American Economic Review 2009, 99:3, 561–571
www.aeaweb/articles.php?doi=10.1257/aer.99.3.561
The Increasing Returns Revolution in Trade and Geography†
By Paul Krugman*
Thirty years have pasd since a small group of theorists began applying concepts and tools from industrial organization to the analysis of international trade. The new models of trade that emerged from that work didn’t supplant traditional trade theory so much as supplement it, creat-ing an integrated view that made n of aspects of world trade that had previously pod major puzzles. The “new trade theory”—an unfortunate phra, now quite often referred to as “the old new trade theory”—also helped build a bridge between the analysis of trade between countries and the location of production within countries.
In this paper I will try to retrace the steps and, perhaps even more important, the state of mind that made this intellectual transformation possible. At the end I’ll also ask about the relevance of tho once-revolutionary insights in a world economy that, as I’ll explain, is arguably more classical now than it was when the revolution in trade theory began.
I. Trade Puzzles
In my first year as an assistant professor, I remember telling colleagues that I was working on international trade theory—and being asked why on earth I would want to do that. “Trade is such
a monolithic field,” one told me. “It’s a finished structure, with nothing interesting left to do.”
Yet even before the arrival of new models, there was an undercurrent of dissatisfaction with conventional trade theory. I ud to think of the propagation of this dissatisfaction as the trade counterculture. There were even some underground classics. In particular, Staffan Burenstam Linder’s An Essay on Trade and Transformation (1961), with its argument that exports tend to reflect the characteristics of the home market, was pasd hand to hand by graduate students as if it were a samizdat pamphlet. And there was also an important empirical literature on intra-industry trade, notably the work of Bela Balassa (1966) and Herbert G. Grubel and Peter J. Lloyd (1975), that cried out for a theoretical framework.0是单项式吗
Why did the trade counterculture flourish despite the apparent completeness of conventional trade theory? Call it the similar-similar problem: the huge role in world trade played by exchanges of similar products between similar countries, exemplified by the massive two-way trade in auto-motive
products between the United States and Canada.
In 1980, this similar-similar trade was still a relatively new phenomenon. Trade in the first great age of globalization—the age made possible by steam engines and telegraphs—was mainly dissimilar-dissimilar: trade in dissimilar goods between dissimilar countries. Comparative advantage, which one may define as the idea that countries trade to take advantage of their dif-ferences, clearly explained most of what was going on. It was only with the recovery of trade after World War II, and especially after the major trade agreements of the 1950s and 1960s, that the more puzzling trade patterns that fed the counterculture became prominent.
† This article is a revid version of the lecture Paul Krugman delivered in Stockholm, Sweden, on December 8, 2008, when he received the Bank of Sweden Prize in Economic Sciences in Memory of Alfred Nobel. This article is copyright © The Nobel Foundation 2008 and is published here with the permission of the Nobel Foundation.
* Woodrow Wilson School of Public and International Affairs, Princeton University, Princeton, NJ 08554 (e-mail: pkrugman@princeton.edu).
561
JUNE 2009
562
中招体育THE AMERICAN ECONOMIC REVIEW Figures 1, 2, and 3 make this point, using data on British trade. Figure 1 shows the commod-ity composition of British exports and imports on the eve of World War I. The pattern of trade made perfect n in terms of classical comparative advantage: Britain, a denly populated nation with abundant capital but scarce land, exported manufactured goods and imported raw materials.
By contrast, Figure 2—which shows comparable data for 1990—offers no comparably easy interpret
ation: Britain both imported and exported mainly manufactured goods. One might have hoped that a look within the manufacturing ctor would reveal a clearer pattern, but this brings us to the issue of intraindustry trade: trade in manufactured goods, especially between countries at similar levels of development, consists to a large extent of two-way exchanges within even narrowly defined product categories.
And trade, as reconstituted after World War II, did take place to a large extent between similar countries, much more so than in the first age of globalization. Figure 3 illustrates this crudely, comparing Britain’s trade with Europe and with rest of the world in 1913 and 1990.
Before World War I, Britain traded remarkably little, by modern standards, with its neighbors, instead focusing on distant lands able to produce what Britain could not—cheap wheat and meat, tea, jute, and so on. By 1990, however, while such trade had by no means vanished, Britain was part of a European economy in which nations emingly made a living by taking in each other’s washing, buying goods that they could, and, at least as far as the statistics indicated, did produce for themlves.
