Douglas Clement - Senior Writer
Published December 1, 2002 | December 2002 issue
Why do nations trade? If two countries are each capable of producing the same set of commodities, why should they go to the trouble and expense of exchanging them? More concretely, how can economists explain why France and Germany, for example, ship agricultural goods, machinery and other products across their mutual border, when each country could reasonably take care of its own needs? And how do these trade patterns help explain where industries will locate?
Since the early 19th century, economists have used the theory of comparative advantage to answer such questions. According to this classic theory, nations are made better off through trade by capitalizing on their inherent differences in natural resource or capital endowments. If two nations differ in land fertility and mineral resources, for example, it makes economic sense for one of them to specialize in growing vegetables and trade some of them to the other country for a commoditysteel, perhapsthat that country can produce more efficiently because of its inherent advantages. (It's a bit trickier than that, actually; stay tuned.)
But over the last 20 years, international trade economics has undergone what some have termed a "revolution" because of a new theory that gives very different answers to the same kinds of questions. This new trade theory is often summed up in the simple words "increasing returns"shorthand for "increasing returns to scale," a term synonymous with "economies of scale."
The new theory, then, is the idea that trade arises to take advantage of economies of scale: Industries in each country can achieve lower unit costs by producing in large volume and spreading the high start-up expenses (for example, research and development, machinery purchases) over many, many units. If confined to the domestic market, an industry might not be able to achieve the highest level of scale economies, but by producing for both domestic and foreign markets, sufficient volumes can be produced to reap greater economies of scale.
According to this theory, two countries could both get cheaper cars and lamps, for instance, if one specialized in automobile manufacturing and the other devoted its resources to making lighting equipment, regardless of the inherent resource endowments of those countries. By taking advantage of increasing returns to scale, each country could produce its specialty at low unit costs and then trade with the otherboth countries will gain through trade, but not because of comparative advantage.
This "revolution" in trade theory, however, is partial and unfinished. Economists generally agree that the new increasing-returns theory has its strengths, but they disagree deeply about how important it is relative to comparative advantage theory. After all, nations often trade products whose manufacture isn't subject to increasing returns to scalethat is, commodities that can be produced at either constant returns (that is, doubling inputs results in exactly twice the output) or diminishing returns (doubling inputs results in less than twice the output). Comparative advantage theory can explain that kind of trade, but increasing-returns theory can't. So how well do the two theories explain what goes on in the real world?
Economists also hotly debate one of the most significant corollaries of increasing-returns theory, the so-called home market effect. This theory helps explain where industries will set up their operationsan issue of great consequence to politicians and labor unions as well as economists. In brief, it says that factories will tend to locate close to their largest markets so that they can minimize overall shipping costs while still taking advantage of economies of scale. That is, car companies will locate in Germany rather than France if it's clear that Germans are likely to buy more cars. That way the company can produce low-cost cars and not have to pay much to ship them to its largest market.
Just a few years ago it appeared that the home market effect hypothesis had been dealt a mortal blow. An insightful critique seemed to have found an Achilles' heel in an ostensibly innocuous assumption that underlay the mathematical model of increasing-returns trade theory. If the assumption were changedas seemed empirically reasonablethe home market effect disappeared.
But with a recent paper, "The Home Market and the Pattern of Trade: Round Three" [Minneapolis Fed Research Department Staff Report 304], two Federal Reserve economists have revived the home market component of increasing-returns trade theory by illustrating that the critique itself rested on an untenable assumption. The paper, by University of Minnesota economics professor Thomas J. Holmes, a visiting scholar at the Minneapolis Fed, and John J. Stevens, an economist at the Fed's Board of Governors, finds that this fundamental outcome of the increasing-returns trade "revolution" remains entirely valid. Yet at the same time, the economists suggest that increasing-returns trade theory and the home market effect may have their greatest explanatory power not at the global level, but much closer to home.
This tale of theoretic death and resurrection requires a bit more background, thougha brief economic primer on trade models, old and new.
According to the classic theory of comparative advantage, nations, regions, cities and even individuals will exchange goods based on their relative levels of labor productivity, which inherently differ because available technologies and natural endowments vary among trading partners.
If England and Portugal, for example, can both produce wool and wine, it might seem there would be little reason for either country to pack up one commodity and ship it off to the other nation. But comparative advantage theory demonstrates that, in fact, both parties can benefit economically by trading those goods. And this is true even if one of themPortugal, say, because of its sunny climateis able to produce both goods using less labor than the other nation (that is, even if it has an absolute advantage in production of both goods).
Describing commodity exchange under comparative advantage is invariably confusing, but in a nutshell, each nation can obtain as much or more of both goods by taking advantage of the countries' differing ratios of labor productivity for the two goods. Each country should determine which productwine or woolit can produce more efficiently than the other nation (relative to its own efficiency in producing the other good) and exchange it for the product that its trading partner produces more efficiently. Put differently, each nation should specialize in production of the good for which its opportunity costs are lower than its trading partner's and import that for which its opportunity costs are higher.
