European Journal of Purchasing & Supply Management 6 (2000) 207}215

Supply strategy and network e!ects * purchasing behaviour in the construction industry
Anna Dubois*, Lars-Erik Gadde
Chalmers University of Technology, Department of Industrial Marketing, S-412 96 Gothenburg, Sweden Accepted 6 April 2000

Abstract Partnering with suppliers and networking are increasingly used as means to improve company performance. This paper explores the occurrence of network e!ects in the construction industry. Bene"ts from network e!ects arise when "rms adapt to one another in terms of technical solutions, logistics or administrative routines. The study "nds such e!ects to be unusual in the construction industry. The main reasons for the absence of adaptation are found to be the current focus on the e$ciency of individual projects and the competitive tendering procedures used. It is concluded that these characteristics are having a hampering e!ect on both e$ciency and innovation in the industry today. 2000 Elsevier Science Ltd. All rights reserved.
Keywords: Network e!ects; Construction e$ciency; Purchasing behaviour

1. Introduction The supply side is of increasing strategic importance in most companies today. This change re#ects a new awareness of the bene"ts to be gained from better utilising resources from beyond the boundaries of "rms. The advantages of partnering with suppliers have been particularly emphasised (see e.g. Carlisle and Parker, 1989). Reaping partnering bene"ts requires a shift in atmosphere and behaviour from traditional arm's length relationships. For example, short-term focus on e$ciency in the individual transaction needs to be replaced with longer-term-oriented relational exchanges based on close buyer}seller relationships. The bene"ts of these constellations are obtained partly from better use of individual suppliers, partly from network e!ects derived out of the combined e!orts of a number of suppliers (Hakansson and Snehota, 1995). The construction industry is heavily dependent on subcontractors and suppliers of building materials. One feature of this industry is `the practice of subcontracting portions of a project to special trade contractors by primary contractorsa (Eccles, 1981a). For example, in
* Corresponding author. Tel.: #46-31-772-11-96. E-mail addresses: [email protected] (A. Dubois), laga@mot. chalmers.se (L.-E. Gadde).

Swedish construction the level of subcontracting has increased substantially over time. The three major Swedish primary contractors are good illustrations of what has been identi"ed as `quasi-"rmsa (Eccles, 1981b), a trend that has increased substantially in recent years. Purchased materials and services account for about 75% of total costs in these "rms. This makes construction ahead of most other industries in terms of outsourcing. However, in other industries increasing specialisation have made buying "rms able to improve e$ciency and e!ectiveness by changing the nature of relationships with suppliers. The aim of this paper is to explore how "rms within the construction industry take advantage of their opportunities to make use of external resources through co-operation.

2. Supply strategy and network e4ects Studies of customer}supplier collaboration have shown that major bene"ts may be achieved when "rms make adaptations to one another (Hines, 1994; Spekman et al., 1999). Three main types of adaptation may be distinguished (Gadde and Hakansson, 1994). First, tech nical adaptations connect the production operations of supplier and customer. These adaptations have to do with the technical content of the products exchanged.

0969-7012/00/$ - see front matter 2000 Elsevier Science Ltd. All rights reserved. PII: S 0 9 6 9 - 7 0 1 2 ( 0 0 ) 0 0 0 1 6 - 2

208

A. Dubois, L.-E. Gadde / European Journal of Purchasing & Supply Management 6 (2000) 207}215

A customer may "nd it appropriate to ask a supplier to develop a product that will be a perfect "t for the needs of the buying "rm, either in terms of technical content or physical features. Another form of technical adaptation is related to the material #ow between supplier and customer. Sophisticated logistics systems, such as just-intime delivery, may enhance e$ciency of operations. The second type of adaptation is in administrative routines. Business transactions are characterised by a great deal of information exchange, regarding inquiries, tenders, orders, delivery noti"cations, invoices, etc. Adaptations in terms of integrated information systems are intended to improve the e$ciency of administrative operations and bureaucracy. The third main type of adaptation is knowledge-based. In close, long-term, relationships customer and supplier develop considerable knowledge about one another's operations. In well-developed partnerships, skills on both sides tend to be connected and not easily separable. These mutual adaptations tend to bring the "rms together and have a particular signi"cance for joint e!orts in technical development. Hakansson and Snehota (1995) present a framework for analysing the characteristics and consequences of adaptations in buyer}seller relationships. The most important dimension of a relationship is its substance. The substance stems from the adaptations undertaken. There are three di!erent kinds of substance that connect the parties in a relationship. The "rst has to do with the linking of the activities of the "rms. Activity links connect the production, logistics and administrative operations of the companies. The best known examples of the bene"ts of activity linking concern the integrated supply-chains in the automotive industry (Lamming, 1993; Hines, 1994). The second substance type relates to the resource ties between the companies. The relationship ties the resource elements on both sides. These resources may be physical facilities, such as machines and equipment, or human knowledge. Resource ties create substance that promotes joint technical development. Bene"ts of supplier involvement in product development have appeared in many industries (see, for example, Wynstra, 1998). The third type of substance is created through interaction between people. Actor bonds are important determinants of trust and commitment, which are, in turn, prerequisites for investments in activity links and resource ties. The substance of a relationship also has an economic aspect. Adaptations are investments that can be used for improving e$ciency in day-to-day operations, as well as for promoting development. Long-term relationships make successive re"nement of activities possible. Adaptations thus contribute to enhanced performance. These adaptation e!ects make close customer-supplier relationships important in themselves. Furthermore, they provide links to other relationships, which can, in turn, produce even greater bene"ts. Therefore it is relevant to consider industrial systems as networks of connected

