The North American Free Trade Agreement, which eliminated most trade barriers between the United States, Canada and Mexico, was adopted in January 1994 after a bruising battle in the U.S. Congress. Now, almost a quarter century later, the pact is on the rocks. As a candidate, Donald Trump called Nafta “the worst trade deal in history”; as president, he forced the parties back to the negotiating table under the threat of unilateral withdrawal. Here, we offer a capsule history of Nafta, its economic and political impact, and alternative scenarios for how the negotiations over its future will play out.

In the Beginning

North American economic integration was underway long before Nafta. Indeed, it was built on a series of earlier efforts to lower trade barriers. The 1965 Canada-United States Auto Pact allowed the big American automakers to operate freely in both countries. The Mexican maquiladora program of the same year set up industrial free trade zones on the Mexican side of the border to encourage U.S. investment in manufacturing. Mexico’s accession to the General Agreement on Tariffs and Trade in 1986 opened much of Mexican industry to imports from across the globe. And the Canada-U.S. Free Trade Agreement (Cusfta) set the two countries on an accelerated path toward economic integration.

Each country had its own goals for Nafta. For Mexico, the pact was a way for President Carlos Salinas to lock in the reforms started by his predecessor, Miguel de la Madrid, in the mid-1980s, which drove the top-down statist economy (kicking and screaming) toward market capitalism. Salinas hoped that free access to U.S. customers, along with a wave of foreign investment and fresh competition for Mexico’s largely protected industrial base, would spur catch-up growth in the country’s long-underperforming economy.

Canada’s objectives were less ambitious. When it became clear in September 1990 that the United States and Mexico would move ahead, with or without Canada, Ottawa decided that it had more to gain by joining Nafta than by staying on the sidelines. Participation enabled the Canadian government to forge a more durable commercial relationship with Mexico and extract concessions from the United States beyond those already secured by Cusfta.

For the United States, Nafta was an economic opportunity to expand exports and a political opportunity to better its prickly relations with Mexico. Before Nafta, U.S. exporters faced Mexican industrial tariffs averaging 12.4 percent, while Mexican firms faced U.S. tariffs averaging just 4.3 percent. Equally important, Mexican regulatory barriers to U.S. investment were daunting.

Nafta was also seen as a vehicle to support political pluralism in Mexico — the country had long endured a one-party system — as well as a response to chronic migration pressures. More jobs in Mexico, the reasoning went, meant fewer economic refugees crossing the border. Last but not least, President George H.W. Bush hoped that the pursuit of regional trade deals would pressure other U.S. trading partners to move ahead on the so-called Uruguay Round of multilateral trade liberalization.

Like all major trade agreements, Nafta was the culmination of complex negotiations both within and among nations. Generating the final text took 14 months of haggling, with side agreements on labor and environmental regulation added later. Not surprisingly, the result is a far cry from a straightforward textbook free trade agreement.

Indeed, more than 100 pages of the agreement enumerate restrictive rules of origin, especially in the textile, apparel and automotive industries, that undermine trade with countries outside the pact. Mexico retained its monopoly for the state oil company, Petróleos Mexicanos (Pemex), a symbol of national sovereignty and the cash cow of Mexican public finance. Free trade in agriculture between the United States and Mexico was phased in over 15 years, while the United States and Canada continued to exclude important farm products from obligations to open trade.

In the United States, ideologically diverse voices — organized labor, the maverick independent H. Ross Perot and the right-wing populist Patrick Buchanan — succeeded in making Nafta a headline issue in the 1992 U.S. presidential campaign. President Bush was badly (arguably fatally) wounded by the one-two punch of a primary challenge by Buchanan and then a third-party run in the general election by Perot, who famously charged that Nafta would cause a “giant sucking sound” of U.S. capital and jobs fleeing south of the border.

Environmental groups piled on, claiming that Mexico would become the pollution haven of North America, attracting firms that wanted to evade higher U.S. and Canadian standards. Bush defended Nafta as the greenest trade agreement ever — but since environmental concerns were not previously incorporated in trade agreements, the bar was low.

