The Competitiveness of Nations

in a Global Knowledge-Based Economy

H.H. Chartrand

April 2002

AAP Homepage

Ron Martin  and Peter Sunley

Paul Krugman’s Geographical Economics and Its Implications for Regional Development Theory:

A Critical Assessment (cont'd)

Economic Geography

Volume 72, Issue 3

July 1996, pp. 259-292


Abstract (Web 1)

Trade, Externalities, and Industrial Localization: The Bases of Krugman’s “Geographical Economics”

The New Trade Theory and Location

Increasing Returns and Imperfect Competition

The Role and Implications of Externalities

Krugman’s Geographical Economics and Economic Geography: A Critical Comparison (Web 2)

The Resurgence of Regional Economies

The New Political Economy of Trade

Krugman’s Model of Economic Integration and Regional Development: The Lessons of the United States for Europe?

Economic Integration and Regional Specialization

Economic Integration and Divergent Regional Growth

Trade and the Regional Policy Issue ( Web 3)

Strategic Trade Policy

Geographical Clustering and Strategic Industrial Policy


References (Web 4)


Krugman’s Geographical Economics and Economic Geography:

A Critical Comparison

Clearly, Krugman shares an interest in regional agglomeration and the geographical consequences of trade with many economic geographers.  At the same time, his treatment of these issues has been significantly different from the approaches pursued in economic geography in recent years.  In this part of the paper we shall examine the most important of these differences and consider the lessons that Krugman and economic geographers can learn from each other.  As we have noted already, a fundamental difference between Krugman’s geographical economics and the various schools of contemporary economic geography is one of method.  Krugman’s reliance on formal models means that his work is rigorous and supported by mathematical proofs.  In his view, the dependence of many of these models on unrealistic assumptions is not a grave problem nor a serious limitation.  Instead, he appears to regard these models as rough metaphors or representations of the core of real world problems (Krugman 1995).  When the models’ results are found to be inadequate, their assumptions can be modified.  In contrast, most contemporary economic geography has abandoned the use of formal modeling and is dominated instead by various types of political economy, which aim, above all, to be “realistic.”  From this perspective, Krugman’s models have an inadequate sense of geographical and historical context.  Knox and Agnew (1994), for example, argue that Krugman’s core-periphery model in Geography and Trade differs from other location models in that it does not suggest a long-term process of conver­


gence.  Instead, “the long term never arrives” (Knox and Agnew 1994, 83).  There are multiple equilibria as concentrations persist for long periods of time but may then be unraveled by new patterns of concentration.  However, Knox and Agnew insist that

concentration somewhere . . . is the perpetual rule.  So though apparently attentive to historical change, this model is static in its assumptions about the operation of economic-locational principles.  The same principles of increasing returns, imperfect competition, and agglomeration are at work in the same way all the time.  From this point of view, geographical outcomes can change but the process driving them does not. (1994, 83; original emphasis)

While this phrasing may be too strong, 12 Krugman (1991a) clearly states that the patterns of concentration that he describes are typical only of some industries under certain conditions; nevertheless, it does identify an important weakness in Krugman’s work.  He claims that the same broad locational forces which explain the growth of nineteenth-century concentrations also underlie the continued tendency to agglomeration.  Indeed, this is one reason why he is reluctant to emphasize technological spillovers as a key determinant of contemporary clusters.  At the same time, however, Krugman makes several passing references to the way in which the nature of agglomeration has changed over time.  Thus he suggests, in one paper, that the railway ad steamship were responsible for the emergence of core-periphery distinctions and that the age of this type of divergence may have passed. but such a “throw-away” suggestion requires a great deal more explanation.  The historical grounding of Krugman’s approach remains unclear and clouded by ambiguity.  What is clear is that his emphasis on continuity in the forces responsible for capital’s agglomeration contrasts with economic geographers’ focus on historical patterns of restructuring.  However, the relative merits of this more historical approach depend on precisely how change is theorized and explained.  It is impossible here to talk about economic geography as a whole; we have therefore selected two relevant areas of work, namely the recent literature in industrial geography on regional agglomeration and recent writing on theorizing the geography of trade.13


The Resurgence of Regional Economies

During the past decade, the most influential approach to industrial organization within economic geography has been the notion of a fundamental transition from Fordist mass production to more flexible production methods, such as flexible specialization.  Scott and Storper (1992a, 1992b), Scott (1988), Storper and Walker (1989), and others have argued that internal economies of scale and scope have been undermined by increased market uncertainty and technological change.  They argue that the response has been horizontal and vertical disintegration, or an externalization of production, which enables a greater ability to meet differentiated demand and a greater adaptability to market forces.  Where a multiplicity of linkages are created which have geographically sensitive transaction costs, externalization is positively related to agglomeration.  In this view, “Agglomeration is a strategy whereby producers ease the tasks of transactional interaction because proximity translates into lower

12. After all, most of the main schools of economic geography and political economy may be criticized on similar grounds.  All, for example, assume that the basic laws of economic development (as they perceive them) remain essentially unchanged as capitalism evolves over historical time.

