UNIVERSITY OF STRATHCLYDE Strathclyde International Business Unit Working Paper 2002/01 Sourcing Patterns of Multinational Subsidiaries in Europe: Testing the Determinants

The paper investigates and tests empirically the determinants of sourcing patterns of foreign multinational subsidiaries in Europe. Starting with the notion that local sourcing constitutes one of the main defining elements of embeddedness and a critical factor in terms of local development impact, a conceptual framework of the main factors influencing local versus foreign sourcing is proposed, followed by rigorous quantitative treatment of such factors. Drawing on a large-scale cross-country, cross-industry dataset including 233 subsidiaries in four EU host countries (namely Portugal, Spain, Ireland and the UK), hypotheses on key determinants of the choice between local and foreign sourcing are tested. These include usually investigated variables such as home country, host country, sector, age of subsidiary and entry mode, but also include novel features such as testing the impact of economic integration and a specific link of MNEs’ sourcing decisions to the increasingly influential subsidiary strategy literature. Using foreign inputs and components as the dependent variable, econometric analysis highlights that economic integration, globalised industries, host country, incentive concession, and subsidiary roles (Rationalised Manufacturer and Product Mandate Subsidiaries) are associated with greater import propensities. A number of important policy issues are discussed. Regarding subsidiary roles, the position of Product Mandates, which are promoted and nurtured for their contribution to dynamic comparative advantage, is particularly noteworthy. The evidence presented is that globalisation and competitiveness pressures may be leading to increased international procurement in these quasi autonomous subsidiaries. If traditional targeting criteria (according to home country, age, entry mode, concession of incentives per se, stimulus to Product Mandates) do not seem to be effective (according to our results), probably the best way ahead for governments really committed to stimulate local sourcing linkages will be to focus on enhancing the capabilities of the local industrial fabric. Contact author: Professor Stephen Young Department of Marketing University of Strathclyde Stenhouse Building 173 Cathedral Street Glasgow G4 0RQ Email: [email protected] Phone: ++ 44 (0)141 548 3146 Fax: ++ 44 (0)141 552 2802 SOURCING PATTERNS OF MULTINATIONAL SUBSIDIARIES IN EUROPE: TESTING THE DETERMINANTS INTRODUCTION The purpose of this paper is to provide explanations for the sourcing patterns of multinational (MNE) subsidiaries in four European countries (Portugal, Spain, Ireland and the UK). Specifically, the article focuses upon the influences on local and foreign input sourcing, and provides new evidence concerning both external and internal (particularly MNE subsidiary) environmental determinants. The topic continues to attract considerable attention from host country governments because of the importance of local linkages in generating externalities and technology spillovers and in supporting the agglomeration of industries (Young, Hood and Peters, 1994; Rodríguez-Clare, 1996; Markusen and Venables, 1999; Birkinshaw and Hood, 2000; Driffield, Munday and Roberts, 2001; Görg and Ruane, 2001). Furthermore, the globalisation of MNEs’ supply chains is creating major competitive pressures for local suppliers, as companies explore globaland single-sourcing options (Young, Hood and Firn, 2002). This paper adds to existing knowledge in isolating the influence of a range of subsidiary characteristics; and in studying input sourcing in four countries of differing characteristics in respect of size, industrial base and level of economic development, and countries which entered the European Union at differing periods. The study involves an econometric analysis of the determinants of local linkages, using imported inputs (rather than the conventional measure of locally-sourced inputs) as the dependent variable. Some explanation of the relationships between local and foreign sourcing is therefore necessary. Studies of local sourcing tend to focus upon sourcing from indigenous as opposed to MNE affiliates in the host country. This is because of the concern with embeddedness, the upgrading of the technological and other capabilities of domestically-owned enterprises and other spillover effects on economic development (Blomström et al., 1995; Hood and Young, 2000). However, the attraction of multinational suppliers may represent a route for the economy into higher value-added and technologyand knowledge-intensive production inputs, which would otherwise be imported. Some foreign-owned suppliers commonly move abroad and co-locate together with their multinational