Global cities and multinational enterprise location … cities and multinational enterprise location strategy ... the global economy (Sassen, ... Global cities and multinational enterprise

Global cities and multinational enterprise

location strategy

Anthony Goerzen1,Christian Geisler Asmussen2

and Bo Bernhard Nielsen2,3

1Queen’s School of Business, Queen’s University,

Kingston, ON, Canada; 2Department of

Strategic Management and Globalization,Copenhagen Business School, Copenhagen,

Denmark; 3University of Technology, Sydney,

Australia.

Correspondence:A Goerzen, Queen’s School of Business,Queen’s University, Goodes Hall Room 339,Kingston, Ontario, Canada.email: [email protected]

Received: 25 November 2011Revised: 19 February 2013Accepted: 20 February 2013Online publication date: 18 April 2013

AbstractWe combine the concept of location derived by economic geographers withtheories of the multinational enterprise (MNE) and the liability of foreignness

developed by international business scholars, to examine the factors that

propel MNEs toward, or away from, “global cities”. We argue that threedistinctive characteristics of global cities – global interconnectedness, cosmo-

politanism, and abundance of advanced producer services – help MNEs over-

come the costs of doing business abroad, and we identify the contingenciesunder which these characteristics combine with firm attributes to exert their

strongest influence. Consistent with these arguments, our analysis of a large

sample of MNE location decisions using a multilevel multinomial model sug-

gests not only that MNEs have a strong propensity to locate within global cities,but also that these choices are associated with a nuanced interplay of firm- and

subsidiary-level factors, including investment motives, proprietary capabilities,

and business strategy. Our study provides important insights for internationalbusiness scholars by shedding new light on MNE location choices and also

contributes to our understanding of economic geography by examining the

heterogeneous strategies and capabilities of MNEs – the primary agents ofeconomic globalization – that shape the nature of global cities.

Journal of International Business Studies (2013) 44, 427–450. doi:10.1057/jibs.2013.11

Keywords: global cities; liability of foreignness; foreign direct investment; alliances andjoint ventures; location strategy; internationalization

INTRODUCTIONThe concept of location – where and why firms place specific acti-vities in particular areas – is a key area of interest in both inter-national business research (e.g., Alcacer & Chung, 2007; Nachum &Wymbs, 2005; Porter, 2001) and economic geography research(e.g., Krugman, 1991; Lorenzen & Mudambi, 2013; Markusen,1996). Yet, despite rising interest in location, our current under-standing of the geographic aspect of multinational enterprise(MNE) behavior remains underdeveloped (McCann, 2011; Ricart,Enright, Ghemawat, Hart, & Khanna, 2004).

Although scholars have begun to demonstrate how MNE strat-egies influence their foreign location decisions, there is little theoreticalor empirical work on the influence of firm characteristics on thespecific location choices of MNE subsidiaries (Beugelsdijk,McCann, & Mudambi, 2010; McCann & Mudambi, 2005). Morespecifically, the attraction of MNEs to “global cities” has beenlargely overlooked by international business scholars. By the same

Journal of International Business Studies (2013) 44, 427–450& 2013 Academy of International Business All rights reserved 0047-2506

www.jibs.net

token, while economic geographers have carefullystudied the concept of location, one strand of eco-nomic geography research has been preoccupiedwith the propensity of firms and workers to clusterin geographic space as regions become more inte-grated, whereas another has focused more specifi-cally on the emergence and evolution of globalcities (Beaverstock, 2002; Friedmann, 1986; Jacobs,1984). Much of this work focuses on the character-istics of economic agglomeration, however, withless attention to the behavior of multinationalfirms (see Beugelsdijk et al., 2010, for an exception),and the way in which their heterogeneous capabil-ities and strategies interact with location. There-fore, while prior research has examined the patternof linkages that tie global cities together in theworld economy, we still know relatively little aboutthe firms that actually create and exploit thoselinkages by establishing, for example, channels ofcommunication between dispersed subsidiaries,and by moving goods, people, and informationacross geographic space. Hence, as suggested byMarkusen (1996: 310), “more intensive study ofmultinational corporations [is necessary] y if amore powerful geographic contribution to progres-sive strategy is to emerge”. Our research, therefore,is designed to improve our understanding of theMNE in geographic space by bridging these diversestreams of thought.

The tradition in international business scholar-ship, in literatures such as those on geographicscope (Asmussen, 2009; Goerzen & Beamish, 2003),expatriates (Belderbos & Heijltjes, 2005), culture(Makino & Neupert, 2000), and entry mode(Tihanyi, Griffith, & Russell, 2005), to name a few,has been to use the nation-state – defined bynational political boundaries – as the basic unit ofanalysis when examining location. Yet, as arguedby previous authors, these large geographical unitsare often too coarse to provide an accurate pictureof the role of MNEs in economic globalization,since the trends of outsourcing and new technol-ogies, as well as economic policies of liberalization,have made nation-states less significant as units ofanalysis (Brown, Derudder, Parnreiter, Pelupessy,Taylor, & Witlox, 2010; Krugman, 1991). Ultimately,firms choose a specific place within a country as thelocation for their investments, and at the country oreven the regional level of analysis these micro-location decisions are obscured; this warrants a morenuanced examination of specific locational features.

More recent research on MNE location strategyhas begun, in fact, to consider the subnational

level, including industry clusters (Gordon &McCann, 2000; Porter, 2001; Pouder & St John,1996) and regions (Ma, Tong, & Fitza, 2012). Yet animportant concept that has scarcely received atten-tion by international management scholars is thatof “global cities”. These cities (of which London,New York, Tokyo, Paris, Chicago, and Frankfurt areprominent examples) are characterized by particu-larly high degrees of centrality and influence in theworld economy, as well as being interconnected inglobal networks that provide an infrastructure forthe global economy (Sassen, 1991, 2012; Wall &van der Knaap, 2011).

As described in more detail below, theory onglobal cities suggests that their natures are distinctfrom regions and industry clusters, or even fromcities with large populations (i.e., megacities) thatdo not exhibit global city characteristics (e.g., Coe,Dicken, Hess, & Yeung, 2010; Luthi, Thierstein, &Goebel, 2010; Markusen, 1996; Scott, 2001). Wepropose, therefore, that a global city perspectivemakes possible significant new insights into MNEbehavior, since these locations are relevant to MNElocation strategy (Nachum & Wymbs, 2005), andare integral to contemporary economic locationpatterns (Beaverstock, Smith, & Taylor, 1999).At the same time, we suggest that prior literatureon global cities has not been concerned with theidiosyncratic nature of the firms that are the raisond’etre of global cities; our research therefore alsocontributes to our understanding of economic geo-graphy by examining the internal resources andcapabilities of MNEs that influence the establish-ment and evolution of global cities.

To shed new light on global cities and MNElocation strategies – a topic that is relevant toscholars, managers, and policymakers – we firstanalyze the key contributions in the internationalbusiness and economic geography literatures. Next,we develop a theoretical argument intended tounpack the conditions under which MNEs are drawntoward, or repelled by, global cities – conditions thatrelate to the motives, idiosyncratic capabilities,international strategy, and organization of the firm.We then test our hypotheses on a large sample ofMNE location choices, providing an analysis anddiscussion of the results before we conclude with theimplications and limitations of our study.

ECONOMIC GEOGRAPHY, GLOBAL CITIES, ANDMNE LOCATION STRATEGY

For almost a century, scholars have been interestedin the emergence of privileged sites for economic

Global cities and multinational enterprise location strategy Anthony Goerzen et al

428

Journal of International Business Studies

development. Beginning with Marshall’s (1920)seminal work, a prolific and influential literatureon the nature of economic agglomeration has cometo the fore in research in both international busi-ness (e.g., Porter, 1998) and economic geography(e.g., Martin & Sunley, 2003; Scott, 1992), whereresearchers have examined various types of agglom-erations, including industrial districts or clusters,regions, and cities. In the management literatureon location strategy, several authors have examinedthe notion that proximity matters, given that localfirms tend to cite each other’s patents more fre-quently (Almeida, 1996; Frost, 2001), and thatknowledge moves slowly beyond geographic bound-aries (Baum & Haveman, 1997). The positive extern-alities stemming from co-location, described byArrow (1962) and Romer (1986), are complementedand intensified by several other interrelated elements,including the dense linkages among co-located buyers,suppliers, and customers (Porter, 1998).

Arguably, the primary perspective of most ofthese studies1 has been that of the industrialdistrict, first proposed by Marshall (1920). Wepropose, however, that a valuable and relativelyunexplored research avenue exists in the inter-section of MNE location strategy and the economicgeography of cities. Among the early theorists tofocus specifically on the nature of cities was Jacobs(1969, 1984, 2000), who built on Marshall’s (1920)work but attributed the growth of cities to thediverse activities within them that are also subjectto economies of scale due to co-location. Asanalyzed by Brown et al. (2010), a city in Jacobs’view can be understood as an ecosystem thatnaturally organizes diverse human activities (i.e.,the “little movements”) to facilitate deliberate aswell as incidental learning and innovation withinan environment that in turn leads to complexrelations with other cities (i.e., the “big wheels” ofcommerce). While the Marshallian and Jacobeantraditions both focus on the concept of geographicproximity, they differ in the sense that “Marshallcentres on belonging to a specialized producercommunity which diffuses the ‘secrets’ of industry,not the kind of cosmopolitan and haphazard citylife described by Jacobs” (Storper & Venables, 2004:353, emphasis added). As we argue below, thisdistinction in the nature of geographic proximityhas important implications for the ways in whichfirms make specific location choices with respect totheir international subsidiaries.

Some observers have suggested that cities arebecoming obsolete (see, e.g., Scott, Agnew, Soja, &

Storper, 2001, for a discussion). This propositionstems from the observation that MNEs, among themost significant forces in the global economy, areacting as centrifugal forces, “offshoring” theiroperations (Bhagwati, Panagariya, & Srinivastan,2004; Harrison & McMillan, 2006), expanding theirworldwide networks of alliances and subsidiaries(Goerzen, 2007; Goerzen & Beamish, 2005), andmoving their back-office operations from urbancenters to outlying suburbs (Sassen, 2001). However,as described by Sassen (1991, 2012), state-of-the-artinfrastructure and the specialized managerial exper-tise required to make international systems andprocesses function appear to be assembling in“global cities”, and these emerging urban phenom-ena are providing a countering centripetal force inthe global economy. Ironically, by enhancing globallinkages that tie these cities together, networktechnologies such as the Internet have onlyaccelerated this force (Sassen, 2002). These observa-tions provide an impetus to re-examine the rela-tionship between global cities and MNE locationstrategy to shed new light on the MNE ingeographic space.

What is a Global City?Major cities have been of interest to researchersfrom a variety of disciplines for years, and manylabels have been proposed to describe them,including great industrial cities, world cities,imperial cities, global capitalist cities, primatecities, and global cities. Essentially, previous workon cities can be decomposed into two approaches: ademographic and a functional tradition (Beaver-stock et al., 1999). Early scholars interested in citiesexamined them in the context of national urbansystems, mostly using demographic data, to devel-op an understanding of urban primacy or hierarchy.This demographic tradition for urban studies isfocused primarily on the human and ecologicalimplications of large human populations, such as“megacities” (Gilbert, 1996). Other researchers,however, reinterpreted the global economic roleof cities (e.g., Cohen, 1981; Friedmann, 1986;Sassen, 1991), giving rise to a new literature thatfocused on their characteristics and interconnec-tions (see, e.g., Brown et al., 2010, for an analysisof this literature). This approach has evolved intoa functional tradition in urban studies (e.g.,Friedmann, 1986; Sassen, 1994, 2012), examiningthe activities that go on within cities by virtue oftheir function in the global economy, and leadingto the development of the concept of “global


429


cities”. Our research is firmly situated in the func-tional tradition (rather than the demographic one2),since, as elaborated below, this tradition highlightsthe connections between geographically separatedactivities, thereby providing an opportunity to offernew insights into the association between geo-graphic location and international management.

An overarching theme in the functional traditionis the notion of global cities as hubs in a broaderweb of global linkages, leading to a co-evolution ofMNE expansion and global city formation3 (Jacobs,Ducruet, & De Langen, 2010). Internationalizingfirms need a global supply of business services tosupport their foreign operations; such services, inturn, are based on high information velocity, andconsequently tend to be highly localized in theiragglomeration patterns (Arzaghi & Henderson,2008). In particular, Dunning and Norman (1983)found that international business service firmslocated their European offices in London, Brussels,and Paris in order to be close to their customers(generally MNEs). This means that both MNEs andtheir business service providers will tend to clusteraround narrowly defined points in geographicspace. This explanation of global city emergence,pioneered by Sassen (1991, 1994), focuses on thespecialization and agglomeration of advanced pro-ducer services (e.g., finance, law, accounting, andadvertising), suggesting that today’s global cities –by virtue of being production spaces for the keyinputs that complex organizations need for theirglobal operations – are command and controlpoints in the organization of the world economy(Sassen, 2012).

