Master Thesis
The Decline of Basic Research in Firm R&D
Investigating the Effect of External Knowledge Sourcing on the Phenomenon
Sebastian Hautkappe University of Groningen Faculty of Economics and Business
S3574997
Supervisor: Dr. Philip Steinberg Co-assessor: Dr. Florian Noseleit
Word count (without Appendix): 12,085
Abstract
Recent research has revealed that firms decreasingly engage in internal basic research. However, there has been no systematic quantitative analysis into potential causes thus far. This thesis examines one of the most frequently named potential reasons behind this
phenomenon: the increasing availability of external basic research knowledge as a substitute for internal activities. While there is an abundance of literature examining the relationship between internal and external R&D strategies, scholars have rarely specifically looked into the case of basic research, which could yield conceivably different findings than generalized studies. Using panel data with 11,084 observations in the years from 2009 to 2015, the relationship between external sourcing and internal basic research was analyzed. The empirical findings show that the frequently emphasized complementarity between internal and external sources also holds in the case of basic research. It is therefore unlikely that firms replace their internal basic research activities through externally sourced knowledge, casting doubts on the assertion that external knowledge sourcing acts as an antecedent to the decline of firm-internal basic research.
Introduction
In the 20th century, firm laboratories were the breeding ground for many
breakthrough innovations, such as the transistor, the laser and the first computer (Arora, Belenzon & Patacconi, 2018; Pisano, 2010). Many of these developments are unimaginable without basic research performed by firms. According to the Frascati manual, basic research1 is thereby defined as “experimental or theoretical work undertaken primarily to acquire new knowledge of the underlying foundations of phenomena and observable facts, without any particular application or use in view” (OECD, 2002, p. 77). Traditionally, basic research is regarded as one of the three building blocks within the linear model of corporate R&D, next to the more commercially oriented applied research and experimental development
(Branscomb & Auerswald, 2002; Cohen, Nelson & Walsh, 2002; Czarnitzki & Thorwarth, 2012).
Various scholars have stressed the importance of basic research for firm performance, value creation and productivity (e.g. Arora et al., 2018; Czarnitzki & Thorwarth, 2012; Griliches, 1986; Zahra, Kaul & Bolívar-Ramos, 2018).
But despite evidence for its substantial importance, several authors have recently reported that firms are increasingly unwilling to perform basic research internally (Arora et al., 2018; Pisano, 2010; George, 2015; Matthews, 2015; The Economist, 2007; Tijssen, 2004). Arora et al. (2018) subsume that “…firms still value the golden egg of science […], but seem to be increasingly unwilling to invest in the golden goose itself…” (p. 3). Several potential reasons for this phenomenon (such as managerial short-sightedness or increasing complexity of basic research) have been suggested but are yet to be tested, as literature on this topic is only emerging.
However, one of the most salient reasons is that firms may instead turn to external knowledge sourcing, since other scholars have simultaneously found a substantial rise for alternative and potentially substitutive (external) sources of knowledge, coinciding with the decline. Numerous scholars have found that external sourcing mechanisms and open
innovation practices are sharply increasing (Arora et al. 2014; Cassiman & Veugelers, 2006; Chesbrough, 2003; Pisano, 2010; Tijssen, 2004). Especially university collaboration warrants more attention, since universities and research institutes have always been the primary source of basic research knowledge (e.g. Andries & Thorwarth, 2014; Branscomb & Auerswald, 2002; Teece, 2010). Yet, external knowledge sourcing has rarely been analyzed in the context of basic research, presenting an interesting research gap. Similarly, Zahra et al. (2018) note that it “would […] be useful to examine the effect of changing trends in [...] open innovation sources, and the rising use of alliances with universities and other firms […]” (p. 168).
This Master thesis attempts to bridge this gap and to link these two interrelated concepts, contributing to an improved understanding of the phenomenon. In order to do so, the focus of this thesis lies on the relationship between external and internal knowledge sourcing in the special context of basic research. Hence, the following research question will be posed:
“What are the effects of external knowledge sourcing expenditures on the investments in internal basic research?”
The results show, that there is a significant positive relationship between internal basic research and external sourcing, consistent with complementarity.
These findings contribute to research within the Knowledge-based View (KBV), which views knowledge “as the most strategically important of the firm's resources” (Grant, 1996, p. 110) and expand previous findings of the complementarity between internal and external knowledge sourcing to the special case of basic research. Since a substitutive relationship needs to be assumed for external knowledge sourcing to be an antecedent to the decline of basic research, the results do not support the view that external sourcing can be considered responsible for the decline, thereby precluding one of the major potential reasons.
The remainder of this thesis is structured as follows: The subsequent section will review relevant literature and presents its key insights. It will serve as the theoretical
grounding for the following development of hypotheses. After that, the research methodology will be explained and data, measurements and analysis methods will be introduced. Next, the result section will present the results of the statistical analyses. Lastly, these results will be connected to theory, hypotheses and the research question in the discussion and conclusion section. This section will also encompass theoretical and managerial implications as well as limitations and opportunities for future research.