Once elected, Bill Clinton (who favored Nafta but was an indirect beneficiary of the backlash) persuaded Salinas and Canadian Prime Minister Brian Mulroney to negotiate side agreements on labor and environmental standards that lured sufficient numbers of Democratic fence-sitters to secure ratification. To further smooth relations with his own party, Clinton attached $90 million (to be sure, a mere token) for adjustment assistance to U.S. workers who lost their jobs because of Nafta.

Beyond these embellishments, Clinton’s primary strategy for gaining Congressional ratification could be summed up in three words: “jobs, jobs, jobs.” Most economists agree that employment levels in a country are actually determined by the vagaries of the business cycle together with fiscal and monetary policy in the short run and work force composition and flexibility in the long run. But both sides of the Nafta debate put imagined job gains or losses at the center of their talking points. The debate did little to convert anyone, but did lay the foundation for President Trump’s wildly inaccurate claim that Nafta was largely responsible for the loss of five million U.S. manufacturing jobs.

The Scorecard

Nafta’s friends and foes stress entirely different metrics in analyzing the pact. Supporters, who include most economists, generally gauge the success of a free trade agreement by how much trade it induces. By this yardstick, Nafta (and Cusfta) were huge successes.

U.S. trade with Mexico is now almost twice as large as could otherwise have been expected. U.S.-Mexican two-way trade in goods and services increased from $100 billion in 1993 to $583 billion in 2016. U.S. two-way trade with Canada spurted from $153 billion in 1988 (the year before Cusfta went into effect) to $240 billion in 1993 (though it did not show an outsized jump after Nafta).

By 2016, Canada and Mexico were the first and second largest markets respectively for U.S. exports. And they ranked just behind China as the sources of U.S. imports. Combined, the Nafta partners accounted for about a quarter of total U.S. exports and imports in 2016. Not surprisingly, the United States is by far the biggest market for Canadian and Mexican exports, absorbing 72 percent and 81 percent, respectively, of their foreign sales.

Foreign direct investment — meaning foreign ownership and control of a host-country enterprise — is another gauge of success since trade and FDI are intimately linked. In 2016, the stock of U.S. FDI in Mexico and Canada totaled $452 billion, up from $85 billion in 1993. Note, though, that this adds up to just 9 percent of U.S. global FDI and has grown less rapidly than U.S. investment in other countries. In fact, Mexico is home to only 2 percent of U.S. FDI — not a giant success, but also evidence that the reality of Perot’s “giant sucking sound” never matched the imagery.

Perhaps Nafta’s greatest achievement in terms of investment has been to increase Can­adian and Mexican FDI in the United States, to a total of $388 billion in 2016 — up from $42 billion in 1993. While Mexican and Canadian FDI in other countries has grown at about the same pace, part of the spectacular growth of total FDI from Canada and Mexico can reasonably be attributed to business lessons learned from Nafta.

Trade and FDI are inextricably linked in regional supply chains for unfinished goods, especially between the United States and Mexico; 40 percent of U.S. imports from Mexico and 75 percent of exports to Mexico consist of such “intermediate inputs.” And a large share of this trade takes place within firms, especially in the transport, electrical and machinery industries. Thus, Ford plants in the United States may import, say, dashboards from Mexico, even as they export transmissions to Ford plants in Mexico.

Total (inflation-adjusted) U.S. trade with its Nafta partners increased by 49.5 percent from 2003 to 2016. Over this same period, U.S. real (inflation-adjust) GDP increased by 25.5 percent. Assuming this GDP growth would have generated a proportional rise in trade without help from Nafta, the “extra” trade generated by Nafta (and, to be complete, the Uruguay Round global trade liberalization) totaled $797 billion. In our research with Lucy Lu at the Peterson Institute, we estimated that this amount of trade contributed $191 billion to U.S. GDP annually by increasing the efficiency of the economy — not a huge figure in an $18 trillion economy, but nothing to sneeze at, either.

