The U.S.-Canada Border Effect: Smaller Than Previously Thought and Becoming Smaller

We revisit the effect of the U.S.-Canada national border on trade, considering to what extent the border reduces trade when observable economic factors are controlled. A reexamination of the data yields estimates of the border effect that are 50% higher than previously expected (Feenstra, 2004); however, the nonlinearity of the estimation and distance measure reduce the effect by 65%. We therefore conclude that the border effect in 1993 had a factor of 4.10. This figure is 15% lower than the effect proposed by previous research. We also calculate the border effect for subsequent years and find that this effect steadily decreased to a low of 3.21 in 2007. Interestingly, a traditional linear methodology cannot be used to identify this decline clearly