customers. When local sourcing is taken to include both indigenous and foreign-owned suppliers, it becomes effectively the mirror image of foreign sourcing. There are still likely to be quality differences between local and foreign inputs, with substantial evidence (UNCTAD, 2001) to show that locally sourced items include lower value, transport cost-intensive production and service inputs. A consideration of the import side of sourcing provides important and different perspectives for host country policy. For example, it leads to greater consideration of the international competitiveness of the local supply base and emphasizes the importance of international benchmarking. It also focuses attention more strongly upon the location of decision-making, since decisions on foreign sourcing will be commonly made at parent MNE level, as part of global supply chain management strategies; whereas the subsidiary is more likely to be an advocate of local sourcing. From an empirical perspective, information obtained on imported inputs may be more reliable than that on local sourcing because of the difficulty of distinguishing on occasions between indigenous and foreign-owned suppliers; and because some ‘local’ suppliers may simply be distributors of imported inputs. THEORY AND CONCEPTUAL FRAMEWORK The theoretical underpinnings for the subject derive, on the one hand, from concepts relating to the MNE, and, on the other, to perspectives from host country economic development. Although theories pertaining to the MNE have traditionally focused upon the parent group, recent work has increasingly recognised the importance of the subsidiary as the unit of analysis (White and Poynter, 1984; Bartlett and Ghoshal, 1986; Jarillo and Martinez, 1990; Birkinshaw, 1997, 2001). Transaction cost theory is most widely known through the work of Buckley and Casson (1976) and Dunning’s OLI paradigm (1977). Although originally emphasising ownership advantages accruing in the home country of the MNE, it is now recognised that ownership advantages may arise anywhere in the group (Hedlund, 1984; Rugman and Verbeke, 1992). In a similar manner, locational factors in the host country may generate ownership advantages for the subsidiary (that can be internalised by the group). The OLI paradigm is thus useful in understanding knowledge acquisition within the subsidiary, as well as decisions such as the choice of in-house v outsourced supply, and supply chain management locational choices. Much subsidiary level research has, however, inclined towards a Network Perspective on the MNE (Andersson, 1997; Birkinshaw and Hood, 1998; Gupta and Govindarajan, 2000). By this view the subsidiary moves from being highly dependent on headquarters decisions and corporate resources to becoming a node in a network, with links to internal and external actors and greater degrees of freedom. These network partners might include suppliers, customers and sister units within the MNE. This perspective helps explain why some subsidiaries become more important than others; and would suggest that MNE subsidiaries would gradually increase local sourcing, as their interactions with other parties and local competence inc reased. There are growing links between Network Perspectives and the Resource-based View (RBV). The latter suggests that a firm’s unique resources and capabilities can generate competitive advantage (Barney, 1991). Competence development may accrue from interaction with other parties and relationship building (Grant, 1996). Andersson et al. (2001) have further proposed that subsidiaries with critical corporate and market relationships will have higher levels of competence. Other relevant competence-based approaches include the dynamic capabilities view (Teece et al., 1997) and Cantwell’s (1989, 2001) work on technological development. All these notions are relevant to the topic of subsidiary evolution and hence to MNEs’ sourcing decisions. The literature in economic development and economic geography highlighting the positive impacts of input linkages on the local economy has a long history (Hirschman, 1958). The existence of input linkages has been shown to generate positive externalities which lead to the agglomeration of interrelated activities in particular regions (Marshall, 1920; Krugman and Venables, 1995). Such work has been incorporated into the MNE literature through studies of inter-firm linkages between multinationals and host country suppliers (Crone, 2000; Görg and Ruane, 2001), and more recently through the notion of regional clusters and the circumstances under which regions are able to remain as ‘sticky places in slippery space’ (Markusen, 1996). Host economy characteristics are thus critical in competence building within the subsidiary. Cantwell and Janne (1999) have distinguished between ‘higher order’ and ‘lower order’ regions, a characteristic