coefficient is not only positive but also insignificant. Controlling for industry-specific fixed
effects in columns (3)-(4), such as differences in initial tariffs, factor intensities, and industry-
specific comparative advantage, seems to have little effect on the sign of β1. It is significant
atthe 10 percent confidence level, suggesting that the reduction in Canadian exports to
Chile was greater in those sectors where the CCFTA tariff cuts were deeper. Therefore,
there is little evidence that tariff preferences granted by Chile for Canadian imports under
the CCFTA have promoted an increase in Canadian agricultural exports to that country.
At the same time, in the specification with Canadian imports from Chile in columns
(1)-(4), the effect of the CCFTA tariff reduction has a strong and significant effect on agri-
cultural trade. The coefficient on the Canadian tariff change toward Chilean imports in
the specification with industry-specific fixed effects is -0.75, which implies a 75 percent
increase in imports in response to a 1 percent tariff reduction. This number seems to be
very large, given that the average tariff change for Chilean agricultural products from 1997
to 2005 was 4.5 percent and the average increase in imports for the same period was only
87 percent. For example, 1 percent tariff reduction for the average country, controlling for
country-specific characteristics, is responsible for only a 28 percent increase in imports. This
suggests that there must be some factors other than tariff changes that affected positively
Canadian imports from Chile (for example, there may be a sharp decline in import quota
restriction going parallel with tariff reduction, and both effects will be attributed to tariffs).
In general the magnitude seems to be overestimated, but the sign and significance are very
robust across specifications. Therefore, the data confirms that Canadian and Chilean agri-
cultural exporters responded differently to trade liberalization made available by the FTA.
17