13. These are two of the leading fields in contemporary (post- Marxist) economic geography.  It would require another paper to consider Krugman’s work in relation to the complete corpus of geographical work on uneven regional development.

269 Index

costs and wider opportunities for matching needs and capabilities” (Scott and Storper 1992b, 13).  In summary, the shift to flexible specialization has been responsible for the rise of new industrial districts and for the new, or renewed, significance of regional agglomeration (Sabel 1989).  While there are many contrasts between this “new industrial geography” and Krugman’s geographical economics (see Table 1), we shall focus on three issues: the treatment of industrial and market structure, of externalities, and of nonmarket transactions and relations.

As we have seen, Krugman tends to rely on several abstract models of monopolistic and oligopolistic market structure.  The assumptions of these models are in some ways unrealistic, but the underlying rationale is that they are useful because of the pervasive presence of imperfect competition.  In contrast, the flexible specialization approach has envisaged a new type of competition involving downsizing and disintegration and therefore a movement back toward perfect competition.  However, the idea that corporate disintegration is a necessary response to uncertainty can be criticized (see, for example, Lovering 1990; Phelps 1992).  Moreover, as Phelps (1992) has argued, Scott’s analysis of the causes of agglomeration pertains primarily to situations approximating to that of perfect competition.  In Phelps’s view, “The assumption of near-perfect competition is otherwise implicit in an analysis which applies to single plant firms and neglects considerations of differential economic power embodied in linkage structures” (1992, 41).  This is especially problematic when the analysis is applied to international trade.  As Markusen (1993, 287) writes, “Most important internationally traded industries are now multinucleated, with large national firms thrust into more spirited competition with similarly sized and politically well-endowed firms from other nations.”  Hence, “the number of players is relatively small, their sizes and clout are varied, and none of them is

Table 1

A Comparison of Krugman’s “Geographical Economics” with the “New Industrial Geography”


unaware of the behaviour of its neighbors.  These are characteristics of oligopolized markets, not perfectly competitive ones.”

Several arguments have been used to support the association of near-perfect competition with agglomeration.  One is the finding that in some industries and places larger producers are located away from local clusters of industry (Hoare 1975; Scott 1986), and another is the observation that the decline of some industrial districts has been associated with the concentration of production into larger firms (Steed 1971).  These are contingent findings, however.  For example, Scott (1992a) notes that large producers are integral to the Southern Californian computer districts.  Even where large firms are not found in local industrial clusters, they may be central to the regional and metropolitan concentrations modeled by Krugman.  It cannot be assumed that internal economies of scale and scope act against agglomeration.  Indeed, the intraindustry trade literature implies that, with increasing product diversity, internal economies and agglomeration become more closely linked.  There is clearly a need to research the relations between market structure and locational dynamics in more detail.

This difference on the issue of competition has important consequences for understanding externalities.  In order to compare Krugman’s representation of externalities with that used in the “new industrial geography,” it is helpful to set both approaches within a general framework.  De Melo and Robinson (1990) argue that three main approaches to externalities are apparent in recent economic literature.  The first is the Marshallian externalities approach, which we discussed above.  They suggest that some parts of endogenous growth theory fall within this approach.  For example, in an article on the externalities arising from human capital formation, Lucas (1988) talked about increasing returns at an economy wide level.  The second type of externality that De Melo and Robinson identify are those that result in uneven rates of growth and occur with imperfect competition.  Again, there are examples from endogenous growth theory: Romer (1990) sees investment in R&D in a situation of monopolistic competition as generating externalities in disembodied knowledge.  The third type of externality arises from demand spillovers between sectors and industries.  Murphy, Schleifer, and Vishny (1989), for example, argue that there are low-level equilibrium traps, where industrialization remains unprofitable.  Industrial production only becomes profitable for individual firms in the context of more general demand linkages.

