Cointegration and Spillover Analysis of Agri-Food Supply Chain

To quantitatively assess the dynamic interrelations among the three principal stages of Uzbekistan's agri-food supply chain (AFSC) — upstream production, midstream processing, and downstream export/consumption — and identify the shock propagation mechanism within the chain. Based on annual data for 2015–2025, the study applies ADF and PP unit root tests, the Johansen cointegration approach, a Structural VAR (SVAR), a VECM(2, r=1) system model, Granger causality tests, Generalized Impulse Response Functions (GIRF), Forecast Error Variance Decomposition (FEVD), the Diebold–Yilmaz spillover index, and Time-Varying Parameter VAR (TVP-VAR). The Johansen test confirms a single cointegration vector (r=1, λ-trace=28.4). The VECM adjustment speed is α₁=−0.427 (p=0.007); the return to equilibrium takes approximately 2.3 years. The production stage is a clear Granger cause of the export stage (p<0.05), while processing and export exhibit bidirectional causality. According to FEVD, the spillover contribution of the production shock reaches 39.9% by period 12. The Diebold–Yilmaz total spillover index is 47.3%, indicating a medium-to-high degree of chain integration. The AFSC is in a stable long-run equilibrium, but top-down shock propagation is dominant; strengthening upstream-midstream integration is essential for chain resilience.