1、Roepke Lecture in Economic GeographyRegional Context and Global TradeMichael StorperInstitut dEtudes Politiquesde Paris, London Schoolof Economics, andUniversity of California,Los Angeles13 rue de luniversit75007 ParisFrancemichael.storpersciences-po.frm.storperlse.ac.ukstorperucla.eduKey words:inno
2、vation diffusiongrowth theoryMAR externalitiesglobal tradecontextbehavioral economicsabstractHow should we think of the role of regions in relationto the global economy? Theory has surprising gapswhen it comes to building a unified vision of thesetwo scales of development.Two contributions to sucha
3、vision are proposed in this article. First, the rela-tionship between geographic concentration andthe regional economic specialization it underpinsand globalization should be theorized as a dynamicprocess. Standard location and trade theory is notadequate for this task; instead, the dynamic relation
4、-ship can be captured through growth theory. But cap-turing this dynamic relationship requires correctinggrowth theory to separate its local and its globalcomponents, which are, respectively, Marshall-Arrow and Romer externalities. Second, the missingelement in all theories of geographic concentrati
5、onand locally specialized development is an elementlabeled “context” here. A theory of context, in turn,raises important new questions about the dynamicwelfare and developmental effects of contemporaryprocesses of fragmenting and relocating productionat a global scale.The Roepke Lecture inEconomic G
6、eography wasestablished to honor thelate Professor Howard G.Roepke, who served on thefaculty of the University ofIllinois at Urbana-Champaign from 19521985.The original lectureseries ran at the annualmeetings of the Associationof American Geographers(AAG) from 19861994.Economic Geography, theUnivers
7、ity of Illinois, andthe AAG Economic Geog-raphy Specialty Groupdecided to resurrect andcosponsor the lectureseries in 2007.1ECONOMICGEOGRAPHY85(1):121.2009ClarkUniversity.www.economicgeography.orgAcknowledgmentsAn earlier version of thisarticle was delivered as theRoepke Lecture in EconomicGeography
8、 at the annualmeeting of the Association ofAmerican Geographers inBoston (April 2008), as wellas at the DRUID SummerConference in Copenhagen(June 2008).This versionbenefited from thecomments made at bothmeetings. I particularly thankGilles Duranton,AndersMalmberg, Jim Love, andStefano Breschi, for a
9、cting asdiscussants on the paper, andPeter Maskell for including itin the DRUID program.YukoAoyama encouraged me toprepare the paper forpublication, and the editorsof Economic Geographyprovided many useful pointsfor clarifying the argument.Markus Perkmann and EwaldEngelen provided detailedcriticisms
10、 that helped merevise the paper, andparticipants at the Societyfor Advancement ofSocio-Economics in San Jos,Costa Rica (July 2008)offered a number ofstimulating new points thatalso helped me restructurethe initial argument.Anyerrors are my soleresponsibility.Regional Variety and the Gainsfrom TradeT
11、he combined effects of the division of labor,regional specialization, and gains from trade arewidely agreed to be one of the two main forces behindworld economic growth since its modern takeoffaround 1820, the other being technological innovation(Mokyr 1990; North 2005). Contemporary debatesaboutthe
12、geographicreshufflingofoutputandemploy-ment through outsourcing and “offshoring” askwhether we have crossed some kind of new threshholdin the world division of labor. There are now increas-ingly fine geographic divisions of labor, not only bybroad “functions” but by “tasks” that affect regionalpatte
13、rns of specialization in ways that may be unprec-edented (Grossman and Rossi-Hansberg 2006). Doessuch fragmentation fundamentally alter the process ofeconomic development, so that it is no longer mean-ingful to speak of local and regional economies asfunctioning systems?Are we on the verge of replac
14、ingthem with a radically new type of geographicallydistributedsystemoforganizingproductionandtrade?In this article, I propose elements of how to thinkabout these questions, asking how they change thefunctional dimensions of the economy and its welfareoutcomes. Trade theory is confident about the pot
15、en-tial welfare effects of a great transformation of thegeography of the economy: there may be new kinds ofadjustment costs from the recomposition of local andnational economies, but there will be gains to special-ization, resulting from the process of fragmentationand trade, for the world economy a
16、s a whole. Adynamic extension of this notion can be found inpolitical economy: that the “whip” of interplace com-petition increases incentives to firms in differentplaces to become more efficient over time (Wolf2004). These effects hold even when a particularcountry or region has historically benefi
17、ted fromstrong localization effects: according to trade theory,by definition, the geographically distributed systemsthat replace local clusters will be more globally effi-cient and hence generate overall welfare gains,although certain localities may not be winners at alltimes. Formally, models of th
