Regional sourcing patterns of multinational

0 downloads 0 Views 100KB Size Report
conduit for knowledge transfers (Crone and Roper, 2001). But the level of regional sourcing by MNEs is also of wider interest in the field of regional development ...
The Determinants of Regional Sourcing by Multinational Manufacturing Firms

September 2002

MIKE CRONE* and H. DOUG WATTS** *Northern Ireland Economic Research Centre, 22-24 Mount Charles, Belfast, Northern Ireland, BT7 1NZ Tel: +44 (0) 28 9026 1820 Fax: +44 (0) 28 9033 0054 e-mail: [email protected] ** Department of Geography, University of Sheffield, Sheffield, S10 2TN, UK. Tel: +44 (0) 114 222 7945 Fax: +44 (0) 114 279 7912 e-mail: [email protected]

Thanks are due to the managers who participated in the survey of MNE plants in Yorkshire and the Humber. The survey was conducted while Mike Crone was supported by an ESRC studentship at the University of Sheffield.

ABSTRACT.

The paper explores the factors influencing the extent of regional sourcing by multinational manufacturing firms using data collected by interview from 50 foreign- and UK-owned plants in Yorkshire and the Humber. Higher levels of regional sourcing are observed where certain demand characteristics coincide with strengths in the regional supply base. A ‘corporate filter’ reduces regional sourcing but there are no differences in regional sourcing between foreign- and UK-owned multinationals. The findings identify barriers to policy aspirations of increased regional sourcing and emphasise that any policy initiatives should target domestic, as well as foreign, MNE plants.

1. Introduction

Large firms continue to play an important role in manufacturing in many urban and regional economies despite a policy focus on small and medium sized enterprises. In 2000, large enterprises (defined by the EU as those with 250 or more employees) accounted for two thirds of the turnover and half the employment in the UK manufacturing sector (Department of Trade and Industry, 2001). Many of these large enterprises are multinational firms with manufacturing operations in more than one country. In the UK, analyses of multinational enterprises (MNEs) within regional economies have tended to focus on foreign-owned MNEs, in spite of the fact that a significant proportion of the UK manufacturing sector is controlled by domestic MNEs. Interest in the regional impacts of MNEs in the manufacturing sector reflects a concern with both their direct impacts (in terms of the number and type of jobs) and their indirect impacts, which can take a number of forms. A multinational plant may, for example, influence wage levels, supply inputs to regional firms, provide a channel for diffusion of new technologies and labour practices and, through regional sourcing, provide a market for regional service and manufacturing firms.

The primary interest in regional sourcing by multinational firms is that regional purchases stimulate the host economy, creating jobs among regional suppliers and acting as a conduit for knowledge transfers (Crone and Roper, 2001). But the level of regional sourcing by MNEs is also of wider interest in the field of regional development because backward linkages are an important indicator of embeddedness in the host economy (Dicken et al, 1994; Phelps, 1997). Traditionally there has been a concern that, in development terms, a multinational plant can be a ‘cathedral in a desert’ (Morris, 1992) with few links to the economy of the host region (e.g. Firn, 1975). This lack of embeddedness is a cause for concern as it is thought to reduce the spin-off benefits to the host region and make it easier for the plant to be shut down in any corporate restructuring.

Over the last decade the issue of embeddedness has been revisited in the context of debates about the regional development implications of ‘globalisation’ (Ohmae, 1995; Yeung, 1998). Notably, some authors have suggested that the imperatives of globalisation will tend to encourage more international sourcing patterns, resulting in lower levels of

1

regional sourcing and embeddedness. Turok (1997), for example, observed the proportion of local sourcing by foreign multinationals in the Scottish electronics industry fell in the 1980s. However, other authors have argued that, despite the pressures of globalisation, some recent developments in corporate organisation may actually favour a localisation of sourcing patterns around major buyers, and greater embeddedness of multinationals in host regions (e.g. Morris 1992; Cooke and Morgan, 1993; Amin et al, 1994). In parallel to these debates, and possibly because of them, the issue of regional sourcing has also moved up the policy agenda in the UK (e.g. Department of Trade and Industry, 1994; Williams, 1994; Welsh Development Agency, 1996; Izushi, 1999). There is now a major concern with ‘tying the firm to the region’ (Phelps, Lovering and Morgan, 1998). It is this context that provides the rationale for a detailed empirical investigation of the issue of regional sourcing by multinational manufacturing firms.

