US and Mexico have been sharing a 3,200 kilometers common border over the last 169 years, as a result of the Treaty of Guadalupe Hidalgo concluding the American-Mexican war of the mid nineteenth century, complemented with a peaceful purchase of territories dedicated to a planned railway a couple of years later. 

It’s difficult to ignore your neighbors for one and a half century – at least this is what trade theory is telling us. At some point you will be tempted to grasp business opportunities and do profitable commerce. It might also be the case that your neighbors ramble in your garden or end up working for you as gardeners. The Nobel Prize Paul Krugman once coined this evidence as “natural trading blocs”.[1] In a nutshell, Canada, the US and Mexico form such natural bloc, as opposed e.g. to US and Jordan (another US free trade agreement).[2] Ronald Reagan apparently shared these views when endorsing in 1987 the US-Canada Free Trade Agreement (CUSFTA) – a follow up to the long standing “Auto Pact” having shaped the patterns of the North-American car industry since the mid sixties.

In contrast, Donald Trump repeatedly opposed geography in his declarations, leading to the cancellation of the visit of President Enrique Peña Nieto. The contentious project of completing the wall between the two neighbors to curb migrations, and the announcement of the renegotiation of the NAFTA put at risk their growing integration over the last two decades. Recall indeed that President Clinton signed the NAFTA in 1994 after having been confronted to Ross Perot statement: “We have got to stop sending jobs overseas (…) there will be a giant sucking sound going south”.

To Donald Trump, NAFTA is just “a terrible deal [and] a total disaster for [the US] since its inception”. To trade and economic geography scholars, NAFTA is just a natural deal given geography. There are excellent arguments to think that getting rid of this preferential relationship would have terrible consequences on both sides and would be inefficient – see recent post by Marc Melitz. Global value chains have been developed with countries Southern border of the US and what is imported in the US is partly US value added embodied in goods assembled in Mexico – and this is the more true as the car industry is concerned (see evidence provided by Keith Head and Thierry Mayer on this: Voxeu blog post and Lettre du CEPII).

Now, the question is, how different is the US relation to Mexico, compared to other trade deals signed by the US administration? What does actually geography tell us? Let’s go for a simple exercise: we follow Egger & Larch (2008)[3] and Baier & Bergstrand (2004)[4] and model the contribution of economic determinant to the probability of formation of a Preferential Trade Agreement (PTA). Our explained variable takes the value of 1 if country i and country j are member of the same PTA in year t and zero otherwise. In more statistical terms, what we estimate is the probability of signing a PTA using a probit estimator, among 159 countries (there are indeed more than ten thousand possible agreements possibly signed: 159*158/2). This probability is explained by simple determinants: the distance between country i and j; the average distance of each country in the pair to all the other destination markets; whether i and j belong to the same continent; the sum of their economic sizes (GDPs); the similarity in the two market size; the difference in their economic development levels (a proxy of their comparative advantage) and vis-à-vis the rest of the world; lastly, the number of previously ratified agreements by the two countries. Based on this estimates, we can predict the probability of signing an agreement between any country pair: this is somehow the desirability (from a geographic and economic point of view) of a bilateral RTA.

Figure 1 reports the share of country pairs (ij) involved in a PTA by year indicating a sharp increase in trade agreements during the last two decades: the fraction of possible pairs covered by a PTA has actually more than tripled.

In such context, what about the two neighbors on North and South of Rio Grande? As shown in Figure 2, Canada and Mexico are the two more “natural” trading partners for the US, followed by Chile and the Central American FTA (CUSFTA), Panama being the threshold agreement (see the vertical bar in the figure). Beyond the latter country, economic and geographic determinants hardly justify signing agreements: more geopolitical considerations then shape the decisions made (like for Jordan, already mentioned).

There is also a dynamic dimension in such reasoning. Figure 3 shows the predicted probability to sign a preferential agreement among the three NAFTA members at different dates. Not surprisingly CAN-USA (83%) seems to be the “natural” pair with a predicted probability far above the cut-off level predicted by the model for the 1995-2014 period (above 11.3%, you should sign a bilateral agreement). When Bill Clinton signed the NAFTA, Mexico was a natural partner (above threshold), but his decision was forward looking: two decades later the pair CAN-MEX (71%) and MEX-USA (76%) is now of the same order of magnitude as CAN-USA. Donald Trump will rightly oppose us that geography did not change over the last two decades; what he misses is that economic conditions (market size, comparative advantages) did change, not only considering the US-Mexico pair, but also comparing this pair to any possible pair of countries. Reagan and Clinton were not wrong in signing the CUSFTA and the NAFTA respectively; Trump would be foolish to dump the latter.
 

[1] Krugman, P. (1991). The move toward free trade zones. Economic Review-Federal Reserve Bank of Kansas City, 76(6), 5.
 

[2] The US-Jordan Free Trade Agreement was fully implemented on January 1, 2010.
 

[3] Egger, P. and Larch, M. (2008). Interdependent preferential trade agreement memberships: An empirical analysis. Journal of International Economics, 76(2):384-399.
 

[4] Baier, S. L. and Bergstrand, J. H. (2004). Economic determinants of free trade agreements. Journal of International Economics, 64(1):29-63.
 

[5] de Sousa, J. (2012). The currency union e_ect on trade is decreasing over time. Economics Letters, 117(3):917-920.