Comparative evaluation of ARFIMA and SARIMA models for forecasting rice production in Nigeria

Accurate forecasting of agricultural output is critical in ensuring effective planning for food security in Nigeria. This study compares and evaluates the forecasting efficiency of the Autoregressive Fractionally Integrated Moving Average (ARFIMA) and Seasonal Autoregressive Integrated Moving Average (SARIMA) models in forecasting rice production in Nigeria. Annual rice production data from 1960 to 2024 were collected from the Central Bank of Nigeria Statistical Bulletin and subjected to time series econometric analysis. The data series was checked for stationarity using the Augmented Dickey-Fuller test before estimating both ARFIMA and SARIMA models. The performance of both models was evaluated using forecast accuracy measures such as Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE). The study found that the SARIMA (1,1,1) model provided a better fit and forecasting efficiency compared to the ARFIMA model in forecasting rice production in Nigeria. The SARIMA model provided lower forecast error values (RMSE = 0.04476; MAE = 0.03842) and a lower MAPE of 1.047%, indicating a forecasting accuracy of 98.95%. The study concluded that rice production in Nigeria is better forecasted using short-memory seasonal structures of the SARIMA (1,1,1) model than using fractional long-memory processes of the ARFIMA model. The study recommends the use of the SARIMA (1,1,1) model for effective forecasting of rice production to support agricultural planning and policy formulation.