Geographic clustering and outward foreign direct investment

Geographic clustering and outward foreign direct investment

International Business Review 21 (2012) 1112–1121 Contents lists available at SciVerse ScienceDirect International Business Review journal homepage:...

276KB Sizes 0 Downloads 78 Views

International Business Review 21 (2012) 1112–1121

Contents lists available at SciVerse ScienceDirect

International Business Review journal homepage: www.elsevier.com/locate/ibusrev

Geographic clustering and outward foreign direct investment§ Gary A.S. Cook a, Naresh R. Pandit b,*, Hans Lo¨o¨f c, Bo¨rje Johansson d a

University of Liverpool Management School, Chatham Street, Liverpool L69 7ZH, United Kingdom Norwich Business School, University of East Anglia, Norwich, NR4 7TJ, United Kingdom The Royal Institute of Technology, Division of Economics, Drottning Kristinasv 30, SE-100 44 Stockholm, Sweden d Jonkoping International Business School, Jonkoping University, Gjuterigatan 5, SE-551 11 Jonkoping, Sweden b c

A R T I C L E I N F O

A B S T R A C T

Article history: Received 29 September 2010 Received in revised form 2 December 2011 Accepted 12 December 2011

This study addresses an important neglected question: To what extent do geographic clusters promote outward foreign direct investment (ODI)? We find evidence that clusters do promote ODI and so support Porter’s argument that advantages gained in clusters can be the foundations of successful internationalisation. Digging deeper, we find that certain cluster incumbents promote more ODI than others, with more experienced firms and firms with stronger resource bases accounting for more ODI. We also find that firms located in clusters within major global nodes/cities engage in more ODI. Finally, we find that both localisation and urbanisation economies promote ODI. However, the former, withinindustry effects, are more important. Overall, this study echoes Dunning’s call for more focus on the ‘L’ component of the ownership, location, internalisation (OLI) paradigm and particularly on the advantages that reside in clusters that make them not only attractive destinations for foreign direct investment (FDI) but also fertile environments from which FDI can spring. ß 2011 Elsevier Ltd. All rights reserved.

Keywords: Clusters FDI MNEs ODI OLI paradigm

1. Introduction Research on both the foreign direct investment (FDI) activities of multinational enterprises (MNEs) and the advantages, disadvantages and processes that arise in geographical clusters have long and rich traditions (Buckley, 2009; Dunning, 2000; Karlsson, 2008; Marshall, 1890). A growing body of research on the interface of these two topics suggests a link between highly productive clusters and FDI (Majocchi & Presutti, 2009). For example, Kozul-Wright and Rowthorn (1998) find that highly productive clusters contain a larger than expected quantity of FDI and Nachum (2003) reports that this imbalance is increasing. Accordingly, highly productive clusters may attract FDI, and FDI may promote cluster productivity, and we know from the work of Blomstrom and Kokko (1998, 2003) that both effects will vary by location and industry. Findings like these have prompted a re-evaluation of the spatial organisation of FDI. Within the International Business (IB) literature, the seminal call for more research on location, and in particular, location in clusters, as a determinant of FDI came from Dunning (1998). He concluded: ‘‘The extent to which MNEs promote, or gravitate to, spatial clusters within a country or region is an under-researched area’’ (1998, p. 58).

§ Disclaimer: This work contains statistical data from ONS which is Crown copyright and reproduced with the permission of the controller of HMSO and Queen’s Printer for Scotland. The use of the ONS statistical data in this work does not imply the endorsement of the ONS in relation to the interpretation or analysis of the statistical data. This work uses research datasets which may not exactly reproduce National Statistics aggregates. * Corresponding author. Tel.: +44 0 1603 592886. E-mail addresses: [email protected] (Gary A.S. Cook), [email protected] (N.R. Pandit), [email protected] (H. Lo¨o¨f), [email protected] (B. Johansson).

0969-5931/$ – see front matter ß 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.ibusrev.2011.12.004

