Issue Brief #238
Two numbers have been introduced recently into the never-ending trade debate in American politics: $1 trillion and $500 billion. The first number ($1 trillion) is a purported measure of how much past trade agreements have added to the U.S. economy. The second number ($500 billion) is a guesstimate of how much future agreements will add to American incomes. The usefulness of these numbers to the cause of maintaining the globalization status quo has unfortunately given them a prominence unjustified given their unreliability.
Even, for example, the office of the United States Trade Representative (USTR), the chief trade negotiating body for the U.S. government, seems to take them seriously.
Today, U.S. annual incomes are $1 trillion higher, or $9,000 per household, due to increased trade liberalization since 1945.1
Both numbers are derived from a 2004 study by Bradford, Grieco, and Hufbauer (BGH), published by the Peterson Institute for International Economics. BGH undertook no original number-crunching in arriving at these numbers (this is no sin—review and interpretation definitely count as research); instead these numbers are derived from their review of a number of independent studies.
A skeptical reading of the BGH results on prospective gains (the more policy-relevant issue) has already been offered in Bivens (2007). The retrospective benefits, while less policy-relevant, are still worth examining given their prominence in recent debates. A fuller treatment of the BGH results, both prospective and retrospective, can be found in Bivens (2007b). This Issue Brief highlights some of their findings about the past benefits of trade liberalization.
• Their estimated gain of $9,000-10,000 per household relies on a consistently and extraordinarily generous reading of a number of independent studies—not one of which actually claims that benefits to the U.S. economy from liberalization are close to the scale expressed by BGH.2
• One of the highest-quality studies reviewed by BGH, besides being applied incorrectly by the authors, shows that trade liberalization from 1982 to the present (a time that saw the passage of the North American Free Trade Agreement (NAFTA), the creation of the World Trade Organization (WTO), and the passage of permanent normal trading relations with China) added only $9 per U.S. household, not $9,000.
• BGH low-ball the costs of trade liberalization by omitting its primary component—slower wage-growth for all U.S. workers who resemble (in terms of skills, education, and credentials) those workers directly displaced by imports.
A generous review of the evidence
The underlying studies that BGH rely upon to estimate the retrospective gains from liberalization are of varying quality (from decent to excellent). BGH give the impression that a large number of separate studies came to strikingly common conclusions, lending robustness to the central finding. This is not true—BGH’s reading of these studies came to a strikingly similar conclusion about each one. In fact, not a single one states benefits to the United States in anything close to the range expressed by BGH. Many of the studies do not even attempt to measure the gains from trade liberalization.
In an important sense, arguing over the BGH numbers about the past benefits of trade liberalization is an absurd exercise. The absurdity rests in their choice of a relevant policy baseline that is universally considered bad for the U.S. economy: the prohibitive Smoot-Hawley tariffs passed in 1930.3 Nobody today argues for anything like a return to Smoot-Hawley level of trade protection.
Nevertheless, the fact that the authors so greatly overstate even the benefits of moving away from the universally condemned Smoot-Hawley trade regime should give rise to skepticism about their more intellectually contested claims. Hence, it is useful to examine each of the BGH claims in some detail. Bivens (2007) provides a critical overview of almost every study that BGH mentions in their paper and provides much more detail about their reading of the evidence. The rest of this Issue Brief will examine just two of the studies reviewed by BGH and will provide a flavor of their over-aggressive reading of evidence.
The OECD study
The first study referenced by BGH is from the Organization of Economic Cooperation and Development (OECD 2003). This study uses growth in the trade share of overall gross domestic product (i.e., imports plus exports divided by GDP) as an explanatory variable in a statistical analysis of the causes of GDP growth, based on a dataset of OECD countries over time.
The OECD authors find a statistically and economically significant correlation between rising trade shares and GDP growth: a 1% increase in trade flows as a share of GDP is associated with a 0.4% increase in GDP per capita. However, as even the OECD authors note, this statistical relationship is an inadequate and misleading way to measure the causal impact of trade on growth, for a couple of reasons.4
First, the causality is far from clear regarding the statistical link between trade and growth. Economic theory provides at least as many reasons to think that higher growth causes higher trade shares as it does the reverse.
