It seems that WonkBlog is picking up some of the Washington Post's bad habits. The Post was a strong supporter of NAFTA when the trade agreement was being debated, virtually closing its news and opinion pages to critics of the deal. In the almost two decades since NAFTA passed the Post has run numerous pieces touting the benefits of the agreement for both Mexico and the United States.
The praise has been especially off the mark in the case of Mexico. In spite of having the lowest per capita growth of any country in Latin America over the last decade, the Post has routinely run pieces highlighting the boom in Mexico and the country's growing middle class (e.g. see here and here). The Post even claimed in a 2007 editorial that Mexico's GDP had quadrupled since 1988. (The actual growth number was 83 percent.)
WonkBlog has a piece that cites a new paper and claims that NAFTA has been good for all involved, showing GDP and wage gains for Canada, Mexico, and the U.S. While this is in fact the conclusion of the study, it would have been worth including some qualifying remarks.
For example, the study explicitly assumes that there is only one type of labor. (The bottom of page 26 explains that in the modeling exercise there is "one wage per
country.") This simplifying assumption can be useful for some purposes, but if the question is whether NAFTA might have hurt less-educated workers (e.g. autoworkers and steelworkers) to the benefit of more highly educated workers (e.g. doctors and lawyers), it cannot be answered with a model where there is one type of labor.
This upward redistribution is exactly what fans of the Stolper-Samuelson theorem would expect from a trade agreement like NAFTA. Therefore this model can not be used to tell us whether NAFTA would have had one of the negative effects predicted by economic theory.
The other big item missing from this model is the impact of stronger patent and copyright protections. NAFTA required Mexico to develop a U.S. style patent system which substantially raised the cost of prescription drugs and other products in Mexico. This model makes no effort to measure the impact of this increased protectionism on the Mexican economy directly, or indirectly on the other two economies. Insofar as this interference with the free market led to higher prices and increased distortions, it would be expected to slow growth, but obviously that effect cannot be picked up in this model.
In short, the model highlighted in this post can be useful for some purposes but it cannot possibly provide a basis for telling us whether NAFTA was on net good or bad for the United States, Mexico, and Canada.
An earlier verison referred to "Wongblog."