Stochastic Modeling of Food Insecurity

Recent advances in food insecurity classification have made analytical approaches to predictand inform response to food crises possible. This paper develops a predictive, statistical framework to identifydrivers of food insecurity risk with simulation capabilities for scenario analyses, risk assessment and forecastingpurposes. It utilizes a panel vector-autoregression to model food insecurity distributions of 15 Sub-Saharan Africancountries between October 2009 and February 2019. Statistical variable selection methods are employed toidentify the most important agronomic, weather, conflict and economic variables. The paper finds that food insecuritydynamics are asymmetric and past-dependent, with low insecurity states more likely to transition to highinsecurity states than vice versa. Conflict variables are more relevant for dynamics in highly critical stages, whileagronomic and weather variables are more important for less critical states. Food prices are predictive for all cases. ABayesian extension is introduced to incorporate expert opinions through the use of priors, which lead tosignificant improvements in model performance.