The role of global agricultural market integration in multiregional economic modeling: Using hindcast experiments to validate an Armington model

The representations of international trade and global market integration play central roles in long-term global agricultural economic modeling. However, the conventional use of the gravity models (i.e., the Armington approach) of trade may not adequately account for the dynamic process of market integration. In this paper, we generalize a logit-based Armington approach by permitting home preference bias erosion to account for a trend of global market integration. We build a simple agricultural economic equilibrium model and conduct hindcast experiments to examine the historical trend of market integration. The results show a significant integrating trend of the global agricultural market implied by converging Armington preference distributions (i.e., home bias erosion) in 1995–2015. We demonstrate important implications of allowing the dynamic trend of market integration implied by bias erosion on estimating trade elasticities. By testing the future market integration trend implied by historical home bias integration in the Global Change Analysis Model (GCAM), we show a high sensitivity of long-term agroeconomic projections to future trade and market integration scenarios. •Armington models may not account for the dynamic process of market integration.•We generalize a logit-based Armington model to permit home bias erosion.•We use hindcast experiments to validate the generalized model.•Agricultural market integration intensified as implied by historical bias erosion.•Global market integration plays a critical role in projecting agroeconomic future.