MODELING MONTHLY EXTERNAL RERSERVE IN NIGERIA: A SEASONAL AUTOREGRESSIVE INTEGRATED MOVING AVERAGE APPROACH

Abstract

Nigeria's external reserves serve as the nation's financial lifeline, shielding the economy from global shocks and maintaining monetary stability - yet their volatile nature makes accurate forecasting a persistent challenge. This study modelled Nigeria's monthly external reserves from January 1981 to December 2023 using Seasonal Autoregressive Integrated Moving Average (SARIMA) techniques. Through rigorous time series analysis, the study identified non-stationarity in the reserve data, which was successfully stabilized through first-order differencing. The comprehensive model selection process, guided by the Box-Jenkins methodology, revealed SARIMA(3,1,3)(2,1,2)₁₂ as the optimal configuration, demonstrating superior performance through minimized Akaike and Bayesian Information Criteria. The model's exceptional diagnostic results, including white noise residuals and high in-sample forecast accuracy (2018-2023), underscore its reliability. The out of sample forecast indicate a cautiously optimistic outlook, with reserves expected to gradually climb to 38,000-40,000 million USD by 2028. These findings equip Nigeria's monetary authorities with a powerful analytical tool for strategic reserve management, while highlighting opportunities for future research to incorporate machine learning approaches for enhanced predictive capability in this crucial economic domain.