Dani Rodrik's Numbers Game
Response to Dani Rodrik's blog
© Peterson Institute for International Economics
Gary Clyde Hufbauer is a senior fellow and Scott C. Bradford is an associate professor at Brigham Young University. Paul L. E. Grieco is a graduate student at Northwestern University's Department of Economics. Agustin Cornejo provided valuable research assistance.
In his weblog for May 7, 2007 (http://rodrik.typepad.com/), Dani Rodrik takes us to task for exaggerating the benefits of globalization in raising US household income and GDP. Professor Rodrik long ago established his reputation as a globalization skeptic; today he is the favorite Harvard economist among the backlash crowd. Eight years ago, Francisco Rodriguez and Rodrik (1999) notched their academic guns against Jeffrey Sachs and Andrew Warner (1995), expressing a skeptical view of the benefits of liberal trade policy to developing countries. As targets of Rodrik's latest outburst of skepticism, we share good company.
To debunk globalization proponents, Rodrik uses arithmetic that can be traced to Frank Taussig (1927) as a "reality check" on our calculations. Taussig was a great economist, but economic science has progressed since 1927. The partial equilibrium formula cited by Rodrik essentially confines the benefits of globalization to the "welfare triangles" created when tariffs are abolished. According to Rodrik, using arithmetic that embodies flawed assumptions, the welfare triangles are small, totaling 0.25 percent or less of US national income, around $35 billion of potential gains if all US tariffs were abolished. In a personal attack on Fred Bergsten and Sebastian Mallaby ("Cheerleaders Gone Wild"), Jared Bernstein and Josh Bivens of the Economic Policy Institute embrace the same flawed arithmetic, claiming that "reasonable estimates" of potential US gains range between $4 billion and $20 billion (http://www.huffingtonpost.com/jared-bernstein-and-josh-bivens/ cheerleaders-gone-wild_b_45405.html). By contrast, our conservative estimate suggests that full global liberalization would ultimately increase US national income by about 4.1 percent of GDP, about $570 billion based on US income in 2007. Even on its own terms, Rodrik's application of welfare triangle analysis is biased downward for reasons spelled out in Appendix A to this note. The more serious problem, however, is that welfare triangle analysis misses the big story.
Done properly, welfare triangles are fine for examining the cost of protecting individual products in otherwise perfectly competitive and undistorted markets. We have used the tool extensively in examining steel quotas, textile and apparel tariffs, sugar duties, and other barriers (e.g., Hufbauer and Elliott 1994). But this sort of analysis completely misses multiple forces that enormously expand the payoff from policy liberalization and technology innovation for a country that participates in a global economy.
What are these forces? "Right sizing" inputs to the needs of industrial producers; lowering the true cost of household purchases below the advertised inflation rate; "sifting and sorting" firms so that the most efficient expand and the least efficient shrink; curtailing the markup margins associated with monopolistic competition (yes, even in the giant US economy); stimulating laggard industries (think autos and steel) to match the productivity of foreign competitors; reducing the enormous international differences that prevail in prices for traded goods; and enjoying the benefits of increasing returns to scale. In our chapter that Rodrik criticizes (Bradford et al. 2005), we used alternative methods to calculate some, but not all, of these effects. The literature since our chapter was published points to additional channels; for example, Arnold, Javorcik and Mattoo (2007) find that services liberalization enhanced the productivity of downstream manufacturing firms in the Czech Republic.
Perhaps sensing that his own back-of-the-envelope calculations cannot be trusted, Rodrik cites the results of a computable general equilibrium (CGE) study carried out by Kym Anderson, Will Martin, and Dominique van der Mensbrugghe (2005, 2006). These respected World Bank economists suggest that global free trade in merchandise would increase US incomes by just 0.1 percent in 2015, an estimate that falls well below the median of other CGE models. Though saying he has no idea whether the World Bank number is right, Rodrik seems to regard the 0.1 percent figure as confirmation for his own "reality check" . So it is worth exploring the World Bank model a bit further.