A clearer example: With his height and athletic ability, Michael Jordan is undoubtedly better at both basketball and window washing than I am. (He has an absolute advantage in both.) But we'd each benefit if he paid me to clean his windows, and I gave up my hoop dreams to watch him specialize in his comparative advantage. He'd be sacrificing much less of his overall productivity by getting me to wash his windows and devoting himself to basketball. And frankly, neither of us would really miss my slam dunk attempts, but I can do a passable job with a squeegee. Through specialization and trade, both parties would come out ahead: clean windows and world-class basketball. [See sidebar for a fuller explanation of comparative advantage.]
Comparative advantage theory has great strength in explaining something not intuitively obvious, and it has held sway in economics since the early 1800s when Robert Torrens and David Ricardo developed it. Ricardo focused on labor productivity relative to trade. Relatively recent elaborations (for example, the Heckscher-Ohlin theory) emphasize the importance of national endowments of capital, as well as labor, and hold that countries will export goods that use their abundant factor intensively and import those that use their scarce factor intensively.
But comparative advantage has its weaknesses as well. It has little ability to explain why regions with similar productivity levels trade to the extent they dowhy Europe and the United States, for example, trade in such great volume. Nor does it shed light on intra-industry trade: the fact that Germany and Japan will trade automobiles with one another. Indeed, much recent discussion suggests that the theory itself may have its best advantage in explaining trade between industrialized and developing nations, but not among countries with similar factor endowments.
In the 1980s, economists Paul Krugman, now at Princeton University, and Elhanan Helpman at Harvard University, among others, began to develop a new theory of trade that redressed the weaknesses of comparative advantage. This "new trade theory" is based on the idea of increasing returns to scale.
Nations with similar factor endowmentsthat is, with little in the way of comparative advantage differencesmay nonetheless find it beneficial to trade because they can take advantage of massive economies of scale, a phenomenon prevalent in a number of economically important industries. In the pharmaceutical and automobile industries, for example, the first unit is very expensive to produce, but each subsequent unit costs much less than the one before because the large setup costs can be spread across all units.
The mathematics behind increasing-returns trade theory are complex, but the idea itself is quite intuitive. A trading partner can develop an industry that has increasing returns to scale, produce that good in great quantity at low average unit costs, and then trade those low-cost goods to other nations. By doing the same for other increasing-returns goods, all trading partners can take advantage of large economies of scale through specialization and exchange.
A key result of the theory is the home market effect: Countries will specialize in products for which there is large domestic demand. The reason here is also intuitive: By locating close to its largest market, an increasing-scale industry can minimize the cost of shipping its products to its customers. (On the other hand, constant-returns-to-scale industries have little need to locate close to large markets. Their unit costs will be similar regardless of quantities produced. So each market can efficiently produce such goods locally.)
But the home market effect also has a disturbing implication. If increasing-scale industries tend to locate near their largest markets, what happens to small market areas? Other things equal, they're likely to become "deindustrialized" as factories and industries move to take advantage of scale economies and low transportation costs. Trade, therefore, could lead to small countries and rural areas becoming peripheral to the economic core, the backwater suppliers of commodities. (As Canadian critics have phrased it, "With free trade, Canadians would become hewers of wood and drawers of water.") But of course, economists caution, other things aren't strictly equal: Comparative advantage effects exist alongside the influence of increasing returns, so the end result of open trade is not a foregone conclusion.
Over the last two decades, new trade theory has had an enormous influence on international trade economics. It has helped explain trade patterns among industrialized nations and within industries. It also sheds light on regional patterns of trade: commodity exchange and industry location within countries. But in 1998, a paper by Harvard economist Donald Davis dealt this flourishing movement what seemed a damaging blow.
Krugman had developed his increasing-returns trade theory by looking at a model with two sectors, one, alpha, with increasing returns to scale and the other, beta, with constant returns. To render the mathematics straightforward, he adopted what he considered a harmless assumption. "For simplicity," Krugman wrote in his seminal 1980 paper, "also assume that beta goods can be transported costlessly."
It was the kind of simplifying assumption that theorists make all the time, abstracting somewhat from the real world in order to increase explanatory power"sacrificing some realism to gain tractability," Krugman wrote. And he believed this assumption, if slightly less than realistic, was basically innocuousnothing vital seemed to hinge on it.
But in his 1998 paper, Davis, now chair of Columbia University's economics department, examined the assumption more carefully. Reviewing data on trade costs (including costs of transportation, nontariff barriers, "border effects" and other costs inherent to trade), Davis concluded that Krugman's simplification was unrealistic: "There is little suggestion that total trade costs are higher" for increasing-returns goods, he wrote. If anything, the cost of trading constant-returns goods is likely to be higher than that for increasing-returns goods.
Even more crucially, Davis found that the simplification was far from innocuous; something important does hinge on it. In fact, if Krugman's model is duplicated with just one small change: Assume that both sectorsnot just the increasing-returns sectorhave positive transportation costs, then Krugman's striking finding about trade and industry location evaporates. When "the industries have identical trade costs, the home market effect disappears," Davis concluded. "Industrial structure then does not depend on market size."
