16
Internalization of R&D outsourcing: An empirical study Sang Yun Han a , Sung Joo Bae b,n a Management of Technology Program, Yonsei University, 50 Yonsei-Ro, Seodaemun-Gu, Seoul 120-749, Republic of Korea b School of Business, Yonsei University, 50 Yonsei-Ro, Seodaemun-Gu, Seoul 120-749, Republic of Korea article info Article history: Received 9 January 2013 Accepted 20 November 2013 Available online 7 December 2013 Keywords: R&D outsourcing Knowledge transfer Organizational composition Absorptive capacity Internalization abstract From the absorptive capacity perspective, this study investigates the extent to which a rm that uses external knowledge attained through R&D outsourcing can increase its performance, and how this effect is moderated by a rm's absorptive capacity via internal R&D efforts and organizational composition of the rm's R&D division. More specically, we use R&D intensity, a traditional measure of absorptive capacity, and ve variables of organizational composition to measure their moderating effect between R&D outsourcing effort and the rm's resulting performance. We use a xed-effect model to analyze panel data from 19,570 Korean manufacturing rms between 2002 and 2007. Our ndings show that the intensity of R&D outsourcing in high technology industries has a direct and positive effect on a rm's performance. We also identify the differences between high- and medium/low-technology industries and analyze how having an R&D staff of highly skilled researchers can moderate the effect of R&D outsourcing on a rm's performance. We nd that for high-technology industries, R&D outsourcing is strongly associated with a rm's positive performance when the ratio of researchers with Ph.D. degrees in R&D organization is high. However, in low- technology industries, our study indicates that while the ratio of researchers in R&D to R&D staff has a direct effect on rm performance, it does not actually moderate the effect of R&D outsourcing on rm performance. We provide an interpretation of these empirical ndings, emphasizing the importance of a rm's absorptive capacity via organizational composition of the R&D division in maximizing R&D outsourcing results. & 2013 Elsevier B.V. All rights reserved. 1. Introduction Scholars and practitioners have acknowledged that R&D outsour- cing is a key and effective strategy to increase a rm's competitiveness by internalizing external expertise. Without absorbing external knowl- edge, the scope of a rm's knowledge, its proprietary technologies, and ability to access external knowledge can be quite limited. There- fore, the generation and transfer of knowledge through external exchanges are essential for a rm to develop and sustain competitive advantage and survive (Foss and Pedersen, 2002; Grant, 1996; Kyläheiko et al., 2011; Mudambi, 2002). Since Chesbrough (2003) introduced open innovation as a key strategy for rm growth, external knowledge has been regarded as an essential element to optimize internal R&D. Many scholars have also noted that external knowledge can be distributed over various players and channels (Acha and Cusmano, 2005; Coombs et al., 2003; Howells et al., 2003; Tether, 2002). In this vein, R&D outsourcing is regarded as one of the most effective strategies for open innovation to make use of external knowledge, among many other options such as technology acquisition, alliances, and R&D cooperation. In the Organization for Economic Co- operation and Development (OECD) countries, business expenditures on external R&D has gradually increased since the 1980s in most developed countries. For instance, in the UK and Germany, business expenditures on external R&D doubled in proportion to total expen- ditures on R&D over a 10-year period (Bönte, 2003; Howells, 1999). In fact, the Continuous Average Growth Rate (CAGR) of internal expen- ditures for R&D was 16% from 2002 to 2007, while the CAGR for funds to outsource R&D from 2002 to 2007 was approximately 8% higher than that of internal R&D (Fig. 1). R&D outsourcing also contributes to a rm's performance. Huang et al. (2009) investigated the impact of outsourcing on a rm in terms of development costs and nancial prots during the new product development (NPD) process. Through an analysis of 121 Taiwanese IT rms, they discovered that R&D outsourcing is effective in lowering development costs and increasing nancial prots. As R&D outsourcing has become a more common practice globally, both the theoretical and empirical literature has advanced in terms of factors that determine the acquisition of external knowl- edge and its effects on a rm's performance. Two of the most important issues identied to date are (1) the determinants of R&D outsourcing and (2) whether external knowledge acquired through Contents lists available at ScienceDirect journal homepage: www.elsevier.com/locate/ijpe Int. J. Production Economics 0925-5273/$ - see front matter & 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.ijpe.2013.12.001 n Corresponding author. Tel.: þ82 2 2123 6578; fax: þ82 2 392 6706. E-mail addresses: [email protected] (S.Y. Han), [email protected] (S.J. Bae). Int. J. Production Economics 150 (2014) 5873

Internalization of R&D outsourcing: An empirical study

Embed Size (px)

Citation preview

Page 1: Internalization of R&D outsourcing: An empirical study

Internalization of R&D outsourcing: An empirical study

Sang Yun Han a, Sung Joo Bae b,n

a Management of Technology Program, Yonsei University, 50 Yonsei-Ro, Seodaemun-Gu, Seoul 120-749, Republic of Koreab School of Business, Yonsei University, 50 Yonsei-Ro, Seodaemun-Gu, Seoul 120-749, Republic of Korea

a r t i c l e i n f o

Article history:Received 9 January 2013Accepted 20 November 2013Available online 7 December 2013

Keywords:R&D outsourcingKnowledge transferOrganizational compositionAbsorptive capacityInternalization

a b s t r a c t

From the absorptive capacity perspective, this study investigates the extent to which a firm that usesexternal knowledge attained through R&D outsourcing can increase its performance, and how this effectis moderated by a firm's absorptive capacity via internal R&D efforts and organizational composition ofthe firm's R&D division. More specifically, we use R&D intensity, a traditional measure of absorptivecapacity, and five variables of organizational composition to measure their moderating effect betweenR&D outsourcing effort and the firm's resulting performance.

We use a fixed-effect model to analyze panel data from 19,570 Korean manufacturing firms between2002 and 2007. Our findings show that the intensity of R&D outsourcing in high technology industrieshas a direct and positive effect on a firm's performance. We also identify the differences between high-and medium/low-technology industries and analyze how having an R&D staff of highly skilledresearchers can moderate the effect of R&D outsourcing on a firm's performance. We find that forhigh-technology industries, R&D outsourcing is strongly associated with a firm's positive performancewhen the ratio of researchers with Ph.D. degrees in R&D organization is high. However, in low-technology industries, our study indicates that while the ratio of researchers in R&D to R&D staff has adirect effect on firm performance, it does not actually moderate the effect of R&D outsourcing on firmperformance. We provide an interpretation of these empirical findings, emphasizing the importance of afirm's absorptive capacity via organizational composition of the R&D division in maximizing R&Doutsourcing results.

& 2013 Elsevier B.V. All rights reserved.

1. Introduction

Scholars and practitioners have acknowledged that R&D outsour-cing is a key and effective strategy to increase a firm's competitivenessby internalizing external expertise. Without absorbing external knowl-edge, the scope of a firm's knowledge, its proprietary technologies,and ability to access external knowledge can be quite limited. There-fore, the generation and transfer of knowledge through externalexchanges are essential for a firm to develop and sustain competitiveadvantage and survive (Foss and Pedersen, 2002; Grant, 1996;Kyläheiko et al., 2011; Mudambi, 2002). Since Chesbrough (2003)introduced open innovation as a key strategy for firm growth, externalknowledge has been regarded as an essential element to optimizeinternal R&D. Many scholars have also noted that external knowledgecan be distributed over various players and channels (Acha andCusmano, 2005; Coombs et al., 2003; Howells et al., 2003; Tether,2002). In this vein, R&D outsourcing is regarded as one of the mosteffective strategies for open innovation to make use of externalknowledge, amongmany other options such as technology acquisition,

alliances, and R&D cooperation. In the Organization for Economic Co-operation and Development (OECD) countries, business expenditureson external R&D has gradually increased since the 1980s in mostdeveloped countries. For instance, in the UK and Germany, businessexpenditures on external R&D doubled in proportion to total expen-ditures on R&D over a 10-year period (Bönte, 2003; Howells, 1999). Infact, the Continuous Average Growth Rate (CAGR) of internal expen-ditures for R&D was 16% from 2002 to 2007, while the CAGR for fundsto outsource R&D from 2002 to 2007 was approximately 8% higherthan that of internal R&D (Fig. 1).

R&D outsourcing also contributes to a firm's performance.Huang et al. (2009) investigated the impact of outsourcing on afirm in terms of development costs and financial profits during thenew product development (NPD) process. Through an analysis of121 Taiwanese IT firms, they discovered that R&D outsourcing iseffective in lowering development costs and increasing financialprofits.

As R&D outsourcing has become a more common practiceglobally, both the theoretical and empirical literature has advancedin terms of factors that determine the acquisition of external knowl-edge and its effects on a firm's performance. Two of the mostimportant issues identified to date are (1) the determinants of R&Doutsourcing and (2) whether external knowledge acquired through

Contents lists available at ScienceDirect

journal homepage: www.elsevier.com/locate/ijpe

Int. J. Production Economics

0925-5273/$ - see front matter & 2013 Elsevier B.V. All rights reserved.http://dx.doi.org/10.1016/j.ijpe.2013.12.001

n Corresponding author. Tel.: þ82 2 2123 6578; fax: þ82 2 392 6706.E-mail addresses: [email protected] (S.Y. Han), [email protected] (S.J. Bae).

Int. J. Production Economics 150 (2014) 58–73

Page 2: Internalization of R&D outsourcing: An empirical study

R&D outsourcing increases performance. Several studies provideinsights on these two issues. Love and Roper (2001) investigatedthe UK's innovative manufacturing industry to discover that the scaleof plant and R&D input, and appropriability conditions are the keydeterminants of R&D boundary decisions. On the effects of R&Doutsourcing, Howells (1999), Caudy (2001), and Watanabe and Hur(2004) pointed out that R&D outsourcing can help maximizeinnovation and overall firm performance when properly plannedand executed.

Although the determinants and effects of R&D outsourcing arediscussed frequently in the literature, organizational factors influen-cing the effects have received relatively little attention from scholars.Rather than focusing on why firms choose R&D outsourcing, weexamine factors that influence the effect of R&D outsourcing on firmperformance. In attempting to do so, we discuss the absorptivecapacity perspective of R&D outsourcing, which helps to explain howan internalizing mechanism can moderate the R&D outsourcingresults. Absorptive capacity is defined as “the ability of a firm torecognize the value of new information, assimilate it, and apply it tocommercial ends” (Cohen and Levinthal, 1990). In this study, wediscuss two sources of absorptive capacity— internal R&D intensity, asource of absorptive capacity traditionally identified, and organiza-tional composition, which is a newly discovered source of absorptivecapacity in this study. The organizational composition of a firm's R&Ddepartment is brought into the discussion of absorptive capacitysince, from the human capital perspective, the real actors intransferring and internalizing the knowledge generated from R&Doutsourcing are internal R&D employees. Therefore, the main goal ofthis study is to investigate the moderating effect of absorptivecapacity via organizational composition, between R&D outsourcingand firm performance. We also examine the direct effect of R&Doutsourcing on firm performance and the moderating effect ofinternal R&D intensity between these main variables, which ties thisstudy to the literature on absorptive capacity and R&D outsourcing.

On the empirical side, this study constructs an accurate frame-work to examine the impact of R&D outsourcing on firm perfor-mance with panel data that can control firm effects and timeeffects. The results enrich our understanding of the relationshipbetween R&D outsourcing and firm performance, and identifythe moderating effect of absorptive capacity via organizationalcomposition.

The remainder of the paper is organized as follows. The nextsection reviews literature on R&D outsourcing strategies as well astechnology transfer and internalization through R&D outsourcing.Next, we propose our conceptual model and hypotheses. Followingthis, in the research methodology section, we address how we

constructed the variable, the data set, the research method andthe empirical models that we built. The final section consists of theempirical results and a discussion of our findings and our study'slimitations.

