Paul Krugman’s Geographical Economics and Its Implications for Regional Development Theory:
A Critical Assessment ***
Volume 72, Issue 3
July 1996, pp. 259-292
Krugman’s Geographical Economics and Economic Geography:
A Critical Comparison
The Resurgence of Regional Economies
The New Political Economy of Trade
Krugman’s Model of Economic Integration and Regional
Development: The Lessons of the
Economic Integration and Regional Specialization
Economic Integration and Divergent Regional Growth
Trade and the Regional Policy Issue
Strategic Trade Policy
Geographical Clustering and Strategic Industrial Policy
Abstract: Economists, it seems, are discovering geography. Over the past decade, a “new trade theory” and “new economics of competitive advantage” have emerged which, among other things, assign a key importance to the role that the internal geography of a nation may play in determining the trading performance of that nation’s industries. Paul Krugman’s work, in particular, has been very influential in promoting this view. According to Krugman, in a world of imperfect competition, international trade is driven as much by increasing returns and external economies as by comparative advantage. Furthermore, these external economies are more likely to be realized at the local and regional scale than at the national or international level. To understand trade, therefore, Krugman argues that it is necessary to understand the processes leading to the local and regional concentration of production. To this end he draws on a range of geographical ideas, from Marshallian agglomeration economies, through traditional location theory, to notions of cumulative causation and regional specialization. Our purpose in this paper is to provide a critical assessment of Krugman’s “geographical economics” and its implications for contemporary economic geography. His work raises some significant issues for regional development theory in general and the new industrial geography in particular. But at the same time his theory also has significant limitations. We argue that while an exchange of ideas between his theory and recent work in industrial geography would be mutually beneficial, both approaches are limited by their treatment of technological externalities and the legacy of orthodox neoclassical economics.
The relationship between economic geography and economics has long been an asymmetric one. In constructing their theories and explanations of regional development, economic geographers have drawn freely on the concepts and perspectives of different schools of economics; but, for their part, economists have tended to accord little if any attention to the role of geography in the economic process. The case of trade theory admirably illustrates this point. Regional devel
* Ron Martin, Department of Geography,
** Peter Sunley, Department of Geography,
*** This is a revised version of a paper presented at
the Special Session on Economic Geography held at the Annual Conference of the
opment theory has always been concerned with the question of interregional trade, because a region’s ability to export goods and services is one of the foundations of local economic growth and employment (Erickson 1989). The typical approach to the study of interregional trade has been to borrow and adapt the ideas and models of comparative advantage (factor endowment) trade theory from economics. Trade economists, however, have invariably regarded the national economy as spaceless, and even international trade typically has been seen as an exchange system devoid of any geography, a world where goods and services move between dimensionless points at zero or uniform transport costs. This lack of a sensitivity to geography by trade theorists partly explains why there is no overall theoretical framework guiding geographical research on international trade (Grant 1994). The absence of such a framework is particularly evident at a time when the “globalization” of economic relations and the continental regionalization of trade are challenging the territorial and regulatory significance of national economic spaces and giving greater prominence to the nature and performance of individual regional and local economies within nations (Dunford and Kafkalas 1992; Anderson and Blackhurst 1993; Gibb and Michalak 1994).
Recently, however, there have been developments within economics which may mark the beginning of a closer relationship with economic geography in general and regional development theory more particularly. Over the past decade, a “new” trade theory and a new economics of competitive advantage have emerged which, among other important features, assign a key significance to the role that the internal geography of a nation may play in determining the trading performance of that nation’s industries. 1 Economists, it seems, are discovering geography. In particular, Paul Krugman, the leading and extraordinarily prolific exponent of the “new” trade theory, 2 has sought to show how trade is both influenced by and in turn influences the process of geographical industrial specialization within nations (for example, Krugman 1991a). In his view, the importance of regional industrial specialization and concentration is such that economic geography should be accepted as a major subdiscipline within economics, “on a par with or even in some senses encompassing the field of international trade” (Krugrnan 1991a, 33). Likewise, from a different, but ultimately related perspective, Michael Porter, the eminent business economist, has argued that the degree of geographical clustering of industries within a national economy plays an important role in determining which of its sectors command a competitive advantage within the international economy (Porter 1990). In a similar vein to Krugman, Porter also argues that there are strong grounds for making economic geography a “core discipline in economics” (Porter 1990, 790).