G.A.S. Cook et al. / International Business Review 21 (2012) 1112–1121

1113

Following Dunning’s call, the study of FDI into clusters (‘inward foreign direct investment’ or IDI), has gathered pace (Cook & Pandit, 2010). However, the study of the promotion of FDI from clusters (‘outward foreign direct investment’ or ODI) has been neglected. For example, two important recent surveys of MNEs and clustering focus exclusively on the location of foreign subsidiaries in clusters, paying no attention to a cluster’s propensity to promote ODI (Dunning & Lundan, 2008; Rugman & Verbeke, 2009). This is surprising given that a central proposition of Porter (1990), which spurred much academic and policy interest in clusters, was that advantages gained in clusters can be the foundations of successful internationalisation. In order to address this gap in the literature, the central question investigated in this study is: To what extent do strong clusters promote ODI? A distinction that has long been made in the clusters literature is between two potential sources of cluster advantages (Hoover, 1948): urbanisation economies, which refer to the advantages of size and diversity within a general geographic concentration of economic activity (an agglomeration) (Jacobs, 1985) and localisation economies which refer to the advantages of size and diversity within the geographic concentration of a particular productive activity (Marshall, 1890). There is a debate regarding the relative importance of urbanisation and localisation economies (Ciccone, 2002; Glaeser, Kallal, Scheinkman, & Shleifer, 1992; Henderson, Kuncoro, & Turner, 1995; Viladecans-Marsal, 2004) which has been neglected in the emerging IB literature on clusters and FDI. This is an important omission because it is localisation economies that the recent work on clusters and FDI focuses on, Porter (1990) being of this ilk, and so the implicit assumption is that localisation economies, not urbanisation economies, are the main source of cluster advantages. This assumption needs to be tested. If it is incorrect, implications, particularly those that are policy related, would need to be modified. Accordingly, this study’s second-order research question is: With respect to ODI, what type of economies are more important, urbanisation economies or localisation economies? The paper is structured as follows. The next section reviews the literature that relates to the study’s questions. The study’s method is then stated leading to the presentation of the evidence which is discussed and evaluated. The final section concludes and draws implications for theory, practice, and policy and suggests avenues for further research. 2. Literature review The idea that firm-specific advantages can be developed in strong clusters has been a mainstay of Porter’s work (1990, 2008) and that such advantages can be leveraged into foreign markets has a long tradition in theories of the MNE (Dunning, 2000). Regarding the development of firm-specific advantages in clusters, beyond so called ‘fixed effects’ (Swann, Prevezer, & Stout, 1998), advantages that exist at a location that are not a function of the co-presence of related firms and institutions (e.g., transportation links, climate and time-zone), there are advantages that are directly related the co-presence of other firms within a cluster. The bulk of the literature acknowledges and builds on the seminal work of Marshall (1890) on these advantages (Gupta & Subramanian, 2008): labour market pooling, which brings the benefits of a deeper division of labour and more highly specialised skills; the emergence of specialised input suppliers; and, technological and knowledge spillovers. Regarding the leverage of cluster-based advantages into foreign markets, an early example of the use of locational variables to explain ODI activities is Vernon’s (1966, 1979) product life cycle model. It explained the observation of earlystage production by corporate parents in their high income homes with subsequent production moving to subsidiaries in lower income and/or lower cost foreign regions. Although the reasons articulated for the US being a privileged location for innovation have clear correspondence with, for example, Porter’s (1990) diamond, namely the existence of sophisticated demand and highly advanced suppliers, the model did not explicitly consider the ability to innovate, the foundation of the ability to sell in foreign markets (Nachum & Zaheer, 2005), as arising from a location in clusters. Post-Vernon, Dunning’s (1981) ‘eclectic’ or ‘ownership, location, internalisation (OLI)’ paradigm emerged to become the most well-known theoretical envelope for explaining FDI. Whilst acknowledging certain locational advantages, the early OLI paradigm was crude in four respects: firstly, locational advantages were conceptualised at the country level. Drawing from economic geography, more recent work has taken a more sophisticated approach and encouraged IB scholars to think of locational advantages at the regional level (Buckley & Ghauri, 2004; Rugman & Verbeke, 2009). Secondly, little attention was given to the role of location in the generation of the parent’s original, domestically-based advantages. Thirdly, O, L and I benefits were often conceptualised discretely without overlap and without interaction. Finally, only tangible and costreducing or demand-enhancing locational advantages were acknowledged. Regarding these last two points, more recent work has added sophistication. Dunning (1998, 2000) emphasises that although analytically distinct, OLI advantages very importantly, do interact with one another and sometimes overlap. Dunning also indicates the growing importance of intangible and revenue generating locational advantages. For example, he maintains that the increased importance of firmspecific knowledge-intensive assets, as a source of competitive advantage has led to the increased appreciation of ‘L’ as a means to enhance ‘O’. Dunning (1998, p. 54) states: ‘‘. . . as strategic asset-acquiring investment has become more important, the locational needs of corporations have shifted from those to do with access to markets, or to natural resources, to those to do with access to knowledgeintensive assets and learning experiences, which augment their existing O specific advantages.’’ In this work, Dunning is clearly focussed on where FDI ends up rather than where it comes from. However, clusters which are attractive to inward investment, may also offer advantages to incumbent firms which enhance ‘O’ advantages and so