2. Literature review

2.1. Why do firms engage in R&D outsourcing?

Based on the recent open innovation paradigm (Chesbrough,2003), we can categorize different strategies that firms employ toacquire and internalize technological knowledge. Firms can eitherchoose to internalize R&D (i.e., develop and support their internalR&D division) to develop their own technology or outsource R&Dto acquire external knowledge. With an external entity, firms canalso opt to form cooperative organizations, such as R&D consor-tiums, R&D joint ventures, research contracts, or licensing agree-ments with other organizations. These strategies aim to increaseR&D efficiency by acquiring and internalizing external knowledge.

Many firms have a keen interest in acquiring external knowledgebecause it helps them establish a competitive advantage throughtechnological innovation, which is becoming increasingly importanttoday (Bierly III et al., 2009). For technology advancement in R&D,using network and cooperation are crucial factors (Von Hippel, 1988).From this perspective, R&D outsourcing is one of the key measures ofopen innovation because external knowledge, which makes innova-tion possible, is usually acquired through a network of firms.

Why firms choose a certain strategy has long been an importantquestion for scholars engaged in the subject of strategy andinnovation. Since vast literature on this subject provides the basisfor the thesis of this paper, we summarize and update the discus-sion in this section.

As we discuss below, ample theoretical literature focuses on thechoice between R&D outsourcing and internal R&D, including theclassical MAKE or BUY decision-making. Literature on this subjectasks why firms engage in R&D outsourcing. Recent research ontheir motivations can be classified into three different perspec-tives. The first is the Transaction Cost theory (Brusoni et al., 2001;Howells, 1999; Narula, 2001; Yasuda, 2005). Originally, transactioncost was first addressed by Ronald H. Coase (1937). The CoaseTheorem suggests that “there must be costs in using the marketthat can be eliminated by using the firm” (Besanko et al., 2009).Costs often involve the time and expense of negotiation, writingand enforcing contracts between buyers and suppliers. Therefore,firms are established to eliminate costs and efficiently transactwith others. From a transaction-cost point of view, sourcingexternal knowledge can be substituted for internal R&D. Inconsidering costs and risks, firms opt for either a make or a buystrategy (Beneito, 2003; Veugelers and Cassiman, 1999). Thus,firms can choose either internal or external innovation strategies,and consequently, they must also decide which technologies todevelop internally or externally (Vega-Jurado et al., 2009).

The second research perspective that addresses the question ofwhy firms engage in R&D outsourcing is the Core Competenceperspective (Prahalad and Hamel, 1990). The core competenceperspective essentially means that firms develop internal R&D orconduct R&D outsourcing to increase their technological competenceand to supplement each other. This view posits that firms with highlevels of R&D competence can more readily enhance their technolo-gical competencies to achieve competitive advantage than firms withlower levels of R&D competence (Brook and Plugge, 2010).

The third perspective is the Resource-Based View (RBV), whichsuggests that firms can increase performance by using theirresources efficiently (Leiblein and Miller, 2003). This perspectiveviews firms as bundles of resources that can outsource R&D when

Fig. 1. Outsourcing funding vs. internal funding for R&D.Source: Report in the survey of Research and Development in Korea-Manufacturing(2002–2007).

S.Y. Han, S.J. Bae / Int. J. Production Economics 150 (2014) 58–73 59

Page 3: Internalization of R&D outsourcing: An empirical study

they need additional resources for innovation that they do notpossess internally.

While these perspectives provide a theoretical standpoint forunderstanding why firms decide to outsource R&D, many subse-quent empirical studies have focused on various factors that affectthe decision-making process to acquire external knowledgethrough R&D outsourcing. Tidd and Trewhella (1997) noted thatR&D outsourcing is a more efficient strategy than building therequired skills internally when internal capabilities are lacking.Veugelers and Cassiman (1999) found that large firms are morelikely than small firms to combine both internal and externalknowledge in their innovation strategy. Yoshikawa (2003) identi-fied key factors such as time pressure and the importance oftechnology that affect the choice to acquire external technology.Two studies (Howells et al., 2008, 2004) found that the lack of in-house R&D and technical expertise are determinants of R&Doutsourcing and that firms that outsource R&D tend to reducedevelopment time and time to market.

Meanwhile, Miyamoto (2007) identified several determinantsof R&D outsourcing. He discovered that firms belonging to a widercorporate group are more active in executing R&D outsourcingactivities. Similarly, diversification strategies such as the expansionof product and sales markets tend to stimulate firms’ R&D out-sourcing behavior.

These various perspectives presented above describe why firmsengage in R&D outsourcing. However, the above literature islacking on how firms can maximize internal competencies andprofits from outsourcing R&D and what role absorptive capacityplays in internalizing the R&D outsourcing outcome. To fill thisgap, we incorporate the concept of absorptive capacity via orga-nizational composition to directly answer these questions. Byinternalization, we mean the overall process of R&D outsourcingwhere a firm first defines the problem and then contracts it out,and then evaluates the results generated by the outsourced firm.The organizational composition of an R&D unit is a mechanismstrongly related to the internalization process. In this way, firmscan increase their absorptive capacity via organizational composi-tion to enhance the results of outsourcing R&D. In the followingsection, we describe the conceptual framework of the internaliza-tion process of R&D outsourcing results and absorptive capacity inR&D outsourcing. This framework underpins our reasoning onwhyand how organizational composition plays an important role ininternalizing R&D outsourcing outcomes.

2.2. Technology transfer and internalization through R&Doutsourcing

Scholars and practitioners have argued that organizations learnnot only from their direct experience but also from the experienceof other organizations (Huber, 1991; Levitt and March, 1988).External knowledge comes from three primary sources. The firstis customers. Generally, customers are the primary source ofexternal knowledge and innovation (Von Hippel, 1988). Thesecond is competitors. Firms usually monitor and analyze theircompetitors’ processes, products and innovations. The result ofthis benchmarking may lead to internal innovation (Bierly III andChakrabarti, 1996; Ghoshal and Westney, 1991). The third source isexternal organizations, which can directly support the focal firmand supply sophisticated knowledge. These external organizationsinclude other firms in the same or different industries, universi-ties, and public research institutes. The external expertise of theseorganizations can enhance the focal firm's skills and competitive-ness (Hamel and Prahalad, 1994; Mowery et al., 1996). For ourstudy, we focus solely on this third source of external knowledge,specifically, R&D outsourcing, which can be described as formally

establishing a partnership with others that have an expertise in aspecific R&D area.

Through R&D outsourcing, technologies — both embodied anddisembodied — are transferred from an external expert to the focalfirm. Argote (1999) also noted that the channel of knowledge hasdifferent sources, including people, technology, and the structure ofthe recipient organization. With R&D outsourcing, demanded knowl-edge or technologies are transferred from outside companies specia-lized in R&D, universities, or government R&D institutions.

However, R&D outsourcing is much more than the meretransfer of knowledge or technology. Fig. 2 shows how weconceptualize R&D outsourcing. In the process of R&D outsourcing,complexity and uncertainty matter when firms define the problemand contract with a supplier (Argote, 1999). Zander and Kogut(1995) found that codified knowledge that could be readily taughtto organizational members transferred more easily than knowl-edge that was not codified or readily taught.

As shown in Fig. 2, from the onset of R&D outsourcing, firmsneed to clearly define the problem they will outsource. This initialstage may be critical in R&D outsourcing because outsourcers maynot be able to properly identify the problem and thus the outcomecan be useless, irrelevant to the firm's needs.

Many studies note that the more the technology can becodified, such as in a blueprint, a process, a formula or rules, theeasier it is to be contracted out (Narula, 2001; Tidd and Trewhella,1997; Yasuda, 2005). Related to this, Kessler et al. (2000) alsosuggested that R&D outsourcing can have hidden costs. One suchcost is the coordination cost that is incurred when firms attempt tointegrate external knowledge into their knowledge base. The costof R&D outsourcing can be high if the external technology isdifficult to interpret or understand (Huang et al., 2009). For thisreason, a certain technology that cannot be easily codified or thatincorporates a high degree of tacit knowledge is more feasiblydeveloped by internal R&D, rather than being outsourced (Narula,2001; Tidd and Trewhella, 1997). Therefore, defining the problemis a significant part of the overall R&D outsourcing process.However, this highly cognitive process can be also very difficultto measure. Therefore, we focus solely on the later stage of theR&D outsourcing — the internalization process.

After the R&D outsourcing supplier solves the problem anddelivers the outcome the recipient firm then begins the inter-nalization process. The firm then interprets it using existinginternal knowledge, or it generates new knowledge combiningexisting and external knowledge, including the outsourced results.When the problem is easy to solve and thus well-defined, theinterpretation process is mainly about combining the outsourcedresults into the ongoing R&D efforts. However, if the problembecomes more complex and ill-defined, the R&D organization ofthe focal company should invest much effort in reinterpretation.We define this process as solution interpretation, a very importantstep for completing the R&D outsourcing project. However, with-out a proper level of internal organizational capability, thisexternal knowledge can be difficult to interpret. During thesolution interpretation process, complexity and uncertainty affectthe success of internalization, and can arise from a firm's inability

Fig. 2. Conceptual framework of technology transfer in R&D outsourcing.

S.Y. Han, S.J. Bae / Int. J. Production Economics 150 (2014) 58–7360

Page 4: Internalization of R&D outsourcing: An empirical study

to understand and interpret external knowledge to generate newinternal knowledge for innovation.

The ability to interpret external knowledge primarily dependson an organization's human resources and organizational struc-ture. Allen (1977) suggested that employees can be the mosteffective interpreters of information because they can restructureand reinterpret information. Therefore, a firm's R&D employeescan help reduce the complexity and uncertainty in R&D out-sourcing through effective solution interpretation. We focus onthe interpretation process that involves a firm's R&D employees’prior knowledge, and we investigate the role of R&D employees’prior knowledge according to Cohen and Levinthal (1990) defini-tion. They noted that absorptive capacity is “the ability of a firm torecognize the value of new information, assimilate it, and apply itto commercial ends,” suggesting that absorptive capacity is largelya function of the firms’ existing knowledge.

With this perspective in mind, we first investigate the role ofabsorptive capacity using this traditional definition. We also discussa new dimension of the absorptive capacity — absorptive capacityvia organizational composition, which occurs when firms engage inR&D outsourcing. In the next section, we present the theoreticaldevelopment of this new dimension of absorptive capacity.

2.3. Absorptive capacity and organizational composition

Since the introduction of absorptive capacity (Cohen andLevinthal, 1990), various conceptualizations of absorptive capacityhave emerged (Lane et al., 2006, 2001; Lane and Lubatkin, 1998;Todorova and Durisin, 2007; Zahra and George, 2002). The initialconcept of absorptive capacity focuses on the ability to recognizethe value of knowledge using other firms’ past experience, andthen assimilate and apply it (Cohen and Levinthal, 1990). Later,Zahra and George (2002) reviewed related literature and reframedabsorptive capacity as a set of organizational routines and pro-cesses that can produce a firm's dynamic organizational capability.Zahra and George laid out four dimensions of absorptive capacity— acquisition, assimilation, transformation, and exploitation. Bymastering these four dimensions, a firm can gain the ability tobetter adapt to changing market conditions for competitiveadvantage and organizational change (Spithoven et al., 2011;Vega‐Jurado et al., 2008). These dimensions are categorized aspotential or realized capacity. Acquisition and assimilation ofknowledge are considered potential capacity, and transformationand exploitation of knowledge are realized capacity. The develop-ment of Zahra and George's (2002) research resulted in thiscategorization based on the social integration mechanism, and isgrounded in the concept that all four dimensions of absorptivecapacity involve social interactions. This suggests that absorptivecapacity can be affected by the interplay of the social integrationmechanism (Spithoven et al., 2011; Todorova and Durisin, 2007;Zahra and George, 2002). In 2007, Todorova and Durisin (2007)introduced a refined model. They first reintroduced “recognizingthe value of” and an alternative understanding of transformationbased on learning theories. The second dimension of Todorova andDurisin (2007)'s model is the supposition about the absorptivecapacity of contingency factors; they propose another contingencyfactor, the power relationship that simultaneously influences thevalue and exploitation of new knowledge. The third dimension isthe feedback loops in a dynamic model of absorptive capacity. Aswe discussed above, the theoretical development of the absorptivecapacity concept has been extended to include two new mechan-isms, a social integration and a political mechanism.