relationships (Cook and Emerson, 1978). To an increasing extent, buying companies encourage collaboration not only with suppliers but also among them. The notion of supply networks is very much in vogue today. Longterm interaction with a network of suppliers should make it possible for a customer to stimulate mutual adaptations and thus network e!ects bene"cial to the performance of the whole system. Eccles (1981b) observed that primary contractor}subcontractor relationships in the US construction industry are fairly stable over time. These relationships are described as `recurrent series of transactions with a small number of subcontractorsa (Eccles, p. 351). In other industries these conditions have been the breeding ground of network e!ects. The "rst research issue is concerned with the kind of relationships there are between primary contractors and their suppliers. The second deals with network e!ects achieved through adaptations in the supply network of a primary contractor. Researching these issues should provide fundamental insights about e$ciency and e!ectiveness in the construction industry and increase our general knowledge about which network e!ects can be attained by applying speci"c supply strategies. As explained by Cox and Thompson (1997), most academic work in these areas has been concerned with `widgeta production within controlled factory environments `where the supply of goods is merely a repeat process o! a production/assembly linea (p. 128). Construction work has other characteristics. Therefore, the results of the study add to our knowledge of network e!ects in general.

3. Methodology and data-set The main empirical basis of this study has been developed from a major case study examining a combined new construction and renovation project. The construction site was located in central Gothenburg (Sweden), which implied some special requirements in terms of logistics and physical distribution. For various reasons the project was under severe time pressure, which a!ected some of the important decisions. This study can be characterized as a retrospective case study covering a four-year period from the "rst planning stages through completion of the building. Most interviews were undertaken in the "nal stage. The "rst part of the data set is an extensive mapping of the suppliers, installation contractors, consultants, and other actors involved in the project. This was carried out in co-operation with representatives of the primary contractor (mainly the project leader and purchasing sta!). The total cost of the project amounted to a sum corresponding to USD 10 million. All in all, the primary contractor used 14 subcontractors and 10 material suppliers. The second part of the empirical data consists of subcases where the roles of "ve main suppliers are analysed.

A. Dubois, L.-E. Gadde / European Journal of Purchasing & Supply Management 6 (2000) 207}215 Table 1 Characteristics of the products studied in the sub-cases Products: Degree of product adaptation High None (standardised) None (standardised) Mix of standardised and adapted products High Contacts with material producer Direct and substantial Direct and through distributor Direct and through distributor Direct and through distributor No Supplier active on site Yes No No Marginal Yes Buyer}seller interaction High Low Low Average High

209

Co-ordination with other suppliers Substantial Low Low Low Substantial

Frames Plasterboards Insulation Windows Electrical installations

Grasping the complexity of the supply processes in the construction industry required application of deep-probing methodology. The cases provide in-depth analysis of the supply chain for "ve selected inputs (the frame, insulation, plasterboards, windows and electrical installation). These sub-cases were developed through interviews with material producers, distributors, specialised contractors and the material producers' suppliers. The materials were selected to represent variation in a number of respects, as illustrated in Table 1. Together the "ve supply chains accounted for 25% of the total costs. Data collection was based mainly on personal interviews. A total of 27 interviews were made with 29 informants, representing 11 companies. The interviews were focused mainly on the particular construction project. However, the interviews also captured similarities and dissimilarities of the project in comparison with other projects in which the suppliers was, or had been, involved. This approach made it possible to discuss and evaluate di!erent ways of organising supply chains and analyse the potential network e!ects. The interviews contained a mixture of structured and semi-structured questions and responses. A detailed, standardised questionnaire was used to collect information concerning the products and services exchanged as well as the adaptations undertaken among "rms. The semi-structured questions aimed at revealing more qualitative and subtle opinions concerning the connection between the "rms' purchasing behaviour on the one hand and the corresponding network e!ects on the other. The scope of this paper makes it impossible to present detailed information from the database and the cases. Instead, general "ndings from the study as a whole have been condensed into four di!erent themes. The "rst theme deals with standardisation and adaptation in construction materials supply. The level of standardisation of activities and resources, in turn, a!ects the roles of the actors, which is the second theme. An actor in a network can either be specialised or broad in terms of activity scope and resource collection. The mixture of the actor