Second, the trade share of GDP is an almost wholly uninformative measure of a nation’s policy stance. There is a very important thing to keep in mind regarding all of BGH’s numbers: they are not about the benefits of expanded trade or globalization writ large. Instead, BGH’s numbers are about the benefits of trade liberalization, that is, using the specific policy lever of reducing domestic barriers to foreign commerce.5 This is a much more restricted ground. Could globalization have added $9,000 to each U.S. household in 2005? Maybe, but probably not. It is a big claim and would be awfully hard to measure. But, the question at issue is: has trade liberalization—the removal of trade barriers—added this much to the American economy? Definitely not.
Falling trade costs: technology or policy
BGH further misreport the implication of the OECD study by attributing reductions in the cost of trade to trade liberalization when in fact they were also driven by other influences (technological change, reduced transport costs, etc.).
BGH cite Cline (1997) for estimates on the role of falling trade costs in driving up trade shares over time. BGH claim the Cline (1997) results as validation for their assumption that half of the total increase in trade shares identified by the OECD study was driven by trade liberalization. However, Cline’s (1997) estimates on this issue included trade costs that fell due to technology (transport costs), not just policy changes.
In short, BGH’s report of findings from their analysis of the OECD study is almost a lesson in how not to analyze the impacts of trade policy on growth. They seem to have failed to heed the cautions of the OECD authors, who wrote:
The possible reverse causality problem in the relationship between trade and economic growth suggests some caution in interpreting empirical results. In particular, we treat the intensity of trade in the growth equation as an indicator of trade exposure – capturing features such as competitive pressures – rather than one with direct policy implications.
Bernard, Jensen, and Schott
Two of the most careful studies cited by BGH are papers by
Bernard, Jensen, and Schott (2003, 2004). In their work, Bernard, Jensen, and Schott (BJS) estimate the effect of falling trade costs (either policy or technology driven) on productivity growth in manufacturing firms. One finding from their research finds that a 1% fall in trade costs results in a 1% increase in productivity.
BGH apply this finding by taking the 40% tariff they posit as characterizing the 1947 U.S. economy and arguing that manufacturing productivity is 40% higher today because of trade agreements signed in the past, which translates into roughly $600 billion in benefits stemming from liberalization.
A careful reading of BJS, however, reveals that this is not a proper interpretation of their results. BJS do not find a statistically significant effect of falling trade costs on aggregate manufacturing productivity. Instead, they find this relationship in a sample that is (a) restricted to trade with OECD (generally rich) countries only, and (b) within industries characterized by high levels of intra-industry trade—a sample that includes only 65 out of a possible 450 industries.
In short, BJS identify the productivity-enhancing effects of falling trade barriers as characterizing only trade with rich nations in industries with lots of intra-industry trade. This is an interesting finding, but it cannot be applied to the total manufacturing sector.
Lastly, and perhaps most telling for assessing the relevance of this work to contemporary debates about globalization, BGH themselves show that literally 99.9% of the gains from trade liberalization they infer from the BJS work occurred before 1982. This makes sense given that by the early 1980s U.S. tariffs were already extremely low. After 1982, the BJS research suggests that trade liberalization has added all of $9 (yes, $9) per U.S. household (see table 2.4 in BGH).
Several significant (and politically charged) changes in trade policy have occurred since 1982, including: the completion of the Uruguay Round of the General Agreement on Tariffs and Trade (GATT); the formation of the World Trade Organization (WTO, which replaced GATT); the passage of the North American Free Trade Agreement (NAFTA, along with many other bilateral agreements); and the permanent normalization of trading relations with China and its entry into the WTO. Today’s political arguments over globalization are almost wholly about what has happened since 1982. Given this, it seems that the more relevant number from BGH regarding trade’s contribution to household incomes in the politically contested era of globalization is $9, not $9,000.
Conclusion
Costs and benefits, net and gross
For those who take trade theory seriously, it is particularly surprising to see that BGH’s characterization of the group who lose from expanded trade (concentrated) and the scale of this group’s gross losses (small relative to net gains) is literally the opposite of what is implied by mainstream trade theory for an economy like the United States. Mainstream theory argues instead that trade (particularly trade between the rich U.S. economy and the poorer members of the global economy) is win-win at the country level but lose-win for productive factors (labor and capital, for example) within countries. For a review of the distributional outcomes of global integration in the United States, see Bivens (2007).
The essential problem with the BGH study is that they only tally the costs borne by those workers directly displaced by trade (for example, steel workers laid off because competition from imported steel forced their plant to close). BGH do not acknowledge the wider wage costs of global integration (lower wages for landscapers as they are forced to compete with laid-off steelworkers). This is an effective (if false) way to minimize the total debit column in the balance sheet of globalization’s impact on American workers.
