<b>Impact of exogenous variables on paddy productivity modeling: an ARIMAX model perspective</b>

Paddy productivity in Malaysia faces escalating difficulties in 2024–2025 due to increasing input costs, unpredictable weather patterns, and rising dependence on rice imports. These problems have intensified the necessity for precise forecasting instruments to inform food security policies and agricultural strategies. The present study examined the effectiveness of ARIMA and ARIMAX models in predicting yearly paddy productivity in the states in the northern region of Peninsular Malaysia, which were Kedah, Perak, Pulau Pinang, and Perlis, based on data from 1981 to 2022. The study investigated whether the inclusion of an exogenous factor, which was planted area, may improve forecasting accuracy beyond conventional univariate models. A systematic approach was employed involving stationarity assessment, model identification, residual diagnostics, and performance evaluation. The results indicated that ARIMAX modeling’s performance consistently surpassed that of ARIMA modeling, with Kedah achieving the best accuracy, closely followed by Perlis, Perak, and Pulau Pinang. The incorporation of exogenous variables markedly enhanced model responsiveness and accounted for structural changes in paddy production. The residuals from the final model for each state exhibited no indications of autocorrelation, hence confirming statistical validity. The study finds that ARIMAX modeling offers a dependable and comprehensible forecasting framework for paddy output and can function as an essential decision-support instrument for agricultural policies, particularly during times of supply instability. The methodology is versatile for many crops and geographies, facilitating extensive applications in agricultural forecasting and food security strategies.