Towards an open R&D system: Internal R&D investment, external knowledge acquisition and innovative performance

Towards an open R&D system: Internal R&D investment, external knowledge acquisition and innovative performance

Research Policy 42 (2013) 117–127 Contents lists available at SciVerse ScienceDirect Research Policy journal homepage: www.elsevier.com/locate/respo...

438KB Sizes 0 Downloads 4 Views

Research Policy 42 (2013) 117–127

Contents lists available at SciVerse ScienceDirect

Research Policy journal homepage: www.elsevier.com/locate/respol

Towards an open R&D system: Internal R&D investment, external knowledge acquisition and innovative performance Luca Berchicci ∗ RSM Erasmus University, Strategy Management & Entrepreneurship, Burg. Oudlaan 50, Office: T07-34, 3062 PA Rotterdam, The Netherlands

a r t i c l e

i n f o

Article history: Received 30 March 2010 Received in revised form 16 April 2012 Accepted 28 April 2012 Available online 23 May 2012 Keywords: Internal and external R&D activities R&D structure R&D configuration R&D capacity Manufacturing firms Innovative performance Complementarity vs substitution

a b s t r a c t To cope with fast-changing business environments, firms are increasingly opening up their organizational boundaries to tap into external source of knowledge. By restructuring their R&D system, firms face the challenge of balancing internal and external R&D activities to profit from external knowledge. This paper examines the influence of R&D configuration on innovative performance and the moderating role of a firm’s R&D capacity. The findings suggest that firms that increasingly rely on external R&D activities have a better innovative performance, yet up to a point. Beyond this threshold, a greater share of external R&D activities reduces a firm’s innovative performance. And such substitution effect is larger for firms with greater R&D capacity. Overall, this paper provides a better understanding of the open innovation paradigm by suggesting that the opportunity cost for further opening up R&D borders is higher for firms with a superior technological knowledge stock. © 2012 Elsevier B.V. All rights reserved.

1. Introduction Over the past years firms have increasingly relied on external sources of knowledge in their R&D processes to develop and profit from innovations (Calantone and Stanko, 2007; Linder et al., 2003). The conventional paradigm of having organizational core R&D activities exclusively in-house is becoming less critical, while more recent models of innovation suggest how firms are ‘opening’ up their R&D borders to tap into external sources of knowledge (Chesbrough, 2003). Tapping into external technology sourcing alleviates some of the challenges firms face such as shorter product life cycles, faster product renewal and increasing R&D costs (Rigby and Zook, 2002). On the other hand, searching for and coordinating an increasing number of new collaborations are activities that require greater investments in time and money. Consequently, higher transaction costs may erode the benefits of new external R&D activities. As firms start to systematically open up their R&D borders, they adapt and fine-tune their R&D configuration – their internal and external R&D processes – to build new or reinforce existing relationships with a diverse range of partners. Given the importance of R&D processes, the difficult task for managers is to find a balance between internal and external R&D activities in order to capture the benefit from external technology sources.

∗ Tel.: +31 10 408 9608; fax: +31 10 408 9013. E-mail address: [email protected] 0048-7333/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.respol.2012.04.017

This paper addresses this issue by investigating how the tradeoff between internal and external R&D processes influences a firm’s innovative performance. In particular, it focuses on how a firm’s internal R&D capacity – internal R&D investment in building stock of knowledge – moderates the relationship between a firm’s R&D structure and its innovative performance. Prior research suggests firms can tap more efficiently into external sources of knowledge by investing in own R&D. Firms that invest in building an internal R&D stock of knowledge are better able to recognize and evaluate external sources and in turn to integrate and use their knowledge (Cohen and Levinthal, 1990). Moreover, they often rely on fewer yet more valuable linkages to achieve greater innovative output (Arora and Gambardella, 1994). Since selection and assimilation of external knowledge depend on a firm’s stock of knowledge (Cohen and Levinthal, 1990), it is relevant to know how internal R&D capacity influences the relationship between the degree of R&D outsourcing and innovation performance. By investigating the moderating role of R&D capacity in balancing internal and external R&D activities, this paper explores the conditions in which the open innovation paradigm matters for greater innovative performance. By doing so, it contributes to the literature in two ways. First, by building on a study by Cassiman and Veugelers (2006), this paper tests the extent to which internal and external R&D activities are complementary or substitute for greater innovative performance. Whereas Cassiman and Veugelers (2006) investigate how each of the distinctive R&D structures (Make, Buy and Make & Buy) influences innovation performance (using

118

L. Berchicci / Research Policy 42 (2013) 117–127

essentially three models with three dichotomous variables), this paper focuses on the degree of R&D outsourcing. By using a continuous approach to the Cassiman and Veugelers’ typology, this paper provides a better understanding of the benefits and drawbacks in opening up a firm’s R&D borders and in trading off internal and external R&D activities. Second, prior research has emphasized the moderating role of internal R&D investment in capturing unintentional external knowledge flows (Escribano et al., 2009) to achieve better innovative output, but has not taken into account how such firm’s R&D capacity influences the knowledge flow through external R&D activities. By examining the moderating role of R&D capacity in this context, this paper provides new insights in the ability of firms to capture value by balancing internal and external R&D activities. More critically, it provides a contextual variable that allows us to better assess the complementarity vs substitution dichotomy. To address these issues empirically, this paper investigates the internal and external R&D configuration of R&D-intensive Italian manufacturing firms. Based on two survey waves, I find that firms with an internal and external R&D system have greater innovative performance. Yet those firms that carry out more external than internal R&D activities perform worse. Moreover, I find that R&D capacity significantly moderates this curvilinear relationship. Firms with greater R&D capacity are able to benefit more from their external R&D activities in terms of innovative output. And they are able to do so by utilizing a smaller share of external R&D activities than those firms with a lower R&D capacity. These findings provide a deeper understanding of the relationship between internal and external R&D that goes beyond the classic opposition between complementarity and substitution. They imply that internal and external R&D activities are complementary up to a point after which they are substitute. More critically, the substitution effect is larger for firms with greater R&D capacity. Overall, these results provide a better understanding of the open innovation paradigm by suggesting that the opportunity cost for further opening up R&D borders is greater for firms with greater internal R&D capacity. This paper is organized as follows. The next section examines the literature on R&D configuration, internal R&D capacity and innovation performance. It proposes a set of hypotheses that drives the analysis. The third section describes the database and the method. Finally, the results are elaborated and discussed.

2. Theoretical framework 2.1. From a closed to an open R&D system To cope with an increasingly competitive environment, firms constantly invest in innovative activities and in creating technological capabilities. Nevertheless, focusing only on internal R&D and the development of internal capabilities and routines is no longer sufficient to cope with increasing costs, shorter product life cycles and greater technological complexities. These drivers have drastically mutated organizations, where the monolithic structure of an internally closed R&D is rapidly fading and shifting from a vertically integrated in-house R&D structure to an open R&D structure by tapping into external sources of knowledge through licensing, alliances and technology agreements (Hagedoorn, 1993). As illustrated by Whittington (1990), the ratio of internal vs external R&D expenditures more than doubled between 1967 and 1986, while R&D partnership has been growing tenfold in the last three decades (Hagedoorn, 2002). The earlier models of innovation depicted it as an internally controlled process. The firm was the locus of innovation and the innovation process was kept away from competitors and other external players to secure that the knowledge was kept in-house.

