Research Paper

The limits of design and engineering outsourcing: performance integration and the unfulfilled promises of modularity

Francesco Zirpoli1 and Markus C. Becker2

1Department of Management, Università Ca’ Foscari Venezia, Fondamenta San Giobbe, 873 Cannaregio, 30121 Venezia, Italy. [email protected] 2Strategic Organization Design Unit, University of Southern Denmark, Campusvej 55, DK-5230 Odense M, Denmark. [email protected]

Outsourcing design and engineering tasks in product development is increasingly popular.

However, firms that outsource design and engineering tasks often experience problems. So far,

no satisfactory answer exists regarding the question to what extent design and engineering

tasks can be outsourced before negative consequences occur. We address this gap. This paper

identifies the limits of design and engineering outsourcing in product development, and the

sources of these limits. To do so, it investigates the organizational challenges that a major

European automotive manufacturer faced when it decided to adopt an extreme form of design

and engineering outsourcing. Based on an in-depth case study, we identify the sources of

problems with outsourcing design and engineering tasks in product development, and shed light

on the limits of design and engineering outsourcing in product development.

1. Introduction: the advantages and limits of design outsourcing

Outsourcing design and engineering tasks in product development is increasingly popu-

lar. From the early nineties, the innovation litera- ture promoted involving suppliers in product development, emphasizing that they could play a key role in successfully developing complex products (Womack et al., 1990; Clark and Fuji- moto, 1991; Wheelwright and Clark, 1992). The focus of attention of this research was on the development practices of Japanese firms, whose success in terms of product development was seen in large part due to the heavy involvement of suppliers in the new product development process (henceforth ‘NPD’) (Clark, 1989; Nishiguchi,

1994; Bensaou, 1999). Considered best practices (Imai et al., 1985), they became models, which were imitated by Western firms (Womack et al., 1990), and the object of study of early supplier involvement (‘ESI’) (Helper, 1991; Smitka, 1991; Lamming, 1993; Nishiguchi, 1994; Helper and Sako, 1995; Sako, 2004).2

In many complex product industries, pressures for involving external sources of innovation have increased due to changes in external conditions (Sturgeon, 2002); the evolution of markets for technology and technological development ser- vices (Arora et al., 2001) has resulted in a general- ized tendency toward vertical disintegration, modularization, outsourcing, and networking in product development (Langlois, 2002; Ches- brough and Crowther, 2006; Howells et al.,

R&D Management 41, 1, 2011. r 2010 The Authors. R&D Management r 2010 Blackwell Publishing Ltd, 21 9600 Garsington Road, Oxford, OX4 2DQ, UK and 350 Main St, Malden, MA, 02148, USA

2008). It is very difficult for one individual firm to master all the knowledge required to design complex products, because the sources of indus- try expertise are so widely dispersed, and the knowledge bases that firms need to draw on for designing and producing complex products are rapidly expanding (Freeman, 1991; Powell et al., 1996; Grandori and Kogut, 2002; Christensen, 2006). Thus, relying on suppliers enables firms to take advantage of the specialist knowledge that suppliers possess. Drawing on suppliers facili- tates spanning the wide range of different tech- nological domains required, and drawing on deep specialist knowledge in each.3 Outsourcing de- sign and engineering tasks to suppliers became one popular way of involving suppliers in pro- duct development so as to draw on their specialist knowledge.

The idea of modularity played an important role regarding the implementation of design out- sourcing. In particular, Sanchez and Mahoney’s (1996) paper had a massive impact on how to organize innovation, focusing primarily on pro- duct architectures as the main lever for managing design and engineering outsourcing. There are several definitions of modularity (Ulrich, 1995; Sanchez, 1997; Baldwin and Clark, 2000). Bald- win and Clark (2000), define modularity as a decomposition scheme that assumes indepen- dence between modules, with interdependencies confined within module boundaries. Modules therefore interact only across standardized inter- faces (Baldwin and Clark, 2000). Ulrich (1995) provided not only a somewhat different but also influential definition of modularity. In his defini- tion, architectures are modular if there is a one- to-one mapping from functional elements to the physical components of the product, which spe- cifies de-coupled interfaces between components (Ulrich, 1995: p. 422).4 The reason why the idea of modular product architectures was linked to efforts of design outsourcing is that modular product architectures provide advantages that are very powerful when design and engineering tasks are outsourced. Modular product architec- tures enable parallel development of modules without the need to coordinate such development – the fit between the different modules is provided because the modules conform to a standardized format of their interfaces. This makes it both easier and more feasible to allocate tasks outside the boundaries of the firm (Siggelkow and Le- vinthal, 2003; Ethiraj and Levinthal, 2004). San- chez and Mahoney (1996: p. 73) went one step further and argued that ‘the creation of modular

product architectures . . . enables the design of loosely coupled, flexible, ‘modular’ organization structures.’ Design outsourcing that leverages the advantages of modular product architecture therefore became wide-spread, in the attempt to leverage the benefits of involving external sources of innovation in the product development process.

Yet, despite the advantages that design and engineering outsourcing promise, empirical re- search shows that many firms developing com- plex products have incurred negative effects when implementing design and engineering outsour- cing. A first limit seems to arise from industry specificities that restrict the degree to which the product architecture can be modularized. Indus- tries such as the PC and other electronic products seem particularly prone to modular architectures, and there are few indications that firms experi- ence problems in organizing their innovation activities (Baldwin and Clark, 2000; Sturgeon, 2002). This does not necessarily hold in other contexts. In the automotive industry, for exam- ple, the beneficial effects of modular product architectures in product development have turned out to be far fewer than expected (not- withstanding benefits in manufacturing and as- sembly) (Takeishi and Fujimoto, 2003; MacDuffie, 2008). The product architecture of the automobile seems to be characterized by persistent integrality (MacDuffie, 2008). Second, the literature that has followed Sanchez and Mahoney (1996) has been criticized for having neglected knowledge-related aspects. Scholars have argued that alignment between product architecture, knowledge, and task partitioning is not always a given fact (Brusoni et al., 2001; Chesbrough and Kusunoki, 2001; Takeishi, 2001). The reason is that the knowledge held by firms is the consequence of learning by doing opportunities connected to the allocation of de- sign and engineering tasks along the supply chain (Fine, 1998; Lincoln et al., 1998). This critique identifies a precise organizational task that firms have to address when leveraging external sources of innovation: to maintain a knowledge base that is broader than the immediate knowledge re- quired for the development of the products and services produced (Brusoni et al., 2001). Building on these insights, the literature on system inte- gration5 has developed advice on how to organize by taking into account the organizational issues arising from modular product architectures and the problems of knowledge erosion triggered by modular product architectures (Brusoni and

Francesco Zirpoli and Markus C. Becker

22 R&D Management 41, 1, 2011 r 2010 The Authors

R&D Management r 2010 Blackwell Publishing Ltd

Prencipe, 2001; Mikkola, 2003, 2006; Ernst, 2005; Frigant and Talbot, 2005; Miozzo and Grim- shaw, 2005; Staudenmeyer et al., 2005).6 This debate, too, cautions against relying on modular product architectures as a guide of how to implement design outsourcing. There is, thus, a shared belief that relying on external sources of innovation is a necessary condition for managing complex product innovation. How to implement the integration of external sources of innovation in the development process, however, has turned out to be a complex issue in the case of products whose architecture is integral (MacDuffie, 2008). Empirical evidence shows that, in this case, firms have problems with implementing the integration of external sources of innovation in the develop- ment process, and that there seem to be inherent limits to outsourcing, beyond which, outsourcing produces negative consequences.

Most of these negative consequences relate to competence accumulation problems. Pisano and Teece (2007: p. 294) highlight this point when they write, ‘vertical disintegration sharpens the need and enhances the difficulties of systems integration, as it requires integrating the activities of parties when there is no common ownership link. Interface standards and modularity, of course, facilitate outsourcing and thereby shar- pen requirements for integration. Failure at in- tegration in turn destroys any benefits associated with outsourcing in the first place. Systemic innovation of course tends to render prior inter- face standards obsolete, and favors vertical in- tegration. Possession of the systems integration capabilities gives high-tech firms outsourcing choices . . .With respect to complex systems, system integration capabilities could indeed be- come the bottleneck asset.’ In this respect, extant literature has highlighted that the main problem with extreme design and engineering outsourcing is that it can weaken firms’ ability to understand the components of a product (Lincoln et al., 1998; Becker and Zirpoli, 2003). This, in turn, can weaken the understanding of the ways in which components are integrated into systems and how to manage systems integration, that is, the firm’s architectural knowledge (Henderson and Clark, 1990). One of the most important risks of extreme outsourcing, hence, is the ero- sion of architectural knowledge and loss of con- trol over product performance. This risk produces a strong conflict between the benefits of design and engineering outsourcing that are the motivation to outsource and its drawbacks (Fine, 1998). On this aspect, the literature indicates

that the link between task partitioning (who does what) and knowledge partitioning (who knows what) cannot simply be managed by relying on modular product architectures (Brusoni, 2005), i.e., as a mechanistic consequence of a given product decomposition scheme. The link is much more complex, and to manage it is a key organizational challenge for firms that develop complex products. It thus becomes important to understand the causes of such complexity. As Takeishi (2002) pointed out, the complexity de- rives from the way in which OEMs7 manage the development of architectural knowledge: product architecture and the related knowledge (Hender- son and Clark, 1990) cannot be controlled with- out underlying component specific knowledge (Takeishi, 2001). Takeishi’s work has contributed to direct scientific attention to the types of knowl- edge – component specific vs architectural – that firms should retain in house, and to the conse- quences that relying on external sources of in- novation has for the competences of OEMs. However, empirical evidence on whether OEMs can successfully retain architectural knowledge when they rely on external component knowledge has, so far, been mixed (Takeishi, 2002). Our knowledge of the precise interaction of this ap- proach to design outsourcing with the firm’s knowledge base remains incomplete (Brusoni et al., 2001; Mahnke, 2001).

In this paper, we tackle these issues by explor- ing why firms that outsource the development of complex products often experience problems in implementing this strategy. As we have seen from the review of prior research, while there are insights into some of the problems, firms still continue to incur negative consequences when outsourcing design and engineering tasks. Identi- fying the sources of these problems is also im- portant in order to better understand how large firms should organize to successfully implement an open innovation strategy (a question specifi- cally identified by Chesbrough et al., 2006). As recent contributions show (e.g. Brusoni and Pre- ncipe, 2006), the detailed understanding of the complex organizational processes involved in im- plementing design and engineering outsourcing holds a key to developing our understanding of what distinguishes successful forms of R&D or- ganization.

The article is structured as follows: Section 2 presents the method, followed by the empirical findings in Section 3. We then turn to discuss the findings in Section 4 and draw our conclusions in Section 5.