structure impacts on what kind of network substance can be generated. The third theme relates to the consequences of one of the most fundamental organisational characteristics in construction: a construction project is a temporary organisation. It can also be regarded as a temporary network within the larger permanent network of the construction industry as a whole. The temporary network thus activates the permanent network in one way or another in every construction project. The way it is activated has a profound impact on the potential network e!ects. A particular aspect in this respect is the nature of the exchange processes applied. As mentioned, two main types of exchange can be distinguished * transactional and relational. The fourth theme of the discussion concerns conditions for substance development depending on the type of exchange implemented.

4. Theme 1: Standardisation versus adaptation The basic research issue was to analyse the occurrence of network e!ects. These e!ects arise from the development of network substances that stem, in turn, from adaptations in the relationships. Adaptations may enhance e$ciency in relationship performance. On the other hand, adaptations may lead to the kind of interdependencies some "rms try to avoid. Furthermore, adaptations are always costly * they require investments of various kinds. For this reason a customer has to make a choice between adapted (customised) solutions and standardised ones. One main "nding from the study is that supply of building materials is primarily characterised by exchange of standardised products. In this project only a few customised solutions were used, the most obvious being the frame, which is more or less unique for every construction project. However, even the frame is composed of di!erent combinations of standardised components. This low degree of customisation seems to apply to construction projects in general. It is quite unusual that material

210

A. Dubois, L.-E. Gadde / European Journal of Purchasing & Supply Management 6 (2000) 207}215

producers develop products for particular construction projects or a speci"c primary contractor. Cox and Thompson (1997) mention some examples of adapted solutions but consider construction to be `inherently a site-speci"c &project-based' activitya (Cox and Thompson, 1997, p. 128). The strong focus on the project and its economy entails a rather short-term perspective emphasising competitive bidding as the main tool in supplier evaluation. Consequently, customer}supplier relationships in construction are generally of the arms-length type rather than being partnerships. This "nding is in accordance with Cox and Thompson (1997) who concluded that competitive tendering assures that sub-contracting is `procured to the lowest-price supplier with little or no guarantee (or even incentive) to future worka (Cox and Thompson, 1997, p. 129). Making use of the price mechanism in this way requires that the buying "rm refrains from adaptations to individual vendors and relies on standardised solutions. At "rst glance this standardisation may seem confusing since a construction project is characterised by a `unique combination of input factors required to complete the projecta (Eccles, 1981b, p. 338). Stinchcombe (1959) observed the strong reliance on standard products in a comparison of the construction industry and the automobile industry. He argued that mass production is dependent on standardised tasks while construction projects are characterised by utilisation of standardised parts. This means that the uniqueness of the particular building must be created at the construction site, resulting in situations where `operative decisions are still very important at the work level [2] modi"cations of tools for special purposes is done by workersa (Stinchcombe, 1959, p. 182). In addition, Gann (1996) analyses similarities and dissimilarities between industrial housing and car production in Japan. A major "nding is that manufacturing principles derived from the car industry have been successfully used to produce customised homes. But, there are limits to which such techniques can be applied to manage the assembly of wide varieties of components and parts. Managers must trade o! the need to achieve economies of scale in the production of standardised factory parts with economies of scope in various stages of assembly (Stinchcombe, 1959, p. 437). A huge number of actors are involved in the modi"cations and adjustments at the construction site. Shirazi et al. (1996) found the main characteristic of construction organisation to be the co-ordination of specialised and di!erentiated tasks at the site level (p. 209). Therefore, the interaction between the actors tends to be quite intense. Their joint e!orts are mainly focused on developing solutions to problems in the particular project. Where development of resources and activities for purposes other than the speci"c project is concerned, however, the interaction is apparently limited. This "nding is in accordance with Hellgren and Stjernberg (1994), who found

