Abstract Modeling trade and transportation costs is an essential part of multiregional or spatial computable general equilibrium models where interregional trade plays an important rolein shaping economic activity. The majority of such models use the iceberg trade cost approach where part of the produced output (representing the material costs of transportation) is assumed to melt away during transportation. There are a few models which employ a more refined approach with an explicit transportation sector providing transportation services which are then used to ship goods between locations. In this paper we show that this approach, although much more convenient than the iceberg approach, still lacks full usability due to the fact that markets, hence prices are defined at the regional level and as a result, transportation costs can not be endogenous at the trade relation level. Moreover, under regional level market clearing the iceberg and the more detailed approach are equivalent. We propose to refine the definition of market equilibrium and move it to the trade relation level. Using this approach we can gain full advantage of the explicit transport sector in the model with respect to trade cost evolution. We show through simulations that refining the way trade costs are modelled indeed gains new insights, and that moving the market definition to the trade relational level leads to qualitative changes in the effect of labor supply shocks on main model variables. The paper also presents a method to estimate a SAM by reallocating data from standard industries to a transportation sector which is then consistent with the model setup. This SAM can be used to calibrate the refined model with a detailed transportation sector.