Debates and Surveys

ion in much of recent geographical writing that simply be bad concepts. Or, they may be addressed to diVerent audiences, meaning something distinctly particularly shortchanges the role of the state in shaping territorial development. LOVERING, 1998, in a comdiVerent in each forum. The fuzziness in this case could be accidental or it could be deliberate. Political pelling critique, demonstrates how the recent economic history of Wales has been rewritten with the help of organizers often look for umbrella concepts that can pull strange bedfellows together ± s̀ustainability’ might what he calls `woolly theory’ , excising the extremely important role of public sector spending and presuming be an example. Or, someone wishing to obscure a 872 Debates and Surveys hegemonic or power relationship might choose to use niques of deconstructionism, constructionism, ethnomethodology and narrative from the ® ne arts and a rhetoric of inclusion ± `market’ is a case in point, and c̀ommunity’ is a word which is sometimes used in philosophy. These are often provocative ± take, for instance, Schoenberger’s fascinating accounts of this fashion (LYNCH and MARKUSEN, 1994). Such fuzziness may be useful for someone writing from a Lockheed and Xerox’s spatial strategies in her new book, The Cultural Crisis of the Firm (1997). However, dissident or relatively powerless position as well. In Dunlap’s analysis of the writing diYculties of MIT and authors of qualitative accounts often fail to make the case that the particular case study is representative or Berkeley graduate students from developing countries, she found that fear of political reprisal or rejection that the ® ndings from it are generalizable. Schoenberger’s two firm studies, for instance, suggest diametrically often led to obfuscation and elliptical conceptualization (DUNLAP, 1990). diVerent outcomes from spatial separation of ® rm operations ± functional in the case of Lockheed, dysIt is not my intention here to engage in a debate on epistemology and the sociology of knowledge. The functional in the case of Xerox. Her conclusion that Lockheed needed to segregate its missiles and space issues touched on tangentially above could be more satisfactorily debated with tools from social theory and division spatially from its aircraft division to succeed prompts one to wonder why another successful partiphilosophy. My point is to show the links between fuzzy concepts and theory, the wane of concern for cipant in both markets, Boeing, apparently did not. Of course, statistical studies which claim representativeness evidence and an increasing distance between intellectual work and policy. are often also riddled with ambiguities of various kinds, but these are generally more transparent and thus more likely to prompt written comment and The slippage in standards of evidence and method critique. Often the quality of case studies is hampered by Perhaps one reason why regional scholars have found it easier to promulgate fuzzy concepts is that the narrowness or b̀linkers’ imposed on the subject by the researcher or by unwarranted selectivity in choice of literature has become increasingly permissive about the quality of, and the necessity to, include evidence in data or interviewees. Scott’s early work on Orange County, for instance, focused on showcasing entreprenpublished research. The term t̀heorist’ is often applied to those who deal mainly in abstractions and abjure eurial electronics firms missed entirely the strong connections between them and the nearby Los Angeles empirical veri® cation, rather than to those who take up knotty problems, hypothesize about their nature military industrial complex and their continued deep dependence on government military contracts and causality, and marshal evidence in support of their views. It is common to hear scholars refer to a divide (SCOTT, 1986; MALECKI , 1987, MARKUSEN et al., 1991). Schoenberger’s powerful account of the struggle between the quantitative people and the theorists, as if those who use data for evidence have no theory and within Lockheed between the aircraft and the missile programme managers omits the parallel and antecedent those who d̀o’ theory have no use for evidence. Evidence, broadly speaking, is fundamental to the struggle within the ® rm’s dominant client, the Air Force, between the pro-pilot, pro-bomber advocates credibility of contending theories of regional development, and yet our standards of evidence seem to have and those who saw unmanned ICBMs as the superior nuclear delivery platform (MARKUSEN et al., 1991). slipped dramatically in the past decade. Many published pieces would bene® t from a map or a table of summary Case studies are diYcult for others to replicate, a timehonoured method of checking the quality of evidence, statistics setting the topic in context. Granted, researchers must contend with inadequate secondary data simply because they are so time-intensive and require on-site interviews. Below, I use the Silicon Valley sources and time and ® nancial constraints on datagenerating original research. But excellent quantitative case, where my colleagues have done ® eld research complementary to Saxenian’s, to illustrate some of accounts are eminently possible in many crucial regional development and policy areas ± I showcase these issues in case study research. The connections between the proliferation of fuzzy several below. Furthermore, new quantitative regional comparative techniques, such as Isserman, Beaumont concepts and devaluation of evidence in regional development scholarship are several. Ill-de® ned conand Rephann’s quasi-experimental methods (ISSERMAN and BEAUMONT, 1989; ISSERMAN and cepts are simply more diYcult to demonstrate empirically. Alternatively, fuzziness could be attractive to some REPHANN , 1995), are breaking new ground. Qualitative techniques have also advanced signi® cantly in writers simply because it excuses them from doing the hard work of documentation and to others because it regional studies, as judiciously chosen and interpreted case studies enrich our understanding of regional enables them to avoid politically diYcult issues, the naming of names or pointing the ® nger. development. Contemporary regionalists are now much more freAlthough I place a great deal of emphasis on empirical veri® cation in what follows, I am not arguing that quently writing narratives or case studies of particular regional development phenomena, borrowing techall concepts must be operationalized to be useful. New Debates and Surveys 873 and valid concepts may be diYcult to operationalize Fordism, that it is characterized by the externalization of transactions and thus the disintegration of large ® rms because we lack appropriate data or evidence. The concept itself may provoke the generation of new data and the rise of smaller, less specialized and locally networked ® rms, and that it creates a tendency toward as, for instance, the notions of aggregate demand and gross national product prompted the creation of spatial re-agglomeration, counteracting the tendency towards dispersion noted in the product/pro® t cycle national income accounting systems in the 1930s, or the big bang theory of the creation of the universe literature. It is mostly strongly articulated by STORPER and SCOTT, 1988: evoked new empirical techniques which eventually supported its protagonists’ remarkable contentions. Indeed, it now seems that a new, hegemonic model of I stress the value of subjecting new concepts to industrialization, urbanization and regional development empirical tests, by which I mean simply evidence of one has been making its historical appearance