The limits of design and engineering outsourcing

r 2010 The Authors

R&D Management r 2010 Blackwell Publishing Ltd

R&D Management 41, 1, 2011 23

2. Method

The research design we adopted was determined by the nature of the research question we tackle. In our review of prior research, above, we saw that existing theory does not yet offer a satisfac- tory answer to the question why firms that out- source the development of complex products often experience problems in implementing this strategy. While extant research has identified some of the factors involved, the complex pro- cesses that generate the negative performance consequences of design and engineering outsour- cing are not yet fully understood. As qualitative data are particularly suited for offering insights into complex social processes (Eisenhardt and Graebner, 2007), and more generally, for ‘why’ questions (Eisenhardt, 1989; Pettigrew, 1990; Yin, 1994), we chose to carry out an in-depth qualita- tive case study in order to contribute to fill the gap.

Our empirical observations have been gathered on the NPD process of a technology-intensive firm developing complex products. In particular, we focused on: (1) the decisions concerning the outsourcing of design and engineering tasks to suppliers; (2) the consequences of such outsour- cing for the knowledge of the firm and its suppli- ers; and (3) the firm’s organization for product innovation (organizational structure, NPD pro- ject management, inter-firm and intra-firm coor- dination mechanisms). In the subsequent sections, we report respectively on how we selected our case, on the data sources and data collection, and on the data analysis process we used.

2.1. Case selection

In selecting the case, we used a theoretical sam- pling approach (Eisenhardt and Graebner, 2007). We carried out case selection in three steps. First, we selected the automotive industry. The auto- motive industry is one of the most complex industries in terms of technologies and players involved in innovation processes (Maxton and Wormald, 2004). Moreover, the automotive in- dustry offers many opportunities to study the relationship of product architecture and the divi- sion of innovation tasks, and how such tasks are coordinated (Takeishi and Fujimoto, 2003). Furthermore, the automotive industry is also one of the industries where the theoretical dis- cussion concerning the role of modularity as a coordination device in inter-firm cooperation

occurs (MacDuffie, 2008: p. 42). Second, we selected the focal firm to be studied. We chose Alpha, a multi-brand auto manufacturer whose product range covers all market segments, from luxury to small cars, from trucks to Formula 1 racing cars. One of the attractive features of Alpha is to provide the opportunity to compare different degrees of design and engineering out- sourcing: Alpha had shifted quite radically from being a fully vertically integrated company (as far as design tasks are concerned) previously to an extreme outsourcer of design, before it subse- quently decided to once more reverse its task allocation scheme. These changes were observa- ble within a time-span of only 10–15 years.8

These clear changes in the allocation of design tasks provide a good opportunity for an in-depth analysis of the triggers and consequences of different ways of dividing design tasks, and thus, for casting light on the research question. Using theoretical sampling, we thus deliberately selected a case that was ‘very special in the sense of allowing one to gain certain insights that other organizations would not be able to provide’ (Siggelkow, 2007: p. 20). Third, the two research centers of Alpha, as well as eight first-tier sup- pliers belonging to Alpha’s value chain, were included in the field work. We selected suppliers on the basis of the following criteria: (1) rele- vance in terms of contribution to Alpha’s devel- opment activities, (2) heterogeneity of their industry, technologies, dimension, ownership, and nationality, and (3) their independence from Alpha. The objective of this third step was to observe the same units of analysis from the different angles of the auto manufacturer, its research centers and its suppliers. This seems appropriate for a study of networked innovation, which by definition comprises the interactions between different actors.

2.2. Data collection

We collected multiple types of data from different sources to ensure an accurate representation and to be able to triangulate our findings between different types and sources of data. We collected (a) industry publications, (b) archival data and company documents, and (c) carried out inter- views with employees of Alpha, its research cen- ters, and eight first-tier suppliers.

(a) The study of industry publications was mainly aimed at obtaining a solid understanding of recent developments in the industry (such as

Francesco Zirpoli and Markus C. Becker

24 R&D Management 41, 1, 2011 r 2010 The Authors

R&D Management r 2010 Blackwell Publishing Ltd

technological developments characterizing the in- dustry, history of the industry, etc.) in order to be able to understand the situation Alpha faced as well as its actions, and in order to formulate precise questions that our interviewees could relate to easily.

(b) Our purpose in collecting archival data and company documents was to obtain a comprehen- sive overview of the formal procedures regulating the NPD process at Alpha, and the tools and processes Alpha uses for managing the involve- ment of suppliers in the product development process. We collected about 2,000 pages of official OEM documents. They included norms and pro- cedures, which range from the description of engineering solutions for technical problem sol- ving, over procedures concerning the organiza- tion of NPD, to the definition of contracts with external sources of innovation, and the segmenta- tion and classification of suppliers.

(c) Finally, we carried out interviews at Alpha and its suppliers. The interviewing process oc- curred as follows. First, we developed an inter- view guideline that comprised theoretical variables of interest that had emerged from a previous review of the scientific literature. Sec- ond, company advisors at Alpha supported the definition of the interview agenda and helped us gain access to Alpha’s top management (as well as to archival sources). We drew up a list of inter- viewees we were interested in. The interviewees on our list were selected according to the relevance of their roles in the innovation process at Alpha and its first-tier suppliers. Table 1 provides an over- view of the interviews, which were carried out in 2006.9 We deliberately involved people in charge of strategic decisions as well as personnel with more operative roles, in order to capture the perspectives of top management as well as the details provided by people involved in the execu- tion of the NPD process. We interviewed the Chief Technology Officer, the Senior Vice Pre- sident of Human Resources, the Vice President of Product Portfolio Management (all three mem- bers of Alpha’s top-level steering committee), and the Director of Vehicle Concept & Integration (i.e., the manager responsible for systems integra- tion for chassis and vehicle), four of the five vehicle line executives (i.e., the engineers respon- sible for the development of cars in the small (A– B), medium (C), and upper (D–E), as well as the commercial vehicle, segments), and the staff func- tions of the Design and Engineering division (the Human Resources Manager and the Division Controller). We thus covered most of the top

managers in charge of the product development process. This set of interviews provided us with a comprehensive picture of Alpha’s perspective on the questions of interest. Regarding the eight suppliers involved in the 2006–2007 interviews, we interviewed ‘account managers’ (responsible for the commercial relationship with Alpha from the pre-offer phase until the end of the project) and ‘project managers’ (responsible for compo- nent or system development). In some cases, we also interviewed the supplier’s CEO (Table 1).

From the overall interview guideline, we se- lected the questions that each interviewee was supposedly most knowledgeable about. Finally, we carried out 80 hr of interviews. They were semi-structured, guided by individual interview guidelines. On average, interviews lasted for 2.3 hr. Many of the interviews were carried out by both authors, but one of us was present during all the interviews. All interviews were recorded and transcribed. At each field visit, field notes were also taken.

2.3. Data analysis

Data collection and analysis were two iterative processes. We started analyzing industry publica- tions and company documents before carrying out the interviews. However, after having com- pleted a first round of interviews and having analyzed them, we went back to re-analyze archi- val sources and eventually asked for additional archival data and information.

(1) The industry publications that we collected have been analyzed in order to identify new technological possibilities and challenges, new industry trends, the strategic and operational opportunities and threats facing Alpha, and Al- pha’s internal strengths and weaknesses. (We subsequently consulted industry experts to dou- ble-check our understanding.) This provided us with a better understanding of the concrete situa- tion Alpha was in when we started the interviews, enabling us to break down and frame the theore- tically inspired research question in a way that was close to the everyday reality of our interviewees. Moreover, it provided us with concrete questions for our interview guideline, such as questions about specific industry events, participants, and so on, which contributed to better operationalize our more abstract research questions.

(2) We systematically analyzed the archival data and company documents we had collec- ted at Alpha, generating from these detailed

The limits of design and engineering outsourcing

r 2010 The Authors

R&D Management r 2010 Blackwell Publishing Ltd

R&D Management 41, 1, 2011 25

documents an overview of (i) the formal system Alpha had in place for structuring, managing, and governing the NPD process, and (ii) the formal system that Alpha had in place for dealing with its suppliers. In order to do so, we built up a grid with the items we intended to observe (e.g. sup- plier segmentation, the timing of supplier involve- ment in the NPD process, the design and

engineering activities that were outsourced, etc.) and the archival data (e.g. Alpha’s NPD manual, Alpha’s NPD projects reports, suppliers’ docu- mentation, etc.), mapped the contents of docu- ments into our categories and verified the consistency of different data sources. Also, this process was iterative as we double checked during the interviews also that, beyond the reliability of

Table 1. Interviews (period January 2006–December 2007)

Company name

Product/role People interviewed Date of interview

Total length of interviews (hrs)

Alpha Cars (European) VP Investor Relations VP Product Portfolio Management VP Human Resources Business Development Manager Director of Vehicle Concept and Integration Manufacturing Director VP Design & Engineering (CTO) Vehicle line executive segment A–B Vehicle line executive segment C Vehicle line executive segment D–E Vehicle line executive segment Commercial Vehicles HR Director for Design & Engineering Director of Engineering&Design Innovation&Pre-Program Concepts

8_02_2006 9_05_2006 23_06_2006 14_07_2006 11_07_2007 13_12_2007

28

Alpha Research Centre

Research Center CEO Business Development Director Technologies Division

8_02_2006 22_03_2006

5

Company A Sealing systems/supplier Plant General Manager, Engineering & Design Technical Director Assistant Quality Manager

21_02_2006 07_06_2006

9

Company B Brakes/supplier Business Development Director Product Engineering R&D Manager Brake Systems

28_03_2006 3

Company C Car design development, turnkey development projects/supplier

Business Strategy Development Manager Project Manager Technical Division Manager Customer Manager Project Leader

23_03_2006 15_05_2006

10

Company D Chassis control (ABS – ESP, etc), power train, car multimedia/supplier

Manager Automotive Technology Product Planning and Marketing Cross Functional Project Manager

29_03_2006 4

Company E Car interiors, seats/ supplier

CEO Senior Program Manager Alpha Account Manager

9_02_2006 23_03_2006

5

Company F Safety systems (airbags, seat belts, brakes, chassis control, ABS – traction control systems, etc.)/ supplier

Account Director, Manager Programs & Application Engineering Inflatable Restraint Systems

9_02_2006 23_03_2006

5

Company G Stamped parts in metal/ supplier

Plant Manager 3_02_2006 5

Company H Thermal systems/supplier Sales & Marketing General Manager, Alpha Sales Manager, R&D Thermal Systems Division Manager

3_04_2006 6

Francesco Zirpoli and Markus C. Becker

26 R&D Management 41, 1, 2011 r 2010 The Authors

R&D Management r 2010 Blackwell Publishing Ltd

the source, our classification of data was correct, and we eventually revisited it. This process helped us to fine tune the questionnaire and contributed to ask specific questions about the formal instru- ments and mechanisms for how Alpha guided, monitored, and managed the NPD process, and the involvement of its suppliers in that process. It was also crucial for being able to understand when interviewees’ descriptions actually did not conform to the formal picture, which in turn enabled us to systematically pursue such diver- gences in the following interviews. Often, such questions were crucial for describing the organi- zation as it was really implemented in practice. Moreover, the description of the formal organiza- tion of NPD enabled us to identify organizational changes in the formal organization, and relate them to external events and internal decisions.

