Trade in dynamic CGE models. Theoretical structure, projections, and challenges

This paper provides an overview of the different approaches to model trade in dynamic computable general equilibrium models. It will address the theoretical structures used, the trade baselines implemented in the projections, and the challenges for dynamic trade modelling. On the theory side we study the import demand specifications employed (Armington, monopolistic competition, and the use of nested preferences) and the export structure (perfect or imperfect transformation between domestic and exported goods). We also provide an in-depth discussion of the approaches to model the trade balance. On the trade structure most models converge on the use of nested Armington preferences with domestic goods and exports being homogeneous. Approaches to modelling the trade balance diverge between constant trade balances, trade balances based on a macro model, trade balances determined by rates of return, and converging trade balances. On projections the main questions are whether and how trade to income ratios are modelled given historical trends in this ratio. The way structural change is modelled also affects trade projections as it impacts the sectoral composition of trade and the possibility to generate Balassa-Samuelson effects (higher price levels in richer countries). Challenges for dynamic trade modelling consist among others of capturing changes in the extensive margin, exploring the impact of new technologies and digitalization on the size and composition of trade, including affiliate sales in projections, creating a database for trade cost trajectories related to the numerous FTAs in place, the appropriateness for dynamic modelling of the theoretical structures and parameters developed for static models, and the role of changes in energy use and energy efficiency in the development of trade balances.