Univariate and Multivariate ARIMA Models for Age-Specific Mortality rates

Document Type : Original Article

Authors

Faculty of commerce- Helwan University

Abstract

Getting  the forecasted values  of the  age-specific  mortality  rates  (ASMR)for  each sex is  very  important  to  build  population  projections. Either  population  projections  are  deterministic  or  probabilistic . the question is how to get the forecasted values of ASMR. One possible technique is to obtain ASMR through the United Nations model life tables. This technique is the common technique in most studies for Egypt. Another possible suggested technique is modeling ASMR themselves. This paper mainly examines the question of which one of the two pervious  techniques would provide us with more accurate results . this question is considered empirically based on integrated Autoregressive Moving  Average (ARIMA) models using annual data on Egypt during the period from 1966 to 2010. This comparative study is conducted between : (a) (b) multivariate ARIMA model to the vector of ASMR for each sex separately. The results of the accuracy measures indicate that multivariate ARIMA model has the ability to clearly represent ASMR. Consequently, multivariate ARIMA model id used for forecasting ASMR in  Egypt  until year 2030.

Keywords