in the US and sort or another, not necessarily quantitative, because it Western Europe . . . (p. 21) is often the best way of revealing inadequacies in Because of the tendency to externalization of the transconceptualization. Even designing such research withactional structures of production, selected sets of proout carrying it out may have a salutary eVect. Having ducers with especially dense interlinkages have a tendency to commit oneself to stating where and when a concept to agglomerate locationally . . . (p. 26) applies and where it does not is often the easiest way Accordingly, the turn towards ̄ exibility has been marked by a to pare fuzzy concepts down to a sharp and clear decisive re-agglomeration of production and the resurgence of pro® le. It may be suYcient to state where and how the phenomenon of the industrial district. . . . (emphasis in other researchers and readers could observe the pheoriginal, p. 27) nomenon if and when we have adequate evidence. The ̄ exible specialization concept has other important facets, particularly regarding production relations inside Policy distance the ® rm and between ® rms. But for regionalists, these spatial contentions have been very in ̄ uential. Under-researched fuzzy concepts are more tolerable the less we expect them to guide action, whether by A rich and engaging literature debates various aspects of the ̄ exible specialization connotation and its the state or groups in civil society. Overall, I would argue, there has been a retreat from engagement with regional embodiment (see, for example, GERTLER, 1988; SAYER, 1989; AMIN and ROBINS, 1990; PARK regional policy since the late 1980s, either at the national or subnational level, on the part of regional and MARKUSEN, 1994). In general, much of this discussion is about the very notion of ̄ exible specializascholars. The relative neglect of the state’s role in regional development exacerbates the distance between tion ± what actually constitutes it and what causes it. Less attention has been paid to the actual empirical theory, research and policy. Uneven regional development continues apace, and state policies contribute to record, on either the pervasiveness of the phenomenon per se or its imputed spatial implications. The relative it in some cases, ameliorate it in others, but current policies (and policy lapses) are less often the focus of dearth of empirical evidence beyond case studies and anecdotes is due in part to the fuzziness of the concept analysis than they were in the 1970s and early 1980s. In what follows, I argue that regional research should itself. Nevertheless, several authors have tested various aspects of the theory. Remarkably, their ® ndings, which be policy-relevant and concerned not only with eYciency but also with normative goals loosely considered cast considerable doubt on the more universalistic claims of the school, have been met with surprising p̀rogressive’: equity; democracy; human rights; environmentally benign development. These are my own indiVerence. In what ways is the concept `̄ exible specialization’ prejudices ± others may favour a diVerent mix. Good science, indeed, can be put to reactionary purposes, as fuzzy? It refers to a multi-faceted process possessing a speci® ed set of features, including shop ̄ oor organizain weapons of mass destruction or celebration of markets in the face of abject poverty. But whatever tion of production, skills and career ladders of employees, aspects of inter® rm transactions and the normative posture, it should be stated. In general, normative statements appear to be increasingly suprelationships, corporate structure, governance mechanisms and regulatory regimes. The process is hypothepressed or absent in recent regional writing, a symptom of policy distance. sized to have both demand and supply side determinants. On the demand side, consumers are hypothesized to be demanding greater variability and Flexible specialization and agglomeration customization in products, while on the supply side, new technologies are supposed to permit greater ̄ exAmong the more highly acclaimed new concepts in economic geography is the notion of ̄ exible specializaibility, smaller production runs and thus lower economies of scale. Few accounts in this literature successfully tion with its associated spatial tendency toward reagglomeration. The contention in its purest form is `unpack’ these various dimensions, and thus it is hard to ascertain just which establishments, industries and that a new process, ̄ exible specialization, is replacing 874 Debates and Surveys ® rms might best be considered `̄ exibly specialized’. and where found in supplier industries, it is associated with greater ineYciency. Luria concludes that union To some, it is production processes themselves which are seen as increasingly specialized and ̄ exible while, avoidance is the most pervasive motivation for outsourcing in the auto industry, more important than the to others, the notion applies to ® rms, while others conceptualize it as applying to workers and yet others externalization of transactions costs. Luria’s ® ndings predict that dispersion rather than re-agglomeration to regions as a whole. It means, therefore, diVerent things to diVerent researchers (HARRISON , 1994, pp. will accompany vertical disintegration. A second important account, especially for 129± 30, distinguishes three diVerent meanings for `̄ exibility’ alone). regionalists, is the work of FELDMAN, 1993, and FELDMAN and MCINTYRE, 1994, on the spatial For the most part, empirical work on the `̄ ex spec’ thesis, especially in the US, has been con® ned to dimensions of ̄ exible production, particularly on the vertical disintegration and re-agglomeration hypoparticular localities and/or to particular industries. An outstanding example is STORPER and CHRISTOtheses. They unbundle the notion of ̄ exible specialization, not a `monolithic concept’, and contrast it to PHERSON ’s, 1987, study of the motion picture industry, which in many ways bears out the ̄ exible global Fordism and its spatial tendencies (1994, pp. 13, 42). In an aggregate analysis of American metropolitan specialization thesis. Storper and Christopherson, without making the case for why this industry might be areas from 1976 to 1984, FELDMAN , 1993, found that àgglomeration in U.S. manufacturing decreased, and the considered a good prototype, conclude that: decrease was the greatest in industries that became The vertical disintegration that lies behind ̄ exible specialmore vertically disintegrated and those with growing ization creates powerful agglomeration tendencies at the employment’, strongly contradicting the ̄ exible speregional level. Flexible specialization itself leads to the cialization thesis. recomposition of the industrial complex, through a new In supplemental nationwide industry case studies, form of horizontal integration of production capacity, carefully selected to capture the range of features associfurther strengthening external economies (p. 115) ated with flexible specialization (an increase in smaller ® rms, a tendency toward vertical disintegration, highBut the movie industry is relatively unique in that it produces a custom product in a one-of-a-kind productech in nature and relatively competitive), FELDMAN and MCINTYRE, 1994, found that none of the industion process. It may thus be an outlier in the contemporary economy. Furthermore, Storper and tries unambiguously exhibits the spatial tendencies posited in ̄ exible specialization theory. The aircraft Christopherson’s own data (Tables 4 and 5, pp. 110± 11) show at best a weak and ambiguous tendency toward industry, for instance, was found to