Comparative Study of Lee Carter and Arch Model in Modelling Female Mortality in Nigeria

Abstract

Using Nigeria mortality data from 2009 to 2020, this study compares and contrasts how well the Lee-Carter and ARCH models performed. Singular value decomposition (SVD) method, Langrage multiplier test, and autoregressive conditional heteroskedasticity (ARCH) effects were examined. Five (5) different ARIMA and ARCH models were fitted together with their criteria, i.e., AIC and BIC in order to determine the best model for Nigeria mortality data. ARIMA (0,1,0) had the lowest AIC and BIC values, and was determined to be the best ARIMA model. The mortality index is then modelled using ARIMA (0,1,0) and plugged back into the Lee-Carter model to predict the future mortality rate. Also ARCH (1) turned out to be the best ARCH model among other candidate models. The performance of Lee-Carter model and ARCH model was compared using error measures. Results obtained revealed that the ARCH model had the minimum RMSE and MAPE when compared with the Lee-carter model, therefore it was concluded that the ARCH model performs better than the Lee-Carter model on Nigeria mortality data.