Cepar

You are here

Bayesian hierarchical multi-population mortality modelling for China's provinces

china

Qian Lu, Katja Hanewald and Xiaojun Wang

Abstract: China has experienced large improvements in mortality rates, but there remain substantial variations at the provincial level. This paper develops new models to project mortality at both the national and provincial levels in China. We propose two models in a Bayesian hierarchical framework based on principal components and a random walk process, and compile a new comprehensive database containing mortality data for 31 provinces over the period 1982–2010. The baseline two-level model with a national–province hierarchy allows for information pooling across provinces, common national factors and consistency conditions. The extended three-level model with a national–region–province hierarchy pools information in the region and also allows for common factors within the region. Both models provide good estimates and reasonable forecasts for China and its provinces. The baseline two-level model provides good fit and reasonable forecasts with equal width intervals for the provinces. The three-level model has a better fit with a lower deviance information criterion and provides forecast intervals reflecting regional uncertainty. The sensitivity analyses show that the forecasts are robust when changing the trend assumptions and regional groups.

Keywords: Mortality modelling, Bayesian framework, Hierarchical models, Coherent mortality projection, China

PDF icon Download (1.71 MB)