Predicting food security through sustainable water in Saudi Arabia: Evidence from Bayesian model averaging

Aird regions face a serious challenge in balancing food security with the sustainable management of their scarce water resources. This study investigates the most influential predictors of food security in Saudi Arabia over the period 2000–2023. Using the minimum dietary energy requirement as a proxy for food security, the analysis incorporates key water sustainability indicators, including agricultural water withdrawal, industrial water use efficiency, irrigated agriculture water use efficiency, service water use efficiency, and total water use efficiency. The Bayesian Model Averaging (BMA) is employed, with Lasso regression used as a robustness check for data analysis. The findings reveal that total, industrial, and irrigated agricultural water use efficiencies are the most significant food security predictors with the highest posterior inclusion probability (PIP) estimated as 1.0, 0.978, and 0.644, respectively. BMA outputs demonstrate strong model fit and high predictive accuracy, with total water use efficiency consistently included in high-probability models and exerting a strong positive influence (PIP = 1.00 and a strong positive effect mean = 0.0585), which is considered the amount of useful output produced per unit of total water consumed. The policy implications indicate the importance of promoting total water use efficiency to enhance food security, particularly in arid regions. Emphasis should be placed on system-wide water productivity improvements rather than isolated sector-specific interventions. This study contributes significantly to the growing field of sustainable development by exploring the complex and ignored link between water use efficiency and food security in water-scarce regions.