This role in command and control has two otherimportant implications for global cities as well.First, as per the focus of early approaches within thefunctional tradition (e.g., Hall, 1966), certain citiesdevelop characteristics that emerge from politics,communications, education, culture, and othersocial factors creating a cosmopolitan environment.Such an environment is closely interlinked to thepooling of specialized managerial capabilitiesrequired by MNEs (Dunning & Norman, 1983),the use of expatriates as a coordination and controlmechanism (Martinez & Jarillo, 1989), and coordi-nation through the intensive “buzz” of face-to-faceknowledge exchange (Storper & Venables, 2004).Second, global cities tend to form external globallinkages, defined as channels for ingoing andoutgoing resource flows (Lorenzen & Mudambi,2013). The cosmopolitan environment is thus com-plemented by infrastructures that are conducive to

the inward and outward mobility of human resources(Bel & Fageda, 2008), and to the establishment ofpersonal relationships between them across geo-graphic space (Bathelt, Malmberg, & Maskell, 2004)– infrastructures that may be physical (e.g., portsand airports) as well as informational (e.g., massmedia). Friedmann (1986) provided the seminalwork on this idea, developing the concept ofprimary and secondary cities based upon ananalysis of several key criteria that related to thisextensive interconnectedness, including the existenceof major financial centers, headquarters for inter-national institutions, the rapid growth of thebusiness services sector, and major transportationnodes.

Taken together, the literature thus suggests thatthe functional tradition has converged on three keyattributes that characterize global cities: a highdegree of interconnectedness to local and globalmarkets; a cosmopolitan environment; and high levelsof advanced producer services. These characteristics,and their distinct implications for MNE locationchoice, make global cities distinct from othersubcountry units of analysis such as megacities orindustrial clusters. Megacities, for instance, arecharacterized by population size with more than10 million inhabitants, such as Calcutta, Karachi,and Dhaka, none of which exhibit global citycharacteristics (Beaverstock et al., 1999). By thesame token, industrial clusters are more industry-specific, and offer technological knowledge spil-lovers and inter-firm industry value chain linkages(Porter, 1990, 1998) of a relatively narrow kind.One example of a global city is Zurich, which isclearly cosmopolitan, extensively interconnectedwith high levels of advanced producer services – yetit is neither very large in population (less than350,000 in 1999) nor centered on just one parti-cular industry. Global cities, then, exhibit distinctattributes and, unlike other subnational geographicunits, are particularly relevant as pre-eminentcultural, political, economic, and social centers(Derudder, Taylor, Witlox and Catalano, 2003;Short, Breitbach, Buckman and Essex, 2000;Taylor, Walker and Beaverstock, 2002) with emer-ging global command and control characteristics(Sassen, 2001, 2012).

How many global cities there are and how theyfit into an international hierarchy are part of anongoing debate (Beaverstock et al., 1999; Derudderet al., 2003; Short et al., 2000; Taylor et al., 2002).Nonetheless, a growing consensus is emerging thatcertain cities are rising in importance as key centers


430


of economic coordination and control by virtue oftheir network-enabling characteristics, and thisnew reality is important for managers and interna-tional business scholars to understand. At the sametime, it is important also for economic geographersto develop a clearer understanding of the interac-tion between global cities and MNEs – the keyagents of global city formation and evolution – sowe now turn to that question.

Global Cities and the Liability of Foreignness(LOF)Hymer (1976) and Kindleberger (1969) first pointedout that foreign firms in host-country markets facecosts over and above those faced by their incum-bent competitors, and a number of empirical studieshave since provided evidence of such costs (Mezias,2002; Zaheer, 1995; Zaheer & Mosakowski, 1997).Known as the LOF in the international businessliterature, these costs are closely related to butmore broadly defined than concepts described byeconomic geographers, such as “border effects”(McCallum, 1995) and the “frictions of distance”(Appold, 1995). Importantly, in the context of MNEstrategy, these liabilities are likely to influence bothex ante geographic location decisions by MNEs andex post performance implications of such decisions(Asmussen, 2009; Zaheer, 1995).

There has been little attention directed towardsubcountry-level determinants of the LOF, how-ever, despite the fact that prior research has begunto show that it varies with the specific invest-ment location (Nachum, 2003). From this perspective,it seems likely that an MNE would not experiencethe same degree of the LOF when locating in, forexample, Shanghai, which has a high degree ofintegration into the world economy as comparedwith another less connected, perhaps rural, part ofChina (Wei & Leung, 2005; Wu, 2000). Thus, assuggested by Nachum (2003: 1202), a global city“gives rise to somewhat different sources of advan-tages and affects directly the existence and strengthof the LOF”.

While the LOF is traditionally defined verybroadly as comprising any additional costs incurredby the foreign firm, Zaheer (1995) identified severalsources as being of prime importance. Specifically:

(1) the challenges directly associated with spatialdistance (e.g., travel, transportation) and coor-dination over distance and across time zones,yielding an overall complexity of operations;

(2) firm-specific costs based on a particular com-pany’s uncertainty due to lack of familiaritywithin a local environment; and

(3) costs resulting from discrimination within thehost-country environment stemming from eco-nomic nationalism and the lack of legitimacy offoreign firms.

These key sources of the LOF – complexity, uncer-tainty, and discrimination – differ depending onwhether the firm chooses to locate within or out-side a global city. In the discussion below, therefore,we elaborate on each of these sources and analyzetheir connections to global cities.

Complexity captures the idea that MNEs sufferfrom the geographic separation between corporateheadquarters and subsidiaries, a challenge thatis not shared by indigenous firms. For example,geographic separation has been shown to inhibittrust and personal relationship-building (Luo,2001), and to lead to significant problems ofinformation asymmetry (Bergen, Dutta, & Walker,1992). As a result, MNEs incur higher communica-tion and coordination costs than do domestic firms(Zaheer & Mosakowski, 1997), making it moredifficult to control foreign subsidiaries or to avoidmoral hazard and agency problems (Gomez-Mejia& Palich, 1997).

Several features of global cities, however, mayserve to reduce this complexity. First, the inter-connectedness of these cities to the global economyshould make the transfer of capital, people, goods,and information to and from local subsidiaries faster,less costly, and more straightforward (Friedmann,1986).4 Second, agglomeration of advanced produ-cer services reduces the need to import suchservices from the home base, and allows the MNEto work with the same service providers acrossborders. Finally, complexity is easier to managebecause the cultural diversity of global citiesfacilitates the use of expatriates (Edstrom & Galbraith,1977; Martinez & Jarillo, 1989). Essentially, theeconomic and social characteristics of the globalcity, including its connectivity to local and globalmarkets, help alleviate some of the added complex-ity costs due to distance.

Uncertainty stems from the foreign firm’s unfami-liarity with the host environment, and may lead itto incur additional search costs, as well as costs ofmistakes including flawed product launches andthe failure to comply with legal and cultural norms


431


(Kostova & Zaheer, 1999; Mezias, 2002). In globalcities, however, availability of global media and anextensive IT infrastructure reduce search costs,because data about the local environment arewidely available and highly codified. Local serviceproviders are accustomed to dealing with foreignfirms, and in fact are often targeting them, makingthe host environment easier for the MNE to deci-pher. Furthermore, the agglomeration of advancedproducer services accelerates learning by supplyingMNEs with highly expert consulting and advice,enabling them to build their businesses withsuppliers who are both locally and globally knowl-edgeable and connected. Taken together, thisreasoning suggests that foreign investment uncer-tainty may be felt less acutely by MNEs wheninvesting within global cities as compared withother locations.

Discrimination may result from the foreign firm’slack of legitimacy in the local environment, andoccurs both formally, such as when foreign firmsare subject to different rules and regulationscompared with domestic firms (Kostova & Zaheer,1999), and informally, such as when consumersand employees prefer to deal with domestic firms(Zaheer & Mosakowski, 1997). In global cities,however, local stakeholders tend to be more cosmo-politan, since they are likely to be more exposed tointernational stimuli – for example, throughexpatriate or diasporic personal relationships trans-cending geographic space (Lorenzen & Mudambi,2013; Sassen, 2002). Riefler, Diamantopoulos, and

Siguaw (2011) define and operationalize cosmo-politanism with three second-order factors:open-mindedness to other countries and cultures,appreciation of diversity, and preference for inter-national consumption. Presumably, such attitudeswill lead foreign firms to be seen as more legitimatein the eyes of customers, suppliers, prospectiveemployees and so on, many of whom themselvesare foreign or have international experience andconnections (Beaverstock, 2002). In addition, theMNE deals with service providers (e.g., accountants,lawyers, advertisers, and bankers) who often haveglobal reach, specializing in serving MNEs, and whomay, in turn, lend local credibility to the MNE, justas government institutions and policies in globalcities may be more conducive to inward investment(e.g., Saito, 2003; Wu, 2000). The three essentialglobal city characteristics are mapped against thekey dimensions of the LOF in Table 1.

The implication of our analysis is that the LOF,ceteris paribus, would be expected to be lower forMNEs locating foreign subsidiaries within globalcities than elsewhere. Since the MNE is underparticularly intense and increasing pressure toimprove the efficiency of its complex structure tojustify the higher transaction costs inherent inoperating across sociopolitical and economic bor-ders (Nohria & Ghoshal, 1997), global citiesrepresent particularly attractive locations for MNEforeign subsidiaries. Given that the expansive net-works of global cities constitute the economicfabric supporting firm internationalization, weexpect MNEs, all else being equal, to gravitate

Table 1 Global cities and the liability of foreignness

Properties of

global cities

Components of the LOF

Uncertainty Discrimination Complexity

International

connectedness

Reduces search costs, because

information about local

environment is widely available and

highly codified, and local experts

are used to dealing with foreigners

Exposes local population to

international stimuli, and ensures

government institutions and policies

that are conducive to foreign direct

investment (FDI)

Makes the transfer of capital,

people, goods, and information

to and from local subsidiaries faster,

less costly, and more accurate

Advanced

producer services

Accelerates learning by supplying

MNEs with consulting and advice,

and allowing MNEs to work with

partners that are both locally and

globally knowledgeable

Legitimizes foreign firms among

service providers who are often

global firms themselves and

specialize in serving MNEs, and who

may in turn lend local credibility to

the MNE

Reduces the need to import services

from home base, and allows the

MNE to work with the same

service providers across borders,

thus reducing coordination costs

Cosmopolitan

environment

Levels the playing field for obtaining

information and gaining familiarity,

by ensuring a diversity of national

cultures

Legitimizes foreign firms in the

eyes of customers, prospective

employees, partners, and so on –

many of whom themselves are

foreign or have international

experience

Facilitates the use of expatriates,

thereby enhancing coordination

and control within the MNE


432


toward these cities in their location decisions,leading us to our first hypothesis:

Hypothesis 1: An MNE is more likely to locate itsforeign subsidiaries within global cities than inother locations.

Global Cities and Investment MotivesWhile there appear to be clear benefits of locatingactivities within global cities, there are also down-sides. One of these is the risk that, in a highinformation velocity environment such as a globalcity, competitors may appropriate knowledge spil-lovers pertaining to the MNE’s sources of competi-tive advantage (Shaver & Flyer, 2000). Furthermore,scholars have described how the economies ofco-location eventually give way to diseconomiescaused by congestion and factor price inflation asthe density of economic activity increases (e.g.,Duranton & Puga, 2003). While global cities varywidely in absolute size, it is likely that somecongestion will occur, at least in the city center,as a natural outcome of the high concentration ofeconomic activity, the narrow agglomeration ofbusiness services, and the clustering of professionaland expatriate labor forces. Hence, even though aManhattan location may provide an MNE with anunrivaled degree of centrality and visibility, expensessuch as rent and salaries may be so high as to renderthese benefits insufficient. This suggests that onlythose MNEs that, for various reasons, reap sufficientbenefits from global city attributes will be likely tolocate in these cities.

One important factor that may lead to thesebenefits is the motive behind the establishment ofthe subsidiary (Enright, 2000). Firms such as high-tech businesses, with strong intellectual propertyand thus concerns over knowledge spill-outs (Gor-don & McCann, 2000), may be disinclined to locatein global cities, preferring to settle, instead, in moreremote locations where firm activities can takeplace at lower cost and with fewer space con-straints. Conversely, it would be worth locating in aglobal city if a high priority of the MNE is to use itssubsidiary as a beachhead to access host-countrymarkets, since the LOF-reducing benefits of globalcities then become highly relevant. Thus an MNE’sattraction to global cities may be contingent on itsinvestment motivation (Nachum & Zaheer, 2005),where, in the terminology of Dunning (1993),MNEs that are “market seeking” will be more likelyto locate in global cities, whereas the otherinvestment motives (i.e., strategic asset seeking,

resource seeking, and efficiency seeking) might bepursued more efficiently outside global cities, otherthings being equal. Global city subsidiaries maythus tend to be “competence-exploiting” units thatare demand driven with a focus on market servic-ing, rather than “competence-creating” units, whichare supply driven with a focus on enhancing pro-duction and research and development competen-cies (Cantwell & Mudambi, 2005, 2011).