When those innovations left the R&D labs, the successful ones were able to finance the subsequent in-house R&D activities. Since these activities were thought to be firm-specific, there was no need for cost sharing with other firms (Chandler and Hikino, 1990). However, this model of innovation was not always very efficient since, as noted earlier by Nelson (1959), it did not prevent spillovers. Firms funded R&D projects whose output was often appropriated and commercialized somewhere else. Instead of closed innovation, one of the most recent models suggests an open innovation paradigm, where the R&D structure should be seen as an open system (Chesbrough, 2003; Chesbrough et al., 2006). This paradigm assumes that “firms can and should use external ideas as well as internal ideas, and internal and external paths to market, as firms look to advance their technology” (Chesbrough, 2003, p. 24). Firms are no longer the exclusive locus of innovation, but external and internal knowledge are equally important. An open R&D system allows firms to outsource R&D projects or technologies with no clear paths to market. By being exposed to external partners, these R&D projects may eventually find their way to market. It also allows firms to in-source external ideas, through the integration of suppliers, customers and external knowledge sources to increase firm innovativeness. However, according to Chesbrough and Teece (1996), openness implies an engagement with external sources, not a total reliance on them. Firms that depend entirely on external partners may lack internal R&D processes themselves and the ability to fully capture and assimilate external knowledge. This literature, however, does not explicitly evaluate the role of R&D capacity in balancing internal and external R&D activities for greater innovative output. The aim of this paper is to investigate this role. The next section discusses theories and relevant empirical research on the relationship between R&D structure and innovative performance, the role of R&D capacity, and offers a set of hypotheses. 2.2. Internal and external R&D processes and innovative performance Various theories of firm behavior explain the shift from closed to open innovation models and the increasing reliance on external R&D activities. The transaction cost of economics (TCE) perspective suggests that the organization of economic activities is driven by the minimization of both production and transaction costs. From this perspective, the rise of R&D labs in the late 1940s could be attributed to lower costs of organizing and managing innovation in-house rather than through the market (Mowery and Rosenberg, 1989), given that R&D activities were considered firm-specific. As costs associated with R&D have been increasing (in TCE terminology, production costs), firms have minimized these costs by sharing them with other firms (Katz, 1986). Firms also realize that some R&D activities in non-core technology areas are not firm-specific and, therefore, they can either have joint R&D activities with other partners or outsource some of them because of the benefits in terms of cost saving and innovative output (Hagedoorn, 2002). Rather than focusing on cost minimization, other theories emphasize how knowledge sharing and inter-firm linkages allow firms to achieve better performance (Dyer and Singh, 1998; Grant, 1996). Given the fast-changing technology environment, the knowledge base of the firm perspective suggests that firms could broaden their existing technology base and access new technology areas by exploring and integrating different specific knowledge areas through internal R&D activities and external technology outsourcing (Kogut and Zander, 1992). Since new developments in non-core technologies are increasingly fast, firms have limited capacity to screen and manage technological knowledge in-house.

L. Berchicci / Research Policy 42 (2013) 117–127

Similarly, the relational view of the firm perspective argues that inter-firm linkages and technology outsourcing allow firms to keep up with novel developments so that they can increase the learning gains from cooperation. The extension of a firm’s technological capabilities increases the chances to develop and realize new products (Becker and Dietz, 2004). Moreover, trust (Zaheer et al., 1998) and long-term idiosyncratic relations (Dyer and Singh, 1998) could also enable greater performance. Beside return on investment, previous scholars have found that inter-firm linkages affect innovative firm performance considerably (Shan et al., 1994; Stuart, 2000). Another stream of research suggests that firms do not necessarily need to have in-house knowledge or capabilities as long as they have access to them (Barney, 1991; Lavie, 2006; Penrose, 1959). Technology outsourcing could still provide firms with opportunities to strengthen their knowledge and capabilities. These scholars argue that external technology sources are beneficial to the firm provided they enrich the firm’s knowledge stock and exploit external specialized resources (Mitchell and Singh, 1996; Mowery et al., 1996; Powell et al., 1996; Steensma and Corley, 2000; Womack et al., 1990), which in turn could enhance innovative performance in terms of product variety and time to market (Eisenhardt and Schoonhoven, 1996; Schoonhoven et al., 1990). The increase in external R&D processes and relational interfirm linkages brings new organizational challenges as well. From a TCE perspective, as a firm pursues a greater share of R&D activities, the costs of searching and selecting suitable R&D partners are likely to rise and additional resources need to be allocated. Second, the shift from closed to open innovation is likely to increase transaction costs since it requires greater effort to coordinate, manage and control the R&D activities of the partners involved (Gulati and Singh, 1998). Due to a greater difference between the focal firm and its R&D partners in terms of information and control systems, and decision processes, coordination costs are likely to rise as well (Dyer and Singh, 1998). Third, as firms move to an open R&D system, their internal R&D structure requires a fundamental transformation, since its role shifts “from discovery generation as the primary activity to systems design and integration as the key function” (Chesbrough, 2005, p. 15). These organizational changes could imply greater costs while the benefits of an open R&D system might only be observable in the long term. Finally, scholars from the Resource-Based View (RBV) of the firm perspective warn against external R&D processes when the external knowledge flow slows down learning-by-doing and impedes the building of pathdependent knowledge stocks inside the firm (Bettis et al., 1992). Despite the rich theoretical explanations, the empirical evidence of the effect of R&D configuration on firm performance is rather limited. One of the first and extensive studies was performed by Arora and Gambardella (1994). In their investigation of collaborations in the biotechnology industry, they found that firms with strong internal technology know-how were able to better utilize external innovation information. Becker and Dietz (2004) found that in the German manufacturing industry R&D collaboration complements internal resources and enhances product innovation implementation. By investigating discrete strategic choices of ‘Make only’, ‘Buy only’ and ‘Make & Buy’, Cassiman and Veugelers (2006) tested, whether the combination of internal and external R&D is associated with a greater firm’s ability to innovate. They found that firms with a ‘Make & Buy’ strategy are more likely to profit from innovation. Their results suggest that neither ‘Make only’ nor ‘Buy only’ lead to greater innovative performance. Other studies, however, suggest a substitution effect. Audretsch et al. (1996) found that internal and external R&D are substitute for firms in low-technology industries. Laursen and Salter (2006) found that a negative effect of over-searching for external knowledge flow. Both theoretical arguments and empirical research suggest that opening up a firm’s R&D system is beneficial for the innovative