This framework provides a means of comparing the two approaches to agglomeration (Table 2).  Krugman’s focus on market-size effects is clearly closest to, and draws most heavily on, pecuniary externalities. 14  As we have seen, his explanation of local clustering also invokes certain types of Marshallian external economy.  Importantly, he has tended to downplay the significance of externalities based on spillovers in technological knowledge.  Krugman (1987c) describes these as “elusive,” preferring to concentrate on externalities that can be modeled.  The difference between his approach and that of the “new industrial geography” is apparent.  In accordance with the reliance on situations of near-perfect competition, Marshallian external economies have been at the forefront of the industrial districts literature.  As Phelps has argued, and as Krugman’s account demonstrates, the external economies that can be used in this approach are only a subset of the range available (Phelps 1992).  To a certain extent this limitation has been weakened by those revisionist studies of industrial

14. Krugman (1993b, 1995) describes this type of external economy as similar to those envisaged in “Big Push” interpretations of industrialization.  He argues that a large-scale program of industrialization can take advantage of external economies and complementarities and so reduce the risk of investment (see Rosenstein-Rodan 1943).

271 Index

Table 2

Comparison of the Treatment of Externalities in Krugman’s “Geographical Economics” and the “New Industrial Geography”

districts which argue that large producers can imitate decentralization.  However, the contradiction between a commitment to perfect competition and the dependence of Schumpeterian models of “creative destruction” and local technological spillovers on imperfect competition cannot be resolved easily. 15

The differences between Krugman’s geographical economics and the recent work in economic geography on regional development are not confined to industrial structure and externalities, but also extend to the question of nonmarket transactions.  Thus another important contrast between Krugman’s approach and those of economic geographers is the manner in which the increasing power of larger producers has been related to contemporary localizations of industry.  In economic geography there has been some dissatisfaction with the way in which the flexible specialization literature has ignored the increasing internationalization of firm structures and globalization more generally (Amin and Robins 1990; Gertler 1992).  Consequently, there has been an interest in the way in which large firms interact with industrial districts.  In contrast to Krugman’s market-size effects, however, the main emphasis has been on the intermingling of firm and local networks (Amin and Thrift 1992; Grabher 1993).  Networks have usually been defined as types of organizational relation that are neither market transactions nor hierarchies, and the term has been used to refer to cooperative and mutually beneficial relationships among producers (Cooke and Morgan 1993).  Using this definition, the boundaries of firms become blurred, and firms and districts become intermingled.  On the one hand,

15. Schumpeterian models of “creative destruction,” technological spillovers, and endogenous growth depend on imperfect competition.  Typically, the incentive for firms to develop new products and processes stems from the temporary monopoly profits which they can earn (Grossman and Helpman 1991; Aghion and Howitt 1993).  This sits uneasily with the new industrial geography’s emphasis on near-perfect competition.


Krugman’s contrary emphasis on pecuniary relations is a reminder to geographers not to lose sight of market effects.  But on the other hand, Krugman’s neglect of externalities that are intangible and leave no paper trail appears too restrictive.  As Jaffe, Trajtenberg, and Henderson (1993) have pointed out, knowledge flows do sometimes leave a paper trail, in the form of citation of patents.

The interest in network forms of organization in economic geography reflects a more general concern to examine the ways in which economic activities are “embedded” in, and made possible by, social and cultural conditions.  This has been applied with particular force by Storper (1992a) to high-technology districts.  As Harrison (1992) notes, this interest in embeddedness has been the distinctive contribution of the recent geographical literature on industrial districts. 16   This stands in complete contrast to Krugman’s rejection of invisible externalities.  However, as Storper (1992b) argues, in the context of increasing market contestability, it is difficult to explain the continuing competitive advantage of certain districts if their conventions, rules of behavior, and implicit accords are not taken into account.  Conversely, the decline of other regions appears to be partly a result of the “lock-in” of outmoded conventions and rules of behavior (Grabher 1993).  Krugman’s rejection of non-market linkages seems to be made primarily on the grounds that if externalities cannot be modeled then they have to be assumed a priori, so that the analyst can say anything she or he likes about types of spillover.  But this rules out other methods of research and more sociological approaches.  Moreover, Krugman’s reluctance to envisage nonmarket linkages seems to conflict with his commitment to new-Keynesian economics, where expectations and conventions are central.  He himself shows (Krugman 1991c) that, under certain conditions, expectations may affect the course of regional development.  But, if they are to be understood, expectations cannot be treated as exogenous “animal spirits.”  Rather, they are an integral part of social conventions and meanings, and their formation should be an important area for regional research.  Our conclusion, then, is that there is a need for a greater exchange of ideas between Krugman’s work and the geographical literature.  But this applies not only to research on regional agglomeration but also to that on trade more generally.


The New Political Economy of Trade

Recent years have seen a growing interest by economic geographers in the spatial patterns of international trade.  While this work has lacked a comprehensive theoretical framework, it has nevertheless been characterized by shared themes revolving around the inability of conventional geographies, based on Ricardian comparative advantage, to explain fully the complex character of contemporary international patterns.  In accord with the “new trade theory,” this geographical revival has stressed the importance of shifts in the world economy and the rise of intraindustry and intracorporate trade.  One of the defining features of this revival has been a call to study the ways in which the geography of trade is shaped by states and by trade regimes.  This emphasis on state policy, and the interpretation of trade on which it is based, contrasts with Krugman’s approach in ways that raise fundamental questions about the effects of trade and its policy implications.