that `activities are orientated towards responding to problems usually in ways that could be described as seeking the simplest and most straightforward solutiona (Hellgren and Stjernberg, 1994, p. 167). Modi"cations at the site can be onerous and costly, in many cases so costly that the savings gained from competitive tendering are more than outweighed by cost increases at the construction site. Love et al. (1999) have analysed the occurrence and consequences of reworking in construction. The lack of attention to quality by the design consultants was costly in time and money to all parties in the supply chain. For example the site management typically spent up to 30}35 hours a week for the "rst 10 weeks of the project checking the architectural and structural drawings to ensure that dimensions etc, coincided with one another. When errors were found the designers had to typically re-design and re-schedule the necessary elements. (Love et al., 1999, p. 8) The conclusion of this discussion is that the resource ties between the resource collections of speci"c "rms tend to be weak in the design phase of a project since adapted solutions are seldom developed. A similar pattern has been identi"ed regarding activities and their inter-linkage. Most activities are integrated through fairly standardised links, which also seem to have been established as norms and common routines for the entire industry. The low level of customisation, however, does not imply that manufacturers of materials have avoided adaptations to customer requirements. On the contrary, very few other industries are so entirely devoted to a speci"c, homogeneous group of customers. In comparison with industries such as the steel and wood industries (characterised by substantial end-user heterogeneity) the di!erence is clear. Standardised products and systems have been developed over time, partly through interaction among the actors in the construction industry, partly as a result of norms and rules lain down by government authorities. In this way the construction industry is characterised by certain &collective' adaptations. Thus, there is great similarity between the product o!erings of di!erent suppliers, both in terms of product features and in terms of price and other commercial conditions. The general pattern is thus that activities in production and distribution are standardised rather than customised, but in parallel to this pattern a slowly increasing trend towards customisation can also be observed. According to Cox and Thompson (1997) there is growing interest among actors in the UK construction industry in developing collaborative relationships. They state that partnering has gained popularity. Similar observations are reported by Baden-Helland (1995), Bennet and Jayes (1995), Cox (1996) and Doree (1997). So far these e!orts H have not been very successful, although `the search for

A. Dubois, L.-E. Gadde / European Journal of Purchasing & Supply Management 6 (2000) 207}215

211

more collaborative contractual relations has become a contemporary theme in the construction industrya (Cox and Thompson, 1997, p. 129). Similar tendencies were observed in this study. Both the plasterboard case and the insulation case provide insights into the bene"ts that can be reaped from using customised materials. In both these cases di!erent kinds of prefabricated solutions have been developed. They illustrate that, in some situations, performance is enhanced when activities are moved from the site backward in the supply chain. Adaptations in the construction industry are typically collective and project speci"c rather than relationship speci"c. A particular feature of these collective adaptations is strong reliance on contracts. However, according to Thompson et al. (1998), most standard forms of contracts actively encourage non-collaborative behaviour. Rather than being a mechanism to unite buyer and seller `the contract was being used as a wedge to drive distance between thema (Thompson et al. p. 36). Contracts thus impact on the degree of standardisation as well as on the role of the actors.

5. Theme 2: Specialisation and role of actors According to Eccles (1981a, p. 449) two prominent characteristics of the construction industry can be identi"ed. The "rst is `the organisation of the production work force into a variety of tradesa. The second is `the practice of subcontracting portions of a project to special trade contractors by primary contractorsa. These characteristics are related, and Eccles shows that specialisation of industry is an important determinant of subcontracting. As mentioned above the construction industry is, in general, characterised by a substantial degree of specialisation. In terms of the role of the actors, however, the situation is somewhat di!erent. The division of labour among the actors varies greatly from project to project and the role of an individual "rm can be very di!erent depending on the project. The cases showed that the contractual form chosen for the project is an important determinant of the organising of actors and, accordingly, of the division of labour in the design and implementation activities. Clients may choose to subcontract activities such as purchasing of installations and materials to a primary contractor. Sometimes, however, clients prefer to take on these activities within their own regime. In doing so, they may outsource the co-ordination of the activities on site either to an installation contractor or a material supplier (in this study the frame producer was an example). Primary contractors are also faced with a number of di!erent roles. They may use their own workforce on the site, or they may subcontract the activities to various trade specialists (bricklayers, plumbers, etc.).

Installation contractors sometimes only install, but they may also be responsible for designing the systems to be installed. The design work may then be undertaken by their own personnel or purchased from consultants. Moreover, there is some overlap between di!erent types of installation contractors. Some of the activities on site a!ect the systems of several installation contractors (e.g. joint solutions for cables, tubes and pipes). Therefore, one installation contractor might take on activities that are undertaken by another contractor in other circumstances. Consultants may be assigned by various actors, e.g. clients, primary contractors or installation contractors. Depending on which assignment they are engaged in the purpose and direction of their activities may change. Manufacturers of materials often supply standardised components only (as in this case). Through prefabrication they may also take on or replace activities that are otherwise undertaken on site. Manufacturers are also involved in physical deliveries * a service normally handled by distributors. Sometimes the materials producers' personnel are engaged in assembling activities on site, for instance when speci"c skills are required. Finally, distributors are involved in other activities than physical distribution. Sometimes they are responsible for materials supply at the construction site. Distributors are also active in certain forms of pre-fabrication and customisation, such as cutting of materials into speci"c lengths. The main conclusion from this discussion is that the roles of the di!erent actors are characterised by substantial variation. This also means that their roles are somewhat di!use. The activity scope of an individual actor tends to be broad, including design, production and distribution in various combinations, which may also vary between di!erent projects. In fact, Hellgren and Stjernberg (1994) have observed that `many actors may be partners with one another in some respects and competitors in othersa (p. 166). The industry as such is characterised by specialisation in di!erent sub-trades. In spite of this, individual "rms seem to have quite a low degree of specialisation. One important determinant is the contractual form that determines the roles of the contractors. This, in turn, may be the result of the extensive use of tendering procedures in the procurement process. There are certain consequences regarding potential network e!ects owing to these characteristics. One observation is that tasks in projects seem to be divided into functional disciplines that operate quite independently. The actors develop their own objectives, goals and value systems without considering the impact on others or the e!ects on project performance (Love et al., 1999, p. 1). Furthermore, the contractual form determines the roles of the actors. However, the main role of the contract does not seem to be to co-ordinate the operations of the various actors. According to Cox and Thompson (1997)