119

performance but only up to a point. Relying heavily on external technology sourcing increases search, coordinating and monitoring costs and could hamper the building of path-dependent knowledge stocks within the firm. Therefore, I hypothesize: Hypothesis 1. An inverted U-shaped relationship exists between the share of external R&D activities and a firm’s innovative performance. 2.3. R&D capacity, R&D configuration and innovative performance Specific firm’s factors can influence the relationship between R&D openness and innovative performance. This section discusses the role of R&D capacity – defined as a firm’s investment in internal R&D to build stock of knowledge – in moderating the effect of R&D configuration on innovative performance. A firm’s stock of knowledge allows firms to perform two crucial activities. First, it enables to develop and produce new products and processes to better compete and survive in the market. Second, and more importantly for the argument of this paper, it allows to evaluate and tap into external source of knowledge. While the first activity is self-explanatory, the latter needs a more exhaustive explanation. Prior research suggests that the ability of the firm to acquire external knowledge is a by-product of a firm’s own internal R&D (Arora and Gambardella, 1994; Cassiman and Veugelers, 2002; Cohen and Levinthal, 1990; Rosenberg, 1990). Through its R&D activities, a firm builds up a stock of knowledge about specific fields of technology, which connects to its products. Over time, a firm becomes skilled in acquiring external knowledge in domains that are close to its own and its investments in processes and procedures facilitate the sharing of the acquired knowledge internally (Cohen and Levinthal, 1990). Beside valuing and assimilating external knowledge, greater R&D capacity allows a firm to recognize new opportunities in the market and to forecast technological trends (Cohen and Levinthal, 1994) and finally to better evaluate opportunities for collaborative R&D projects. Thus, firms with greater R&D capacity have a developed technology base that allows them to both produce new knowledge and better evaluate the knowledge offered by the external environment. Empirical evidence confirms these arguments. Firms better utilize R&D cooperation when they have a dedicated internal R&D department and personnel (Veugelers, 1997). By focusing on involuntary knowledge flows, Escribano et al. (2009) found that higher internal R&D investment allows firms to tap into external knowledge sources more efficiently and, in turn, stimulate innovative output. The aforementioned theoretical and empirical discussion brings us to a key question: how do different levels of R&D capacity influence the relationship between R&D configuration and innovative performance? Firms that complement internal R&D activities with external R&D activities and have a high level of R&D capacity can better evaluate and assimilate knowledge from external environment. New knowledge can be combined with a solid existing technology knowledge base, which could bring additional opportunities and insights for new products and markets. Furthermore, these firms better understand technological trends and future opportunities and are better equipped to recognize the value of external knowledge, to assimilate it internally and to apply it for innovative outcomes (Cohen and Levinthal, 1990). Given an internal R&D configuration, a firm with greater R&D capacity incurs lower search costs than a similar firm with lower R&D capacity because the former is able to better select promising projects and partners. Arora and Gambardella (1994) found that “firms with better ability to evaluate (collaborative R&D projects) are more selective and focus on fewer but more valuable linkages” (p. 109). By establishing

120

L. Berchicci / Research Policy 42 (2013) 117–127

privileged relationships with a limited and valuable number of partners, these firms can reduce coordinating and monitoring costs. Since it is difficult to transfer knowledge between firms (Grant, 1996; Mowery et al., 1996), acquiring knowledge from external R&D partners requires an interacting process over time. A greater degree of involvement with fewer, valuable partners enables firms with a robust stock of knowledge to build a common understanding of their R&D collaborative activities and to make fine adjustments in coordination and task processes. Moreover, these valuable interfirm linkages may develop into long-term idiosyncratic relations, which allow firms to keep up with novel developments so that they can increase the learning gains from cooperation (Dyer and Singh, 1998). In turn, a strong technological base allows the extension of a firm’s technological capabilities and increases the odds of developing and realizing new products (Becker and Dietz, 2004). On the contrary, firms engaging in both internal and external R&D activities, yet not investing in an internal stock of knowledge, may experience relatively higher search costs. With a weak stock of knowledge, the ability to recognize valuable linkages is less developed and consequently relatively more time is needed to select useful partners. As R&D partnerships are established, the limited ability of these firms to value and assimilate external knowledge leads to an increase of transaction and coordination costs, because greater effort is required to manage R&D activities with partners and ultimately to benefit from them. At the other extreme, firms that rely mainly on acquiring external R&D tend to have a great range of R&D partners with a limited stock of internal knowledge. With a limited basic R&D capacity within the firm, new external knowledge is hard to screen, recognize, exploit and finally benefit from (Arora and Gambardella, 1994; Rosenberg, 1990). In summary, a high level of R&D capacity allows firms that rely on internal R&D to recognize and select valuable linkages and to capture the know-how of the partners more efficiently. Thus, I hypothesize that R&D capacity influences the relationship between R&D outsourcing and performance. Given the curvilinear relationship, firms with a higher level of R&D capacity show a higher point of maximum efficiency than firms with lower level of R&D capacity. And I expect that such maximum point for firms with high level of R&D capacity is reached at the smaller share of external R&D due to their ability to benefit from fewer and more valuable collaborations. Hypothesis 2a. The inverted U-shaped relationship between the share of external R&D activities and innovative performance is moderated by R&D capacity in such a way that greater R&D capacity is associated with a higher point of maximum efficiency in the inverted U-shaped curve. Hypothesis 2b. The inverted U-shaped relationship between the share of external R&D activities and innovative performance is moderated by R&D capacity in such a way that greater R&D capacity is associated with a higher point of maximum efficiency in the inverted U-shaped curve for a smaller share of external R&D.

Table 1 General description of the final sample.

Number of observations (firms) Firms that reported sales from new or significantly improved products (%) R&D expenditure Rate of external R&D Average number of employees Average number of employees in R&D

1998–2000

2001–2004

All years

1528 46.24%

1377 68.70%

2905 (2,537) 56.90%

1.82% 27% 117 7.4

1.69% 20.40% 148 7.8

1.76% 23.80% 135 7.5

conducted in 1992 (covering the period 1989–1991), the second survey in 1995 (covering 1992–1994), the third survey in 1998 (covering 1995–1997), the fourth survey in 2001 (covering 1998–2000), and the last in 2004 (covering the period 2001–2003). All firms with more than 500 employees were included in the survey, whereas those with fewer than 500 employees were selected using a sampling design stratified by geographical area, industry and firm size.3 From these five surveys, this study merged data from the 2001 and 2004 surveys to construct an unbalanced panel data. Since this paper focuses on R&D configuration, the final sample includes those firms that performed internal R&D activities only, those with external R&D activities only and those with a combination of both internal and external activities. Hence, from a total of 8,969 observations (and 6,872 firms), the final sample consists of 2,905 observations and 2,537 firms. These firms represent 21 industries (at the two-digit level) of the manufacturing sector – according to the ATECO classification system.4 The majority of the firms come from the industries “Machinery and Equipment,” (21%) “Fabricated Metal Products, Except Machinery and Equipment,” (10%) “Food Products and Beverages,” (8%) “Chemical and Chemical Products” (7%) and “Textiles,” (7%) which together account for more than 50% of the entire sample. Table 1 illustrates some of the characteristics of the final sample. It is noteworthy that the average of external R&D activities decreased from 27% in the 1998–2000 period to 20.40% in the 2001–2003 period. 3.1. Dependent variable To measure a firm’s innovative performance, the SIMF reports the share of turnover from new or significantly improved products. Thus the Share of innovative sales variable is used as dependent variable, which measures the share of turnover in the last year of the survey due to new or significantly improved products introduced in the previous three years. The variable is expressed in percentage going from zero (no turnover from selling innovative products) to 100%. This type of variable is often used in innovation studies (Cassiman and Veugelers, 2006; Escribano et al., 2009; Laursen and Salter, 2006; Tsai, 2009), because it directly measures the success of new or significantly improved products in the market.

