Changes in transport and non-transport costs: Local vs global impacts in a spatial network

We develop a multi-country Dixit-Stiglitz trade model and analyze how industry location and welfare respond to changes in: (i) transport frictions (e.g., infrastructure, transportation technology); and (ii) non-transport frictions (e.g., tariffs, standards and regulations). We show that changes in non-transport frictions, which are usually origin-destination specific, do not allow for any clear prediction as to changes in industry location and welfare; whereas changes in transport frictions, which are usually not origin-destination specific, may allow for such predictions. In particular, we show that reductions in transport frictions occurring at links around which the spatial network is locally a tree are Pareto welfare improving.