have increased its employment dramatically over the period while average agglomeration on the part of both employment and establishments. Interestingly, this case belies the strong establishment size declined, as did the proportion of all ® rms serving national markets; nevertheless the induscontention of STORPER and SCOTT , 1988, that reagglomeration is most likely to take place in virgin try has not re-agglomerated either in its original Connecticut Valley locale nor anywhere else. Instead, the locations, since Storper and Christopherson contend that it is in Los Angeles that the industry’s reindustry has experienced a slow process of spatial diVusion, motivated by both lower labour costs and agglomeration is most apparent. Two accounts are notable for their eVorts to test the union avoidance, on the one hand, and by the desire to penetrate international markets (via oVset agreevarious causal links embedded in the `̄ ex spec’ literature. LURIA, 1990, marshals evidence for the American ments) on the other, outcomes which support the predictions of pro® t cycle theory (MARKUSEN , 1985). economy as a whole and for the auto industry. On the demand side, examining the contention that products Feldman and McIntyre conclude that these case studies yield greater support for the global Fordist characterizahave become more highly diVerentiated over time, Luria ® nds that this may be true for apparel but not tions than for the ̄ exible specialization thesis. Recent work on Seattle, Colorado Springs, Silicon for a wide range of other industries, including food, chemicals, oil, steel, major gas and electrical appliances Valley and a group of fast growing cities in Brazil, Korea and Japan (MARKUSEN , 1996c; MARKUSEN et and computers. Although production runs are smaller in the auto industry, he ® nds that real product diVerenal., 1999) suggests that ̄ exible specialization is not pervasive, even among this set of relatively new, `untaintiation has actually declined. Luria oVers an alternative demand side theory of growing product diVerentiation ted’ and highly competitive locales. At least four diVerent spatial con® gurations are possible, only one of ± that it is caused by a worsening income distribution ± and oVers plausible evidence in support of it. which is related to ̄ exible specialization, and the latter is the least common of the four. Furthermore, nonOn the supply side, Luria ® nds that the number of single-focus plants is actually on the rise in the auto local embeddedness ± strong links between establishments and firms at a distance from one another ± industry and that these plants are more eYcient than `̄ exible’, or multi-focus plants. He ® nds, furthermore, appears to be more important than local networks in many of the cities studied. Research on other that batch production is not on the rise in auto assembly, Debates and Surveys 875 agglomerations also casts doubt on the posited corresas formulated by Sassen, is that t̀he territorial dispersion of economic activity associated with globalization crepondence between vertical disintegration and reagglomeration. In an intensive study of northeastern ates a need for expanded (my emphasis) central control and management if this dispersal is to occur along with Ohio, CLARKE and HOLLY, 1996, conclude: `Thus, vertical disintegration as a characterization of produccontinued economic concentration’ (SASSEN , 1994, p. 24). tion cannot be said to dominate this territorial complex of ® rms in high technology sectors as it may in other `World city’ is at least a thrice-fuzzy concept. It can mean diVerent things to diVerent observers and authors regions’ (p. 138). The doubt that these studies cast on ̄ exible special(see also GOTTMAN , 1989). First, it may refer to the concentration of key roles in international transactions ization and its hypothesized agglomerative tendencies leads one to wonder why this body of work has been within certain cities, a notion that implies domination or at least leadership at the international scale. Second, so in ̄ uential, widely read and cited. Apart from its original, Third Italy version, associated with progressive it may refer to an external orientation on the part of cities, in that a greater share of its economic activity is politics in its native Italian context and from progressive eVorts (including by Luria) to apply it to mature directed to international rather than domestic markets. Third, it may refer to the ranking of a city in the global industrial regions like Detroit in the US, the odd thing about the American `̄ ex spec’ is that it has largely hierarchy of cities, encompassing cities which dominate large national economies and are the major entrepoà ts for been applied, misleadingly I would argue, to new industrial regions like Orange County, Los Angeles and interaction with external economies. Other conceptions are possible ± operating as a gateway for immigSilicon Valley. Its policy content is discouraging ± much of what is celebrated in Scott’s `new industrial spaces’ rants, for instance. Some usages combine one or more of these, but rarely all ± take, for example, the SASSEN , implicitly treats mature industrial regions with heavily unionized workforces and good income distributions 1994, glossary de® nition of global cities: `cities that are strategic sites in the global economy because of their as if they are past history. Interestingly, the more highly pro® led American `̄ ex spec’ districts are all highly concentration of command functions and high-level producer-service ® rms oriented to world markets; more dependent on military R&D and procurement spending, heavily white and male at the top of their generally, cities with high levels of internationalization in their economy and in their broader social structure’ occupational hierarchies, characterized by dualized labour forces and staunchly non-union (Hollywood is (p. 154). As a result of this fuzziness, the labelling of cities as an exception in this latter regard). Concentration on the more successful regions also diverts attention away `world’ will vary depending upon the connotation intended. For instance, if it is a city’s key role in from heightened interregional competition and devolution of responsibility for economic development and international transactions, broadly construed, that matters, surely Washington DC counts as a world city, social welfare from national to subnational governments in ways that seem implicitly to accept if not applaud since it is the military hub of the international security system and the site of the World Bank, IMF and these trends (MARKUSEN et al., 1999). Federal Reserve System headquarters. Geneva might also be included in the group under such a de® nition, World cities: in a class of their own? as might Silicon Valley as a nexus of world cybernetic innovation. If it is the external orientation of a city’s The phrase `world city’ was ® rst coined by Patrick Geddes in 1915 and resuscitated by Peter Hall (1966) economy that matters, cities like Hong Kong and Singapore would ® t neatly into a club with New York. as a device to set apart a select number of cities for special status in regional studies. The triumvirate of If it is dominance of a large national economy and control over its relations with the international comNew York, London and Tokyo is grouped together and studied most frequently as members of this club. munity, cities such as Beijing, Mexico City and Moscow would surely belong. If it refers to an unusual Gaining currency in the 1980s (e.g. FRIEDMANN and WOLFF, 1982), a number of intensive studies were incidence of international residents, San Antonio, San Francisco, Miami and