Some years ago, Johnston (1989) called for trade to be explained as part of a holistic theory of uneven development that combines the logics of capitalism and the policies of states.  To some degree, his

16. Amin and Thrift (1994) describe this embedding as best summed up by the phrase “institutional thickness.”  This is defined by a strong institutional presence in a local area, high levels of interaction among these institutions, strong social structures, and a collective awareness of common enterprise.

273 Index

plea for an enlarged research agenda has begun to be recognized.  Grant summarizes recent developments as follows:

The unifying theme in newer approaches is their study of the interactions between governments and firms and their connections to trade and industrial policy within the context of a politically and economically competitive world economy, one in which governments attempt to “create” the most advantageous environment for national business.  Accordingly, approaches recontextualize comparative advantage to include an understanding of developments in the trade-industrial policy arena, (1994, 301)

In line with this theme, Grant focuses on the role of governments, especially the formation of regional blocs, and the role of firms as the bases of a more comprehensive theory.  Moreover, he argues that high-technology trade occupies a key place in any new theory, as success in high-technology bestows national benefits on productivity and high-wage job creation (see also Drache and Gertler 1991).  Likewise, in their recent study of trade in textiles and clothing, Glasmeier, Thompson, and Kays (1993) contend that it is necessary to understand how the actions of the state influence the structure of global competition.  Indeed, they conclude that state actions have superseded market forces as the regulator of the industry’s geographical evolution.

The conceptual movement away from orthodox comparative advantage explanations has been most fully spelled out in Trading Industries, Trading Regions, edited by Noponen, Graham, and Markusen (1993).  Here again it is argued that success in trade is fundamentally shaped by government intervention.  In a chapter in that volume, Howes and Markusen claim that governments have played a key role in creating and maintaining industrial leadership, and that “in a world with governments’ successfully conducting such industrial and trade policies, open economies without such efforts will find themselves the targets of import penetration and potential export market shrinkage”  Howes and Markusen 1993, 4).  In this view, factor endowments can be used to explain trade in minerals, agricultural goods, and some labor-intensive consumer goods, but the majority of trade between developed market economies can only be explained by a “dynamic revisionist” theory.  This has four major tenets that contradict orthodox trade theory.  First, the mix of sectors matters, as some industries have greater growth and productivity differentials.  Second, growth is not constrained by factors but by demand for the product.  Third, in some industries rapid growth leads to continuing success due to increasing returns.  Fourth, because of the existence of increasing returns, comparative advantage may conceivably be created by strategic intervention on the part of nation-states and regional authorities.  On this basis they argue that the orthodox view that free trade means growth for all regions is mistaken; instead, “there is some danger that the unfettered pursuit of free trade will actually depress wages and employment and lower world living standards” (Howes and Markusen 1993, 35).  Furthermore, Markusen (1993) argues that in the United States free trade and laissez-faire strategies have produced persistent unemployment and a waste of infrastructure.

While this “dynamic revisionist” theory shares an emphasis on “new trade theory” with Krugman, it more closely resembles the strategic trade views of authors such as Tyson (1992) and Reich (1991), whom Krugman (1994a, 1994c) has recently criticized. 17  As we have noted, Krugman remains convinced that the mutual benefits of greater international trade outweigh the costs.  Moreover, in his opinion, comparative advantage is not just a sector-specific theory, it remains a general principle that explains the beneficial

17. For a debate on Krugman’s critique see the discussion on “The Fight over Competitiveness” in Foreign Affairs (1994a), Friedman (1994), and The Economist (“The Economics of Meaning” 1994).


consequences of trade.  The concept makes clear that absolute productivity advantage in some areas is not necessary for a country to gain from economic integration.  Trade, therefore, is not a zero-sum game, so that concerns about national competitiveness are misplaced and unfounded.  Krugman (1987b) concedes that the intellectual case for free trade has been weakened and that it is not an absolute ideal, but he believes that it is still the best general policy or rule of thumb.  But Krugrnan’s position faces several key questions.  The first is the extent to which this continuing use of comparative advantage is compatible with his own emphasis on the pervasive presence of increasing returns.  Kaldor (1985), for example, argues that the presence of increasing and diminishing returns conflicts with the basic tenets of Ricardian comparative advantage.  Simply put, he contends that diminishing returns may mean that the resources released by trade will not necessarily be employed in other sectors, so that there is a real possibility of absolute loss (a “negative sum” game).  Conversely, increasing returns in some industries may inhibit the transfer of resources elsewhere.  Krugman’s economic geography pays insufficient attention to these problems.  This is reflected by his insistence that it is pointless to try to identify high-return sectors, so that the mix of sectors does not really matter. 18  Given his insistence on the importance of productivity, it is surprising that he devotes little attention to the extent to which high-technology sectors do generate the productivity spill-overs which some authors have suggested (for example, Hanink 1994).