Drawing on these studies, we consider who —what players in thefirm — conduct and participate across the entire process of absorp-tive capacity. Employees of the firm's R&D organization are consid-ered the main actors in R&D outsourcing because the transferred

external knowledge is merged and internalized with related internalknowledge mainly by R&D employees of the firm. Hence, we nowmove on to the discussion of the literature on organizationalcomposition of the firm's R&D division.

Although the firm's R&D division's composition can be a crucialfactor for absorptive capacity, research examining the role of R&D'sorganizational composition is scarce. Instead, many studies havefocused on R&D intensity relative to sales (Cohen and Levinthal,1990; Escribano et al., 2009; George et al., 2001; Kostopoulos et al.,2011; Rothaermel and Alexandre, 2009; Stock et al., 2001; Tsai,2001; Xia, 2013; Zahra, 1996; Zahra and Hayton, 2008) or numberof patents (Austin, 1993; Cohen and Levinthal, 1990; Lin et al.,2012; Zahra and George, 2002); others investigate firms’ level ofability a firm's R&D division's ability to value and apply knowledgeas a proxy of absorptive capacity using survey method (Bagchiet al., 2013; Chen, 2004; Clausen, 2013; Lund Vinding, 2006;Spithoven et al., 2011). A notable exception is Rothwell (1992)'sstudy that found that a firm's links to external scientific andtechnical knowledge sources were effective in absorbing knowl-edge only if the organization was well prepared and had a skilledscientific and technical staff.

In reality, a firm's R&D organization usually consists of employ-ees with various levels of skills, education, experience, andresponsibilities. For example, distinctions exist between aresearcher and research assistant, and a researcher with a Ph.D.degree vs. a master's degree, and full-time vs. part-time employ-ees. We take this a step further. We hypothesize that not only is afirm's financial performance affected by the firm's absorptivecapacity via R&D intensity, but also by its R&D division's organiza-tional composition in terms of the above mentioned distinctions.Possessing a higher level of skill and knowledge will give the firm'sR&D organization more cognitive capabilities to internalize theexternal knowledge gained from R&D outsourcing. The rest of thispaper investigates these issues empirically in depth.

In the following section, we describe the conceptual frameworkand hypotheses that we generated to investigate the role of thefirm's R&D division's organizational composition plays in linkingR&D outsourcing to firm performance.

3. Conceptual framework and hypotheses

3.1. Effects of R&D outsourcing on firm performance

The empirical investigation we conduct in this study looks atthe direct effect of R&D outsourcing on firm performance. Athorough review of published empirical studies on R&D outsour-cing effects offer mix findings (Bergman, 2011).

The first perspective is that R&D outsourcing increases firmperformance. Before reviewing the literature on the relationshipbetween R&D outsourcing and performance, we established therelationship between technological knowledge and firm performanceespecially since R&D outsourcing is the effort to increase the firm'stechnological knowledge via an external entity. This causal linkage isbased on firm-level economic theories of technological change, suchas the endogenous growth theory that suggests that a firm's produc-tivity growth is an outcome of expanding technological knowledge(Griliches, 1986). Firm-level theories of technological change suggestthat innovation is an outcome of an increase in a firm's knowledgebase via investments in internal R&D (Collis, 1994; Hall, 1992; Lenoxand King, 2004; Pakes and Schankerman, 1984; Schmidt, 2010). TheSchumpeterian theory posits that R&D is a crucial contributor toincreasing the firm's productivity (Aghion and Howitt, 1992; Moweryand Oxley, 1995). We identified three key sources of performancegrowth — R&D-induced innovation, technology transfer, and R&D-based absorptive capacity (Mangematin and Nesta, 1999). Many

S.Y. Han, S.J. Bae / Int. J. Production Economics 150 (2014) 58–73 61

Page 5: Internalization of R&D outsourcing: An empirical study

studies have extensively investigated the relationship between thefirm's knowledge investment and its performance (Carter, 1989; Collis,1994; Schmidt, 2010).

A firm's technological knowledge can also derive from externalentities including cooperative projects and outsourcing. Cohen andLevinthal (1990) noted that as a firm expands its internal knowl-edge and technological capability, it also enhances its ability toabsorb and utilize external knowledge (Schmidt, 2010). However,firms can also use various strategies to acquire external knowledgesuch as R&D outsourcing, R&D alliances, R&D cooperation, and thedirect buying of technology. Thus, R&D outsourcing can positivelyaffect the firm's performance by increasing its level of technolo-gical knowledge. As a result, the external knowledge gainedthrough R&D outsourcing may increase the firm's performanceby increasing its level of technological knowledge.

Many scholars have empirically examined this relationship andsupport our view that external knowledge increases the firm'sperformance by adding complementary resources and technologycapabilities gained from external expertise (Chesbrough, 2003;Kessler et al., 2000; Nohria and Garcia-Pont, 1991; Teece, 1986;Tidd and Trewhella, 1997; Yasuda, 2005). Bönte (2003) investi-gated the productivity effects of investment in external vs. internalR&D utilizing 26 samples of German manufacturing industriesusing total factor productivity estimation analysis from 1980 to1993. The results provide strong evidence of a positive relationshipbetween the ratio of external R&D to total R&D and productivity.Bönte (2003) also examined the productivity impact of internaland external R&D using an industry-level panel data set and founda positive relationship between the share of external R&D andproductivity. Guellec and Van Pottelsberghe de la Potterie (2004)also estimated the long-term impact of R&D outsourcing on themulti-factor productivity growth of 16 countries from 1980 to1998, and found that R&D outsourcing was a significant factor indetermining the rate of long-term productivity growth. With aslightly different outcome measure, Schmiedeberg (2008) foundthat contracted R&D is related to the focal firm's patenting, with alarger effect than internal R&D. The regression is conducted withcross-sectional 689 firm level data of the German manufacturingsector using objective performances such as patents and sales ofnew products.

On the contrary, just a few empirical studies found that R&Doutsourcing may not be related to a firm's performance. Gilley andRasheed (2000) and Kessler et al. (2000) found that R&D out-sourcing may not increase a firm's profitability or performance.Gilley and Rasheed (2000), using a regression analysis with usingsurvey data of 90 manufacturing firms, found that performancewas mainly measured by stability of staff, process innovation,product innovations, and employee compensation. Kessler et al.(2000) studied 75 new product development projects from 10large, U.S.-based companies in several different industries usingthe survey method. They measured performance based on innova-tion speed and competitive success of projects in each firm andfound that external sourcing negatively affected innovation speedand competitive success based on when the knowledge wastransferred. Cassiman and Veugelers (2002) investigate the effectof external technology sourcing on firms’ performance based on asample from the Taiwanese Technological Innovation Survey thatincludes 753 firms of low and medium technology. Using aregression analysis, they found that external technology outsour-cing does not contribute significantly to performance. Cassimanand Veugelers’ performance measure was the firm's turnoverattributable to technologically improved or new products.

In summary, our extensive survey of published empiricalstudies on the relationship between R&D outsourcing and firmperformance produced mixed results; in some studies, externalR&D was found to have a large positive effect than internal R&D,

but in other studies, it was smaller, and often not as significant asin others (Bergman, 2011). This suggests that the relationshipbetween R&D outsourcing and firm performance remainsinconclusive.

In addition, a limited number of studies use panel data tocontrol for firm effects and time effects. In most studies, hypoth-eses on positive and negative relationships are investigatedprimarily with secondary data. Most of these studies asked surveyrespondents whether firm performance had increased by R&Doutsourcing. Some issues arose that could have caused bias, suchas respondents’ memory loss, recent effect, and R&D time lag. Forexample, respondents could have had different perceptions of thetime lag from R&D outsourcing to performance. We designed ourempirical analysis to overcome these limitations.

Therefore, we present the effect of R&D outsourcing on firmperformance as the first hypothesis to explore, based on theendogenous growth theory and other studies that assumed apositive relationship between R&D outsourcing and firm perfor-mance. We also assume that R&D outsourcing can be a crucialfactor for improving firm performance.

Hypothesis 1. (H1): Firms with high levels of R&D outsourcingachieve greater performance than those that engage in compara-tively small levels of or no R&D outsourcing.

3.2. Moderating effect of absorptive capacity for internalization

The literature clearly indicates that firms are more likely to takeadvantage of external knowledge when they have high levels ofabsorptive capacity (Cohen and Levinthal, 1990). In other words,for successful internalization of external knowledge, firms mustpossess enough ability to understand it and integrate it with theircurrent knowledge (Clausen, 2013; Mellat-Parast and Digman,2008; Schneider, 1987; Tsai, 2001). In this vein, many studies havefound that a high level of absorptive capacity strengthens thefirm's competitive advantage and leads to valuable organizationaloutcomes such as learning (e.g., additional knowledge), innovation(e.g., new products and processes), and increased financial perfor-mance (George et al., 2001; Mellat-Parast and Digman, 2008;Mowery et al., 1996; Spanos and Voudouris, 2009).

Drawing on this perspective, many studies investigate the mod-erating effect of absorptive capacity on performance when firmsintegrate external knowledge (Jones et al., 2001; Tsai and Wang,2008; Zahra and Hayton, 2008). Studies that have analyzed the directeffect and moderating effect of absorptive capacity on a firm'sperformance use the intensity of internal R&D investment relativeto sales as a proxy of absorptive capacity (Cohen and Levinthal, 1990;Escribano et al., 2009; Jones et al., 2001; Lin et al., 2012; Tsai andWang, 2008). Related to this, Kessler et al. (2000) explained that R&Doutsourcing is in fact an external learning process. They argued thatthe internalization process — how firms interpret external knowl-edge to generate new ideas for innovation using their existinginternal knowledge — is significant. This means that firms mayexperience difficulties in learning and internalizing the externalknowledge without absorptive capacity (Chen, 2004). Mowery(1984) presented a similar view by pointing out that firms canabsorb the output of external knowledge better if they performappropriate level of internal R&D investment. In the process ofidentifying and using external technological knowledge, internalR&D effort plays a positive role to enhance the process (Cohen andLevinthal, 1989; Kim, 1999; Lane and Lubatkin, 1998). Tsai and Wang(2008) explored the extent to which external technology acquisitionaffected firms’ performance, and how this effect is moderated byinternal R&D efforts. They focused on internal R&D input which isdescribed as ‘absorptive capacity’. Zahra and Hayton (2008) alsonoted that absorptive capacity moderates the relationship between

S.Y. Han, S.J. Bae / Int. J. Production Economics 150 (2014) 58–7362

Page 6: Internalization of R&D outsourcing: An empirical study

level of external knowledge and firm performance using data from217 global manufacturing firms.

In our study, these findings provided a crucial foundation uponwhich to conceptualize the moderating roles of absorptive capa-city, via internal R&D, on the relationship between externaltechnology acquisition and firm performance. To investigate thisrelationship, we initially proposed that the absorptive capacity viainternal R&D effort plays a significant role in moderating the effectof R&D outsourcing on firm performance.

Hypothesis 2-1. (H2-1): R&D outsourcing is more strongly asso-ciated with firm performance when the level of absorptivecapacity via internal R&D effort is higher.

As mentioned above, the ability of the firm to internalizeexternal knowledge via absorptive capacity can influence theextent to which it can achieve higher performance from R&Doutsourcing. This absorptive capability relies on a firm's internalcapabilities, such as internal R&D, production experience, andtechnical expertise. In reality, the internalization of externalknowledge through R&D outsourcing is performed by a firm'sR&D staff who conduct an internal R&D after R&D outsourcingresults are delivered. Thus, R&D employees are the key determi-nants of R&D outsourcing outcomes. In this vein, the performanceof R&D outsourcing might depend on not only the level of R&Doutsourcing but also on the composition of employees in the firm'sR&D division.