Toronto would qualify. There done on such cities, also called global cities, prominent among them the work of SASSEN , 1989, 1991, 1994, may be yet other connotations as well ± certain cities might be considered world cities because they play and FAINSTEIN , 1990. The work continues. In the early 1990s, the Social Science Research Council unique roles for far̄ ung constituents and/or are revered globally ± Jerusalem, Rome, Mecca are funded a major study of New York, Tokyo and London, the papers for which will be published this coming examples. Since the de® nition is fuzzy, attempts to operayear. Such comparative analyses of world cities stress the commonalties among them and tend to downplay tionalize `world cities’ often result in the marshalling of disparate kinds of data and evidence, much of which the roles that national systems of cities and the condition of host national economies play in their success. illuminates only one facet. If the point is to show that a city plays key roles in organizing the international The general theory underlying the rise of global cities, 876 Debates and Surveys economy, various indicators from the percentage of the Others show that New York City’s manufacturing output is twice as apt to be exported as consumed world’s stock market transactions to the number of new patents to the size of the military forces comnationally, and exports shipped through New York area ports continue to grow rapidly (NEW YORK STATE, manded would be appropriate. If the exercise is aimed at demonstrating externalization, shares of city output 1990, p. 16; PORT AUTHORITY OF NEW YORK AND NEW JERSEY, 1991, p. 23.) Yet others note that foreign or services traded internationally will suYce. If status is accorded on the basis of command over, and repreownership of business and assets has increased dramatically (TELESIS, 1989, p. 9). A second type of evidence sentation of, a considerable hinterland, absolute size and presence of primate city functions would be relevant. is aimed at demonstrating that New York is more specialized in, and hosts absolutely larger shares of, Politically, the stylization of a city as a `world city’ is important because it often bears with it prescriptions certain producer services than do other American cities (DRENNAN , 1992, p. 229, Table 11; SASSEN , 1991). for economic development that bear strongly on the prospects for its residents. Claims that a city belongs Such evidence supports the ® rst two connotations of world city ± that New York plays a key leadership in the world city ranks might be used to support commitments of regional public expenditures and infrarole in at least some sectors, particularly ® nance, and that many of its activities are more internationally structure in ways that would enhance that status, with diVerential consequences for diVerent groups of residoriented than those of other American cities. The primacy of New York within the American hierarchy ents. World city claims may also be used within national debates over infrastructure and revenue-sharing to jusof cities and its pre-eminence as an American entrepoà t is generally not addressed in these accounts, probably tify skewed distributions, with adverse impacts on other regions. because they would be diYcult to support. Evidence supporting an alternative interpretation of Fuzzy thinking and inadequate data with problematic political results can be demonstrated by re ̄ ecting on New York ± as one of a half dozen major specialized centres in a nation with a multi-polar and progressively the case of New York City. Recent scholars of New York have chronicled the `unhinging’ of New York ̄ attening urban hierarchy ± casts doubt on the validity of this third connotation of world city, at least as it from its domestic hinterland and located its future and dynamism in its increasingly international orientation. applies to New York. New York is surprisingly underrepresented in many producer services categories, and DRENNAN, 1987, considers New York to be t̀he major commercial and ® nancial city in the world’s its position in many others has eroded over time vis-a Á vis cities like Boston, Washington, Chicago, Los largest economy’, arguing that while it plays a smaller role in producing, shipping and selling goods, it plays Angeles and San Francisco (BEYERS, 1989; MARKUSEN and GWIASDA, 1993). New York diVers a larger one in the world economy, particularly in ® nance and other business services. Although analysts from London and Tokyo in having to compete within its national space with at least these ® ve other cities, like Sassen, Fainstein, Mollenkopf and Castells deplore various developments associated with this externaleach of which serves as a `capital’ ± educational, political, industrial, high tech, military industrial ± in its ization, especially dualization and poverty, they are as a group quite sanguine about New York’s continued own right. Two points can be made about the consequences of pre-eminence. Its future is widely considered to be ensured by its strength in producer services, both this competition. First, New York’s apparent specialization in services is in part a function of its relatively ® nancial and non-® nancial, and by its heavy international orientation (MOLLENKOPF , 1985, p. 251; poor performance in manufacturing in recent decades. In 1960, New York region led the nation in total NOYELLE and PEACE, 1988; HALL, 1990; and DRENNAN , 1991, p. 20.) As one study puts it: `Especially as numbers of jobs and mirrored the nation’s economic structure ± New York’s manufacturing share of all its importance within the national economy recedes, the scope and character of New York’s economy will increasingly be jobs exactly matched the national average of 31%. Subsequently, New York’s manufacturing jobs evaporde® ned by the role it plays in the emerging global marketplace’ (O’NEILL and MOSS, 1991, p. 9, emphasis in original). ated at a more rapid pace than the nation’s. Measures of concentration such as location quotients which show New York is paired with Tokyo, Paris and London by most of these authors (DRENNAN, 1987, pp. 26, 28; New York increasingly specialized in services are thus re ̄ ecting the absence of manufacturing as much as the VOGEL, 1988; FAINSTEIN , 1990, pp. 120, 125; HALL , 1990, p. 13; SASSEN, 1991). presence of services. Furthermore, New York has not added service jobs as fast as have its competitors with Two types of evidence are oVered for the claim that New York is a world city, implicitly the American faster rates of growth in manufacturing. The robust manufacturing growth regions of Los Angeles, San world city. The ® rst type consists of evidence that New York’s activities are more externally oriented than other Francisco and Washington, DC, all experienced producer service job growth rates between 80% and 90%, American cities and are increasingly so over time (DRENNAN, 1985, pp. 35-4; MOLLENKOPF , 1985; while New York’s service sector jobs grew by only 34% (BEYERS , 1989, pp. 193± 94). WARF, 1987, pp. 3± 12; SASSEN , 1991, chapter 4.) Debates and Surveys 877 Second, in quite a few large and important producer ® nancial services (FAINSTEIN , 1990, pp. 129, 132). New York’s relatively narrow specialization renders it service sectors, several other American cities were more specialized and increased their specialization over time vulnerable to slowdowns in the pace of globalization, such as that which began in 1998, and to reversals in at New York’s expense (MARKUSEN and GWIASDA , 1993). Although New York still leads in securities, the ® nancial sector, as in the aftermath of 1987 and again as the century draws to a close. A more diversi® ed