Anticipating arguments from the other side, Nafta supporters contend that the United States is bound to run an overall trade deficit with the rest of the world because that trade deficit is the flip side of the reality that U.S. financial savings by households, business and government are negative — as they have been since 1971. Trade agreements may thus slightly change the pattern of bilateral surpluses and deficits, but exert little or no impact on the size of the United States’ overall deficit with the rest of the world.

The U.S. bilateral trade deficit with Mexico has indeed grown, going from a surplus of $4 billion in 1993 to a deficit of $58 billion in 2016. But this was not because of a “giveaway” by U.S. trade negotiators. As noted above, at the time Nafta became law, the average U.S. tariff on imports from Mexico was 4.3 percent, while the average Mexican tariff on imports from the United States was 12.4 percent. Since both tariff averages soon went to zero, the “loser,” as measured by tariff concessions, was Mexico.

It’s also important to put the bilateral deficit in context. From 1993 to 2016, the U.S. goods deficit with the world (excluding petroleum) expanded from $121 billion to $952 billion. This reflected the fact that the United States gradually altered its status from small net borrower from the rest of the world to a huge net borrower, a phenomenon largely driven by rising federal budget deficits and falling U.S. household savings. Mexico’s contribution was about 13 percent of the $952 billion total — about the same as Mexico’s share of U.S. merchandise imports. In short, Mexico simply does not account for a disproportionate slice of the U.S. trade deficit.

Nafta opponents, for their part, stress two other alleged adverse economic effects: loss of jobs (especially manufacturing jobs) in industries that compete with imports from Mexico, and suppression of U.S. wages owing to the fear that wage militancy would move production south of the border. Both effects are real; the crux of the debate, though, is their quantitative impact.

Lost jobs. An average of 52,000 Americans applied for aid under Nafta’s trade adjustment assistance program for workers displaced by imports from Canada and Mexico between 1994 and 2002 (the TAA database was not collected on a country basis after 2002). In these years, Nafta displacement represented less than 2 percent of total U.S. job displacements, which averaged over 3 million a year. Meanwhile, the U.S. economy as a whole did well in the first decade of Nafta, with annual growth of 3.3 percent and generally low unemployment.

The impact of Nafta on jobs after 2002 can be inferred using a different method. Between 2001 and 2014, a decrease of $1 billion in U.S. output (measured in 2016 prices) that corresponded to the composition of U.S. imports would have reduced U.S. employment (direct and indirectly) by about 8,300. On the other side of the ledger, an increase of $1 billion in representative U.S. exports over this period would have raised the job count by 6,900. Applying these estimates to bilateral trade with Mexico, rising U.S. imports displaced an average of 70,700 jobs per year between 2001 and 2016, while rising U.S. exports created 45,200 new jobs per year. By this metric, then, the net increase in the trade deficit with Mexico cost around 25,500 jobs a year — traumatic, no doubt, for many of the 25,500 who couldn’t quickly find comparable jobs, but a drop in the ocean in an economy in which 35 million workers change employment in an average year.

Consider, too, the other side of the ledger. For each of the 707,000 jobs displaced over a decade, the gains to the U.S. economy owing to enhanced productivity of the work force, a broader range of available goods and services, and lower prices at the checkout counter equaled about $270,000.

Reduced wages. A powerful charge leveled by Nafta critics is that trade with Mexico has enabled U.S. firms to contain the pace of wage gains — and in some cases allowed them to cut wages. Their argument is based on common sense. The current average manufacturing wage is $4.50 per hour in Mexico, compared with $19.50 in the United States. For the auto sector, the comparison is yet more stark: $7.79 versus $37.38. Taken at face value, this seems a powerful incentive to save on wages by shifting work to Mexico — or, at least, to threaten to move if their American workers demanded more pay.