Literature Review
This literature review will set the frame for the subsequent development of the hypotheses. Firstly, key concepts will be defined and relevant terms will be distinguished. Secondly, the role of basic research is described according to previous research findings, which entails a contextualization and evaluation of the benefits and challenges for firms that engage in internal basic research. Thirdly, literature describing the phenomenon, i.e. the declining engagement by firms in basic research, is introduced. Fourthly, academic ideas about potential reasons for the decline are retraced. This is narrowed down to the focus of this thesis, which is also one of the most frequently mentioned potential explanations: the
increasing trend towards external knowledge sourcing. Lastly, the concept of external knowledge sourcing is described and relevant findings regarding the relationship to internal knowledge sources are detailed.
Definition of Concepts
theoretical work undertaken primarily to acquire new knowledge of the underlying
foundations of phenomena and observable facts, without any particular application or use in view” (OECD, 2002, p. 77). Basic research “analyses properties, structures and relationships with a view to formulating and testing hypotheses, theories or laws” and its outcomes are “not generally sold but are usually published in scientific journals” (OECD, 2002, p. 77f). Yet, the Frascati manual makes an even finer distinction: between oriented basic research and pure basic research. This is differentiation is made to take the fact into account that although basic research is per definition not immediately application-directed, it can nevertheless be actively steered towards fields with a high potential of future applications (OECD, 2002, p. 77f). While pure basic research is strictly not application-oriented and serves solely the advancement of knowledge, oriented basic research is “carried out with the expectation that it will produce a broad base of knowledge likely to form the basis of the solution to recognised or expected, current or future problems or possibilities” (OECD, 2002, p. 77f).
Now that the term “basic research” is somewhat refined, it is furthermore necessary to define the terms “basic science” and “corporate science” as well as to establish whether there are meaningful differences between the concepts, since all are used in the relevant literature. Interestingly, the terms “basic science” and “corporate science” are not defined by the Frascati manual, which are used by several authors such as Cohen et al. (2002), Zahra et al. (2018) and Simeth & Cincera (2016).
To fully comprehend the semantic differences, it is also useful to consult the Oxford Dictionary for the terms of “research” and “science”. Research is thereby defined as “The systematic investigation into and study of materials and sources in order to establish facts and reach new conclusions” (Oxford Dictionary, 2018a). The definition of science is: “The intellectual and practical activity encompassing the systematic study of the structure and behaviour of the physical and natural world through observation and experiment” (Oxford Dictionary, 2018b).
However, it is evident that differences are only minor: If a firm is involved in basic science, it automatically is simultaneously involved in basic research, since a “systematic investigation” in the sense of science cannot be called “systematic” if appropriate “materials and sources” are not studied (research).
do admit that “…corporate science focuses primarily on basic research…” (p. 157) and use the terms “basic research” seven times and the term “basic science” ten times in the following without further definitions and elaborations.
Due to the ambiguous usage of terms and the semantic analogy between them, it can therefore be inferred that findings about intra-firm (basic) science are equally valid in the context of basic research. Basic research shall therefore be used synonymously for basic science and corporate science findings henceforth. However, the concepts basic research and basic science should nevertheless be distinguished from their applied counterparts.
The other term that needs to be clarified is “external knowledge sourcing”. In this case a distinction between external knowledge sourcing and external knowledge acquisition needs to be made. This distinction is advocated by Dahlander & Gann (2010) who view acquisition as “pecuniary inbound innovation” and contrast it with sourcing, which they define as “non-pecuniary inbound innovation”. According to Dahlander & Gann (2010), acquisition entails “acquiring inventions and input to the innovative process […]” whereas sourcing encompasses the “sourcing [of] external ideas and knowledge from suppliers, customers, competitors, […]” (p. 706). However, according to the Oxford Dictionary, “acquisition” has not necessarily a pecuniary connotation: it is either defined as “an asset or object bought or obtained…” or as “the learning or developing of a skill, habit, or quality” (2018c). Furthermore, other authors have not made this distinction: E.g., Kang & Kang (2009) analyse three ways of external knowledge sourcing with one of which being
technology acquisition. Similarly, Arora, Cohen & Walsh (2016) use “source” and “sourcing” in a pecuniary context. Carayannopoulos & Auster (2010) also use “external knowledge sourcing” for both alliances and acquistions. “Sourcing” is in all these instances used as an umbrella term, with “acquisition” being a part of it. This view seems also appropriate for the frame of this thesis, since both mechanisms are providing access to external knowledge in one way or the other.
Following these authors, the term “external knowledge sourcing” should henceforth be understood as an umbrella term for accessing external knowledge, encompassing both its pecuniary and non-pecuniary dimension.