banking, manufacturing administration, accounting and advertising, other American cities are more specialized, economic development policy should be favoured over one preoccupied with ® nance and international relaand growing faster, in other business services such as legal services, computing and data processing, managetions. It should, as New York has belatedly been doing in the 1990s, seek to stabilize and retain manufacturing, ment and public relations, research and development, and membership organizations. Evidence is particularly and it should target not just export markets but the large domestic market in its backyard. strong for the hypothesis that the capital city role attracts a disproportionally large contingent of producer There is a deeper, more troubling policy issue associated with grandiose interpretations of New York’s services ± Washington, DC, is more specialized than New York in a plurality of signi® cant producer service status and dynamism. If New York is seen principally as having a stake in the success of economic activities functions. A contemporary world city could be more modestly facilitating accelerated globalization, this suggests that national policies such as free trade, international ® nanand unambiguously de® ned as one which is successfully competing for dominant status in at least one of the cial mediation, a continued role for the US as global policeman, in ̄ ation control and de® cit reduction are several important functions (® nance, international market regulation, information management) of integin New York’s interests. From this vantage point, New York is assumed to have interests at odds with other rating the transnational capitalist economy in a neomercantilist world in which gains from trade and American regions which may have a greater stake in managed trade, a slower rate of international economic ® nance dominate over gains from production. In our own work, New York meets this de® nition because of integration, infrastructure investments and a Keynesian full employment policy. Yet the New York ® nancial its clear and relatively unchallenged dominance in banking and ® nance. It is the primary locus of the sector is heavily dependent on domestic savings and capital funnelled into the stock market and other ® nagents involved in these speci® c functions, as Sassen comprehensively shows. As pro® t has been increasingly ancial instruments ± an eroded domestic economy could rebound negatively on New York’s economic generated by the international movement, rather than creation, of capital, technology, commodities and sersuccess. And if the ® nancial and business services sectors are privileged in either theory or practice, at either the vices over the two past decades, the agents of merchant and ® nance capital, collectively known as `Wall Street’ regional or national levels, New York’s blue-collar ® rms and workers will continue to face displacement. in the US, have also become politically ascendant in national and international policy making, especially in the free trade and deregulation debates. Networking and co-operative competition in industrial districts But New York is not a world city in the fuzzier, more expansive sense employed in most accounts from A third example of fuzzy conceptualization is the emphasis on networks and co-operative competition in the 1990s. It does not rank on a par with Tokyo and London, because the latter are the undisputed ® nancial, geographers’ and planners’ work on industrial districts. Contrasting industrial districts to those dominated by cultural, political, educational, research and industrial capitals of their countries. New York can only claim large ® rms or whose ® rms operate in atomistic markets, researchers attribute the relative success of the Third to be the ® rst of these ± arguably, Los Angeles with its pre-eminence in TV and ® lm, has become the nation’s Italy, Baden WuÈ rttemberg and Silicon Valley to strong, reciprocal social networks among large numbers of (and the world’s) cultural capital. The political and policy implications of this more relatively small ® rms which facilitate innovation and collective management of crisis (e.g. COOKE and modest and empirically grounded interpretation of New York’s global role are sobering for New York MORGAN , 1994; SAXENIAN, 1994). Firms in this extensive literature engage in `co-operative competienthusiasts and constitute a critique of its globallyoriented economic development strategy. The domestic tion’ and are often characterized as having porous or f̀uzzy’ boundaries. market is still very important to New York, and yet New York is losing market share to other domestic as The emphasis on networking among ® rms in this literature is welcome and addresses what GRANOVwell as international regions. A decades-long indiVerence to manufacturing decline, in both scholarship ETTER, 1985, characterizes as the missing social relations operating at the level in between neo-classical and policy, was unwarranted. New York City mayoral regimes, from Lindsay onward, did little to stop the economics’ individual consumer and ® rm, on the one hand, and Marxist class and amalgamated capital on out ̄ ow of manufacturing and port-related jobs and much to encourage the rapid growth of business and the other. Nevertheless, in most regional accounts, 878 Debates and Surveys networks are presented generically and extolled without on ® rm interviews do not write up their methodology or address issues of interview selectivity, informant examining the motivations of participants, mapping who might be included and excluded, analysing veracity, interviewer neutrality and the generalizability of ® ndings, despite lively debate on these issues in the unequal power relationships among members or gauging the durability or fragility of relationships. A recent economic geography literature (SCHOENBERGER, 1991; HEALEY and RAWLINSON , 1993; similar critique can be levelled against the notion of social capital ± I have yet to read an account of it, MARKUSEN, 1994). Second, because industrial district research is con® ned within the borders of the district, including the pioneering PUTNAM , 1995; COLEMAN , 1988; and heavily mathematicized BECKER, 1996, ® rm and establishment, ties to other ® rms and organizations, particularly the state, outside the region are versions, which distinguishes whether it is a stock or a ̄ ow concept or how it is produced and accumulated inappropriately eliminated from the analysis. This may lead to the mistaken impression that a region’s eco(since it is an analog to physical capital). The companion notion of co-operative competition nomic dynamism is endogenously driven. A number of papers have been written purporting to ® nd netis a fuzzy and inherently non-operational concept because it connotes tension between two types of works and industrial districts where on closer examination, their existence is questionable. Other researchers behaviour. What types of behaviour, which actions, exactly, are competitive and which are co-operative? have documented the non-local embeddedness of establishments and ® rms in far̄ ung networks and the Co-operative behaviour may enhance the bottom line, but it may also result in one’s competitors stealing a dominance of large ® rms in regional production systems (STORPER and HARRISON , 1991; MARTIgood idea or pilfering key employees away. This calculus must be present in every act of interchange between NELLI and SCHOENBERGER, 1991; HARRISON, 1994; PARK and MARKUSEN , 1994). Third, students ® rms and employees within them. In the Silicon Valley case, for instance, two sets of researchers, sometimes of industrial