Executive Summary

Trade policies can affect the environment at a resolution that goes beyond what CGE models can model. Models with technological and spatial resolution that is sufficient to capture regional environmental impacts, in contrast, may not properly capture relevant aspects of international trade, such as the impacts on factor markets. We bridge the gap between CGE models, by the example of MIRAGE, of trade and the regional agricultural supply models of CAPRI by developing a top-down link that is modular in the sense that it is straightforward to implement in CGE models that rely on the GTAP database and that is minimally invasive in the CAPRI model.
We implement the module for the pair MIRAGE-CAPRI and apply it to study the impacts of the EU-Canadian trade agreement “CETA” on nutrient surpluses in the EU.
The results indicate that the economic impacts of CETA are small, and so are the environmental impacts. Nevertheless, our simulations indicate that CETA can have regionally differentiated impacts on the environment when it comes to dairy production. Dairy exports to Canada expand to fill the enlarged tariff rate quotas, and
this expansion stimulates production in regions already specializing in dairy production, such as Northern Italy. In such regions, the manure application and crop nutrient surpluses can increase, making it more difficult to comply with environmental regulation such as that under the Water Framework Directive. Moreover, the
simulations indicate that not all dairy producing regions are similarly affected.
Depending on the sectoral use of endowments such as labour and capital and how those prices change in each country following CETA, the dairy sector may not expand at all. This is the case for the Netherlands in our simulations.
From a technical/methodological point of view, our simulations indicate some lessons for the future: Firstly, the modular implementation chosen appears feasible. Secondly, the linked models sometime give conflicting results for the economic development of the same sector, which we interpret as a sign that either the cost shares of inputs are different in the models or that the link is incomplete. In particular, land rents and use is endogenously modelled in both models and thus cannot easily be linked. Finally, it would be interesting to differentiate the price reactions coming from trade to the regional models in relation to the extent to which they supply the products that actually expand. With the current set-up, all regional models receive the same price
signal for the same broad category of primary output, e.g. “raw milk”, whereas the trade deal may entail expansion of particular types of cheese.