3. Method 3.2. Independent variables The data used for this study derive from the Surveys of Italian Manufacturing Firms (SIMFs) conducted by the Research Department of Capitalia Banking Group.1 Five surveys were carried out from 1992 to 2004 using questionnaires sent to a representative sample of Italian manufacturing firms.2 The first survey was

1 On October 1, 2007, Capitalia was acquired and merged into Unicredit Group, one of the largest banks in the world. 2 The questionnaire and methodology for the survey related to the innovation and technology section is similar to that adopted for the Community Innovation Survey (CIS).

The first independent variable of interest is External R&D, which captures the extent to which firms engage in external R&D activities. In the surveys, managers are asked to indicate which percentage of R&D activities is outsourced and which percentage is performed in-house. The value of the External R&D variable goes

3 The respondents are usually owners and CFOs of SMEs and heads of accounting of larger firms. 4 The Italian version of the North American Industry Classification System (NAICS) or the NACE European Classification.

L. Berchicci / Research Policy 42 (2013) 117–127

from zero (when R&D activities are in-house only) to 100% (when R&D activities are fully outsourced). The second variable of interest is R&D capacity. Its measure should capture the inherently cumulative process that the development of a firm’s stock of knowledge entails. The measure assessing it should not be strongly correlated to external R&D activities but should capture such inherent cumulativeness. In this study R&D capacity is calculated as the number of employees working in the R&D department divided by the total number of employees in the firm (Cassiman and Veugelers, 2002; Escribano et al., 2009; Veugelers, 1997). Prior research has used this variable as proxy of absorptive capacity (see e.g. Cassiman and Veugelers, 2002; Escribano et al., 2009), – defined as a firm’s ability to value, assimilate, and commercially utilize new, external knowledge (Cohen and Levinthal, 1990). Although absorptive capacity is intertwined and linked with R&D capacity, its measurement presents a number of identification problems in the current setting. First, to be able to capture absorptive capacity, one needs to clearly separate the “two faces of R&D” (Cohen and Levinthal, 1989), the internal production of new knowledge and the external acquisition of new knowledge. Without such separation it is difficult to fully evaluate the effect of absorptive capacity. Second, absorptive capacity was originally measured in a single industry (Cohen and Levinthal, 1989). Since the current data cover a broad range of industries it is difficult to capture absorptive capacity due to the high level of firm’s heterogeneity. Instead, R&D capacity directly measures the effort of a firm to build a stock of knowledge, which allows one to produce and acquire new knowledge across industries. 3.3. Control variables Prior research suggests that there are specific factors that influence the innovative performance of a firm, which need to be controlled for. For example, the size and age of a firm are wellknown factors that affect a firm’s innovative output. Firm’s Size is included as the logarithmic form of the number of firm employees. Age is calculated as the logarithmic form of the number of years from the firm’s establishment. Export is another firm characteristic included as a control variable and identifies whether the firm engages in export activities. Firms operating in an international market are more likely to develop innovative products than those active in a domestic market due to stronger competition (Basile, 2001). Finally, Wave 2004 captures the survey conducted in 2004. Although not reported in the following tables, each model includes dummy variables for each industry that is present in the sample. 3.4. Models The dependent variable has two peculiar characteristics. The first is the percentage of sales from innovative products, which ranges between 0 and 100. Second, since many firms reported no sales from innovative products, one-third of the observations have a value equal to zero for the dependent variable. For these two reasons, scholars suggest that it is appropriate to use the Tobit estimation model (Gujarati, 1995). Since 367 firms are present in both waves (for a total of 734 observations) the cluster option in Tobit estimations takes into account that some observations are not independent by clustering the standard errors at a firm level. By doing so, it controls for the correlation of residuals to obtain robust standard errors. Following previous research, the dependent variable is used in its logarithmic form since it reduces the problem of non-normality of the residuals (Greene, 2003; Laursen and Salter, 2006). To test the above-mentioned hypotheses, two models are proposed. The first model investigates the effect of external R&D on innovative performance for firm i that perform R&D (internal R&D

121

only, external R&D only or a combination of both internal and external R&D). Share of innovative salesi = ˇ0 + ˇ1 R&D capacityi + ˇ2 External R&Di + ˇ3 External R&D2i + ˇ4 ln(Sizei ) + ˇ4 ln(agei ) + ˇ5 Exporti + ˇ6 wave + Industry dummies + εi (1) The second model investigates how R&D capacity moderates the relationship between the share of external R&D activities and innovative performance. The equation adds a linear by linear interaction between R&D capacity and External R&D to Eq. (1). To better infer the result from the conditional effect of R&D capacity, the equation is displayed to facilitate its interpretation such as: Share of innovative salesi = (ˇ2 + ˇ7 R&D capacityi ) × External R&Di + ˇ3 External R&D2i + (ˇ1 R&D capacity + ˇ4 ln(sizei ) + ˇ4 ln(agei ) + ˇ5 Exporti + ˇ6 wave + Industry dummies + ˇ0 ) + ε

(2)

where in Eq. (2) (ˇ2 + ˇ7 R&D Capacity) indicates the overall linear trend of Y (share of innovative sales) on External R&D at one or more values of R&D capacity. If (ˇ2 + ˇ7 R&D Capacity) is positive, the Tobit estimation has an overall upward linear trend, otherwise it has an overall downward linear trend. 4. Results Table 2 presents the descriptive statistics and shows a negative and weak correlation between R&D capacity and External R&D (−0.18). It suggests that firms with a relatively larger R&D department tend to perform R&D activities mainly in-house. Table 3 presents three Tobit models with robust estimations. Model 1 shows the relationship between the control variables and innovative performance. Firm’s Size and Export positively affect innovative performance, while the contribution of firm’s Age is not significant. Turning to R&D capacity, its coefficient is positive and significant. This finding suggests that building a stock of knowledge strongly influences a firm’s innovative output. Model 2 adds the External R&D variable, which shows a negative yet not significant effect on innovative performance. Model 3 investigates whether External R&D has a non-linear effect and includes its squared term. The External R&D variable has a positive and significant coefficient (p value < 0.001) while External R&D squared has a negative and significant coefficient (p value < 0.001). Model 3 shows also a larger chi-square value than Model 2 or Model 1. This suggests that the addition of the main effect of External R&D and its squared term increases the explanatory power of the model. Taken together, these two effects suggest that External R&D has a curvilinear effect on innovative performance. To ease the interpretation, these effects are shown graphically. Fig. 1 illustrates an inverted U-shaped line that captures the relationship between External R&D and the share of innovative sales. It implies that firms that engage in external R&D activities have greater benefits in term of innovative performance. Yet they do so only up to a point. Beyond this threshold, greater external R&D reduces a firm’s innovative performance. This finding supports Hypothesis 1. Turning to hypotheses 2a and 2b, Model 4 illustrates the conditional effect of R&D capacity on the relationship between External R&D and innovative performance by applying Eq. (2). The

122

L. Berchicci / Research Policy 42 (2013) 117–127

Table 2 Descriptive statistics.