18、e gains to specializationand trade are based on optimizing locations so as tomaximize aggregate gains in productivity, with tech-nology held constant. External shocks (technologicalchange among them) can then change the optimalgeographic pattern.ECONOMIC GEOGRAPHY2Developmenteconomistshavenonetheles
19、sidentifiedsomehighlycontestedexceptionstotheassumptionthattheoveralllevelofgeographicconcentrationanddiffusionisalwaysand everywhere optimal: economies of scale create advantages for bigger markets andpenalize smaller ones when trade costs are high; the timing of entry is important, so thatearly mo
20、vers can “fill up” the available slots for agglomeration and specialization,especially when economies of scale are strongly positive; sequencing is important, in thatlateentrantsorbigentrantscansufferdeterioratingtermsoftrade;andinfantindustriesaregenerally not diffused to less developed places but
21、require local resources and originality(Bardhan1971;Bruton1998).Thus,thereisnotenoughdiffusionordelocalization,nottoomuch.As a result, less developed countries and regions have insufficient opportunities togetintotheworldeconomy,andconvergencethroughtheglobalizationofproductionfacessignificant barri
22、ers (in addition to those from imperfections in capital markets anddifferences in institutions) (cf. Helpman 2004). Notice that this critique comes from adynamic view, rather than the optimization problem derived from standard trade andlocation theory.Buttherearesomedynamicperspectivesonlocalization
23、-globalizationthathaveflawsaswell.Themostprominentofthemisanideathathasinfluencedmuchregionalandnationaldevelopment policy. It consists in claiming that the more the supply chain is localized, themoredevelopmentalbenefitsforalocalitycanbecapturedovertimethroughexpansionofthe activity. This idea expr
24、esses a fundamental bias against long-distance linkages andcommodity chains that are highly fragmented over different territorial jurisdictions. Insimple accounting terms, it has some short-term empirical validity, in that the morelocalized the value chain, the more of it will be captured locally fo
25、r a given increment ofincreaseinoutputoftheactivityinquestion.Thiswayofthinkingisfundamentallyflawedwithrespecttoeconomicdynamics,however(PugaandVenables1999).Thereisreciproc-ityintrade,soifeveryoneadoptspoliciestocapturethesechains,ultimatelythemechanismof comparative advantage will be blocked. Pro
26、tecting local content also frequentlygeneratespoliticaleconomyproblems,thatis,signalingthroughsuchpoliciestolocalfirmsthat they can be lazy about quality and innovation.1The dynamic process of economicdevelopmenthaslesstodowithhowmuchofaninput-outputchainisconcentratedlocallythanwithwhethertheongoin
27、grecompositionoflocaloutput(throughsectoralandactivity“succession” and intrasectoral innovation, improvements in quality, and vertical differen-tiation), as certain activities are fragmented and relocated to other places, is sufficient toincrease local productivity levels enough to permit increasing
28、 local real relative factorprices over the long run.2In other words, long-term growth depends on local learning,innovation, and adjustment, not on the mercantilist “capture” of supply chains. Butdevelopmenteconomics,ontheonehand,andregionaldevelopmentstudiesindevelopingcountries, on the other hand,
29、show that there is nothing automatic about this process.Themereexistenceofthewhipoflong-distancecompetitiondoesnotcallitforth.Itdependsonintricate socioeconomic processes at the local level and appropriate broader institutionalenablingconditions(suchasmarketrulesandpropertyrightsbutalsocertainnonsta
30、ndardinstitutional innovations; cf. Rodrik 2007).All in all, then, there remains considerable confusion about how we should conceptu-alize the relationship of regional economic development to the globalization process:1This is the “whip of competition” point made earlier (Wolf 2004). However, some o
31、f the research on EastAsias successful economies suggests that this is not always the case, since it depends on the politicalgovernance of firms and markets (Wade 1990; Amsden 2001; Rodrik 2007).2That is, if the region were a sovereign currency area, its real equilibrium exchange rate would apprecia
32、teover the long run.Vol. 85 No. 1 20093REGIONALCONTEXTANDGLOBALTRADEshould it be static or dynamic, and if the latter, then what kind of dynamic theory? In thisarticle, I address the relationship of regional development to globalizationcentering onthe processes of geographic concentration, fragmenta
33、tion, and tradein two principalways. First, I show that the standard approach to gains to trade should be complementedby a dynamic, evolutionary view that is rooted in growth theory. To do so, however, thegeographic dimension of endogenous growth theory needs to be sorted out, which is whatI attempt