Paul Krugman’s work, especially, is worthy of closer interest by geographers. Krugrnan has written on a wide range of issues that impinge on the regional development question: trade, externalities, the localization of industry, strategic industrial policy, globalization, the role of history and “path dependence,” and the implications of economic and monetary integration for regional growth. One of the key thrusts of his work is that in order to understand trade we need to understand the process of regional development within nations. A number of his writings have thus sought to explain why industrial development is likely to be geographically
The set of ideas referred to as the
“new trade theory” was originally expounded in a series of papers by Dixit and
2. Such has been Krugman’s influence within the economics profession that Paul Samuelson (1994, vii) refers to him as “the rising star of this century and the next.”
uneven. To this end, he draws on a range of economic and geographical ideas, from Alfred Marshall’s account of localization economies, through traditional location theories, to notions of cumulative causation. For Krugman, economic geography - by which he means uneven regional development - is a central part of the process by which national economic prosperity and trade are created and maintained.
Our aim in this paper is to provide a critical assessment of Krugman’s “geographical economics” and its implications for contemporary economic geography. An exchange of ideas between his theory and recent work in economic geography would be mutually beneficial. Such an exchange is not easy to engineer, however, as there are several significant obstacles, on both sides. First, Krugman’s ideas are far from static. Indeed, his views seem to change continuously over time - sometimes in a self-criticizing way - so it is important to base any evaluation on a range of his works. Second, throughout Krugman’s writings there is a strong distinction between what is theoretically possible and what is empirically and practically important, so that his conclusions have to be read carefully and closely. Most important, however, Krugman’s geographical economics and contemporary economic geography are very different academic genres, with different methodological styles and conventions of analysis and writing (Krugman 1993a). Krugman’s method is to start with a real world problem and then build a model to capture the “essence” of that problem (Krugman 1989, 1992). The model, which is usually mathematically specified, is made as simple as possible to remove unnecessary clutter, although in most cases he also gives a highly readable narrative account of the model. The mathematical aspect of his methodology may well explain the strong location-theory flavor to much of his geographical economics. 3 However, this methodological and theoretical disposition is unlikely to appeal to many economic geographers, who have abandoned formal models and rigorous exegesis for a more discursive approach, in which broad master concepts (like “flexible specialization” and “post-Fordism”) are mingled with anecdotal spatial stereotypes (“industrial spaces” and “industrial districts”).
These differences probably largely explain why Krugman’s
writings have thus far had a limited impact on economic geography, and why they
have been summarily dismissed by certain geographers.
3. This location-theory orientation has if anything become even more pronounced in his two most recent books on “spatial economics,” Development, Geography and Economic Theory (Krugman 1995) and The Self-Organising Economy (Krugman 1996). In the former, lamenting economic geographers’ “retreat” from quantitative models into Marxist and regulationist concerns with “post-Fordism,” Krugman resurrects what he calls the five “exiled traditions” of economic geography: location theory, social physics, cumulative causation, land use modeling, and Marshallian local external economies. In the latter, von Thunen’s model and central place theory occupy a key role in his theorization of the “self-organizing” space economy.