The Role of Basic Research
knowledge source which has been proven to increase firm performance (Andries & Thorwarth, 2014; Arora et al., 2018; Czarnitzki & Thorwarth, 2012; Griliches, 1986; Mansfield, 1991; Zahra et al., 2018). Basic research can also be seen as a necessity to safeguard a firm’s ability to compete in the future marketplace: firms “undertake basic research, with a view to preparing for the next generation of technology” (OECD, 2002, p. 77f). Relatedly, basic research has therefore been described as “fuel that powers innovation” (Czarnitzki & Thorwarth, 2012, p. 1556). This “fuel” also accelerates new product
development: Mansfield (1991) finds that without basic research, approx. 10% of new products and processes could not have been developed as fast as they were.
However, despite these repeated emphases on its importance, engaging in in-house basic research is not exclusively advantageous but rather entails serious detriments, too. Findings about both advantages and disadvantages are portrayed in the following to contextualize findings about the decline of firm-internal basic research.
Advantages associated with Basic Research
Disadvantages associated with Basic Research
However, despite these numerous benefits, authors have also recognized the difficulties and challenges of basic research. Perhaps the most striking argument against performing in-house basic research is that it is hard to appropriate and that discoveries are difficult to monetize due to a high risk of unintended spillovers and imperfections of knowledge markets (Cohen & Levinthal, 1989, 1990; Czarnitzki & Thorwarth, 2012; Matthews, 2015; Pisano, 2010; Rosenberg, 1990; Simeth & Cincera, 2016; Teece, 2010; Tijssen, 2004; Trajtenberg, Henderson & Jaffe, 1992). Matthews (2015) concisely sums this up: “Nobel Prizes won by corporate research labs may have earned their parent companies prestige, but they don’t always make money. Sometimes it’s competitors that benefit—think Steve Jobs’ appropriation of Xerox inventions like computer desktop icons”. Related to the weak appropriability is another major drawback of basic research: the high Knightian uncertainty regarding strategic benefits or monetary gains (Czarnitzki & Thorwarth, 2012; Rosenberg, 1990; Tijssen, 2004). Scholars point out that the translation of basic research outcomes into useful technologies is intricate, with immediate outputs barely possessing intrinsic economic value (Chen et al., 2016; Czarnitzki & Thorwarth, 2012; Simeth &
Cincera, 2016). Furthermore, Gittelman & Kogut (2003) find that science and innovation are distinct from each other, with each governed by a different logic. Internal (basic) research may also lead to the Not-invented-here syndrome, negatively biasing the firm against external knowledge (Arora et al., 2014; Katz & Allen, 1982). Additionally, commercializing basic research takes a substantial amount of time despite efforts to accelerate development times (Czarnitzki & Thorwarth, 2012; George, 2015). But not only time and risks are a problems – engaging in basic research is also costly due to the ever-increasing complexity that co-arises with scientific advances (Branscomb & Auerswald, 2002; Tijssen, 2004).
The Decline of Basic Research
by corporate researchers has declined by 52% from 1980 to 2006, while patenting and R&D intensity have been increasing. They also find (through stock market value regressions) that the value attributed to scientific research has dropped significantly, while the value of technological knowledge has remained stable (Arora et al., 2018). Similarly, Tijssen (2004) has found that despite overall increases in patenting and patent citations to scientific
literature, (co-)authorship has declined between 1996 and 2001.
The consequences of this decline of firm-internal basic research are often portrayed as negative and quite severe: For example, Matthews (2014) sees the decline as the “Death of American Research and Development”. George (2015) writes that the United States “…will continue to lose well-paying jobs, further hollowing out the middle class, and creating even greater income inequality”. Likewise, Arora et al. (2018) are concerned about the future of the US innovation system: “The fact that one of its key components — the large corporate lab — is in decline can be seen as a reason for concern” (p. 28).
Reasons for the Decline
To comprehend the decline, researchers have posited several hypotheses regarding potential reasons – however, they are by and large untested.
Several arguments can be derived from the difficulties associated with basic research as detailed above, with appropriability issues named frequently (e.g. Andries & Thorwarth, 2014; Matthews, 2015; Teece, 2010).
In addition, various authors blame the decline of basic research on short-termism of shareholders and managers, since basic research needs a long time horizon to yield returns (George, 2015; Matthews, 2015; Pisano, 2010; Tijssen, 2004). Arora et al. (2018) provide a comprehensive theoretical analysis of potential reasons: Comparing the shares of basic and applied research with development, they conclude that firms are moving away from research towards more development. They state that the decline does neither stems from changes in the publication practices nor from rising costs of basic research. Instead, they find that the value of basic research has declined for firms, which they attribute to a policy shift. This is in line with Tijssen (2004), who also identifies a policy shift of corporations towards
appropriation and commercialization.
Branscomb & Auerswald (2002) conjecture that the decline of firm-internal basic research may be inevitable due to the rising complexity: “Today, even the large corporations with the largest R&D budgets have difficulty putting together all the elements required for in-house development and commercialization of science-based technologies“ (p. 10).
Andries & Thorwarth (2014) blame the press: “It has even been advocated – particularly in the popular press – that in-house basic research is no longer economically warranted […]” (p. 303).