212

A. Dubois, L.-E. Gadde / European Journal of Purchasing & Supply Management 6 (2000) 207}215

the standard forms of contracts are `nothing more than instruments used by the parties to seek strict liability and attach blame to events as they occura (Cox and Thompson, 1997, p. 132). The authors argue that when such contracts are linked with the prevailing adversarial culture the parties are easily led away from trust towards self-interest.

6. Theme 3: The network within the network The most signi"cant characteristic of the construction industry and the construction process is the project organisation which has been described as a `temporary multiple organisationa (Cherns and Bryant, 1983). The organising of actors and the contractual forms applied have profound e!ects on the division of labour in design and implementation, as well as on the roles of the actors. The impact of the project organisation, however, is even more far-reaching. In the foregoing section it was mentioned that one e!ect of the prevailing division of labour is that two "rms may be both partners and competitors. Within a speci"c project, however, the situation is very di!erent. In the project the actors involved are partners in a larger context, the common objective being to erect a speci"c building within a set time frame. When the project network has been established it seems that the requirements of the project and the partners become even stronger than those emanating from the "rms. The interdependence within the project organisation tends to be strong, while it is quite weak among the "rms in the permanent construction network. Therefore, it may be appropriate to consider a project as a speci"c temporary network within the permanent network. In Fig. 1 "rms A, B and C are involved together in a construction project. They provide input in terms of people, equipment and other resources (A1, B1 and C1). At the same time A, B and C are involved in other projects. The characteristics of the construction process as presented above then imply rather strong interdependencies among the activities and resources of A1, B1 and C1. None of these actors will be able to ful"l its tasks without the support of the others. Therefore, throughout the project, co-operation between A1, B1 and C1 must be strong to handle the interdependencies. For the people from A involved in this project the relationships with people in B and C seem to be more important than those with other people in A. In fact, the relation between A1 and A2 may even to some extent be competitive if these require the same resources at the same time. Two major reasons seem to explain the strong interdependence in the project. First, as discussed above, substantial adjustments have to be undertaken on site. In addition, the bulk of products delivered are standardised.

Fig. 1. A construction project in its context.

Therefore, the participating actors need to interact and compromise extensively to be able to make the necessary adjustments. The second cause of interdependence is related to time. In most construction projects there are severe time restrictions. For some projects time is clearly the dominant restriction. The conditions observed in this study seem to be relevant to construction projects in general. According to Eccles (1981b) time is an important determinant of the unique input factor combination of individual construction projects. The need for adjustments is also a!ected by `unpredictable site conditionsa, and Eccles concludes that `on-site production exposes the production process to environmental conditionsa (Eccles, 1981b, p. 338). The uncertainties posed by these conditions are taken as arguments for extensive use of subcontracting (Eccles, 1981a, b). However, they may also be relevant in explaining the need for interaction and adjustments among the subcontracting "rms. This section on `the network within the networka has brought up two important issues for further discussion. The "rst is related to the fact that there are two network layers to consider when network e$ciency is to be analysed. One layer is the `permanenta network of actors and resources. The other is the temporary network formed around each construction project. The project net-