So what was going on?
关于诸葛亮II. Increasing Returns and Trade
It shouldn’t have been that hard to make n of similar-similar trade. Indeed, for some economists it wasn’t. In his minal paper on the ri of intraindustry trade in Europe, Balassa (1966) stated it quite clearly: each country produced only part of the range of potential products within each industry, importing tho goods it did not produce, becau “specialization in nar-rower ranges of machinery and intermediate products will permit the exploitation of economies of scale through the lengthening of production runs.”
100%50%
0%
Exports Imports
Figure 1. Composition of British Trade circa 1910
Source: Richard E. Baldwin and Philippe Martin (1999).
pdf怎么复制VOL. 99 NO. 3563
KRUgMAN: THE INCREASINg RETURNS REVOLUTION IN TRAdE ANd gEOgRApHy Yet this straightforward-eming explanation of similar-similar trade was not at all part of the standard corpus of international trade theory circa 1975. It was not so much that the ideas were rejected as that they emed incomprehensible. Why?
The answer was that unexhausted economies of scale at the firm level necessarily imply imper-fect competition, and there were no readily usable models of imperfect competition to hand. Even more to the point, there were no general equilibrium models of imperfect competition readily to hand—and trade theory, perhaps more than any other applied field of economics, is built around general equilibrium analysis.
100%
50%
0%
Exports Imports
100%
50%
0%
circa 1910 1990s
街舞breakingFigure 2. Composition of British Trade in the 1990s
Source: Baldwin and Martin (1999).
Figure 3. Destination of British Exports
Source: Baldwin and Martin (1999).鸟巢体育馆
564THE AMERICAN ECONOMIC REVIEW
JUNE 2009 The result was the state of affairs almost triumphantly described by Harry Johnson (1967): “The theory of monopolistic competition has had virtually no impact on the theory of interna-tional trade.”
Then came the new monopolistic competition models: Kelvin J. Lancaster (1979), Michael Spence (1
976), and above all Avinash Dixit and Joph Stiglitz (1977). All of the papers were intended by their authors as ways to address the classic welfare questions about whether monopolistic com-petition led to inefficient scale, or perhaps to production of the wrong mix of products. But when I learned about the new literature (in a short cour taught by Robert Solow in 1976), I—like a number of other people working independently, including Victor Norman (1976) and Lancaster (1980) him-lf—quickly saw that the new models provided “gadgets,” ways to think about the role of increasing returns in a variety of contexts. And there was, in particular, a near-perfect match between simple models of monopolistic competition and the stories already circulating about intraindustry trade.
It quickly became apparent (Norman 1976; Krugman 1979; Lancaster 1980) that one could u monopolistic competition models to offer a picture of international trade that completely bypasd conventional arguments bad on comparative advantage. In this picture, countries that were identical in resources and technology would nonetheless specialize in producing different products, giving ri to trade as consumers sought variety. A natural extension—although, like many things that em obvious in retrospect, it was surprisingly hard at first to figure out how to do it—was to bring comparative advantage back in. This was most easily done by assum-ing that all the differentiated products within an industry were produced with the same factor proportions; one could
then explain inter industry specialization in terms of Heckscher-Ohlin, with an overlay of intra industry specialization due to increasing returns. And this extension, as reprented for example by Elhanan Helpman (1981) and Dixit-Norman (1980), in turn meant that the new models offered an intellectually satisfying explanation of similar-similar trade: similar countries had little comparative advantage with respect to each other, so their trade was dominated by intraindustry trade caud by economies of scale.
What was really needed to get peoples’ attention, however, was a “killer ap”: a demonstration that the new view offered a fundamentally different insight into something that mattered. I found that killer ap in an empirical insight by Balassa (1966), who pointed out that trade liberaliza-tion among industrial countries had proved surprisingly nondisruptive, belying fears that, for example, there would be a major rearrangement of Europe’s industrial landscape after the forma-tion of the Common Market, and possibly large effects on income distribution. Becau trade expansion had taken the form of intraindustry specialization rather than interindustry specializa-tion, Balassa noted, “the fears expresd in various member countries of the Common Market concerning the demi of particular industries have not been realized. There are no examples of declining manufacturing industries in any of the member countries.”