1 2 3 4 5 6 7

Variables

Mean

Standard deviation

Min

Max

Share of innovative sales R&D capacity External R&D Size Age Export Wave 2004

1.68 0.08 23.80 4.08 3.18 0.85 0.47

1.61 0.08 32.19 1.10 0.63 0.36 0.50

0 0 0 2.48 1.10 0 0

4.62 0.63 100 9.06 5.17 1 1

1

2

3

4

5

6

7

1 0.13 −0.06 0.08 0.02 0.10 0.11

1 −0.18 −0.25 −0.09 −0.05 −0.03

1 −0.14 −0.06 −0.07 −0.10

1 0.20 0.21 0.19

1 0.06 0.15

1 0.09

1

Table 3 External R&D and innovative performance and the moderating effect of R&D capacity. Independent variables

Share of innovative sales as dependent variable Model 1

Model 2

***

R&D capacity Size Age Export Wave 2004

Model 3

***

Model 4

***

4.317 (0.69) 0.211*** (0.051) 0.0360 (0.083) 0.614*** (0.15) 0.837*** (0.11)

4.336 (0.71) 0.212*** (0.052) 0.0363 (0.083) 0.615*** (0.15) 0.838*** (0.11) 0.000227 (0.0017)

3.870 (0.72) 0.185*** (0.052) 0.0290 (0.083) 0.578*** (0.15) 0.867*** (0.11) 0.0185*** (0.0053) −0.000212*** (0.000058)

Yes −2.509*** (0.38) 2905 0.0271 −4827.3 367 1251 39

Yes −2.524*** (0.39) 2905 0.0271 −4827.3 367 1251 39

Yes −2.428*** (0.39) 2905 0.0285 −4820.7 367 1251 39

External R&D External R&D squared R&D capacity × External R&D Dummy variables for each industry (2-digit code) Constant Observations Adjusted R-squared Log likelihood function Number of firms in both waves Left censored Right censored Robust standard errors in parentheses. * p < 0.1. ** p < 0.05. *** p < 0.01.

1.6

Innovative Performance

1.4 1.2 1 0.8 0.6 0.4 0.2 0 0

10

20

30

40

50

60

70

80

External R&D Fig. 1. The relationship between external R&D and innovative performance.

90

100

5.631*** (0.93) 0.189*** (0.052) 0.0204 (0.084) 0.561*** (0.17) 0.871*** (0.10) 0.0305*** (0.0065) −0.000296*** (0.000065) −0.0959*** (0.029) Yes −2.542*** (0.40) 2905 0.0297 −4814.5 367 1251 39

L. Berchicci / Research Policy 42 (2013) 117–127

123

1.6 Low R&D capacity

1.4

Avg. R&D capacity High R&D capacity

Innovative Performance

1.2

1

0.8

0.6

0.4

0.2

0

0

10

20

30

40

50

60

70

80

90

100

External R&D Fig. 2. The moderating effect of R&D capacity on the relationship between external R&D and innovative performance.

interaction variable (External R&D × R&D capacity) is negative and significant (p value < 0.001). This result shows that R&D capacity not only influences a firm’s innovation performance directly but also the relationship between the share of external R&D activities and a firm’s innovative output. However, it is not immediately apparent from the model how its conditional effect works on the given relationship. As suggested by Aiken et al. (1991), one approach is to graph the main effects given the conditional effect under study. Based on Eq. (2), Fig. 2 shows three linear trends of innovative performance on External R&D at three values of the R&D capacity variable. ‘Average R&D capacity’ captures the R&D capacity variable at its mean value (line with a triangle mark), ‘low R&D capacity’ at half standard deviation below its mean (continuous line) and ‘high R&D capacity’ at half standard deviation above the mean (dotted line). When the effect of R&D capacity on the relationship is at its mean value, the curve in Fig. 2 and the curve in Fig. 1 show a similar trend. Relevant findings are found when R&D capacity moves away from the mean. As shown in Fig. 2, higher R&D capacity is associated with greater innovative performance at a lower level of external R&D. For these firms with high R&D capacity, the maximum efficiency – the optimal value of External R&D where the maximum performance value is achieved – is at 26.7% of External R&D, while for firms with low R&D capacity the equilibrium point is at 42% of External R&D. This suggests that firms with greater R&D capacity require relatively less exposure to external knowledge to efficiently recognize and assimilate it. As the External R&D variable increases, the three curves tend to converge. This confluence may reconfirm that a greater share of external R&D is detrimental for innovative performance regardless of the R&D capacity of the firm. Taking a closer empirical look at the curves, we discover that the number of firms with high R&D capacity decreases as the share of external R&D activities increases. Moreover, there is no high R&D capacity firm with more than 80% of External R&D. This observation is not surprising since firms with larger R&D departments relative to their size are less likely to substantially outsource their R&D activities. Taken together, these findings suggest that both hypotheses 2a and 2b are supported. As robustness checks, alternative estimations are performed. First, to reduce risk of multicollinearity (Cronbach, 1987), Aiken

et al. (1991) suggest that the two variables that form the interaction term should be transformed into mean centered variables. The findings are confirmed since the coefficients are much alike. Second, alternative measures of R&D investment in building stock of knowledge are used to test whether the findings are strictly dependent on the idiosyncratic nature of the R&D capacity variable. Prior research suggests that R&D expenditure relative to the firm’s sales and R&D expenditure relative to the number of employees are two common proxies to measure internal R&D investment. When either one or the other is estimated, similar findings are achieved as Table 4 suggests. Third, an additional test is to exclude those firms that outsource most of their R&D activities. Since their internal R&D activities are marginal, they may lack strong technological knowledge. Without such a strong technological base, these firms may tend to focus less on innovative activities and, consequently, underperform in terms of innovative output. Therefore, these potential underperforming firms may bias the estimations by accentuating a downward curvilinear trend. Table 5 replicates the estimations in Table 3, excluding firms that perform more than 80% of external R&D activities. The sample drops to 2,573 observations. Model 1 confirms that the curvilinear relationship between the share of external R&D activities and innovative performance, and Model 2 corroborates the conditional effect of R&D capacity on the relationship between R&D outsourcing and innovative performance. These results confirm and strengthen the explanatory power of the model. Forth, I test the consistency of the role of R&D capacity across industries rather than across firms. By defining each industry by its industry technological effort according to the OECD technology industry classification (Hatzichronoglou, 1997), this test allows one to separate firms by industry type in two sub-samples. The first subsample includes firms in low/medium-low technology industries, while the second one encompasses firms in high/medium-high technology industries. The findings suggest that the role of R&D capacity is equally significant for firms in both types of industries, but its effect is stronger for firms in high/medium-high technology industries. The results of this test are omitted in the interest of parsimony, but are available from the author upon request.

