effects varies substantially across studies. Cernat (2001) considered many FTAs, and found
that some were trade creating (AFTA, EU, SADC and COMESA) while others were trade
diverting (MERCOSUR and Andean Community). For the EU, he estimated 20 percent
trade creation measured as a share of total EU imports, while for the NAFTA the results are
mixed. Frankel and Wei (1995) report 15 percent trade creation for the EU and found no
evidence of trade diversion for the period 1972-92. The above literature highlights differences
across FTAs in its effect on trade, and emphasizes the need to analyze each FTA individually.
A large number of papers starting with McCallum (1995), Helliwell (1998), Hillberry
(1998), Anderson and van Wincoop (2001) assess the effects of borders for the bilateral trade
flows between FTA partners and conclude that borders matter. Regarding the particular
issue of agricultural trade, Furtan and Van Melle (2004) estimate the border effects of trade
in agricultural products between Canada and its NAFTA partners: US and Mexico between
1992 and 1998. The authors conclude that the non-tariff frictions in Canada-US agricultural
trade are significant (Furtan and van Melle (2004)).
Vollrath and coauthors (2006) adopt a gravity approach when they analyze trade in
processed and staple agri-food products among a wide cross-section of 69 countries biennially
from 1996 to 2002. Among others, they find that differences in per capita income affect trade
in manufactured food but not trade in commodity foods, that is consistent with the HO
theory of international trade - the land/labor ratio is an important determinant of trade in
food products and that the EU, NAFTA and MERCOSUR have all increased the intra-bloc
trade in food products beyond that which would have taken place in the counterfactual. In a
more specialized study focusing on Canadian pork exports to the US market, Prentice et. al.