Non-tradable goods and the border effect puzzle

The surprisingly high Canada-U.S. border effect estimated by McCallum has been puzzling trade economists in the last ten years. We argue in this paper that conventional estimates of the border effect without consideration of non-tradable goods can overstate the trade reducing effect of the national border and the impacts can be considerable. We then explore the Canada-U.S. case with a numerical general equilibrium model with parameters calibrated to 2001 data. Our counterfactual experiment results suggest that after adjusting for effects of non-tradable goods the Canada-U.S. border effect is reduced to 2.11.