124

L. Berchicci / Research Policy 42 (2013) 117–127

Table 4 Alternative measures of R&D capacity. Independent variables

R&D expenditure over sales

Share of innovative sales Model 1

Model 2

8.518*** (2.49)

11.37*** (2.48)

R&D expenditure over employees Size Age Export Wave 2004 External R&D External R&D squared

0.112** (0.050) 0.0324 (0.084) 0.577*** (0.17) 0.889*** (0.11) 0.0199*** (0.0055) −0.00024*** (0.000060)

R&D expenditure over sales × External R&D

0.110** (0.050) 0.0296 (0.084) 0.566*** (0.17) 0.892*** (0.11) 0.0232*** (0.0058) −0.00026*** (0.000061) −0.140* (0.082)

Model 3

Model 4

0.0624*** (0.012) 0.112** (0.050) 0.0431 (0.084) 0.543*** (0.17) 0.857*** (0.10) 0.0193*** (0.0055) −0.00024*** (0.000060)

0.0848*** (0.014) 0.108** (0.050) 0.0412 (0.084) 0.537*** (0.17) 0.862*** (0.10) 0.0226*** (0.0057) −0.00025*** (0.000060)

Yes −2.003*** (0.39) 2905 0.0284 −4821 367 1615 39

−0.000908** (0.00037) Yes −2.042*** (0.39) 2905 0.0290 −4818 367 1615 39

R&D expenditure over employees × External R&D Dummy variables for each industry (2-digit code) Constant Observations Adjusted R-squared Log likelihood function Number of firms in both waves Left censored Right censored

Yes −1.971*** (0.39) 2905 0.0278 −4828 367 1615 39

Yes −2.002*** (0.39) 2905 0.0281 −4822 367 1615 39

Robust standard errors in parentheses. * p < 0.1. ** p < 0.05. *** p < 0.01.

Table 5 External R&D and innovative performance (firms with more than 80% of external R&D are excluded). Independent variables

R&D capacity Size Age Export Wave 2004

Share of innovative sales Model 1

Model 2

Model 3

Model 4

4.309*** (0.75) 0.225*** (0.053) 0.0205 (0.087) 0.455** (0.18) 0.741*** (0.11)

4.325*** (0.75) 0.226*** (0.053) 0.0215 (0.086) 0.441** (0.18) 0.768*** (0.11) 0.00610** (0.0028)

4.137*** (0.75) 0.206*** (0.053) 0.0196 (0.086) 0.420** (0.18) 0.775*** (0.11) 0.0312*** (0.0087) −0.000478*** (0.00016)

Yes −2.215*** (0.41) 2573 0.0235 −4339.4 315 1066 35

Yes −2.327*** (0.41) 2573 0.0241 −4336.9 315 1066 35

Yes −2.261*** (0.41) 2573 0.0251 −4332.1 315 1066 35

5.620*** (0.96) 0.206*** (0.053) 0.00986 (0.086) 0.407** (0.18) 0.779*** (0.11) 0.0405*** (0.0093) −0.000509*** (0.00016) −0.0965*** (0.037) Yes −2.333*** (0.42) 2573 0.0261 −4328.0 315 1066 35

External R&D External R&D squared External R&D × R&D capacity Dummy variables for each industry (2-digit code) Constant Observations Adjusted R-squared Log likelihood function Number of firms in both waves Left censored Right censored Robust Standard errors in parentheses. * p < 0.1. ** p < 0.05. *** p < 0.01.

L. Berchicci / Research Policy 42 (2013) 117–127

125

Table 6 Fixed effects estimations and Arellano–Bond GMM estimations for those firms present in both waves. Independent variables

Fixed Effects

Arellano–Bond GMM estimator

Share of innovative sales

R&D capacity Size Age Export Wave 2004 External R&D External R&D squared

Model 1

Model 2

Model 3

Model 4

1.218 (1.44) 0.166 (0.35) −0.304 (0.26) −0.235 (0.39) 0.590*** (0.11) 0.0241*** (0.0087) −0.000270*** (0.000094)

2.842* (1.69) 0.159 (0.35) −0.290 (0.26) −0.220 (0.39) 0.583*** (0.11) 0.0348*** (0.010) −0.000338*** (0.00010) −0.0842* (0.046) 1.655 (1.69) 734 367

2.519 (1.77) 0.0938 (0.098) −0.219 (0.19) 0.188 (0.30) 0.524*** (0.11) 0.0190* (0.010) −0.000214* (0.00012)

3.677** (1.79) 0.0747 (0.092) −0.203 (0.18) 0.162 (0.30) 0.526*** (0.11) 0.0298** (0.012) −0.000277** (0.00012) −0.101* (0.054) 1.498 (2.31) 734 367

R&D capacity × External R&D Constant Observations Number of firms

1.843 (1.69) 734 367

1.575 (2.39) 734 367

Standard errors in parentheses. * p < 0.1. ** p < 0.05. *** p < 0.01.

Fifth, to control for unobserved time-invariant heterogeneity, a fixed-effect estimation is performed for those firms that are present in both waves. Since fixed-effect estimations are not suitable for non-linear models such as Tobit models, a standard fixed-effect regression for panel data is used instead. Although the number of observations drops to 734 (and 367 firms), Model 1 and Model 2 in Table 6 suggest that the inverted U-shaped relationship between a greater share of external R&D activities and innovative performance is confirmed and the moderating role of R&D capacity is still negative and significant (p < 0.10). The above-mentioned empirical tests assume that the main variables of interest are exogenous – they are not correlated with the error term. However, this may not be the case since the external R&D variable and the interaction term may not be strictly exogenous. Thus, this final test assumes instead that our main independent variables are endogenous and therefore requires an estimation method that includes instrumental variables (since Tobit or fixed effects estimations are potentially biased). The challenging task is to find a set of valid instruments that are uncorrelated with the errors, but correlated with the endogenous variables and with the dependent variables. With weak instrument variables the results are likely to be biased in the same way of Tobit or OLS estimations. To cope with this problem, I use the Arellano–Bond difference GMM estimator (Arellano and Bond, 1991) (from now, AB estimator). The AB estimator combines instrument variables with first-differences transformations of all the variables improving the overall efficiency of the models. Moreover, the AB estimator is suitable for panel data with few time periods and a great number of observations of individuals or firms (Roodman, 2006). In Table 6, Model 3 and 4 show the AB estimations.5 Overall, the results are consistent to Model 1 and 2 and confirm the suggested hypotheses.

5 The main instrumental variables encompass industry-level variables like the average R&D expenditure per industry and the normalized standard deviation of external R&D per industry and firm-level variables such as the concentration of the different types of R&D partners.

