Evidence that national borders continue to matter has been accumulating since 1995. It is
not that these lines on the map cause trade distortion by themselves; they merely
delineate the areas within which trade is relatively more intense than between them.
These areas are coincidentally known as nations. As long as these different trade
intensities persist, implying higher trade costs for goods that are traded internationally,
nations (and by extension, national borders) matter for trading purposes. In M cCallum
(1995), after taking into account economic size and distance, a Canadian province was
found to be 22 times more likely to trade with another province than with a U.S. state.
Subsequent research in this area has confirmed significant and robust border effects
between the U.S. and Canada. Helliwell (2001) estimates the border effect at 11; the
same ratio o f intensity o f Canadian trade to Canada-U.S. trade is implied in Anderson and
van Wincoop (2001). These and most other estimates, like 15.2 in Anderson and Smith
(1999), are found in the range between 10 and 22. The existence o f the border effect has
also been confirmed for OECD and EU trade by Wei (1996), Nitsch (2000) and Head and