1114

G.A.S. Cook et al. / International Business Review 21 (2012) 1112–1121

provide a springboard for ODI. This proposition has been neglected in the IB literature and needs to be tested. For completeness, such a test would need to acknowledge that clusters, because they are usually expensive and congested locations (Swann et al., 1998), also confer disadvantages to incumbents. Within clusters, certain types of firm may be more associated with ODI than others and this relates to the more general challenge to cluster theory to account for why some firms appear to benefit more than others from cluster membership. Tallman, Jenkins, Henry, and Pinch (2004) meld insights from strategic management and economic geography to argue that membership of key clusters can be the foundation for sustained competitive advantage. They place particular emphasis on knowledge-based resources and the link to differential performance within clusters is made by arguing that there exist cluster-level knowledge systems, which some firms are better able to exploit than others. Those firms then experience superior growth and so come to occupy the top tier of the size distribution within the cluster. This idea sits comfortably with the increasingly influential Resource-Based View (Barney, 1991; Peng, 2001), not least because of the seminal work of Penrose (1959), which is a key intellectual antecedent of the Resource-Based View, on why some firms have disproportionately high growth. A proxy for such advantages therefore is firm size and so we expect that larger firms, with stronger ‘O’ advantages, will engage in more ODI. Further theoretical reasoning on the link between clusters and ODI can be gained from the literature on the internationalisation of small and medium sized firms (SMEs). This literature has evolved with three distinct strands: The Stages Model (Johanson & Vahlne, 1977, 1990); the International New Venture School (Oviatt & McDougall, 1994); and the Business Network Approach (Coviello, 2006). The Stages Model is behavioural and predicated on the assumption that risk-averse agents pursue profit under bounded rationality and imperfect information. This leads to the central proposition that firms, especially SMEs, internationalise incrementally, acquiring learning in an experiential manner. This learning may occur more rapidly in clusters in which rich circuits of information reside (Storper & Venables, 2004). On this basis, the model predicts that firm age will influence ODI. The Stages Model has long been subject to the basic objection that not all firms internationalise in small steps and that the sequential model is too mechanistic (Melin, 1992). The International New Venture School is premised on this basic thought and is motivated in part by evidence that firms are increasingly establishing international relationships and operation from a very early stage. In particular, the ability to exploit experience in international business is closely related to the ability to internationalise rapidly. What cluster theory contributes is the idea that such experience is more abundantly available at some locations than others both because in stronger clusters there will be more firms which have international experience and also because the labour markets, particularly at major ‘global nodes’ (Amin & Thrift, 1992), are highly international in composition. Similarly, the Business Network Approach begins with the premise that all firms are embedded in networks comprising suppliers, customers and competitors. The extent to which firms are embedded in networks which are international in scope, can affect the ability to successfully internationalise. From a clustering perspective, networking skills, which are known to be important in international business, can be effectively developed in a cluster that not only contains dense internal connections but also is a ‘global node’ and as such has strong connections with other global nodes/clusters. Dicken (2011) points out that MNEs have a clear tendency to site both global and regional HQs in major cities, an outstanding example of which is London, which is in the elite of world cities. London has consistently been ranked number 1 overall in the European Union’s European Cities Monitor as a place to do business, with other British cities clearly lagging. In a major government report on the competitiveness of UK cities (ODPM, 2004), London (and by extension the South East region within which it was framed) eclipsed the other cities under consideration in almost every aspect, and stands out for its highly cosmopolitan labour pool, rich in experience of business conditions abroad. Finally, a distinction that has long been made in the clusters literature is between two potential sources of cluster advantages: urbanisation economies, which refer to the advantages of size and diversity of economic activity within an agglomeration; and localisation economies which refer to the advantages of large scale in a particular industry (Hoover, 1948). Whilst localisation economies relate to the classic Marshallian advantages stated earlier, Jacobs (1985) argues for the greater importance of urbanisation economies as critical to city-region dynamism and innovation by pointing to the importance of economic diversity which enables resource variety and the interchange of different ideas, both of which are seen as important innovation inputs. The key process by which cities enter into a cycle of dynamic growth and capability is the production of goods and services it once imported leading to the export those goods and services. This process of diversifying import replacement feeds a virtuous circle in two ways. Firstly, the emergence of new industries in the city creates skills and knowledge which can be applied to other lines of activity in novel ways, many of which cannot be foreseen in advance. Secondly, the new activity, and especially its export component, provides the means to pay for a greater volume and variety of imports which can themselves provide the opportunity for yet more industries to emerge to replace the new imports and so on. Jacobs was not concerned about FDI, however the essential thrust of her argument is consistent with the idea that locations favourable to innovation will support the development of resource-based or ownership advantages, to use contemporary nomenclature, supporting not just exports but also FDI, the choice of internationalisation mode being influenced by the particular characteristics of the line of activity the firm is engaged in. As Capello (2002) recognises, both urbanisation and localisation economies may be available in large metropolitan regions which can support the clustering of activity in several industries. There is no clear cut evidence either way regarding their relative importance (Ciccone, 2002; Glaeser et al., 1992; Henderson et al., 1995; Viladecans-Marsal, 2004). To date, the emerging IB literature on clusters and MNE investment flows has neglected this debate. As stated, it is important to address this neglect because the recent work on clusters and FDI implicitly assumes that localisation economies, not urbanisation

G.A.S. Cook et al. / International Business Review 21 (2012) 1112–1121

1115

Table 1 Study propositions. Proposition

Theoretical basis

Empirical test

Strong clusters promote ODI

Dunning (1981, 2000) Porter (1990, 2008) Vernon (1966, 1979) Barney (1991) Tallman et al. (2004) Peng (2001) Johanson and Vahlne (1977, 1990) Amin and Thrift (1992) Dicken (2011) Jacobs (1985) Porter (1990, 2008)

Relationship between cluster strength and ODI

Larger firms are more likely to engage in ODI

Firms will engage in ODI incrementally Firms in clusters containing major global nodes will engage in more ODI Localisation and urbanisation economies promote ODI with the former being more important than the latter

Relationship between firm size and ODI

Relationship Relationship node cluster Relationship Relationship

between between and ODI between between

firm age and ODI location in global localisation and ODI urbanisation and ODI

economies, are the main source of cluster advantages. This assumption needs to be tested so that appropriate policy implications can be drawn. Table 1 summarises the propositions that are tested in this study. 3. Method The dataset from which this study draws is the UK’s Annual Foreign Direct Investment (AFDI) survey which includes firmlevel data on outward direct investment flows. The AFDI data were merged with additional databases, the Annual Respondents Database and the Business Structure Database, maintained by the UK’s Office for National Statistics (ONS), to provide further information on firm-specific variables. Using the AFDI database, 3011 firms were identified which had engaged in ODI between 2003 and 2005 (2005 is the most recent year for which data are available). Of these, 1895 were matched to the other two databases to allow richer analysis. 3.1. Econometric models 3.1.1. The propensity to engage in ODI A set of models estimated logit regressions based on a 1,0 dependent variable depending on whether the firm was engaged in ODI or not. The basic models have the form: Yi ¼ b1 Sizei þ b2 Size2i þ b3 Agei þ b4 Age2i þ b5 Locquoi þ b6 Locquo2i þ b7 Ownempi þ b8 Ownemp2i þb9 Totempi þ b10 Totemp2i þ b11 Regdiv þ b12 Foreign þ