Competence-exploiting activities may take manyforms, and be exposed to the LOF in different ways.To the extent that they target host-country con-sumers or firms, demand-driven subsidiaries areparticularly vulnerable to two of the aforemen-tioned components of the LOF: discrimination maymake indigenous customers less positive toward theMNE’s brand; and uncertainty reinforces this effectby, for example, making it more difficult for theMNE to identify customer preferences accurately inthe host country. As explained above, global citieshelp MNEs overcome these dimensions of the LOF.For example, an MNE may use the global city asa location for accessing national media and otherbranding channels, as well as tapping into thephysical infrastructure and linkages extending fromthese cities. This suggests that MNEs that arefocused on acquiring access to customers in thehost country – henceforth local market seeking – willreap particularly high benefits from the aforemen-tioned global city attributes.

Nevertheless, local market seeking is only onetype of demand-driven investment motive. Anothertype occurs when the subsidiary is intended as aninitial location from which subsequently to reachan even larger geographic area by acting as a hubfor distribution. Such a function, often operationa-lized as a regional or global mandate (Roth &Morrison, 1992), would require extensive pipelinesfor people, goods, and information to move bothforwards and backwards through the value chain.A good example of this type of strategy is anAmerican firm that locates in London or Paris inorder to serve the European market. The inter-connectivity and infrastructure of global cities,consisting of both global links to distant locationsand national and regional links to more proximatemarkets, mean that these locations may be parti-cularly attractive to firms that aim to establishdistribution networks – henceforth referred to asglobal market servicing. While the motives describedabove are different in nature, they can both becharacterized broadly as demand driven and com-petence exploiting, and their implications for


433


global city attraction are similar. We thereforesummarize them in the following hypothesis:

Hypothesis 2a: An MNE’s propensity to locate itsforeign subsidiaries in a global city is greater if itsinvestment motive is demand driven and compe-tence exploiting (local market seeking or globalmarket servicing).

Of these demand-driven motives, targeting localmarkets is likely the most challenging, since itrequires new customers to be won over in directinteraction across cultural and institutional bar-riers. While the LOF-reducing characteristics ofglobal cities may help MNEs overcome the chal-lenge of locating abroad, the ultimate success of agiven firm will depend on its resource base enablingit to use those characteristics to its advantage.Similar to the notion of absorptive capacity (Cohen& Levinthal, 1990), firms that aim to understandand target local customer preferences need high-level knowledge and capabilities to do so. By thesame token, working closely with market-orientedproducer service providers such as advertisingagencies requires an understanding of the productsand processes of these firms. Such marketing cap-abilities are likely to reside at MNE parent level,embedded in previous experiences, such as adver-tising campaigns and distribution agreements. Tothe extent that MNEs can leverage these capabilitiesacross foreign markets when establishing newsubsidiaries, costs relating to their foreign opera-tions are likely to be lower. Thus, firms reap ahigher return on their marketing capabilities if theylocate their local-market-seeking subsidiaries inglobal cities, because they then maximize the fitbetween their strategies and resources, as reflectedby our next hypothesis:

Hypothesis 2b: The positive relationship betweenan MNE subsidiary’s motive for targeting localmarkets and its propensity to be located in aglobal city are positively moderated by the parentfirm’s marketing capabilities.

As opposed to competence-exploiting subsidiaries,competence-creating and supply-driven subsidiariesare less likely to be attracted to the particularproperties of global cities, whether they locate inforeign markets in order to establish produc-tion or seek inspiration for new product develop-ment. Production activities are typically associatedwith efficiency-seeking motives (Dunning, 1993;

Enright, 2000) related to access to raw materialsand particular types of labor. The substantialproperty costs associated with often large produc-tion facilities may, however, deter MNEs fromlocating their subsidiaries within global cities. Thusthe potential LOF-reducing benefits of global citieswould play a lesser role as firms focus on develop-ment of scale and scope advantages in productionrather than servicing a local market via sales ordistribution. Therefore, while complexity would stillarise as a function of the spatial division of labor inproduction activities, firms would be unlikely tobenefit from the availability of advanced producerservices. Nor does the attractiveness of global citiesto expatriates play a great role, as productionefficiency typically is driven by local employment.Moreover, as more firms find that the additionalcosts of locating within a global city offset thepotential advantages, production facilities andadjacent activities (such as R&D) start to gather inthe areas outside global cities, thus further increas-ing the attractiveness of such co-location.

R&D represents another competence-creatingactivity that can be characterized as a strategicasset, that is, knowledge-seeking motive (Dunning,1993; Enright, 2000). Since R&D activities areknowledge intensive, firms seek to locate suchactivities in close geographic proximity to: (1)suppliers of highly skilled knowledge, such asuniversities; and (2) firms with complementarytacit knowledge that are cognitively distant enoughto avoid unintended spillover (Suire & Vicente,2009). Tellingly, the three most famous examples ofUS high-tech industrial clusters – Silicon Valley,Route 128, and the Research Triangle – are notlocated within global cities but rather outside them,suggesting that they benefit from different types ofagglomeration economies (Malecki, 1984). In addi-tion, R&D activities are often scale sensitive,leading to MNEs establishing only a few globalcenters worldwide (Yip, 1995). Consistent with thisview, Alcacer (2006) found R&D facilities to bemore concentrated than sales and productionsubsidiaries. Large-scale R&D facilities requirespace, and the costs and inconveniences of beinglocated within a global city are likely to outweighany potential benefits. Indeed, many of the veryfeatures of a global city that serve to reduce the LOFfor demand-driven, competence-exploiting activ-ities may be of less importance or even detrimental(e.g., cost-increasing) to supply-driven, compe-tence-creating activities. Together with the signifi-cant cost of the space differential between locating


434


within or outside a global city, we therefore expectsupply-driven activities to be more likely to locateoutside the global cities, leading to the followinghypothesis:

Hypothesis 2c: An MNE’s propensity to locate itsforeign subsidiaries outside a global city is greaterif its investment motive is supply driven andcompetence creating (establishing productionnetworks or product development).

Global Cities and MNE integrationWhile we have argued above that demand-drivenmotives make MNEs more sensitive to the LOFelements of discrimination and uncertainty in theirexternal environment, the third component of theLOF – complexity – may be subject to differentcontingencies. As explained above, the addedcomplexity of organizing business internationallyforces MNEs to grapple with coordination andcontrol challenges within and among their sub-sidiaries that are clearly over and above those facedby their indigenous competitors. Part of the attrac-tion of global cities is that they alleviate these LOFproblems, as their supply of advanced businessservices and their extensive interconnectedness toglobal networks of transportation and communica-tion make it easier for MNEs to monitor theirsubsidiaries and keep them closely integrated with-in the corporate network. Global cities are alsoparticularly well suited to the deployment ofexpatriates, a key coordination and control mech-anism in MNEs (Belderbos & Heijltjes, 2005).

These benefits – which have to be weighedagainst the incremental costs of locating in globalcities – may not be equally important to all MNEs.Even within particular industries, there is a sig-nificant strategic heterogeneity among firms in theextent to which they prioritize global integrationover local responsiveness (Bartlett & Ghoshal, 1989;Prahalad & Doz, 1987). Whereas globally integratedMNEs would be particularly vulnerable to thecoordination problems that result from internationalcomplexity, locally responsive MNEs would be less so– indeed, too much coordination and control wouldbe detrimental to MNEs following a strategy of localresponsiveness. Hence an MNE’s international strat-egy is an important factor in determining its ability toreap the benefits of a global city location, as capturedin the following hypothesis:

Hypothesis 3: An MNE’s strategy of globalintegration (vs local responsiveness) is positively

related to the propensity of its subsidiaries to belocated within global cities.

Global Cities and MNE Alliance StrategyEven for a globally integrated MNE, however, somesubsidiaries are more likely than others to posecoordination and control challenges (Nohria &Ghoshal, 1994). One element that may play aparticularly important role is the presence of a jointventure ( JV) partner in the subsidiary. As noted byGeringer and Hebert (1989: 239), “in comparisonwith wholly owned subsidiaries, the exercise ofeffective control over international JVs may repre-sent a more difficult proposition for the parent”.When MNEs enter foreign markets with JVs, theynot only have to control their own countrymanagers, but also need to protect their intellectualproperty from appropriation, in addition tomanaging other types of interest conflictsbetween partner firms. Considering the complex-ity of “double-layered acculturation” inherent ininternational JVs (Barkema, Bell, & Pennings,1996), it is not surprising that high dissolutionrates have been reported for these types of opera-tions (e.g., Beamish & Lupton, 2009; Hennart,Roehl, & Zietlow, 1999).

While the literature is rich with analysis of theseproblems (e.g., Goerzen, 2007; Schilke & Goerzen,2010; Zollo, Reuer, & Singh, 2002), the geographiclocation of the JV remains an understudied ante-cedent (Globerman & Nielsen, 2007). However, theaforementioned unique challenges of the interna-tional JV suggest that it would be even moreimportant to place such a subsidiary in a strategiclocation to reduce the LOF – complexity in parti-cular. Global cities constitute such a strategiclocation where there is easy access to information,a more level playing field between the partners, andglobal connectivity so that problems can be quicklyidentified and addressed by geographically distantcorporate managers. Moreover, by virtue of theircosmopolitan environment, global cities are cultu-rally diverse, which reduces some of the issuesrelated to cultural distance. Furthermore, sinceinternal JV struggles may otherwise interfere withthe global strategy of the MNE, the drive to locateJVs in global cities would be stronger for globallyintegrated firms than for firms that pursue localresponsiveness strategies, and are therefore lessattentive to subsidiary control. This is con-sistent with Kumar and Seth’s (1998) findingthat the strategic interdependence between JVand parent stimulates the need for coordination


435


and control mechanisms, leading to our finalhypothesis:

Hypothesis 4: The positive relationship betweenan MNE’s strategy of global integration and itspropensity to be located within global cities ispositively moderated by the presence of localsubsidiary JV partners.

METHODS

DataTo assess the extent to which global cities arerelevant to MNE location strategy, we compiled adata set on the locational choices of JapaneseMNEs. Japan is the world’s third largest economy(passed by the Chinese economy only in 2010) and,as such, deserves scholarly attention (Gedajlovic &Shapiro, 2002). Furthermore, according to the WorldInvestment Report (UNCTAD, 2012), Japanese MNEshave long been among the world’s largest foreigninvestors in both developing and developed eco-nomies. They are thus ubiquitous in the globaleconomy, have created many highly successfulbrands, and are well known for their managementcapability, making their foreign location choiceshighly relevant. The core of our data was collectedin a survey by Kaigai Shinshutsu Kigyou Souran,a publication of Toyo Keizai Shinposha (2001).The surveys, which were sent to the subsidiariesthrough their parent firms, were completed witha response rate of 60% by the subsidiary generalmanagers. The survey requested basic facts such assubsidiary location, foundation date, industry,annual revenue, and capital invested, and oursample consisted of 8541 subsidiaries.

Dependent Variable

City class of subsidiary locationOur data on MNE subsidiaries were coded accordingto whether or not they were located in a given city.As noted elsewhere, global cities differ widely frommegacities, since size is not a determinant ofthe former, and cosmopolitan and interconnectedenvironment not of the latter. To distinguish globalcities from other locations, we therefore adoptBeaverstock et al.’s (1999) theoretically transparentand empirically rigorous classification of worldcities, consisting of a hierarchy of 10 a globalcities, 10 b, 35 g, and 67 d global cities, based onSassen’s (1991) concept of “global capacity”. Notable

alternatives to this list have been proposed byMasterCard, Mori, and AT Kearney (to which we returnin our discussion below), but, unlike Beaverstock et al.(1999), these other city classifications do not matchtemporally with our firm-level sample. A full list of theglobal cities we use is provided in Table 2.

Since there are several classes of global cities, as arobustness test we used two alternative cutoffpoints to create our list. The first includes citiesthat Beaverstock et al. (1999: 455) characterize asbona fide global cities (i.e., the 55 a, b, and g citiesshown in Table 2) plus those that show “evidence ofworld city formation processes” (i.e., the 67 d citiesin Table 2), creating a total of 122 cities. As asensitivity analysis, we also used a more restrictivedefinition of global cities in which only the 55 a, b,and g cities are included.

Another empirical question is how fine-grained aclassification of location choices to use. We initiallyused official city boundaries to distinguish betweenglobal cities and other locations, based on our beliefthat the dynamics ascribed to them are highlylocalized in nature. For example, Gordon andMcCann (2000) found that, unlike the surroundingmetropolitan area, the inner City of London wascharacterized by a concentration of finance andmedia industries, and a high density of regionaland domestic headquarters. Nevertheless, previousresearchers have made different distinctionsbetween the inner city and its surrounding areas(e.g., Sassen, 2001; Scott, 2001). Friedmann (1986),for example, argued that the territorial basis ofworld cities comprises not only the central city butalso the whole economic space of the surroundingregion, whereas Sassen (1991, 2010) suggested thatthere is a new type of extension of a space ofcentrality into older social geographies, such as thesuburb or the metropolitan area. Therefore, toprovide new and more nuanced insights into theassociation of location and MNE investment,we include the metropolitan city-region as adistinct third category in our empirical analysis.This resulted in three categories for our dependentvariable, where the location of each subsidiary iscoded as: 0, being within a global city proper; 1, inthe metropolitan area surrounding a global city; or2, in the periphery (i.e., anywhere else).