Reality is more nuanced. A 2013 study of the U.S. labor market as a whole by David Autor et al. found that increased imports from Mexico and Central America had no significant effect on overall U.S. wages in manufacturing. A subsequent study targeting local labor markets by Shushanik Hakobyan and John McLaren did find that about 1.3 million American workers experienced wage growth retardation of at least 5 percent during the 1990s because of import competition from Mexico. Remember, though, these Nafta-impacted workers represented only about 1 percent of U.S. workers in 2000. Their losses were real, but they are tiny relative to the economy-wide gains from Nafta.

Political Impact

As with the economic side of the Nafta saga, supporters and opponents apply different metrics for assessing political effects. Supporters stress the constructive changes within Mexico and in U.S.-Mexican relations. Opponents emphasize the role of globalization and Nafta in stealing the American dream.

For the United States, possibly the biggest Nafta payoff was the creation of a new foundation for U.S.-Mexican relations. Unlike U.S.-Canadian relations, prior U.S.-Mexican relations can be characterized as cool at best. Nafta greatly improved the picture. Dramatic evidence of the fresh start: President Clinton’s leadership in assembling $50 billion in financial assistance for Mexico in the wake of the collapse of the exchange rate of the peso in 1994. By 1996, it’s worth noting, Mexico had repaid the loans in full, with interest.

Later, as drug-related violence in Mexico escalated (fueled, of course, by U.S. demand for illicit drugs), the United States provided — and Mexico welcomed — intelligence assistance and military aid to fight the cartels. The wisdom of waging the “war on drugs” can (and should) be questioned, but not the extent of cooperation between senior officials in Mexico City and Washington. Cooperation, incidentally, also improved on other issues ranging from agricultural inspection to border crossings.

Why did candidate Trump’s characterization of Nafta as a “total disaster” resonate so well with the voters in the 2016 election? The explanation can’t be found in opinion polls. In December 2002, nine years after Nafta went into force, a bit less than half (48 percent) of Americans thought the United States was better off with the pact. In 2017, after a long election campaign in which free traders were put on the defensive by both Trump and Bernie Sanders, public opinion remained much the same. Indeed, Americans viewed Nafta with slightly more favor: 51 percent said Nafta has been a good thing.

What did change in the decade preceding the 2016 presidential election were three broad features of the American economic landscape. Wages of American men who did not attend college stagnated. Demand for blue collar labor fell, with the manufacturing work force declining from 14 million in 2006 to 12 million a decade later. And after a pause in the middle of the decade for the financial crash, the share of household income going to the top 1 percent resumed its climb, reaching 21 percent in 2016. Meanwhile, the U.S. trade deficit, concentrated in manufactured goods, hovered around 3 percent of GDP and totaled a half-trillion dollars in the year of the election.

Candidate Trump skillfully channeled these facts into an indictment of U.S. trade policy. In his telling, Nafta was responsible for the bilateral deficit with Mexico (some $58 billion in 2016), killing tens of thousands of good American jobs and laying waste to the expectation that each generation would fare better in the labor market than their parents. As argued above, his indictment doesn’t hold much water. But in an environment of blue-collar frustration, it played well in Midwestern swing states.

Alternative Futures

In the quick-march negotiations that President Trump forced on Mexico and Canada, the diplomats should be able to agree on the text of several “modernization” themes, such as new rules for state-owned enterprises and digital trade, and tougher chapters on labor and environmental obligations. But the Trump administration has other demands that won’t be easy for Ottawa and Mexico City to swallow:

• Bilateral trade in goods should be “balanced,” implying that Mexico should unilaterally reduce its merchandise trade surplus with the United States, perhaps by government initiatives to restrict exports or to purchase more American goods.

• Tighter rules of origin should be imposed — this is, tougher limits on the portion of inputs made outside North America that could be included in a product to qualify for lower tariffs when entering the United States. (The goal: promoting U.S. production, especially of autos and parts, and textiles.)

• Canada and Mexico should open federal, state and provincial procurement to U.S. firms, with no comparable opening imposed on U.S. federal and state governments.

