This paper analyzes the qualitative properties of a multisectoral, multiregional computable general equilibrium model where some industries include heterogeneous firms as in Melitz (2003). The model, formulated according to Roson and Oyamada (2014), adds endogenous productivity effects to a standard Walras-Ricardian framework. We argue that the inclusion of such effects changes the magnitude and distribution of welfare benefits obtainable by reductions in trade barriers, due to of comparative advantages. We illustrate the point through a numerical example, in which alternative model formulations are assessed. A standard neoclassic GE model, a basic Melitz model and a hybrid model are then compared. The three model versions are all calibrated with the same data set and an identical simulation experiment (a 50% reduction of transport costs between two regions) is carried out in the three cases. The results show that the hybrid model displays the largest welfare gains, as it combines Ricardian comparative advantages with Melitz average productivity improvements. However, they also show that new effects, not present in the original Ricardo and Melitz frameworks, are at a work.