networks often con ̄ ate their existence with regional economic success without acknowquoting the same ® rms, paint entirely diVerent pictures of inter-® rm dynamics. SAXENIAN, 1994, depicts ̄ ows ledging other forces at work in the regional economy. They do not carefully establish the links between of personnel and information from one ® rm to another in a friendly, unmonitored atmosphere, bene® ting all. individual informants, their organizational roles, the establishment, the ® rm, the industry and the region in FLORIDA and KENNEY, 1990, point to the high incidence of law suits as evidence that the boundaries making causal inferences (MARKUSEN , 1994). To illustrate the diYculties of establishing empirically of ® rms are vigilantly policed and that there are winners as well as losers from this process, as when one or the the role of networks and co-operative competition in industrial districts and the extent to which they account other co-operative partner moves more quickly on shared information. Both tendencies may be present, for regional prosperity, let us re ̄ ect on one of the very best accounts ± SAXENIAN ’s, 1994, study of Silicon but the concept of co-operative competition does not help us to distinguish between them and leaves planners Valley. Saxenian presents Silicon Valley as à regional network-based system that promotes collective learning with little to go on. Can ® rm boundaries be fuzzy, even in theory? and ̄ exible adjustment among specialist producers of a complex of related technologies’ (p. 2). She characterCapitalist ® rms, whether large or small, operate in a highly regulated environment in which property rights, izes activity in the Valley as an `unusual mix of cooperation and competition’ (p. 57); t̀he functional stock ownership, bank loans and other equity investments are very carefully watched and tended to. The boundaries within ® rms are porous in a network system, as are the boundaries between ® rms themselves boundaries of a ® rm are, ultimately, not at all fuzzy ± they are written down in asset, cost and revenue and between ® rms and local institutions such as trade associations and universities’ (p. 3). Saxenian’s is a rich statements that owners and managers, whether private or public, scrutinize carefully every quarter. If the regional portrait in which strategic relationships are local and face-to-face, so that the region and its culture bottom line is not positive for most quarters, or in the case of a start-up within a few years, ® rms will cease are central to industry performance. The success of the Silicon Valley district is counterposed to that of Route to exist. Firm survival rates may be low and personnel turnover high, but as organizations and institutions, 128 near Boston, which is depicted as verticallyintegrated, hierarchical, rigid, and too traditionally ® rms are clearly bounded. Because our theory is underdeveloped, the nature of family-oriented to compete with the western upstart. Unlike their eastern counterparts, Silicon Valley engin® rm interactions in various types of regions is largely an empirical question. Firms’ interactions do matter, eers and technicians are committed to one another and to the region, not to companies or even industries. and network analysis has enriched economistic regional studies. But at least three problems encumber empirical They are even, Saxenian suggests, less motivated by money than by the drive to innovate. work on industrial districts. First, there has been remarkably little discussion about what makes for a Regional Advantage is written as a narrative, with examples and evidence woven into the text. Saxenian g̀ood’ network study. Most regional researchers relying Debates and Surveys 879 oVers more documentation of her method than most role in the evolution of the Valley has been well told by SCHOENBERGER, 1997, and by Saxenian in her authors of regional case studies who rely on interviews. In an appendix, she states the following: earlier work (1985). Lockheed and the other defence contractors are signi® cant customers of the Valley’s The research for the book was ethnographic in nature, smaller ® rms, many of whom specialize in defence with the empirical material accumulated over the course electronics. They also have considerable non-local ties of nearly a decade living in and observing the two regional to Washington and to other defence contractors and economies. The core of the argument is built from more divisions of their own companies elsewhere ± than 160 in-depth interviews with entrepreneurs, industry Lockheed, for instance, is currently headquartered in leaders, corporate executives, and representatives of local Bethesda, Maryland, a recent shift from Los Angeles. business associations, governmental organizations and uniAlso missing from this account are the many branch versities in Silicon Valley and Route 128 (p. 209) operations of foreign, especially Asian ® rms, acting as èyes and ears’ for their home-based operations. and lists the ® rms and positions of the individuals interviewed. Because her method is more transparent, Third, it is not clear that Saxenian’s sample captures the experience of losers as well as winners and of it is paradoxically easier to raise questions about the comprehensiveness of the account. Several points may outmigrants as well as those committed to staying in the Valley. If the set of those interviewed was generated be made. First, the interviews in Regional Advantage appear to by a snowballing technique, where one ® rm mentions or recommends others, and/or by frequent mention in be heavily concentrated within a subset of industrial sectors in the Valley. The electronics and computing secondary press and other accounts, then ® rm success, membership in networks, degrees of co-operation and sectors appear to be over-represented, with fewer proportionally in the missiles and space, communications positive evaluation of such relationships are apt to be overstated. If the point is to showcase `best practice’, equipment and instruments sectors. The former two sectors account for about 2/3 or 120,000 jobs in the this may be appropriate, but Regional Advantage implies that this is prevalent Valley culture and experience. high-tech manufacturing sectors (Saxenian’s own data, p. 208). Some 60,000 additional jobs are concentrated Similarly, and perhaps this is asking too much, no mention is made of those ® rms and individuals who in the latter three highly defence-dependent sectors; indeed, the Valley is more heavily specialized in missiles have chosen to relocate outside of the Valley to places like Seattle and Colorado in what has been a marked and space activity than in electronics as measured by location quotients (GRAY et al., 1999). Furthermore, outmigration over the past decade and where they continue to compete successfully in these same sectors an undetermined share of the Valley’s computing and electronics innovative activity (as opposed to sales), despite leaving the agglomeration. Finally, the causal link implied between the coespecially leading edge research, is oriented towards and underwritten by military missions. For decades, operatively competitive culture described (but not measured) in the book and the success of Silicon Valley as the federal government has awarded more than US$5 billion annually in real terms in military and space a regional economy is not thoroughly established. A number of very important and relatively unique attribprime contracts to Silicon Valley, the nation’s premier defence electronics and avionics region. Furthermore, utes of the Valley are left out of the analysis. These include the Valley’s embeddedness in the