A Beta Regression Approach to Modelling Country-Level Food Insecurity

Food insecurity remains a persistent global challenge, despite significant advancements in agricultural production and technology. The main objective of this study is to identify and quantitatively assess some of the structural determinants influencing country-level food insecurity and provide an empirical background for policy-making aimed at achieving the Sustainable Development Goal of Zero Hunger (SDG 2). This study employs a beta regression model in order to study moderate or severe food insecurity across 153 countries, using a cross-sectional dataset that integrates economic, agricultural, political, and demographic independent variables. The analysis identifies low household per capita final consumption expenditure (β = -9 × 10^-5, <i>p</i> < 0.001), high income inequality expressed as a high GINI coefficient (β = 0.047, <i>p</i> < 0.001), high long-term inflation (β = 0.0176, <i>p</i> = 0.003), and low economic globalization (β = -0.021, <i>p</i> = 0.001) as the most significant predictors of food insecurity. Agricultural variables such as land area (β = -1 × 10^-5, <i>p</i> = 0.02) and productivity per hectare (β = -9 × 10^-5, <i>p</i> = 0.09) showed limited but statistically significant inverse effects (lowering food insecurity), while factors like unemployment, political stability, and conflict were not significant in the model. The findings suggest that increased economic capacity, inequality reduction, inflation control, and global trade integration are critical pathways for reducing food insecurity. Future research could employ beta regression in time-series and panel analyses or spatial models like geographically weighted regression to capture geographic differences in food insecurity determinants.