A. Dubois, L.-E. Gadde / European Journal of Purchasing & Supply Management 6 (2000) 207}215

213

work activates resources in the permanent network to perform the activities required for completion of the building. According to our observations the permanent network is characterised mainly by standardisation in terms of products and routines and, consequently, a low degree of interdependence among individual actors. In the project network, on the other hand, there is a substantial degree of interdependence and a consequent need for intense interaction among the actors. Both network layers impact on the e$ciency of a project. The permanent network imposes restrictions regarding the way it can be activated. Within these limits, however, many di!erent alternatives are available. A number of authors have argued that more co-ordination among the "rms in the permanent network would enhance productivity and performance. Love and Gunasekaran (1998) advocate concurrent engineering as a means for improving development processes. Agapiou et al. (1998) illustrate the bene"ts of logistics management and conclude that these systems need to be based on `partnering-typea arrangements (p. 136). Low and Mok (1999) studied a construction project where just-in-time principles were applied. The case study shows that it is possible to apply various just-in-time techniques to improve productivity at the site. The main problems concerned the roles of suppliers and subcontractors. The authors conclude that it takes long time to develop a long term and trusted relationship with a supplier `who is able to cater from all the needs of the company, including numerous meetings, committees and sta! other than those in the purchasing departmenta (Low and Mok, 1999, p. 667). The second issue concerns the complexity of individual projects as well as the construction industry in general. The resources of a "rm are simultaneously activated in a number of projects. Both resources and activities therefore need be co-ordinated. Four dimensions of co-ordination can be identi"ed in Fig. 1. First, individual projects must be co-ordinated * that is the co-ordination between A1, B1 and C1. Second, each "rm has to coordinate its activities and resources among the di!erent projects in which the "rm is involved (A1, A2, etc.). According to Kornelius and Warmelink (1998) a main problem for contractors `lies in the fact that many of its participants work in several projects simultaneously, and set their own priorities regarding the use of capacity and/or materialsa (p. 201). Third, some activity and resource co-ordination takes place on the "rm level, i.e. between A, B and C, and is thus not related to speci"c projects. Finally, co-ordination is needed between these sub-contractors and their suppliers, for example supplier C and supplier D. This study showed that the most prominent co-ordination dimensions are the project level and the "rm level. The relationship level has not attracted as much interest as in other industries, but would merit further exploration.

7. Theme 4: Relational exchange versus transactional exchange One of our main "ndings is that transactional exchange is the dominant form of business in the construction industry. Thompson et al. (1998) made similar observations and concluded that the business exchange tended to be contractual rather than relational. Furthermore, Gann (1996) found that management of supply chains in traditional craft housing production are often `typi"ed by market-based, short-term interactions between independent businessesa (p. 445). The underlying reason for this is the heavy reliance on tendering procedures. Supplier competition in each transaction is assumed to be the most appropriate means of securing e$ciency in operations. However, competitive tendering has other e!ects as well. The prevailing procedures guarantee that the actor constellations change all the time, making it di$cult to utilise experience gained in previous projects. According to Cox and Thompson (1997) this `creates particular cost ine$ciencies for the client as a new learning curve is climbed by the supplier each timea (Cox and Thompson, 1997, p. 128). It is always argued that each construction project is unique. It should be observed, however, that this uniqueness develops to some extent because of the purchasing behaviour of primary contractors. According to our "ndings the focus on individual projects and competitive tendering makes each project far more unique than is necessary. The supply strategy chosen thus clearly present contractors with certain drawbacks. In spite of this it has a very strong foothold throughout the industry. Purchasing behaviour based on arm's length relationships is supposed to make it possible to take advantage of the `market forcesa. However, there is some doubt as to whether this is really the case. Most supply `marketsa show few signs of being real markets, because in many cases only a few suppliers are available. Furthermore, the network as a whole is burdened with substantial costs for playing the market. The prevalent tendering procedure represents considerable sunk costs owing to the huge number of hours spent on design, planning and calculations that are never used. Some of our informants stated that, on average, they are able to win one tender out of 10. In one way or other the costs of the remaining nine have to be paid for. Other "rms estimated that they spend between 4 and 7% of their turnover on tendering. In addition, market mechanisms may be appropriate when there are no interdependencies among the solutions applied, but in construction solutions from di!erent suppliers have to be adjusted and combined into speci"c buildings on every occasion. Owing to this interdependence, products cannot be developed and exchanged independently. There is a strong connection between how companies buy and what they can buy. In a long-term perspective it