Independent Variables

Marketing capabilitiesAs suggested by previous authors, it is important toaccount for an MNE’s proprietary assets (Dess,


436


Table 2 A roster of world cities

a world cities (n¼10) b world cities (n¼10) g world cities (n¼35) d world cities (n¼67)

Group a1

(n¼4)

Group a2

(n¼6)

Group b1

(n¼4)

Group b2

(n¼4)

Group

b3 (n¼2)

Group g1

(n¼15)

Group g2

(n¼5)

Group g3

(n¼15)

Group

d1(n¼11)

Group d2

(n¼26)

Group d3

(n¼30)

London Chicago San Francisco Brussels Moscow Amsterdam Bangkok Atlanta Auckland Abu Dhabi Adelaide

New York Frankfurt Sydney Madrid Seoul Boston Beijing Barcelona Dublin Almaty Antwerp

Paris Hong Kong Toronto Mexico City Caracas Rome Berlin Helsinki Athens Arhus

Tokyo Los Angeles Zurich Sao Paulo Dallas Stockholm Buenos Aires Luxembourg Birmingham Baltimore

Milan Dusseldorf Warsaw Budapest Lyon Bogota Bangalore

Singapore Geneva Copenhagen Mumbai Bratislava Bologna

Houston Hamburg New Delhi Brisbane Brasilia

Jakarta Istanbul Philadelphia Bucharest Calgary

Johannesburg Kuala Lumpur Rio de Janeiro Cairo Cape Town

Melbourne Manila Tel Aviv Cleveland Colombo

Osaka Miami Vienna Cologne Columbus

Prague Minneapolis Detroit Dresden

Santiago Montreal Dubai Edinburgh

Taipei Munich Ho Chi Minh City Genoa

Washington Shanghai Kiev Glasgow

Lima Gothenburg

Lisbon Guangzhou

Manchester Hanoi

Montevideo Kansas City

Oslo Leeds

Rotterdam Lille

Riyadh Marseille

Seattle Richmond

Stuttgart St Petersburg

The Hague Tashkent

Vancouver Tehran

Tijuana

Turin

Utrecht

Wellington

Source: Beaverstock et al., 1999.

Glo

bal

citie

san

dm

ultin

atio

nal

en

terp

riselo

catio

nstra

teg

yA

nth

on

yG

oerze

net

al

43

7

Journ

al

of

Inte

rnatio

nalBusin

ess

Stu

die

s

Gupta, Hennart, & Hill, 1995). Therefore, followingprior research (e.g., Caves, 1996; Kotabe, Srinivasan,& Aulakh, 2002), we have operationalized an MNE’sproprietary marketing capabilities using advertisingintensity (i.e., advertising spending divided byrevenue).

Global integrationThe use of expatriates has been linked to thecontrol and coordination capacity of MNEs(Edstrom & Galbraith, 1977; Martinez & Jarillo,1989), and thus contains an important signal aboutfirm integration. Therefore we measure globalintegration by “expatriate intensity”, that is, theratio of expatriates to total MNE employees. Thismeasure, which is normalized for firm size bydividing the number of expatriates by parentemployees, is particularly well suited to gauge theglobal integration of our sample of firms, given thatJapanese firms are known to rely heavily onexpatriates as the primary means of globallyintegrating their subsidiaries (Belderbos & Heijltjes,2005).

Demand-driven motivesWithin the survey, each subsidiary was also askedabout the original purpose(s) of establishing a givensubsidiary. Among the common responses was “accessto local markets”, which is consistent with Dunning’s(1993) typology of internationalization motives,and is considered a demand-driven motive. Anothercommon answer was “establishment of global distri-bution network”, which can also be seen as ademand-driven motive. We use the former item tocapture local market seeking and the latter tomeasure global market servicing.

Supply-driven motivesRespondents to the survey often indicated “estab-lishment of a global production network” as themotive for establishing a foreign subsidiary, and weused this item to capture production-related, supply-driven motives. Similarly, others reported “productdevelopment and planning” as a motive behind thesubsidiary, and we used this as our measure of R&D-related, supply-driven motives.

JVWe operationalized JV as a dummy variable, wheresubsidiaries with equity JV partners were codedas 1 and wholly owned MNE subsidiaries werecoded as 0.

Control VariablesSince certain MNEs are very large, and have accessto significant internal pools of managerial exper-tise, financial capital, and network connections,our model includes parent size, defined by theMNE’s total number of employees. We also con-trolled for international experience, operationalizedas average subsidiary age. This measurement hasthe advantage that it normalizes for firm size, andtherefore is orthogonal to our previous controlvariable in the sense that it emphasizes thetime dimension of experience. We controlled forwhether the MNEs operated primarily in high-techor low-tech industries, given that previous empiri-cal research has found an effect of this distinctionon global city location (Gordon & McCann, 2000).Using the main two-digit Standard IndustrialClassification (SIC) codes reported by the compa-nies, high-tech industries were defined as the com-puter, electronics, communications, and softwareindustries, consistent with the definition by theAmerican Electronics Association.5 At the subsidi-ary level, we believe the investment motive specificto the subsidiary to be a potentially importantfactor. In addition to the above-mentioneddemand- and supply-driven motives, two moreitems from the survey’s list of motives wereidentified as being relevant to global cities: informa-tion collection and following customers/affiliated firms.Both of these variables were included as dummyvariables.6 In addition, we controlled for a numberof other potentially relevant variables, such as thesize and regional location of the subsidiary (Arregle,Beamish, & Hebert, 2009). These latter controls,however, were not significant, and were droppedfrom the final models (Snijders & Bosker, 1999).

Econometric ApproachWe tested our Hypotheses 2–4 with a multilevelmultinomial logistic regression model. In interna-tional business research, the influence of higher-level (firm) factors on subsidiary-level outcomes hastypically been investigated based on the principleof disaggregation (for each micro-unit within amacro-unit the higher level factor is recordedrepeatedly). The main problem with disaggregationis that it is statistically incorrect, as sample size isdramatically exaggerated (Snijders & Bosker, 1999),and the possibilities of both Type I and Type IIerrors are elevated (Short, Ketchen, Bennett, & duToit, 2006). Since the subsidiaries of one of theJapanese MNEs in our sample are likely to share


438


certain characteristics, such as access to commonheadquarters managerial resources, the independenceassumption underlying multiple regression wouldbe violated, as there would be a correlation betweenthe error terms of subsidiaries from the same MNE.Ignoring the hierarchical nesting may not onlypose serious statistical problems but also potentiallyconstrain the conceptual development of morecomprehensive models looking at interactionsbetween Level 1 and Level 2 variables (Raudenbush& Bryk, 2002). As noted by Chan, Makino, andIsobe (2006), multilevel modeling is highly recom-mended to further our understanding of key inter-national business phenomena.

The research design, where subsidiaries are nestedwithin MNEs (i.e., headquarters), resulted in ahierarchical data structure with two levels ofrandom variation: variation between subsidiarieswithin MNEs (Level 1) and variation between MNEs(Level 2). Data sets with a nested structure areusually not adequately represented by the prob-ability model of ordinary least squares regressionanalysis. Instead, a hierarchical linear model (HLM),which is an extension of multiple regression to amodel that includes nested random coefficients, isrecommended (Raudenbush & Bryk, 2002).

Since our dependent variable is categorical withthree possibilities, we modeled the choice of loca-tion (global city vs metropolitan area vs periphery)for each foreign subsidiary using multinomial HLM(Raudenbush, 2004). This is an extension of theBernoulli model with more than two possibleoutcomes. The default estimator in two-level multi-nomial HLM is the restricted “penalized quasi-likelihood” (PQL) estimation approach. Comparedwith other potential estimators (e.g., Laplace), PQLis less computationally intensive and gives accep-table estimations (Raudenbush, 2004: 108).

Although some approximation methods for vari-ance partitioning have been suggested for a multi-level logistic analysis, an available method formultilevel multinomial models is yet to be devel-oped. As a result, no study has yet conclusivelydemonstrated variance partitioning in a multilevelapproach with multinomial outcomes (Steele &Goldstein, 2004). Therefore, in the current model-ing, the variance component is not quantified.

RESULTSThe final sample for our multilevel models con-sisted of 6955 subsidiaries nested within 318 MNEs.Table 3 provides means, standard deviations, andcorrelations for all variables. Our independent T

ab

le3

Corr

ela

tion

san

dd

esc

rip

tive

statist

ics

Vari

ab

leM

ean

Sta

nd

ard

Devi

ati

on

.1

23

45

67

89

10

11

12

1Lo

cati

on

1.1

60.9

1—

2Lo

calm

ark

et

seeki

ng

0.3

80.4

80.0

4*

—

3G

lob

alm

ark

et

serv

icin

g0.1

00.3

00.0

6*

0.2

5*

—

4G

lob

alp

rod

uct

ion

netw

ork

0.1

50.3

6�

0.1

2*

0.2

4*

0.0

7*

—

5Pro

duct

deve

lop

men

tan

dp

lan

nin

g0.0

40.2

1�

0.0

10.1

0*

0.1

0*

0.0

7*

—

6In

form

ati

on

colle

ctio

n0.1

50.3

60.0

8*

0.3

4*

0.1

9*�

0.0

2*

0.1

7*

—

7Fo

llow

ing

cust

om

ers

0.0

30.1

7�

0.0

3*

0.0

9*

0.0

5*

0.0

9*

0.0

2*

0.0

7*

—

8JV

0.4

10.4

90.0

2*

0.0

8*�

0.0

2*

0.1

9*�

0.0

1�

0.0

5*

0.0

6*

—

9Pare

nt

size

(000)

21.9

130.3

10.0

1�

0.2

3*�

0.1

0*�

0.1

4*

0.0

6*�

0.1

2*�

0.0

4*

0.0

6*

—

10

Inte

rnati

on

alexp

eri

en

ce10.6

82.8

90.0

10.0

6*�

0.0

1�

0.0

3*�

0.0

20.0

1�

0.0

2*�

0.0

7*�

0.0

7*

—

11

Hig

h-t

ech

ind

ust

ry0.1

90.3

9�

0.0

5*�

0.0

0�

0.0

3*�

0.0

5*�

0.0

6*�

0.0

8*�

0.0

7*�

0.1

5*�

0.1

9*

0.1

0*

—

12

Mark

eti

ng

cap

ab

ility

0.0

10.0

2�

0.0

0�

0.0

00.0

10.0

10.0

1�

0.0

3*�

0.0

4*�

0.0

9*�

0.2

4*�

0.0

1�

0.0

3*

—

13

Glo

balin

teg

rati

on

0.0

20.0

30.0

8*

0.0

9*

0.0

0�

0.0

4*

0.0

2*

0.0

7*

0.0

2*�

0.0

5*�

0.1

7*�

0.0

4*

0.0

20.0

5*

n¼

6955

sub

sid

iaries

from

318

MN

Es;

*sh

ow

ssi

gn

ific

an

ceat

the

po

0.0

5le

vel.


439


variables are not highly correlated; indeed, thehighest pairwise correlation is between local marketseeking and information collection, at 0.34. More-over, analysis of variance inflation factors of eachpredictor variable revealed that all were below 2,which is well below the suggested threshold of10 (Hair, Anderson, Tatham, & Black, 1995) or eventhe more conservative one of 5. Together, theseanalyses alleviate any concerns about multicollinearity.

Hypothesis 1 proposes that MNEs are more likelyto locate their foreign subsidiaries in global citiesthan in other locations. To determine whetherthere is support for this first hypothesis, it ispossible to examine a frequency table as shown inTable 4, where it appears that global cities areclearly related to MNE location strategy. The MNEsin our sample located 3010 subsidiaries, or 35% ofthe total, in the set of 10 a world metropolitan citiesalone (i.e., Chicago, Frankfurt, Hong Kong, London,Los Angeles, Milan, New York, Paris, Singapore, andTokyo). Further, according to Beaverstock et al.’s(1999) definition of bona fide global cities (i.e., a, b,and g cities), the MNEs in our sample located 77%of their subsidiaries (i.e., 6610 sites) in this group of55 cities, with the remaining 23% of the sub-sidiaries located in “all other locations” (an infinitelist of all possible sites outside global cities includ-ing, as shown in Table 4, some megacities that donot possess global city characteristics7).

Further, the MNE subsidiaries within global citiescontained 66% of the MNEs’ foreign employees,with only 34% of these employees working outsideglobal cities. To put this latter distribution intoperspective, the same cities contain only 3.4% ofthe world’s population. In other words, MNEs havea disproportionate propensity to locate subsidiariesin global cities, choosing these cities to a muchlarger extent than would be expected by thecombined size of those cities (as shown in Table 4).This is consistent with our theoretical framework,in which the attractiveness of global cities stemsnot from their size but from their unique attributes,which include local and global linkages and theresulting ability to reduce the LOF.

