8
JournalInsurance: Mathematics and Economics
Volume71
DOIs
Publication statusPublished - 1 Nov 2016

Fingerprint

Complex number
Mortality
Modeling
Singular value decomposition
Model
Forecast
Framework

Keywords

  • Mortality forecasting
  • Complex-valued process
  • Coherent forecasts
  • Lee-Carter model
  • Life expectancy
  • COMMON FACTOR MODEL
  • LIFE EXPECTANCY
  • POPULATIONS
  • PROJECTION
  • EXTENSION
  • COUNTRIES
  • DECLINE
  • PATTERN
  • TRENDS
  • RISK

Cite this

@article{2806510813104a6996426b92b6dbd2d5,
title = "Coherent modeling of male and female mortality using Lee–Carter in a complex number framework",
abstract = "Forecasts of female and male mortality that are conducted independently run the risk of projecting implausible sex differentials and fail to exploit correlations that are known to exist between the sexes. We present a new model for the simultaneous modeling of female and male mortality. The model casts mortality as a complex-valued process where the real and imaginary parts correspond to female and male mortalities, respectively. Calculations proceed similarly to the usual Lee–Carter model, via the singular value decomposition, albeit in complex form. Initial applications suggest that the complex Lee–Carter gives fits that are broadly comparable to independent real fits, while offering the advantage of explicit modeling of the relationship between the sexes. Furthermore, model parameters are informative and easily-interpretable.",
keywords = "Mortality forecasting, Complex-valued process, Coherent forecasts, Lee-Carter model, Life expectancy, COMMON FACTOR MODEL, LIFE EXPECTANCY, POPULATIONS, PROJECTION, EXTENSION, COUNTRIES, DECLINE, PATTERN, TRENDS, RISK",
author = "{de Jong}, Piet and Leonie Tickle and Jianhui Xu",
year = "2016",
month = "11",
day = "1",
doi = "10.1016/j.insmatheco.2016.08.003",
language = "English",
volume = "71",
pages = "130--137",
journal = "Insurance: Mathematics and Economics",
issn = "1873-5959",
publisher = "Elsevier",

}

Coherent modeling of male and female mortality using Lee–Carter in a complex number framework. / de Jong, Piet; Tickle, Leonie; Xu, Jianhui.

In: Insurance: Mathematics and Economics, Vol. 71, 01.11.2016, p. 130-137.

Research output: Contribution to journalArticleResearchpeer-review

TY - JOUR

T1 - Coherent modeling of male and female mortality using Lee–Carter in a complex number framework

AU - de Jong,Piet

AU - Tickle,Leonie

AU - Xu,Jianhui

PY - 2016/11/1

Y1 - 2016/11/1

N2 - Forecasts of female and male mortality that are conducted independently run the risk of projecting implausible sex differentials and fail to exploit correlations that are known to exist between the sexes. We present a new model for the simultaneous modeling of female and male mortality. The model casts mortality as a complex-valued process where the real and imaginary parts correspond to female and male mortalities, respectively. Calculations proceed similarly to the usual Lee–Carter model, via the singular value decomposition, albeit in complex form. Initial applications suggest that the complex Lee–Carter gives fits that are broadly comparable to independent real fits, while offering the advantage of explicit modeling of the relationship between the sexes. Furthermore, model parameters are informative and easily-interpretable.

AB - Forecasts of female and male mortality that are conducted independently run the risk of projecting implausible sex differentials and fail to exploit correlations that are known to exist between the sexes. We present a new model for the simultaneous modeling of female and male mortality. The model casts mortality as a complex-valued process where the real and imaginary parts correspond to female and male mortalities, respectively. Calculations proceed similarly to the usual Lee–Carter model, via the singular value decomposition, albeit in complex form. Initial applications suggest that the complex Lee–Carter gives fits that are broadly comparable to independent real fits, while offering the advantage of explicit modeling of the relationship between the sexes. Furthermore, model parameters are informative and easily-interpretable.

KW - Mortality forecasting

KW - Complex-valued process

KW - Coherent forecasts

KW - Lee-Carter model

KW - Life expectancy

KW - COMMON FACTOR MODEL

KW - LIFE EXPECTANCY

KW - POPULATIONS

KW - PROJECTION

KW - EXTENSION

KW - COUNTRIES

KW - DECLINE

KW - PATTERN

KW - TRENDS

KW - RISK

UR - http://www.scopus.com/inward/record.url?scp=84988519849&partnerID=8YFLogxK

U2 - 10.1016/j.insmatheco.2016.08.003

DO - 10.1016/j.insmatheco.2016.08.003

M3 - Article

VL - 71

SP - 130

EP - 137

JO - Insurance: Mathematics and Economics

T2 - Insurance: Mathematics and Economics

JF - Insurance: Mathematics and Economics

SN - 1873-5959

ER -