larger San there are good reasons to believe that, with the increasing signi® cance of electronics, guidance, computing Francisco regional agglomeration, with strong ties to the downtown San Francisco business services sector and software in the cost of any single weapons platform, the Valley receives a greater than proportional share of and, again, the stimulative role of defence spending with its annual infusion of billions in contracts and subcontracts from contractors elsewhere. Second, not all important actors in the Valley are long term investment in human capital and technology. Saxenian’s study is an important, fascinating and included in the study. Although Saxenian does an excellent job analysing Hewlett-Packard, other large accurate interpretation of a subset of sectors, ® rms and individuals in Silicon Valley. Activities in the ® rms and ® rms and international ® rms do not receive their due (TEECE, 1992; HARRISON , 1994; GRAY et al., 1999). sectors may indeed be disproportionately responsible for Silicon Valley’s success. I would like to stress that Conspicuous for their absence, for instance, are the three largest defence contractors ± Lockheed, FMC her account is among the very best-researched and documented in this genre. My point is that even Corporation and Loral Western Division (formerly Ford Aerospace and now part of Lockheed/Martin) ± research questions best approached with qualitative methods such as interviewing require rigour and transas well as many medium-sized ones, including IDS, the successful defence electronics ® rm founded by parency in methodology. My team’s interviews with large American corporations, defence contractors and former Secretary of Defense, William Perry. Lockheed is the single largest industrial employer in the Valley ± foreign operations of Korean ® rms in the Valley suggest a quite guarded attitude toward other ® rms and an with 16,000, it is twice the size of Sun Microsystems, and the history of its location in the Valley and seminal absence of f̀uzzy boundaries’ . We also found substantial 880 Debates and Surveys linkages external to the region, both up and down clearly and causally. In my own view, the emphasis on process, while a welcome new dimension, has been stream, counterbalancing local connectedness and questioning the extent of endogeneity in the region’s overdone and tends to obfuscate agency and responsibility. If the focus is on processes of change over time, growth. In our view, it is not possible to conclude that the success of the Valley is due principally to locally the implication is that actors, even powerful ones, are caught up in such processes and have little control over embedded co-operative competition and not also to extraordinarily valuable federal defence contracts, conshaping their own or collective destinies. American ® nancial and industrial elites, however, are actively nections with other actors (such as Microsoft) outside the region, and the proximity and dynamism of the engaged in fashioning the new world order through their dominance in ® nancial, economic, trade and neighbouring San Francisco business services sector. Relative causality of each of these factors remains to foreign policy. I believe, however, that it is principally the recession of political movements and the dilemmas be demonstrated. A nuanced interpretation of Silicon Valley is politicof achieving equity and stability in a world dominated by American ® nancial and military elites, with their ally important. The Valley is increasingly viewed as an icon of success in the increasingly global economy. It attendant ideologies, that make the intellectual project so diYcult. has been widely studied by economic development planners in many countries and is often cited as an Even given this political reality, it is hard to understand why there is so little good empirical work beyond example of what regions can do to distinguish themselves in a world of heightened spatial competition and anecdotes or case studies. The latter are often published without re ̄ ection on choice, representation or generalof eroding regional policy. Accolades for Silicon Valley are common among those advocating decentralization izability. Empirical work can be laborious, and secondary data analysis is often more time-consuming and of responsibility for economic development from national governments to localities. To the extent that less enjoyable than interviewing. Most American case studies appear to have been done by researchers on Silicon Valley’s culture is credited with its success, then implicitly, as in the culture of poverty literature, other industries, ® rms and establishments in their own back yard. While understandable given resource constraints, regions’ relative poverty of culture must explain their relatively poor performance. such p̀roximity’ research may constitute a form of provincial boosterism, conscious or not ± most case Yet many localities in the US and elsewhere have attempted to incubate high-tech entrepreneurs or repstudies of southern California, Silicon Valley, and New York are suspect in this regard. The policy impact of licate the networks heralded in the Valley, often failing. Why? Two points are important. First, it is likely these studies, especially for regions elsewhere and at the national and international level, will be limited if that a new complex of industries ± electronics and computing in this case ± will take root in one or a few this is the case. Overall, the political implications of much of the centres in early stages and remain clustered as long as innovation continues at a rapid pace (MARKUSEN , literature discussed here are sobering. By and large, the studies pro® led in the three bodies of work oVer 1985). That complex will take root somewhere, but it does not follow that other regions can replicate the complimentary and optimistic accounts of the ability of managers and entrepreneurs on Wall Street, in highculture and expect to reap similar returns, because an industry may be able to support only a few such centres tech electronics and computing ® rms, in the movie and multi-media industries and in small businesses of innovation. Second, if continued large Pentagon commitments to technology-intensive weapons play an generally to guarantee the future growth and competitiveness of their host regions. Yet the places, industries ongoing role in the Valley’s success (military R&D spending has been constant in real terms since the end and enterprises celebrated include those with a low incidence of unionization (Hollywood, again, of the Cold War and has increased in the defence electronics sector in which the Valley specializes), then excepted), relatively polarized occupational structures, hectic and non-community oriented work lives, chie ̄ y would-be high-tech regions might design their strategies diVerently, targeting federal R&D money. Since white male hierarchies, and relatively reactionary organizational positions on issues such as universal defence R&D spending is slated for a real decline presently, it might be preferable to target other categorhealth care, social security, privatization of education, devolution and funding for community development ies, especially the growing National Institutes of Health budget. and other social programmes. Regional studies is in need of a soul-searching about the quality of its theorizing, the rigour of its research Conclusion methods and the policy and political implications of its work. All three are connected. Insulation from policy In regional studies, fuzzy concepts may have proliferated because it is more diYcult to see clearly what pressures invites fuzzier