Another way to provide an objective perspective onthis phenomenon is to look at the number of sub-sidiary employees per capita in each type of location.These figures, which are normalized measures of“relative MNE attraction” to a given location, arereported in the last column in Table 4. Clearly, theglobal cities have an exceptionally high representa-tion among MNE subsidiary employees, with a scoreof 4605 vs 239, which is the world average (and we T

ab

le4

MN

Ein

vest

men

tlo

cation

ing

lob

alci

tiesa

Num

ber

of

sub

sid

iari

es

Perc

en

tag

eSub

sid

iary

em

plo

yees

(mill

ion

)

Perc

en

tag

eC

ity

pop

ula

tion

(mill

ion

)

Perc

en

tag

eEm

plo

yees

per

cap

ita

b

aw

orl

dci

ties

(n¼

10)a

3010

35

0.2

918

48

0.7

6042

bw

orl

dci

ties

(n¼

10)a

749

90.2

616

56

0.8

4643

gw

orl

dci

ties

(n¼

35)a

2851

33

0.5

032

125

1.9

4000

Tota

lglo

bal

citi

es(n¼

55

)a6

61

07

71

.05

66

22

83

.44

60

5

Non

-glo

balm

eg

aci

ties

(n¼

10)

103

10.0

32

140

2.1

209

All

oth

er

loca

tion

s(n¼N

)1828

22

0.5

232

6339

94.5

82

Tota

lw

orl

d8

54

11

00

1.6

01

00

67

07

10

0.0

23

9

aO

nth

eb

asi

sof

city

pro

per

defin

itio

ns,

excl

ud

ing

metr

op

olit

an

are

as.

bEm

plo

yees

per

cap

ita¼

sam

ple

sub

sid

iary

em

plo

yees

per

mill

ion

inh

ab

itan

ts.


440


can even see the Beaverstock et al., 1999, hierarchyemerge among the global cities, with the a cities ontop, followed by the b and the g cities). Furthermore, ahigh MNE attraction factor is limited solely to globalcities, and does not extend to non-global megacities,which have only 209 subsidiaries per million inhabi-tants, that is, less than one-twentieth of the score ascompared with global cities, and in fact lower than theaverage MNE attraction of the whole world. Together,these figures reinforce the picture of global cities beinghighly attractive to MNEs, and thereby lend furthercredibility to our Hypothesis 1.

To examine the next set of hypotheses it isnecessary to analyze two-level multinomial HLMmodels, as described above. Table 5 reports theresults for the inclusive definition of global cities,including the a, b, g, and d cities from Beaverstocket al.’s (1999) classification. We estimated thelogistical probability that MNEs would locate theirsubsidiaries in either a metropolitan area or aperipheral area as opposed to within a global city.Models 1 and 2 show the main effects for bothcategories of locations vis-a-vis global cities.

Consistent with Hypothesis 2a, when FDIs aremotivated by demand-driven considerations suchas local market seeking or global market servicing,subsidiaries are more likely to locate in global citiesthan in peripheral areas (b¼�0.47; p o 0.001 andb¼�0.32; p o 0.01, respectively). At the same time,demand-driven motives do not significantly influ-ence the relative probability of subsidiaries beinglocated within the metropolitan area vs the globalcity center, suggesting that even locating nearglobal cities may help to overcome the barriers tomarket entry associated with discrimination anduncertainty. In addition, our results in Model 1 forthe establishment of production network or pro-duct development and planning (b ¼ 0.74; p o 0.01and b ¼ 0.49; p o 0.01, respectively) reveal thatsupply-driven motives are more likely to lead to asubsidiary being located outside the global city,particularly in the peripheral area, thereby supportingHypothesis 2c. Production networks are sometimeslocated in the metropolitan areas surrounding a globalcity, as indicated in Model 2. However, the potentialbenefits of locating in global cities are seemingly

Table 5 Multilevel multinomial logistic regression models of global city location

Model 1 2 3 4 H

Description Main effects only Including interactions

Periphery vs

global city

Metro vs

global city

Periphery vs

global city

Metro vs

global city

Fixed effects

Level 1 (subsidiary level):

Local market seeking (sales) �0.47*** 0.15 �0.50*** 0.14 H2a

Global market servicing (distribution) �0.32** 0.01 �0.32** 0.01 H2a

Global production networks 0.74** 0.57* 0.74*** �0.25 H2c

Product development and planning 0.49** �0.24 0.50** 0.56* H2c

Information collection �0.49*** �0.44** �0.50*** �0.43**

Follow customers/affiliated 0.34 �0.38 0.31 �0.36

JV 0.07 �0.87*** 0.06 �0.92***

Level 2 (parent MNE level):

Intercept �0.37*** �1.02*** �0.36*** �1.02***

Parent size �0.00 �0.00 �0.00 �0.00

International experience �0.03 0.02 �0.02 0.02

High-tech industry 0.32*** 0.48*** 0.31*** 0.50***

Marketing capabilities �3.40 2.79 �0.17 1.19

Global integration �6.75*** �2.19 �6.24*** �0.63 H3

Level 2 – Level 1 interactions

Marketing capabilities�Market-seeking �16.48** 8.73 H2b

Global integration� JV �2.39 �8.77* H4

Random effects

Variance 0.20*** 0.40*** 0.19*** 0.41***

Correlation 0.58 0.63

*, **, *** show significance at the p o 0.05, p o 0.01, and p o 0.001 level, respectively.


441


outweighed by the diseconomies of co-location andthe emphasis on sales and services of the city center.

Moreover, we also find support for Hypothesis 3,which points to the important association of MNEorganization (i.e., global integration) and subsidiarylocation. As indicated in Model 1, firms are morelikely to locate their foreign subsidiaries in globalcities than in other locations when following a globalintegration strategy (b¼�6.75; p o 0.001). As shownin Model 2, there is no discernible differencebetween global city and metropolitan area in termsof location associated with a global integrationstrategy; it thus appears that the metropolitan regionis tied closely enough to the global city, as was thecase for the demand-driven motives, to convey thebenefits discussed in our theory section.

In order to test Hypotheses 2b and 4, weproceeded to add cross-level interactions. Sincethe location decision of foreign subsidiaries is likelyto be influenced by the resources of the parent firm,we first tested the extent to which a match betweenparent marketing capabilities and subsidiary market-seeking motives influence the location choice. Asshown in Model 3, we found support for Hypo-thesis 2b (b¼�16.48; p o 0.01) that the propensityof market-seeking subsidiaries to locate in globalcities as opposed to the peripheral area is higherwhen the parent firm is in possession of strongmarketing capabilities. Once more, the interactionhas no explanatory power for the distinctionbetween metropolitan region and city center.

With respect to Hypothesis 4, we find only partialsupport for the notion that subsidiary JV partnersmay increase the likelihood of subsidiaries beinglocated within global cities when MNEs are globallyintegrated. On the one hand, as indicated in Model2, the likelihood of subsidiaries locating in globalcities as opposed to the metropolitan region whenJV partners are present is statistically significant(b¼�0.87; p o 0.001 for the main effect), and thiseffect is reinforced by the MNE’s global integrationstrategy (b¼�8.77; p o 0.05 for the interaction), asshown in Model 4. Yet the main effect of globalintegration is statistically insignificant for thisdecision, suggesting that, in the choice between aglobal city center and the surrounding metropoli-tan region, global integration is important only inthe case of JVs. On the other hand, when consider-ing the choice between the periphery and theglobal city center (Models 1 and 3), the JV modeand its interaction with global integration becomeinsignificant, while the direct effect of global inte-gration is highly significant. This may suggest that

the types of coordination and control problemsthat are specific to JVs require the direct connecti-vity of the global city center, whereas the issuesstemming from global integration of wholly owned(i.e., non-JV) subsidiaries can be managed on ageographically more dispersed basis that extends tothe metropolitan region.

Meaningful results emerge for the control vari-ables as well. First, an information collectionmotive is related to the probability that a subsidiarywill be located in a global city. The coefficients onthis variable suggest that the city center is moststrongly related to information collection, and thatthis motive is associated with the metropolitanregion only slightly, if at all, lending furthercredibility to the notion of global cities as a placeto reduce uncertainty and the LOF associated withinformation asymmetry. Hence the high informa-tion velocity in global city centers may benefit notonly business service firms (Dunning & Norman,1983) but also internationalizing firms in a broaderarray of industries. Second, while some firms mayfollow their competitors or customers into newmarkets, this motive surprisingly seems not torelate to the actual location decision of foreignsubsidiaries at a subnational level. While parent sizeand international experience exhibit no consistentrelationship with global city location propensity,firms from high-tech industries seem to prefer tocluster outside city centers but, in particular, inmetropolitan regions of global cities. This seems tocorroborate earlier studies (Gordon & McCann,2000), and may reflect a compromise by thesefirms in order to get some of the connectivitybenefits of the global city center while avoiding themost severe agglomeration diseconomies.

As described earlier, we also ran our multilevelmodels with a more restrictive definition of globalcities, that is, excluding the d cities that merelyhave “global city potential”. These results were verysimilar to the ones we have already reported, withthe only substantial change being that local marketservicing was not significant in the model with themore restrictive definition. Perhaps this means thatthe degree of agglomeration taking place in the topglobal cities is unnecessary, or even counterproduc-tive to the establishing of distribution networks.The significance level of the other variables did notchange, but the coefficients on information collec-tion and global integration became stronger withthe more restrictive definition. This might reflectthe fact that the information velocity in the topglobal cities is higher, and that the knowledge


442


found in these cities is more valuable and fungible,and easier to transfer internally within the MNE.Together, our results provide evidence of a complexinteraction between parent and subsidiary strategyand resources in determining the micro-location offoreign subsidiaries, underscoring the importanceof looking at the interaction between location andfirm-specific characteristics when studying MNElocational choice.

DISCUSSIONThe dominant choice among the MNEs in oursample (i.e., 77%) was to locate their foreignoperations within global cities, clearly pointing tothe importance of considering subnational levelswhen analyzing investment location decisions. Ourresults show that the MNE in geographic space isclearly attracted toward global cities that providespecific micro-locational advantages that appear tohelp mitigate the negative effects of the LOF. Morespecifically, the international connectedness, advancedproducer services, and cosmopolitan environment serveto alleviate the incremental costs (i.e., “friction”)associated with the uncertainty, discrimination, andcomplexity of doing business in a foreign environ-ment. That the attraction of global cities arises fromtheir attributes, such as linkages, rather than fromtheir size, is an assertion that is borne out by thedisproportionately high number of subsidiaryemployees per capita we found in those cities,particularly compared with the dramatically lowerratio observed in non-global megacities.

While global city attributes reduce the costsassociated with the LOF for MNEs that locatewithin them, our findings imply that this relation-ship varies with the underlying motive of the MNE,a result that reinforces Nachum and Zaheer’s (2005)conclusions. Further, drawing on Dunning (1993)and Cantwell and Mudambi (2005), we distin-guished between demand-driven (competence-exploiting) and supply-driven (competence-creat-ing) activities. Consistent with our theory, theresults show that while demand-driven market-seeking and market-serving activities, such as salesand distribution, are more likely to locate in globalcities, supply-driven efficiency-seeking and asset-seeking activities, such as production and R&D, aremore likely to be located outside the global cities,either in the metropolitan areas surrounding thesecities, or in the peripheral rural areas. Together,these findings provide further evidence for theimportance of accounting for the nature of acti-vities of subsidiaries in conjunction with locational

attributes when studying subsidiary location, sup-porting Enright’s (2009) contention that MNEsinvest on an activity-by-activity basis.

Our results suggest that MNE investment behavioris contingent upon a mixture of micro-locationalfactors as well as firm-level and subsidiary-levelattributes. This is important, because, while eco-nomic geographers have highlighted the crucialimportance of both place and space alongsideexplicit notions of distance and connectivity, littleattention has been paid in the literature to the roleof firm-specific idiosyncrasies that may give rise to –or extinguish – particular locales as primary centersof economic and social activity.

Our findings inform our understanding of globalcities as we integrate insights from MNE literatureon firm-level heterogeneity in strategies and cap-abilities with location theory pertaining to thelocation of economic activities in particular spatiallocales. MNE’s spatial division of activities results inspecific patterns of co-location in and aroundglobal cities that, in turn, influence the evolutionand development of such localities: that is, MNEsand global cities co-evolve.

While it may be the particular characteristics ofglobal cities that, in the first place, attract MNEs tolocate particularly competence-exploiting activitieswithin them, such activities further influence thecosmopolitan and interconnected nature of suchcities. At the same time, the surrounding metropo-litan and peripheral areas develop in differentways as a result of the migration of demand-driveninvestments to such areas. Thus, our work builds onMarkusen (1996), who highlighted the importanceof improving our understanding of the role ofindividual firms in creating locational characteris-tics. Our study therefore offers a theoreticallydistinctive and empirically verified view of theMNE in geographic space that allows for the simul-taneous attention to place, space, and organizationas well as spatial transaction costs associated withlocating specific activities in particular micro-loca-tions (Beugelsdijk et al., 2010).

