摘要：With the Doha Round of negotiations having come to a standstill, more countries opt for preferential trade agreements with only a limited number of partners. Starting two recent negotiations, the Trans-Pacific Partnership and the EU-US trade deal, might mark the beginning of a new era in multilateral trade negotiations in a sense that they connect the largest but geographically distant players of the world market. The impact of preferential agreements on welfare and trade patterns has been subject to economic investigation for decades. Applied equilibrium models are key analytical tools in the ex ante assessment of trade negotiations, but are often criticized as being sensitive with regard to underlying assumptions and input data. For trade related impact assessment, assumptions relating to the aggregation and presentation of border protection instruments are of specific interest. This study contributes to the assessment of equilibrium modelling techniques with a focus on tariff rate quotas (TRQ) by systematically comparing simulated impacts on traded volumes and welfare under different implementation of TRQs. In the equilibrium modelling literature TRQ instruments are either modelled explicitly (linking the variable tariff rate and the fill rate of the quota threshold) or transformed into an ad valorem equivalent (AVE) tariff rate. In the standard Vinerian framework of welfare analysis, trade diversion occurs when imports from low cost producers in the rest of the world are displaced by exporters benefitting from trade preferences. The simulated shift in imports in an equilibrium model depends on the third country policy representation. With binding tariff rate quotas in the initial point, for example, shifts in traded volumes will be significantly different if the TRQ instrument is modelled explicitly or by its AVE tariff rate. This study demonstrates the sensitivity of simulated results by both developing a simple three country model of international trade and by implementing an illustrative EU-US trade deal scenario with the Common Agricultural Policy Regionalised Impacts (CAPRI) modelling system. The focus is on whether the choice of modelling TRQ instruments with third countries explicitly or by their AVE tariff rates has a significant impact on simulation results. In default, most policy instruments in CAPRI – including border protection and market intervention mechanisms – are modelled explicitly. Tariffs subject to quota limits are approximated with a smooth function mimicking the switching mechanism between preferential and out of quota rates. For the sake of this study this mechanism is optionally replaced with the AVE representation. CAPRI is then calibrated under both TRQ representations and the results of the same trade deal scenario are compared.