The second question is whether Krugman underestimates the significance of adjustment costs and the obstacles to regional adjustment.  On the one hand, Krugman is committed to a nonequilibrium view of economic geography in which there is no process of convergence to a spatial equilibrium where all factors are equally rewarded.  He rejects the neoclassical faith in the efficiency of markets on the grounds that the collective result of individual choices may be to “lock-in” a bad result.  On the other hand, in a methodological sense, Krugman (1993a) insists that all economic models should contain a well-specified equilibrium.  By this he means that they should specify how individuals behave and show how market outcomes emerge from the interaction of these individual behaviors (Krugman 1993a, 115-16).  He holds these two opposing convictions together, it seems, through a commitment to a “new Keynesian” brand of economics.  According to this, economic trends and patterns are the products of innumerable individual decisions, but these decisions are not perfectly rational and informed.  Instead they are frequently both near-rational and individually reasonable and sensible.  However, in imperfectly competitive markets the aggregate result will be unstable and irrational.  In his words, “What look like highly irrational outcomes in the marketplace are caused by the interaction between imperfectly competitive markets and slightly less than perfectly rational individuals” (Krugman 1994c, 213).  But if emphasis is placed on the second of these factors, then the position is readily reassimilated into a neoclassical view of the economy.  It lends itself to the view that markets would adapt efficiently and rapidly if only people would behave rationally.  This is exemplified, perhaps, by Krugman’s (1993e) argument that Eurosclersosis, or the problem of a persistently high level of unemployment in Europe, has been caused by the effects of welfare states on labor markets.  The whole

18 Krugman’s (1994c) argument is that it is wrong to assume that high-technology sectors such as computers and aerospace are the sectors with highest value added per worker.  In fact, he notes that in the United States the real high-value industries are extremely capital-intensive sectors, such as cigarettes and oil refining.  This says nothing, however, about the possibility of positive spillovers from the high-technology sectors.


question of adjustment to the effects of trade is one that Krugman has recently considered explicitly in terms of the impact of economic integration on regional development, particularly in the European Union, and it is to this aspect of his work that we now turn.


Krugman’s Model of Economic Integration and Regional Development:

The Lessons of the United States for Europe?

The regional consequences of European economic integration is an issue that has attracted surprisingly little attention from economic geographers.  At the heart of this issue is the question of what the impact of progressive economic and monetary integration in the European Union (EU) will be on regional patterns of economic growth, employment, and income across member states.  Economists have offered two opposing answers to this question.  On the one hand, there are those who believe that the free movement of goods, services, and capital associated with European economic and monetary integration (EMU) should lead to regional convergence, not only in factor returns and economic performance but also in economic structure.  To the extent that wages and other costs are lower in the less productive and slower-growing regions, the removal of barriers to trade and factor movements, it is argued, should enable industries and services in these regions to better exercise this comparative advantage and to attract increased flows of capital investment. 19   This optimistic scenario is, on balance, the view taken by the European Commission (Commission of the European Communities 1991, 1994).  In contrast, others argue that economic integration will, intensify rather than reduce regional imbalances in growth and income across the European Union.  Instead of leading to equalizing centrifugal movements of firms and investment toward depressed and peripheral regions within the European Union, economic integration is likely to stimulate a spatial reconfiguration of economic activity in favor of growth regions precisely because these are the areas that already enjoy greater comparative advantage in terms of access to markets, inputs, expertise, and business infrastructure. 20

Krugman falls into the second of these two camps, although he appears to subscribe to two somewhat different models of regional divergence.  In an earlier paper (Krugman and Venables 1990), he follows a core-periphery argument not unlike that in Geography and Trade.  Although the removal of barriers to trade and movement of capital and labor within the European Union will increase the inflow of capital into, and the relative competitiveness of, the low-wage peripheral regions, given transport costs this centrifugal process is on balance likely to be outweighed by further concentration of industry and employment in the high-wage core regions, because these areas have the largest markets, well-developed external economies and infrastructures, and a comparative advantage in terms of

19. Additionally, economic integration represents a major supply shock to such regions, since it exposes them to the full force of competition elsewhere in the system.  Such shocks, the argument continues, should (allowing for adjustment lags) eliminate inefficient firms, work practices, and products in depressed regions and improve their supply-side competitiveness and flexibility.