The basic flaw in the World Bank CGE model is that it omits virtually all the important channels of growth opened by policy and technology liberalization-the forces we have already enumerated-and it also omits services from the story. Services are the fastest growing component of global commerce and promise the biggest payoff from future liberalization. The only "dynamic" feature in the model is its ability to calculate year-by-year results. In reality, the World Bank CGE model is a glorified version of Taussig's framework, made vastly more complex thanks to computer technology, but limited to merchandise trade under static conditions. For a truly dynamic CGE model that reflects forces set in motion when an economy liberalizes, and covers services as well, interested readers should instead consult the University of Michigan model, designed by Drusilla Brown, Alan Deardorff, and Robert Stern (e.g., Kiyota and Stern 2007). Similar to our calculations, the Michigan modelers estimate that global free trade in goods and services would increase US national income by 3.4 percent.
Rodrik makes a big deal of increasing returns to scale and implicitly accuses us of using this feature to cook the books. Increasing returns may be important in the real world, but they play a minor role in our 4.1 percent figure. Rodrik further claims that our calculated price gaps are caused mostly by things other than trade barriers. He provides no evidence for this claim even though published studies make the case that trade barriers are substantial (Bradford 2003, Bradford and Lawrence 2005).
To embellish his argument, Rodrik credits us with several straw men and ridicules the whole lot. Supposedly, according to Rodrik, we believe that globalization "will yield full price convergence"; supposedly we urge the United States to harmonize its laws with those of its trading partners and join them in a global currency union; and supposedly we welcome 80,000 pages of EU-style regulations. All this is nonsense.
Finally, Rodrik takes us to account for misstating Andrew Rose's (2003) gravity model as to the increase in merchandise trade created by a regional trade agreement (RTA). As we explained (Bradford et al. 2005, p. 93, fn. 60), we used the most conservative RTA coefficient calculated by Rose (middle column, Rose's table 1), namely 0.78 (exp0.78 - 1.00 = 118%), which we then discounted by our own estimate of trade diversion. Subsequent gravity model studies, reflecting additional free trade agreements besides those included in Rose's original data set, buttress our expectation that global free trade would substantially boost world commerce (e.g., DeRosa and Gilbert 2005). Speaking of gravity models, these modern econometric "work horses" demonstrate that free trade agreements enlarge bilateral commerce to a much greater extent than predicted by Frank Taussig's framework. The payoff of FTAs, in terms of boosting two-way merchandise trade flows, are often two to three times as great as the standard textbook trade theory might suggest.
It makes an enormous policy difference whether you think the gains to the United States and to the world at large, from total free trade in goods and services, would be only 0.1 percent of GDP, or instead could reach 4 percent of GDP. If gains are as small as 0.1 percent, why bother?-the skeptics might ask. Forget about the Doha Development Round and put those pesky free trade agreements (300 and counting) on the shelf. If gains are as large as 4 percent, the payoff is handsome: That's right at the top of what governments can deliver through economic policy reform.
In this debate, we are glad to be on the same side as Fed Chairman Bernanke and Treasury Secretary Paulson. Free trade remains our prescription.
Appendix A Technical Flaws in Rodrik's Formula
Rodrik's formula depends on three crucial assumptions that bias the result downward. First, the formula assumes that governments use only tariffs. The protection package usually contains nontariff barriers (NTBs) such as agricultural subsidies and procurement restrictions. The more these NTBs are used, the more the formula's key number is biased downward. Second, the formula assumes that protection is the same for all goods and services. When properly accounting for the actual peaks and valleys in the protection profile across sectors, the estimated cost of barriers becomes much higher, typically two or three times as high. Third, the elasticity of import demand tends to increase as protection levels rise and import quantities fall. Taking account of this feature can also yield higher estimates of protection costs. In short, using a crude formula as a "reality check" on careful applied calculations is like using a yardstick to check the tolerance of a wristwatch.