4. It is not difficult to see how economic geographers might take offense at Krugman’s view of their work. In Geography and Trade (1991a, 3-4) he writes: “The decision by international economists to ignore the fact that they are doing geography wouldn’t matter so much if someone else were busy… looking at localization and trade within countries. Unfortunately, nobody is. That is, of course an unfair statement. There are excellent economic geographers out there… However,… economic geographers proper are almost never found in economics departments, or even talking to economists… They may do excellent work, [but it does not inform or influence the economics profession.” It could equally be argued, of course, that it is the economists who have failed to talk to economic geographers and that, as a result, like Krugman they are largely ignorant of the major developments that have taken place in economic and industrial geography over the past decade or so. Equally irritating is Krugman’s comment, in Development, Geography and Economic Theory (1995, 88), that “in the end, we [i.e., economists] will integrate spatial issues into economics through clever models (preferably but not necessarily mine) that make sense of the insights of the geographers in a way that meets the standards of the economists.” Whether economists have any such monopoly over analytical or theoretical standards may most certainly be questioned.] HHC: [bracketed] display on p. 262 of original.
in his review of the same book, Hoare (1992, 679) criticizes the particular economic geography used by Krugman as “dated, historically and intellectually” and his analysis as based on the “flimsiest of empirical support.” However, Krugman’s remarks are leveled primarily at his own colleagues’ failure to admit that “space matters” (Krugman 1991a, 8), and he should at least be congratulated for wanting “to bring geography back into economic analysis,” even if the particular form of geography he uses - essentially a form of regional science - is open to criticism. Furthermore, Geography and Trade gives only a partial glimpse into Krugman’s analyses, and any considered judgment as to the significance of Krugman’s work for economic geography must also be based on his numerous other writings in the field.
We too have criticisms to make of Krugman’s treatment of economic geography, although we also believe that his work raises some interesting issues for contemporary regional theory. We begin by outlining what we take to be the essential arguments and components of his “geographical economics,” focusing on his interpretation of the relationships between location and trade, the role of increasing returns and externalities in the localization of industry, and the significance of history, “lock-in,” and path dependence for regional development. The subsequent section examines these ideas in closer, more critical detail, and compares Krugman’s theories with those that have emerged from the “new industrial geography” in the past few years. We then examine his arguments about the impact of economic integration on regional development, especially his prognoses of the regional implications of integration within the European Union and his views on regional stabilization and industrial policy. We conclude the paper by drawing together the main strengths and weaknesses of Krugman’s approach to economic geography.
The Bases of Krugman’s “Geographical Economics”
Krugman’s geographical economics and theorization of uneven regional development are firmly rooted in his contributions to the “new trade theory.” Conventional trade economics is based on Ricardian comparative advantage theory (especially in its Heckscher-Ohlin-Samuelson versions), which argues that under conditions of perfect competition, and given the relative immobility of one or more factors of production, nations will specialize in those industries in which they have comparative factor advantages (favorable resources of raw materials, cheaper labor, and so forth). The relative factor endowments of different nations is thus the main reason for international trade and specialization. The principle of comparative advantage, then, predicts that countries with dissimilar resource endowments will exchange dissimilar goods. The theory does not and cannot, however, predict what sort of goods will be exchanged by countries that have similar resource endowments. But much of world trade, and most of Organization for Economic Cooperation and Develop-
ment (OECD) trade, is between countries with similar factor endowments, and they exchange predominantly similar products. Such intraindustry trade has been expanding rapidly in recent decades, even though countries have been converging in skill levels and per capita endowments of capital (OECD 1994). The “new trade theory” is an attempt to account for this form of trade. The new trade theory acknowledges that differences among countries are one reason for trade, but it goes beyond the traditional view in four main ways (Krugman 1990). 5
First, it argues that much trade between countries, especially intraindustry trade between similar countries, represents specialization to take advantage of increasing returns to scale rather than to capitalize on inherent differences in national factor endowments. Contrary to the assumptions of perfect competition and constant returns to scale that underpin the basic Ricardian theory of comparative advantage and trade, according to the new theory imperfect competition and increasing returns are pervasive features of contemporary industrial economies. 6 If specialization and trade are driven by increasing returns and economies of scale rather than by comparative advantage, the gains from trade arise because production costs fall as the scale of output increases. Second, with this view of the world, specialization is to some extent a historical accident. The specific location of a particular microindustry is to a large degree indeterminate, and history-dependent. But once a pattern of specialization is established, for whatever reason, that pattern gets “locked in” by the cumulative gains from trade. There is thus a strong tendency toward “path dependence” in the patterns of specialization and trade between countries: history matters. Third, the patterns of demand for and rewards to factors of production under conditions of imperfect competition and intraindustry trade will depend on the technological conditions of production at the micro level, and nothing can be said a priori about the evolution of factor demands. Fourth, whereas under the Ricardian model free trade is assumed to be the appropriate policy stance, the new trade theory argues that the existence of imperfect competition and increasing returns opens up the possibility of using trade policies strategically to create comparative advantage by promoting those export sectors where economies of scale - and particularly external economies - are important sources of rent. In other words, strategic trade policy may enable a nation to shift the pattern of international economic specialization in its own favor (Krugman 1980).
