MODELING AND FORECASTING INFANT DEATHS IN ZIMBABWE USING ARIMA MODELS

Abstract

Employing annual time series data on total infant deaths in Zimbabwe from 1960 to 2018, the study models and forecasts total infant deaths over the next decade using ARIMA models. Diagnostic tests indicate that the Zimbabwe annual infant deaths series is an I (2) variable. Guided by Mishra et al. (2019), the study uses the “minimum AIC criteria” to select the optimal model, the ARIMA (1, 2, 5) model. The ADF test of the residuals, the correlogram of the residuals as well as the inverse roots of the AR/MA polynomials; all indicate that the presented model is stable and suitable for forecasting annual infant deaths in Zimbabwe. The study, whose results are not surprising, indicates that the number of infant deaths per year, over the out-of-sample period, will follow a downward trend. A five-fold policy implication has been put forward in order to reduce infant deaths in the country.