An underlying assumption in this area is that absorptivecapability involves socially complex routines that can be valuableorganizational resources (Collis, 1994; Hall, 1992, 1982). Forexample, Mangematin and Nesta (1999) argue that highly edu-cated employees in firms will increase the knowledge stock of anorganization. Similarly, Carter (1989) also argue that employeeswith high levels of education are the main contributors to know-how trading, because they possess a high level of knowledge. Inthis context, operationalizing internal R&D efforts with internalR&D investment is quite limited since various organizationalarrangements will be overlooked if the internal R&D investmentis the sole proxy for internal R&D efforts. This suggests thatspecific organizational variables must be considered simulta-neously when investigating the effect of R&D outsourcing on firmperformance.

Since Cohen and Levinthal (1990)'s study, absorptive capacityhas been operationalized in various ways: as R&D intensity relativeto sales (Cohen and Levinthal, 1990; Escribano et al., 2009; Georgeet al., 2001; Kostopoulos et al., 2011; Rothaermel and Alexandre,2009; Stock et al., 2001; Tsai and Wang, 2008; Tsai, 2001; Xia,2013; Zahra, 1996; Zahra and Hayton, 2008), number of patents(Austin, 1993; Cohen and Levinthal, 1990; Lin et al., 2012; Zahraand George, 2002), whether or not a firm has an R&D department(Becker and Peters, 2000; Cassiman and Veugelers, 2002; Nietoand Quevedo, 2005), and the level of ability to value and applyknowledge through a survey (Bagchi et al., 2013; Chen, 2004;Clausen, 2013; Lund Vinding, 2006; Spithoven et al., 2011). Joneset al. (2001) explored the moderating effects of internally availableresources on the relationship between external technology acqui-sition and firm performance. However Jones et al.'s study did notspecify what constitutes internally available resources.

There have been increasing critiques on the operationalizationof absorptive capacity (Spithoven et al., 2011). They assert thatabsorptive capacity is a multidimensional concept and should beoperationalized as such (Lenox and King, 2004; Schmidt, 2010).Accordingly, some studies have focused on the human capitalinvolved in the internalization of external knowledge (LundVinding, 2006). Nevertheless, only a few empirical studies usethe features of human capital in internalizing external knowledge.

Although some studies have examined the relationship betweenabsorptive capacity from the human capital perspective and firms’performance, they have done so in a very limited manner. Somestudies only suggest the concept of absorptive capacity from thehuman capital perspective (Glass and Saggi, 1998; Keller, 1996),while others simply measure the number of employees with auniversity education (Grimpe and Sofka, 2009; Liu and White,1997) or the ratio of R&D employees to the total number ofemployee (Spanos and Voudouris, 2009). In light of this shortfall,we focus on the composition of a firm's R&D staff for the optimalinternalization of external knowledge from a holistic perspective.We posit that how a firm arranges its R&D organization foreffective internalization of external knowledge affects the out-come of R&D outsourcing.

Following this line of reasoning, what we attempt to investigateempirically is the moderating effect of absorptive capacity viaorganizational composition on the relationship between R&Doutsourcing and its outcome. To investigate the moderating roleof absorptive capacity via organizational composition, we focus onthe abilities of the R&D staff. Logically, when organizationalmembers have high levels of knowledge and responsibilities, theyare more dedicated to their work and have greater capabilitieswhich in turn positively affect their organization. The responsi-bility of employees can be differentiated according job types andposition; for example, full-time vs. part-time, a researcher vs.research assistant. A full-time researcher may hold more jobresponsibilities and have greater knowledge than a part-timeresearcher. A full-time researcher with a high level of responsi-bility (compare to a part-time researcher) can be involved in thedecision-making process about what needs to outsourced andwhere best to contract the work.

Therefore, we assume that absorptive capacity via organiza-tional composition plays an important role in moderating theeffect of R&D outsourcing on firm performance.

Hypothesis 2-2. (H2-2): R&D outsourcing is more strongly asso-ciated with the firm performance when the absorptive capacity viaorganizational composition is higher.

4. Research methods

4.1. Model

Fig. 3 shows the conceptual framework proposed in this paper.The model indicates that R&D outsourcing has a direct effect onfirm performance. We also hypothesize that the relationshipbetween R&D outsourcing and firm performance is moderatedby a firm's absorptive capacity via internal R&D effort and byorganizational composition due to the internalization processdescribed in the previous section.

Fig. 3. Conceptual framework.

S.Y. Han, S.J. Bae / Int. J. Production Economics 150 (2014) 58–73 63

Page 7: Internalization of R&D outsourcing: An empirical study

4.2. Data

We used a firm-level merged data set composed of financialdata from the Korea Investors Service (KIS) and the R&D Survey inScience and Technology. We used firm identifying numbers formerging these two data sets. KIS is financial information servicefor Korean firms. We utilized data from the Korean Ministry ofEducation, Science and Technology's 2002–2007 R&D Survey inScience and Technology in accordance with the OECD FrascatiManual for the equivalent years.

First, the survey of R&D in Science and Technology wasconducted from 2002 to 2007. The survey data includes 59,911companies with separate R&D departments. We matched 97,407firms’ financial data from KIS with the above R&D survey inScience and Technology (2002–2007) using firms’ registered busi-ness number. Finally, a data pool for this study was generated byintegrating these two data sets from 19,570 firms (Table 1).

4.3. Variables and measures

The variables used in the analyses are defined as follows. Firmperformance as a dependent variable is measured by sales amountin this study, because the purpose of R&D outsourcing is toenhance their sales by developing new technology and products.R&D outsourcing is an independent variable, measured by theintensity of R&D outsourcing. This is calculated by the amount ofR&D outsourcing divided by the amount of sales.

We used four control variables — firm size, financial soundness,level of market competition, openness, and year dummy. Firm sizewas calculated with two variables — the firm's number of employ-ees and total amount of capital stock defined by the InternationalFinancial Reporting Standards (IFRS). The level of market competi-tion was measured by calculating the total market share of thefour largest firms based on the Korean Standard Industrial Classi-fication (KSIC) 2 digit. We measured financial soundness based onthe capital adequacy ratio, calculated by stockholder's equitydivided by each firm's total assets. The openness of firms forcontrolling the level of internalization is the amount of exportdivided by amount of sales.

The R&D intensity of firms is used as the proxy for absorptivecapacity via internal R&D effort (Cohen and Levinthal, 1989;Griliches, 1998; Stock et al., 2001). This is calculated by R&Dexpenditure divided by the amount of sales. Variables of absorp-tive capacity via organizational composition are measured byvarious human resource ratios/in each firm? In each firm's R&Ddivision. We will explain these variables in detail below.

Fig. 4 shows how we classified the staff of each firm's R&Ddivision. R&D employees in an organization include researchers andresearch assistants who support research through testing, measur-ing, experimenting and other supporting activities. Researchers arecategorized using two different criteria — whether a researcher is afull-time employee or not and whether a researcher has a Ph.D. or a

master's degree. The limitation of our data set is that it cannotidentify the working type (full-time vs. part-time) and type ofdegree simultaneously (Table 2).

We assumed that the internalization of external knowledgedepends on the quality of a firm's R&D employees, which weconceptualized as the absorptive capacity via organizational compo-sition. So we use the ratio of R&D employees calculated by thenumber of R&D workers divided by firm's total employees as oneindicator of absorptive capacity of an R&D organization. Anothermeasure is the ratio of R&D researchers to total R&D employees. Wealso investigate whether high-level academic degrees can be asignificant factor in the internalization process. For this investigation,we use (1) the ratio of Ph.D. researchers to total R&D employees, and(2) the ratio of master's degree researchers to total R&D employees.The level of responsibilities of R&D staff can be a crucial factor forR&D outsourcing since much effort is needed to internalize externalknowledge in a way that it can positively affect the organization. Wecan infer that a full-time R&D employee holds more responsibilitythan a part-time R&D employee. Thus, the ratio of full-time researchemployees to R&D employees is used.

Table 3 provides the descriptive statistics, including means, stan-dard deviations, and the minimum and maximum values of the vari-ables. To check the multicollinearity, we check the variance inflationfactors (VIFs). The highest individual VIF score among all the variablesis 3.893, and the mean VIF score is 1.292. As Luo and Deng (2009)state, a VIF score of 10 or less is a widely used guideline for such a test,thus the multicollinearity of our variables is not a problem.

Table 3 has some noteworthy points from the perspective oforganizational composition. First, the ratio of R&D employees tototal staff does not correlate highly with the ratio of Ph.D.researchers to overall R&D staff. The correlation coefficient of theR&D employee ratio to master's degree researcher ratio is higherthan that of R&D employee ratio to the ratio of Ph.D. researcher.A similar pattern is also found for the correlation between theratio of researchers to total R&D employees and the ratio ofresearchers with Ph.D.s. to total R&D employees. This finding canbe explained by cost issues and the characteristics of the manu-facturing industry. First, Ph.D. researchers are core assets for R&Dorganizations; however, their labor cost is higher than researcherswith lower degrees. Therefore, firms may hire relatively few Ph.D.degree researchers for their R&D divisions. Additionally, in themanufacturing industry, a master's degree researcher focusedon engineering may be more in demand than a researcher witha Ph.D. degree who is focused on scientific and engineeringresearch.

Our explanation is further illuminated in Table 5, which showsa comparison of the descriptive statistics of the major variables

Table 1Description of data set (2002–2007).

Financial statement Data R&D activity survey Merged dataset

Year No. of firms Year No. of firms Year No. of firms

2002 14,108 2002 7178 2002 27932003 14,973 2003 6991 2003 29882004 15,588 2004 8300 2004 27952005 16,567 2005 9837 2005 32212006 17,788 2006 12,639 2006 39082007 18,083 2007 14,966 2007 3865

Total 97,407 Total 59,911 Total 19,570

Fig. 4. Classification of human resources in R&D.

S.Y. Han, S.J. Bae / Int. J. Production Economics 150 (2014) 58–7364

Page 8: Internalization of R&D outsourcing: An empirical study

within our sample. We divided the full sample of 19,570 firms intofour groups based on the International Standard Industrial Classi-fication (ISIC) Rev. 3 of technology intensity defined by the OECD(shown in Table 4). According to the OECD, manufacturing indus-tries are classified based on their level of R&D intensity — high,medium-high, medium-low, and low technology industries. Thus,we use the ISIC 2–4 digit industry code to identify four groups offirms based on their level of technological sophistication.

Table 5 shows that the standard deviations for all variables arerather large. This suggests that the data distribution has a highdegree of dispersion. Whenwe compare the means, the ratio of Ph.D. researchers to overall R&D staff is smaller than the ratio ofmaster's degree researchers to overall R&D staff in all fourcategories. For example, in high technology industries, the ratioof Ph.D. researchers to total R&D employees is 4.1% but the ratio ofmaster's degree researchers to overall R&D employees is 25.8%.The same pattern is found in other technology-level categories.When we compare the mean of the ratio of Ph.D. researchers in allfour categories, more highly sophisticated technology industrieshire more Ph.Ds. The same pattern occurs for master's degreeresearchers with the exception of low-technology industries inwhich more master's degree researchers are hired as discussedabove. This pattern also appears in the ratio of full-time employee(FTE) researchers with respect to hiring Ph.Ds.

Second noteworthy point from the organizational compositionperspective is that the CR4 is positively correlated with (1) the ratio ofR&D employees to overall staff, and (2) the ratio of researcher in R&Demployee, but negatively correlated with the ratio of Ph.D. researchersto R&D employee. Consequently, in more competitive industries(i.e. high CR4), firms hire fewer R&D employees and researchers,but compensate by hiring higher degree (Ph.D.) researchers.

4.4. Empirical model

We have examined the hypotheses with an unbalanced paneldata set. Panel data is most useful when the outcome variabledepends on explanatory variables, which are not observable butcorrelated with the observed explanatory variables. If suchomitted variables are constant over time, panel data estimators

allow us to consistently estimate the effect of the observedexplanatory variables (Schmidheiny, 2011).

Consider the multiple linear regression model for firm i¼1 toN, which is observed at each year, t¼1 to T.

yit ¼ αþx′itβþυiþεit ;

i¼ 1;2;…;N

t ¼ 1;2;…; T

εit � i:i:d:ð0; s2e Þ

Here, yit is the dependent variable, xit represents the indepen-dent variables excluding the constant, α is the intercept, β is aparameter, υi is an unobservable individual and firm-specific effectas a time invariant, and εit is an idiosyncratic error term.