In Krugman’s view, these developments in the “new trade theory” both necessitate and facilitate a rapprochement between trade theory and location theory. In recent work he has compared the contrasting assumptions underlying these two, hitherto largely separate, sets of economic literature (Krugman 1993a). His geographical economics is a hybrid of the two. It combines the models of imperfect competition and scale economies used in new trade theory with location theory’s emphasis on the significance of transport costs. The interaction of external economies of scale with transport costs is the key to his explanation of regional industrial concentration and the formation of regional “centers” and “peripheries” (Krugman 1991a; Krugman and Venables 1990). His model suggests that high
5. There are in fact several different versions of the new trade theory, but the various strands all subscribe to the basic elements elaborated by Krugman in his Rethinking International Trade (1990).
6. Of course, the idea that increasing returns and economies of scale could be alternatives to comparative advantage as explanations of international specialization and trade goes back to Ohlin (1933), if not to Adam Smith. But while their importance has been recognized in principle, they invariably have been assigned a subsidiary or supplementary role in formal trade theory. The novelty of the “new trade theory” is that increasing returns and economies of scale are moved into the mainstream.
transport costs will act to prohibit the geographical
concentration of production. With
some reduction in transport costs, however, firms will want to concentrate in
one site to realize economies of scale both in production and in transport. In Krugman’s words, “Because of the
costs of transacting across distance, the preferred locations for each
individual producer are those where demand is large or supply of inputs is
particularly convenient - which in general are the locations chosen by other
producers” (1991a, 98). If
transport costs continue to fall, the model suggests that the need to locate
near to markets will disappear and production may disperse. However, given that some transport costs
will remain, the circular relation, or positive feedback, between production and
demand means that regions which have a head start in manufacturing, typically as
a result of accidental good fortune, will attract industry and growth away from
regions with less favorable initial conditions. Krugman (1991a, 1991d) argues that this
model explains the rise of the manufacturing belt in the
On the basis of this location model, Krugman (1993b)
argues that large-scale regions are more significant economic units than
nation-states. He writes that a
satellite image of the world at night shows regional agglomerations rather than
national concentrations. Furthermore, in his view, “The best
evidence for the practical importance of external economies is so obvious that
it tends to be overlooked. It is
the strong tendency of both economic activity in general and of particular
industries or clusters of industries to concentrate in space” (Krugman 1993b,
173). This tendency, he argues,
provides a decisive refutation of the competitive model of economic equilibrium,
for when one turns to the location of production in space the “irrelevance of
equilibrium economics” is compelling and there are multiple possible
equilibria. An economy’s form is
determined by contingency, path dependence, and the initial conditions set by
history and accident. Forward and
backward linkages mean that once an initial regional advantage is established it
may become cumulative. There is
therefore no automatic tendency toward an optimum solution, as apparently
“irrational” economic distributions may be “locked in” through increasing
returns. So that while Krugman
associates economic geography with path dependence, or what he calls “the
economics of qwerty,” 7
he does not neglect the possibility
of reversal and change. Rather, he
argues that when change in regional fortunes occurs it will be sudden and
unpredictable. He repeatedly uses
the example of
7. This term derives from the first line of letters (QWERTYUIOP) on the keyboard of a typewriter or word processor. That this order is the same today as it was on the first mechanical typewriters of the nineteenth century, even though more efficient sequences are possible, represents a form of “lock-in” and persistence that has analogous parallels in the economy.