In Krugman (1981) I built a special version of the emerging style of model to encapsulate this phenomenon. What the model showed was that the classic Stolper-Samuelson result, in which trade liberalization hurts scarce factors, can emerge—but only if comparative advantage is strong and/or economies of scale weak. In the rever ca, which emed to describe the growth of trade among industrial countries, trade was win-win.
There was one more significant insight from the application of Dixit-Stiglitz–bad models to trade: Burenstam Linder was right! Once one added transport costs to the model, it was straight-forward to show that countries would, other things equal, tend to become exporters in the indus-tries in which they had large domestic markets. As is so often the ca, the logic of this result was obvious once the result had been devid, but not at all obvious beforehand. Indeed, I began the rearch that led to Krugman (1980) with the strong presumption that countries would not tend to export goods for which they had a large home market—I came to bury Burenstam Linder, not to prai him. But the algebra said otherwi, and the intuition followed. Increasing returns
KRUgMAN: THE INCREASINg RETURNS REVOLUTION IN TRAdE ANd gEOgRApHy VOL. 99 NO. 3565
蟋蟀和蛐蛐一样吗
provide an incentive to concentrate production of any one product in a single location; given this incentive to concentrate, transport costs are minimized by choosing a location clo to the largest market, and this location then exports to other markets.
Initially, the “new trade theory” emed to consists of a ries of special-purpo, incompat-ible models. It turned out, however, that it was possible to create a common ground for many though not all of the models, and extend that common ground to much traditional trade theory as well, using an insight originally due to Paul Samuelson (1949). In explaining factor price equal-ization, Samuelson reverd the usual way we think about trade, as a process of coming together. Instead, Samuelson thought of trade as the result of a process of coming apart. He envisaged a Tower of Babel scenario, in which an angel descends from Heaven and breaks up a previously unified economy: factors of production suddenly find themlves with national labels, and are able to work only with other factors that have the same national label. Samuelson pointed out that factor price equalization would take place if and only if the international distribution of factors of production was such that it was possible, even while obeying the angel’s new limits, to reproduce the production of the pre-angel integrated economy—and that in such circumstances specializa-tion and trade could be viewed as being about reproducing the integrated economy.
Helpman and I (1985) ud the same approach to think about trade involving both comparative advantage and increasing returns. The key insight was that in order to reproduce the integrated economy, it was necessary to locate all production of each good subject to economies of scale within one country. This approach united factor-proportions-bad comparative advantage and specialization due to economies of scale: both could be viewed as part of how a world economy curd by Samuelson’s angel undid the damage.
This approach also, more or less en passant, made it clear that increasing returns ordinarily reinforce, rather than call into question, the argument that there are gains from trade. To be sure, in cas where the integrated economy is not reproduced, it’s possible to conjure up examples in which countries are wor off with trade than without, in a way that isn’t possible in pure com-parative advantage models. But the clear presumption is that trade is a good thing under increas-ing returns—indeed, better than previously thought.
By the mid 1980s, then, the “new trade theory” had integrated increasing returns more or less amlessly into our understanding of international trade. The impossible complexity that had previously daunted economists contemplating a major revision of trade theory had vanished, replaced by a surprisingly simple and elegant structure.
But how did that happen? Why did the problems facing the trade counterculture em to melt away? I’d argue that at the heart of the story was an attitude shift on the part of international economists.
贺信格式III. Some Meta Reflections
The emergence of the new trade theory was, in the first place, made possible by the new mod-els of monopolistic competition. But it did not remain confined to tho models; by the mid-1980s recognizably “new trade” approaches had been taken to trade involving external economies, Cournot and Bertrand oligopoly, even contestable markets. What made it all possible was a shift in attitude among trade theorists, mainly consisting of two changes. First, there was a new willingness to explore the implications of illuminating special cas rather than trying to prove general results given some broad upfront assumptions. Second, there was a change in focus from detailed predictions—which country produces each specific good—to system-level or aggregate descriptions of the pattern of trade.
On the first point: in the late 1970s many trade theorists thought of themlves as theorem-provers. Given big initial assumptions such as constant returns, so many factors, etc., what could