We used the panel analysis to correct the estimation bias fromunobservable exogeneity rather than utilize a cross-sectional analysis.For the panel analysis, it matters if υi is correlated with theindependent variables. An unobservable individual and firm-specificeffect usually does not change over time. We can examine the analysiswith a random effect model if we assume that υi is uncorrelated withthe independent variables. But if υi is correlated with independentvariables, the random effect model is not suitable for estimating theefficient estimates. Thus we performed the Hausman specification testto assess whether υi is correlated with the independent variables. Thetest result indicated that we have to adopt the fixed effect model.

We used a 2-year time lag for R&D outsourcing intensity andR&D intensity, since it takes an estimated 1–4 years for R&D toaffect firm performance (Kay, 1988). Researchers have found theseimpacts to be time lagged. For example, Ravenscraft and Scherer(1982) investigated the lag time between R&D and its impact onfirms’ financial performance and found a time gap of 4 years. Manystudies that use Korean manufacturing industry data assume a 1-to 3-year time lag. Therefore, we applied a 2-year time lag for R&Doutsourcing intensity and R&D intensity.

5. Results

Analytical processes in this study are conducted by hierarchicalregression procedures (Cohen, 2003), and the data is analyzed bythe panel data analysis with a fixed-effect model. Tables 4–6 listthe results for our initial analysis before we split the sample for a

Table 2Definition of variables and data sources.

Category Variables Definition Data source

Dependent variableFirm performance Sales ln (amount of sales) KIS

Independent variablesLevel of R&D outsourcing R&D outsourcing intensity Amount of R&D outsourcing/total

R&D expenditureR&D activitysurvey

Absorptive capacityAbsorptive capacity via internal R&D effort R&D intensity R&D expenditure/amount of sales KISAbsorptive capacity via organizationalcomposition

Ratio of R&D employee No. of researcher and assistant/total employee R&D activitysurvey

Ratio of researcher in R&D employee No. of researcher/R&D employeeRatio of Ph.D. researcher No. of Ph.D. researcher/R&D employeeRatio of master's degree researcher No. of researcher with master's degree/R&D

employeeRatio of Full time employee (FTE)researcher

No. of FTE of researcher/R&D employee

Control variablesLevel of internationalization Openness Amount of export/amount of sales KISSize Capital ln (total capital)

No. of employee ln (no. of employee)Financial soundness Capital adequacy ratio Stockholders’ equity/total assetLevel of market competition CR4 The sum of market share of 4 largest firms

in KSIC 2 digitYear dummy Year dummy Year

S.Y. Han, S.J. Bae / Int. J. Production Economics 150 (2014) 58–73 65

Page 9: Internalization of R&D outsourcing: An empirical study

Table 3Descriptive statistics and correlation matrix of variables (N¼19,570).

Variables Mean Std.dev.

Min. Max. Sales R&Doutsourcingintensity

R&Dintensity

Ratio ofR&Demployee

Ratio of researcherin R&D employee

Ratio ofPh.D.researcher

Ratio of Masterdegreeresearcher

Ratio of FTEresearcher

Openness Capital No. ofemployee

capitaladequacyratio

CR4

Sales 10.222 1.598 0.909 17.961 1.000R&D outsourcing

intensity0.066 0.147 0 0.997 0.009 1.000

R&D intensity 0.025 0.056 0 0.997 �0.298nnn 0.035nnn 1.000Ratio of R&D

employee0.166 0.161 0 1.000 �0.421nnn �0.007 0.355nnn 1.000

Ratio ofresearcher inR&Demployee

0.831 0.201 0 1.000 �0.081nnn 0.036nnn 0.046nnn �0.007 1.000

Ratio of Ph.D.researcher

0.036 0.097 0 1.000 �0.015n 0.051nnn 0.095nnn 0.047nnn 0.108nnn 1.000

Ratio of masterdegreeresearcher

0.216 0.223 0 1.000 0.042nnn 0.090nnn 0.136nnn 0.093nnn 0.303nnn 0.154nnn 1.000

Ratio of FTEresearcher

0.846 0.224 0 1.000 �0.040nnn 0.011 0.023nn 0.014n 0.323nnn 0.040nnn 0.099nnn 1.000

Openness 0.005 0.039 0 0.987 0.043nnn 0.283nnn 0.027nnn �0.001 0.006 0.007 0.026nnn 0.001 1.000Capital 7.965 1.506 3.912 17.082 0.706nnn 0.034nnn �0.091nnn �0.220nnn �0.016n 0.063nnn 0.164nnn �0.018n 0.042nnn 1.000No. of employee 4.740 1.188 0.693 11.367 0.854nnn �0.001 �0.185nnn �0.471nnn �0.056nnn 0.007 0.103nnn �0.041nnn 0.042nnn 0.690nnn 1.000capital adequacy

ratio0.464 0.310 �9.523 1.000 0.056nnn 0.016n 0.015n 0.035nnn �0.015n 0.006 0.073nnn 0.002 0.004 0.042nnn 0.045nnn 1.000

CR4 0.356 0.177 0.078 0.975 0.059nnn �0.037nnn 0.068nnn 0.098nnn 0.022nn �0.027nnn �0.005 �0.001 0.020n 0.100nnn 0.083nnn �0.061nnn 1.000

n po0.05; nn po0.01; nnn po0.001.

S.Y.Han,S.J.Bae

/Int.J.Production

Economics

150(2014)

58–73

66

Page 10: Internalization of R&D outsourcing: An empirical study

more detailed analysis. A 2-year time lag for R&D outsourcingintensity and R&D intensity was applied because the outcome ofR&D manifests some time lag.

Model 1 investigates Hypothesis 1, whether R&D outsourcingcan improve firm performance. But the result is not significant inall samples. As expected, larger firms can improve performance.Financial adequacy is also significantly related to firm perfor-mance. Models 2 and 3 include control variables and variables ofabsorptive capacity via internal R&D effort and organizationalcomposition, as moderating variables. Examining adjusted R2

values across all models suggests that the full model providesthe best fit to the data. The result of Model 3 shows that R&Doutsourcing is not significant, but the estimated coefficient of theresearcher to R&D employees ratio is positive (t¼2.84, po0.01) atthe 5% significance level. So, a greater ratio of researchers to R&Demployees has a direct, positive effect on firm performance.

Among the moderating variables, only Hypothesis H2-1 isconfirmed. In other words, the absorptive capacity with internalR&D effort has a positive effect (t¼4.44, po0.001) not directly onfirm performance but indirectly as a moderating effect when firmsoutsource R&D. On the other hand, the rest of variables are notsignificant in the full sample. These results confirm that a higherlevel of internal R&D input improves a firm's ability to utilizeexternal knowledge (Cohen and Levinthal, 1990; Gambardella,1992; Helfat, 1997; Mowery et al., 1996; Zahra and Hayton, 2008).

To examine our hypotheses in more details, we divide the sampleaccording to the OECD level of technological sophistication. Dependingon the technological level, the level of sophistication in internalizationalso differs. For example, a firm that produces highly complexproducts will require a highly sophisticated organizational composi-tion compared to a firm that produces products through mereassembly line and simple work. This rationale suggests the possibilitythat the internalization mechanism role also differs based on the levelof technological sophistication. Thus we investigate our hypotheses ata more detailed level using the sample division method.

The results of the split-sample analysis are listed and Model 3 isused to discuss Tables 7–10. Within the four groups, only the high-technology group had a significantly positive coefficient (t¼0.27,po0.05) of R&D outsourcing intensity.

The speed of change in the high-technology industry is most rapidamong all industries. What we can infer from this result is that hightechnology firms must cope with environmental and technologicalchanges in their market. Thus, firms invest a large amount of internalR&D to increase their innovation capability, and need to seek andtransfer external technology through R&D outsourcing.

In the examination of the moderating effect of absorptive capacitybetween R&D outsourcing and firm performance, only high- and low-technology industries had some significantly positive coefficients. First,in high technology industries (Table 7), the absorptive capacity viainternal R&D effort (H2-1) has a significantly positive (t¼5.51, po0.001) effect between R&D outsourcing and firm performance. Thisresult is consistent with previous research (Tsai and Wang, 2008) thatfound that external technology acquisition does not significantlyenhance firm performance per se; however, the positive impact ofexternal technology acquisition on firm performance increases withthe level of internal R&D investment (as an absorptive capacity).Therefore, we propose that by investing more resources in R&D input,firms can achieve higher levels of performance in the R&D outsourcingsetting.

For Hypothesis H2-2, the ratio of Ph.D. researchers to R&D staff hasa positive significant effect (t¼1.97, po0.05) on R&D outsourcingregarding a firm's performance. In addition, the ratio of FTE research-ers (H2-2e) has a notable negative effect (t¼�1.99, po0.05). Thesetwo empirical findings support only part of H2-2. The results showthat high-level technology firms need to increase the quality (e.g.,education and responsibility levels) of R&D personnel when they useR&D outsourcing to gain external knowledge. In other words, high-level technology firms should have higher internal R&D input andmore highly educated researchers (i.e., Ph.D.s. and master's degrees)when they choose to outsource R&D and use absorptive capacity to

Table 4ISIC REV. 3 technology intensity definition of OECD.

High-technology industries Medium-high-technology industries

Aircraft and spacecraft Electrical machinery and apparatus, n.e.c.Pharmaceuticals Motor vehicles, trailers and semi-trailersOffice, accounting and computing machinery Chemicals excluding pharmaceuticalsRadio, TV and communications equipment Railroad equipment and transport equipment, n.e.c.Medical, precision and optical instruments Machinery and equipment, n.e.c.

Medium-low-technology industries Low-technology industries

Building and repairing of ships and boats Manufacturing, n.e.c.; RecyclingRubber and plastics products Wood, pulp, paper, paper products, printing and publishingCoke, refined petroleum products and nuclear fuel Food products, beverages and tobaccoOther non-metallic mineral products Textiles, textile products, leather and footwear

Table 5Comparison of descriptive statistics of major variables within the sample.

Variable Full sample(N¼19,570)

High technologyindustries (N¼5749)

Medium-high technologyindustries (N¼7812)

Medium-low technologyindustries (N¼2851)

Low technologyindustries (N¼3158)

Mean S.D. Mean S.D. Mean S.D. Mean S.D. Mean S.D.

Sales (USD. Mil.) 198.947 1,473.684 161.053 2,105.263 155.789 1024.211 342.105 1715.789 92.000 220.000R&D intensity 0.025 0.056 0.040 0.075 0.033 0.040 0.029 0.020 0.025 0.055R&D outsourcing intensity 0.066 0.147 0.071 0.143 0.053 0.121 0.062 0.130 0.046 0.118Ratio of R&D employee 0.166 0.161 0.226 0.195 0.146 0.113 0.098 0.086 0.186 0.194Ratio of researcher in R&D employee 0.831 0.201 0.849 0.174 0.821 0.197 0.809 0.209 0.840 0.220Ratio of PhD. researcher 0.036 0.097 0.041 0.096 0.038 0.086 0.036 0.106 0.035 0.100Ratio of master's degree researcher 0.216 0.223 0.258 0.233 0.186 0.205 0.172 0.205 0.219 0.222Ratio of FTE researcher 0.846 0.224 0.854 0.214 0.845 0.221 0.828 0.226 0.852 0.230

S.Y. Han, S.J. Bae / Int. J. Production Economics 150 (2014) 58–73 67

Page 11: Internalization of R&D outsourcing: An empirical study

internalize the external knowledge from this strategy. For example,Hoffman et al. (1998) noted that the most important determinants ofinnovation and economic success for high-tech firms are the scientist,engineer and owner-manager. Namely, highly-educated R&D staff is akey factor for innovation in the high technology industry. From thisfinding, we can state that the ratio of Ph.D. to R&D staff indicates howwell firms are organized to utilize external knowledge, given that theratio of highly educated employees is a determinant for synthesizing,translating and internalizing external knowledge. In particular, high-tech firms that develop state-of-the-art technology and internalize itssubsequent knowledge will have a stronger effect from hiring highlyeducated researchers.