According to Storper and Scott (2009), theinteresting theoretical question about the genesisof cities is not so much where the initial seed ofdevelopment is planted (e.g., from first-natureresources, via an initial technological/commercialbreakthrough, or at random), but how the seedthen unfolds in a process of growth and develop-ment. Our study points to the co-evolution of MNElocation strategies and the emergence of certainlocales as centers of particular types of economic


443


activity. Indeed, our study may inform local policy-makers and city planners about how to attract(or repel) particular types of foreign investment,based on the city’s ability to reduce costs of the LOF.

Our finding that parent-marketing capabilitiesenhance a market-seeking foreign subsidiary’spropensity to locate in global cities provides animportant insight into the parent role in establish-ing the geographic configuration of the MNE.Market-seeking foreign subsidiaries are naturallydrawn to global cities, owing to the locationaladvantages of international connectedness, cosmo-politan environment, and essential infrastructure;however, the ability to overcome the often highbarriers to entry and compete successfully insuch hypercompetitive markets, where the maininternational competitors are also likely to locateby virtue of the global city attractiveness, is to alarge extent contingent upon the existing market-ing capabilities of the parent. Hence the extentto which MNE headquarters contribute withpositive value-added resources to the subsidiary(Mudambi & Swift, 2011) may impact its abilityto overcome location-specific LOFs, thereby put-ting a premium on parent–subsidiary relationsin location decisions (Dellestrand & Kappen,2012).

Another important finding from our study is thecross-level interaction between parent global inte-gration strategy and the existence of subsidiaryforeign JV partners. Our results suggest that globallyintegrated firms often combine this strategy withlocal partnering strategies in order to reduce thenegative consequences of the LOF. This is in linewith the literature on international JVs, suggestingthat local JV partners may help reduce uncertaintiesassociated with cultural distance, lack of localmarket knowledge, and risk of investment (Hennart& Zeng, 2002). The fact that MNEs seeminglycombine global integration strategy with local JVpartnering in global cities points to a complex setof decisions underlying specific entry modes inspecific micro-locations. Clearly, more researchis needed to determine how entry mode choice isrelated to micro-locations, and the cost and benefitconsequences thereof. Together, these results pointto the importance of simultaneously consideringparent-level and subsidiary-level strategies whendeciding on investment locations. Such theorizingand analysis respond to recent calls for moreconsideration of multilevel issues in internationalbusiness and economic geography research (Arregleet al., 2009).

LIMITATIONS AND FUTURE RESEARCHOur study has a number of limitations that providefertile avenues for future theoretical and empiricalrefinement. First, while our data examine onlyJapanese MNEs, it may be that firms originatingfrom other countries have varying needs, andtherefore may have different responses to theattractions of global cities. Hence, extending ourstudy to a multi-country context would be aparticularly fruitful area for future research. Further,our study focuses on only two levels (headquartersand subsidiary), and we acknowledge that MNEactivities are often embedded in additional layers ofrelationships; future research may seek to capturethe complexity of subsidiary networks at the global,regional, and local levels. Also, our cross-sectionaldata does not allow us to investigate the dynamicsof subsidiary location choice. For example, the LOF-reducing properties of global cities suggest that anMNE’s initial entry into a given host country maybe within such a city, whereas subsequent entries,enriched with host-country experience, may takeplace outside it. Consistent with this idea, weobserve in our data that “global-city-centric” MNEs(defined as those that have located more than 75%of their total number of subsidiaries in global cities)tend to have only 1.6 subsidiaries in each countrywhere they operate, whereas non-global-city-cen-tric firms have 2.5 subsidiaries per country. Perhapsthis is an indication that firms locate their initial“beachhead” subsidiaries in global cities, and asthey add more subsidiaries to their portfolio, moveoutside those cities. Future studies could providemore direct tests of this portfolio idea with paneldata on subsidiary location choices.

Another potential limitation of our study is thatwe focus on the commonalities between globalcities rather than the variance among them. Forexample, global cities may vary in their industrialtraditions, and this may influence their attrac-tiveness to firms from different sectors (Sassen,2012). We have left this variance for future studiesto explore, because we are interested in howglobal cities, as a phenomenon, influence MNEbehavior and because, arguably, the properties wediscuss (see, e.g., Table 1) are somewhat fungible(Montgomery & Wernerfelt, 1988; Teece, 1982),in the sense that they may be useful to firms froma variety of industries. Nevertheless, we acknowl-edge that the extent to which global city networksform along industrial divides is ultimately an empiri-cal question that requires appropriate data to beresolved.


444


In addition, although the typology of citiesprovided by Beaverstock et al. (1999) used in thispaper is a systematic, multi-sector assessment ofcities, this list may be inadequate or incomplete. Asmentioned in the methods section, our mainimpetus was to match the timing of our firmsample with the timing of the global city list, butwe also point to the fact that more recent lists areavailable for MNE managers contemplating inter-national expansion. While a systematic longitudi-nal study of the evolution of the global cityhierarchy is beyond the scope of this paper, ourpreliminary comparisons suggest that the globalcity hierarchy does evolve over time, as one wouldexpect, but that these changes are more gradualthan abrupt, and reflect general economic devel-opment patterns at the macro level8 (see alsoDerudder et al., 2010).

There may be other locational attributes that aremeaningful to certain MNEs, given their particularneeds, and which therefore have a significant effecton MNE location strategy. In some countries,furthermore, government policies and legislationmay constrain MNEs in their subcountry locationchoices, making these attributes less important.It is also possible that MNE choices are driven notonly by objective locational attributes but alsoby legitimacy and uncertainty concerns (Henisz& Delios, 2001; Henisz & Zelner, 2005), so thatMNEs may flock to global cities through a processof “mimetic isomorphism” or locational herd-ing behavior (DiMaggio & Powell, 1983; Suire &Vicente, 2009). For example, Bell, Filatotchev, andRasheed (2012) propose that MNEs choose loca-tions for foreign listings, such as stock markets inHong Kong, Toronto, New York, and London, inpart based on such rationales.

Even if the logic of MNE investment appears to beinfluenced by the characteristics of locations, it isan empirical question for future research as to theperformance benefits for those entities that arewilling and able to locate in global cities. In theory,MNEs may receive a performance boost fromlocating in global cities, not only because thesegreatly mitigate transactions costs but also becauseof flexibility and information effects. Also, creativ-ity and innovation in global cities may beenhanced because of the variety and diversityof skills and experiences within the labor force(Feldman & Audretsch, 1999; Feldman & Florida,1994; Jaffe, Trajtenberg, & Henderson, 1993; Scottet al., 2001). On the other hand, these benefits maycome at additional costs, if MNEs’ convergence

toward global cities leads to increased competitionfor limited local resources (Stuart & Sorenson,2003) or inbred knowledge linkages that areinsulated from ideas arising outside the global citynetwork (Pouder & St John, 1996). While our studyfocused on the choice of location, future studiesmay seek to relate such choices to corporate andsubsidiary performance.

Consistent with investment location theories,investors should take into consideration economic,geographic, and institutional factors when decidingon the location of their foreign subsidiaries. Ourresearch focused on the choice between globalcities, metropolitan, and peripheral areas as afunction of the interplay between parent firmresources and subsidiary characteristics. Futureresearch may seek to add country-level predictors,such as logistical infrastructure, formal and infor-mal institutions, economic development, and pro-duction factors.

We note that our findings are based on the directinvestment location choices of MNEs; yet MNEsubsidiaries are usually prominent firms withintheir local economies, and through their local andglobal linkages these firms often form broader hubsof local value creation (Agrawal & Cockburn, 2003).In so doing, MNE subsidiaries generate substantialknowledge spillovers into the local economicenvironment through the absorption of MNEknowledge by local firms (Lorenzen & Mudambi,2013). Thus the actual influence of an MNE in agiven location is significantly greater than simplythe direct investment, or the employment gener-ated by that investment. This suggests that ourstatistical tests are probably conservative, as we donot account for an MNE’s satellite system of buyersand suppliers – an important extension availablefor future research.

Another significant area for future inquiry relatesto the socioeconomic ramifications of the propen-sity of MNEs to invest in global cities. Whereas thecollective location choices of MNEs encourage aprocess by which favored locations (such as globalcities) become ever more favored, alternative loca-tions would fall further behind in a type of“Matthew effect”,9 where the rich get richer andthe poor get poorer. Specific centers of businessreceive massive investments in real estate andtelecommunications, while outlying areas arestarved of resources. Connected, global enterprisesproduce abnormal profits, while others barelysurvive. Highly skilled global city employees seetheir incomes rise to stratospheric levels, while


445


others see their wages and opportunities driftdownwards. Through this process of unequaleconomic development fueled by MNEs (first raisedby Hymer, 1972), global cities may be disconnect-ing from their regions as they connect ever moretightly with one another (Sassen, 1997). As a result,Toronto or New York City might have more incommon with Tokyo, for example, than with othermore proximate communities. Thus, vast territoriesmay become more peripheral, excluded from theprocesses that are increasingly important in fuelingeconomic growth and global integration. Thismeans that global cities may be bending the fabricof geographic space as it applies to MNEs, warrant-ing new theorizing about the concept of distanceitself, as also called for by Zaheer, Schomaker, andNachum (2012). Also, although there has been anextensive literature on the impact of MNEs on hosteconomies that include employment, training,technology transfer, stimulation of entrepreneur-ship, and knowledge spillovers (see, e.g., Dunning,1993), there are untapped opportunities for futureresearch to investigate the connection betweenpublic micro-policy (i.e., governance of globalcities) and actions of private firms such as MNEs.

CONCLUSIONOne of the basic ideas that underpins the global cityconcept in its functional role in the global eco-nomy is that city attributes create forces thatencourage (or discourage) the co-location of peo-ple, companies, and institutions. Yet, as pointedout at the outset of this paper, previous streams ofresearch on geographic scope have generally usedthe political boundaries of nations (or regions) as abasic unit of analysis. On the basis of our analysis,however, it appears clear that different echelons ofurban agglomerations within countries vary intheir impact on MNE location choices. One possibleresult of the tension between globalization andurban agglomerations such as global cities is thatwe might end up with firm configurations thatinclude tight coupling across geographically sepa-rated agglomerations, so that a few linked locationscould dominate the world (Storper, 1991; Zaheer &Manrakhan, 2001). Since the location strategies ofMNEs overwhelmingly favor global cities, anempirical question that arises is whether or notMNEs are, in fact, finding ways to connect thesecities (Laud, Grein, & Nachum, 2009). Further, ourfocus on global cities as the unit of analysis forinvestment location choice contributes to theongoing debate regarding increasing economic

and geographic embeddedness associated withglobalization.

Location has been described as the central ques-tion that defines international business research.The results in this paper, coupled with the work ofSassen (1991, 1994, 2001) and others, suggestclearly that urban agglomerations such as globalcities have a significant impact on the locationstrategies of MNEs. Not only does this have abearing on the global movement of capital; thesefindings also have implications for the inter-national movement of people. Therefore, it isimportant, both for MNE managers responsible forthe location strategy of their subsidiaries and forscholars interested in international business, todevelop a better understanding of the influence ofglobal cities as well as the opportunities to use themicro-locational attributes to advantage. Further,through the process of globalization via cities, theinterrelationships between local social issues, suchas wage disparity, skill development, and theeconomic renewal of urban centers to MNE loca-tion choice are critical issues to public policy-makers. The concept of global cities shouldtherefore be explored further in future research, inorder to improve our understanding of the MNEwithin geographic space.

ACKNOWLEDGEMENTSOur thanks go to the seminar participants at Queen’sUniversity, the University of Leeds, Tilburg University,and the University of Leuven for guidance and supporton earlier drafts of this paper, and to Claus Vistesenand Zijun Zhang for excellent research assistance. Wealso acknowledge Mark Lorenzen, Steve Tallman, andthe participants at the JIBS Special Issue Conference atTemple University, who provided valuable input, aswell as the three anonymous JIBS reviewers. Finally, weare grateful to Guest Editors Ram Mudambi and SjoerdBeugelsdijk for their guidance during the reviewprocess, and for organizing this Special Issue. Allremaining errors and omissions are our own.

NOTES1We found that most studies in this literature stream

cite Marshall (1920), who centers on specializedproducer communities that diffuse the “secrets” ofindustry, but do not mention Jacobs (1969, 1984,2000), who describes a cosmopolitan and haphazardcity life.

2While we do not apply the demographic traditionin this paper, we distinguish between megacities andglobal cities in the empirical section.


446


3This can be considered a special case of the moregeneral notion of co-evolution between MNEs andtheir environments (e.g., Cantwell, Dunning, &Lundan, 2009).

4For example, Boeh and Beamish (2012) demon-strate that travel time between headquarters andsubsidiary is a more powerful predictor of firm gover-nance and location choices than is geographic distance.

5For more information, visit http://www.aeanet.org/Publications/IDMK_definition.asp

6Note that the investment motives are not mutuallyexclusive; this enables us to include all dummieswithout falling into the “dummy variable trap”.

7These cities, which are included for comparisonpurposes, are Guangzhou, Lagos, Calcutta, Dhaka,Karachi, Delhi, Mumbai, Cairo, Tehran, and Rio deJaneiro. All of these have more than 10 millioninhabitants, yet they are not classified as global citiesin the Beaverstock et al. (1999) list.