5. Discussion This study investigates the role of R&D configuration in depth and provides a number of important findings. First, R&D openness provides both benefits and costs. With moderate level of external R&D firms are able to capture and exploit the intentional knowledge flow and improve innovative performance. However, firms carrying out more external than internal R&D activities see a decline in their innovative performance. These results imply that restructuring a firm’s R&D configuration to reach a greater number of partners means incurring search and coordination costs. Managers need to allocate time to search and select new R&D partners whose activities fit with their own focal firm. As new partnerships are formed, coordination and monitoring costs need to be considered. These costs are even more relevant when a shared knowledge base and a common understanding of the tasks need to be developed. Second, this study reaffirms the central role of the internal R&D capacity in balancing external and internal R&D activities for greater innovative output. The results demonstrate that the benefits and costs associated with R&D outsourcing are not homogeneous for all firms, but depend on and are intertwined with a firm’s own ability to build internal stock of knowledge. Thus, how a firm manage its own stock of knowledge influences the way in which the R&D system operates to achieve greater innovative performance. As the findings suggest, firms with greater R&D capacity perform systematically better than those with a lower level of R&D capacity. And they capture the optimum of innovative performance with less external R&D than those firms with a lower level of R&D capacity. In other words, at a given output, the former are able to rely more on internal R&D than other firms, as Fig. 2 illustrates. These findings imply that firms with a high level of R&D capacity are more efficient in recognizing and assimilating crucial knowledge from external sources. They are able to select fewer valuable partners from which they assimilate useful knowledge. These results provide an empirical test for Arora and Gambardella’s model (1994), which suggests that innovation involves the exchange of technological information amongst firms, and those better equipped with in-house technological assets have

126

L. Berchicci / Research Policy 42 (2013) 117–127

superior ability to select and assimilate information from fewer partners. Having fewer valuable linkages implies lower coordination and monitoring costs as well. As firms manage fewer partners with high added value, establishing shared processes and task routines requires relatively less effort. Beside lower coordination and monitoring costs, high R&D capacity enables the focal firm to create an organizational ‘fit’ with its R&D partners and efficiently assimilate and use their knowledge (Berchicci, 2011; Rothaermel and Deeds, 2006). This study provides important contributions to open innovation literature by exploring circumstances in which open innovation matters for greater performance and by teasing out the complementary/substituting effect of external R&D. First, it shows that framing the complementarity/substitution issue as a dichotomy does not help us to fully grasp the phenomenon under study. This study suggests that external R&D is complementary to internal R&D up to a point after which it has a substituting effect. By focusing on external R&D, it complements prior studies that indicate both complementary and substitution effects in other settings such as information search (Laursen and Salter, 2006) and alliance portfolios (Deeds and Hill, 1996; Rothaermel and Deeds, 2006). Second, the investigation of the role of R&D capacity provides a specific context to further examine the complementary and substituting effects of external R&D. As Fig. 2 suggests, the substitution effect is larger for firms with greater R&D capacity. But a complementarity effect still exists: even firms with greater R&D capacity need to get external R&D to achieve greater performance. This result implies that firms investing in building high internal R&D capacity do so by substituting external knowledge. And firms investing in their internal R&D capacity perform better than firms that invest in external outsourcing. In other words, the opportunity cost for further opening up R&D borders is higher for firms with a superior technological knowledge stock. Third, by including a wide range of low and high intensive R&D firms, this study provides a richer test of the open innovation paradigm than prior studies that often focused on high intensive R&D firms only (Cassiman and Veugelers, 2006; Arora and Gambardella, 1994). Thus there are important implications for low intensive R&D firms as well. Fig. 2 suggests that firms with lower internal R&D capacity make use of more external R&D, which could imply that greater R&D outsourcing is carried out by ‘weaker’ R&D firms – they invest in R&D outsourcing rather than in building internal R&D capacity. Thus, they are more inclined to R&D openness than those firms with greater R&D capacity.

6. Conclusion Due to a fast-changing innovation environment, firms are increasingly turning their R&D labs into R&D open systems to be able to tap into external sources of knowledge. The effective organization of the R&D system is a crucial challenge for a firm’s future innovative activities. To better understand how firms organize their R&D system with external knowledge partners, this study examines the influence of R&D configuration on innovative performance and the moderating role of a firm’s R&D capacity. In summary, this paper contributes to our understanding of the effect of the trade-off between internal and external R&D processes on a firm’s innovative performance. Firms that move the boundaries of their R&D configuration by engaging in external R&D activities need to balance the benefits from tapping into external sources and the costs of searching, coordinating and monitoring linkages. This paper highlights how a focal firm’s technological capabilities and its internal stock of knowledge influence such a balance. Firms with a high level of R&D capacity are better able to capture and exploit external knowledge through R&D collaborations in terms of innovative output, by investing relatively less in external R&D

activities than other firms. Overall, this study provides a test to the open innovation paradigm by exploring under which conditions greater R&D openness benefits innovative performance. The findings provide some managerial implication as well. This study indicates that the average external R&D percentage (23%) of Italian firms is below the average optimal percentage (34%) in relation to innovative performance (see Fig. 1). This suggests that managers of firms with relatively low external R&D could reap greater benefits in terms of innovative performance by increasing the percentage of their external R&D. They, however, need to take into account their own technological knowledge base and R&D capabilities to capture value from R&D collaborations in an effective manner. This study thoroughly examines the role of R&D configuration and R&D capacity on a firm’s innovative performance, yet it faces some important limitations. First, the sample encompasses only two waves of Italian manufacturing firms. A greater longitudinal set could provide greater exploratory power. Second, the results could be generalized to industry systems that are similar to the Italian ones, where SMEs are predominant. Third, it focuses on the role of external R&D activities without investigating the characteristics of R&D partners. Examining the types of R&D partners and the linkages with the focal firm merits further inquiry. How, for example, do firms structure their external R&D activities? How diverse is their R&D collaboration portfolio? Finally how does R&D partnership diversity influence a focal firm’s innovative performance? Future research could focus on the nature of R&D collaborations providing an even more thorough examination of the role of R&D system on a firm’s performance. References Aiken, L.S., West, S.G., Reno, R.R., 1991. Multiple Regression: Testing and Interpreting Interactions. Sage Publications, Newbury Park, Calif. Arellano, M., Bond, S., 1991. Some tests of specification for panel data – Monte-Carlo evidence and an application to employment equations. Review of Economic Studies 58 (2), 277–297. Arora, A., Gambardella, A., 1994. Evaluating technological information and utilizing it – scientific knowledge, technological capability, and external linkages in biotechnology. Journal of Economic Behavior & Organization 24 (1), 91–114. Audretsch, D.B., Menkveld, A.J., Thurik, A.R., 1996. The decision between internal and external R&D. Journal of Institutional and Theoretical Economics 152 (3), 519–530. Barney, J., 1991. Firm resources and sustained competitive advantage. Journal of Management 17 (1), 99–120. Basile, R., 2001. Export behaviour of Italian manufacturing firms over the nineties: the role of innovation. Research Policy 30 (8), 1185–1201. Becker, W., Dietz, J., 2004. R&D cooperation and innovation activities of firms – evidence for the German manufacturing industry. Research Policy 33 (2), 209–223. Berchicci, L., 2011. Heterogeneity and intensity of R&D partnership in Italian manufacturing firms. IEEE Transactions on Engineering Management 58 (4), 674–687. Bettis, R.A., Bradley, S.P., Hamel, G., 1992. Outsourcing and industrial decline. Academy of Management Executive 6 (1), 7–22. Calantone, R.J., Stanko, M.A., 2007. Drivers of outsourced innovation: an exploratory study. Journal of Product Innovation Management 24 (3), 230–241. Cassiman, B., Veugelers, R., 2002. R&D cooperation and spillovers: some empirical evidence from Belgium. American Economic Review 92 (4), 1169–1184. Cassiman, B., Veugelers, R., 2006. In search of complementarity in innovation strategy: internal R&D and external knowledge acquisition. Management Science 52 (1), 68–82. Chandler, A.D., Hikino, T., 1990. Scale and scope: the dynamics of industrial capitalism. Belknap Press, Cambridge, Mass. Chesbrough, H., 2005. Open innovation: a new paradigm for understanding industrial innovation , DRUID Summer conference, Copenhagen, June 27–29. Chesbrough, H.W., 2003. Open Innovation: The New Imperative for Creating and Profiting from Technology. Harvard Business School Press, Boston, MA. Chesbrough, H.W., Teece, D.J., 1996. When is virtual virtuous? Organizing for innovation. Harvard Business Review 74 (1), 65–73. Chesbrough, H.W., Vanhaverbeke, W., West, J., 2006. Open Innovation: Researching a New Paradigm. Oxford University Press, Oxford; New York. Cohen, W.M., Levinthal, D.A., 1989. Innovation and learning: the two faces of R & D. The Economic Journal 99 (397), 569–596. Cohen, W.M., Levinthal, D.A., 1990. Absorptive-capacity – a new perspective on learning and innovation. Administrative Science Quarterly 35 (1), 128–152. Cohen, W.M., Levinthal, D.A., 1994. Fortune favors the prepared firm. Management Science 40 (2), 227–251.