of the former being general technological excellence which may generate substantial spillovers from the local environment for subsidiaries located in such regions. There have been numerous empirical studies on the subject of MNEs and backward linkages, reviewed recently in the World Investment Report 2001 (UNCTAD, 2001). This identified the following influencing factors: investment motives and strategies; technology and market position; affiliate role; affiliate age; mode of establishment; and affiliate size. Host country characteristics, particularly the local supply base, were also shown to have major influences on sourcing patterns. There is thus some concordance between theory and empirical evidence, and Figure 1 attempts to provide an overview of the factors conditioning the choice between local and foreign sourcing. The figure draws upon the Birkinshaw and Hood (1997) organizing framework for subsidiary development in distinguishing MNE parent & subsidiary and host country influences; however, industry characteristics and environment are added as a further conditioning factor in the choice between local and foreign sourcing. FIGURE 1 . Factors Conditioning the Choice between Local and Foreign Sourcing MNE strategy ? Corporate objectives, culture and governance ? Centralization v autonomy ? Globalisation v localization ? Inter-organisational relationships ? In-house/outsourcing/alliance policies Subsidiary strategy and role ? Performance and evaluation criteria ? Management autonomy and entrepreneurship ? Subsidiary value chain ? Subsidiary characteristics: age, size, entry motivation & mode, and importance within MNE group Industry characteristics and environment: ? Product ? Technology ? Competition ? Degree of globalisation/ localization including supplier & customers Host country policy and environment: ? FDI policy ? Cluster strategies/ local sourcing and supplier development policy ? Economic development conditions and policies ? Supplier base and competencies Local v foreign sourcing Major influences on subsidiary operations, including sourcing decisions, derive from the decisions of parent MNEs. Factors influencing parent strategies in the area of sourcing include broad corporate objectives, which themselves are affected by corporate culture and corporate governance. MNE corporate attitudes towards centralisation / decentralisation and globalisation / localisation, as well as inter-organisational networking relationships will also influence sourcing. Similarly, decisions on the choice between in-house production, outsourcing and joint ventures or partnerships will impact upon the potential for local sourcing at subsidiary level. Despite the constraints imposed by their parent company, subsidiaries may have the capacity and capability to influence their own destiny in a variety of areas including local sourcing, and this is reflected in Figure 1. Other sourcing determinants such as subsidiary age, size, and entry mode are also shown. The discussion above highlighted the potential mediating effects of the local environment on subsidiary development especially in innovative milieux (Aydalot, 1986) where strong regional clusters of technological activities exist. Policies to support such cluster development, as well as FDI policies per se (ranging from subsidies to after-care support [Young and Hood, 1994]) will impact upon sourcing decisions. Finally, industry characteristics will also influence the local v foreign sourcing option through, for example technology, levels of competition and the global or multidomestic characteristics of the sector. MNE corporate and subsidiary strategy will impact upon the local v foreign sourcing choice directly, as will both industry characteristics and host country policy and environment. However, the level of local sourcing achieved will also induce government policy initiatives given the importance of backward linkages in economic development. Similarly there are likely to be twoway linkages between industry and host country factors and corporate/subsidiary strategy. LITERATURE REVIEW AND HYPOTHESES The present study aims to provide explanations for the foreign sourcing propensities of multinational (MNE) manufacturing subsidiaries in Portugal, Ireland and the UK. Data collected in the questionnaire survey included a number, although not all, of the variables identified in the conceptual framework above. Host country characteristics have been shown to have major influences on sourcing patterns. Features of importance include the size and level of economic development of the country and the availability of an established and diversified industrial fabric, which will influence the size and capabilities of the local supplier base (Crone, 2000); as well as FDI policies in respect, for example, of incentives, and cluster and supplier development