Market Integration and Forecasting in Sustainable Citrus Supply Chains in Türkiye

Sustainable fresh food supply chains depend heavily on effective coordination under uncertain market conditions. Citrus export systems are particularly sensitive to perishability, seasonality, and cold-chain constraints. Under these conditions, reliable price forecasts are important for export planning, logistics efficiency, and sustainable supply-chain management. This study examines the relationship between orange and mandarin export prices in Türkiye using monthly data from 2016 to 2025. Export prices are proxied by real unit values derived from official trade statistics. The analysis applies Augmented Dickey–Fuller tests, Johansen cointegration analysis, and Vector Error Correction Models (VECMs). Forecast performance is evaluated using a rolling-origin framework and compared with ETS, SARIMA, XGBoost, and a seasonal naïve benchmark. The results identify one cointegrating relationship between the two export markets. The estimated long-run coefficient is 0.92, indicating near one-to-one price co-movement. Adjustment toward equilibrium is asymmetric, with orange prices responding faster (ECT = −0.44) than mandarin prices (ECT = −0.21). Forecasting results show that VECM-based models outperform all alternative specifications. The robust VECM achieves the lowest forecast errors (MAPE = 8.3%), compared with 9.8% for XGBoost, 10.6% for SARIMA, 11.5% for ETS, and 14.1% for the seasonal naïve benchmark. Diebold–Mariano tests confirm that these improvements are statistically significant. The findings indicate that orange and mandarin export prices should be analyzed jointly rather than separately. In closely connected citrus supply chains, cointegration-based forecasting models provide more reliable forecasts and a stronger analytical basis for sustainable market coordination.