Forecasting of Male Life Expectancy in Nigeria: Box-Jenkins Approach

Abstract

Over the past few decades, Males’ life expectancy has been observed to be less than that of females. In the literature, little has been done to forecast the Life expectancy at birth for Males (LEM) in developing countries, especially in Nigeria. To understand the forecasting dynamics of LEM in Nigeria, this work therefore examines the series using the Box-Jenkins (1976) methodology. Pre-examination tests conducted on the series revealed that it is not only a difference stationary process of order one {I(1)} but it is also best fitted by ARIMA (1, 1, 1) model. Moreover, diagnostic checking carried out on the residuals of the fitted model confirmed that the residual is a white noise error term in that it fulfils the assumptions of homoscedasticity, stationarity, and autocorrelation. To obtain a good forecast, the LEM data spanning 1960 to 2014 (55 years) were used to train the model while the remaining LEM data spanning 2015 to 2020 (6 years) were used to test the data. For the test data, the forecasted LEM has been shown to give approximate values of the actual LEM. Findings showed that the LEM is expected to increase slightly by 0.64% within the forecasted periods spanning 2021 to 2030. Diagnostic checking, Forecasting, Homoscedasticity, Life expectancy, Male, Training sets, Testing sets