214

A. Dubois, L.-E. Gadde / European Journal of Purchasing & Supply Management 6 (2000) 207}215

is important to recognise that purchasing behaviour determines product development and innovation (Araujo et al., 1999). Strong reliance on standardised products and standardised interfaces between "rms clearly does not foster technical development. When standardised products are used in a context where interdependence among them prevails this problem is accentuated. Therefore, it is not surprising that the only traces of product development found in this study were related to the input side of the material producers. The raw materials used for two of the products (plasterboards and insulation) had been changed considerably * without any impact on the functionality of the products. Such changes can be undertaken without problems since the manufacturing of materials is independent of the rest of the construction process. As long as the standardised interfaces are not a!ected the content of the products can be changed. On the other hand, changes impacting on the interfaces to other products used on site would necessarily require adaptations by other "rms and are therefore avoided. Therefore, it is most likely that current co-ordination mechanisms hamper product development. On the other hand they provide bene"ts in other respects. Kornelius and Warmelink (1998) discuss the general development of industry towards specialised "rms and `virtual corporationsa. Their conclusion is that industry in general will have no choice but to evolve co-ordination mechanisms `similar to those that have existed in construction for long timea (p. 200). In the construction industry thus the co-ordination of projects is emphasised while co-ordination among "rms is left to transactional exchange. Focusing on one single dimension of e$ciency while neglecting others may cause problems because some aspects of relevance for total e$ciency are never considered. Torvatn (1996) discusses the importance of system limitations in general. He argues that problems tend to appear when certain system boundaries are chosen and then used for too long time. Kreiner (1995) points out the danger of a one-dimensional view, stating that `the fact that projects occupy only a bracket in time and thus have neither history or future, allows evolutionary processes little scope for improving performancea (p. 345). In a similar way Uzzi (1997) highlights the need for alternative units of analysis. His main argument is that when using the relationship as the unit of analysis the focus is shifted from the qualities of the transaction to the qualities of the relationship. Recently a number of studies have pointed out the need for variety in supplier relationships. Spekman et al. (1999) observed that `not all suppliers are treated equally, nor should they bea (p 114). Bensaou (1999) found that "rms balance a portfolio of di!erent types of relationships rather than relying on one type. In the same vein Dyer et al. (1998) advise "rms to avoid a `one-size"ts-alla strategy for procurement, and advocate supplier segmentation. Variety in relationships is important be-

cause customers can make use of di!erent skills and capabilities of their suppliers, and cope di!erently with di!erent interdependency situations. This is of particular importance where embedded networks are concerned. It is interesting to observe the di!erences between the prevailing type of exchange in the construction industry and the sample of 22 supply chains in "ve selected industries studied by Spekman et al. (1999). They conclude that `true supply chain partners have abandoned their unilateral, win}lose perspectives and advocate an approach that favours mutual gain, shared value and total system costsa (Spekman et al., 1999, p. 113). However, the main managerial conclusion of the authors is that di!erent contexts demand di!erent types of exchange. They argue that `some supplier relationships are still best managed as an &open-market' exchangea (Spekman et al., 1999, p. 114). What seem to be exceptions in other industries thus tend to be the norms in the construction industry. If the potential bene"ts of di!erent types of relationships were recognised by primary contractors the prerequisites for adaptations and development of network substance would be considerably enhanced. E!orts to increase involvement with suppliers should encourage co-operation among suppliers as well. In a study of learning e!ects in a construction project it was shown that the extent of "rm learning was highly correlated with the existence of connections between di!erent relationships (Hakansson et al., 1999). However, the importance to unlearn the old habits has also been stressed. Shammas-Toma et al. (1998) analysed the obstacles to implementation of total quality management systems in construction. Their main recommendation is `to unlearn the habits of competitive tendering, to take a holistic view of the process which is at odds with the current situation where a set of "xed and "nite requirements is expected to be passed from party to party in the delivery chaina (p. 189).

8. Concluding discussion The aim of the study was to analyse the occurrence and consequences of network substance in the construction industry. For this purpose it was found appropriate to make a distinction between two network layers * one permanent and one temporary. The permanent network is characterised by long-term recurrent exchange of products and services among a limited number of "rms. However, relationship adaptations in terms of activity links and resource ties still seem to be negligible in comparison with other industries. The implication is that there is little network substance to bene"t from. This has also been observed in the UK construction industry where the conclusion of one study is that the "rms `traditionally paid very little attention to the relational elements of business transactionsa (Thompson et al., 1998, p. 36).

A. Dubois, L.-E. Gadde / European Journal of Purchasing & Supply Management 6 (2000) 207}215

215

In the temporary network, on the other hand, co-ordination is substantial. The activities on the construction site require massive interaction among the actors. The adjustments and modi"cations undertaken provide huge opportunities for joint learning. These conditions should provide a breeding ground for network e!ects. This study discovered seeds of network substance in terms of jointly adjusted solutions that facilitated work for several parties on the site. However, there is little bene"t gained from this shared learning as long as relationships are con"ned to the duration of the individual project. This observation is also in accordance with "ndings from UK construction (Thompson et al., 1998), the main reason being that the constellation of "rms involved in a temporary network does not have joint plans beyond the project. The conclusion of the study is that the prevailing focus on transactional exchange hampers the development of network substance both in the permanent and the temporary network. Increasing reliance on relational exchange would enhance conditions for adaptations among "rms in the permanent network. Customised products and services could improve both e$ciency and innovation for all actor categories. Firstly, adapted solutions would reduce the need for adjustments thus increasing e$ciency of site operations. Secondly, customised solutions would stimulate development of di!erentiated o!erings. Increasing relational exchange would impact on the temporary network as well. This type of involvement implies that "rms recognise joint plans beyond the individual project. If they do so, it is likely that they will "nd ways to make better use of the incipient of network e!ects that appear in the temporary network. References
Agapiou, A., Clausen, L., Flanagan, R., Norman, G., Notman, D., 1998. The role of logistics in the material #ow control process. Construction Management and Economics 16, 131}137. Araujo, L., Dubois, A., Gadde, L.-E., 1999. Managing interfaces with suppliers. Industrial Marketing Management 28 (5), 497}506. Baden-Helland, R., 1995. Project Partnering: Principle and Practise. Thomas Telford, London. Bennet, J., Jayes, S., 1995. Trusting the team: the best practise guide to partnering in construction with the partnering task force of the Reading construction forum. University of Reading. Bensaou, M., 1999. Portfolios of buyer}supplier relationships. Sloan Management Review, 40, Summer, 35}44. Carlisle, J., Parker, R., 1989. Beyond negotiation. Redeeming customer supplier relationships. Wiley, Chichester, UK. Cherns, A., Bryant, D., 1983. Studying the client's role in construction management. Construction Management and Economics (2). Cook, K., Emerson, R., 1978. Power, equity and commitment in exchange networks. American Sociological Review 43, 721}739. Cox, A., 1996. Relational competence and strategic procurement management. European Journal of Purchasing and Supply Management 2 (1), 57}70. Cox, A., Thompson, I., 1997. &Fit for purpose' contractual relations: determining a theoretical framework for construction projects.