concepts. Fuzzy concepts make the job of coming up with evidence much more a progressive spatial strategy might be under capitalism and perhaps also because it is easier than thinking diYcult. Poverty of evidence results in toleration of Debates and Surveys 881 supposed to refer to the (plural) subjects of the sentence, fuzzy concepts and misguided policy. The work of and strictly speaking, tables and ® gures cannot be authors reviewed here is welcome in scale and ambiindicators. tion, counterpoised against the narrowly empiricist 5. See, for example, the acknowledgement of state strucexercises and overly formalistic models which have tures and actions in our analysis of the formation of dominated regional science. I hope that this summary American new industrial districts in MARKUSEN et al., of a selective set of highly regarded works will help 1991, and MARKUSEN , 1991. to prompt a lively conversation. Particularly in the 6. I am indebted to Nancy Ettlinger for raising this possibilpreparation of students who will be the future teachers ity with me. and researchers, we must ensure that we are working 7. Consider the extraordinary success of the concept `parawith powerful, plainly stated theories which can be digm’ as developed by Thomas Kuhn in his work on operationalized and which oVer clear guidance for scienti® c revolutions (1962). Although stimulating a great deal of debate and furthering our understanding of those with the responsibility to shape the future of fundamental shifts in scienti® c outlook, Kuhn himself cities and regions. We owe our practitioner community used the word p̀aradigm’ in at least 20 diVerent ways the same. (WEINBERG, 1998), inviting the progressive appropriation of the word for such minor shifts as to render ANN MARKUSEN the term virtually meaningless. Hubert H. Humphrey Institute of Public Affairs 8. I am grateful to Najwa Makhoul for pressing me to Humphrey Center, Minneapolis make these distinctions. 9. Roger Hayter suggests that the disdain for empirical work in the tradition of veri® cation in much recent Acknowledgements ± My thanks to Matt Drennan, Nancy geographical writing may be because evidence is seen Ettlinger, Susan Fainstein, Marsh Feldman, Bennett Harrison, simply as something convenient to support an established Roger Hayter, Charlie Hoch, Arnoud Lagendijk, David position. Lewis, Najwa Makhoul, Marla Nelson, Michael Neuman, 10. I am indebted to Roger Hayter for this point. Laura Powers, Annalee Saxenian, Michael Storper, Mike 11. I am grateful to Rachel Weber for this point. Teitz, Rachel Weber and two anonymous reviewers for 12. Despite Luria’s important ® ndings for autos, a signi® cant, feedback on earlier versions of this article. growing and income-elastic industry, little attention has been paid to them in the geographic literature. STORPER, 1997, for instance, seems not to have engaged Notes Luria on his work, citing only the pre-publication version of his paper, not the revised version published 1. I became interested in fuzzy concepts when I began in 1990. noticing that graduate students were increasingly mim13. Elsewhere in the literature, case studies seem to have icking what they read, particularly in the new geography been chosen without careful consideration of representaliterature, writing lengthy treatises with new fuzzy contiveness, and perhaps because the industry was at hand. cepts embedded in long sentences in the passive voice. 14. The following discussion draws upon MARKUSEN and Such writing alienates practioners and lay readers. GWIASDA, 1993. Furthermore, it is unnecessary. Paraphrasing Robert 15. Much of what Fainstein and Sassen focus on are the Musil in his novel The Man Without Qualities (1953), similarities among London, New York and, in Sassen’s any time there is something truly new and important to case, Tokyo, in the internal structure of these ® nancesay it can be said in a very simple and succinct way. dominated cities and their common tendencies towards Sociologist Howard Becker makes the case eloquently dualization and inner city redevelopment, not addressed in his Writing for the Social Sciences (1986), and historian here. However, their contention of dualization is quesPatty Nelson Limerick echoes it in her `Dancing with tioned by Drennan and others, who have shown in Professors’ in the New York Times Book Review (1993). cross-sectional studies of cities that those with higher Of course, my students, as did Becker’s, complain that producer services specializations also have relatively it may be all right for established professors to write better income distributions, show improvement in those simply but that making one’s name requires pretentious distributions over time, and have posted a decline in the presentation. I do not believe this to be true. Furthernumber of and size of populations in ghetto Census more, it is unfair to graduate students not to demand tracts ( JARGOWSKY, 1994; DRENNAN et al., 1996). rigour and clarity. FAINSTEIN , 1998, oVers a more nuanced view of con2. Sustainability, noted Michael Teitz at a recent ACSP tradictory evidence for dualization. discussion of this article, is the ultimate fuzzy concept. 16. SOJA , 1991, compares New York and Los Angeles. 3. WHATMORE, 1994, also calls for increased debate about While he characterizes them diVerently than do social agency, more detailed empirical work and political Markusen and Gwiasda, his work also demonstrates the relevance. signi® cance of within-country comparisons. 4. Consider for instance this summary sentence introducing 17. New York, in 1962, handled 61% of all cargo value a paragraph in COOKE and MORGAN, 1994, p. 96): Às coming in and out of the US, but by 1987, its share had an indicator of the pressure to respond to competition dropped to 40%. In contrast, the west coast ports by moving production into the supplier base rather increased their share from 13% to 46% in the same than producing in-house, Fig. 5.1 and Table 5.3 are instructive’. The (singular) noun in the ® rst phrase is period (CAMPBELL, 1992, p. 14). New York’s port 882 Debates and Surveys 20. The large and ongoing role of military demand and never recovered from containerization and jet travel; jobs on the docks and railroads fell from 100,000 in 1958 to R&D in the Valley is well-documented elsewhere (YUDKEN and SIMONS, 1989; MARKUSEN and 22,000 in 1983 (DRENNAN , 1985, p. 89) 18. I have not been able to identify the origins of the phrase YUDKEN , 1992; GRAY, 1998) and it belies the overwhelmingly civilian orientation conveyed by Saxenian, c̀o-operative competition’. HARRISON, 1994, pp. 88, 111) uses it in his chapters synthesizing the literature on who has written seminal accounts of the early role of defence spending but writes as if this is an historical and the Italian industrial districts and Silicon Valley. 19. SCOTT , 1988, for instance, identi® es Colorado Springs not contemporary phenomenon. 21. See, for instance, the debate between PORTER, 1996, as a ̄ exibly specialized industrial district, but work by MARKUSEN et al., 1991, and MARKUSEN and GRAY, and MARKUSEN , 1996a, 1996b. Saxenian’s policy discussion focuses on how the Valley might shore up 1999, ® nd no evidence of this before the mid 1990s and then only among new, mass-based Christian organizaits regional advantages by creating more co-ordinated governance structures, aÁ la the Third Italy. tions. In another example. W ILLOUGHBY, 1993a, 1993b, ® nds a networked biotechnology district on 22. For a fuller discussion of the economic development implications, see HARRISON , 1994. the East Coast, while GRAY, 1998, ® nds very little networking among biotech ® rms either there or else-