A multi-developing-country comparison of the lee-carter model for mortality and life expectancy projections

Abstract

This study evaluate the Lee-Carter model for projecting age and gender specific mortality rates and life expectancy at births Thailand and Singapore. Although the Lee many developed countries, little information is obtained on the suitability of the model for developing countries. We fit age and gender specific mo country from 1960 to 1989 into the model and forecast over the evaluation period from 1990 to 2015. The forecast values are then compared with actual rates over the same period to estimate the out-sample forecast errors. Th that the Lee-Carter model works better in Malaysia, Indonesia and Thailand, than in Singapore. In terms of life expectancy at births, the Lee accurate for females than that of males. Carter model for projecting age and gender specific mortality rates and life expectancy at births in four developing countries namely Malaysia, Indonesia, Thailand and Singapore. Although the Lee-Carter model has been successfully applied in many developed countries, little information is obtained on the suitability of the model for developing countries. We fit age and gender specific mortality rates of each country from 1960 to 1989 into the model and forecast over the evaluation period from 1990 to 2015. The forecast values are then compared with actual rates over the same period to sample forecast errors. The mean forecast errors of log mortality rates show Carter model works better in Malaysia, Indonesia and Thailand, than in Singapore. In terms of life expectancy at births, the Lee-Carter model consistently more f males. life expectancy; Lee-Carter model; population ageing. , Selangor, Malaysia Carter model for projecting age and gender specific mortality Malaysia, Indonesia, Carter model has been successfully applied in many developed countries, little information is obtained on the suitability of the model for rtality rates of each developing country from 1960 to 1989 into the model and forecast over the evaluation period from 1990 to 2015. The forecast values are then compared with actual rates over the same period to e mean forecast errors of log mortality rates show Carter model works better in Malaysia, Indonesia and Thailand, than in Carter model consistently more