However, we found that the ratio of R&D researchers to overallR&D staff and the ratio of master's degree researchers to overallR&D staff are not significant for all split groups. Based on the resultsshown in Table 10, the coefficient of the ratio of R&D employee tooverall staff in low-technology industries is significantly positive(t¼2.06) at the 5% level. Thus, firms in low-technology industriesmust invest resources to increase the ratio of R&D employees tooverall staff. They can reap benefits from hiring more R&D employ-ees, without consideration of their level of knowledge, education orresponsibility level, since internalization may not require sophisti-cated (i.e. education level) and responsible (i.e. FTE) R&D personnelin this setting.

Table 6Results of panel analysis (full sample, n¼19,570).

Variables Model 1 Model 2 Model 3

Dependent variable: SalesOpenness 0.368(0.143)† 0.373(0.143)nn 0.375(0.143)nn

Capital 0.106(0.016)nnn 0.104(0.015)nnn 0.101(0.016)nnn

No. of employees 0.550(0.017)nnn 0.571(0.018)nnn 0.570(0.018)nnn

Capital adequacy ratio 0.087(0.021)nnn 0.084(0.02)nn 0.084(0.021)nnn

CR4 0.059(0.200) 0.040(0.200) 0.033(0.200)R&D outsourcing intensity (t�2) 0.034(0.036) 0.036(0.036) 0.017(0.037)R&D intensity (t�2) 0.037(0.119) 0.025(0.119)Ratio of R&D employee 0.016(0.038)n 0.014(0.038)Ratio of researcher in R&D employee 0.196(0.069) 0.195(0.069)nn

Ratio of Ph.D. researcher �0.033(0.50) �0.031(0.050)Ratio of master's degree researcher 0.025(0.037) 0.022(0.037)Ratio of FTE researcher 0.002(0.021) 0.002(0.021)R&D outsourcing intensity (t�2)�R&D intensity (t�2) 2.673(0.602)nnn

R&D outsourcing intensity (t�2)� ratio of R&D employee �0.070(0.198)R&D outsourcing intensity (t�2)� ratio of researcher in R&D employee 0.265(0.217)R&D outsourcing intensity (t�2)� ratio of Ph.D. researcher 0.126(0.354)R&D outsourcing intensity (t�2)� ratio of master's degree researcher 0.186(0.165)R&D outsourcing intensity (t�2)� ratio of FTE researcher 0.072(0.135)Year dummy Included Included IncludedAdjusted R2 0.250 0.252 0.256F-Value (P) 16.50nnn 15.35nnn 15.40nnn

Estimated standard errors are in parentheses.† po0.10; n po0.05; nn po0.01; nnn po0.001.

Table 7Result of panel analysis (high-technology industries, N¼5749).

Variables Model 1 Model 2 Model 3

Dependent variable : SalesOpenness 0.279(0.237) 0.306(0.238) 0.329(0.235)Capital 0.072(0.035)n 0.069(0.035)n 0.066(0.034)No. of employees 0.695(0.034)nnn 0.717(0.037)nnn 0.708(0.036)nnn

Capital adequacy ratio 0.326(0.048)nnn 0.321(0.048)nnn 0.323(0.048)nnn

CR4 �0.241(0.684) �0.338(0.688) �0.535(0.681)R&D outsourcing intensity (t�2) 0.170(0.077)n 0.175(0.077)n 0.023(0.084)n

R&D intensity (t�2) �0.026(0.188) 0.083(0.187)Ratio of R&D employee �0.049(0.088) �0.094(0.088)Ratio of researcher in R&D employee 0.188(0.128) 0.129(0.126)Ratio of Ph.D. researcher �0.105(0.114) �0.093(0.115)Ratio of master's degree researcher 0.015(0.084) �0.005(0.083)Ratio of FTE researcher 0.043(0.050) 0.043(0.050)R&D outsourcing intensity (t�2)�R&D intensity (t�2) 5.264(0.956)nnn

R&D outsourcing intensity (t�2)� ratio of R&D employee �0.728(0.532)R&D outsourcing intensity (t�2)� ratio of researcher in R&D employee 0.196(0.379)R&D outsourcing intensity (t�2)� ratio of Ph.D. researcher 1.605(0.824) n

R&D outsourcing intensity (t�2)� ratio of master's degree researcher 0.184(0.382)R&D outsourcing intensity (t�2)� ratio of FTE researcher �0.733(0.368)n

Year dummy Included Included IncludedAdjusted R2 0.336 0.338 0.360F-value (P) 10.23nnn 9.15nnn 9.39nnn

Estimated standard errors are in parentheses.n po0.05; nn po0.01; nnn po0.001.

S.Y. Han, S.J. Bae / Int. J. Production Economics 150 (2014) 58–7368

Page 12: Internalization of R&D outsourcing: An empirical study

6. Conclusion

6.1. Summary and implication

R&D outsourcing is one of the most commonly used strategiesthat firms employ to gain and utilize external knowledge. As wediscussed earlier, most of the previous studies have focused on thequestion of whether R&D outsourcing impacts the firm's overallperformance Research also highlights the role of absorptive capacityvia internal R&D effort as a moderating variable when a firm enjoysthe benefit of external knowledge via various open innovationstrategies. However, how a firms sets up its R&D organization

(e.g., ratio of R&D staff to total employees; R&D staff skills andeducation levels) to use external knowledge efficiently is worthy ofinvestigation. Based on the literature, we had assumed that a firm'sabsorptive capacity via internal R&D effort and organizationalcomposition are key elements of the entire R&D outsourcing processsince the primary actors — those who plan R&D outsourcing andinternalize external knowledge — are internal R&D employees.Through the empirical testing, this assumption has been testedwith additional insights from considering different levels of tech-nological sophistication. Our study adds a new perspective to thecomposition of the R&D organization in firms while maintaining thecurrent knowledge on absorptive capacity.

Table 8Result of panel analysis (medium-high-technology industries, N¼7812).

Variables Model 1 Model 2 Model 3

Dependent variable : SalesOpenness 0.532(0.280)n 0.530(0.280) 0.507(0.281)†

Capital 0.076(0.029)nn 0.075(0.029)nn 0.075(0.029)nn

No. of employee 0.379(0.030)nnn 0.383(0.035)nnn 0.383(0.035)nnn

Capital adequacy ratio 0.144(0.067)n 0.144(0.067)n 0.144(0.067)n

CR4 �2.262(0.585)nnn �2.270(0.587)nnn �2.272(0.588)nnn

R&D outsourcing intensity (t�2) 0.121(0.069)† 0.127(0.069) 0.117(0.070)†

R&D intensity (t�2) 0.268(0.257) 0.318(0.279)Ratio of R&D employee 0.075(0.065) 0.074(0.065)Ratio of researcher in R&D employee 0.011(0.158) 0.015(0.160)Ratio of Ph.D. researcher �0.200(0.101)n �0.178(0.104)†

Ratio of master's degree researcher �0.129(0.071) �0.125(0.072)†

Ratio of FTE researcher �0.038(0.036) �0.030(0.040)R&D outsourcing intensity (t�2)�R&D intensity (t�2) �0.784(1.418)R&D outsourcing intensity (t�2)� ratio of R&D employee �0.262(0.368)R&D outsourcing intensity (t�2)� ratio of researcher in R&D employee 0.155(0.664)R&D outsourcing intensity (t�2)� ratio of PhD. researcher 0.888(0.988)R&D outsourcing intensity (t�2)� ratio of master's degree researcher 0.138(0.354)R&D outsourcing intensity (t�2)� ratio of FTE researcher 0.397(0.297)Year dummy Included Included IncludedAdjusted R2 0.189 0.194 0.196F-Value (P) 12.18nnn 11.76nnn 11.71nnn

Estimated standard errors are in parentheses.† po0.10; n po0.05; nn po0.01; nnn po0.001

Table 9Result of panel analysis (medium-low-technology industries, N¼2851).

Variables Model 1 Model 2 Model 3

Dependent variable : SalesOpenness 0.234(0.310) 0.243(0.307) 0.338(0.320)Capital 0.147(0.039)nnn 0.134(0.039)nn 0.132(0.039)nn

No. of employee 0.590(0.048)nnn 0.626(0.056)nnn 0.620(0.056)nnn

Capital adequacy ratio 0.237(0.084)n 0.227(0.084)nn 0.228(0.084)nn

CR4 �1.080(0.588) �1.299(0.593)n �1.260(0.598)n

R&D outsourcing intensity (t�2) �0.109(0.080) �0.096(0.080) �0.120(0.100)R&D intensity (t�2) �0.204(0.662) �0.095(0.691)Ratio of R&D employee �0.155(0.079)n �0.164(0.080)nn

Ratio of researcher in R&D employee 0.193(0.266) 0.229(0.271)Ratio of Ph.D. researcher 0.196(0.088)n 0.200(0.089)n

Ratio of master's degree researcher 0.282(0.072)nnn 0.286(0.072)nnn

Ratio of FTE researcher 0.110(0.050)n 0.108(0.050)n

R&D outsourcing intensity (t�2)�R&D intensity (t�2) �0.957(3.875)R&D outsourcing intensity (t�2)� ratio of R&D employee 0.473(0.453)R&D outsourcing intensity (t�2)� ratio of researcher in R&D employee �0.516(0.847)R&D outsourcing intensity (t�2)� ratio of Ph.D. researcher 0.723(0.845)R&D outsourcing intensity (t�2)� ratio of master's degree researcher 0.050(0.380)R&D outsourcing intensity (t�2)� ratio of FTE researcher �0.084(0.360)Year dummy Included Included IncludedAdjusted R2 0.379 0.399 0.403F-Value (P) 24.34 23.26 23.00

Estimated standard errors are in parentheses.n po0.05; nn po0.01; nnn po0.001.

S.Y. Han, S.J. Bae / Int. J. Production Economics 150 (2014) 58–73 69

Page 13: Internalization of R&D outsourcing: An empirical study

This study examined these issues. The longitudinal sampleanalysis allowed us to control several important variables — firmsize, firms’ financial soundness, openness and level of marketcompetition, which led to convincing evidence on the significanceof organizational composition in maximizing the effect of R&Doutsourcing. A merged data set consisting of financial data fromKIS and the survey of R&D in Science and Technology 2002–2007led us to conduct this empirical study on 19,570 firms, a largeenough data set to establish the external validity of our findings.

The major comparison of two radically different levels of tech-nology (i.e., high- and low-technology) is summarized in Table 11.The effect of R&D outsourcing on firm performance is confirmedonly for high-level technology industries. Internal R&D efforts suchas absorptive capacity were found to moderate R&D outsourcing ona firm's performance in a full sample, and again in the spilt sampleof high-technology industries. In the case of high technologyindustries, the ratio of Ph.D. researchers to overall firm staff had apositive moderating effect on the relationship between R&D out-sourcing and a firm's performance, while the ratio of FTE researchershad the opposite effect.

Overall, our findings indicate that the quality (i.e. educationallevel) of researchers is more important than mere quantity ofresearchers in terms of impacting the firm's performance. Speci-fically, high technology firms should employ more highly educatedresearchers if their aim is to get better results from internalizingexternal knowledge gained via outsourcing. The resource-basedview focuses on the technology capacity for innovation as anintangible resource and identified knowledge of expertise, experi-ence, skill and culture of organization as essential technologycapacities (Hall, 1992). Many studies have pointed out thatexperienced employees with a advanced education and skills canbe a determinant of innovation in firms (Koschatzky et al., 2001;Romijn and Albaladejo, 2002). Our results also coincide with thesefindings. The entire R&D outsourcing process needs internalexpertise. As Cohen and Levinthal (1990) pointed out, the experi-enced employee with a high level of education plays a major rolein developing new knowledge by understanding, absorbing, andutilizing external knowledge; our study supports these results.Therefore, a firm should have a strong division of R&D employeeswith advanced experience and existing knowledge if it aims to

Table 10Result of panel analysis (low-technology industries, N¼3158).