20. Because the gains foreseen from completion of the internal market are thought to be generated mainly endogenously, the various processes of resource allocation are bound to cumulate resources in the leading core regions.  It is the historically established competitive advantage of the growth regions which enables them to capture a disproportionate share of the benefits of economic integration.  As for the depressed and lagging regions, economic integration is seen as bringing prolonged problems of adjustment and the need for greater levels of spending on regional policies.

276 Index

relative accessibility.  His second approach is more emphatic, but different in its specific arguments.  In his paper on the “Lessons of Massachusetts for EMU,” he supports the movement toward European economic integration as “a generally good thing,” but argues that it will lead to greater regional instability and divergence of regional growth rates (Krugman 1993d, 241).  In developing this thesis he begins by drawing on his earlier ideas on trade and the localization of industries that we have discussed above:

For regional issues … in the EC, … the key aspect of regional specialisation is the dependence of regional economies on export clusters held together by Marshallian external economies … Are such regional clusters more likely to form in a more integrated economy?  The answer is definitely yes. (Krugman 1993d, 244)

These ideas are then used in a somewhat different way from his earlier work to produce a theoretical account that not only carries over some of the problems we have already highlighted, but also introduces additional elements of contention.

The gist of this second model may be summarized as follows.  First, given the existence of increasing returns, the expansion of interregional trade that EMU will bring about will lead to greater regional industrial concentration and specialization along essentially arbitrary lines.  Once under way, there will be a tendency for this regional specialization process to become “locked in” by the operation of location-specific external economies.  Second, Krugman argues that this increased regional specialization will render the European regions much more subject to random, idiosyncratic demand and technology shocks, so that region-specific recessions and crises will be more likely to occur. 21  Third, when combined with the increased factor mobility that integration will promote, such region-specific shocks will lead to divergent long-term regional growth paths.  Thus, fourth, given that under EMU member states will no longer be able to use the exchange rate mechanism as a policy instrument (see also Krugman 1989), the only way regional adjustment problems can be ameliorated is by transferring a significant part of national budgets to the European Union to allow fiscal federalism to function as an automatic stabilizer.

Thus, in contrast to his previous work - for example, in Geography and Trade (1991a) and Krugman and Venables (1990) - Krugman argues that the process of uneven regional development that EMU may be expected to produce will not be one of cumulative divergence into a core-periphery pattern.  He believes that the forces generating this form of uneven regional development have probably reached their limit in advanced industrial nations; indeed, he suggests that in both the United States and Europe industrial activity is becoming much more evenly distributed geographically (Krugman 1993d).  Rather, in his view the process will be one of increasing regional export specialization, with the result that the pattern of regional growth and decline will be more unpredictable, dependent on the particular incidence of random demand shocks.  Hence, unlike the argument in Geography and Trade, past regional success need not be self-reinforcing, and even prosperous regions may experience sudden reversals of fortune.

Another distinctive feature of Krugman’s exposition is the method he uses to support his theory empirically.  The United States is taken to be the sort of integrated economic and monetary unit which the European Union is seeking to emulate, so that regional experiences in the former are considered to be a good guide as to what to expect in the latter.

21 In an earlier paper, Krugman (1989) stressed that increasing interdependence in Europe acts as a buffer against regional and national shocks, but this buffering effect only acts against locally generated recessions such as those caused by investment slumps.


Using simple measures of the dispersion of economic structure, Krugman (1991a, 1993d) finds that the broad regions of the United States are more industrially specialized than are European countries.  Furthermore, a comparison of Belgium with the state of Ohio is used to suggest that the regional employment growth rates in the United States are more unstable than in the European Union.  He then examines the disparities in long-term growth rates between certain states in the United States and among the main EU countries and finds that these disparities are larger in the United States than in Europe.  In addition, he uses the recent economic slump in the New England region of the United States as an illustration of how, in a monetary union, regional industrial specialization can give rise to pronounced local instability in the face of region-specific demand shocks, and how such shocks can lead to permanently lower levels of employment (Krugman 1993d).  These various empirical results are taken as lending support to his thesis that increased market integration in the European Union will lead to more regional specialization and unequal growth.  In our view, however, Krugman’s empirical examples and findings are not of themselves sufficient to prove his case, and several features of his analysis are problematic.


