Econometric Modelling and ARIMA Forecasting of GDP Growth, Inflation Dynamics, and Exchange Rate Volatility in Nigeria

Abstract

Macroeconomic forecasting is essential for economic planning, particularly in developing economies where growth instability, inflationary pressures, and exchange-rate uncertainty complicate decision-making. This study evaluates the adequacy and forecasting performance of econometric and Autoregressive Integrated Moving Average (ARIMA) models for Nigerian GDP growth, changes in inflation(DINF), and exchange-rate volatility (EXRV). Annual data covering 1990–2024 were analyzed. Dynamic regression models were estimated using 33 observations(1992–2024) to examine the relationships among lagged GDP growth, inflation changes, and exchange-rate volatility. Model adequacy was assessed using diagnostic tests for multicollinearity, heteroscedasticity, serial correlation, residual normality, and specification errors, with HC3 robust standard errors applied where necessary. The forecasting analysis employed ARIMA models estimated using the 1990–2019 training period and evaluated through fixed-origin out-of-sample validation for 2020–2024. Forecast accuracy was measured using MAE, RMSE, MAPE, and MASE. Results showed differences in model behaviour across indicators. The GDP growth regression displayed stronger within-model explanatory performance, with an adjusted R² of 0.327 and a significant HC3 robust joint test (p = 0.046); however, information criteria were interpreted only within individual model specifications and were not used for cross-model ranking. The selected ARIMA models were ARIMA(0,0,2) for GDP growth, ARIMA(0,1,0) with drift for inflation level forecasting, and ARIMA(0,0,0) for exchange-rate volatility. Validation results showed reduced forecast accuracy during periods of economic disruption. Forecasts for 2025–2029 are presented as conditional baseline projections with prediction intervals, reflecting uncertainty rather than guaranteed future outcomes. The study highlights the importance of combining diagnostic evaluation, forecast validation, and uncertainty assessment in Nigerian macroeconomic forecasting