Zahra et al. (2018) offer yet another perspective: Instead of focusing exclusively on the firm’s role in the decrease of internal basic research, they adopt a more global process perspective. They conclude that the actual problem is rather that the commercialization of basic research fails, instead of an internal or external lack thereof.
External Knowledge Sourcing as a Potential Driver
Despite these different and at times diverging explanation attempts, there seems to be far-reaching consensus amongst scholars that firms are increasingly using external knowledge sourcing and open innovation practices as a likely substitute for internal basic research
(Andries & Thorwarth, 2014; Arora et al. 2014; Arora et al., 2018; Cassiman & Veugelers, 2006; Chesbrough, 2003; Pisano, 2010; Tijssen, 2004; Zahra et al., 2018). Arora et al. (2018) concisely sum up this notion: “… large firms are moving to a bare minimum of internal research expenditures that allow them to tap into externally generated knowledge. […] Long-term investments in science may be less desirable with the growing availability of external inventions…” (p. 27f.).
How does this mechanism work in the case of basic research? Knowledge and especially basic research are partially public goods, in that intellectual property rights (IPR) mechanisms are often seriously deficient in preventing unintended knowledge spillovers (Chen et al., 2016; Teece, 2010). While involuntary inside-out spillovers may pose risks for companies, firms may actually benefit from outside-in knowledge sourcing. Recently,
Both, open innovation as a new paradigm and external knowledge sourcing as one of its components have been increasingly embraced by firms (Andries & Thorwarth, 2014; The Economist, 2007; Zahra et al., 2018). The extent of this increase is considerably large: E.g., the formation of inter-firm R&D partnerships has increased from around 10 per year in the 1960s to approx. 500 by the end of the 1990s (Hagedoorn, 2002). Similarly, Vega-Jurado et al. (2009) cite that firms’ external R&D expenditures have doubled compared to total R&D expenditures over a 10-year period. Arora et al. (2016) report that as much as 49% of innovating firms have relied on external sources for their most important new product.
This rapid adoption of external sourcing practices can be attributed to its advantages: As Noseleit & de Faria (2013) point out, firms sometimes even have no choice but to rely on external knowledge sources, as internal R&D efforts may be insufficient: “In many
innovation projects firms search for external knowledge sources since they cannot rely solely on their own R&D efforts” (p. 2000). Several scholars have furthermore highlighted the benefits of utilizing external knowledge sourcing mechanisms (Brunswicker &
Vanhaverbeke, 2015; Kang & Kang, 2009; Lakhani, Jeppesen, Lohse, & Panetta, 2006). For example, Faems, van Looy & Debackere (2005) have found a positive relationship between inter-organizational collaboration and firm performance. In the context of basic research, external knowledge sourcing may help to overcome some (but not all) of the challenges associated with basic research: For instance, collaborations with URI’s and other firms enable firms to share the costs and risks of basic research and achieve synergistic benefits. By
accessing external sources, firms have the chance to overcome the restrictions of their own knowledge base, resource constraints and the liability of smallness (Brunswicker &
Vanhaverbeke, 2015; Chesbrough, 2003). However, the specific benefits of external knowledge sourcing are contingent on the particular method that is used (Kang & Kang, 2009).
2007; Thursby & Thursby, 2006; Zahra et al., 2018). This is especially promising due to the fact that universities have always been the primary locus of basic research (Trajtenberg et al., 1992; Rafferty, 2007) and the fact that URI’s engage increasingly in commercialization (Cohen et al., 2002; Fabrizio, 2006; Rafferty, 2007). Fabrizio (2006) illustrates the high relevance of universities for the private sector: “Many industries owe their technological foundation to … research performed in university labs” (p. 2). An official report by the OECD (2015) has found that firms “draw upon scientific results and resources from the publicly funded science base through contract R&D projects, collaborations and partnerships that facilitate acquisition or licensing strategies, or even by attracting highly qualified
personnel” (p. 4). For small and medium-sized firms, Brunswicker & Vanhaverbeke (2015) assert that URI’s are an important source for inventive and pre-industrial knowledge. Faems et al. (2005) also stress that especially the explorative collaboration with URI’s is beneficial for the creation of new products.
Relationship between Internal and External Knowledge Sourcing
Due to the trend towards external knowledge sourcing, a large amount of literature has investigated the relationship between internal and external sources. However, in deciding whether the two mechanisms can be seen as substitutes or complements, scholars have come to contradictory results. These divergent results can be largely explained by the disparate theoretical lenses utilized by scholars (Vega-Jurado et al., 2009).
Transaction cost economists view external knowledge sourcing as substitutes where firms either follow a make or buy-strategy depending on the outcome of considerations regarding costs and risks (Vega-Jurado et al., 2009). Laursen & Salter (2006) find an inverted U-shape relationship between external knowledge sourcing and innovation performance, negatively moderated by internal research which points to a substitutive relationship. Cohen et al. (2002) find that “...public research may partly substitute for a firm's own R&D" (p. 14). Arora et al. (2014) also note a negative relationship between internal research and external knowledge acquisition, hinting at substitutive effects. Scholars often refer to the NIH-syndrome that may occur if firms use both mechanisms simultaneously. In this light, Grigoriou & Rothaermel (2017) find that an internal knowledge stock may diminish the benefits of external knowledge sources.
labor between universities and firms regarding basic research (Branscomb & Auerswald, 2002; Arora et al., 2018; Teece, 2010; Zahra et al., 2018).
