Free Trade Doesn't Work

by Ian Fletcher

  One of the clearest signs in the U.S. of the inevitability of industrial policy is that while Washington has fiddled, the states have rushed in to fill the gap. Research parks—most famously Research Triangle Park in North Carolina—based upon links with state universities are the most obvious case, followed by the huge financial incentives states are awarding to lure foreign corporations. Alabama, for example, in order to win a Mercedes-Benz plant in 1993, agreed to provide tax abatements, train the workers, clear the site, upgrade utilities, install infrastructure, and buy 2,500 of the cars produced. This cost $153 million—between $150,000 and $200,000 per job created.728 The problem is that state-level policy like this can easily spend public money merely to shift investments from one state to another, with no net gain to America as a whole.

  Meanwhile, the federal government continues to stick its head in the sand. For example, it allowed the SEMATECH semiconductor research consortium to be effectively dismantled in 1996 by opening it up to foreign manufacturers.729 The competitive difficulties of the American semiconductor industry in the late 1980s were treated as a one-off anomaly requiring merely tactical intervention, rather than as a symptom, destined to be repeated, of the difficulties experienced by an American industry trying to compete on its own against foreign industries backed by effective state industrial policy. America’s tax credit for research and development, once the world’s most generous, is now surpassed by 17 other nations.730 This is despite the fact that, according to one rigorous 1988 study:

  A substantial gap exists between the private and social returns [to R&D] despite the availability of patents. The social rate of return is between 50 and 100 percent, so to be conservative we will say that the excess return to R&D is 35 to 60 percent above the return to ordinary capital.731

  The George W. Bush administration abolished the only program specifically designed to increase the competitiveness of American industry by funding development of technologies that the private sector would not fund on its own: the aforementioned Advanced Technology Program. Free market ideologues repeatedly tarred this program as corporate welfare despite the fact that an audit by the respected National Academy of Sciences vindicated its claim to generate economic benefits far exceeding its cost.732 One single $5.5 million grant, for example, seeded development of the small disk drive industry, which enabled creation of the iPod, the iPhone, TiVo and the Xbox.733 It was replaced by an alternative carefully trimmed to avoid such accusations: the Technology Innovation Program. This program is well run, but pitifully underfunded at a mere $65 million per year.734

  The Obama administration has proved only slightly better than the Bush administration. Although not blinded by an ideological fetish for free markets, its priorities for allocating serious money are decidedly elsewhere. Thus the giant stimulus package it passed in 2009 included money for every Congressional pork barrel under the sun, but nothing for one of the industrial-policy programs with the best track record of saving and creating jobs, the Manufacturing Extension Partnership,735 despite a campaign promise to double the program’s funding.736 This program maintains a network of centers in every state designed to help American manufacturers adopt innovative technologies. One evaluation found that it generated $1.3 billion a year in cost savings for manufacturers and $6.25 billion in increased or retained sales, all for an annual federal outlay of only $89 million.737

  As a result of America’s neglect of industrial policy, there is a starvation of basic and applied research in areas such as biocomputing, computer architecture, software, optoelectronics, aeronautics, advanced materials, factory automation, sensors, energy conversion and storage, nanomanufacturing, and robotics. The U.S. will pay a serious price for this in the decades ahead.

  Chapter 10

  The Multiple Equilibrium Revolution

  We saw in the previous chapter how profoundly the real origins of economic growth contradict free-market, free-trade economics. Real-world growth is path dependent, reliant upon scale economies and “good” industries, and inefficient by the standards of pure free markets. This last point is the most important, because it means that a purely free-market approach to economic policy, which rejects protectionism abroad and industrial policy at home, will necessarily underperform a competent embrace of these policies. But although this has been known and successfully applied by governments (including our own) for centuries, its underlying economics has never been properly mathematized. This has been the Achilles’ heel of this knowledge in post-WWII America, because academic economists have therefore not taken it seriously. Until now. Finally, someone has found a way to translate this eminently practical wisdom into the abstruse mathematics economists are prepared to consider “serious” economics. This is, in fact, the intellectual innovation that well may eventually end economists’ faith in free trade for good.738

  The theoretical breakthrough in question was made by economists Ralph Gomory and William Baumol, starting around 1991 and reaching its crowning synthesis in their 2000 book Global Trade and Conflicting National Interests.739 Ralph Gomory, currently Research Professor at the Stern School of Business at New York University in New York City, holds a PhD in mathematics from Princeton, spent 30 years with IBM, managed its world-famous Research Division, and was its senior vice-president for science and technology. (Two Nobels were won under his direction.) William Baumol, currently Professor of Entrepreneurship at the Stern School, is Professor Emeritus of economics at Princeton and a former president of the American Economic Association. Prior to his work in trade economics, he was best known for having carved out a place in economic theory for entrepreneurs.

