Was NAFTA Behind the Mexican Export Boom (1994-2000)?

The North America Free Trade Agreement (NAFTA) took effect in January 1994. From that year until 2000, Mexican exports grew three times faster (15%) than in the previous six years. This remarkable showing has encouraged Mexico and other countries to pursue more accords. Nevertheless, the literature on the early effects of NAFTA is rather scant and seems supportive of the wide-spread belief (based on simulations of CGE models) that the Agreement would have little effect on Canada and the USA but would bestow large benefits on Mexican foreign trade. On the contrary, I find that the behavior of Mexican exports has not been affected in any meaningful way by that accord and that the evidence in the case of imports is ambiguous at best. I estimate simple long-run functions for imports and exports using monthly data and show, with proper tests for regressions with I(1) variables, that they are stable and their parameters are constant from 1990 to 2000. The short-run disequilibrium models are stable and constant as well. The absence of a NAFTA effect remains even when more disaggregated data are analyzed. These results leave the strong American economic expansion as the only explanation for the outstanding performance of Mexican foreign trade (it accounts for 96% of the growth of the equilibrium level of Mexican exports since 1994). Therefore, countries should not indulge in unrealistic hopes when signing these agreements although in no way are my findings an argument against the promotion of free trade-just the opposite. Extending the analysis to include the 80's suggests that the really important institutional change that allowed Mexican exporters to take advantage of the favorable external conditions was the abandonment of an erratic trade policy in favor of the opening up of the economy following the adherence of Mexico to the GATT. The econometric analysis is very interesting on its own since the data provide an excellent case to test the performance of several techniques to estimate cointegration vectors (residual-based, Johansen, single-equation ECM and FM-LS) in the presence of structural changes likely to be found in developing economies.