Regional sourcing by MNE plants includes, conventionally, the purchases of material and component inputs and services. The material and component inputs, which typically account for four or five times as much expenditure as service inputs (PA Cambridge Economic Consultants, 1995), are the focus of the present paper. The data are drawn from a sample of large MNE-owned plants in the Yorkshire and the Humber region of the UK in the mid-1990s. The study is distinctive in a number of ways. It examines both domestic and foreign multinationals. It examines the sourcing behaviour of MNEs in an older industrial region where recent flows of inward investment (both domestic and foreign) have been lower than in other parts of the UK such as Wales, Scotland, and North East England. Conceptually, perhaps its major contribution is a detailed analysis of the influence of corporate context on the level of regional sourcing by MNE plants. In its approach the study is essentially cross-sectional, with no attempt being made to trace changes in regional sourcing behaviour over time (but see Crone and Watts, 2002).

The discussion is in four parts. It begins with a theoretical and conceptual discussion of the factors that may influence the level of regional sourcing by MNE-owned manufacturing plants and this is followed by a brief review of the methodology. The third substantive section describes the overall pattern of regional sourcing of material and component inputs by MNE-owned plants and explores the influence of various factors on the level of regional sourcing by MNE plants in Yorkshire and the Humber. This is done

2

using a framework that takes account of the demands of individual plants, the supply potential of the host region, and the ways in which plant level outcomes are modified by a corporate filter. The concluding section considers the implications of the findings, including the lessons for MNE-focused regional policy initiatives.

3

2. Regional Sourcing by Multinationals

A potential framework for the analysis of the level of regional sourcing by MNE plants is provided in Figure 1. It is based on the understanding of the behaviour of MNEs and their

Figure 1. Analytical framework. DEMAND Industrial sector Product complexity Product standardisation Method of production Plant size Plant age

CORPORATE FILTER Nationality of parent firm Geographical extent of parent firm Corporate sourcing strategies Intra-firm sourcing Centralised buying Single/dual sourcing JIT

SUPPLY Sectoral mix Size of region Size of plants Competitiveness

regional sourcing patterns which has been built up over the last three decades. The framework focuses on the nature and volume of demand for material and component inputs at individual MNE plants and the supply potential of the host region relative to other locations. A third element, which intervenes between the nature of demand and regional supply potential, is the corporate context in which the plant operates. In essence this is a straightforward demand and supply model with a ‘corporate filter’ which modifies the level of regional sourcing. 4

2.1 Demand

Research over the past thirty years has identified a number of factors influencing a plant’s demand for material and component inputs from regional sources. These have included industrial sector and closely related factors such as the degree of product complexity, the extent to which the products are standardised and the method of production used by the plant.

Industry sector has often been found to be a predictor of variations in regional sourcing (e.g. Lever, 1974; Reid, 1995). There has, however, been a lack of consistency in the findings for specific industrial sectors and few authors have managed to derive any generally applicable conclusions as to which sectors should be expected to have high or low levels of regional sourcing (Hoare, 1985). In part this reflects lack of clarity concerning what it is about certain industries that influence their sourcing patterns. It is therefore important to look closely at both product and production technology.

The technological complexity of a product can be usefully measured by manager selfratings (such as high, medium or low technology). Arguably plants producing low-tech products will use relatively simple inputs, which are more likely to be available from regional suppliers because they are established technologies with low barriers to market entry (Phelps, 1996). In contrast, specialised inputs are less likely to be sourced from regional suppliers because fewer suppliers will possess the technical capability to manufacture complex inputs, so there will be fewer regional sourcing opportunities. Also, the higher value-to-weight ratio of technologically complex inputs means they can bear the higher transportation costs associated with purchasing from distant suppliers (O’Farrell and O’Loughlin, 1981). These arguments were endorsed by evidence from a study of US-owned branch plants in Ontario, Canada (Britton, 1976). More basic inputs tended to be sourced in Ontario whereas more complex or specialist inputs tended to be sourced from the US parent firm or from other suppliers in the US.