(c) As far as the analysis of the interviews is concerned, we generated two types of documents from the interviews: interview transcripts and field notes with observations from the field, which extended beyond the conversation of the interview to include our own reflections and remarks, as well as occasional observations, for instance about the environment where the inter- views occurred, or objects such as prototypes that we were shown. We created a case study database (Yin, 1994) where we stored the interview guide- lines, the interview audio files, the interview transcripts, field notes, and separately from the data, all documents that pertained to our inter- pretation and analysis (explained in more detail below). Regarding the field notes, immediately following each of the field visits both authors compared their individual field notes and dis- cussed their impressions of the field visit. As for the interviews, we analyzed them as follows. Once the transcripts were available, we shared them and carefully reviewed each transcript in- dividually, each author for himself. In doing so, each of us did two things: First, we summarized the interview, to render what each of us thought was the gist of the interview. These summaries served to build up a case description, with regard to the aspects we were interested according to the research question (Yin, 1994). Second, we as- signed codes to passages in the text, categorizing them according to topics they provided informa- tion on. We focused attention on data that seemed to be particularly useful for casting light on the research question (Yin, 1994). Based on this coding of the text, each author then identified the answers that seemed most interesting with regard to the overarching research question. The

overarching research question was then broken down in many ways in the interview guideline. In the process of identifying the answers, we thus summarized and translated the detailed answers to individual questions in the interview back to the overarching research question. Having com- pleted the individual review, we then compared and discussed both our summaries of the inter- view. This served to check our understanding of the content of the interview. Where we did not agree, we considered the discrepancies in what we thought was the gist of the interview, until we arrived at a common understanding. We then discussed our coding of the interview, and what each considered the most interesting content in each category that we had used for coding the interview. Discussing the meaning of the inter- views for answering the overarching research question, we implemented our analytical strat- egy, i.e., to build an explanation by identifying sets of causal links (Yin, 1994). It led to the identification of causes of phenomena that we had either asked about directly or discovered in the interviews (for instance, the causes of low project performance or the reasons for reorgani- zation or a change in a relationship with a supplier). It also led to formulating provisional propositions, which we would pick up in subse- quent interviews. Those where we found no contradicting evidence but where supporting evi- dence accumulated, later crystallized into the main findings. As the outcome of this process of discussing an interview, one of us wrote a note with the result of this discussion, which we subsequently shared. We repeated the process, thus generating interview summaries, coded ver- sions of the transcripts, and notes that captured the condensed result of our discussion of the analysis of each interview. In the next step, we then compared, summarized, and identified the ‘big lines’ in the responses over all the interviews where there was consistence, and identifying parts of the story where we could not identify consistence. From these ‘big lines’ of the data, we then developed the answers to the research ques- tion by iterating among the case data, emerging theory, and literature (Eisenhardt and Graebner, 2007). Having identified the preliminary answers to the research question from the interviews, we then engaged in triangulation. In particular, we triangulated between different respondents, and between different types of data. First, we used archival data to reconcile diverse views we ob- tained through the interviews (we used the afore- mentioned grid to perform this task). In this

The limits of design and engineering outsourcing

r 2010 The Authors

R&D Management r 2010 Blackwell Publishing Ltd

R&D Management 41, 1, 2011 27

respect, we extensively used the rich documenta- tion of its NPD process that Alpha provided us with. In the NPD process handbook, for exam- ple, we found both technical norms (e.g. how a component should be engineered, what compe- tences and knowledge bases were Alpha’s prior- ity, etc.) and organizational procedures (e.g. company structure, processes milestones, sup- plier segmentations, type of contracts and re- sponsibilities of suppliers, etc.). We constantly verified the consistency of information gathered from these two types of sources (company docu- ments and interviews). Second, we triangulated data by systematically comparing the informa- tion gathered from staff belonging to different organizations. The baseline for comparison was established by asking the same questions on the same unit of analysis. Given the type of research question, confronting the suppliers’ and OEM’s views on the allocation of component specific knowledge, architectural knowledge, and their development over time was fundamental to building a comprehensive and articulate repre- sentation of reality. In fact, if we had confined our analysis to Alpha’s perspective, we would not have achieved a comprehensive representa- tion of reality. In this respect, the possibility of gathering and triangulating data from eight different suppliers was of paramount importance. This allowed us to obtain a more precise and articulate picture of the strategy Alpha has actu- ally implemented in practice, as far as the alloca- tion of design activities along the supply chain is concerned.

Our method is subject to limitations. Our sample could be biased by the fact that we only interviewed companies located in one country, and belonging to one industry. However, the vast majority of companies in the sample are local branches of multinational corporations. More- over, technological heterogeneity somewhat counterbalances industry specificity. As shown in Table 1, the companies analyzed belong to completely different sectors (from pure mechan- ical engineering to electronics, engineering con- sultancy, and rubber). Exposure to multiple technological domains, product development priorities, communities of practice, technical com- plexity, and system, component and module in- tegration characteristics during the research process provided a source of learning and con- tributed to conclusions that are well informed and empirically grounded (although they can, of course, be neither normative nor generally applicable).10

3. Empirical findings

In this section, we first describe the evolution of Alpha’s engineering and design strategy and how its implementation were organized, then turn to identify the consequences produced by this strat- egy and by how it was implemented.

3.1. The evolution of Alpha’s vertical structure for innovation

3.1.1. Phase 1: supply base rationalization (1987– 1993) Around the beginning of the 1990s, Alpha’s supply base was essentially limited to the domes- tic automotive components industry, and Alpha was still a highly vertically integrated company. Very few if any suppliers were involved in the design process: about 3,450 of the 3,500 technical drawings that make up the design of a car were carried out by Alpha’s engineers. Moreover, Al- pha dealt with more than 3,000, mostly quite small, first-tier suppliers. In those years, rationa- lization needs were a priority and many efforts were dedicated to building a local supply base that was competitive in terms of quality and production costs.11

3.1.2. Phase 2: toward extreme design outsourcing (1993–2001) Until the mid-1990s, supplier involvement in Alpha’s NPD was still limited to the phase of industrialization and production of components and systems. With the re-engineering of the NPD process in 1996, Alpha’s outsourcing strategy gained speed, however. Alpha decided to move from outsourcing manufacturing tasks to suppli- ers (a practice already consolidated at that time) to outsourcing the design and engineering of components and systems. At the same time, Alpha began to frame suppliers’ contribution also in terms of strategic partnerships to develop complementary capabilities and to outsource components design. Alpha’s target for the end of 1990s was to outsource complete systems of components, and finally, complete pre-assembled modules. The aim was to deal with just two types of suppliers during the NPD phase: systems and module suppliers. Alpha did indeed reach an exceptional level of outsourcing: up to 85% of the total value of a car was engineered by suppli- ers at the beginning of the 1990s. By doing so, Alpha could leverage on a newly shaped tiered supplier structure and was able to find suppliers

Francesco Zirpoli and Markus C. Becker

28 R&D Management 41, 1, 2011 r 2010 The Authors

R&D Management r 2010 Blackwell Publishing Ltd

to build and pre-assemble modules. Between 1996 and 2001, Alpha became one of the firms with the highest degree of outsourcing of design engineer- ing in the automotive industry, becoming most probably the one which pushed design outsour- cing further than any other OEM in the industry.

Why did this happen? In the same way as other car makers did, in those years Alpha needed to cope with new exogenous12 technologies. At the same time, the group Alpha belongs to did not have sufficient resources to invest in the auto division. Moreover, at least at the beginning of the out- sourcing implementation, Alpha started appreciat- ing the benefits of increased strategic flexibility, reduction of development costs and lead time, and quality improvement yielded by the early involve- ment of suppliers in the NPD process. Ironically, for top management this confirmed the idea that no new investments in the auto business were required, reinforcing a vicious circle the consequences of which only became clear later on.

3.1.3. Phase 3: a new pattern toward in-sourcing (2002–2006) When we interviewed the manager at Alpha responsible for product development in 2001, he admitted that Alpha had lost its competence to design dashboards, suspensions, and occupant safety systems. In the interview, he also reported, however, that Alpha had started a process of re- building some competences in house, which would have taken an estimated 5 years. Our interviews in 2006 confirmed this information:

When I arrived [in 2005] the situation was disastrous from the point of view of technical competences and from an organizational point of view. The business units were completely out of touch. Through a very strange strategy of outsourcing, the firm had systematically destroyed the technical heart of the firm. The engineering division was just another engineer- ing supplier that by accident was part of the firm. Nothing more, nothing less. (Chief Technology Officer, 2006)

Our 2006 interviews, moreover, allow us to trace how the recovery strategy was put into action. It had two important stages. First, Alpha realized that learning from suppliers had to become an objective in itself, and a justification for involving suppliers in the product development process (i.e., the main objective was to learn, rather than have support in designing a particular component or system). Suppliers were selected on the basis of

their competence and with an explicit agreement of sharing knowledge.13 In this respect, the de- creasing emphasis on cost aspects was a real novelty for Alpha’s managers. Second, Alpha started hiring staff knowledgeable in the areas in question, and to enable internal development and learning (the most important areas in which Alpha started acquiring competences that were lost previously were dashboards, suspensions, occupant safety systems, and virtual development tools). At the moment, Alpha has engineers that possess all the necessary competences to design and engineer a whole new car. It currently out- sources 50% of the design and engineering of new systems, down from 85%. Note that 50% is considered ‘natural’ for the auto industry by Alpha’s managers – it being standard industry practice to outsource the design and engineering of components whose technology is mature (i.e., the calliper for a brake), or not within the scope of their activities (i.e., electronic hardware).

Our interviews, however, show that not all the problems in acting as a system integrator related to the level of outsourcing. Indeed, the problems were not solved simply by increasing the number of design tasks carried out in house. One of the main consequences of the outsourcing strategy described above, in fact, was that Alpha found itself in the very difficult situation of competence erosion. Apparently, it had pushed outsourcing too far. The interesting question is: Precisely, which problems were triggered by too high a degree of engineering outsourcing? To cast light on this, the following sub-section describes how systems integration was organized at Alpha. This provides an occasion to ‘read off’ the challenges raised by going beyond the limit of outsourcing.