it echo Ohlin (1933), Hirschman (1958), and Myrdal (1957), it strongly resembles Weber’s (1929) model of the overlaying of transport costs on agglomeration economies. Whereas Weber identified spatial overlaps, the Krugman-Venables model adds the general level of transport costs as a variable that can fluctuate over time. Given these predecessors, we should consider whether there is anything really new in Krugman’s geographical economics. In several places he himself states that he is simply retelling an old story in a more rigorous way. It would be tempting to conclude, as some critics have done, that there is nothing new in this. However, this conclusion overlooks the way in which Krugman’s reading of agglomeration has been shaped both by the developments in trade theory and by recent models of industrial organization. One of the main reasons for trade theory’s traditional neglect of the advantages which arise from increasing returns and economies of scale was the difficulty of modeling market structure. In one sense, recent developments in modeling market structure with nonconstant returns have facilitated the new trade theory (Helpman 1984; Krugman 1983a; Buchanan and Yoon 1994; Smith 1994). Hence the best place to start, in order to understand Krugman’s interpretation of increasing returns, is with these models. Two approaches are particularly relevant to Krugman’s account of geographical concentration, namely the Marshallian and Chamberlinian models.
The Marshallian approach to understanding increasing
returns is already familiar in economic geography. It is based in a long tradition that sees
economies of scale as primarily external, as arising from the specialization of
the social division of labor (Young 1928; Stigler 1951). Typically, economies of scale have been
taken to be purely external, so that the assumptions of perfect competition may
be retained (Chipman 1970). While
Krugman is aware of this long tradition, he suggests that recent advances in the
modeling of such external economies (see, for example, Romer 1990) have given
them a new tractability. He argues,
in one paper (Krugman 1981), that external economies at a national level are the
key to the uneven development of countries. Yet, increasingly, Krugman has been
reluctant to treat nations as economic units and has emphasized the significance
of external economies at a local and regional scale. Indeed, in Geography and Trade his
account of the localization of industries and agglomeration at a relatively
small scale is based on
Further reasons for Krugman’s lack of emphasis on technological spillovers become apparent when we turn to the second model of market structure that has been influential in new trade theory, namely the Chamberlinian model (Chamberlin 1949). This model of market equilibrium envisages competition among similar firms producing differentiated products which are close but not perfect substitutes. Each firm faces a downward sloping demand curve and has some monopoly power. The entry of new firms producing slightly different products eliminates monopoly profits and means that there are many little monopolists. Many explanations of intraindustry trade
by new trade theorists have been developed from this model, with the assumption of economies of scale that are internal to firms. 8 According to Helpman and Krugman (1985), these internal economies are easy to justify. They argue that firms could both achieve economies of scale and meet a demand for differentiated products from other producers and consumers by locating at one site and engaging in intraindustry trade (Krugnian 1989). Krugman sees this approach as especially relevant to intermediate products and components, where the scope for differentiation is high and the market often too small for an exhaustion of scale economies. Moreover, he argues that
where intermediate goods produced with economies of scale are not tradeable, the result will be to induce the formation of “industrial comp1exes” - groups of industry tied together by the need to concentrate all users of intermediate goods in the same country. In this case the pattern of specialisation and trade in the Chamberlinian world will come to resemble the pattern in the Marshallian world described previously. (Krugman 1987c, 319; compare Losch 1967, 109)
Both of these models of competition are implicit in Krugman’s discussion of regional and local externalities.