Even using a simplistic formula, Rodrik's analysis compares apples and oranges. His "reality check" of 0.25 percent of GDP reflects the static cost of US tariff barriers imposed only on merchandise trade. Our number of 4.1 percent (which Rodrik compares with his 0.25 percent) reflects a comprehensive analysis that includes US NTBs on merchandise trade, US barriers on services trade, foreign barriers of all kinds, and multiple dynamic effects thwarted by assorted barriers here and abroad. If all these features of our analysis are stripped away, we would be left with an estimate near 0.5 percent of US GDP. This is about what one would expect from a proper static formula that accounts for differences in protection across sectors. But why focus on a misleading comparison when more robust analysis is available?
Anderson, Kym, Will Martin, and Dominique Van der Mensbrugghe. 2005. Market and Welfare Implications of Doha Reform Scenarios. In Trade Reform and the Doha Agenda, ed. K. Anderson and W. Martin. Washington : World Bank.
Anderson, Kym, Will Martin and Dominique van der Mensbrugghe. 2006. Doha Merchandise Trade Reform: What's at Stake for Developing Countries? World Bank Policy Research Working Paper 3848 (February).
Arnold, Jens Matthias, Beata Smarzynska Javorcik and Aaditya Mattoo. 2007. Does Services Liberalization Benefit Manufacturing Firms? Evidence from the Czech Republic . World Bank Policy Research Working Paper 4109 (January).
Bernstein, Jared, and Josh Bivens. 2007. Cheerleaders Gone Wild. (http://www.huffingtonpost.com/jared-bernstein-and-josh-bivens/ cheerleaders-gone-wild_b_45405.html).
Bradford, Scott C. 2003. Paying the Price: Final Goods Protection in OECD Countries. The Review of Economics and Statistics 85, no. 1: 24-37 (February).
Bradford, Scott C., and Robert Z. Lawrence. 2004. Has Globalization Gone Far Enough? The Costs of Fragmented Markets. Washington : Institute for International Economics.
Bradford, Scott C., Paul L. E. Grieco, and Gary Clyde Hufbauer. 2005. The Payoff to America from Global Integration. In The United States and the World Economy: Foreign Economic Policy for the Next Decade, C. Fred Bergsten and the Institute for International Economics. Washington : Institute for International Economics.
DeRosa, Dean A., and John P. Gilbert. 2005. Predicting Trade Expansion under FTAs and Multilateral Agreements. Working Paper 05-13 (October). Washington : Peterson Institute for International Economics.
Kiyota, Kozo, and Robert M. Stern. 2007. Economic Effects of a Korea-U.S. Free Trade Agreement. Korea Economic Institute of America (KEI) Special Studies Series 4. Washington (April). Available online at http://www.keia.org/4-Current/KORUSFTASternWebFinal.pdf.
Hufbauer, Gary Clyde, and Kimberly Ann Elliott. 1994. Measuring the Costs of Protection in the United States. Washington : Institute for International Economics.
Rodríguez, Francisco, and Dani Rodrik. 1999. Trade Policy and Economic Growth: A Skeptic's Guide to the Cross-National Literature. NBER Working Paper 7081 (April).
Rodrik, Dani. 2007. The Globalization Numbers Game (May 7). Dani Rodrik's Weblog http://rodrik.typepad.com.
Rose, Andrew. 2003. Which International Institutions Promote International Trade? Review of International Economics Manuscript # 3147. Available online at http://faculty.haas.berkeley.edu/arose/Comparer.pdf.
Sachs, Jeffrey, and Andrew Warner. 1995. Economic Reform and the Process of Global Integration. Brookings Papers on Economic Activity 1995: 1: 1-118.
Taussig, Frank W. 1927. International Trade. Reprint by Augustus M. Kelley Publishers, New York (1966).