i¼n X

j¼n X

i¼1

j¼1

bRi Regioni þ

bSIC j Industryi þ ui

where Yi is a latent variable. The dummy variable Yi takes the value 1 if Yi > 0; 0 otherwise. In line with Cooke and Morgan (1998), Scott (2001) and Storper (1997) the region is employed as the key spatial unit at which cluster processes operate most strongly. What distinguishes this study from much of the literature (see Beaudry & Schiffauerova, 2009) is the fact that two distinct proxies for cluster strength are used. Locquo is the location quotient of the region in which the firm is located, a region being an English Government Office Region, Wales or Scotland. This is the ratio of total employment in the firm’s 3-digit SIC industry in the region to that of total employment in that industry across the UK divided by the ratio of total employment in the region to all employment in the UK. The location quotient thus represents a relative measure of the extent of localisation economies in the region. A quotient above 1 indicates that the region has a disproportionate share of employment in a particular industry relative to its total employment. The assumption then is that there is some process underlying the development of this disproportionate concentration in a particular industry. In some cases, oil being the leading example, it is simply a question of the fact that natural resource deposits are concentrated in particular places. In other cases, net cluster advantages will have set off a positive feedback loop whereby a particular location develops superior productivity and/or innovation which attracts resources which allow for further productivity/ innovation gains and so on. In line with the first and fifth propositions in Table 1, the expectation is that the coefficient will be positive. Ownemp is the log of total employment in the firm’s 3-digit industry in the firm’s region, logs being taken because of the high positive skew in the levels of the data. This captures the absolute scale dimension of cluster strength within the firm’s industry, as distinct from the location quotient which is a relative measure. The absolute scale of the cluster is significant for several reasons. Firstly, specialisation is apt to grow as scale grows, as suggested by Smith (1776), who advanced the proposition that the division of labour is limited by the extent of the market. Labour market pooling is also subject to positive feedback (David & Rosenbloom, 1990), not least because better opportunities to do the most challenging and well-remunerated work are likely to arise in larger clusters (Cook & Pandit, 2007). Thirdly, it is in human capital that the

G.A.S. Cook et al. / International Business Review 21 (2012) 1112–1121

1116

source of many key, hard-to-imitate strategic and knowledge-based assets lie (Colbert, 2004). Again, in line with the first proposition in Table 1, the expectation is that the coefficient will be positive. There are also absolute and relative measures of urbanisation economies. Totemp is total employment in the region. This proxies the scale of urbanisation economies in the region, which is likely to influence the concentration of demand as well as variegated and specialised supply across a range of industries. In line with the fifth proposition in Table 1, the expectation is that the coefficient will be positive. This variable is measured in levels as its distribution was not skewed, unlike Ownemp. Regdiv, which is a relative measure of urbanisation economies, is a Herfindahl index of the sum of squared market shares of 3digit industries by employment within the firm’s region. This is a measure of diversity, and, in line with the fifth proposition in Table 1, is expected to take a negative sign as the smaller the value, the more diversified the regional economy. Following the theoretical reasoning in Section 2, certain control variables are included in the model. Size was measured by the natural log of numbers of employees, due to the strong positive skew. In line with the second proposition in Table 1, the expectation is that the coefficient will be positive as the ability to engage in international activity increases as firm resources increase. Age is the age in years of the firm since first registration. For subsidiaries of foreign MNEs, this is based on the age of the subsidiary. Following Johanson and Vahlne’s (1977, 1990) arguments regarding the importance of experiential learning, and in line with the third proposition in Table 1, the expectation is that the coefficient will be positive. Following the fourth proposition in Table 1, a set of regional dummies were included to capture any regional fixed effects. The expectation is that London in particular will have a positive influence on the likelihood of a firm engaging in ODI. Similarly, Wales and Scotland may also be associated with a higher propensity to engage in ODI, since they too contain capital cities, albeit ones which are far less significant in the global economic system. Finally, a set dummies were included to control for principal line of activity at the 2-digit SIC level (using 3-digit SIC dummies did not alter the conclusions), given the well known influence of line of activity on the propensity to become multinational (Dunning & Lundan, 2008) and a dummy indicating foreign ownership was included. Most foreign-owned enterprises are associated with inward investment flows, but a small number (114) also engage in outward direct investment. The sign is expected to be negative, since so few of the foreign-owned companies send outward investment flows. Size, Age, Locquo, Totemp and Ownemp were entered in quadratic form, a squared term being included to allow for either the possibility of exponential increase beyond some critical mass, or possibly diminishing returns. Inspection of the simple correlation matrix (see Table 2) did reveal strong correlations between these variables and their squared terms. However, this did not give rise to serious problems of multicollinearity, due to the high number of observations available. Coefficients were stable and not unduly influenced by either the inclusion or exclusion of the squared terms. 3.1.2. The volume of ODI flows and stocks An additional set of models were run to examine the principal influences on the volume of ODI flows and stocks. Both were measured in 2005. As flows relate to current cluster strength, stocks were investigated as a robustness check, given that there may be idiosyncratic factors influencing flows in any particular year. The Heckman (1979) two-step procedure is the preferred method as observations of ODI are censored, arising only for firms which actually undertake such investments. Failing to take into account the fact that firms have made this prior choice leads to biased estimates. The second step equation has the same form as the model set out in Section 3.1.1 and details of the selection equation are given as a footnote to Table 4. 3.1.3. Geographic diversity of ODI flows The final set of models explore the propensity to engage in ODI in heterogeneous markets, a further dimension of the degree of internationalisation, and are based on the count of the number of separate markets each firm was engaged in. The appropriate modelling technique was negative binomial regression as the presence of a small number of firms investing in a large number of countries meant that the over-dispersion test of Cameron and Trivedi (1990) rejected the restriction implicit in the Poisson model that mean and variance be equal. The dependent variable is the count of the number of markets to which the firm sends ODI and the independent variables are the same as in the logistic regressions described Section 3.1.1. Table 2 Correlation matrix.

