Brad Delong’s Case for NAFTA: Based on Assumptions, Not on Data

Writing at the Equitable Growth blog, Brad DeLong takes on Jeff Faux’s assessment of NAFTA. DeLong tries to make a “cosmopolitan” argument for NAFTA; he claims that NAFTA cost the United States fewer than 350,000 jobs, and says that Faux got “the analysis wrong” when he reported that NAFTA resulted in a net loss of 700,000 jobs. But DeLong’s “analysis” is based on faulty data. He then goes on to assert that the jobs gained through increased exports pay more than the jobs lost due to growing imports, which also turns out to be wrong. Lastly, he ignores the larger and much more important negative impact of growing trade with low-wage countries on the wages of all U.S. workers without a college degree.

President Bill Clinton and the economists he relied on built their case for NAFTA on the assertion that the United States would gain jobs as a direct result of the agreement. He claimed that NAFTA would create an “export boom to Mexico” that would create 200,000 jobs in two years and a million jobs in five years, “many more jobs than will be lost” due to rising imports. This is the central argument used to justify NAFTA by its proponents, so it’s entirely appropriate to evaluate its impact by examining its net impacts on U.S. employment.

The economic logic behind Clinton’s argument was clear: Trade creates new jobs in exporting industries and destroys jobs when imports replace the output of domestic firms. Exports support domestic jobs and production, and imports displace domestically produced goods, displacing existing jobs and preventing new job creation. Clinton assumed, without much evidence, that the net employment effect would be positive. But unfortunately, it wasn’t.

DeLong begins his jobs analysis by noting that trade would have grown with or without NAFTA. But this ignores the status quo ante. U.S.-Mexico trade was roughly balanced in the pre-NAFTA period. If exports and imports had both doubled, our bilateral trade would still have been balanced. Instead, imports grew faster than exports, so the United States developed a significant, job destroying trade deficit with Mexico. We differ over the size of this effect, which I’ll get to in a moment, but the key question is why that trade deficit developed. As I’ve shown elsewhere, the growth of outsourcing, and a near-tripling of foreign direct investment (FDI) in Mexico, were the principle causes. NAFTA created a unique set of investor protections that encouraged multinationals to shift production from the United States to Mexico. The growth of FDI in Mexico overshadowed any impact of tariff cuts (which were larger in Mexico than the United States).

DeLong develops ballpark estimates of the numbers of jobs gained and displaced by U.S. trade with Mexico after NAFTA, using gross trade data from the St. Louis Fed for total U.S. exports and imports of goods to and from Mexico. I have estimated that trade deficits with Mexico displaced 682,900 U.S. jobs in 2010 (the United States had a small trade surplus with Mexico in 1993, before NAFTA took effect, so our best estimate is that growing trade deficits displaced about 700,000 U.S. jobs between 1993 and 2010, as noted by Faux). Our analysis is based on trade data from the U.S. International Trade Commission (USITC) and an employment requirement matrix from the U.S. Bureau of Labor Statistics.

My estimates of the jobs supported by U.S. exports are based only on U.S. domestic exports to Mexico—goods produced in the United States with U.S. labor. Total exports also include a growing share of transshipments, or foreign exports (aka re-exports), goods produced in other countries (e.g., China) that are imported into the United States and re-exported to other countries (e.g., Mexico). Foreign exports do not support domestic employment. Data on transshipments are almost always excluded from analyses of the impacts of trade and trade agreements, such as those prepared by the USITC, as noted by agency economists. And the exclusion of transshipments makes a big difference in NAFTA trade, as shown in the graph below (my jobs analysis is also based on imports for consumption rather than general imports, but this distinction has no significant impact on job loss estimates).

U.S. exports to Mexico, both total and domestic, have increased since NAFTA took effect, but the gap between them (i.e., foreign exports) has increased much more rapidly. Imports (not shown) have also increased faster than exports, and as a result, the U.S. trade deficit with Mexico has increased steadily since NAFTA took effect.

