This paper reaches two conclusions. First, in evaluating PTAs, no spec
ial consideration need be given to transport costs over other types of
costs. Unless one focuses on cases in which PTA and global free-trade
outcomes are identical, it does not matter whether differences in del
ivered prices result from transport costs or from differences in techn
ology or factor endowments. Second, gravity models are inherently inca
pable of giving any guidance on whether a specific grouping is desirab
le or whether transport costs make regional PTAs desirable. The gravit
y equation is consistent with virtually any trade model. The welfare i
mplications of PTAs, however, depend crucially on the specific model.
JEL no. F02, F15.