L. Berchicci / Research Policy 42 (2013) 117–127 Cronbach, L.J., 1987. Statistical tests for moderator variables: flaws in analyses recently proposed. Psychological Bulletin 102 (3), 414–417. Deeds, D.L., Hill, C.W.L., 1996. Strategic alliances and the rate of new product development: an empirical study of entrepreneurial biotechnology firms. Journal of Business Venturing 11 (1), 41–55. Dyer, J.H., Singh, H., 1998. The relational view: cooperative strategy and sources of interorganizational competitive advantage. Academy of Management Review 23 (4), 660–679. Eisenhardt, K.M., Schoonhoven, C.B., 1996. Resource-based view of strategic alliance formation: strategic and social effects in entrepreneurial firms. Organization Science 7 (2), 136–150. Escribano, A., Fosfuri, A., Tribo‘, J.A., 2009. Managing external knowledge flows: the moderating role of absorptive capacity. Research Policy 38 (1), 96–105. Grant, R.M., 1996. Toward a knowledge-based theory of the firm. Strategic Management Journal 17, 109–122. Greene, W.H., 2003. Econometric Analysis, 5th ed. Prentice Hall, Upper Saddle River, NJ. Gujarati, D.N., 1995. Basic Econometrics, 3rd ed. McGraw-Hill, New York. Gulati, R., Singh, H., 1998. The architecture of cooperation: managing coordination costs and appropriation concerns in strategic alliances. Administrative Science Quarterly 43 (4), 781–814. Hagedoorn, J., 1993. Understanding the rationale of strategic technology partnering – interorganizational modes of cooperation and sectoral differences. Strategic Management Journal 14 (5), 371–385. Hagedoorn, J., 2002. Inter-firm R&D partnerships: an overview of major trends and patterns since 1960. Research Policy 31 (4), 477–492. Hatzichronoglou, T., 1997. Revision of the high-technology sector and product classification, STI Working Papers, 1997/2. OECD, Paris. Katz, M.L., 1986. An analysis of cooperative research and development. The RAND Journal of Economics 17 (4), 527–543. Kogut, B., Zander, U., 1992. Knowledge of the firm, combinative capabilities, and the replication of technology. Organization Science 3 (3), 383–397. Laursen, K., Salter, A., 2006. Open for innovation: the role of openness in explaining innovation performance among UK manufacturing firms. Strategic Management Journal 27 (2), 131–150. Lavie, D., 2006. The competitive advantage of interconnected firms: an extension of the resource-based view. Academy of Management Review 31 (3), 638–658. Linder, J.C., Jarvenpaa, S., Davenport, T.H., 2003. Toward an innovation sourcing strategy. MIT Sloan Management Review 44 (4), 43–49. Mitchell, W., Singh, K., 1996. Survival of businesses using collaborative relationships to commercialize complex goods. Strategic Management Journal 17 (3), 169–195.

127

Mowery, D.C., Oxley, J.E., Silverman, B.S., 1996. Strategic alliances and interfirm knowledge transfer. Strategic Management Journal 17, 77–91. Mowery, D.C., Rosenberg, N., 1989. Technology and the pursuit of economic growth. Cambridge University Press, Cambridge [England], New York. Nelson, R.R., 1959. The simple economics of basic scientific research. Journal of Political Economy June, 297–306. Penrose, E.T., 1959. The Theory of the Growth in the Firm. Basil Blackwell, Oxford [Engl.]. Powell, W.W., Koput, K.W., SmithDoerr, L., 1996. Interorganizational collaboration and the locus of innovation: networks of learning in biotechnology. Administrative Science Quarterly 41 (1), 116–145. Rigby, D., Zook, C., 2002. Open market innovation. Harvard Business Review 80 (10), 80–89. Roodman, D., 2006. How to do Xtabond2: An Introduction to Difference and System GMM in Stata. Center for Global Development Working Paper No. 103. Rosenberg, N., 1990. Why do firms do basic research with their own money. Research Policy 19 (2), 165–174. Rothaermel, F.T., Deeds, D.L., 2006. Alliance type, alliance experience and alliance management capability in high-technology ventures. Journal of Business Venturing 21 (4), 429–460. Schoonhoven, C.B., Eisenhardt, K.M., Lyman, K., 1990. Speeding products to market – waiting time to 1st product introduction in new firms. Administrative Science Quarterly 35 (1), 177–207. Shan, W.J., Walker, G., Kogut, B., 1994. Interfirm cooperation and startup innovation in the biotechnology industry. Strategic Management Journal 15 (5), 387–394. Steensma, H.K., Corley, K.G., 2000. On the performance of technology-sourcing partnerships: the interaction between partner interdependence and technology attributes. Academy of Management Journal 43 (6), 1045–1067. Stuart, T.E., 2000. Interorganizational alliances and the performance of firms: a study of growth and innovation rates in a high-technology industry. Strategic Management Journal 21 (8), 791–811. Tsai, K-H., 2009. Collaborative networks and product innovation performance: toward a contingency perspective. Research Policy 38 (5), 765–778. Veugelers, R., 1997. Internal R&D expenditures and external technology sourcing. Research Policy 26 (3), 303–315. Whittington, R., 1990. The changing structures of R&D: from centralization to fragmentation. In: Loveridge, R., Pitt, M. (Eds.), The Strategic Management of Technological Innovation. John Wiley, Chichester, West Sussex, England. Womack, J., Jones, D.T., Roos, D., 1990. The Machine that Changed the World. Free Press, New York, NY. Zaheer, A., McEvily, B., Perrone, V., 1998. Does trust matter? Exploring the effects of interorganizational and interpersonal trust on performance. Organization Science 9 (2), 141–159.