Income Tiers, Expenditure Shares, and Satiation: A Fully Interacted Fractional Logit Analysis of Household Meat Consumption in Türkiye

ABSTRACT Although food demand in Türkiye is widely researched, existing studies mostly rely on aggregate demand systems and provide limited insights into income‐specific meat expenditure shares. Using microdata from the 2022 Household Budget Survey, this study analyzes household meat spending shares via a unified fractional logit interaction model estimated with quasi‐maximum likelihood and HC3 robust standard errors. Formal model testing strongly rejects parameter homogeneity across income groups ( p < 0.001), confirming systematic slope variation and supporting a fully interacted specification over restricted pooled models. The findings reveal critical behavioral differences across income strata. Disposable income significantly increases meat budget shares only among low‐income households, an effect that vanishes for middle‐ and high‐income groups, providing empirical evidence for the household‐level satiation hypothesis. Predictive simulation trajectories further illustrate structural asymmetry: low‐income meat budget shares rise sharply from 3.0% to 4.1% (a relative increase exceeding 36%), confirming high‐protein foods as aspirational goods under tight budget constraints. Conversely, medium‐ and high‐income trajectories flatten and plateau between 5.7% and 6.0%, indicating complete income inelasticity and a saturation ceiling. Socio‐demographic factors—including education, employment, gender, tobacco use, and dining out—also display notable income‐dependent relationships, gaining economic significance once structural interactions are accounted for. Although based on a single cross‐section during high inflation—limiting dynamic causal inferences about inflation—the study moves beyond average population effects by verifying structural behavioral shifts across income tiers. These results suggest that food assistance and social welfare policies must account for income‐specific behavioral dynamics rather than relying on uniform aggregate measures.