3.2. Organizational aspects of system integration

3.2.1. The scope of the knowledge base According to the managers we interviewed at Alpha, the main reason why Alpha’s system integration strategy failed was a considerable lack of technological competence in key areas and the way Alpha interpreted system integration. This observation is further substantiated by a concise statement made by the Director of Vehicle Concept and Integration:

It is naı̈ve to believe you can integrate a system without holding an in depth and detailed knowledge of the components that are going

The limits of design and engineering outsourcing

r 2010 The Authors

R&D Management r 2010 Blackwell Publishing Ltd

R&D Management 41, 1, 2011 29

to affect the performance of the whole car. Managing each system performance does not, in fact, automatically result in effective system integration. The performance is the ultimate objective, not systems. (Director of Vehicle Concept and Integration, 2006)

The Chief Technology Officer also underlined that:

. . . what matters is the cost of performance, not the cost of systems. (Chief Technology Officer, 2006)

An illustration of the difference between integrating systems and performances and of the role of specific component knowledge is provided by the following quote from an interview at a first-tier supplier of safety systems (airbags, seat belts, brakes, chassis control, ABS – traction control systems, etc.):

Before the integration of the occupant safety system was outsourced to us, despite the fact that we did not control the design of all the (sub-)systems involved. In fact, we were not responsible (or even competent) for the design of the chassis, the engine layout, and packa- ging, i.e. of components and systems that affect the performance of the occupant safety system. After the successful frontal crash test of a key vehicle for Alpha which obtained the maxi- mum score on the EuroNCAP test, we met with Alpha people in order to consolidate the lessons learnt and draw some best practices to extend to future projects. In the meeting we could not point to the reasons for success because neither Alpha nor we fully dominated the whole system performance, i.e. the interac- tions of all the components. After this, Alpha realized that they could not leave the fate of the next occupant safety system developments to serendipity. Alpha re-internalized the respon- sibility for managing the overall Occupant Safety System. As a consequence, we have been relieved of being responsible for the whole system performance and, now, develop only parts and components of the system. (Alpha Account Director at the supplier, 2006)

This example shows to what extent Alpha had outsourced system integration, and with it, full control of key vehicle performances. Alpha’s man- agers came to realize that acting as a system integrator without underlying component-specific knowledge was just not possible: without this underlying component-specific knowledge, systems

integration competence was difficult to achieve. It was precisely this kind of knowledge, however, which had been lost by focusing – a little too much it seems – exclusively on the competence of systems integration. As a consequence, this affected its ability to reach given vehicle performances in areas considered highly relevant for customers (dash- board design or safety-NCAP test).

3.2.2. The organization of NPD One of the basic problems was that due to the outsourcing of certain design and engineering tasks, there were simply no longer people in- house with competences to design dashboards, suspensions, and occupant safety systems at the level that was required.

To appreciate the situation, it is useful to look at the changes in the internal structure adopted by Alpha in 1996–2001. Alpha had gradually adapted it to fit its outsourcing strategy. Functional units (where component and system technologies were developed) lost many engineers that were perma- nently staffed at the ‘platforms’ (in Alpha jargon, ‘platforms’ are organizational units whose task is to develop vehicles, all the way from concept to production ramp-up – not to be confused with the usual meaning of ‘product platform’). Platforms became the most important units of the company. They performed most of the design and engineering of the vehicle, heavily leveraging on suppliers’ competences, and were responsible for development projects’s execution. The design and engineering division accordingly became very ‘light-weight.’ In the words of the Chief Technology Officer:

Engineers in the company just delegated design and engineering to somebody else, usually outside the company. This was not acceptable for a technology firm. (Chief Technology Officer, 2006)

The assumption underlying delegation was that the role of Alpha’s engineers was to put together the systems developed by competent suppliers. The work of Alpha’s engineers – as they inter- preted it – was, hence, basically to coordinate suppliers and manage the development projects. Such an interpretation did not lack a certain consistency with a ‘light-weight’ design and en- gineering division (although that might or might not have been intended). This task consisted mainly in assigning (and monitoring) systems /components performances to suppliers, to be achieved in a specified time schedule at given costs. In many cases, the level of specification of systems/components characteristics did not go far

Francesco Zirpoli and Markus C. Becker

30 R&D Management 41, 1, 2011 r 2010 The Authors

R&D Management r 2010 Blackwell Publishing Ltd

beyond the setting of target price and perfor- mances. There was a problem, though. As the manager responsible for system integration put it:

In so doing, [Alpha] was losing the ability to set performance targets to suppliers and moni- tor their work. Most importantly, [Alpha] was losing the ability to manage performance trade-offs. Managing each system performance does not, in fact, automatically result in effec- tive system integration. (Director for Vehicle Concept and Integration, 2006)

The managers we interviewed on the issue justi- fied the lack of competences in managing perfor- mances trade-offs at the systems and components level as being the direct consequences of not carrying out design activities concerning those systems and components any more. In particular, the managers we interviewed confirmed that it is difficult (if not outright impossible) to integrate systems without mastering the technology under- lying each system in depth. Moreover, the grow- ing importance of the ‘platforms’ within the organization, together with the need to coordi- nate external suppliers, contributed to congesting many development projects. In a recent interview (2006), a vehicle line executive told us:

Before [period until 2005], you had to be a ‘magician’. Everything was under the control of the manager responsible for the vehicle. It was just impossible to cope with all the aspects of product development, from technical to economic issues.

(Vehicle Line Executive, 2006)

The problem was that platform managers were responsible both for the economic and project

management aspects, and at the same time, for all the technical coordination and responsibility for the engineering solutions. The latter task included systems integration. All this was simply too much. Interestingly, the reason why people at Alpha did not foresee, and hence, avoid the problem was that they were convinced that the involvement of competent suppliers would have eased the need for technical coordination. In reality, economic and project management aspects are strictly inter- twined with technical coordination issues.

And, in fact, this example shows that the consequences of extreme outsourcing also af- fected the efficiency and effectiveness of project management and the organization. The loss of control of the important pre-development phase (Figure 1) and Alpha’s peculiar interpretation of co-design14 are both relevant consequences of the choice to implement a networked innovation strategy that deserves further attention. We now turn to analyzing them.

Owing to the lack of manpower in some areas, and of the level of engineers’ competences in designing particular systems, Alpha chose to in- volve suppliers in the pre-development phase in order to let them carry out large parts of it. For example, suppliers started to assist Alpha also in the analysis of what component technologies and technical solutions could better suit the vehicle targets, including the definition of key perfor- mance attributes. Involving and governing such supplier involvement are not trivial, however. Because of the lack of competences, it had pro- blems for instance in setting precise component specifications (interview with manager responsi- ble for systems integration, July 14, 2006). More- over, suppliers were given the responsibility for (and the task of organizing and carrying out) in- tegrating sub-systems (defining main component

Figure 1. Pre-development and development phases.

The limits of design and engineering outsourcing

r 2010 The Authors

R&D Management r 2010 Blackwell Publishing Ltd

R&D Management 41, 1, 2011 31

targets and specifications). This process was not ‘explicit,’ in the sense that Alpha continued to set targets and specifications. The point, in fact, was that the design solutions and systems perfor- mances were not fully under Alpha’s control. This is why, implicitly, suppliers gradually took over parts of the pre-development phase. More- over, the pre-development phase was not orga- nized in the best way possible, resulting in bottlenecks in the decision process. As the man- ager responsible for system integration observed:

When system integration is carried out in the ‘development’ phase, the overall product per- formances are tested at the end of the process: there remains little time for fine tuning and any change is costly and may result in product quality problems. (Director for Vehicle Con- cept and Integration, 2006)

This is the result of black-box sourcing: suppliers in charge of component concept development (pre- development phase) hand over the component to Alpha when necessary for the physical integration in the vehicle, i.e. in the late stages of development.

We turned to modules as a means to move from a situation in which we had to manage 5000 components to a situation in which, once the interfaces had been designed, we could leave everything to 5 system suppliers. Once the interfaces were defined, suppliers were supposed to develop the best systems. In reality the best was the best for them. We experienced an increase in costs. In fact, suppliers did not develop the best components for what we needed, but supplied the best available compo- nents. We did not have the lenses to see inside the module.(Alpha Innovation and Methodol- ogies Manager, 2007)

It is worth noting that the loss of control occur- ring during the pre-development phase is particu- larly important: the pre-development phase decides on the ‘amount’ and difficulty of trade- offs that necessarily will occur down the line in the development phase. They are ‘programmed’ in the pre-development phase. Not dominating that phase but letting suppliers take important decisions (with a focus on their sub-systems), and not focusing attention on this phase sow the seeds for a whole string of problems in the development phase. When solicited on this issue in order to gather information on how they decided to cope with the problem, Alpha’s managers reported that

they acted on two levers. Beyond the tendency described above to increase the amount of in- house design, Alpha started differentiating the involvement of suppliers in its NPD process in order to re-gain control of the pre-development phase. In particular, Alpha increased the amount of in-house design during the pre-development phase, reversing the black-box sourcing strategy as it was carried out before. This also implied a change in the type of skills that Alpha’s engineers are requested to have. In this respect, it is im- portant to report that one of the main concerns expressed by the manager of human resources of the division in charge for products engineering and design was a particular shortage of staff with skills in virtual simulation technology. These skills are key for performing pre-development activities efficiently (with as few prototypes as possible) and effectively (with reliable knowledge of the performance the vehicle will have once in production). The Chief Technology Officer con- firmed that this shortage was a serious limitation for organization design; because such staff were few, it was impossible to establish decentralized virtual simulation groups in the various organiza- tion units where virtual simulation was needed. Rather, there was little choice but to have a central group where all virtual simulation was carried out complicating the re-focusing on the pre-development phase. The following quote drawn from an interview with Director of ‘In- novation & Pre-Program Concepts’ in the Engi- neering & Design Division exemplifies this point:

We realized that technical interdependences must be tackled at the beginning of the process, not at the end. Before we integrated the vehicle at the end of the process with inevitable and costly re-design. This was a consequence of the way we outsourced design activities and of the balance within the firm of design responsibil- ities. At the moment my division carries out the pre-development for all the vehicles under development. This has changed our organiza- tion for NPD substantially relieving product platforms and suppliers of most pre-develop- ment activities.

4. Discussion

Our study was designed to cast light on the question why firms that outsource the develop- ment of complex products often experience

Francesco Zirpoli and Markus C. Becker

32 R&D Management 41, 1, 2011 r 2010 The Authors

R&D Management r 2010 Blackwell Publishing Ltd

problems in implementing this strategy. To an- swer the question, we investigated the conse- quences of design and engineering outsourcing on the firm’s competences and knowledge do- mains, in order to identify the sources of pro- blems with outsourcing design and engineering tasks in product development. We also analyzed how the way in which design outsourcing is implemented in practice (organizational processes and structures) can alleviate or worsen problems arising from design outsourcing.

Our data, moreover, add new insights into the choices concerning the balance between architec- tural and component specific knowledge and on how to maintain and develop the knowledge base necessary to leverage external sources of innova- tion. These issues seem to be among the most important operational details that require atten- tion in innovation management. At the same time, they also seem to be the operational details that have the most important strategic conse- quences.