34、 to do in the second section.The second principal task is then to delve back downinto the regional level and to ask which types of fragmentation, delocalization, and tradeare likely to generate dynamic welfare gains for the world economy. In this vein, thesecond main argument is that the effects of
35、decisions to fragment and relocate productionare not just manifest in outputs and productivity levelsthey also involve the creation,loss, and change of contexts.A theory of context is proposed in the third section. Contextsaffect future development potential as well as current performance. Decisions
36、 about theorganization and geography of production may therefore have unintended long-termwelfare effects.This is a particularly important contemporary question if globalization isbringing about the replacement of traditional regional contexts by territorially distributed“global” contexts for econom
37、ic action.Why We Need Geographic Fragmentation and Trade:There Are No “MAR Externalities”As I noted earlier, the case for trade rests on gains in productivity from comparativeadvantage and dynamic effects of competition that are due to greater openness. There isanother way to see the positive effect
38、s of specialization and trade, however. It comes fromgrowth theory, rather than from location theory, and centers on the geography of positiveexternalities and increasing returns. The Romer endogenous growth model establisheseconomy-wide increasing returns as the principal source of long-run economi
39、c growthunder resource constraints (Romer 1990). Knowledge and technology are nonrival andgenerally only weakly excludable over time; hence, they can be infinitely reused withoutloss and tend to spread their effects among communities of users (industries, for example)and geographic areas.These chara
40、cteristics of knowledge liberate it from the constraint ofdiminishing returns. Knowledge can therefore become a source of limitless growth intotal factor productivity through changes in techniques. This point is reconciled withstandard theory because the different specific activities to which innova
41、tions are appliedtend to become perfectly competitive in the long run (Romer 1986, 1990, 1994).The linkbetweenthesetwoseeminglycontradictoryprocessesisthatmonopolyrentstoinnovationare bid away through diffusion and entry into each sector, but at the economy-wide level,the recombination and reuse of
42、technology (across firms, sectors, and users) createsincreasing returns.Applied to the geography of economic growth, the Romer theory is frequently alliedwith earlier contributions by Marshall (1920) and Arrow (1962), respectively, abouttechnology spillovers at the regional scale (“the secrets of in
43、dustry are in the air”) and“learning by doing.” Both capture key mechanisms of increasing returns: those of reuse,spillover, and improvement.3Many such Marshall-Arrow processes occur at definiteterritorial scales (local, regional, national innovation systems, and spillovers) (Acs 1997,3Since these f
44、ounding ideas, the literature on the socioinstitutional sources of regional positive externalitieshas expanded to include the advantages of “thick labor markets,” local labor pooling, local rules andconventions, trust and social capital in economic transactions and technological spillovers, and regi
45、onalsystems of innovation.ECONOMIC GEOGRAPHY42002; Feldman 1994).There is a lively ongoing debate over whether such externalities arelocated principally within sectors, owing to specialization, or between different sectors incognate or complementary knowledge fields, with the latter said to give the
46、 edge todiversity as a key to dynamic regional development and the former to high levels ofspecialization (Duranton and Puga 2002; Jacobs 1969; Glaeser, Kallal, Scheinkman, andShleifer 1992). Notwithstanding these differences, there is widespread adherence to thenotion that “Marshall-Arrow-Romer (MA
47、R) externalities” at the regional scale are aprincipal source of growth-enhancing increasing returns in the economy as a whole.4Yet there is a key contradiction between this claim and a central point of the Romertheory, which is that increasing returns can be reconciled with allocative efficiencythr
48、ough perfect competition. If the only source of increasing returns were technologies towhich access was restricted to a (regional) community of producers, then these insiderswould, in effect, price their outputs to capture all of the returns to their monopolizedinnovations, in the form of localized
49、technology or knowledge rents. We know that suchrents exist when knowledge is sufficiently complex or uncodified that some kind ofproximity (geographic, member of a network, often both) is necessary to gain access tothe know-how required to use it properly.That such rents allow regional wages and profitsto rise above the economy-wide norm, at least for a time, and hence contribute to (at leasttemporary) successes in regional development is widely admitted among scholars ofregional development and development economics.5But if this were the end of the story, then there would be n