strategies. Conversely, barriers to local sourcing in some host countries (including developed nations), include the absence of indigenous supply capabilities, especially for meeting the new product / process development requirements of MNEs (UNCTAD, 2001). Further common criticisms include the inability of local suppliers to meet the quality / price / delivery specifications of multinationals; the small size of supplier enterprises, and their management and financial resource limitations, meaning an inability (or unwillingness) to invest in long-term partnerships with MNEs. In respect of the present country sample, the expectation is that: Hypothesis 1: Import propensities among multinational affiliates in Portugal and Ireland will be higher than in the UK and Spain. Home country. Historically, studies have indicated that Japanese-owned subsidiaries had a higher propensity to import (Kotabe and Omura, 1989; Dunning, 1986); but UNCTAD (2001) reports evidence that Japanese affiliates increased their local procurement in all host countries during the 1990s (primarily by purchasing from other Japanese inward investors). In the case of EU firms, there is some evidence from Pearce and Papanastassiou (1997) of greater local sourcing propensities. The lack of strong evidence suggests that: Hypothesis 2: No home country effect will be apparent in terms of foreign sourcing propensities. Sector. Included within the sample are MNE subsidiaries within sectors which are commonly viewed as ‘global industries’ (such as automobiles, chemicals, electronics, and pharmaceuticals [Makhija et al., 1997)). The expectation is for significant import propensities in these sectors. By comparison the machinery and especially metal industries are low or intermediate technology with significant localisation features relating to customer requirements and transport costs. MNEs operating in such sectors with mature, non-proprietary technologies will be able to readily identify pools of local suppliers and utilise their bargaining power to ensure competitive prices The expectation is for lower import propensities. The textiles, clothing & footwear industry, also considered, is global in respect of its sales and supply chains, but its low technology characteristics indicate a greater ability to procure many inputs locally. The expectations regarding foreign sourcing are thus uncertain. The literature reviewed in UNCTAD (2001) highlights other influences, however. Thus local sourcing is likely to be lower in capital intensive and continuous process industries, and in higher value-added and technologically advanced sectors. Overall, it is anticipated that: Hypothesis 3(a): High foreign input sourcing propensities will exist in the autos, chemicals, electronics and pharmaceutical sectors. Hypothesis 3(b): Conversely, low foreign input sourcing propensities will exist in the machinery and metals sectors. Hypothesis 3(c): Foreign input sourcing propensities are uncertain in the textiles industry. Age of subsidiary. There is evidence in the literature to indicate that older subsidiaries source more locally and hence import less (McAleese and McDonald, 1978; Turok, 1997; Driffield and Mohd Noor, 1999). This reflects the evolution of the subsidiary itself and the accumulation of knowledge concerning supplier capabilities. Assistance from government organisations in the host region is a further positive influence. Some more recent work has questioned such a simple relationship between age and sourcing, however. Thus Görg and Ruane (2001) found that the extent of linkages between MNEs and Irish indigenous firms was related in a nonlinear fashion to the length of time the firm was located in Ireland, suggesting the existence of other influences. Given that no further evidence is known relating affiliate size and sourcing patterns for this crosscountry sample, it is expected that: Hypothesis 4(a): The age of the subsidiary is negatively associated with import propensity. Date of establishment and the impact of EU integration. Radical environmental changes such as economic integration, or the formation of regional trading blocs, are thought to impact significantly on various aspects of multinationals’ strategies (Cuervo-Cazurra, 1999). This effect is supposed to be more pronounced in small, peripheral economies such as Portugal and Ireland (Tavares and Pearce, 2001; Barry and Bradley, 1997) , where it can be expected that economic integration will lead to considerable new investment, and to strategic changes in existing affiliates towards more interdependent (mainly rationalised) roles. However, the specific impact of economic integration upon foreign sourcing is not clear, and well explored in existing research. On the one hand, there are pervasive factors encouraging stronger MNE affiliate integration within the EU, meaning greater intra-plant trade and higher