4.1. Incomplete decomposability of product performance and its consequences

One of the key insights that emerge clearly from the case study is the crucial importance of the difference between the integration of physical systems and the integration of the performance of such systems (a process known as ‘performance management’). Simply speaking, a car is designed in order to provide a certain performance (hand- ling, fuel consumption, certain crash test para- meters, and so on). This performance is generated by physical components and systems (such as the occupant safety system composed of brakes, air- bags, seats, seat belts etc.). The practical problem engineers have to solve is to design a vehicle that can systematically generate a certain level of overall performance (such as, for instance, 5 stars in the NCAP crash test, handling characteristics, noise, etc). This task is accomplished by designing individual physical components and systems. What really matters, though, is the performance of the vehicle as a whole. Importantly, however, performance cannot necessarily be decomposed in the same way as the components. The reason is that some performances are generated by more than one system (Sosa et al., 2003). The occupant safety system is a good example. Brakes, seat belts, etc. are important in determining the safety characteristics of a car. In a frontal crash, how-

ever, the position, weight, and structure of the engine, and the chassis characteristics, will also influence the impact absorbed by the vehicle and the consequences on the driver. Vehicle-level performances, thus, are not fully decomposable and different aspects of vehicle-level performance cannot always be attributed completely to parti- cular components and systems. In our example, in order to design a car so that it has high perfor- mance values regarding occupant safety, it is not sufficient to design just a good occupant safety system. From this insight, it now becomes clear that the automobile is a persistently integral product (MacDuffie, 2008). Another limit to the decomposition of performances is posed by multi- ple performance dimensions (e.g. speed, noise, vibration, harshness, etc.), which are interdepen- dent in the overall vehicle-level performance. For these two reasons, there remain reciprocal inter- dependencies between the contribution of the individual components and systems to the overall performance (Sosa et al., 2003). As a conse- quence, there are limits to what extent one can fully specify the contribution of an individual component or system to overall vehicle-level performances ex ante. Extant research on search under complexity (e.g. Levinthal, 1997) shows that in such situations, finding the combination with the global performance optimum is not a minor issue.

The existence of reciprocal technical interde- pendences between components, component spe- cific performances and the overall performance of the product thus force managers to consider performance trade-offs. This is precisely what we saw was a major challenge in this case: as the manager in charge of systems integration explained, the key challenge in system integration therefore is to manage performance trade-offs between the various systems.

4.2. The limits of leveraging modular product architecture for achieving coordination

As seen in the literature review section, modular- ity is a central concept in the scholarly debate on organizing networked innovation. What is the role of modularity in organizing the development of complex products? As described above, from the mid-1990s onwards, Alpha attempted to pur- sue a system-integration strategy with the aim of outsourcing whole modules. The analysis of the case clearly shows that modular product design is

The limits of design and engineering outsourcing

r 2010 The Authors

R&D Management r 2010 Blackwell Publishing Ltd

R&D Management 41, 1, 2011 33

not the most appropriate way to deal with the issue of integrating performances – even though it is appealing for physical integration. The reason is that the benefits attributed to modularity (San- chez, 1997) are not able to address the problem of integrating performances. Modular product de- sign is beneficial in solving the task of coordinat- ing suppliers as they independently develop components and systems that need to be inte- grated physically (thanks to standardized inter- faces). For this task, modular product design is effective, as it diminishes the coordinative efforts required to coordinate with suppliers. As seen, this is the case when it is possible to both define the interfaces between elements of the whole ex ante, and to assign the development of a specific part to a supplier. Coordination then would follow as a natural consequence. The interface definition, in fact, would embed all the necessary information for the parties as Sanchez and Ma- honey (1996) predicted. However, our case study shows that for purposes of designing vehicle-level performances, it is necessary either to reduce the reciprocal interdependencies between the perfor- mances of the different components and systems ex ante, or to specify them. There are limits to the former, and in order to do the latter, component- specific knowledge is required. Defining standar- dized physical interfaces does not mean standar- dizing the performance contribution of a module. It does not, therefore, diminish reciprocal inter- dependencies between component- and systems- performances.

This is why firms need to tackle performance trade-offs. And in tackling performance trade- offs, standardized interfaces simply have no trac- tion. In other words, even if a one-to-one map- ping between performance and component is achieved (see for example Ulrich, 1995), a white spot remains due to the fact that the coordination properties implicit in standard definitions of modularity do not apply to the case of perfor- mances that cannot be decomposed in the same way as the components. Modular product design as an ex ante integration mechanism is not, there- fore, always effective for the integration of per- formances. This is an important addition and limit to the consequences of modular product design that is not usually mentioned in the influential literature on modular product design (e.g. San- chez and Mahoney, 1996; Baldwin and Clark, 2000).

Consequently, the upshot is that modular pro- duct architecture does not necessarily lower the coordination cost among different parties in-

volved in developing complex products (for simi- lar evidence, see Brusoni, 2005). The reason is that modular product architecture simply cannot directly address the dimension that is important, i.e. performance. This becomes clear once perfor- mance, rather than the physical dimension that generates such performance, is the focus of atten- tion. The hopes placed in modular product archi- tecture to ease the involvement of external sources of innovation in the development of complex products might therefore be largely misplaced. On a final note, one might apply a coordination cost perspective on the involvement of external sources of innovation. Such a perspective would decide how many of such sources to involve (and which ones) based on coordination cost, driven by coordination challenges. Our argument in this section is that modular product architecture can- not lower the coordination cost in the context of performance integration at all. Moreover, the drivers of coordination cost in the context of performance integration are very different from modular product architectures, i.e., learning by doing processes.

4.3. Component-specific knowledge and architectural knowledge

Our empirical evidence indicates that for making performance trade-offs, a profound knowledge and understanding of the interaction between the subsystems (e.g., seat belts, airbags, ABS system, engine, body etc. in a frontal crash) is indispen- sable. This is neither surprising nor new. It is an insight well established in the literature, both in complex product systems more generally and in NPD more specifically. Such knowledge is some- times termed ‘architectural knowledge’ (Baldwin and Clark, 2000). Our empirical contribution is to identify the prerequisites for acquiring pro- found knowledge and understanding of the in- teraction between the subsystems. As described in Section 4, one of the main causes of Alpha’s difficulties were problems in setting cross-system functional requirements. Problems in setting such requirements, in turn, were caused by insufficient knowledge of the underlying systems. Previous research has argued that knowledge of the underlying systems is required to identify the consequences of different trade-offs and make the best decision regarding overall product per- formances (Brusoni et al., 2001; Takeishi, 2001). According to this argument, component-specific knowledge is a prerequisite for architectural

Francesco Zirpoli and Markus C. Becker

34 R&D Management 41, 1, 2011 r 2010 The Authors

R&D Management r 2010 Blackwell Publishing Ltd

knowledge. This is why component-specific knowledge has such an important role for de- signing the performances of complex products. This importance has been recognized also be- yond the specific context of development of complex products. In their work on knowledge recombination, for instance, Taylor and Greve (2006) reach a conclusion that is fully consistent but even stronger. One of their conclusions from an analysis of innovation in the comic book industry is that combining knowledge requires a deep understanding of knowledge, rather than just information scanning or exposure (Taylor and Greve, 2006: p. 735). Such findings from a different context raise additional questions (pointing to matters of great interest for opera- tional implementation). For instance, to what degree can shadow engineering, listening posts, and the monitoring of technological advances fulfil the requirement to ‘keep abreast’ of the developments of component-specific knowledge? Taylor and Greve’s (2006) finding indicates that understanding, not just possessing the informa- tion, is required. The evidence in our case corro- borates this idea. At Alpha, we have seen that component-specific knowledge fulfils a very pre- cise function in solving performance trade-offs. The evidence showed very clearly that to fulfil that function, a certain depth of understanding is required. Depth of understanding technical spe- cificities (for instance, pertaining to component- specific technology) holds the key to the compe- tence of taking performance trade-offs. Thus, the level of knowledge of component technologies represents a limit to design and engineering out- sourcing.

As it turns out, such depth of understanding appears best provided by learning by doing. Several instances in the Alpha case provide sup- port for the idea that actively carrying out design tasks is crucial for maintaining design compe- tences, through learning by doing. This came to the fore clearly in the case in several instances: for example, in the pre-development phase, where Alpha had relinquished its deep involvement in many technical discussions; in the case of the safety system supplier, where it became clear that no party had a real understanding of the performance attained; and most importantly, in the loss of capability of taking performance trade- offs. The evidence of the Alpha case therefore supports the argument that loss of learning by doing beyond the degree required to maintain component-specific knowledge represents a limit to design and engineering outsourcing.

Our empirical evidence also supports the argu- ment established by previous research that com- ponent-specific knowledge is a complement to the firm’s architectural knowledge (Takeishi, 2002). At the same time, our evidence adds the insight that component-specific knowledge is also a means to strengthen architectural knowledge and provides the opportunity to maintain it even in the face of technological change (Brusoni et al., 2001). Our empirical evidence goes beyond pre- vious research by indicating that maintaining component-specific knowledge in important areas in-house can hold the key to the successful in- volvement of external sources of innovation in the development of complex products. The empirical evidence supports the idea that it will be difficult, if not impossible, to maintain systems integration capabilities – for which performance integration is key – without mastering the underlying compo- nent-specific knowledge. Interestingly, mastering such knowledge is impossible without going into operational details such as the organization of component-development tasks. In fact, as emerged from our interviews, the problems with performance integration that we described did not manifest themselves right from the start, i.e. when the outsourcing strategy was being implemented at the beginning of the 1990s. The reason was simply that Alpha’s engineers had only just handed over components and systems engineering to suppliers. At that point of time, their knowl- edge of underlying systems was still intact, and they could still easily guide the engineering work of suppliers, also during the pre-development phase. Over the next couple of years, however, these competences steadily eroded. By the end of the nineties, they had become almost non-exis- tent, at least for some systems.

The arguments for the importance of compo- nent-specific knowledge we referred to above help to better understand what happened, and the implications. They therefore help to understand the sources of limits in design and engineering outsourcing.

First, because component design was out- sourced, Alpha’s engineers did not have access to learning opportunities any more. They were cut off from learning by doing in component design. Above, we have argued that understanding, rather than just information about components and how they work is required, and that learning by doing is a powerful mechanism that can foster understanding (Levitt and March, 1988; Gavetti and Levinthal, 2000). The case supports the idea that learning by doing opportunities are

The limits of design and engineering outsourcing

r 2010 The Authors

R&D Management r 2010 Blackwell Publishing Ltd

R&D Management 41, 1, 2011 35

important for developing understanding. Cutting engineers off from learning by doing opportu- nities, meant that their understanding of compo- nent specific knowledge started to decay.

Moreover, the case also provides interesting indications on the question to what extent learn- ing by doing opportunities can be substituted. Consider the following: during the pre-develop- ment phase, suppliers (in particular suppliers of important systems) were actually co-located at Alpha’s design and engineering division, provid- ing the occasion for intense communication, for building trustful and personal relations, and for transferring tacit knowledge and competences. In the literature, co-location (Lamming, 1993) is seen as conducive to knowledge transfer and absorp- tive capacity (Cohen and Levinthal, 1990). How- ever, despite such favorable conditions in some systems, the ability of Alpha’s engineers to dom- inate the pre-development phase eroded. Cutting off Alpha’s engineers from both components underlying technology and from learning by doing negatively impacted their ability to design overall vehicle performance in the pre-develop- ment phase. The important insight is that lack of learning by doing can weaken the firm’s absorp- tive capacity, and subsequently, the ability to integrate external knowledge (in this case pro- vided by the suppliers) into its own product development process. The absence of learning by doing opportunities therefore has a double negative consequence.

The example also provides a powerful indica- tion that learning by doing seems to be very difficult to substitute. Learning by doing seems to be particularly able to assure the maintenance of specific design competences over time. Our data show that delegating all component devel- opment work to suppliers weakened the OEM’s ability to take design decisions early on in the development process.