Over the past decade . . . it has become a familiar point that in the presence of imperfect competition and increasing returns, pecuniary externalities matter; for example, if one firm’s actions affect the demand for the product of another firm
8. Hanink (1988, 1994) describes these approaches as the theory of differentiated markets, and uses an extended Linder model to explain the consequences of geographical product differentiation.
9. The term “pecuniary externalities” was used by Scitovsky (1954) to refer to externalities arising from market imperfections of both demand and supply. Market-size effects are an important form of pecuniary external economy; the larger the market, the more individual firms can increase their output without having to cut prices. Increasing market size permits increasing returns. Such market-size effects may operate at various geographical scales, from the international to the local. Technological externalities refer to the situation where there are spillovers from the production function of one firm into those of other firms, for example when a firm makes an innovation that other firms can imitate.
whose price exceeds marginal cost, and this is as much a “real” externality as if one firm’s research and development spills over into the general knowledge pool. At the same time, by focusing on pecuniary externalities, we are able to make the analysis much more concrete than if we allowed external economies to arise in some invisible form. (Krugman 1991b, 485)
This focus on pecuniary externalities shapes Krugman’s
Krugman’s analysis emphasizes that externalities operating within and between industries in these regional agglomerations make a difference to the competitive advantage of the constituent firms. In this sense, then, Krugman’s work conveys a sense of regional competitiveness. 11 At the same time, his recent writings on the international economy have criticized certain popular definitions and uses of competitiveness, and it is important to set his regional work within the context of his more general understanding of the consequences of trade (Krugman 1994a, 1994b, 1994c, 1994d). The primary issue here is that Krugman sees all forms of international economic integration, including trade and capital and labor mobility, as essentially beneficial. For example, the specialization produced by trade raises the efficiency of the world economy as a whole and produces mutual benefits to the trading nations (Krugman 1994d). This view is partly founded on his belief that comparative advantage still remains important and useful. It is also based on his belief that the “new” trade theory’s recognition of externalities and imperfect competition highlights the potential gains from economic integration. Increased trade may allow greater economies of scale through rationalization and, in other situations, it may have a beneficial effect on oligopolistic markets by increasing competition.
The complexity of much of Krugman’s work also reflects the fact that the existence of significant externalities and nonconstant returns also opens up possible ways in which increased trade and integration may have adverse effects. Krugman (1989) highlights two main sources of adverse effects. The first is the possibility of the uneven distribution of benefits associated with the existence of excess returns in imperfectly competitive industries. A country that gains a disproportionate share of high-returns industries can gain at others’ expense, raising the possibility that trade policies designed to
10. Krugman also praises Fujita (1989) for his emphasis on market-size effects as explanations of urban agglomeration.
11. In a similar fashion, Porter (1990) argues that the geographical concentration of leading industries often reinforces and intensifies their competitive advantage.
foster these industries will lead to trade conflict. Hence, “While the possibility of actual losses from trade is probably purely academic, there is a real issue of conflict over the division of the gains” (Krugman 1989, 361). (For a global example, see Krugman and Venables 1994.) There is clearly also a regional dimension to this problem of uneven distribution. As a result of the importance of external economies and the accumulated, path-dependent advantages of certain regions, it is possible that these leading regions will capture a disproportionate share of the benefits of increased integration. A major obstacle to integration, in this view, is that its benefits are not equally shared across regions within countries.
This forms the basis of the second set of adverse
consequences identified by Krugman, namely adjustment costs. He argues that although it is costly for
capital and labor to shift into new industries, these costs represent a type of
investment. They may be deserving
of compensation, but they are not reasons to prevent or delay change. On the other hand, where these
adjustments involve significant social costs, most importantly unemployment, he
concedes that this may provide a case against moving too fast. The possibility of adjustment costs
becoming real social costs should not be dismissed lightly. One ameliorating factor that Krugman
notes is that the growth of trade between the industrialized nations since the
Second World War, especially within