In contrast, scholars of the resource based view rather consider the two mechanisms as complements, especially stressing the importance of internal efforts for absorptive capacity (Vega-Jurado et al., 2009). Numerous studies have adopted this view and have come to the result that internal and external knowledge sourcing are complements (Arora et al., 2001; Caloghirou, Kastelli & Tsakanikas, 2004; Cassiman & Veugelers, 2006; Grimpe & Kaiser, 2010).
As mentioned earlier, despite the abundance of research into the interaction of internal and external sources, the special case of basic research has rarely been addressed.
Interestingly, these findings rather point to complementarity between the two mechanisms (Cassiman & Veugelers, 2002; Fabrizio, 2006) which would run contrary to the assumption that the decline of in-house basic research can be attributed to the rise of external knowledge sourcing.
Hypotheses Development
Based on the literature outlined above, I presuppose that there is both a decline of internal basic research as well as an increase of external knowledge sourcing, and that these two knowledge sourcing mechanisms are related to each other (e.g. Arora et al., 2018; Andries & Thorwarth, 2014; Zahra et al., 2018).
Recall, that the aim of this thesis is to test the alleged notion that external knowledge sourcing is able to replace internal basic research, therefore contributing to the decline. By doing this, it is necessary to hypothesize a substitutive (and therefore negative) relationship. As outlined above, this is not unreasonable, as various scholars have previously supported this view. Hence, I pose the following hypothesis:
being the primary source of basic research, it is very likely that collaborations between URI’s and firms are set up with the goal of sourcing basic research knowledge (Branscomb & Auerswald, 2002; Trajtenberg et al., 1992; Rafferty, 2007; Tijssen, 2004). In contrast, firm collaborations are more likely to be related to applied research and development. This results in the following effect: A higher proportion of URI collaborations should lead to less internal basic research2.
H2: ”The type of external knowledge sourcing mechanism moderates the relationship between external knowledge sourcing and internal basic research. URI collaborations are
thereby expected to intensify the underlying relationship (H1).” Regarding firm size, I argue the following: Increasing firm size intensifies the negative relationship between external knowledge sourcing and internal basic research. This argument is made due to the fact that the literature investigating the decline of basic research has focused almost exclusively on large corporations (Arora et al., 2018; George, 2015; Matthews, 2015; The Economist, 2007). This seems logical, as larger firms may have been the only ones in the position to stem the costs and risks associated with internal basic research in the first place. It may therefore very well be that a decline is only noticeable in large firms. Cohen et al. (2002) conclude that large firms and start-ups benefit more from external
knowledge (in the form of public research) than smaller firms. To sum up, it can be inferred that larger firms have (a) been previously the only ones conducting basic research internally and (b) stand to benefit more from a switch to the absorption of externally available basic research knowledge. Following Andries & Thorwarth (2014) who include “firm size as a moderator for the relationship between in-house or outsourced basic research” (p. 304), the same is replicated here. The underlying relationship may therefore only hold if firm size is large, leading to a hypothesized moderating effect3:
H3: ”Firm size moderates the relationship between external knowledge sourcing and internal basic research, in that the underlying relationship (H1) is significantly stronger for
larger firms.”
2 Since Laursen & Salter (2004) find that firms who use open innovation practices are more likely to draw from universities, an interaction effect shall also be investigated.
Methodology
As previously stated, empirical and quantitative research regarding the decline of basic research within firms is still in its infancy and was thus far only indirectly measured. More importantly, assumptions about causes for the phenomenon have not been adequately tested. This thesis therefore utilizes a multivariate statistical analysis in order to test one of the most promising potential antecedents of the decline: the increasing use of external knowledge sourcing mechanisms. Panel data is used to capture the longitudinal effects of external knowledge sourcing on internal basic research.
The following section will describe the data used in this study as well as the
composition of the sample. Furthermore, the measurements of the variables will be detailed. An overview about the analysis methods will conclude this section.
Dataset and Sample
For this thesis, an already existing database is used: Data is sourced from the
Stifterverband für die Deutsche Wissenschaft e.V. (SV), which biannually conducts extensive R&D surveys via its affiliated SV Wissenschaftsstatistik GmbH and on behalf of the German Federal Ministry of Education and Research. The survey covers multiple dimensions and provides detailed information on important R&D figures such as firm employees, R&D staff, firm revenue and internal/external R&D expenditures. Most importantly, it fits the research intent of this thesis, since it measures the share of R&D expenditures spent on basic research, applied research and experimental product/service development as well as data on R&D expenditures on several types of collaboration.