  So this is serious economics, albeit cutting-edge and thus highly controversial. It is not a crank theory, an ideological shibboleth, special-interest pleading, or an academic fad. It is, above all, not going to go away any time soon, but will almost certainly challenge the existing free trade consensus until one or the other gives way or a new synthesis emerges.740

  MATHEMATIZING SCALE ECONOMIES

  The easiest way to understand Gomory and Baumol’s work is to add another assumption to the list in Chapter Five of the dubious assumptions of the theory of comparative advantage:

  Dubious Assumption #8: There are no scale economies.

  Gomory and Baumol assume instead that there are. This is both true and, as we saw in the previous chapter, a fact with vast implications. Scale-economy industries are where the action is when it comes to economic growth, and they are the key trade battlegrounds for advanced economies. And because they are the highly capitalized and knowledge intensive industries that most strongly defy Ricardian economics, Gomory and Baumol’s analysis is thus highly appropriate for understanding those industries where free trade isn’t America’s best move. This makes theirs an excellent analysis for anyone who wants to criticize free trade—and figure out what the rational alternative might be.

  WHAT ARE RETAINABLE INDUSTRIES?

  Gomory and Baumol’s analysis is founded on the implications they draw from a key fact about scale-economy industries. Because, by definition, their cost per unit goes down as their output goes up, which nation is the world’s low-cost producer is, other things being equal, a function of which nation is producing more. So whichever nation reaches large production volume first in a scale-economy industry thereby becomes the world’s low-cost producer in that industry. The winner’s cost advantage then locks other nations out.

  Under these circumstances, the only way a challenger can succeed is to start on day one at a production volume equal to the incumbent’s. This is rarely feasible. Not only would the challenger have to match the investments that incumbent had already made to reach high-volume production, but the payoff would be a Pyrrhic victory: head-to-head competition with an entrenched incumbent. As this sort of competition tends to drive profits toward zero, it is rarely financially viable to challenge an incumbent under these circumstances. Gomory and Baumol call industries that behave this way “retainable” industries and if a
nation can acquire them, it can generally hang onto them.

  The important thing here is the lockout phenomenon. Even if another nation hypothetically might have been an even lower-cost producer, the first arrival is so entrenched that the latecomer never gets a chance. This is the opposite of what happens in Ricardo’s model. In his model, historical accidents of which nation reaches high-volume production first certainly happen, but they don’t matter because they get washed away afterwards by competitive forces. Potentially superior latecomers realize their potential superiority and win, so if one nation reaches high-volume production before another, this has no particular significance. Ricardo didn’t allow for the effects of scale economies, and in their absence, a head start doesn’t permanently lower a nation’s costs below its rivals. So without entrenched scale economies, there can be no lockouts.

  This is, of course, how the world ended up with half its large passenger aircraft being built in Seattle and two-thirds of its fine watches being made in Switzerland. Ricardian comparative advantage is useless for explaining why Bangladesh exports many t-shirts and few soccer balls, while Pakistan exports the reverse. Why does South Korea export so many microwave ovens, and almost no bicycles, while Taiwan is the other way around? There is nothing intrinsic about South Korea that makes it a good place to build microwave ovens. Entrenched scale economies are the reason.

  GOOD-BYE PERFECT COMPETITION

  As noted in the previous chapter, scale economies are incompatible with perfect competition. Instead, firms that possess them will be (at least partly) sheltered from competition and will therefore realize quasi-monopoly profits. When two or three firms in an industry all have scale economies, this will tend to make that industry an oligopoly. Imperfect competition is generally a happy thing for those who own and work in such industries, as these industries can reap and pay higher-than-normal profits and wages. Of particular interest to residents of advanced industrial nations, these industries are sheltered against cheap foreign labor if their scale economies are large enough.

  From the general public’s point of view, the best thing about these industries is that they pay higher wages to their employees. Hypothetically, they could just return greater profits to their owners, but the empirical data actually suggests that workers receive a greater benefit. The most authoritative study on this question is by Harvard economists Lawrence Katz and Lawrence Summers (former president of Harvard and Secretary of the Treasury, then chief economic advisor to President Obama). They found that “variations in labor rents [extra wages] across industries are at least two to three times as important as variations in the rents accruing to shareholders [extra profits].”741 Among other things, their analysis changes the evaluation of industrial policy; as they note with respect to the European aircraft industry:

  Once labor rent considerations are recognized, the overall assessment of the Airbus program for European welfare turns from marginally negative to strongly positive. Even in the less favorable case, the subsidy generates a welfare gain representing about half its cost.…Policies promoting domestic production that appear undesirable without taking account of labor market imperfections yield large gains once the existence of these imperfections is acknowledged.742

  Imperfect competition has other benefits. As previously noted, it supports innovation, as the only way innovation can be profitable is if firms which finance it get at least quasi-exclusive rights to sell its application. And because knowledge itself leads to scale economies as the cost of an innovation is amortized over a rising number of units of product sold, this is a self-reinforcing process. More innovation leads to more scale economies, which lead to more profits and more money to finance innovation. Therefore scale economies, in the presence of a healthy financial system with long time horizons, tend to drive an economy to endlessly seek out innovation.