An alternative way of conceptualising product characteristics is to distinguish between standardised and customer-specific (unstandardised) products. It has been argued that standardised inputs, associated with standardised products, can be readily obtained from

5

distant suppliers once the initial deal has been struck as only order size needs to be discussed. Conversely, close proximity to suppliers is thought to be advantageous for unstandardised inputs because of the need to frequently renegotiate supply contracts and amend specifications (Amin and Malmberg, 1992). This argument was endorsed by a study of auto parts manufacturers in the USA (Glasmeier and McCluskey, 1987). Standardised inputs tended to be sourced over longer distances from peripheral region branch plants whereas unstandardised inputs tended to be procured from nearby specialist suppliers in the traditional mid-west heartland of US automobile production. A similar argument can be applied to the method of production, which is often closely related to the degree of product standardisation in that customer-specific products tend to be associated with unit or small-batch production whereas standardised products tend to be produced by large-batch or mass production (Scott, 1983). There is some evidence that unit or smallbatch production plants display higher levels of regional sourcing than plants using largebatch or mass production methods (Marshall, 1979) although this relationship has not always been proven (Reid, 1995).

Whilst the nature of demand will primarily reflect the type of product, the volume of demand for material and component inputs will closely reflect the size of the buying plant. For the most part there seems fairly strong evidence that larger plants have a lower dependence on the host region as a source of inputs. This is found in the more traditional studies (Lever, 1974; Britton, 1976; Marshall, 1979) as well as in more recent work (Barkley and McNamara, 1994; Gorg and Ruane, 2001). The suggested explanation is that larger plants may have lower levels of regional sourcing because regional suppliers lack the scale required to supply the larger volumes of inputs required. However, the plant size effect has not been identified by all studies (O’Farrell and O’Loughlin, 1981; O’hUallachain, 1984; Reid, 1995).

Even in a cross-sectional analysis it is important to recognise the possibility that regional ties may be strengthened or weakened over time, and also that present sourcing patterns may reflect inertia and established sourcing behaviour. A number of studies of foreign MNEs have observed that the level of regional sourcing is positively related to the age of the plant (e.g. O’hUallachain, 1984; Gorg and Ruane, 2001). The basic rationale here is that there is a ‘learning curve’ for both newly-established MNE plants and regional

6

suppliers whereby they become more aware of one another’s needs and capabilities over time (O’Farrell and O’Loughlin, 1981). However, as Phelps (1997) has argued, an observed ‘age effect’ may also reflect inertia in sourcing patterns. For example, plants established in more recent decades may be more ‘in tune’ with contemporary (perhaps more internationalised) patterns of sourcing than plants of older ‘vintage’, which may be slow to adjust new sourcing imperatives and retain stronger ties to the regional economy. Linked to the age factor is the view that MNEs which enter a region by acquisition will have a higher level of regional sourcing than MNEs who establish a new greenfield plant because a regional purchasing network is already established in the former case (e.g. Barkley and McNamara, 1994). In contrast newly established plants may initially rely on familiar sources of inputs (other group plants or suppliers in the home country).

2.2 Supply

The term regional supply potential refers to the ability of the host region to supply appropriate inputs to MNE plants. Although this factor has been acknowledged in conceptual discussions of local sourcing by MNEs (e.g. Dunning, 1993; Dicken, 1998) it has received surprisingly little attention in empirical studies. Regional supply potential can be seen as a product of a number of elements including the regional industrial structure (i.e. sectoral mix) and regional supply capacity, which itself reflects the size of the region (i.e. number of plants) and the size of individual plants.

The influence of the regional industrial structure on the level of regional sourcing was recognised many years ago by Gilmour (1974). In many instances, there were no Canadian producers of the complex inputs required by US-owned subsidiaries in Canada. More recently, Kirchner (2000, 608) has reported that some German firms in north-east England would ‘like to source more regionally’ but the materials or equipment were not available. In a more wide ranging study Barkley and McNamara (1994) attempted to take account of sectoral variations in regional supply opportunities in their analysis of the sourcing patterns of branch plants in Georgia and South Carolina. They found that plants in industrial sectors with an established regional presence, which were assumed to have a more developed regional supply base, displayed higher levels of regional (intra-state) sourcing than plants in industrial sectors with a more limited regional presence. Thus it is

7

implied that the level of regional sourcing will be partly determined by the extent to which there is an appropriate regional supply base.

If regional supply capacity is partly determined by region size, with ceteris paribus a larger region being more able to supply both a wider range of inputs and greater quantities of inputs, we might expect to observe a higher level of regional sourcing in larger regions. This assertion is supported by Twomey and Tomkins (1996a, 1996b). They demonstrated, using UK input-output data, that smaller GB regions have greater deficiencies in their supply bases than larger regions. However, region size was dismissed as an influence on the level of regional sourcing in a study of Japanese plants in the United States (Reid, 1995). In any case, it is not possible to verify possible region size effects in a study of a single region such as this (but see Crone, 2002, for further discussion of this issue in respect of the UK regions). A further constraint on regional supply capacity may operate through the plant size mix of the region. Smaller regional supplier plants may not have the capacity individually to provide inputs in the quantities demanded by MNE plants.