8Specifically, we compared the Beaverstock et al.(1999) list with a 2010 version of the same list, a 2008list from MasterCard, a 2011 list from the Morifoundation, and a 2012 list from AT Kearney. Out ofthe top 30 cities in the 1999 list, five did not showup in the top 30 of any of the more recent lists

(i.e., Caracas, Dusseldorf, Johannesburg, Prague, andSantiago). However, all of these five cities did appearbelow the top 30 in more recent lists, suggesting thatthey did not cease to be global cities, but perhapsmerely have been relegated by other emerging citiesto a lower position in the hierarchy. Similarly, relativeto the 1999 top 30, there was only one new “new”city that appeared consistently in the more recent lists– that is, Shanghai – which advanced from the 42ndplace to the 7th place in the GaWC hierarchy,following the development of the Chinese economy(Guthrie, 2009). Although not yet in the top tier,change of a similar magnitude could be observed forVienna, which did not appear at all on the 1999 list butwas at 38th place in the 2010 GaWC list, most likelyreflecting the Austrian capital’s development over theprevious decade to become a regional hub by virtue ofa de facto position as bridgehead between eastern andwestern Europe (Musil, 2009).

9The term “Matthew effect” was coined by Merton(1968), based on the biblical Gospel of Matthew: “Forwhoever has will be given more, and they will have anabundance. Whoever does not have, even what theyhave will be taken from them” (Matthew 25:29, NewInternational Version).

REFERENCESAgrawal, A., & Cockburn, I. 2003. The anchor tenant hypoth-

esis: Exploring the role of large, local, R&D-intensive firms inregional innovation systems. International Journal of IndustrialOrganization, 21(9): 1227–1253.

Alcacer, J. 2006. Location choices across the value chain: Howactivity and capability influence collocation. ManagementScience, 52(10): 1457–1471.

Alcacer, J., & Chung, W. 2007. Location strategies and knowl-edge spillovers. Management Science, 53(5): 760–776.

Almeida, P. 1996. Knowledge sourcing by foreign MNEs: Patentcitation analysis in the US semiconductor industry. StrategicManagement Journal, 17(Winter Special Issue): 155–165.

Appold, S. 1995. Agglomeration, interorganizational networks,and competitive performance in the US metalworking sector.Economic Geography, 71(1): 27–54.

Arregle, J.-L., Beamish, P., & Hebert, L. 2009. The regionaldimension of MNEs’ foreign subsidiary localization. Journal ofInternational Business Studies, 40(1): 86–107.

Arrow, K. 1962. The economic implications of learning bydoing. Review of Economic Studies, 29(3): 155–173.

Arzaghi, M., & Henderson, J. 2008. Networking off MadisonAvenue. Review of Economic Studies, 75(4): 1011–1038.

Asmussen, C. 2009. Local, regional, or global? Quantifying MNEgeographic scope. Journal of International Business Studies,40(7): 1192–1205.

Barkema, H., Bell, J., & Pennings, J. 1996. Foreign entry, culturalbarriers, and learning. Strategic Management Journal, 17(2):151–166.

Bartlett, C., & Ghoshal, S. 1989. Managing across borders: Thetransnational solution. Boston: Harvard Business School Press.

Bathelt, H., Malmberg, A., & Maskell, P. 2004. Clusters andknowledge: Local buzz, global pipelines and the processof knowledge creation. Progress in Human Geography, 28(1):31–56.

Baum, J., & Haveman, H. 1997. Love thy neighbor? Differentia-tion and agglomeration in the Manhattan hotel industry,1898–1990. Administrative Science Quarterly, 42(2): 304–338.

Beamish, P., & Lupton, N. 2009. Managing joint ventures.Academy of Management Perspectives, 23(2): 75–94.

Beaverstock, J. 2002. Transnational elites in global cities: Britishexpatriates in Singapore’s financial district. Geoforum, 33(4):525–538.

Beaverstock, J., Smith, R., & Taylor, P. 1999. A roster of worldcities. Cities, 16(6): 445–458.

Bel, G., & Fageda, X. 2008. Getting there fast: Globalization,intercontinental flights and location of headquarters. Journal ofEconomic Geography, 8(4): 471–495.

Belderbos, R., & Heijltjes, M. 2005. The determinants ofexpatriate staffing by Japanese multinationals in Asia: Control,learning and vertical business groups. Journal of InternationalBusiness Studies, 36(3): 341–354.

Bell, G., Filatotchev, I., & Rasheed, A. 2012. The liability offoreignness in capital markets: Sources and remedies. Journalof International Business Studies, 43(2): 107–122.

Bergen, M., Dutta, S., & Walker, O. 1992. Agency relationshipsin marketing: A review of the implications and applications ofagency and related theories. Journal of Marketing, 56(3): 1–24.

Beugelsdijk, S., McCann, P., & Mudambi, R. 2010. Place, spaceand organisation: Economic geography and the multinationalenterprise. Journal of Economic Geography, 10(4): 485–493.


447


Bhagwati, J., Panagariya, A., & Srinivastan, T. 2004. Themuddles over outsourcing. Journal of Economic Perspectives,18(4): 93–114.

Boeh, K., & Beamish, P. 2012. Travel time and the liability ofdistance in foreign direct investment: Location choice andentry mode. Journal of International Business Studies, 43(5):525–535.

Brown, E., Derudder, B., Parnreiter, C., Pelupessy, W., Taylor, P.,& Witlox, F. 2010. World city networks and global commoditychains: Towards a world-systems’ integration. Global Net-works, 10(1): 12–34.

Cantwell, J., & Mudambi, R. 2005. MNE competence-creatingsubsidiary mandates. Strategic Management Journal, 26(12):1109–1128.

Cantwell, J., & Mudambi, R. 2011. Physical attraction and thegeography of knowledge sourcing in multinational enter-prises. Global Strategy Journal, 1(3–4): 206–232.

Cantwell, J., Dunning, J., & Lundan, S. 2009. An evolutionaryapproach to understanding international business activity: Theco-evolution of MNEs and the institutional environment.Journal of International Business Studies, 41(4): 567–586.

Caves, R. 1996. Multinational enterprise and economic analysis,2nd edn. Cambridge, UK: Cambridge University Press.

Chan, C., Makino, S., & Isobe, T. 2006. Interdependentbehavior in foreign direct investment: The multi-level effectsof prior entry and prior exit on foreign market entry. Journal ofInternational Business Studies, 37(5): 642–665.

Coe, N., Dicken, P., Hess, M., & Yeung, H. 2010. Makingconnections: Global production networks and world citynetworks. Global Networks, 10(1): 138–149.

Cohen, R. 1981. The new international division of labour,multinational corporations and urban hierarchy. In M. Dear &A. Scott (Eds), Urbanisation and urban planning in capitalistsociety: 287–315. London: Methuen.

Cohen, W., & Levinthal, D. 1990. Absorptive capacity: A newperspective on learning and innovation. Administrative ScienceQuarterly, 35(1): 128–152.

Dellestrand, H., & Kappen, P. 2012. The effects of spatial andcontextual factors on headquarters resource allocation to MNEsubsidiaries. Journal of International Business Studies, 43(3):219–243.

Derudder, B., Taylor, P., Ni, P., De Vos, A., Hoyler, M., Hanssens,H., Bassens, D., Huang, J., Witlox, F., Shen, W., & Yang, X.2010. Pathways of change: Shifting connectivities in the worldcity network, 2000–2008. Urban Studies, 47(9): 1861–1877.

Derudder, B., Taylor, P., Witlox, F., & Catalano, G. 2003.Hierarchical tendencies and regional patterns in the world citynetwork: A global urban analysis of 234 cities. Regional Studies,37(9): 875–886.

Dess, G., Gupta, A., Hennart, J.-F., & Hill, C. 1995. Conductingand integrating strategy research at the international, corpo-rate, and business levels: Issues and directions. Journal ofManagement, 21(2): 357–393.

DiMaggio, P., & Powell, W. 1983. The iron cage revisited:Institutional isomorphism and collective rationality in organi-zational fields. American Sociological Review, 48(2): 147–157.

Dunning, J. 1993. Multinational enterprises and the globaleconomy. New York: Addison-Wesley.

Dunning, J., & Norman, G. 1983. The theory of the multi-national enterprise: An application to the multinational officelocation. Environment and Planning A, 15(5): 675–692.

Duranton, G., & Puga, D. 2003. Micro-foundations of urbanagglomeration economies, NBER Working Paper Series No.W9931, http://ssrn.com/abstract¼439613.

Edstrom, A., & Galbraith, J. 1977. Transfer of managers as acoordination and control strategy in multinational organiza-tions. Administrative Science Quarterly, 22(2): 248–263.

Enright, M. 2000. Globalization, regionalization and the knowl-edge-based economy of Hong Kong. In J. Dunning (Ed),Regions, globalization and the knowledge-based economy:391–406. Oxford: Oxford University Press.

Enright, M. 2009. The location of activities of manufacturingmultinationals in the Asia-Pacific. Journal of InternationalBusiness Studies, 40(5): 818–839.

Feldman, M., & Audretsch, D. 1999. Innovation in cities:Science-based diversity, specialization, and localized competi-tion. European Economic Review, 43(2): 409–429.

Feldman, M., & Florida, R. 1994. The geographic sources ofinnovation: Technological infrastructure and product innova-tion in the United States. Annals of the Association of AmericanGeographers, 84(2): 210–229.

Friedmann, J. 1986. The world city hypothesis. Development andChange, 17(1): 69–83.

Frost, T. 2001. The geographic sources of foreign subsidiaries’innovations. Strategic Management Journal, 22(2): 101–123.

Gedajlovic, E., & Shapiro, D. 2002. Ownership structure andfirm profitability in Japan. Academy of Management Journal,45(3): 565–575.

Geringer, M., & Hebert, L. 1989. Control and performance ofinternational joint ventures. Journal of International BusinessStudies, 20(2): 235–254.

Gilbert, A. 1996. The mega-city in Latin America. Tokyo: UnitedNations University Press.

Globerman, S., & Nielsen, B. 2007. Equity versus non-equityinternational strategic alliances involving Danish firms: Anempirical investigation of the relative importance of partnerand host country determinants. Journal of InternationalManagement, 13(4): 449–471.

Goerzen, A. 2007. Alliance networks and firm performance: Theimpact of repeated partnerships. Strategic ManagementJournal, 28(5): 487–509.

Goerzen, A., & Beamish, P. 2003. Geographic scope andmultinational enterprise performance. Strategic ManagementJournal, 24(13): 1289–1306.

Goerzen, A., & Beamish, P. 2005. The effect of alliance networkdiversity on multinational enterprise performance. StrategicManagement Journal, 26(4): 333–354.

Gomez-Mejia, L., & Palich, L. 1997. Cultural diversity and theperformance of multinational firms. Journal of InternationalBusiness Studies, 28(2): 309–336.

Gordon, I., & McCann, P. 2000. Industrial clusters: Complexes,agglomeration and/or social networks? Urban Studies, 37(3):513–532.

Guthrie, D. 2009. China and globalization: The social, economicand political transformation of Chinese society. New York: Taylor& Francis.

Hair, J., Anderson, R., Tatham, R., & Black, W. 1995. Multivariatedata analysis, 4th edn. Englewood Cliffs, NJ: Prentice Hall.

Hall, P. 1966. The world cities. London: Heinemann.Harrison, A., & McMillan, M. 2006. Dispelling some myths

about offshoring. Academy of Management Perspectives, 20(4):6–23.

Henisz, W., & Delios, A. 2001. Uncertainty, imitation and plantlocation: Japanese multinational corporations, 1990–1996.Administrative Science Quarterly, 46(3): 443–475.

Henisz, W., & Zelner, B. 2005. Legitimacy, interest grouppressures, and change in emergent institutions: The case offoreign investors and host country governments. Academy ofManagement Review, 30(2): 361–382.

Hennart, J.-F., & Zeng, M. 2002. Cross-cultural differences andjoint venture longevity. Journal of International Business Studies,33(4): 669–716.

Hennart, J.-F., Roehl, T., & Zietlow, D. 1999. “Trojan horse” or“workhorse”? The evolution of US–Japanese joint venturesin the United States. Strategic Management Journal, 20(1):15–29.

Hymer, S. 1972. The multinational corporation and the lawof uneven development. In J. Bhagwati (Ed), Economicsand world order from the 1970 s to the 1990 s : 1–17. NewYork: Collier-Macmillan.

Hymer, S. 1976. The international operations of national firms: Astudy of direct foreign investment. Cambridge, MA: MIT Press.


448


Jacobs, J. 1969. The economy of cities. New York: Vintage.Jacobs, J. 1984. Cities and the wealth of nations. New York:

Random House.Jacobs, J. 2000. The nature of economics. New York: Vintage.Jacobs, W., Ducruet, C., & De Langen, P. 2010. Integrating

world cities into production networks: The case of port cities.Global Networks, 10(1): 92–113.

Jaffe, A., Trajtenberg, M., & Henderson, R. 1993. Geographiclocalization of knowledge spillovers as evidenced by patentcitations. Quarterly Journal of Economics, 108(3): 577–598.

