The Performance of The Lee-Carter Model on Heterogeneous Adult Mortality Data in a Limited Data Situation

Abstract

Adult mortality in developing Countries remains one of the greatest challenges for monitoring, analyzing and projecting the health situation of a large proportion of the world's population. Because of limited resources, many developing Countries, especially in sub-Saharan Africa lack vital registration systems that could reliably and continuously collect information on adult mortality. However, periodic reviews of mortality rate and pattern using mortality models could show new trends and may provide information for the planning of a Country’s health care delivery services and a host of other programs.

Previous studies have found that the Lee-Carter model works well with homogenous mortality data. In this current study, we want to see the performance of the model on heterogenous adult mortality data in a limited data situation. A modified version of the Lee-Carter method is used to model sex-combined adult mortality data of Nigerians aged 15-84 years for the time periods 1990, 2000 and 2012. The model's parameters are estimated using the approach proposed by Lee and Carter (1992) based on the singular value decomposition technique, while the mortality index is predicted using the approach developed by Li et al. (2002). Our results show that on the overall, the model follows the mortality pattern very well for most of the ages despite the heterogonous nature of the data used. Forecast values of the mortality index show a gradual decline in mortality from 2013-2025 in Nigeria.

KEYWORDS: Adult mortality; Age; Lee-Carter model; Nigeria.