By undermining the home market effect, Davis' finding struck at the heart of new trade theory, so he was careful to note that further empirical and theoretic investigation was called for. "The importance of the home market hypothesis for production and trade structure is such that more extensive inquiry is in order," he wrote.
Holmes and Stevens answer the call for further investigation by subjecting Davis' critique to the same close scrutiny he gave Krugman's model. And similarly, they find that "a seemingly innocuous simplifying aspect of the Davis model is actually crucial." Just as Krugman had derived his model using an assumption of conveniencethat the constant-returns sector had no transportation costsDavis had built a model with a simplifying assumption: that here are just two types of industries, one with constant returns to scale and the other with increasing returns.
Holmes and Stevens alter this assumption by creating a more flexible model, one that allows for a range of industry types that vary in their degree of increasing returns. Clearly, all industries aren't either of a constant scale or a single increasing scale; industries vary tremendously. The Holmes-Stevens model thus provides a broader, more realistic description of an economy. And when they use this model to explore the effects of trade, they find that Krugman's initial finding reasserts itself: "The pattern of trade here does depend on country size, just like the original Krugman result."
Specifically, Holmes and Stevens assume (as did Davis) that transportation costs are positive and identical in all industries, and they find that industries with low levels of scale economies aren't traded, while those with medium or high degrees of increasing returns are traded. According to their model, smaller countries will export medium increasing-returns products and larger countries will export high economies-of-scale goods.
As an example, they say, think about wool as a constant-returns industry, food processing as a medium economies-of-scale industry and automobiles as high economies of scale. "We find that opening of trade between New Zealand [a small market] and Japan [a large market] causes New Zealand to shift into food processing out of autos, to export food processing and import autos," they wrote. The wool industrybecause neither nation has an economies-of-scale advantage in itwill not trade internationally (or at least, not because of increasing returns).
Thus, Holmes and Stevens give new life to a threatened theory. But the question remains, to what degree is this theory relevant to the real world? The increasing-returns explanation for trade never aspired to replace the comparative advantage explanation; it just supplemented it. But is increasing returns a minor or major factor in determining patterns of trade? And where does it have its greatest impact?
In a recent interview, Holmes said that new trade theory may be both more powerful and more easily tested at the regional level than at the international level. Because the theory looks at the influence of market size on trade patterns and industry structure, it should have its strongest measurable effects across markets of widely varying sizes and densities.
"If you look at the European Union and the United States, you see they are fairly similar in terms of market size and overall population density," Holmes said. "But if you compare Minneapolis and some place in North Dakota, you see population density differences of many orders of magnitude." So industry locations and trade at the regional level are where increasing returns "is going to have its biggest bite."
For example, in another recent paper, Holmes looked at data for nearly 30,000 branch sales offices of manufacturers in the United States and found that the offices were highly concentrated in large cities, not spread out evenly across the nation. His preliminary analysis indicated that the home market effectlocating offices in the largest markets so as to capitalize on returns to scale and minimize transportation costsaccounts for a third to a half of that concentration.
A number of studies have looked at international patterns of trade to gauge the relative importance of comparative advantage and increasing returns in explaining trade among nations. Davis, in a 1996 paper with David Weinstein, then at the University of Michigan, looked at trade patterns in Organisation for Economic Cooperation and Development member countries and found that the data strongly rejected the increasing-returns explanation in favor of a comparative advantage explanation. But a few years later, Davis and Weinstein used what they called a "richer" geographic model on the same data and reversed their position, finding compelling evidence of the influence of increasing returns on trade. A 2002 study by Werner Antweiler at the University of British Columbia and Daniel Trefler at the University of Toronto looked at trade flows among 71 countries from 1972 to 1992 and reached similar conclusions.
Notwithstanding this research, Holmes suggests that data limitations have prevented the kind of detailed analysis necessary to reach a conclusive understanding of the importance of increasing-returns effects at the international level. "To look at these issues, you really need to get data that is disaggregated to a fine level of industry detail," he said. "But most of the research in this literature uses international trade data sets where industries are highly aggregated to broad levels like 'transportation equipment' or 'chemicals.' That impedes progress in trying to understand how this is working."
Increasing-returns models, after all, emphasize the importance of factory specialization and product differentiation. If the data don't measure that differentiation, they can't fully gauge the effect of increasing returns. Nations taking advantage of economies of scale may trade, for example, in Toyotas and Fords. Both autos, true, but they're very distinct products and most trade data don't reflect the difference.
"In terms of applying this to international [trade], I think the theory is ahead of the empirical work," observed Holmes. Data limitations and the relatively weak influence of increasing returns at the international level may stand in the way of rapid progress in verifying the theoretical models.
"It's really a hard question to address, trying to sort out the importance of size from comparative advantage," said Holmes. And ironically, this economist who helped breathe new life into increasing-returns trade theory remains quite cautious about its significance for understanding world trade. "At the international level, in terms of applied work, I think we're still in its infancy," he said. "We don't yet have a smoking gun that this is what's really important."