An Empirical Time Series Modelling of Death Rate in Nigeria Using Box–Jenkins Approach: Evidence from Annual Data (1960–2025), Historical Trends and Future Projections
Nnaemeka Martin Eze *
Department of Statistics, University of Nigeria, Nsukka, Nigeria.
Oluchukwu Chukwuemeka Asogwa
Department of Mathematics & Statistics, Federal University Ndufu Alike, Nigeria.
Charity Uchenna Onwuamaeze
Department of Statistics, University of Nigeria, Nsukka, Nigeria.
*Author to whom correspondence should be addressed.
Abstract
This study examines the historical behaviour and future trajectory of Nigeria’s annual crude death rate using the Autoregressive Integrated Moving Average (ARIMA) model. The study employed annual secondary data on Nigeria’s crude death rate, measured as the number of deaths per 1,000 population, covering the period 1960–2025. The study contributes to the understanding of long-term changes in mortality patterns, provides an empirical assessment of the historical behaviour of Nigeria's crude death rate and shows the application of the Box–Jenkins ARIMA methodology to demographic research. The data were obtained from the Macrotrends demographic database and analysed using the Box–Jenkins methodology. Descriptive statistics, time series plots, the Augmented Dickey–Fuller (ADF), Phillips–Perron (PP), and Kwiatkowski–Phillips–Schmidt–Shin (KPSS) tests were used to examine the characteristics and stationarity of the series. The results indicated that the original series was non-stationary but became stationary after second-order differencing. Competing ARIMA specifications were subsequently estimated and compared using the Akaike Information Criterion (AIC), corrected Akaike Information Criterion (AICc), and Bayesian Information Criterion (BIC). Based on these criteria, ARIMA(1, 2, 1) was selected as the preferred model, with AIC, AICc, and BIC values of −181.56, −181.16, and −175.08, respectively. The estimated autoregressive and moving-average parameters were statistically significant at the 5% level. Residual diagnostic results, including the Kolmogorov–Smirnov and Shapiro–Wilk tests, provided no significant evidence against residual normality, while the Ljung–Box test (Q = 1.0791, p = 0.8980) indicated no significant residual autocorrelation at the tested lags. The forecasts suggest that Nigeria’s crude death rate may continue its downward trajectory, reaching approximately 8.28 deaths per 1,000 population by 2040. However, the widening prediction intervals indicate increasing uncertainty over the longer forecast horizon. The study concludes that ARIMA(1, 2, 1) provides a useful statistical structure for describing the historical pattern and generating conditional forecasts of Nigeria’s crude death rate. The findings may support demographic assessment, healthcare planning, and evidence-based public policy.
Keywords: Crude death rate, time series analysis, Box–Jenkins methodology, ARIMA model, forecasting