ODI Age Age2 Size Size2 Locquo Locquo2 Totemp Totemp2 Ownemp Ownemp2

ODI

Age

Age2

Size

Size2

Locquo

Locquo2

Totemp

Totemp2

Ownemp

Ownemp2

1.000 0.029 0.020 0.014 0.011 0.001 0.001 0.003 0.009 0.005 0.006

1.000 0.887 0.249 0.208 0.033 0.031 0.023 0.021 0.165 0.170

1.000 0.186 0.174 0.022 0.021 0.013 0.012 0.093 0.095

1.000 0.945 0.077 0.072 0.032 0.029 0.140 0.142

1.000 0.086 0.081 0.024 0.022 0.110 0.111

1.000 0.872 0.004 0.008 0.085 0.081

1.000 0.018 0.025 0.009 0.009

1.000 0.995 0.030 0.029

1.000 0.002 0.001

1.000 0.996

1.000

G.A.S. Cook et al. / International Business Review 21 (2012) 1112–1121

1117

Table 3 Logit regression for engaging in outward direct investment. Variable

Coefficient

Z

Marginal effect

Size Age Locquo Locquo2 Totemp Totemp2 Ownemp Ownemp2 Regdiv London location South East location Wales location Scotland location Foreign

0.9314 0.0047 0.1698 0.0221 1.78  e7 6.41  e15 0.7820 0.0418 30.89 0.4315 0.1631 0.1706 0.2329 0.7457

55.95 1.32 1.84 1.62 1.73 1.62 4.51 4.49 0.86 2.61 1.25 0.71 1.44 7.80

0.0007*** 0.000003 0.0001* 0.00002 1.32  e10* 4.75  e18 0.0006*** 0.00003*** 0.0229 0.0004*** 0.0001 0.0001 0.0002 0.0004***

Industry dummies

Included

No. of observations Wald x2 (55) Pseudo-R2

546,778 Highly significant 0.2532

* ***

Significant at 10%. Significant at 1%.

Since most firms in the sample do not engage in ODI, a zero-inflated negative binomial estimation procedure was used in order to take account of the fact that there are two different types of firm in the population, those which choose to engage in ODI and those which do not, and the zero inflated model is preferred to a hurdle model as it is more parsimonious (Cameron & Trivedi, 2005). 4. Results and discussion 4.1. The propensity to engage in ODI The results in Table 3 indicate that the strongest influence on ODI amongst the cluster strength variables is the absolute scale measure of total employment in the region in the firm’s own industry (Ownemp), which captures localisation economies. The positive coefficient on the square is consistent with the idea that a cluster may experience rapid growth once it grows above a certain ‘critical mass’, although it is implausible that a cluster would grow exponentially, without limit. This finding is also consistent with ownership or resource-based advantages that are based on a large labour pool. It could equally indicate that there might be some advantage of a particular location which is not related to the co-presence of other firms, but which nonetheless attracts a high volume of activity, such as resource deposits or a favourable location to serve a wide market. The measure used here cannot distinguish between these two alternative interpretations, however it is consistent with the weight of theoretical argument and evidence in the literature that a large labour pool is important in dynamic cluster processes. The location quotient (Locquo) is positively and significantly associated with the probability of engaging in ODI. The result is robust to alternative specifications of the model (not reported here). In quadratic form, the negative marginal effect of the square of the location quotient is on the borderline of statistical significance (p = 0.106) and suggests diminishing returns to cluster size, which is plausible due to worsening problems of congestion as a cluster grows. Urbanisation economies, as proxied by total regional employment (Totemp), have a small statistically significant influence with the signs of the coefficients implying an inverted-U relationship, also consistent with an intuitively reasonable congestion effect, particularly in the context of the UK, which is densely populated in its main urban areas. This is again consistent with the arguments that large regions may support both strong demand and diverse innovation activity. The sign on regional diversity (Regdiv) is in line with expectation, implying greater diversity is associated with a higher probability of the firm being multinational, although it is quite a long way from statistical significance. There is a large and highly statistically significant coefficient on the dummy for London. London is a major global node in the international economy, therefore it plausibly provides a fertile environment from which to expand internationally. The dummies for Scotland and Wales are signed as expected and Scotland is not very far from significance at the 10% level. This makes sense given that Edinburgh is higher up the ranking of world cities than Cardiff, ranking third in terms of international connectivity in the UK (Taylor, 2010). The positive and highly statistically significant coefficient on firm size and the positive and close to significant (p = 0.185) coefficient on age are both reasonable as proxies of resource strength and accumulated experience respectively. The much stronger effect of firm size is consistent with the idea that firm-specific resources and capabilities are more important than age and accumulated experience. 39 2-digit SIC dummies were included in the model, 38 of which were highly statistically significant and the 39th was just outside conventional levels. The existence of these industry effects on ODI activity is as expected.

1118

G.A.S. Cook et al. / International Business Review 21 (2012) 1112–1121

Table 4 Volume of ODI flows and stocks. Variable

ODI flows

ODI stocks

Coefficient

Z

Coefficient

Z

Age Age2 Size Size2 Totemp Totemp2 Locquo Locquo2 Ownemp Ownemp2 Regdiv London location South East location Wales location Scotland location Constant

0.0565 0.0056 97.6720 0.4488 0.00003 1.23  e12 38.7413 4.4731 60.7699 2.7586 7152.909 71.3921 36.6653 19.3067 44.5837 832.3023

0.04 0.34 3.54*** 0.39 2.03** 1.96** 2.58*** 2.01** 2.18** 1.90* 1.45 2.54** 1.81* 0.58 1.86* 2.77***

1.7075 0.0219 316.1459 8.3598 0.0001 3.87  e12 147.3863 14.5915 196.5740 9.3738 6668.432 255.8687 112.0624 15.6016 71.6478 3558.227

1.24 1.31 3.49*** 2.64*** 2.13** 2.05** 2.65*** 1.99** 2.60*** 2.49** 0.46 2.58*** 1.89* 0.16 1.07 3.20***

Industry dummies Selection model

Included Included

Included Included

Rho Sigma Lambda

0.9371 315.7863 295.9223

0.9957 1131.039 1126.226

No. of observations Censored observations Wald x2 (25) Wald test of indep. equations x2 (1)

546,778 54,5306 Highly significant*** 28.75***

546,778 54,5013 Highly significant*** 61.45***

Included in the selection model: Size, Size2, Totemp, Totemp2, Locquo, Locquo2, Ownemp, Ownemp2, Regdiv, Foreign, Regional dummies, Industry dummies. * Significant at 10%. ** Significant at 5%. *** Significant at 1%.