In conclusion, our empirical evidence indicates that some component development activities should be carried out in-house. The reason is that doing seems to be the most important means for acquiring and maintaining component-specific knowledge. In turn, such knowledge plays a key role for deciding performance trade-offs, which are crucial for the development of complex pro- ducts. Once again, our empirical evidence points to operational details (providing guidance on how to implement in practice), which hold the answers to questions of great strategic importance. To what extent, however, component-specific knowl- edge has to be retained in house is a critical issue.

The next section presents the indications on this question that we derived from our field work.

4.4. Implications for firm boundaries

As shown, component-specific knowledge plays a crucial role for being able to make integration mechanisms work. This insight leads to new questions, however. If component-specific knowl- edge is crucial and designing and engineering in- house components cannot be easily substituted as a means for acquiring and maintaining the rela- tive knowledge, which component-specific knowl- edge should be maintained in-house (by keeping component development tasks in-house)? Our empirical evidence provides an answer to this question – a new criterion (beyond that already cited in the literature) as to what component specific knowledge to retain in house. The analy- sis of our empirical evidence indicates a two-fold criterion of what type of knowledge firms that develop complex products should retain in house: develop in house all the specific component tech- nologies that (1) have a direct impact on key product performances and (2) present a high degree of reciprocal interdependences with the key technologies contributing to the overall pro- duct performances. The first criterion refers to the relevance of the component or subsystem, the second indicates when the knowledge concerning how to design and engineer that component or subsystem should be retained in house in order to achieve system integration efficiently and effec- tively. OEMs can choose which component tech- nology to invest in according to their assessment on what the overall performances that matter most for their customers are.

An example drawn from our observations can clarify this point. In the development of a sports car, ‘handling’ is usually one of the key perfor- mance dimensions. Our first criterion would lead Alpha to develop and maintain in-house compo- nent-specific knowledge connected to the ‘hand- ling’ of sports cars. It would not be possible, in fact, to master a sports car architecture (i.e., to have the correct architectural knowledge) without a full command of the components and systems affecting the handling performances. However, these performances are determined by numerous components (e.g. tires, suspensions, steering sys- tem, etc.). Having assessed the importance of handling for the overall product performance achievement (first criterion), our second criterion would suggest that Alpha acquires component-

Francesco Zirpoli and Markus C. Becker

36 R&D Management 41, 1, 2011 r 2010 The Authors

R&D Management r 2010 Blackwell Publishing Ltd

specific knowledge only for the components whose performance present a high degree of interdependence with the rest of the product (in the case of low interdependence, it is possible to rely on black box sourcing, it being possible to specify the interface between the component and the rest of the product and manage the integra- tion of performances either ex ante, through standard (modular) interfaces, or ex post, after minor adjustments). With reference to handling, in fact, Alpha re-acquired some component- specific knowledge about suspensions and not about tires.15

In our view, this argument adds another, more general, criterion that goes beyond the criterion indicated by Takeishi (2002). The key criterion for deciding on which components firms should de- velop component specific knowledge, according to Takeishi, is ‘the nature of component develop- ment project in terms of technological newness’ (Takeishi, 2002: p. i).16 Our criterion includes Takeishi’s notion but indicates that knowledge partitioning demands overlaps between Alpha and the supplier not only in the case of technolo- gical newness but also in all the cases in which knowledge about the component is required for managing complex technical interdependences. In this latter case, as in the case of technological newness, the outcomes of component integration within the product are uncertain and difficult to specify ex ante. This is not only due to the supplier’s lack of knowledge, as suggested by Takeishi in the case of technological newness but also due to the intrinsic complexity of overall product performance integration. We have thus been able to indicate, based on the empirical findings, an operative criterion for where to draw the boundary of the firm’s knowledge (thus contributing to casting light on the mechan- isms by which product architecture and firm boundaries influence each other, Fixson et al., 2005). This operative detail holds the key to unlocking (or not) important strategic conse- quences.

4.5. The role of organizational decisions on limits of outsourcing

Our empirical evidence points to in-sourcing (or keeping in house) component-specific knowledge as the key to the ability to integrate overall vehicle performances. So what insights and advice do we have to offer on how firms can organize in order

to successfully maintain component-specific knowledge and learning opportunities?

As seen in Section 3, Alpha was forced to re- organize the NPD activities on the basis of a new criterion for allocating design responsibilities to suppliers and a different structure of the internal NPD process. This is why the case provides such strong and clear indications for our conceptual framework. Even after re-acquiring component specific knowledge, focusing organizational ef- forts on the integration of physical systems and components was not a sufficient condition for integrating overall systems performance. This forced Alpha to thoroughly re-organize its NPD activities. This re-organization revolved around the distinction between integrating physical sys- tems vs integrating systems performances. The assumption Alpha started with was the following: in designing performances on the level of the overall vehicle, trial-and-error and experimenta- tion with the vehicle as a whole is indispensable.17

(Such an assumption makes sense on the basis of research such as Pisano, 1994 and Gavetti and Levinthal, 2000). We can now appreciate why experimentation and the use of virtual simulation started to play such an important role. Without such experimentation with the vehicle as a whole, there is no guarantee that reciprocal interdepen- dences between the individual performances of systems and components are addressed comple- tely. In such a situation, having a tight hold on the design of the overall product performance is difficult (Sosa et al., 2003). This point links back to the incomplete decomposability of overall product performance, and the impossibility to fully identify the contribution of each individual component and system to overall performance. Because most of the vehicle performance involves reciprocal interactions between many compo- nents and systems that are usually developed by different suppliers, rich interaction and ongoing coordination with suppliers are required. Our findings thus support the literature in this respect (Sako, 2003; Takeishi and Fujimoto, 2003). Moreover, because ongoing coordination, rich interaction, and component-specific knowledge are required, performance integration needs to be recognized as an important organizational task in its own right. This in turn involves, for in- stance, a job position responsible for accomplish- ing it, an organizational unit dedicated to systems integration, important resources, or at the very least making it a dominating criterion in structur- ing the firm. While our findings are consis- tent with recent articles on systems integration

The limits of design and engineering outsourcing

r 2010 The Authors

R&D Management r 2010 Blackwell Publishing Ltd

R&D Management 41, 1, 2011 37

(Brusoni and Prencipe 2006 and the citations therein), they go beyond the level of detail with which the organizational implications are spelled out. A further organizational decision that has important strategic consequences is the point of time at which integration efforts occur. The reason is that the negative consequences of an inappropriate and untimely management of the key performance interdependencies increase as the design process progresses and more specifica- tions become more concrete. In the later stages of a project, in fact, design and engineering con- straints grow as the propagation costs of any component and system re-design increase.18 An- ticipating the design and definition of the decom- position scheme into the pre-development phase therefore enables coordination mechanisms and efforts to intervene at a stage where interdepen- dencies are less strong. Consequently, interdepen- dencies can be dealt with at a point of time when they are less strong than later on. The simple conclusion is that it makes sense to intervene early on in the process. The earlier the problem is tackled, the lesser the need for late re-design to match expected performances (Thomke 1998a, b; Thomke and Fujimoto, 2000). Obviously, this is a matter of interest only in the presence of severe time and cost constraints.

5. Conclusion

This paper has provided a description of the reasons why firms that outsource the development of complex products often experience problems in implementing this strategy. We have identified what limits Alpha, as a manufacturer of a com- plex product, encountered: the erosion of compo- nent-specific knowledge led to the loss of the capability to take performance trade-offs regard- ing performance of the product as a whole. The most important source of the loss of component- specific knowledge was the lack of learning by doing.

Identifying the most important source of limits to design and engineering outsourcing enables deriving insights regarding how to deal with the challenges raised by design and engineering out- sourcing. The main insights in this regard are two.

First, given the incomplete decomposability of complex products, architectural knowledge and the related competence of taking performance trade-offs are achieved if the OEM retains some component specific knowledge in-house. In this respect, our finding shows that the reliance on

modular product architectures cannot be a sub- stitute for maintaining in-house component knowledge development. This result, by itself, shows that there can be intrinsic limitations to the extent to which firms can pursue design and engineering outsourcing. We thus provide a cri- terion according to which firms can choose what component specific knowledge should be held in house. This insight is of relevance both for theory and for practice. The case study provides support for the importance of learning by doing for maintaining competences, and for the comple- mentary role of component-specific and architec- tural knowledge. Managers cannot assume that the firm is able to develop architectural knowl- edge independently from investments in compo- nent-specific knowledge.

Second, we show that even for a given bound- ary of the firm, i.e., when the extent of design and engineering outsourcing is defined, there are still organizational decisions (such as the timing of supplier involvement, the allocation of engineer- ing resources on different phases of the product development process, the use of new technologies in product development, etc.) that have an impact on the effects of outsourcing design and engineer- ing tasks. Management can benefit from a sys- tematic assessment of how and why organizational decisions might influence, repre- sent an obstacle or even make it impossible to implement the strategic choices concerning design and engineering outsourcing. The implication is that in order to reap the benefits of leveraging external sources of innovation in the development of complex products, how the outsourcing of design and engineering tasks is organized is es- sential. In practice, as far as the extent to which the drawbacks of outsourcing are concerned, it makes all the difference whether the outsourcing firm continues to be involved in learning by doing regarding component design or not. These deci- sions can be at least as important – if not far more important – than decisions on product architec- ture and might contribute to restrict the available strategic options (i.e., make or buy choices).

These contributions stimulate further research aiming at a systematic connection and integration of strategic aspects of innovation management and its operational aspects. Of course, our find- ings are somewhat specific to the industry we examined. The most relevant specificities are that cars are not fully decomposable complex systems and that, in the automotive sector, there is a substantial lack of common industry stan- dards or standard architectures for most of the

Francesco Zirpoli and Markus C. Becker

38 R&D Management 41, 1, 2011 r 2010 The Authors

R&D Management r 2010 Blackwell Publishing Ltd

systems involved in a car (MacDuffie, 2008). Our interviews show that in some cases, suppliers have even been forced to change the basic design of a component just because the testing protocol was different from that of the OEM for which the component was originally designed: with different testing procedures, the component would not have been approved although it delivered the expected performances. This makes the issue of system integration more idiosyncratic to each company, and the strategic concern of controlling the industry architecture via a dominating plat- form (Jacobides et al., 2006) less relevant, at least for the time being. As a consequence, our conclu- sions cannot necessarily be applied to contexts where industry standards play an important role in both the development of product architectures and components (e.g. the PC or the telecommu- nication industry). However, our results contri- bute to show that much still remains to be included in current theories, particularly in an integrated theory that covers both strategic and organizational aspects. It is also highly relevant for innovation management practice, as glossing over dimensions and factors that matter can have serious consequences.

References

Arora, A., Fosfuri, A. and Gambardella, A. (2001)

Markets for Technology: The Economics of Innova-

tion and Corporate Strategy. Cambridge MA: The

MIT Press.

Baldwin, C.Y. and Clark, K.B. (2000) Design Rules:

Volume 1. The Power of Modularity. Cambridge,

MA: MIT Press.