  So, despite myths about high technology being the ultimate arena of pure competition and free markets being the master key to innovation, high technology actually tends to be an area in which pure free market principles do not operate. In the words of one OECD study:

  Oligopolistic competition and strategic interaction among firms and governments rather than the invisible hand of market forces condition today’s competitive advantage and international division of labor in high-technology industries.743

  Ralph Gomory got a good look at this fact during his time at IBM, which once held 70 percent of the market for mainframe computers and faced a Department of Justice antitrust investigation.744

  GOOD-BYE RICARDO’S EFFICIENCIES

  In a Ricardian world, as noted when we first examined the theory of comparative advantage in Chapter Five, every industry automatically migrates to the lowest-possible-cost producer.745 But in a Gomory-Baumol world, it may or it may not—because historical accidents can interfere. In fact, in a Gomory-Baumol world there isn’t even just one answer to the question of which nation is the lowest-possible-cost producer. Instead, there are an infinite number of possible answers, depending on which nation reaches large production volume first.

  This has profound implications. For one thing, it means that some possible outcomes are “bad” outcomes, in which the winning nation locks out potentially more-efficient rivals. Would the world enjoy cheaper fine watches if Japan had captured this industry, rather than Switzerland? We’ll never know, because the accidental equilibrium that placed this industry in Switzerland is now entrenched. Would the world have better small cars if Brazil had captured this industry, rather than Japan? Same problem. If the “wrong” nations win the race in various industries, the world economy may get stuck incurring opportunity costs it could have avoided if other nations had won.746 Thus it will waste opportunities and be less productive than it could have been. So all bets are off about free trade necessarily producing the best possible outcome for the world economy as a whole.747

  The same goes for individual nations. Would Germany be a richer country if it had a commanding position in airplanes, rather than cars? Maybe, but maybe not. The fact that the free market steered Germany’s factors of production into the car industry does not guarantee that this was the best possible use for them.748 Because lockout can interfere, free trade won’t necessarily allocate every nation its most appropriate industries, and mere bad luck can deprive a nation of the industries that would actually have been best for it. And the Ricardian assumption that departing industries will automatically be replaced by better ones becomes even more problematic than already noted with Chapter Five’s dubious assumption #2 (factors of production move easily between industries) because lockout can interfere with breaking into the right replacement industries. Michigan is not going to neatly segue into the helicopter industry.

  WEIGHING THE POSSIBLE OUTCOMES

  If the Ricardian contention that free trade outcomes are always the best possible is no longer tenable, the obvious next question is just how good are the various outcomes that can occur, and for which nations? Is there an intelligible pattern to the quasi-arbitrary outcomes Gomory and Baumol seem to imply, with historical accidents tripping up predictable efficiency as often as not? We need some kind of map of the possible distributions of retainable industries among nations, showing who gets which industries when—and how big are the economic benefits when they do.

  These possible distributions of retainable industries among nations are the subject of a computer model constructed by Gomory and Baumol on the basis of their theory. This model runs out different possible scenarios of industry assignment between two imaginary nations A and B. In some scenarios, A wins the automobile, aircraft, and semiconductor industries, plus two others, while B wins the rest. In other scenarios, B wins those industries, three others, and A wins the rest. With a model containing just two nations (the UN recognizes 194) and 10 industries (the Commerce Department recognizes about 1,800),749 there are over 1,000 possible distributions of industries between nations. So this model is highly simplified, compared to the real world. But it still suffice
s to draw out the key implications of Gomory and Baumol’s ideas.

  Whenever a nation captures an industry, that industry’s output will become part of its GDP, so the nation will be richer for it. Thus one might, at first sight, conclude that a nation’s best move is simply to capture as many retainable industries as it can, by winning the race to high-volume production in each. But if one looks closely at the graph on the next page generated by the computer model, there is clearly more to the story.750

  Each dot on the graph on the next page represents one scenario, that is one possible distribution of retainable industries between A and B. Non-retainable industries will be distributed Ricardo-fashion, and nontraded industries will not be (directly) affected by trade at all, but will form a base of unchanging economic activity at the bottom of the graph. The horizontal axis shows what percentage of the world’s retainable industries A has captured (B gets the rest); the vertical axis shows A’s GDP. So the further to the right a dot is, the more retainable industries A has captured, and the higher a dot is, the richer A is in that scenario.

  GREEDY, BUT NOT TOO GREEDY

  The graph on the next page shows that, in general, the more retainable industries A captures, the higher its GDP. No surprise there. But there’s also a surprise: it is not nation A’s best move to capture all the retainable industries in the world! If it does, it actually ends up poorer than if it captures, say, 70 percent. Thus the cluster of dots is actually lower at the far right side of the graph than if we move a little to the left. So A’s best move is apparently to be greedy, but not too greedy.