Incorporating Modern Trade Theory into CGE Models

We propose a parsimonious way to incorporate three workhorse models in the modern trade literature into the computable general equilibrium model (CGEs) GTAP. Furthermore we estimate (a part of) the parameters in the model structurally. We show that the Ethier-Krugman monopolistic competition model, and the Melitz firm heterogeneity model can be defined as an Armington model with generalized marginal costs, generalized trade costs, and a demand externality. We also incorporate the Eaton-Kortum in the standard GTAP-code. As already known in the literature in both the Ethier-Krugman model and the Melitz model generalized marginal costs are a function of the amount of factor input bundles. In the Melitz model generalized marginal costs are also a function of the price of the factor input bundles. Lower factor prices raise the number of firms that can enter the market profitably (extensive margin), reducing generalized marginal costs of a representative firm. For the same reason the Melitz model features a demand externality: in a larger market more firms can enter. The Eaton-Kortum model deviates from the Armington model because aggregate industry prices do not vary by origin in this model. We implement the different models in the CGE model GTAP. The simulations show that the largest welfare gains are generated in varying models, depending upon the type of trade liberalization and the number of sectors which are modelled as Ethier-Krugman or Melitz. Effects are largest in the Ethier-Krugman model under nationwide trade liberalization and all sectors modelled as either Ethier-Krugman, Melitz or Armington. With only some sectors modelled as Melitz or Ethier-Krugman, the largest welfare effects are produced in the Melitz model. We also provide a detailed and intuitive description of incorporation in the GEMPACK-code of the three different models.