4.2. The volume of ODI flows and stocks The flow and stock models show a very similar pattern of results (see Table 4) and underline the importance of cluster strength influences. Both localisation economy proxies, the location quotient (Locquo) and own industry employment (Ownemp), show significant quadratic relationships with the size of ODI flows and stocks. The location quotient has an inverted-U relationship, implying diminishing returns, whereas the scale variable own industry employment implies exponential increase, again implausible if extrapolated. As with the logit model in the previous section, this result is consistent with the existence of classic cluster advantages, although it does not measure them directly. The total size of the region (Totemp) is also significant and takes an inverted-U form, which implies, as expected, that beyond some point the balance between cluster advantages and disadvantages (congestion effects) will become increasingly unfavourable. Regional diversity (Regdiv) has the expected sign, implying as the index falls (indicating increasing diversity), ODI flows (and stocks) will rise. It is just outside conventional significance levels (p = 0.146) in the flow model, though much further from significance in the stock model. There are some important regional effects. London and the South East both have significant positive coefficients relative to the reference group of other English regions. Scotland also has a positive and significant coefficient in the flow model, though positive and not close to statistical significance in the stock model. London has a much larger coefficient than the other included regions, which is plausible given its status as a major node in the global economy (Taylor, 2010). Regarding the firm specific variables, age is not statistically significant, yet size is positive and significant. Size proxies resource strength and ownership advantages that enhance the ability to compete in international business. Age plays a less certain role. It is plausibly associated with a process of learning, however part of the reason firms grow is that they have other capabilities complementary to this learning, which may include superior routines (Nelson & Winter, 1982) or absorptive capacity (Cohen & Levinthal, 1989). As would be expected, there were several important and significant industry-specific effects. 4.3. Geographic diversity of ODI flows Again, the main evidence of cluster effects relates to the localisation variables (see Table 5). Own industry employment (Ownemp) and its square are both statistically significant. The coefficients imply an exponentially increasing relationship beyond some point, the most plausible interpretation of which again is that there is a critical mass a cluster must achieve before these clustering effects exert a substantial influence. The location quotient (Locquo) also has a U-shaped relationship,

G.A.S. Cook et al. / International Business Review 21 (2012) 1112–1121

1119

Table 5 Zero-inflated negative binomial model of the number of countries to which ODI is sent. Variable

Coefficient

Z

Age Age2 Size Size2 Totemp Totemp2 Locquo Locquo2 Ownemp Ownemp2 Regdiv London location South East location Scotland location Wales location Constant

0.0316 0.0009 0.0495 0.0376 3.38  e8 1.65  e15 0.5801 0.0402 1.1761 0.0781 5.4156 0.0532 0.2141 0.1661 0.1312 2.8229

0.86 0.76 0.34 2.26** 0.11 0.14 2.82*** 1.08 2.19** 2.67*** 0.06 0.11 0.61 0.36 0.23 0.79

Industry dummies Inflation model

Included Included

No. of observations No. of nonzero observations Wald x2 (26)

546,778 1754 Highly significant***

Included in the selection model: Size, Size2, Totemp, Totemp2, Locquo, Locquo2, Ownemp, Ownemp2, Regdiv, Foreign, Regional dummies, Industry dummies. ** significant at 5%. *** significant at 1%.

although the squared term is not significant. There is little evidence in favour of the urbanisation economies as proxied by total size of the region (Totemp) and regional diversity (Regdiv), both being of negligible significance. The regional fixed effects which were evident in modelling both the propensity to engage in ODI and the volume of the ODI flows are absent. Of the firm-specific variables, only the square of size is significant and implies a positive exponential relationship (Table 5). Taking these findings together, one possibility is that the ability to enter a larger range of countries may be closely related to the capabilities of particular firms and it is the case that a comparatively small number of firms are successful in entering a large number of foreign locations. This speculation bears on the issue of the fact that there is a comparatively small number of firms which develop resource strengths well in excess of the majority of others. There may also be a reverse causality. As Dunning and Lundan (2008) argue, the experience of engaging in higher levels of international activity will itself increase the ownership advantages the firm possesses and reinforce its ability to be successful in additional markets (Yang, Mudambi, & Meyer, 2008). 5. Conclusion Overall, this study finds strong and consistent evidence that clusters promote ODI and therefore it supports Porter’s (1990) proposition that advantages gained in clusters can be the foundations of successful internationalisation. The generally positive coefficients on size and age also support the idea that cluster strength promotes ODI asymmetrically: it may be a tide which lifts all boats, but a small number are able to prosper more than the rest. In line with complementary thinking in IB (Peng, 2001) and strategic management (Barney, 1991) larger cluster incumbents with stronger resource bases are more likely to engage in ODI. This process of internationalisation itself may then strengthen the firm’s ‘O’ advantages. Similarly, in line with the Stages Model, this study suggests that cluster incumbents engage in ODI incrementally over time, although the evidence here was weaker and this is in line with known problems of the Stages Model (Melin, 1992). The proposition that firms located in strong clusters within major global nodes/cities will engage in more ODI receives strong endorsement. Finally, there is evidence that both localisation and urbanisation economies are important. It is, however, the former, within-industry effects, which are more important. The IB literature needs to be more careful to distinguish between these two effects, which imply different processes whereby firms build capabilities and resource strengths. They also imply different prescriptions for policy makers interested in promoting clusters and/or innovation. The broader discourse within which the paper is situated is the regionalisation versus globalisation debate (Clark & Knowles, 2003; Flores & Aguilera, 2007). One implication of the evidence presented in the paper is to reinforce the general point made by Rugman and Verbeke (2004, 2007) that the geographic pattern of multinational activity is very unevenly spread. Rugman and Verbeke (2004), in common with much of the IB literature, also acknowledge the importance of the subnational scale when considering where MNEs will choose to locate abroad. This study indicates the sub-national scale is also highly relevant in explaining the source of FDI.