Economic Integration and Regional Specialization

The first problem concerns the evidence on regional specialization.  What is the “regional” scale being referred to?  The “regions” used by Krugman in his comparisons of regional specialization and regional growth rate disparities in the United States and the European Union are extremely aggregate ones: the four “Great Regions” and individual states in the former and whole countries in the latter.  Krugman argues that these spatial units are of roughly similar size, and thus broadly comparable.  That may be so, but they do not necessarily represent the geographical scale at which local external economies and the processes leading to industrial clustering actually operate.  The basic point is that the analysis of localization economies requires an identiflcation of the relevant regions as economic areas and the relevant level of industrial disaggregation at which to measure the extent of geographical concentration and specialization.  The geographical literature on “new” flexible industrial districts indicates that such clusters are in fact quite localized, and far smaller than the broad spatial units used by Krugman.  Certainly in the European Union, local differences in economic structure and economic growth rates within member countries (for example, at the so-called NUTS1 and NUTS2 level regions) are much larger than the disparities between countries (Collier 1994; Dunford 1993; Dunford and Kafkalas 1992).  Likewise, as von Hagen and Hammond (1994) argue, the metropolitan rather than the state or broad regional level is the most meaningful one for analyzing geographical differences in industrial concentration and localization economies in the United States, a view to which, as we have already noted, Krugman has elsewhere subscribed.  These findings imply that Krugman’s method of comparing the European Union with the United States will generate different results according to the geographical scale used to define economic regions in the two areas.  Indeed, it may even be that at some geographical scales regional specialization is not in fact greater in the more-integrated economy of the United States than in the European Union.

In any case, is increasing regional industrial specialization an inevitable outcome of economic integration?  While the existence of external economies and localization economies in the European Union could well lead to the increased regional specialization that Krugman predicts (Baldwin and Lyons 1990; Cabellero and Lyons 1990, 1991; Martin and Rogers 1994a, 1994b), some observers have argued that product market integration in the European Union will increase the


scope of intraindustry trade there still further, and that this is likely to render regional industrial structures increasingly similar over time (Commission of the European Communities 1991; Eichengreen 1993; Emerson, Anjean, and Catinat 1988).  Indeed, possible evidence for this effect is provided for the United States by Krugman: as he shows, U.S. statistics indicate that regional specialization there has actually been declining since the Second World War (Krugman 1991a, Chap. 3).  He suggests that this may be a statistical illusion, in that specialization may have become more difficult to measure but may not necessarily be less in fact.  However, there is also evidence from Europe that economic integration and increased trade lead to regional industrial diversification rather than specialization (Peschel 1982).  Indeed, both the definition of regional economic “specialization” and the question of how specialization actually influences regional instability are not straightforward issues. 22  As a number of writers have shown, the empirical patterns of regional shocks in both the United States and the European Union appear to be more complex than those posited by Krugman (see, for example, Bayoumi and Eichengreen 1993; Palmini and Cray 1992; von Hagen and Hammond 1994).  There seems to be no simple relationship between economic integration, regional specialization, and regional shocks.  Both the pattern and severity of shocks will depend not only on the degree and geography of regional industrial specialization, but also on how such shocks are transmitted between regions (for example, through interregional input-output linkages and the impact of government policies) and on how flexible regional labor markets are in adjusting to disturbances.  In short, much more theoretical and empirical analysis of regional industrial specialization within both the United States and the European Union is required before the former can be taken as a guide of what to expect in the latter.


Economic Integration and Divergent Regional Growth

This last point links with the third element of Krugman’s thesis, that demand shocks in an integrated Europe will have permanent regional growth effects, in the same way that temporary policies may have long-term implications (Krugman 1987a).  Suppose a region experiences a decline in the demand for its clusters of export industries.  This would put downward pressure on relative wages and other factor costs in the region.  If relative wages and other costs fall, this would help to restore the region’s competitiveness vis--vis other, higher-cost regions, so that new industries would be attracted there and demand and growth should be restored.  As Krugman puts it,

Regions that have been unlucky in their heritage of industries from the past will have lower costs than lucky regions, and will therefore be more likely to break into industries in the future.  We would expect this process to put limits on the extent of regional divergence in growth. (Krugman 1993d, 248)

Unfortunately, however, according to

22.  There is a sizable literature on this topic, although it is not referred to by Krugman (for example, see Barth, Kraft, and Wiest 1975; Conroy 1975; Brewer 1984; Jackson 1984; Kurre and Weller 1989).  Much of this is based on what is called a “portfolio” approach to the analysis of regional industrial specialization.  This type of analysis, first applied to regional economics by Conroy (1975), borrows the concepts of expected return and risk from theories of the optimal diversification of financial portfolios developed by Markowitz (1959).  The regional industrial structure may be conceptualized as a “portfolio” which provides “returns” to the region in the form of employment, income, and tax revenues.  These returns are associated with risk - arising from demand and technology shocks - as represented by the variance and covariance in the returns.  It is this measure of risk, “the portfolio variance,” which measures the degree of instability of the region.