Variables Model 1 Model 2 Model 3

Dependent variable : SalesOpenness �0.456(0.633) 0.429(0.635) �0.234(0.647)Capital 0.038(0.037) 0.035(0.037) 0.044(0.037)No. of employees 0.502(0.045)nnn 0.538(0.047)nnn 0.539(0.047)nnn

Capital adequacy ratio �0.078(0.067) �0.080(0.066) �0.082(0.066)CR4 �0.790(0.802) �0.839(0.800) �0.810(0.800)R&D outsourcing intensity (t�2) 0.002(0.112) 0.001(0.112) �0.007 (0.128)R&D intensity (t�2) 0.524(0.261)n 0.461 (0.270)†

Ratio of R&D employee �0.027(0.103) 0.015(0.106)Ratio of researcher in R&D employee 0.447(0.146)nn 0.451(0.146)nn

Ratio of Ph.D. researcher 0.029(0.128) �0.003(0.216)Ratio of master's degree Researcher �0.065(0.106) �0.037(0.108)Ratio of FTE researcher �0.039(0.051) �0.043(0.052)R&D outsourcing intensity (t�2)�R&D intensity (t�2) �1.983(1.758)R&D outsourcing intensity (t�2)� ratio of R&D employee 1.017(0.493)n

R&D outsourcing intensity (t�2)� ratio of researcher in R&D employee �0.104(0.632)R&D outsourcing intensity (t�2)� ratio of PhD. researcher �0.611(2.870)R&D outsourcing intensity (t�2)� ratio of master's degree researcher 0.670(0.570)R&D outsourcing intensity (t�2)� ratio of FTE researcher �0.241(0.465)Year dummy Included Included IncludedAdjusted R2 0.223 0.243 0.250F-Value (P) 19.53 17.29 17.25

Estimated standard errors are in parentheses.† po0.10; n po0.05; nn po0.01; nnn po0.001.

Table 11Summary of results.

Variables Full Sample High-technologyindustries

Low technologyindustries

Independent variable R&D outsourcing (H1) (þ)n

Moderating variablesAbsorptive capacity with internal R&D effort R&D intensity (H2-1) (þ)nnn (þ)nnn

Absorptive capacity with organizationalcomposition

Ratio of R&D employee (H2-2a) (þ)n

Ratio of researcherin R&D employee (H2-2b)Ratio of Ph.D. researcher (H2-2c) (þ)n

Ratio of Master degree researcher (H2-2d)Ratio of FTE researcher (H2-2e) (�)n

Estimated standard errors are in parentheses.The results of medium-high and medium-low technology industries are omitted because they do not have any significant coefficients.n po0.05; nn po0.01; nnn po0.001.

S.Y. Han, S.J. Bae / Int. J. Production Economics 150 (2014) 58–7370

Page 14: Internalization of R&D outsourcing: An empirical study

internalize knowledge from outsourced R&D. However, our find-ings indicate that for low technology industries, the quantity ofemployees (i.e., the ratio of R&D staff to overall employees), butnot the quality of R&D employees moderates the effect of R&Doutsourcing on a firm's performance. Together with the findingson high-technology industries that we discussed above, we con-clude that there should be differentiated approach to R&D staffingwith regards to R&D outsourcing, and that single rule does notapply to all settings.

To give a clearer view to the reader, we summarized all theempirical results in Table 11. In this table, we can explain why theempirical results are differentiated in Table 4. Here, we canobserve the difference of effects between absorptive capacity viainternal R&D efforts and via organizational composition. Thisindicates that not only does absorptive capacity via internal R&Defforts but also via organizational composition does make adifference, and that they have different effects on firms of variouslevels of technological sophistication.

We can infer some managerial implications for practice. All ofthe ratios of R&D employees, researchers in R&D employees, Ph.D.and master's degree researchers between high and low technologyindustries are very different except the FTE researcher ratio. We caninfer that these distinctions in organizational composition producea moderating effect between R&D outsourcing and a firm's perfor-mance. Therefore, firms should optimize the R&D division's com-position especially when their level of technological sophisticationis growing. By focusing on the organizational composition, they canmaximize the R&D outsourcing effect by increasing the absorptivecapacity for internalizing external knowledge.

6.2. Limitations and future research

This study has several moderate limitations. First, we used thelevel of human capital as the proxy for our focal firms’ absorptivecapacity via organizational composition. While this is not anunreasonable proxy, we need to conduct a further study —

preferably a more micro-level field study — to confirm thisassumption. Case studies with some representative firms couldhelp us better understand the nature of the internalization processwe describe to see how education level works as the moderator.This could also provide us with opportunities to theorize on howa firm's micro-level mechanisms help it internalize externalknowledge through R&D outsourcing. For example, Kessler et al.(2000) found that R&D outsourcing is detrimental to competitiveadvantage during the idea generation stage of new product andtechnology development and significantly lengthened the projectcompletion time during this stage.

Secondly, even though we found evidence on our hypotheses,the results are varied in different industries. This calls for addi-tional studies of the same nature but in other countries so as tocompare results across nations.

This study also highlights the need for more in-depth studies onorganizational composition factors as potential moderating variablesof R&D outsourcing effects on a firm's performance. For example,diversity of R&D division can be an important factor that could berelated to absorptive capacity via organizational composition. Moreresearch is needed to investigate the impact of diversity as amoderating effect of R&D outsourcing on a firm's performance.

Additionally, the type of R&D a firm does — basic, applied, anddevelopmental research — can impact the relationship betweenR&D outsourcing and firm performance in different ways(Lichtenberg and Siegel, 1991). As hinted from the findings ondifferent level of technology sophistication in this study, thevarious types of R&D research that firms conduct can havedifferent effects on the causal linkages between external knowl-edge and internalization that we identified.

Finally, several organizational composition factors require inves-tigation, such as the structure of a firm's communication, itsorganizational culture which affects how it obtains, processes anduses external knowledge, and the scope of a firm's external net-work, which may differ based on the technology level of companies.

References

Acha, V., Cusmano, L., 2005. Governance and co-ordination of distributed innova-tion processes: patterns of R&D co-operation in the upstream petroleumindustry. Econ. Innov. New Technol. 14, 1–21.

Aghion, P., Howitt, P., 1992. A model of growth through creative destruction.Econometrica 60, 323–351.

Allen, T.J., 1977. Managing the Flow of Technology: Technology Transfer and theDissemination of Technological Information Within the R&D Organization. MITPress, Cambridge, MA.

Argote, L., 1999. Organizational Learning: Creating, Retaining, and TransferringKnowledge. Springer, Netherlands.

Austin, D.H., 1993. An event-study approach to measuring innovative output: thecase of biotechnology. Am. Econ. Rev. 83, 253–258.

Bönte, W., 2003. R&D and productivity: Internal vs. external R&D-evidence fromwest german manufacturing industries. Econ. Innov. New Technol. 12, 343–360.

Bagchi, P., Lejeune, M.A., Alam, A., 2013. How supply competency affects FDIdecisions: some insights. Int. J. Prod. Econ..

Becker, W., Peters, J., 2000. Technological Opportunities, Absorptive Capacities, andInnovation. Volkswirtschaftliche Diskussionsreihe, Institut für Volkswirtschaft-slehre der Universität Augsburg.

Beneito, P., 2003. Choosing among alternative technological strategies: an empiricalanalysis of formal sources of innovation. Res. Policy 32, 693–713.

Bergman, K., 2011. Internal and External R&D and Productivity—Evidence fromSwedish Firm-level Data. Lund University Working Papers 2011, p. 27.

Besanko, D., Dranove, D., Shanley, M., Schaefer, S., 2009. Economics of Strategy, 5thEd. Wiley, New York.

Bierly III, P.E., Chakrabarti, A.K., 1996. Technological learning, strategic flexibility,and new product development in the pharmaceutical industry. IEEE Trans. Eng.Manage. 43, 368–380.

Bierly III, P.E., Damanpour, F., Santoro, M.D., 2009. The application of externalknowledge: organizational conditions for exploration and exploitation.J. Manage. Stud. 46, 481–509.

Brook, J.W., Plugge, A., 2010. Strategic Sourcing of R&D: The Determinants ofSuccess. Global Sourcing of Information Technology and Business Processes,pp. 26–42.

Brusoni, S., Prencipe, A., Pavitt, K., 2001. Knowledge specialization, organizationalcoupling, and the boundaries of the firm: why do firms know more than theymake? Adm. Sci. Q., 597–621.

Carter, A.P., 1989. Knowhow trading as economic exchange. Res. Policy 18, 155–163.Cassiman, B., Veugelers, R., 2002. R&D cooperation and spillovers: some empirical

evidence from Belgium. Am. Econ. Rev. 92, 1169–1184.Caudy, D.W., 2001. Using R&D outsourcing as a competitive tool. Medical Device

and Diagnostic Industry Magazine.Chen, C.J., 2004. The effects of knowledge attribute, alliance characteristics, and

absorptive capacity on knowledge transfer performance. R&D Manage. 34,311–321.

Chesbrough, H.W., 2003. Open Innovation: The New Imperative for Creating andProfiting From Technology. Harvard Business Press, Boston, MA.

Clausen, T.H., 2013. External knowledge sourcing from innovation cooperation andthe role of absorptive capacity: empirical evidence from Norway and Sweden.Technol. Anal. Strateg. Manage. 25, 57–70.

Coase, R.H., 1937. The nature of the firm. Economica 4, 386–405.Cohen, J., Cohen, C., West, S.G., Aiken, L.S., 2003. Applied Multiple Regression/

Correlation Analysis for the Behavioral Sciences. Lawrence Erlbaum,Mahwah, NJ.

Cohen, W.M., Levinthal, D.A., 1989. Innovation and learning: the two faces of R&D.Econ. J. 99, 569–596.

Cohen, W.M., Levinthal, D.A., 1990. Absorptive capacity: a new perspective onlearning and innovation. Adm. Sci. Q., 128–152.

Collis, D.J., 1994. Research note: how valuable are organizational capabilities?Strateg. Manage. J. 15, 143–152.

Coombs, R., Harvey, M., Tether, B.S., 2003. Analysing distributed processes ofprovision and innovation. Ind. Corp. Change 12, 1125–1155.

Escribano, A., Fosfuri, A., Tribó, J.A., 2009. Managing external knowledge flows: Themoderating role of absorptive capacity. Res. Policy 38, 96–105.

Foss, N.J., Pedersen, T., 2002. Transferring knowledge in MNCs:: the role of sourcesof subsidiary knowledge and organizational context. J. Int. Manage. 8, 49–67.

Gambardella, A., 1992. Competitive advantages from in-house scientific research:the US pharmaceutical industry in the 1980s. Res. Policy 21, 391–407.

George, G., Zahra, S.A., Wheatley, K.K., Khan, R., 2001. The effects of allianceportfolio characteristics and absorptive capacity on performance: a study ofbiotechnology firms. J. High Technol. Manage. Res. 12, 205–226.

Ghoshal, S., Westney, D.E., 1991. Organizing competitor analysis systems. Strateg.Manage. J. 12, 17–31.

Gilley, K.M., Rasheed, A., 2000. Making more by doing less: an analysis ofoutsourcing and its effects on firm performance. J. Manage. 26, 763–790.

S.Y. Han, S.J. Bae / Int. J. Production Economics 150 (2014) 58–73 71

Page 15: Internalization of R&D outsourcing: An empirical study

Glass, A.J., Saggi, K., 1998. International technology transfer and the technology gap.J. Dev. Econ. 55, 369–398.

Grant, R.M., 1996. Toward a knowledge-based theory of the firm. Strateg. Manage. J.17, 109–122.

Griliches, Z., 1986. Productivity, R&D, and Basic Research at the Firm Level in the1970s. National Bureau of Economic Research, Cambridge, MA.

Griliches, Z., 1998. R&D and Productivity: The Econometric Evidence. University ofChicago Press, Chicago, IL.

Grimpe, C., Sofka, W., 2009. Search patterns and absorptive capacity: low-and high-technology sectors in European countries. Res. Policy 38, 495–506.