Becker, M.C. and Zirpoli, F. (2003) Organizing knowl-

edge integration and co-ordination in New Product

Development. International Journal of Automotive

Technology Management, 3, 1–2, 30–46.

Bensaou, M. (1999) Portfolios of buyer-supplier rela-

tionships. Sloan Management Review, 40, 35–44.

Bidault, F., Despres, C. and Butler, C. (1998) The

drivers of cooperation between buyers and suppliers

for product innovation. Research Policy, 26, 719–

732.

Brusoni, S. (2005) The limits to specialization: pro-

blem-solving and coordination in modular networks.

Organization Studies, 26, 12, 1885–1907.

Brusoni, S. and Prencipe, A. (2001) Unpacking the

black box of modularity: technologies, products

and organizations. Industrial and Corporate Change,

10, 1, 179–205.

Brusoni, S. and Prencipe, A. (2006) Making design

rules: a multi-domain perspective. Organization

Science, 17, 2, 179–189.

Brusoni, S., Prencipe, A. and Pavitt, K. (2001) Knowl-

edge specialization, organization coupling, and the

boundaries of the firm: why do firms know more than

they make? Administrative Science Quarterly, 46, 4,

597–625.

Chesbrough, H. (2003) Open Innovation. New York:

Free Press.

Chesbrough, H. and Crowther, A.K. (2006) Beyond

high tech: early adopters of open innovation in other

industries. R&D Management, 36, 3, 229–236.

Chesbrough, H. and Kusunoki, K. (2001) The mod-

ularity trap: innovation, technology phases shifts and

the resulting limits of virtual organisations. In: Non-

aka, I. and Teece, D. (eds), Managing Industrial

Knowledge. London: Sage Press, pp. 202–230.

Chesbrough, H., Vanhaverbeke, W. and West, J. (2006)

Open Innovation: Researching a New Paradigm. Ox-

ford: Oxford University Press.

Christensen, J.F. (2006) Wither core competency for

the large corporation in an open innovation world?

In: Chesbrough, H., Vanhaverbeke, W. and West, J.

(eds), Open Innovation: Researching a New Paradigm.

Oxford: Oxford University Press, pp. 35–61.

Clark, K.B. (1989) Project scope and project perfor-

mance: the effect on parts strategy and supplier

involvement in product development. Management

Science, 35, 1247–1263.

Clark, K.B. and Fujimoto, T. (1991) Product Develop-

ment Performance. Boston, MA: Harvard Business

School Press.

Cohen, W.M. and Levinthal, D.A. (1990) Absorptive

capacity: a new perspective on learning and innova-

tion. Administrative Science Quarterly, 35, 1, 128–

152.

Davies, A., Hobday, M. and Prencipe, A. (2005)

Systems integration: a core capability of the modern

corporation. Industrial and Corporate Change, 14, 6,

1109–1143.

Eisenhardt, K. (1989) Building theories from case study

research. Academy of Management Review, 14, 4,

532–550.

Eisenhardt, K.M. and Graebner, M.E. (2007) Theory

building from cases: opportunities and challenges.

Academy of Management Journal, 50, 1, 25–32.

Enrietti, A., Follis, M. and Whitford, J. (2002) Im-

proving performance at the second tier of the auto-

motive supply chain. Supply Chain Governance and

Regional Development in the Global Economy,

University of Wisconsin, September 10.

Ernst, D. (2005) Limits to modularity: reflections on

recent developments in chip design. Industry and

Innovation, 12, 3, 303–335.

Ethiraj, S.K. and Levinthal, D.A. (2004) Modularity

and innovation in complex systems. Management

Science, 50, 159–173.

Fine, C.H. (1998) Clockspeed: Winning Industry Con-

trol in the Age of Temporary Advantage. Reading,

MA: Perseus Books.

The limits of design and engineering outsourcing

r 2010 The Authors

R&D Management r 2010 Blackwell Publishing Ltd

R&D Management 41, 1, 2011 39

Fixson, S., Ro, Y. and Liker, J.K. (2005) Modularisa-

tion and outsourcing: who drives whom? A study of

generational sequences in the US automotive cock-

pit industry’. International Journal of Automotive

Technology & Management, 5, 2, 166–183. Special

issue on ‘Knowledge and Task Partitioning in the

Auto Industry: Coordination, Governance and

Learning in New Product Development’.

Freeman, C. (1991) Networks of innovators: a synthesis

of research issues. Research Policy, 20, 499–514.

Frigant, V. and Talbot, D. (2005) Technological deter-

minism and modularity: lessons from a comparison

between aircraft and auto industries in Europe.

Industry and Innovation, 12, 3, 337–355.

Gavetti, G. and Levinthal, D. (2000) Looking forward

and looking backward: cognitive and experiential

search. Administrative Science Quarterly, 45, 1,

113–137.

Grandori, A. and Kogut, B. (2002) Dialogue on orga-

nization and knowledge. Organization Science, 13,

224–231.

Helper, S. (1991) How much has really changed be-

tween U.S. automakers and their suppliers? Sloan

Management Review, 32, 15–28.

Helper, S.R. and Sako, M. (1995) Supplier relations in

Japan and the United States: are they converging?

Sloan Management Review, 36, 77–84.

Henderson, R.M. and Clark, K.B. (1990) Architectural

innovation: the reconfiguration of existing product

technologies and the failure of established firms.

Administrative Science Quarterly, 35, 9–30.

Howells, J., Gagliardi, D. and Malik, K. (2008) The

growth and management of R&D outsourcing: evi-

dence from UK pharmaceuticals. R&DManagement,

38, 2, 205–219.

Iansiti, M. and Levien, R. (2004) The Keystone Advan-

tage. Boston: Harvard Business School Press.

Imai, K., Nonaka, I. and Takeuchi, H. (1985) Mana-

ging the new product development process: how

Japanese learn and unlearn. In: Clark, K.B., Hayes,

R.H. and Lorenz, C. (eds), The Uneasy Alliance:

Managing the Productivity-Technology Dilemma.

Boston: Harvard Business School Press, pp. 337–376.

Jacobides, M., Knudsen, T. and Augier, M. (2006)

Benefiting from innovation: value creation, value

appropriation and the role of industry architectures.

Research Policy, 35, 8, 1200–1221.

Lamming, R. (1993) Beyond Partnership. London:

Prentice Hall Europe.

Langlois, R. (2002) Modularity in technology and

organization. Journal of Economic Behavior and

Organization, 49, 1, 19–38.

Levinthal, D.A. (1997) Adaptation on rugged land-

scapes. Management Science, 43, 934–950.

Levitt, B. and March, J. (1988) Organizational learn-

ing. Annual Review of Sociology, 14, 319–340.

Lincoln, J.R., Ahmadjian, C.L. and Mason, E. (1998)

Organizational learning and purchase-supply rela-

tions in Japan: hitachi, matsushita and toyota com-

pared. California Management Review, 40, 241–264.

MacCormack, A.D., Rusnak, J. and Baldwin, C.Y.

(2008) Exploring the duality between product and

organizational architectures: a test of the mirroring

hypothesis. Scancor/DRUID Conference, Stanford

University, Palo Alto, CA, March 27–28.

MacDuffie, J.P. (2008) Technological and Organiza-

tional Barriers to Modularity: Persistent Integrality

in the Global Automotive Industry. Wharton: Mimeo.

Mahnke, V. (2001) The process of vertical dis-integra-

tion: an evolutionary perspective on outsourcing.

Journal of Management and Governance, 5, 353–379.

Maxton, G. and Wormald, J. (2004) Time for a Model

Change: Re-Engineering the Global Automotive In-

dustry. Cambridge: Cambridge University Press.

Mikkola, J. (2003) Modularity, component outsour-

cing, and inter-firm learning. R&D Management, 33,

4, 439–454.

Mikkola, J. (2006) Capturing the degree of modularity

embedded in product architectures. Journal of Pro-

duct Innovation Management, 23, 2, 128–146.

Miles, M.B. and Huberman, A.M. (1994) Qualitative

Data Analysis: An Expanded Sourcebook. Thousand

Oaks, CA: Sage Publications.

Miozzo, M. and Grimshaw, D. (2005) Modularity and

innovation in knowledge-intensive business services:

IT outsourcing in Germany and the UK. Research

Policy, 34, 9, 1419–1439.

Nishiguchi, T. (1994) Strategic Industrial Sourcing.

New York: Oxford University Press.

Pettigrew, A.M. (1990) Longitudinal field research on

change: theory and practice. Organization Science, 1,

3, 267–292.

Pisano, G.P. (1994) Knowledge, integration, and the

locus of learning: an empirical analysis of process

development. Strategic Management Journal, 15, 85–

100.

Pisano, G.P. and Teece, D.J. (2007) How to capture

value from innovation: shaping intellectual property

and industry architecture. California Management

Review, 50, 1, 278–296.

Powell, W.W., Koput, K.W. and Smith-Doerr, L.

(1996) Interorganizational collaboration and the

locus of innovation: networks of learning in biotech-

nology. Administrative Science Quarterly, 41, 116–

145.

Puranam, P. and Jacobides, M.G. (2006) The Dynamics

of Coordination Regimes: Implications for Organiza-

tional Design, Mimeo, London Business School.

Sako, M. (2003) Modularity and outsourcing. The

nature of co-evolution of product architecture in

the global automotive industry. In: Prencipe, A.,

Davies, A. and Hobday, M. (eds), The Business of

Systems Integration. Oxford: Oxford University

Press, pp. 229–253.

Sako, M. (2004) Supplier development at Honda,

Nissan and Toyota: comparative case studies of

Francesco Zirpoli and Markus C. Becker

40 R&D Management 41, 1, 2011 r 2010 The Authors

R&D Management r 2010 Blackwell Publishing Ltd

organizational capability enhancement. Industrial

and Corporate Change, 13, 2, 281–308.

Sanchez, R. (1997) Modular architectures in the mar-

keting process. Journal of Marketing, 63, 92–111.

Sanchez, R. and Mahoney, J.T. (1996) Modularity,

flexibility, and knowledge management in product

and organization design. Strategic Management

Journal, 17, 63–76.

Siggelkow, N. (2007) Persuasion with case studies.

Academy of Management Journal, 50, 1, 20–24.

Siggelkow, N. and Levinthal, D.A. (2003) Temporarily

divide to conquer: centralized, decentralized, and

reintegrated organizational approaches to explora-

tion and adaptation. Organization Science, 14, 650–

669.

Smitka, M.J. (1991) Competitive Ties. New York:

Columbia University Press.

Sosa, M.E., Eppinger S.D. and Rowles, C.M. (2003)

Identifying modular and integrative systems and

their impact on design team interactions. Journal of

Mechanical Design, 125, 2, 240–252.

Staudenmeyer, N., Tripsas, M. and Tucci, C. (2005)

Interfirm modularity and its implications for product

development. Journal of Product Innovation Manage-

ment, 22, 4, 303–321.

Sturgeon, T.J. (2002) Modular production networks: a

new American model of industrial organization.

Industrial and Corporate Change, 11, 3, 451–496.