1120

G.A.S. Cook et al. / International Business Review 21 (2012) 1112–1121

Key policy and practitioner implications flow. The importance of cluster advantages as an influence on multinational activity is supportive of the idea of cluster promotion in policy circles. Policy thinking has been strongly influenced by the idea of flexible specialisation where clusters are composed of agile and highly-networked SMEs, exemplified by the Third Italy and Baden Wurttemberg. In this study, MNEs emerge as important hub firms which are central to the dynamism of regional clusters. Policy thinking should take this more explicitly into account. Our findings on localisation versus urbanisation economies suggest that policy-making should favour the cluster over the agglomeration as the unit of analysis. For practitioners there are two main implications. Firstly, access to cluster advantages is an important aspect of the location decision, although realising potential advantages is not automatic and requires competence and effort. Secondly, the problems of congestion in major clusters, hinted at in the results, mean a critical view needs to be taken of which activities are best placed or retained within a particular cluster. This study has limitations. It would be desirable to incorporate a wider range of controls for differences in regional characteristics, the regional fixed effects dummies being crude proxies for what may be multifarious sources of regional advantage or disadvantage. Also, the econometrics afford no insight into the strategic orientation of firms, nor how firms create and leverage advantages from locating within clusters: the empirical proxies for resource strength are limited. These weaknesses could be addressed by qualitative case-study research. Beyond complementary qualitative research, there are two clear directions for future research. In general, the evidence presented is consistent with the cluster providing substantial advantages to MNEs. However, the relative importance of cluster advantages and internal resource strengths on performance has only been touched upon and this provides one avenue for further research. Deeper understanding is also needed of how MNEs capitalise on advantages available in the cluster and then disseminate them throughout their global operations.

References Amin, A., & Thrift, N. (1992). Neo-Marshallian nodes in global networks. International Journal of Urban and Regional Research, 16, 571–587. Barney, J. B. (1991). Firm resources and sustained competitive advantage. Journal of Management, 17, 99–120. Beaudry, C., & Schiffauerova, S. (2009). Who’s right, Marshall or Jacobs? The localisation versus urbanisation debate. Research Policy, 38, 318–337. Blomstrom, M., & Kokko, A. (1998). Multinational corporations and spillovers. Journal of Economic Surveys, 12, 1–31. Blomstrom, M., & Kokko, A. (2003). The economics of foreign direct investment incentives. National Bureau of Economic Research, Working Paper 9489. Buckley, P. J. (2009). Business history and international business. Business History, 51, 307–333. Buckley, P. J., & Ghauri, P. N. (2004). Globalisation, economic geography and the strategy of multinational enterprises. Journal of International Business Studies, 35, 81–98. Cameron, A., & Trivedi, P. K. (1990). Regression-based tests for over-dispersion in the Poisson model. Journal of Econometrics, 46, 347–364. Cameron, A., & Trivedi, P. K. (2005). Microeconometrics: Methods and applications. Cambridge: Cambridge University Press. Capello, R. (2002). Entrepreneurship and spatial externalities: Theory and measurement. The Annals of Regional Science, 36, 387–402. Ciccone, A. (2002). Agglomeration effects in Europe. European Economic Review, 46, 213–227. Clark, T., & Knowles, L. L. (2003). Global myopia: Globalization theory in international business. Journal of International Management, 9, 361–372. Cohen, W. M., & Levinthal, D. A. (1989). Innovation and learning: The two faces of R&D. The Economic Journal, 99, 569–596. Colbert, B. A. (2004). The complex resource based view: Implications for theory and practice of strategic human resource management. Academy of Management Review, 29, 341–360. Cook, G. A. S., & Pandit, N. R. (2007). Service industry clustering: A comparison of broadcasting in three city regions. Service Industries Journal, 27, 453–469. Cook, G. A. S., & Pandit, N. R. (2010). International linkages and entrepreneurship in media clusters: Evidence from the UK. In C. Karlsson, B. Johansson, & R. Stough (Eds.), Entrepreneurship and regional development: Local processes and global patterns. Cheltenham: Edward Elgar. (Chapter 6). Cooke, P., & Morgan, K. (1998). The associational economy. Firms, regions and innovation. Oxford: Oxford University Press. Coviello, N. (2006). The network dynamics of international new ventures. Journal of International Business Studies, 37, 713–731. David, P. A., & Rosenbloom, J. L. (1990). Marshallian factor market externalities and the dynamics of industrial localisation. Journal of Urban Economics, 28, 349– 370. Dicken, P. (2011). Global shift. Mapping the changing contours of the world economy. London: Sage. Dunning, J. H. (1981). International production and the multinational enterprise. London: Allen & Unwin. Dunning, J. H. (1998). Location and the multinational enterprise: A neglected factor? Journal of International Business Studies, 29, 45–66. Dunning, J. H. (2000). The eclectic paradigm as an envelope for economic and business theories of MNE activity. International Business Review, 9, 163–190. Dunning, J. H., & Lundan, S. M. (2008). Multinational enterprises and the global economy. Cheltenham: Edward Elgar. Flores, R. G., & Aguilera, R. V. (2007). Globalization and location choice: An analysis of US multinational firms in 1980 and 2000. Journal of International Business Studies, 38, 1187–1210. Glaeser, E. L., Kallal, H. D., Scheinkman, J. A., & Shleifer, A. (1992). Growth in cities. Journal of Political Economy, 100, 1126–1152. Gupta, V., & Subramanian, R. (2008). Seven perspectives on regional clusters and the case of Grand Rapids office furniture city. International Business Review, 17, 371–384. Heckman, J. J. (1979). Sample selection bias as a specification error. Econometrica, 47, 153–161. Henderson, V., Kuncoro, A., & Turner, M. (1995). Industrial development in cities. Journal of Political Economy, 103, 1067–1090. Hoover, E. M. (1948). The location of economic activity. New York: McGraw-Hill. Jacobs, J. (1985). Cities and the wealth of nations. Principles of economic life. Harmondsworth: Penguin. Johanson, J., & Vahlne, J.-E. (1977). The internationalisation process of the firm—A model of knowledge development and increasing foreign market commitments. Journal of International Business Studies, 8, 23–32. Johanson, J., & Vahlne, J.-E. (1990). The mechanism of internationalization. International Marketing Review, 7, 11–24. Karlsson, C. (Ed.). (2008). Handbook of research on cluster theory. Cheltenham: Edward Elgar. Kozul-Wright, R., & Rowthorn, R. (1998). Spoilt for choice? Multinational corporations and the geography of international production. Oxford Review of Economic Policy, 14, 74–92. Majocchi, A., & Presutti, M. (2009). Industrial clusters, entrepreneurial culture and the social environment: The effects on FDI distribution. International Business Review, 18, 76–88. Marshall, A. (1890). Principles of economics. London: Macmillan. Melin, L. (1992). Internationalization as a strategy process. Strategic Management Journal, 13, 99–118 (Special issue Fundamental Themes in Strategy Process Research). Nachum, L. (2003). Liability of foreignness in global competition? Financial service affiliates in the City of London. Strategic Management Journal, 24, 1187–1208.