This database is regarded as a reputable source for research since it follows
international standardized rules, is part of the official reporting of German R&D data to the EU and the OECD and has been previously used in academia (e.g. Schmid et al., 2014; Steinberg et al., 2017). For the research within this thesis, anonymized firm-level data is used. The final sample consists of 11,084 observations of 5,393 firms in the survey years 2009, 2011, 2013, 2015.
Measurements
selected based on the best fit between the literature and measurements provided by the data source. The independent and dependent variables are both continuous variables and are measured in absolute R&D expenditures, as expenditures are one of the officially recognized R&D input measurements (OECD, 2002, p. 21). Furthermore, the use of these measurements ensures comparability and reliability: “For the purposes of international comparison, the breakdown should be based on current expenditures only” (OECD, 2002, p. 77).
Dependent Variable
Unlike previous literature (e.g. Tijssen, 2004) which has used the somewhat indirect measure of corporate (co-)authorship of scientific publications as a measurement of Basic research, I take advantage of the fact that the SV survey measures the percentage of internal R&D expenditures spent on basic research more directly (item 81). With the aid of item 31 measuring internal R&D expenditures, these percentages can be converted into absolute amounts.
Independent Variable
To operationalize external knowledge sourcing, I am using the External R&D expenditure as the independent variable, according to item 40 of the SV survey. External R&D expenditure is thereby defined as the cost of R&D services provided outside of companies such as the award of research contracts to other companies, universities, government research institutions (e.g. the Max Planck Society or the
Fraunhofer-Gesellschaft) or institutions for joint research. Although this measurement may be somewhat noisy, it has been used before (Vega-Jurado, 2009). The literature differs here in that it typically uses Likert scales (Brunswicker & Vanhaverbeke, 2015; Kang & Kang, 2009).
Moderating Variables
Regarding the type of knowledge sourcing mechanism, I distinguish between URI and firm collaboration. URI collaboration consists of firm’s expenditures in contract R&D carried out by “state research institutes” and “university institutes and professors” and divided by the total external R&D expenditures, as measured by the items 34, 35 and 40 of the SV survey.
of employees, there is ambiguity since other institutions such as the EU still use turnover as a determinant of firm size. Theoretical considerations also justify a measurement in terms of revenue: Since basic research entails high costs, firms with larger revenues may be better positioned to stem these expenditures. However, the SV data provide both these
measurements – therefore it has been decided to examine both measurements for the
hypothesized moderating effect, with the better fitting one used in the regression models. In this case turnover was chosen. Item 15 of the SV survey thereby provides the absolute
number of domestic employees at the end of the business year, whereas item 16 provides data about turnover.
Control Variables
Due to the fact that there are several other factors that may also affect the dependent variable, I included several control variables.
Drawing on previous findings, I assume that a firm’s R&D intensity has an effect on the relationship specified in the conceptual model (see Appendix C). Cohen et al. (2002) find that public research is only critical to a small number of industries. These differences can potentially be explained by heterogeneous levels of research intensity across industries. For instance, Czarnitzki & Thorwarth (2012) find quantitative proof that conducting in-house basic research only yields a large premium on a firm’s output for firms in high-tech sectors, while there is no such premium for companies in low-tech industries. Laursen & Salter (2004) observe that a higher level of a firm’s R&D intensity also increases the likelihood for sourcing external (university) knowledge (similarly: Tijssen, 2004). The literature therefore assumes that a rising R&D intensity increases both internal basic research efforts and the likelihood of engaging in external collaboration, pointing to complementarity of the
have a higher R&D capacity and vice versa. For R&D intensity, turnover is measured in item 16, while data about the R&D expenditure is captured by item 41 of the SV survey. For R&D capacity, the number of R&D employees is given by item 194 whereas the total number of employees is measured by item 15. Again, since both measurements seem appropriate and are available, both have been examined in the analysis and the regression model utilizes the better fitting one – in this case R&D capacity.
Industry dummies on the basis of two-digit NACE Rev.2 industry codes as well as Year dummies are also incorporated in line with previous research (e.g. Steinberg et al., 2017).
Analytical Methods
As outlined above, a continuous dependent variable is used in this thesis. Following the decision tree for multivariate analyses proposed by Tabachnick & Fidell (2014), a sequential (also known as hierarchical) multiple regression will be applied, as it is beneficial for testing hypothesis while the researcher stays in control of advancements.
Due to the fact that developments over time play an important role in the conceptual model of this thesis and due to the availability of cross-sectional time-series data, I am using a panel data. This entails that unobserved heterogenerity terms need to be controlled for and implies that pooled OLS estimators may be both biased and inconsistent (Schmidheiny, 2018). To avoid these issues, there are typically two different strategies that can be used: Either through introducing fixed- or random-effects estimators (Mundlak, 1978). Since the choice of either strategy depends on whether there is covariance between unobserved time-stable factors and the independent variable, a Hausman test is applied to determine the more appropriate final models.
Furthermore, a time-lag was introduced to somewhat minimize the incidence of reversed causality4. This strategy can be helpful, as one prerequisite for causality is that for x being a cause of y, x must precede y (Cohen, Cohen, West & Aiken, 2003).