• U.S. firms should be protected by an independent arbitration system operating under international jurisprudence when they invest in Canada or Mexico. But firms from those countries that invest in America should have recourse solely to U.S. law.

• The special Nafta arbitration system for settling allegations of unfair trade practices should be abolished, and the current prohibition on invoking “safeguard” exceptions that would allow the United States to bar imports (even when imports from Mexico and Canada are very small) should be revoked.

We see four possible outcomes to a negotiation in which the United States’ objectives are vehemently opposed by Canada and Mexico.

Capitulation by U.S. partners. Conceivably — but only barely — Canada and Mexico might cave. This unlikely outcome is made more unlikely by the fact that Andrés Manuel Lopez Obrador, the leading candidate for the Mexican presidency in the July 2018 election, would surely reject this outcome given his lifelong populist credentials. Indeed, any significant concession offered by the sitting Mexican president, Enrique Pena Nieto, would very likely be renounced by whoever follows him, just as President Trump renounced the Trans-Pacific Partnership negotiated by Barack Obama. Nor could Canadian Prime Minister Justin Trudeau accept the U.S. demands and retain a firm grip on his Liberal Party.

Capitulation by the United States. A converse outcome (equally unlikely): the Trump administration abandons its protectionist objectives and instead concentrates on “modernizing” Nafta. That scenario would include stronger enforcement of labor and environmental standards, requirements that state-owned enterprises follow the same rules as private producers, commitments to the free

flow of digital commerce, and liberalization of trade in services such as finance, telecommunications, broadcasting, education and health. Many of these changes would serve the interests of U.S. labor as well as consumers, and thus would fit into a rational populist agenda. But since President Trump has condemned Nafta and made gutting it part of a broader anti-Mexico agenda, he could hardly applaud a merely “modernized” Nafta.

Termination. Trump threatened during the campaign (and again on April 27, 2017) to withdraw from Nafta if the negotiators didn’t meet his demands. Without Nafta, Canada could still rely on the Canada-U.S. Free Trade Agreement of 1989, which was suspended but never revoked when Nafta took effect in 1994. Even so, investment in Canada would suffer since firms would wonder whether President Trump’s next step, after terminating Nafta, might be terminating Cusfta.

Mexico would be hit much harder than Canada in terms of investment inflows since a great deal of FDI in Mexico — now running about $30 billion annually — is oriented toward production for the U.S. market and since Mexico’s only fallback agreement on the trade front (if Nafta were terminated) is the World Trade Organization rules that would deter the United States from specifically discriminating against Mexican exporters. Termination would presumably also jeopardize cooperation with the United States on terrorism, drug trafficking, immigration and other important issues.

But the economic and political consequences for the United States of a flat-out break with Nafta would also be substantial. Termination would disrupt the commercial lifeblood of numerous firms and communities that directly depend on commerce with Mexico or are indirectly dependent on the intricate supply chains that serve industry in all three countries. Many of these firms and places, moreover, are in states that supported candidate Trump. And it is difficult to imagine a scenario in which Congress casually allowed the Trump administration to trash the interests of a whole swath of American business.

Muddle through. Between these extreme scenarios is a “muddle through” scenario — one that papers over differences and avoids a confrontation or a surrender. There’s precedent in the bitter (but inconclusive) trade battles with Japan during the 1980s and early 1990s.

In a muddle-through scenario, Mexico and Canada would offer concessions that would not only please the Trump team but also serve their own broad economic interests — for example, larger purchases of natural gas by Mexico and fewer restrictions on dairy imports by Canada. By combining such “deliverables” with modernization themes, and by making announcements of progress from time to time in the months ahead, Canada and Mexico might elicit patience from the Trump team. At some point, the logic goes, both President Trump and his angry blue-collar constituents would become distracted, allowing the U.S. negotiators to declare victory and retire from the field of battle with much sound and fury signifying little.

This is not a scenario that would satisfy anyone. But politics and diplomacy are full of problems for which the most practical solution is no solution at all. This is probably one of them.