Guellec, D., Van Pottelsberghe de la Potterie, B., 2004. From R&D to productivitygrowth: do the institutional settings and the source of funds of R&D matter?Oxford Bull. Econ. Stat. 66, 353–378.

Hall, R., 1992. The strategic analysis of intangible resources. Strateg. Manage. J. 13,135–144.

Hall, R.H., 1982. Organizations: Structure and Process. Prentice-Hall, EnglewoodCliffs, NJ.

Hamel, G., Prahalad, C.K., 1994. Competing for the future. Harvard Business SchoolPress, Boston, MA.

Helfat, C.E., 1997. Know-how and asset complementarity and dynamic capabilityaccumulation: the case of R&D. Strateg. Manage. J. 18, 339–360.

Hoffman, K., Parejo, M., Bessant, J., Perren, L., 1998. Small firms, R&D, technologyand innovation in the UK: a literature review. Technovation 18, 39–55.

Howells, J., 1999. Research and technology outsourcing. Technol. Anal. Strateg.Manage. 11, 17–29.

Howells, J., Gagliardi, D., Malik, K., 2008. The growth and managementof R&D outsourcing: evidence from UK pharmaceuticals. R&D Manage. 38,205–219.

Howells, J., James, A., Malik, K., 2003. The sourcing of technological knowl-edge: distributed innovation processes and dynamic change. R&D Manage. 33,395–409.

Howells, J., James, A.D., Malik, K., 2004. Sourcing external technological knowl-edge: a decision support framework for firms. Int. J. Technol. Manage. 27,143–154.

Huang, Y.A., Chung, H.J., Lin, C., 2009. R&D sourcing strategies: determinants andconsequences. Technovation 29, 155–169.

Huber, G.P., 1991. Organizational learning: the contributing processes and theliteratures. Organ. Sci., 88–115.

Jones, G.K., Aldor, L.J.R., Teegen, H.J., 2001. Determinants and performance impactsof external technology acquisition. J. Bus. Ventur. 16, 255–283.

Kay, N., 1988. The R&D function: corporate strategy and structure. Tech. ChangeEcon. Theory, 283–294.

Keller, W., 1996. Absorptive capacity: on the creation and acquisition of technologyin development. J. Dev. Econ. 49, 199–227.

Kessler, E.H., Bierly, P.E., Gopalakrishnan, S., 2000. Internal vs. external learning innew product development: effects on speed, costs and competitive advantage.R&D Manage. 30, 213–224.

Kim, L., 1999. Building technological capability for industrialization: analyticalframeworks and Korea's experience. Ind. Corp. Change 8, 111–136.

Koschatzky, K., Bross, U., Stanovnik, P., 2001. Development and innovation potentialin the Slovene manufacturing industry: analysis of an industrial innovationsurvey. Technovation 21, 311–324.

Kostopoulos, K., Papalexandris, A., Papachroni, M., Ioannou, G., 2011. Absorp-tive capacity, innovation, and financial performance. J. Bus. Res. 64,1335–1343.

Kyläheiko, K., Jantunen, A., Puumalainen, K., Luukka, P., 2011. Value of knowledge—technology strategies in different knowledge regimes. Int. J. Prod. Econ. 131,273–287.

Lane, P.J., Koka, B.R., Pathak, S., 2006. The reification of absorptive capacity: acritical review and rejuvenation of the construct. Acad. Manage. Rev. 31,833–863.

Lane, P.J., Lubatkin, M., 1998. Relative absorptive capacity and interorganizationallearning. Strateg. Manage. J. 19, 461–477.

Lane, P.J., Salk, J.E., Lyles, M.A., 2001. Absorptive capacity, learning, and performancein international joint ventures. Strateg. Manage. J. 22, 1139–1161.

Leiblein, M.J., Miller, D.J., 2003. An empirical examination of transaction‐and firm‐

level influences on the vertical boundaries of the firm. Strateg. Manage. J. 24,839–859.

Lenox, M., King, A., 2004. Prospects for developing absorptive capacity throughinternal information provision. Strateg. Manage. J. 25, 331–345.

Levitt, B., March, J.G., 1988. Organizational learning. Annu. Rev. Sociol., 319–340.Lichtenberg, F.R., Siegel, D., 1991. The impact of R&D investment on productivity—

new evidence using linked R&D—LRD data. Econ. Inq. 29, 203–229.Lin, C., Wu, Y.-J., Chang, C., Wang, W., Lee, C.-Y., 2012. The alliance innovation

performance of R&D alliances—the absorptive capacity perspective. Technova-tion 32, 282–292.

Liu, X., White, R.S., 1997. The relative contributions of foreign technology anddomestic inputs to innovation in Chinese manufacturing industries. Technova-tion 17, 119–125.

Love, J.H., Roper, S., 2001. Location and network effects on innovation success:evidence for UK, German and Irish manufacturing plants. Res. Policy 30,643–661.

Lund Vinding, A., 2006. Absorptive capacity and innovative performance: a humancapital approach. Econ. Innov. New Technol. 15, 507–517.

Luo, X., Deng, L., 2009. Do birds of a feather flock higher? The effects of partnersimilarity on innovation in strategic alliances in knowledge‐intensive indus-tries. J. Manage. Stud. 46, 1005–1030.

Mangematin, V., Nesta, L., 1999. What kind of knowledge can a firm absorb? Int. J.Technol. Manage. 18, 149–172.

Mellat-Parast, M., Digman, L.A., 2008. Learning: the interface of quality manage-ment and strategic alliances. Int. J. Prod. Econ. 114, 820–829.

Miyamoto, D., 2007. Determinants of R&D outsourcing at Japanese firms: transac-tion cost and strategic management perspectives. In: Proceedigns of WorldAcademy of Science, Engineering and Technology, vol. 24, pp. 46–51.

Mowery, D.C., 1984. Firm structure, government policy, and the organization ofindustrial research: Great Britain and the United States, 1900–1950, Bus. Hist.Rev., pp. 504–531.

Mowery, D.C., Oxley, J.E., 1995. Inward technology transfer and competitiveness:the role of national innovation systems. Cambridge J. Econ. 19, 67–93.

Mowery, D.C., Oxley, J.E., Silverman, B.S., 1996. Strategic alliances and interfirmknowledge transfer. Strateg. Manage. J. 17, 77–91.

Mudambi, R., 2002. Knowledge management in multinational firms. J. Int. Manage.8, 1–9.

Narula, R., 2001. Choosing between internal and non-internal R&D activities: sometechnological and economic factors. Technol. Anal. Strateg. Manage. 13,365–387.

Nieto, M., Quevedo, P., 2005. Absorptive capacity, technological opportunity,knowledge spillovers, and innovative effort. Technovation 25, 1141–1157.

Nohria, N., Garcia-Pont, C., 1991. Global strategic linkages and industry structure.Strateg. Manage. J. 12, 105–124.

Pakes, A., Schankerman, M., 1984. The Rate of Obsolescence of Patents, ResearchGestation Lags, and the Private Rate of Return to Research Resources. In:Griliches, Z. (Ed.), R & D, Patents, and Productivity. University of Chicago Press,Chicago IL, pp. 73–88.

Prahalad, C.K., Hamel, G., 1990. The Core Competence of the Corporation. HarvardBusiness School Press, Boston, MA.

Ravenscraft, D., Scherer, F.M., 1982. The lag structure of returns to research anddevelopment. Appl. Econ. 14, 603–620.

Romijn, H., Albaladejo, M., 2002. Determinants of innovation capability in smallelectronics and software firms in southeast England. Res. Policy 31, 1053–1067.

Rothaermel, F.T., Alexandre, M.T., 2009. Ambidexterity in technology sourcing: themoderating role of absorptive capacity. Organ. Sci. 20, 759–780.

Rothwell, R., 1992. Successful industrial innovation: critical factors for the 1990s.R&D Manage. 22, 221–240.

Schmidheiny, K., 2011. Panel Data: Fixed and Random Effects. Short Guides toMicroeconometrics. Retrieved from ⟨http://www.schmidheiny.name/teaching/panel2up.pdf⟩.

Schmidt, T., 2010. Absorptive capacity-one size fits all? A firm-level analysis ofabsorptive capacity for different kinds of knowledge. Managerial Decis. Econ.31, 1–18.

Schmiedeberg, C., 2008. Complementarities of innovation activities: an empiricalanalysis of the German manufacturing sector. Res. Policy 37, 1492–1503.

Schneider, B., 1987. The people make the place. Pers. Psychol. 40, 437–453.Spanos, Y.E., Voudouris, I., 2009. Antecedents and trajectories of AMT adoption: the

case of Greek manufacturing SMEs. Res. Policy 38, 144–155.Spithoven, A., Clarysse, B., Knockaert, M., 2011. Building absorptive capacity to

organise inbound open innovation in traditional industries. Technovation 31,10–21.

Stock, G.N., Greis, N.P., Fischer, W.A., 2001. Absorptive capacity and new productdevelopment. J. High Technol. Manage. Res. 12, 77–91.

Teece, D.J., 1986. Profiting from technological innovation: implications for integra-tion, collaboration, licensing and public policy. Res. Policy 15, 285–305.

Tether, B.S., 2002. Who co-operates for innovation, and why: an empirical analysis.Res. Policy 31, 947–967.

Tidd, J., Trewhella, M.J., 1997. Organizational and technological antecedents forknowledge acquisition and learning. R&D Manage. 27, 359–375.

Todorova, G., Durisin, B., 2007. Absorptive capacity: valuing a reconceptualization.Acad. Manage. Rev. 32, 774–786.

Tsai, K.H., Wang, J.C., 2008. External technology acquisition and firm performance:a longitudinal study. J. Bus. Ventur. 23, 91–112.

Tsai, W., 2001. Knowledge transfer in intraorganizational networks: effects ofnetwork position and absorptive capacity on business unit innovation andperformance. Acad. Manage. J. 44, 996–1004.

Vega-Jurado, J., Gutiérrez-Gracia, A., Fernández-de-Lucio, I., 2009. Does externalknowledge sourcing matter for innovation? Evidence from the Spanish man-ufacturing industry. Ind. Corp. Change 18, 637–670.

Vega‐Jurado, J., Gutiérrez‐Gracia, A., Fernández‐de‐Lucio, I., 2008. Analyzing thedeterminants of firm's absorptive capacity: beyond R&D. R&D Manage. 38,392–405.

Veugelers, R., Cassiman, B., 1999. Make and buy in innovation strategies: evidencefrom Belgian manufacturing firms. Res. Policy 28, 63–80.

von Hippel, E., 1988. The Sources of Innovation. Oxford University Press, New York.Watanabe, C., Hur, J.Y., 2004. Resonant R&D structure for effective technology

development amidst megacompetition–an empirical analysis of smart coop-erative R&D structure in Japan's transport machinery industry. Technovation24, 955–969.

Xia, T., 2013. Absorptive capacity and openness of small biopharmaceutical firms—aEuropean Union–United States comparison. R&D Manage. 43, 333–351.

S.Y. Han, S.J. Bae / Int. J. Production Economics 150 (2014) 58–7372

Page 16: Internalization of R&D outsourcing: An empirical study

Yasuda, H., 2005. Formation of strategic alliances in high-technology industries:comparative study of the resource-based theory and the transaction-costtheory. Technovation 25, 763–770.

Yoshikawa, T., 2003. Technology development and acquisition strategy. Int. J.Technol. Manage. 25, 666–674.

Zahra, S.A., 1996. Technology strategy and new venture performance: a study ofcorporate-sponsored and independent biotechnology ventures. J. Bus. Ventur.11, 289–321.

Zahra, S.A., George, G., 2002. Absorptive capacity: a review, reconceptualization,and extension. Acad. Manage. Rev. 27, 185–203.

Zahra, S.A., Hayton, J.C., 2008. The effect of international venturing on firmperformance: the moderating influence of absorptive capacity. J. Bus. Ventur.23, 195–220.

Zander, U., Kogut, B., 1995. Knowledge and the speed of the transfer andimitation of organizational capabilities: an empirical test. Organ. Sci. 6,76–92.

S.Y. Han, S.J. Bae / Int. J. Production Economics 150 (2014) 58–73 73