Modeling Full Mortality Schedule from Child Mortality: A Flexible Model Life Table Approach Based on Similar Mortality Profiles

Tianyu Shen, Australian National University
Xueqing Wang, Princeton University, office of Population Research
Ugofilippo Basellini, Max Planck Institute for demographic Research
Monica Alexander, University of Toronto
Irena Chen, Max Planck Institute for Demographic Research (MPIDR)

In countries lacking vital registration systems, granular age-specific mortality rates for all age groups may be derived from model life tables. In such settings, child mortality is often available from targeted surveys. Existing methods typically draw on the fundamental relationship between child mortality and mortality at other ages to predict the latter when only the former is known. In this paper, we show that child mortality becomes a weaker predictor of mortality at both prime adult ages, i.e. 20-45, and older ages, i.e. 70+. We improve the performance of existing models by bringing in auxiliary information: population change. Drawing from decennial censuses readily available even in data-scarce contexts, population change contains useful information about a population’s mortality schedule. We propose an innovative approach that matches the target country with the observed mortality schedules based on similarity in population change before prediction. We demonstrate that this method improves predictions of adult mortality.

Keywords: Data and Methods, Mortality and Longevity

See extended abstract.