279 Index

Krugman labor mobility prevents the wage flexibility mechanism from bringing regional growth rates into balance in this self-correcting way.  To the contrary:

An unfortunate region will not have lower factor prices for very long: capital and labour will move to other regions until payments are equalized.  This means, however, that there is no particular reason to expect a region whose traditional industries are faring badly to attract new industries.  It can simply shed people instead.  The implication is that relative output and employment of regions should look more like a random walk than like a process that returns to some norm. (Krugman 1993d, 248)

In developing this argument, Krugman draws on Blanchard and Katz’s (1992) study of patterns of growth among U.S. states.  According to these authors, while employment growth rates differ consistently across U.S. states, unemployment rates and wages vary much less, suggesting that when states are hit by demand shocks workers react by relocating (see also Barro and Sala-i-Martin 1992).  There is no discernible tendency for states to recover lost jobs: relative regional unemployment returns to normal through the out-migration of workers.  This would seem to be in contrast to the adjustment process in the European Union, where historically factor mobility has tended to be far lower than in the United States and regional unemployment disparities appear to be characterized by greater hysteresis (Eichengreen 1993).  Krugman draws the obvious implication that if Europe moves toward U.S. levels of regional specialization and factor mobility, disparities in economic growth rates among countries and regions may be expected to increase.

Labor mobility is thus central to Krugman’s model of divergent regional growth.  In this respect his analysis is similar to local “endogenous growth” models, in which labor mobility intensifies local disparities in the accumulation of human capital and hence long-term development (Grossman and Helpman 1991; Bertola 1993).  In this respect we find it somewhat curious that Krugman is at pains to distinguish his model of uneven regional development in the European Union not only from “core-periphery” models of cumulative concentration but also from “local endogenous growth” models (Krugman 1993d).  His own model implies a similar cumulative divergent growth mechanism, at least in the sense that interregional shifts in labor prevent the reequilibration of regional growth rates.  The question mark over his analysis is exactly how far labor mobility will increase in an integrated Europe.  Although interregional migration across national borders will in principle be unrestricted, there are further reasons to doubt whether labor mobility will ever reach the levels found in the United States.  The marked cultural and language differences across Europe will continue to be a significant barrier to migration for many groups of workers.  But if this form of adjustment to regional shocks remains slow, where does this leave Krugman’s view of regional long-term growth differences in an integrated European Union?

The implication of his model is that if labor mobility is low, then local (downward) relative wage flexibility will serve to restrain the degree of divergence between regional growth rates.  Unfortunately, wages in the European Union do not seem to be particularly flexible:  European labor markets appear to be more rigid or “sclerotic” than their American counterparts, a point highlighted by Krugman (1993e).  In the European regions, adverse sectoral demand shocks trigger greater unemployment, without the equilibrating mechanisms of labor migration or downward relative wage movements (the rigidity of the latter possibly reflects the considerably higher rates of institutionalized wage setting among workers and the availability of more generous unemployment benefits in the EU countries compared to the United States).  The Commission of the European Communities (1990) argues that EMU, by increasing the credibility of fiscal authorities’ commitment not to bail


out depressed regions, should encourage workers in such areas to moderate their wage claims, thus imparting greater local wage flexibility.  In practice, little is known about how far regional relative wages would have to fall in order to stimulate capital inflows and the restoration of employment.  Equally, we still know little about interregional productivity and technology spillovers, which may offset the need for wage reductions (Jaffe, Trajtenberg, and Henderson 1993; Audretsch and Feldman 1994).  In short, it is by no means obvious whether increasing integration in the European Union will lead to convergence or divergence of regional growth.  The evidence so far would seem to suggest that “club convergence” may be the most likely outcome, with convergence within the northern, core regions, on the one hand, and within the southern and peripheral regions, on the other, but little or no convergence between these subsets (Button and Pentecost 1993; Chatterji 1993; Neven and Gouyette 1994).

Thus, though suggestive, Krugman’s arguments about the impact of economic integration on regional trade, specialization, instability, and long-term growth disparities in the European Union are problematic and limited.  Comparison between the United States and the European Union in terms of “regions” and their structures, shocks and reactions to them is not, perhaps, as valid as Krugman and others (such as Eichengreen) assume.  We do not have a counterfactual history for the United States - that is, a picture of what regional development would be like if the United States was not an economic and monetary union.  Nor do we know what would have happened to the regions of the European countries in the absence of the formation of the European Community and its recent movement toward EMU.  Finally, what of Krugman’s views on the regional policy implications of European integration?  To assess this aspect of his analysis we need to look at the policy debate within the new trade theory more generally.


previous page

next page

AAP Homepage