G.A.S. Cook et al. / International Business Review 21 (2012) 1112–1121

1121

Nachum, L., & Zaheer, S. (2005). The persistence of distance? The impact of technology on MNE motivations for foreign investment. Strategic Management Journal, 26, 747–767. Nelson, R. R., & Winter, S. G. (1982). An evolutionary theory of economic change. Cambridge. MA Harvard University Press. Office Of The Deputy Prime Minister. (2004). Competitive European cities: Where do the core cities stand? London: ODPM. Oviatt, B. M., & McDougall, P. P. (1994). Toward a theory of international new ventures. Journal of International Business Studies, 25, 45–64. Peng, M. W. (2001). The resource-based view in international business. Journal of Management, 27, 803–829. Penrose, E. T. (1959). The theory of the growth of the firm. Oxford: Basil Blackwell. Porter, M. E. (1990). The competitive advantage of nations. London: Macmillan. Porter, M. E. (2008). Clusters and competition: New agendas for companies, governments and institutions. In M. E. Porter (Ed.), On competition (updated edition, pp. 213–304). Cambridge, MA: HBS Press. Rugman, A. M., & Verbeke, A. (2004). A perspective on regional and global strategies of multinational enterprises. Journal of International Business Studies, 35, 3–18. Rugman, A. M., & Verbeke, A. (2007). Liabilities of regional foreignness and the use of firm-level versus country-level data: A response to Dunning et al. Journal of International Business Studies, 38, 200–205. Rugman, A. M., & Verbeke, A. (2009). Location and competitiveness. In A. Rugman (Ed.), The Oxford handbook of international business. Oxford: Oxford University Press. Scott, A. J. (Ed.). (2001). Global city-regions. Oxford: Oxford University Press. Smith, A. (1776). An inquiry into the nature and causes of the wealth of nations. New York: Modern Library. (1937). Storper, M. (1997). The regional world: Territorial development in a global economy. New York: Guilford. Storper, M., & Venables, A. (2004). Buzz: Face-to-face contact and the urban economy. Journal of Economic Geography, 4, 351–370. Swann, G. M. P., Prevezer, M., & Stout, D. (Eds.). (1998). The dynamics of industrial clustering: International comparisons in computing and biotechnology. Oxford: Oxford University Press. Tallman, S., Jenkins, M., Henry, N., & Pinch, S. (2004). Knowledge, clusters and competitive advantage. Academy of Management Review, 29, 258–271. Taylor, P. (2010). UK cities in globalisation. In P. J. Taylor, P. Ni, B. Derudder, M. Hoyler, J. Huang, & F. Witlox (Eds.), Global urban analysis. London: Earthscan. Vernon, R. (1966). International investment and international trade in the product cycle. Quarterly Journal of Economics, 80, 190–207. Vernon, R. (1979). The product cycle hypothesis in a new international environment. Oxford Bulletin of Economics and Statistics, 41, 255–267. Viladecans-Marsal, E. (2004). Agglomeration economies and industrial location: City-level evidence. Journal of Economic Geography, 4, 565–582. Yang, Q., Mudambi, R., & Meyer, K. E. (2008). Conventional and reverse knowledge flows in multinational corporations. Journal of Management, 34, 882–902.