For an account on the history in question, search for the 'Bracero Program', 'maquiladora' and 'NAFTA'.
P.96c*:
The conclusion of the standard trade model is that migration and trade are substitutes in both the short and long run (Heckscher, 1949; Ohlin, 1933; Mundell, 1957; Stolper and Samuelson, 1941; Krauss, 1976).However, as this conclusion relies on 5 assumptions, any violation of the assumptions will make migration and trade complements, giving rise to a short-term migration 'hump', as depicted in the graph below.
This is a replication of Figure 5.1 in Martin2001. I hope that I am not violating any copyrights here...Assumption 1: 'The two countries share identical production technologies':
If this assumption is not met, as was indeed the case in corn production in Mexico and the U.S., exports from the country with more advanced technology will drive employment in the other country down.
In the mid-1990s, the U.S. produced about 10 times more corn than did Mexico, and could export to Mexico for a cheaper price. However, during that time, half of the man-days worked in Mexican agriculture were employed for corn production, and those jobs were eliminated as a consequence of free trade. Wages in Mexico faced downward pressure also. Clearly, emigration became more desirable.
Assumption 2: 'The two countries use the same factors of production (factor homogeneity)':
However, in reality, productivity is often high in the economically advanced country. Due to public and private offering of education, transportation, communications system, etc., the same Mexico worker may become more productive in the U.S. than in the home country. This again, encourages emigration.
Assumption 3: 'Technologies exhibit constant returns to scale in production (there are no scale economies)':
If the U.S. enjoys economy of scale, trade liberalization will lead to the expansion of U.S. production, thereby increase the demand for migrant workers (Mexicans, in our case).
Assumption 4: 'Adjustment to changes in international market is instantaneous':
In fact, those driven out of work cannot be re-employed to other positions of comparative advantage very quickly. And in this case, about 40 percent of Mexico's jobs were in the maquiladoras, which tended to hire women; yet those displaced from agriculture were mainly men. Therefore, many men migrated to work on U.S. farms.
[I believe those who migrated were those that lost their agricultural jobs in Mexico, as this is the explanation that involves the least transaction cost; however, there is no logical link indicating that this is the case, and an inquiry into who actually migrated can never be redundant work.]**
Assumption 5: 'There is perfect competition, with full employment and complete markets in both countries':
One instance against this assumption is that migration might involve high costs (because smuggling is necessary). Trade liberalization brings wages up, increasing people's ability to pay the smugglers, hence they are more likely to migrate. Martin suggests that this 'seems to have happened in the south-western Chinese province of Fujian.'
The NAFTA case in reality:
In stead of fine-tuning the 5 assumptions, Martin brings the factor of population in: the Mexico-U.S. migration may fall faster than predicted because population growth in Mexico is going down, while new jobs are created.
* The letter 'c' means the lower part of the page. Similarly, 'a' denotes the upper part, and 'b' the middle. This is a style borrowed from Encyclopaedia Britannica, IIRC.
** [] indicates my own thoughts, as per the style recommendation by Dr. Gareth.