Takeishi, A. (2001) Bridging inter- and intra-firm

boundaries: management of supplier involvement in

automobile product development. Strategic Manage-

ment Journal, 22, 403–433.

Takeishi, A. (2002) Knowledge partitioning in the

inter-firm division of labor: the case of automotive

product development. Organization Science, 13, 321–

338.

Takeishi, A. and Fujimoto, T. (2003) Modularization

in the car industry: interlinked multiple hierarchies of

product, production, and supplier systems. In: Pre-

ncipe, A., Davies, A. and Hobday, M. (eds), The

Business of Systems Integration. Oxford: Oxford

University Press, pp. 254–278.

Taylor, A. and Greve, H.R. (2006) Superman or the

fantastic four? Knowledge combination and experi-

ence in innovative teams. Academy of Management

Journal, 49, 723–740.

Teece, D.G. (1986) Profiting from technological inno-

vation: implications for integration, collaboration,

licensing and public policy. Research Policy, 15, 285–

395.

Thomke, S.H. (1998a) Simulation, learning and R&D

performance: evidence from automotive develop-

ment. Research Policy, 27, 55–74.

Thomke, S.H. (1998b) Managing experimentation in

the design of new products. Management Science, 44,

6, 743–762.

Thomke, S.H. and Fujimoto, T. (2000) The effect of

‘front-loading’ problem-solving on product develop-

ment performance. Journal of Product Innovation

Management, 17, 128–142.

Ulrich, K. (1995) The role of product architecture in

the manufacturing firm. Research Policy, 24, 419–

440.

Wheelwright, S. and Clark, K. (1992) Revolutionizing

Product Development. New York: Free Press.

Womack, J.P., Jones, D.T. and Ross, D. (1990) The

Machine that Changed the World, Rawson Ass, New

York.

Yin, R.K. (1994) Case Study Research: Design and

Methods. Thousand Oaks/CA: Sage.

Notes

1. Research funding by the Italian Ministry of Higher

Education and Research (MIUR) under the PRIN

2004 program (project 2004135057_004) and by the

Agence Nationale de Recherche (ANR), France

(‘Jeunes Chercheuses et Jeunes Chercheurs’ pro-

gram, grant no. JC05_44029) is gratefully acknowl-

edged. For purposes of formal assignment,

Sections 1, 4, and 5 were written jointly, Francesco

Zirpoli wrote Section 2.1 – case selection, 2.3 – data

analysis and Section 3.2, and Markus Becker wrote

Section 2.2 – data collection and Section 3.1.

2. For a comprehensive review of the advantages of

involving suppliers in the NPD process, see Bidault

et al. (1998).

3. Chesbrough (2003) has synthesized and popular-

ized the superiority of business models that hinge

on the contribution of external sources of innova-

tion (‘open innovation’). Innovating firms are ad-

vised to invest in their capabilities of drawing on

external sources of innovation. In the same vein,

Iansiti and Levien (2004) observe that the sustain-

ability of a firm’s profitability is linked to the

success of the ecosystem it belongs to. Firms thus

should attempt to achieve an important (‘key-

stone’) position within the ecosystem. Iansiti and

Levien (2004) concur with Chesbrough (2003) that

firms should adapt their business models in order

to make use of external resources and competences.

Expanding the seminal contribution of Teece

(1986), Pisano and Teece (2007) add to the litera-

ture on strategic aspects of innovation management

by pointing to the importance of value appropria-

tion conditions as a guide for a profitable network

innovation strategy.

4. In this article, we follow Baldwin and Clark’s

(2000) definition.

5. As Davies, Hobday and Prencipe write, systems

integrators are firms that ‘bring together high-

technology components, subsystems, software,

skills, knowledge, engineers, managers, and techni-

cians to produce a product in competition with

other suppliers’ (2005, p. 1,110). This requires

‘design[ing] and integrat[ing] systems, while mana-

The limits of design and engineering outsourcing

r 2010 The Authors

R&D Management r 2010 Blackwell Publishing Ltd

R&D Management 41, 1, 2011 41

ging networks of component and subsystem sup-

pliers’ (Davies et al., 2005, p. 1,109).

6. Sanchez and Mahoney’s (1996) idea that organiza-

tion structure follows product architecture has also

been criticized as deterministic. Sako (2003: p. 230)

argued that there is ‘no simple deterministic link

between the type of product architecture and

organization architecture.’ Puranam and Jacobides

(2006), in the same vein, have criticized the ‘carry-

over of ideas’ on technological architecture (includ-

ing interfaces) to organizational architecture.

7. OEM stands for Original Equipment Manufac-

turers. Examples of OEM companies in the auto-

motive industry, the focus of the empirical work

presented in this paper, include firm such as Gen-

eral Motors, Ford, and Toyota. In some industries,

such as electronics, OEMs build products or com-

ponents used in products sold by another company

(often called a value-added reseller, or VAR). In

other industries, they are identified as Original

Design Manufacturer. Here, we refer to the OEM

as the leader of its value chain, i.e. as the final

system integrator.

8. The case, therefore, offers at least as much insight

into industry evolution as the electronics industry –

the paradigmatic example of an industry where

changes are observable in a relatively short time

span (Fine, 1998). This point is a matter of interest

in itself and makes the case intriguing for analyzing

industrial and organizational change. For this

reason, we believe our sampling choice seizes, as

good single-case research does, the ‘opportunities

to explore a significant phenomenon under rare or

extreme circumstances’ (Eisenhardt and Graebner,

2007, p. 27).

9. One of the authors has been carrying out research

on Alpha and its suppliers since 1997. The inter-

views and data-gathering activities carried out

between 1997 and 2006, hence, provided a solid

starting point for a more fine-grained and in-depth

analysis in the empirical work we present in this

paper. Moreover, this background allowed us to

trace back the evolution of the OEM’s strategy,

organization, and competences for product devel-

opment according to the evolution of its design

outsourcing strategy.

10. The intrinsic limitations of case study research

more generally do, of course, also apply (see

Miles and Huberman, 1994, for a comprehensive

discussion.)

11. Especially between 1990 and 1992, the OEM ex-

perimented with so-called driven growth pro-

grammes. Suppliers, which demonstrated the

potential, in terms of quality and cost of their

products, to comply with the new OEM standard,

were provided with continuous managerial support

from the OEM together with tailored contractual

agreements, such as long-term contracts and steady

increase of volumes purchased. In exchange, these

suppliers had to respect an improvement schedule

(Enrietti et al., 2002).

12. Thai is, exogenous to the automotive industry.

13. Most often, the purpose of knowledge sharing was

knowledge acquisition on the part of the OEM.

14. Co-design was interpreted as ‘black box’ sourcing

(Lamming, 1993) and not as a form of joint

development. In the development of a new vehicle,

two major phases can be distinguished: (1) the

pre-development phase: the vehicle is ‘set up,’

including the definition of component and system

characteristics (design architypes and expected

performances), and (2) the development phase:

the design and engineering of the vehicle is spe-

cified in detail. Between the two phases lies the

freeze of the design, and the go-ahead for the

full set of investments required in order to indus-

trialize the vehicle with those particular design

parameters.

15. Please note that for sports cars, tires are not a

completely standard option as could be intuitively

thought. An extreme case is the need for tight

integration between the tires and the rest of the

vehicle in Formula 1, an example of how important

the component-specific knowledge on tires could

be.

16. For regular projects, it is more important for

automakers to have a higher level of architectural

knowledge (how to coordinate various components

for a vehicle) than of component-specific knowl-

edge, which is supposed to be provided by suppli-

ers. However, when the project involves new

technology for the supplier, it is important for the

automaker to have a higher level of component-

specific knowledge to solve unexplored engineering

problems together with the supplier. In innovative

projects, effective knowledge partitioning seems to

demand some overlaps between an automaker and

a supplier, rather than efficient and clear-cut

boundaries (Takeishi, 2002: p. i).

17. In order to see why, take the example of an

automobile and try to put together suspensions,

chassis, engines, etc. and drive the car. The car

would work, but the handling would require

further rounds of trial and error and integration

activities to reach the handling that is desired

(handling can be defined as the dynamic beha-

vior of a vehicle on the road and the related

driving experience this behavior generates for the

customer).

18. Propagation cost identifies the percentage of sys-

tem elements that can be affected, on average,

when a change is made to a randomly chosen

element (MacCormack et al., 2008:p. 12). Our

data confirm that the cost in terms of design and

engineering hours (and eventual delay on the

project schedule) that would follow the change in

a component or system are by far higher in the late

stages of a project.

Francesco Zirpoli and Markus C. Becker

42 R&D Management 41, 1, 2011 r 2010 The Authors

R&D Management r 2010 Blackwell Publishing Ltd

Francesco Zirpoli is an Associate Professor of Management at the Department of Business Eco- nomics and Management at Università Ca’ Fos- cari Venezia and a Research Associate at the International Motor Vehicle Program. He re- ceived his PhD in management from Cambridge University and from Università Federico II di Napoli. His research focuses on the strategic organization of innovation, make or buy strate- gies, and supply chain management with a specific focus on the NPD process. His works have appeared in Research Policy, Journal of Eco- nomic Behavior and Organization, European Management Review, International Journal of Technology Management, International Journal of Operations and Production Management, Journal of Analytical and Institutional Econom-

ics, and International Journal of Automotive and Technology Management.

Markus C. Becker is a Professor at the Strategic Organization Design Unit, University of South- ern Denmark, Odense. He received his PhD in Management from Cambridge University. His works have appeared in Cambridge Journal of Economics, Industrial and Corporate Change, International Journal of Automotive and Tech- nology Management, International Journal of Operations and Production Management, Jour- nal of Business Research, Journal of Economic Behavior and Organizations, Journal of Manage- ment Studies, Research Policy, and others. His current research focuses on strategic organization of innovation and organizational routines.

The limits of design and engineering outsourcing

r 2010 The Authors

R&D Management r 2010 Blackwell Publishing Ltd

R&D Management 41, 1, 2011 43

Copyright of R&D Management is the property of Wiley-Blackwell and its content may not be copied or

emailed to multiple sites or posted to a listserv without the copyright holder's express written permission.

However, users may print, download, or email articles for individual use.

OUS 2- outsourcing , competitive capabilities and performance.pdf

• Outsourcing, competitive capabilities and performance: an empirical study in service firms
• Introduction
• Related literature
• Outsourcing
• Theoretical perspectives on outsourcing
• Outsourcing and firm performance
• Model and hypotheses
• Benefits of outsourcing and its impact on the firm’s competitive capabilities
• Impact on competitive capabilities and their relation to the determination of firm performance
• Methodology--empirical study
• Universe, sample, and type of investigation
• Main scales
• Benefits of outsourcing
• Impact on competitive capabilities
• Measuring firm performance
• Results
• Discussion
• Relationship between the benefits of outsourcing and the impact of outsourcing decisions on competitive capabilities
• Relation between the impact of outsourcing decisions on external competitive capabilities and firm performance
• Importance of this study
• Conclusions
• Academic implications
• Limitations and future lines of research
• References