import propensities; on the other, MNEs may relocate within the EU with a view to access cheaper inputs locally. It is, therefore expected that: Hypothesis 4(b): The effect of economic integration upon the propensity to import is uncertain. Entry mode. Lower import propensities are anticipated in subsidiaries established through mergers and acquisitions, at least in the early years of operation, because of the local sourcing patterns established by the previous indigenous company owners (UNCTAD, 2000, 2001). However, if existing linkages are inefficient, acquired firms may switch to foreign suppliers (UNCTAD, 2000). Joint ventures are also thought to be more prone to source locally their inputs. On balance it is expected that: Hypothesis 5: Subsidiaries established as new greenfield ventures are likely to have higher foreign sourcing propensities than those established by takeover or joint venture. Size of subsidiaries. Evidence on size of affiliate is contained in Schachmann and Fallis (1989) and Barkley and McNamara (1994). Large subsidiaries have been found to procure a lower proportion of inputs locally than smaller ones (although the absolute level of local sourcing may be much higher). Local companies may lack the capacity to supply large volumes. On the other hand, large subsidiaries will have greater capacity to invest in the identification of and relationship-building with local suppliers. It may, therefore, be expected that: Hypothesis 6: Large multinational subsidiaries will import a higher proportion of inputs than small subsidiaries. Motivation. Within the questionnaire survey the classification of motivations for setting up in the host country largely follows Dunning (1993), distinguishing resource-seeking, marketseeking, efficiency-seeking and strategic asset-seeking. However, the resource-seeking motivation is used in a narrower sense to mean the exploitation of locally-bound natural resources; as such, the expectation would be for a lower propensity to import inputs. Aside from this group, expectations are uncertain. For example, market-seekers focusing upon a particular host country could be expected to use local suppliers to reflect local tastes and conditions. However, with the growth of the EU, the market-seeking motivation tends to include firms established to supply wider European markets which, in turn, may be reflected in wider sourcing patterns. Efficiency-seekers may access low costs in the host country market and increase local sourcing as part of this; on the other hand, they may search for cost reduction throughout the world which could lead to increased imports. Again, strategic asset-seeking subsidiaries encompass not only those firms acquiring the assets of foreign firms, but also those establishing subsidiaries in a host country location in order to access local scientific and technological competencies. Incentives for FDI were also considered. They represent a distinct motivation, and it is thought that they were important in justifying the set up of a considerable number of subsidiaries, especially in the Irish case. On the one hand, relevant literature has highlighted the lack of embeddedness of subsidiaries established mainly due to incentives (Driffield, 2001). This would lead to an expectation of a high import propensity. On the other hand, incentives may have conditions attached, for instance regarding local sourcing requirements, which would diminish the propensity to buy inputs and components abroad. This kind of requirement was more frequent in the past than nowadays. The expectation is that: Hypothesis 7(a): Resource-seeking subsidiaries will have a lower propensity to import inputs. Hypothesis 7(b): The foreign sourcing propensity of Market-seeking, Efficiency-seeking and Strategic asset-seeking subsidiaries is uncertain, as well as that of subsidiaries established mainly due to incentives. Subsidiary roles. There is now considerable support for the notion of the MNE as a differentiated network of subsidiaries with distinct strategic roles and scope. In the present study a distinction was made between Miniature Replica, Rationalised Manufacturer and Product Mandate subsidiaries. While numerous typologies of MNE subsidiaries have been developed in the literature, this represents a revised version of the categorisation originally proposed in the pioneering work of White and Poynter (1984); similar three-way typologies have been applied in other empirical studies (Young et al., 1988; Crone and Roper, 2001; Pearce, 2001). The issue of subsidiary roles has taken on increasing significance from a policy perspective because of the presumed economic benefits associated with more autonomous subsidiaries (Young et al., 1994); but there is relatively little information on the independent effect on sourcing of subsidiary type per se. The local market-oriented Miniature