Non-Linear Spatial Price Transmission on the Turkish Wheat Market

Agricultural production is spread all over Turkey and the considerably different climatic and topographical conditions among the provinces lead to highly diversified agricultural production. Thus, is it reasonable to assume an integrated market all over Turkey? This paper analyzes the spatial price transmission among 28 Turkish provinces. Applying a bivariate threshold vector error correction model allows to account for transaction costs without having direct information of these. The results suggest that there is an inner cluster of several provinces that can be treated as one large integrated market. However, the degree of market integration decreases in the outer provinces of Turkey. Although transaction costs are significant for nearly all provinces, Ankara clearly stands out, showing transaction costs with every other province. The observed economic variables fail to explain this behavior. Hence, social networks as relevant factor for the interactions on the Turkish wheat market come into play. A band of non-adjustment is only established for a minority of those pairs that show a threshold. Thus, one has to critically assess the applicability of such a threshold model to analyze the market behavior of spatial price transmission.