In addition, potential multicollinearity was detected by checking variance inflation factors (VIFs). All variables were standardized. Firm size, External R&D expenditures and
Basic research were additionally transformed by taking their natural logarithm to counter skewness and to ensure normality.
Six models are considered. The first model includes both the control and moderator variables only. The next model 2 adds the independent variable. Model 3, model 4 and model 5 each add one interaction effect at a time. The final model 6 includes the independent
variable as well as all moderators, controls and interaction terms.
Results
In the following, the statistical results are presented. As a first step, descriptive statistics and correlations will be displayed and commented. On that basis, inferential the main regression models will be detailed.
Descriptive Statistics and Correlations
Table 1 (p. 20) displays the descriptive statistics and correlations between the
variables. As noted, there were 11,084 observations. Unfortunately, significance levels were not retrieved at the data center.
The correlations in the table are by and large unremarkable, as they are mostly weak to moderate (r < 0.6), which does not raise concerns about multicollinearity. However, there is one peculiarity: External R&D expenditures appears to be highly correlated to Basic Research (r > 0.6), which is not surprising given the hypothesized relationship.
To further alleviate doubts about multicollinearity, Variance Inflation Factors (VIFs) were reported in Table 2 (p. 20). As evident, all VIFs are below the recommended maximum threshold of 10. However, while most VIFs do not come close to this boundary, the VIF of external R&D expenditure in Model 6 with a value of 9.97 is a notable exception. This should not be of concern: the external R&D expenditure of Model 6 will not be theoretically
Table 1: Correlations
Variable Mean Standard Deviation (1) (2) (3) (4) (5)
(1) Basic research .1705288 .4552411 1.0000
(2) Firm size 3.098985 1.964045 0.5127 1.0000
(3) R&D capacity .2158336 .2411365 0.0278 -0.4733 1.0000
(4) URI collaborations 3863.612 16149.33 -0.0133 0.0100 -0.0177 1.0000
(5) Ext. R&D expenditures .1444055 .4787758 0.6470 0.4434 0.0425 -0.0484 1.0000
Note: N = 11,084 observations. Significance levels were not inquired at the data center. Table 2: Variance Inflation Factors
Variable Model 1 Model 2 Model 3 Model 4 Model 5 Model 6
VIF 1/VIF VIF 1/VIF VIF 1/VIF VIF 1/VIF VIF 1/VIF VIF 1/VIF
Firm size 1.47 0.682 1.98 0.505 2.02 0.496 1.98 0.504 2.09 0.479 2.20 0.455
R&D capacity 1.64 0.609 1.75 0.572 1.75 0.572 1.29 0.775 1.77 0.565 1.78 0.560
URI collaborations 1.29 0.777 1.29 0.775 3.25 0.308 1.75 0.572 1.29 0.774 3.26 0.307
Ext. R&D expenditures - - 1.58 0.635 2.60 0.385 1.64 0.611 6.28 0.159 9.97 0.100
Ext. R&D expenditures X URI collaborations - - - - 3.78 0.264 - - - - 3.80 0.263
Ext. R&D expenditures X R&D capacity - - - 1.06 0.941 - - 1.44 0.696
Ext. R&D expenditures X Firm size - - - 4.97 0.201 6.73 0.149
Mean VIF 1.23 1.26 1.43 1.26 1.52 1.80
Table 3: Panel data regression models with time lag, random effects
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6
Control Variables
R&D capacity 0.920*** 0.107 0.115 0.107 0.363*** 0.479***
(0.173) (0.149) (0.146) (0.151) (0.115) (0.114)
Industry Dummies Yes Yes Yes Yes Yes Yes
Year Dummies Yes Yes Yes Yes Yes Yes
Moderator Variables Firm size 1.902*** 0.105 0.137 0.105 0.658*** 0.961*** (0.321) (0.251) (0.234) (0.255) (0.179) (0.169) URI collaborations -0.017 0.115*** -0.116 0.115** 0.055* -0.313 (0.032) (0.044) (0.201) (0.045) (0.033) (0.256) Independent Variable
Ext. R&D expenditures 3.148*** 2.981*** 3.150*** -0.491 -2.309***
(0.764) (0.682) (0.783) (0.704) (0.867)
Interaction Effects
Ext. R&D expenditures X URI collaborations -0.829 -1.223
(0.725) (0.883)
Ext. R&D expenditures X R&D capacity -0.007 1.255***
(0.426) (0.467)
Ext. R&D expenditures X Firm size 1.549*** 2.099***
(0.511) (0.570) R² within 0.0006 0.0030 0.0036 0.0029 0.0024 0.0066 R² between 0.1082 0.3303 0.3299 0.3304 0.4432 0.4649 R² total 0.0856 0.2173 0.2178 0.2173 0.2762 0.2978 Observations 11,084 11,084 11,084 11,084 11,084 11,084 Number of Firms 5,393 5,393 5,393 5,393 5,393 5,393
Regression Results
Table 3 (p. 21) presents the result of the panel data regression using random effects estimators. Before arriving at these models, various other models have also been considered (see Appendix A) and Hausman tests were used. All Hausman tests have yielded a p-value below 0.05, which would imply to reject the null hypothesis and therefore to use fixed effects models. However, it has appeared that the within-R² is too low (and the panel too
unbalanced) for the Hausman test to be conclusive. E.g. Hahn, Ham & Moon (2011) caution against applying the Hausman test: “…if the within variation is small, the fixed effects estimates may not be asymptotically normal, which may invalidate the basic premise of the Hausman test” (p. 1). This may cause the Hausman test to over-reject the null hypothesis. Therefore, random effects models were preferred instead, since the unobserved effect is assumed to be uncorrelated with all explanatory variables (Wooldridge, 2013).
