Compositional Data Analysis – Coherent Forecasting Mortality Model with Cohort Effect

In this paper, an extension of the Coherent forecasts of mortality with compositional data analysis (CoDa) model of Bergeron-Boucher et al. (2017) to cohort effect is proposed applied to data from six African countries. The process of fitting this model starts by adapting the Renshaw and Haberman (2006) to compositional data analysis (CODA) as suggested by Bergeron-Boucher et al. (2017). The proposed CoDa-cohort model generally fits the data better than the original cohort model of Renshaw and Haberman (2006). To get the full CoDa-cohort-coherent model the multiple population factor is included in CoDa-cohort model. Then a comparison between CoDa -coherent and CoDa-cohort-coherent models revealed that they have similar accuracy for the selected countries in West Africa but not for countries in East Africa based on Aitchinson distance (AD). But for merged populations like male and female, the new model, CoDa-cohort-coherent, has generally better fits for Kenya mortality data.