Kindleberger, C. 1969. American business abroad: Six lectures ondirect investment. New Haven, CT: Yale University Press.

Kostova, T., & Zaheer, S. 1999. Organizational legitimacy underconditions of complexity: The case of the multinationalenterprise. Academy of Management Review, 24(1): 82–98.

Kotabe, M., Srinivasan, S., & Aulakh, P. 2002. Multinationalityand firm performance: The moderating role of R&D andmarketing capabilities. Journal of International Business Studies,33(1): 79–97.

Krugman, P. 1991. Geography and trade. Cambridge, MA: MITPress.

Kumar, S., & Seth, A. 1998. The design of coordination andcontrol mechanisms for managing joint venture–parentrelationships. Strategic Management Journal, 19(6): 579–599.

Laud, R., Grein, A., & Nachum, L. 2009. Gaining advantagethrough global learning hubs. Journal of Practical GlobalBusiness, January.

Lorenzen, M., & Mudambi, R. 2013. Clusters, connectivity, andcatch-up: Bollywood and Bangalore in the global economy.Journal of Economic Geography 13(3): 501–534.

Luo, Y. 2001. Antecedents and consequences of personalattachment in cross-cultural cooperative ventures. Adminis-trative Science Quarterly, 46(2): 177–201.

Luthi, S., Thierstein, A., & Goebel, V. 2010. Intra-firm and extra-firm linkages in the knowledge economy: The case of theemerging mega-city region of Munich. Global Networks,10(1): 114–137.

Ma, X., Tong, T., & Fitza, M. 2012. How much does subnationalregion matter to foreign subsidiary performance? Evidencefrom Fortune Global 500 Corporations’ investment in China.Journal of International Business Studies, 44(1): 66–87.

Makino, S., & Neupert, K. 2000. National culture, transactioncosts, and the choice between joint venture and whollyowned subsidiary. Journal of International Business Studies,31(4): 705–713.

Malecki, E. 1984. High technology and local economic devel-opment. Journal of the American Planning Association, 50(3):262–269.

Markusen, A. 1996. Sticky places in slippery space: A typology ofindustrial districts. Economic Geography, 72(3): 293–313.

Marshall, A. 1920. Principles of economics, 8th edn. London:Palgrave Macmillan.

Martin, R., & Sunley, P. 2003. Deconstructing clusters: Chaoticconcept or policy panacea? Journal of Economic Geography,3(1): 5–35.

Martinez, J., & Jarillo, C. 1989. The evolution of research oncoordination mechanisms in multinational corporations. Jour-nal of International Business Studies, 20(3): 489–514.

McCallum, J. 1995. National borders matter: Canada–USregional trade patterns. American Economic Review, 85(3):615–623.

McCann, P. 2011. International business and economic geo-graphy: Knowledge, time and transactions costs. Journal ofEconomic Geography, 11(2): 309–317.

McCann, P., & Mudambi, R. 2005. Analytical differences in theeconomics of geography: The case of the multinational firm.Environment and Planning, 37(10): 1857–1876.

Merton, R. 1968. The Matthew effect in science. Science,159(3810): 56–63.

Mezias, J. 2002. Identifying liabilities of foreignness andstrategies to minimize their effects: The case of labor lawsuit

judgments in the United States. Strategic Management Journal,23(3): 229–244.

Montgomery, C., & Wernerfelt, B. 1988. Diversification,Ricardian rents, and Tobin’s q. RAND Journal of Economics,19(4): 623–632.

Mudambi, R., & Swift, T. 2011. Leveraging knowledge andcompetencies across space: The next frontier in internationalbusiness. Journal of International Management, 17(3): 186–189.

Musil, R. 2009. Global capital control and city hierarchies: Anattempt to reposition Vienna in a world city network. Cities,26(5): 255–265.

Nachum, L. 2003. Liability of foreignness in global competition?Financial service affiliates in the city of London. StrategicManagement Journal, 24(12): 1187–1208.

Nachum, L., & Wymbs, C. 2005. Product differentiation,external economies and MNE location choices: M&A in globalcities. Journal of International Business Studies, 36(4): 415–434.

Nachum, L., & Zaheer, S. 2005. The persistence of distance? Theimpact of technology on MNE motivations for foreigninvestment. Strategic Management Journal, 26(8): 747–767.

Nohria, N., & Ghoshal, S. 1994. Differentiated fit and sharedvalues: Alternatives for managing headquarters-subsidiaryrelations. Strategic Management Journal, 15(6): 491–502.

Nohria, N., & Ghoshal, S. 1997. The differentiated network:Organizing multinational corporations for value creation. SanFrancisco: Jossey-Bass.

Porter, M. 1990. The competitive advantage of nations. NewYork: Free Press.

Porter, M. 1998. Clusters and the new economics of competi-tion. Harvard Business Review, 76(6): 77–90.

Porter, M. 2001. Regions and the new economics of competi-tion. In A. Scott (Ed), Global city-regions: Trends, theory, policy:139–152. Oxford: Oxford University Press.

Pouder, R., & St John, C. 1996. Hot spots and blind spots:Geographic clusters of firms and innovation. Academy ofManagement Review, 21(4): 1192–1225.

Prahalad, C., & Doz, Y. 1987. The multinational mission: Balancinglocal demands and global vision. New York: The Free Press.

Raudenbush, S. 2004. HLM 6: Hierarchical linear and nonlinearmodeling. Chicago: Scientific Software International.

Raudenbush, S., & Bryk, A. 2002. Hierarchical linear models:Applications and data analysis methods. Thousand Oaks, CA:Sage.

Ricart, J., Enright, M., Ghemawat, P., Hart, S., & Khanna, T.2004. New frontiers in international strategy. Journal ofInternational Business Studies, 35(3): 175–200.

Riefler, P., Diamantopoulos, A., & Siguaw, J. 2011. Cosmopo-litan consumers as a target group for segmentation. Journal ofInternational Business Studies, 43(3): 285–305.

Romer, P. 1986. Increasing returns and long-run growth. Journalof Political Economy, 94(5): 1002–1037.

Roth, K., & Morrison, A. 1992. Implementing global strategy:Characteristics of global subsidiary mandates. Journal ofInternational Business Studies, 23(4): 715–735.

Saito, A. 2003. Global city formation in a capitalist develop-mental state: Tokyo and the waterfront sub-centre project.Urban Studies, 40(2): 283–308.

Sassen, S. 1991. The global city: New York, London, Tokyo.Princeton, NJ: Princeton University Press.

Sassen, S. 1994. Cities in a world economy. Thousand Oaks, CA:Pine Forge/Sage.

Sassen, S. 1997. Global cities: A challenge for urban scholarship.http://www.columbia.edu/cu/21stC/issue-2.4/sassen.html,accessed on 18 May 2005.

Sassen, S. 2001. Global cities and global city-regions: Acomparison. In A. Scott (Ed), Global city-regions: Trends,theory, policy: 78–95. Oxford: Oxford University Press.

Sassen, S. 2002. Global cities and diasporic networks: Micrositesin global civil society. In H. Anheier, M. Glasius, & M. Kaldor(Eds), Global civil society: 217–240. Oxford: Oxford UniversityPress.


449


Sassen, S. 2010. Global inter-city networks and commoditychains: Any intersections? Global Networks, 10(1): 150–163.

Sassen, S. 2012. Cities in a world economy. Los Angeles: Sage.Schilke, O., & Goerzen, A. 2010. Alliance management capability:

An investigation of the construct and its measurement. Journalof Management, 36(5): 1192–1219.

Scott, A. 1992. The role of large producers in industrial districts:A case study of high technology systems houses in SouthernCalifornia. Regional Studies, 26(3): 265–275.

Scott, A. 2001. Global city-regions: Trends, theory, policy. Oxford:Oxford University Press.

Scott, A., Agnew, J., Soja, E., & Storper, M. 2001. Global city-regions. In A. Scott (Ed), Global city-regions: Trends, theory,policy: 11–30. Oxford: Oxford University Press.

Shaver, J.M., & Flyer, F. 2000. Agglomeration economies, firmheterogeneity, and foreign direct investment in the UnitedStates. Strategic Management Journal, 21(12): 1175–1193.

Short, J., Breitbach, C., Buckman, S., & Essex, J. 2000. Fromworld cities to gateway cities: Extending the boundaries ofglobalization theory. City, 4(3): 317–340.

Short, J., Ketchen, D., Bennett, N., & du Toit, M. 2006. Anexamination of firm, industry, and time effects on perfor-mance using random coefficients modeling. OrganizationalResearch Methods, 9(3): 259.

Snijders, T., & Bosker, R. 1999. Multilevel analysis: An introductionto basic and advanced multilevel modeling. Thousand Oaks, CA:Sage.

Steele, F., & Goldstein, H. 2004. A general multilevel multistatecompeting risks model for event history data with anapplication to a study of contraceptive use dynamics.Statistical Modelling, 4(2): 145–159.

Storper, M. 1991. The limits to globalization: Technologydistricts and international trade. Economic Geography, 68(1):60–93.

Storper, M., & Scott, A. 2009. Rethinking human capital,creativity and urban growth. Journal of Economic Geography,9(2): 147–167.

Storper, M., & Venables, A. 2004. Buzz: Face-to-face contactand the urban economy. Journal of Economic Geography, 4(4):351–370.

Stuart, T., & Sorenson, O. 2003. The geography of opportunity:Spatial heterogeneity in founding rates and the performanceof biotechnology firms. Research Policy, 32(2): 229–253.

Suire, R., & Vicente, J. 2009. Why do some places succeed whenothers decline? A social interaction model of cluster viability.Journal of Economic Geography, 9(3): 381–404.

Taylor, P., Walker, D., & Beaverstock, J. 2002. Firms and theirglobal service networks. In S. Sassen (Ed), Global networks,linked cities: 93–116. New York: Routledge.

Teece, D. 1982. Towards an economic theory of the multi-product firm. Journal of Economic Behavior and Organization,3(1): 39–63.

Tihanyi, L., Griffith, D., & Russell, C. 2005. The effect of culturaldistance on entry mode choice, international diversification,and MNE performance: A meta-analysis. Journal of Interna-tional Business Studies, 36(3): 270–283.

Toyo Keizai Shinposha. 2001. Kaigai shinshutsu kigyou souran-kuni betsu. Tokyo: Toyo Keizai Ltd.

UNCTAD. 2012. World investment report. Geneva: UnitedNations.

Wall, R., & van der Knaap, G. 2011. Sectoral differentiation andnetwork structure within contemporary worldwide corporatenetworks. Economic Geography, 87(3): 267–308.

Wei, Y., & Leung, C. 2005. Development zones, foreigninvestment, and global city formation in Shanghai. Growthand Change, 36(1): 16–40.

Wu, F. 2000. The global and local dimensions of place-making:Remaking Shanghai as a world city. Urban Studies, 37(8):1359–1377.

Yip, G. 1995. Total global strategy. Englewood Cliffs, NJ: PrenticeHall.

Zaheer, S. 1995. Overcoming the liability of foreignness.Academy of Management Journal, 38(2): 341–363.

Zaheer, S., & Manrakhan, S. 2001. Concentration and disper-sion in global industries: Remote electronic access and thelocation of economic activities. Journal of International BusinessStudies, 32(4): 667–686.

Zaheer, S., & Mosakowski, E. 1997. The dynamics of the liabilityof foreignness: A global study of survival in financial services.Strategic Management Journal, 18(6): 439–463.

Zaheer, S., Schomaker, M., & Nachum, L. 2012. Distancewithout direction: Restoring credibility to a much-lovedconstruct. Journal of International Business Studies, 43(1):18–27.

Zollo, M., Reuer, J., & Singh, H. 2002. Interorganizationalroutines and performance in strategic alliances. OrganizationScience, 13(6): 701–713.

ABOUT THE AUTHORSAnthony Goerzen is the D. R. Sobey Professor ofInternational Business at Queen’s School of Businessof Queen’s University. His areas of research interestrelate to the behavior and performance of multi-national corporations, with a particular focus onlocation strategy and cooperative strategies (e.g., jointventures, alliances, and networks). Emerging areas ofinterest pertain to social and environmental sustain-ability in the context of international business.

Christian Geisler Asmussen is an Associate Profes-sor of Strategic Management and InternationalBusiness at Copenhagen Business School. Drawingon a background in economics, but applying aninterdisciplinary approach, his research and teachingrevolve around competitive advantage, globalization,and the expansion trajectories of multinationalcorporations.

Bo Bernhard Nielsen is a Professor of StrategicManagement at the Department of Strategic Man-agement and Globalization at Copenhagen Busi-ness School. His research is at the intersection ofstrategy and international business, with a specificfocus on strategic collaboration, firm internationa-lization, and knowledge management across borders.

Accepted by Ram Mudambi and Sjoerd Beugelsdijk, Guest Editors, February 2013. This paper has been with the authors for three revisions.


450


Documents

Global cities and multinational enterprise location … cities and multinational enterprise location strategy ... the global economy (Sassen, ... Global cities and multinational enterprise