ECONOMIC DEVELOPMENT QUARTERLY / November 2001
combined exports to Canada and Mexico increased from $680 million to $1.59 billion (an increase of 133%), suggesting a trade-related employment increase of 18,000 new jobs (using the Com- merce Department’s multiplier). Over the same period, the Commerce Department noted that approximately 2,500 jobs in WNY were lost as a direct result of import competition from Mexico and Canada, giving a net gain of approximately 15,500 jobs. Estimates regarding job losses come from trade adjustment assistance (TAA) claims that are filed by employers or individual workers with the U.S. Department of Labor. Although this illustration considers jobs created or lost as a result of trade with Mexico and Canada, it does not say very much about the role of NAFTA itself. Even if one were to fully accept the accuracy of the Commerce Department’s multiplier, estimates of trade-related job creation would need to be deflated with regard to trade that would have taken place anyway.
Not surprisingly, impact estimates that come from the Commerce Department’s approach (or variants of that approach) have been criticized by trade economists on a number of grounds. The most obvious problem is that a substantial proportion of WNY’s increased trade with Mexico or Canada would have taken place regardless of NAFTA (in which case, the basic task would be to assess NAFTA’s contribution to the increase). Even if this could be done, there are a number of other problems with the multiplier approach. For example, the multiplier that links employment with exports is rather old (last calibrated in 1993); the multiplier represents an aggregate (national) parameter that may not apply to specific regions as a result of structural differences in the composi- tion of exports at the metropolitan or county levels; and more fundamentally, the multiplier does not account for the possibility that export growth can take place in the absence of any employment growth at all. (We illustrate this point later.)
On the import side, the TAA approach is also problematic. For instance, TAA certifications are issued on the basis of self-reported employment effects by firms and workers. Verification stan- dards are far from rigorous in that the Department of Labor does not have a methodology for assessing the legitimacy of trade-induced losses (NAID, 1996). A related problem is that jobs lost to import competition might well have disappeared anyway (regardless of whether NAFTA was in place). Will (2000) reported that TAA certifications have not increased dramatically in the post- NAFTA period and that there is no correlation between import growth rates from Mexico or Can- ada and TAA certifications with regard to either of these two nations.
. . . it would seem that any assessment of the impact of NAFTA ought to address the contingency issues inherent to the current debate. Specifically, what would have happened if NAFTA had not been approved in the first place?
A more sophisticated approach toward impact assessment comes from the NAID-Armington methodology (NAID, 1996), which differentiates the employment effects of export and import change by industry sectors that were liberalized under NAFTA compared to sectors that were not. This approach attempts to measure the degree of substitutability (and complementarity) between imports and domestic production (Armington elasticities). The most recent results from this tech- nique reveal that the sectors with the largest employment gains (electronic equipment, industrial machinery, apparel, and transportation equipment) are also the sectors that are most vulnerable to import competition. On the basis of Armington elasticities, NAID ranked Erie County (the single largest part of WNY) 65th in the nation in terms of NAFTA sensitivity (i.e., the potential for job loss). An important disclaimer in the NAID study is that reliable impact assessments at the county level are unlikely to emerge until a multiperiod, multiregional, dynamic general equilibrium model of trade, capital, and labor flows linking all three nations is constructed (see Kouparitsas, 1997). An operational model of this type has yet to be developed.
From the sketch outlined above, it would seem that any assessment of the impact of NAFTA ought to address the contingency issues inherent to the current debate. Specifically, what would have happened if NAFTA had not been approved in the first place? In an effort to answer this ques- tion, Gould (1998) employed a multiperiod gravity model of trade that included incomes, prices, and exchange rates. Once the fundamental determinants of trade flows were accounted for in the pre- and post-NAFTA periods, extraordinary U.S.-Mexico flows were attributed to the free trade agreement. Although Gould’s results suggested a net gain for the United States as a whole, his model was not framed to capture regional effects, nor did it consider specific sectors. To the best of our knowledge, regional applications of Gould’s approach have yet to appear in the literature. This is unfortunate, if only because Gould’s methodology could be combined with a multiplier approach to estimate jobs created (or lost) as a direct result of trade legislation.