European Journal of Purchasing and Supply Management 3, 127}135. Doree, A., 1997. Construction procurement by Dutch municipalities. H Journal of Construction Management 3 (3), 78}88. Dyer, J., Cho, D., Chu, W., 1998. Strategic supplier segmentation: the next 'best practice' in supply chain management. California Management Review 40 (2), 57}76. Eccles, R., 1981a. Bureauctratic versus craft administration: the relationship of market structure to the construction "rm. Administrative Science Quarterly 26, 449}469. Eccles, R., 1981b. The quasi-"rm in the construction industry. Journal of Economic Behaviour and Organization 2, 335}357. Gadde, L.-E., Hakansson, H., 1994. The changing role of purchasing  * reconsidering three strategic issues. European Journal of Purchasing and Supply Management 1 (1), 38}45. Gann, D., 1996. Construction as a manufacturing process? Similarities and di!erences between industrialized housing and car production in Japan. Construction Management and Economics 14, 437}450. Hellgren, B., Stjernberg, T., 1994. Networks in design and networks in project implementation. Proceedings of the INROP Conference on Temporary Organizations and Project Management, Lycksele, Sweden, pp. 155}180. Hines, P., 1994. Creating World Class Suppliers: Unlocking Mutual Competitive Advantage. Financial Times/Pitman Publishing, London. Hakansson, H., Havila, V., Pedersen, A.-C., 1999. Learning in networks. Industrial Marketing Management 28 (5), 443}452. Hakansson, H., Snehota, I., 1995. Developing Relationships in Business Networks. Routledge, London. Kornelius, L., Warmelink, J., 1998. The virtual corporation: learning from construction. Supply Chain Management 3 (4), 193}202. Kreiner, K., 1995. In search of relevance: project management in drifting environments. Scandinavian Journal of Management 11 (4), 335}346. Lamming, R., 1993. Beyond Partnership * Strategies for Innovation and Lean Supply. Prentice-Hall, Englewood Cli!s, NJ. Love, P.E.D., Gunasekaran, A., 1998. Putting an engine into re-engineering: toward a process-oriented organisation. International Journal of Operations and Production Management 18 (9}10), 931}949. Love, P.E.D., Li, H., Mandal, P., 1999. Rework: a symptom of a dysfunctional supply-chain. European Journal of Purchasing and Supply Management 5, 1}11. Low, S.P., Mok, S.H., 1999. The application of JIT philosophy to construction: a case study in site layout. Construction Management and Economics 17, 657}668. Shammas-Toma, M., Seymour, D., Clark, L., 1998. Obstacles to implementing total quality management in the UK construction industry. Construction Management and Economics 16, 177}192. Shirazi, B., Langford, D., Rowlinson, S., 1996. Organizational structures in the construction industry. Construction Management and Economics 14 (3), 199}212. Spekman, R., Kamau!, J., Spear, J., 1999. Towards more e!ective sourcing and supplier management. European Journal of Purchasing and Supply Management 5 (2), 103}116. Stinchcombe, A., 1959. Bureaucratic and craft administration of production. Administrative Science Quarterly 4, 168}187. Thompson, I., Cox, A., Anderson, L., 1998. Contracting strategies for the project environment. European Journal of Purchasing and Supply Management 4, 31}41. Torvatn, T., 1996. Productivity in industrial networks. A case study of the purchasing function. Dissertation. Norwegian University of Science and Technology, Trondheim. Uzzi, B., 1997. Social structure and competition in inter"rm networks: the paradox of embeddedness. Administrative Science Quarterly 42, 35}67. Wynstra, F., 1998. Purchasing involvement in product development. Dissertation. Eindhoven University of Technology, Eindhoven.