Another dimension of regional supply potential that should be considered in an analysis of regional sourcing behaviour is the competitiveness of regional suppliers relative to suppliers in other locations. This may lead to MNEs sourcing from outside the host region, even where there appear to be appropriate regional supply opportunities. The study by Kirchner (2000) of German firms (noted above) observed that in some cases potential north-east suppliers were too expensive or produced goods of poorer quality that suppliers elsewhere. A final related, and particularly perceptive, point was raised by Hoare (1985) who noted that decision makers’ perceptions of the availability, suitability and competitiveness of regional suppliers may be as important as the actual facts in determining the level of regional sourcing.

2.3 The corporate filter

The demand and supply factors discussed above might apply to any plant, regardless of ownership. An MNE plant, however, is set within a wider corporate context, which may exert an additional influence on its sourcing patterns. Specifically, each MNE plant is set within its own distinctive corporate geography. MNEs have their head offices in different

8

countries, vary in the geographical extent of their international production activities and vary in their corporate sourcing strategies. These potential influences on the level of regional sourcing are conceptualised as a ‘corporate filter’ in the analytical framework that underpins this paper (Figure 1).

In the 1970s, a number of studies in the UK, Ireland and Canada identified lower levels of regional sourcing among foreign-owned plants than among indigenous firms (Britton, 1976; Stewart, 1976). However, in most cases, the indigenous firms in question were not multinationals themselves. Nevertheless, it could be argued that we might expect higher levels of regional sourcing among UK-owned MNEs than foreign-owned MNEs in any given UK region. UK MNEs might be more familiar with regional suppliers, since the centre of decision making is closer to these suppliers, and UK MNEs may show some favouritism towards home country suppliers, including some in the host region. These arguments are open to debate, however, as such parochialism is surely unsustainable in an increasingly competitive, and increasingly global, marketplace. Significantly, more recent studies by Phelps (1997) and Barkley and McNamara (1994) have found no difference in levels of regional sourcing between domestic and foreign-owned MNE plants in Wales and Georgia/South Carolina respectively. Further arguments also draw distinctions between the activities of the MNEs of different nationalities because of the different histories and culture within which the firms originate (Porter, 1990, 16-21; Ruigrok and Van Tulder, 1995). Dicken, Forsgren and Malmberg (1994, 33-35) have made similar assertions in the geographical literature, suggesting different nationalities may have different dispositions towards regional sourcing.

In addition to nationality of ownership, a geographically extensive network of international operations will widen the firm’s knowledge of worldwide sourcing opportunities and possibly reduce levels of regional sourcing. Compared with purely domestic firms, Dunning (1993, 452) argues, MNEs have ‘better information about world wide prices and quality of parts and components’. Since MNEs vary widely in the geographic spread of their operations, we might then expect plants owned by geographically extensive organisations to have lower levels of regional sourcing than plants owned by less geographically extensive organisations. This is because of better information about, and access to, a wider range of potential non-regional sources.

9

The remaining corporate filter variables arise from corporate sourcing strategies. Indeed, the role of supply chain management strategies have been highlighted in a number of studies but these have tended to draw primarily on the automotive sector (Wells and Rawlinson, 1992; Sadler, 1997). The strategies MNEs adopt in relation to sourcing of materials and components can vary markedly and can be assessed in a number of different ways.

Some firms attempt to internalise their sourcing activities and where this occurs plants may be strongly tied into corporate production systems. In these cases the plant may be characterised by strong backward and forward flows with other group plants. A number of earlier studies have demonstrated that plants which are closely integrated into corporate production systems have low levels of regional sourcing (Britton, 1976; Marshall, 1979).

Some firms will centralise some or all of their input purchasing at one location so as to achieve economies of bulk buying (Dunning, 1993) and many plants will have little autonomy in making purchasing decisions. In such circumstances, individual MNE plants may be prevented from regional sourcing as the optimal supplier for the firm may not be the same as the optimal supplier for a single plant. Also, the larger volumes required by pooled purchasing might militate against regional sourcing because there is less chance that a sufficiently large supplier will be found regionally. Another variant of centralised buying is where the parent firm operates a list of approved or preferred suppliers. This strategy may result in lower regional sourcing because smaller or weaker regional suppliers may struggle to satisfy the stringent requirements of the approved supplier list.