Lastly, and most important, mainstream trade theory predicts that the larger the net gains from rich/poor trade, the larger the re-distribution of income and the larger the gross losses suffered by the (majority) losing group. Even worse, this theory predicts that more income will be redistributed than created because of trade. Thus, it is grossly inaccurate to refer to the losses caused by trade as either small or concentrated.
Not $1 trillion
The BGH claim that past trade liberalization has added $1 trillion to the U.S. economy cannot be sustained by the evidence. Good policymaking requires honest and open debate based on solid empirical foundations. BGH’s numbers do not fit in this debate and should be retired from public discussion. Their estimates are derived from a consistently aggressive interpretation of a small sub-sample of the literature on the gains from trade liberalization. It is telling that they ignore some of the most famous and respected empirical investigations of the question (such as Rodriguez and Rodrik (2001), Frankel and Romer (1999), and Sachs and Warner (1995)). These studies tend to show growth benefits from liberalization, but they are nowhere in the neighborhood of the BGH reported results.
Finally, even their own numbers argue that the economic gains from the primary issue at stake in today’s globalization debate—the integration of the U.S. economy with a much poorer global economy—are actually very small. Finding a way to manage this integration in a way that helps workers on both sides of the equation will be as difficult as it is important.
Endnotes
2. In popular discussion, the $1 trillion number has often been expressed as either $9,000 or $10,000 per U.S. household. There are 120 million U.S. households, so the correct quotient of $1 trillion divided 120 million is closer to $9,000. That is the number this paper will use.
3. For those readers interested in a brief overview of Smoot-Hawley Tariff Act of 1930, this Department of State Web site gives a decent summary: http://www.state.gov/r/pa/ho/time/id/17606.htm
4. For exhaustive reasons why, see Rodriguez and Rodrik (2001) and Frankel and Romer (1999).
5. As a concrete example of why using trade shares of GDP as a measure of policy-openness is ill-advised, consider the comparison between Vietnam and the United States. The former has a trade share that is 200% larger than the United States, yet nobody would argue that the United States is a bastion of protectionism relative to Vietnam—in fact, average tariff rates are roughly five to six times as high in Vietnam as in the United States, and the USTR has noted that non-tariff barriers are rife in the Vietnamese economy.
References
Bernard, Andrew B., J. Bradford Jensen, and Peter Schott. 2003. Falling Trade Costs, Heterogeneous Firms and Industry Dynamics. IFS Working Paper 03/10. London: Institute for Fiscal Studies.
Bernard, Andrew, and J. Bradford Jensen. 2004. Exporting and Productivity in the U.S. Tuck School of Business at Dartmouth, National Bureau of Economic Research, and the Institute for International Economics. Photocopy (July).
Bivens, L. Josh. 2007. Globalization and American Wages: Today and Tomorrow . EPI Briefing Paper #196. Washington, D.C.: Economic Policy Institute.
Bivens, L. Josh. 2007b. The gains from trade: How big and who gets them? EPI Working Paper. Washington, D.C.: Economic Policy I
nstitute.
Bradford, Scott, Paul Grieco, and Gary Hufbauer. 2005. “The Payoff to America from Global Integration,” in C. Fred Bergsten, ed., The United States and the World Economy: Foreign Economic Policy for the Next Decade. Washington, D.C.: Peterson Institute for International Economics.
Cline, William. 1997. Trade and Income Distribution. Peterson Institute for International Economics Policy Brief. Washington, D.C.: Peterson Institute for International Economics.
Frankel, Jeffrey A., and David Romer. 1999. Trade and growth: an empirical investigation. American Economic Review. Vol. 89, No. 3. June. pp. 379–99.
OECD. 2003. The Sources of Economic Growth in OECD Countries. Paris: Organization for Economic Cooperation and Development.
Rodriguez, Francisco, and Dani Rodrik. 2001. “Trade Policy and Economic Growth: A Skeptic’s Guide to Cross-National Evidence,” in Ben S. Bernanke and Kenneth Rogoff, eds., NBER Macroeconomics Annual 2000. Cambridge, Mass.: MIT Press.
Sachs, Jeffrey, and Andrew Warner. 1995. Economic Reform and the Process of Global Integration. Brookings Papers on Economic Activity. Vol. 1, pp. 1-118.