The U.S.-Mexico trade deficit in 2013 (and hence the jobs displaced by trade) when properly calculated, using domestic exports, was $96.2 billion, nearly twice as large as the deficit estimated with total U.S. exports ($54.4 billion). The difference in estimated trade balances largely explains why my estimate of the jobs displaced by NAFTA (700,000) is twice as large as DeLong’s (350,000). The centerpiece of his critique of Faux’s analysis is based on faulty data, but that’s just where the problems begin.

Job losses are just the tip of the iceberg for domestic workers

DeLong asserts that even if NAFTA did result in job losses, some workers gained because the jobs gained through increased exports are “better jobs with higher pay” than the jobs lost due to higher imports. But this assertion is not backed up with any data. I have compared jobs created and displaced by NAFTA and found that the facts do not support this assumption. In a 2006 report I showed that jobs displaced by imports from Mexico paid $813 per week, while jobs supported by exports to Mexico paid only $799 per week, 1.8% percent less. More importantly, with respect to the 700,000 net jobs lost, even when re-employed in non-traded industries, workers earned only $683 per week, 19 percent less than the jobs displaced by imports. Wage losses associated with trading good jobs in import-competing industries for lower paying jobs in exporting and non-traded goods industries cost U.S. workers $7.6 billion in 2004.

My results for Mexico trade are not anomalous. I found similar and even stronger results for U.S.-China trade. Average weekly wages of jobs displaced by imports from China between 2001 and 2011 were 17.0 percent higher than average wages in jobs supported by exports to China. Workers in non-traded industries earned 9.4% less than workers in exporting industries. Growing trade deficits with China cost 2.7 million jobs between 2001 (which China entered the WTO) and 2011, resulting in $37 billion in lost wages in 2011 alone.

The analysis of wages paid to workers in import, export and non-traded industries holds everything else constant. It does not reflect the much more important, economy-wide impacts of trade on the wages of working Americans, the well-known Stolper-Samuelson effects. Direct wage losses for workers displaced by NAFTA and China trade are just indicative of the larger impact of trade on wages for working Americans. Most traded goods are manufactured products, production of which disproportionately employs non-college-educated workers. Trade with low-wage countries like Mexico and China puts manufacturing workers, and everyone with a similar skill set, into competition with low-wage workers abroad.

Josh Bivens has shown that growing trade with less developed countries (LDCs) has been responsible for large shares of the rapidly growing college–non-college wage gap. That overall wage gap increased 22.0 percentage points between 1979 and 2011. Bivens’ model estimates the broad impacts of trade on all non-college educated workers in the United States, who number about 100 million. His study shows that growing LDC trade “lowered wages in 2011 by 5.5 percent—or by roughly $1,800—for a full-time, full-year worker earning the average wage” for such workers. Trade with low-wage countries can explain roughly a third of the overall rise since 1979 in the wage premium earned by workers with at least a four-year college degree relative to those without one. However, trade with low-wage countries explains more than 90 percent of the rise in this premium since 1995. China and Mexico are the United States’ two largest low-wage trade partners.

It’s important to note that trade was not the only cause of growing inequality over the past three decades or so (since the late 1970s). Growing inequality in this era was the result of policy choices on behalf of corporate interests including macroeconomic policy (e.g., too-tight monetary policy leading to an increase in average unemployment levels), trade agreements, deregulation of the financial sector, attacks on the legal foundations of organized labor, declines in the real minimum wage, deregulation of many industries and privatization of the public sector, and other policies that have helped some workers and hurt others. Many economists and policy makers blame technology instead (more specifically, skill biased technical change), but there is mounting evidence against this view. Rapid technological progress has been a hallmark of the American economy for generations, but massive growth in inequality began only in the late 1970s, and is strongly correlated with the policy choices noted above.