There are also two further strategic considerations. Some MNEs have focused on single sourcing strategies where the building of quality relationships with one or at the most two suppliers of a particular commodity offset the loss of competition between suppliers. A number of writers have suggested single sourcing strategies are becoming more important (Morris and Imrie, 1993; Imrie and Morris 1992) and single sourcing may well work against regional suppliers except where the a regional plant is selected as the single source (Phelps, 1993). Finally, in MNEs where the strategic emphasis is on just in time

10

(JIT) systems with strict controls over on-site inventory levels and, as a result frequent deliveries, there may be a tendency to favour regional suppliers (Reid, 1995, 346)

Overall, the evidence available suggests that regional sourcing will be lower among plants belonging to MNEs with geographically extensive operations, with strong intracorporate flows, centralised or pooled buying systems and a strategy of single sourcing. In contrast, the use of JIT or ‘JIT-like’ sourcing may increase the propensity to source regionally. The likely impact of the nationality of the parent firm is uncertain.

In sum, the extent of regional sourcing appears to be related to the availability of regional supply and the level of demand although the level of demand is modified through a corporate filter. In the remainder of the paper we explore the influence of demand, supply and the corporate filter variables on the levels of regional sourcing amongst the large plants of UK-owned and foreign-owned MNEs in Yorkshire and the Humber.

11

3. Methodology

The data upon which this study is based are drawn primarily from a sample of MNEowned plants in Yorkshire and the Humber. It is an area characterised as an older industrial region of the UK occupying an intermediate location between the more prosperous southern parts of the country and the more peripheral areas (such as Scotland, Wales and Northern Ireland) in which significant areas have been eligible for regional aid. The region has just over 0.4 million employees in manufacturing making it the fourth largest UK region in terms of manufacturing employment. Its dependence on manufacturing is above the national average (22 per cent of employees compared with a UK average of 18 per cent).

Enterprise data are difficult to obtain at regional level but an indication of the importance of large firms can be provided by exploring the importance of large plants. The region had 437 large manufacturing plants (with 200 or more employees) in the mid-1990s representing just around half of the manufacturing employment in the region - a proportion very close to the national average (Business Monitor PA1003, 1993). In the absence of any comprehensive data set on MNE plants a database of large plants by ownership was built up from local authority records and cross-checked against national and local directories. Overall, 413 large plants were identified (95 per cent of those listed in the Business Monitor) of which 224 plants (54 per cent) were owned by MNEs. This group was made up of both UK-owned MNEs (31 per cent) and foreign-owned MNEs (23 per cent).

The data discussed in the paper were collected through face to face interviews with a senior manager at each MNE plant, using a standardised questionnaire format. Before sampling, the twelve plants with 1000 or more employees were excluded since the data collection methods were thought to be inappropriate to handle the complexity of what were often multi-activity sites. From the remaining 212 plants a random stratified sample was drawn to provide a distribution of respondents across the region. The overall response rate was 62 per cent and the representativeness of the 50 respondents by sector and nationality of ownership was examined using chi-squared for goodness of fit. Although there was a small over-representation of overseas-owned plants in the sample

12

there was no significant difference between the observed and expected frequencies and thus the sample can be argued to be broadly representative of the total population, at least in terms of nationality and industrial sector.

Non-parametric statistical techniques are employed in the quantitative analysis including the Spearman rank-order test, ANOVA test, Mann-Whitney U test, Kruskal-Wallis test, chi-squared test for two-by-two contingency tables and Fisher’s exact test, where chisquared is inappropriate due to the small number of observations. In the tests, various determinants (potential explanatory variables) constitute independent variables and two alternative measures of the level of regional sourcing constitute the dependent variables. The first of these dependent variables is the percentage of material and component inputs, by value, obtained from suppliers in Yorkshire and the Humber (also referred to as the ‘regional percentage’). The second measure is the number of ‘key suppliers’ of material and component inputs in Yorkshire and the Humber (‘regional key suppliers’). The determinants of the regional percentage (ratio data) can be examined using Spearman, ANOVA, Mann-Whitney and Kruskal-Wallis tests whereas the determinants of the number of regional key suppliers (ordinal data) can only be examined using chi-squared or Fisher’s exact tests. Although the tests examine only bivariate statistical relationships, some attempt is made to take account of possible interaction effects although this difficult due to the small size of the sample. In all of the tests, four levels of significance are employed (p