Replica should have a lower propensity to import, given its focus on the usually less demanding domestic market and perhaps for reasons of goodwill. This effect may be moderated by entry into the EU and the operation of the EU’s Single Market Programme which provides increased opportunities for sourcing internationally as discussed above. It would be expected that the Rationalised Manufacturer would have a high import propensity, since its position within an integrated MNE group would suggest high levels of intrafirm trade (exports and imports). The Product Mandate subsidiary has autonomy and significant resources for development, production and marketing. Since the company has local R&D and understands the local market this may encourage local sourcing (Holm and Pedersen, 2000). However, the company’s network relationships and its focus on competitive performance would suggest a significant import propensity, thereby leading to an uncertain foreign sourcing propensity overall. The expectations are thus that: Hypothesis 8(a): Miniature Replica subsidiaries will have a low propensity to import inputs. Hypothesis 8(b): Rationalised Manufacturers will have a high foreign sourcing propensity. Hypothesis 8(c): The foreign sourcing propensity of Product Mandate subsidiaries is uncertain . DATA AND METHODOLOGY Data The following empirical analysis is based on the results of a survey of foreign manufacturing subsidiaries operating in four EU economies (Portugal, Spain, Ireland and the UK). A focus on the EU was chosen since it represents the most accomplished and developed regional integration experiment (Thomsen and Woolcock, 1993), hence allowing for corporate strategies to have developed as a reaction to key environmental changes which occurred in the last decades, such as the Single Market Programme (Young, Hood and Dunlop, 1991). The choice of these four countries reflects an interest in exploring cross-country comparisons and in confronting findings on both large and small EU economies. A postal questionnaire was deemed the most appropriate data collection method given the intention of gathering information on a large number of subsidiaries in order to undertake rigorous quantitative analysis and to allow results to be generalised (Taggart, 1999). The survey instrument was pre-tested by interviews with subsidiaries’ top managers and several academics were also consulted, which led to some changes to the wording of the original questionnaire. The pre-tested and amended questionnaire was then mailed to the managing director of each selected subsidiary. A total of 1092 questionnaires were sent. The selection of the subsidiaries was based on various publications listing the largest MNEs in each country (e.g. Dun and Bradstreet, Who’s Who, Jordan and Kompass), and on directories provided by Chambers of Commerce and Industry, Embassies and various Ministries and Government agencies. Distinct sources were compared to increase reliability. Only majority-owned manufacturing subsidiaries were chosen. The survey was administered in 1999, and for each country the survey was sent twice, the second time four weeks after the first. 265 replies were received (a response rate of 24.3 per cent, which compares favourably with that obtained in similar studies [Harzing, 1997]), and a total of 233 valid replies were considered (as 32 replies were deemed incomplete or incorrectly classified). Appendix 1 presents the descriptive data relating to the subsidiaries in the sample, distinguishing proportions of firms in the categories 0-50.0 per cent and 50.1-100 per cent imported inputs. The mean level of imported inputs was 55 per cent, with median and mode values of 60 per cent and 90 per cent respectively. The evidence shown in Appendix 1 lends preliminary support to some key hypotheses suggested above. This aspect is further discussed when analysing subsequent empirical modelling findings. Method Reflecting the central aim of this study, a formal test of the determinants of subsidiaries’ sourcing patterns was undertaken, based on a recent and unique database resulting from the cross-country survey described above. An ordered probit model was used, given the nature of the dependent variable (consisting of a finite set of categories with a compelling ordered structure, as explained in the next sub-section). This type of statistical method models the probability of an event (Liao, 1994), in this case the propensity of a given subsidiary to import a certain percentage of inputs/components. Since the dependent variable is categorical, conventional regression methods were inappropriate (Machin and Stewart, 1990; Stewart, 1991), thereby leading to the choice of a limited dependent, qualitative response method. Further considerations about this type of methodology can be found in Amemiya (1981, 1985), McFadden (1984) and Greene (1997).