DeLong argues that the United States, as a “hyperpower,” has a “strong moral obligation to the world” as a whole to support trade and investment deals like NAFTA. We can certainly agree that NAFTA was a much bigger deal for Mexico than it was for the United States. But it was a bad deal for Mexico. DeLong asserts that NAFTA “boosted employment in Mexico by 1.5 million” (3 percent of the Mexican labor force). But most of the jobs gained must have been in manufacturing, since about 80% of Mexico’s exports are manufactured products. However, Mexico only added about 400,000 manufacturing jobs between 1993 and 2013, and this is probably an overestimate of the employment gains from trade because it doesn’t take into account job losses in Mexican agriculture due to opening of grain trade.  Furthermore, rates of growth in per capita GDP in Mexico fell from 3 percent per year in the period between the 1940s and the 1970s to 1 percent after it liberalized trade in the late 1980s and then joined NAFTA. Again, the data conflict with DeLong’s theoretical assumptions. If NAFTA was a bad deal for workers in Mexico, why should policymakers in the United States support trade and investment deals that cost jobs and reduce wages for most working Americans?

DeLong expresses consternation at the “energy the American left poured and pours into the anti-NAFTA cause.” He, of course, is energetically still trying to defend policies that he supported while serving as a deputy assistant secretary of the U.S. Treasury between 1993 and 1995, when NAFTA (and the WTO) were created. But it’s clear that NAFTA failed to measure up to the claims on which it was sold to the American people and to Congress and has been a significant contributor to growing inequality, one of the most pernicious problems of the past two decades. Furthermore, DeLong does not dispute the fact that NAFTA cost jobs in the United States, and only quibbles about the numbers. This only raises the question, if trade hurts the domestic economy, why should we supercharge it?

DeLong also argues that if NAFTA increased unemployment and lowered wages, then we should just fix our macroeconomic policies, including “exchange rate policies” to change them. But this illustrates just how flawed NAFTA was—a two thousand page document that allowed corporations to sue host governments over any laws that might possibly threaten their profits, without regard to national interest. Somehow, nothing was included in the NAFTA, WTO, or other major, U.S. trade and investment deals about exchange rate policy, perhaps the single most important determinant of trade balances (and imbalances), resulting in the loss of millions of jobs due to growing trade deficits with Mexico and China, alone.

The continued opposition of “the American left” to NAFTA and similar deals is easily explained. It’s because the self-styled technocratic center insists on shoving these agreements through the system every chance they get. In 2009 and 2010, in the absolute teeth of the Great Recession (or as DeLong calls it, the Lesser Depression) we got the Korea, Panama and Columbia trade and investment deals shoved down the throats of Congress and “the left.” And it’s well to recall that in 1993, Clinton decided to make NAFTA, rather than health care, the centerpiece of his policy agenda for what became a notably unproductive 8-year term. Despite the well-known failures of past trade and investment deals like NAFTA, the Obama Administration has made negotiating new, job killing agreements like the U.S. Korea Free Trade Agreement Free Trade Agreement and the proposed Trans-Pacific Partnership a centerpiece of its economic policies, based on its claim that such deals will create tens of thousands of “American jobs through increased exports alone.”

NAFTA has had real and significant costs for the vast majority of working Americans, and it never delivered the promised benefits for Mexico. Working Americans and “the left” have very good reasons for opposing such deals and should continue to do so.


  • benleet

    The minimum wage was recently posted on Nov. 2013, at $5.08 per day in region A, and $4.63 in region B. The average worker salary is $12.50 a day, or $3,258 a year or $1.56 an hour — the average not the minimum which is 63 cents and hour. Purchasing Power Parity may lift this by 40%. Mexican per capita GDP was $10,307 (World Bank figure 2012), and that also equals to around $30,000 per worker per year, or $15 an hour or $120 a day — not a $12.50 a DAY. See these articles: http://www.jorada.unam.mx/ultimas/2013/11/27/promedio-salarial-en-mexico-es-de-2-5-veces-el-minimo-unt-4180.html — and this one: http://www.usleap.org/usleap-campaigns/labor-rights-mexico

    “Approximately 90% of collective bargaining agreements in Mexico are signed by protection unions, which are company-backed and prevent real unions from forming.”