By looking at the total and between R²-values, it is evident that the explanatory power successively increases from model 1 to model 6, especially between model 1 and model 2. This means that the predictive power is considerably better once the independent variable is included.
In H1, I have hypothesized that external R&D expenditures will negatively impact expenditures on basic research. In order to check this hypothesis, model 2 needs to be consulted. As evident, it displays a highly significant positive value. Since models 3-6 all incorporate interaction terms, the main effect is likely to be heavily distorted, making
interpretation attempts without in-depth analyses very risky (cf. discussion on: ResearchGate, 2016, 2018). On the basis of model 2, H1 shall therefore be rejected.
H2 has assumed that URI collaboration intensifies the underlying effect (H1) and thus should bear a negative sign. However, as seen in model 2, URI collaboration is positive and highly significant, albeit to a relatively small magnitude. Model 3 and model 6 fail to find a significant influence for its interaction (Ext. R&D expenditure X URI collaboration) on basic research. Therefore, H2 needs to be rejected.
Analogous to H2, H3 has stated that firm size intensifies the underlying effect (H1) and should consequently be negative. While firm size has no direct effect on basic research (model 2), its interaction (Ext. R&D expenditure X Firm size) has a relatively large positive and highly significant impact on basic research (models 5, 6). H3 also need to be rejected.
Discussion & Conclusion
This section discusses and contextualizes the empirical results. First, the results will be interpreted and connected to the hypotheses and the research question, putting them into the theoretical context. Second, some theoretical and managerial implications will be derived. And third, some of the most important research limitations will be highlighted which
simultaneously serve as avenues for future research.
Interpretation of the Results
Interestingly and as described above, the regression models have yielded very different results than expected by the conceptual model – which could be due to mainly two reasons: theoretical considerations and methodological limitations.
In terms of a theoretical interpretation, recall that from the outset there were two divergent theoretical perspectives about the relationship between external knowledge sourcing and internal basic research. While both have their merits, it has been chosen to assume a substitutive relationship – that way, it was possible to test the notion made by several scholars that the former mechanism is superseding the latter, thus contributing to a large part to the decline of in-house basic research. It is therefore interesting but not very surprising that all hypotheses have been rejected. H1 was not just simply rejected, it was also shown that the opposite should be assumed: External sourcing mechanisms appear to
complement internal basic research efforts. This finding is in line with earlier studies that emphasize complementarity (e.g. Cassiman & Veugelers, 2006; Fabrizio, 2009).
However, a word of caution is in order: The findings in this thesis are purely
descriptive findings, i.e. firms tend to simultaneously de- and increase internal and external research expenditures. Normative statements cannot be derived. In other words: Firms merely appear to be treating both sources as complements – whether they are right to do so, remains another question. A recommendation to view both mechanisms as complements cannot be given as the appropriateness of such an assumption and its consequences on firm
performance were not analyzed. Therefore, the findings in this thesis do neither support previous literature in saying that firms benefit from employing both mechanisms simultaneously as complements, nor does it disprove findings highlighting substitutive characteristics of the two sources.
expenditures simultaneously entail higher levels of internal basic research expenditures and vice versa. The generalized relationship is the following: A rise in one leads to the rise in the other, a decline in one leads to a decline in the other. By presupposing this relationship, it can be inferred that the increasing external R&D expenditures cannot be blamed for a potential reduction in in-house basic research spending.
Now that the main effect has been interpreted, what about the influence of controls and moderators? As for URI collaboration, the results show that it only plays a marginal role. This may point to the fact, that collaboration between URI’s and firms is still a rare
occurrence (Cohen et al., 2002). Firm size has a positive interaction effect with external knowledge sourcing. This seems logical, since scholars have found that larger firms simply collaborate more often (Cohen et al., 2002; Laursen & Salter, 2004; Veugelers & Cassiman, 2005). Lastly, R&D capacity exhibits a positive interaction effect: This is also not surprising, as a moderating role has already been established (Berchicci, 2013).
Further Considerations
By investigating one of the major potential causes for the decline of firms’ in-house basic research, the oft-asserted presence and importance of this phenomenon has been taken for granted. However, as shown in the following, these previous findings (on which this thesis is based) may not be as conclusive as they appear to be.
Regarding the decline, there are some indications that would counter this assertion. Although not primarily investigated in this thesis, a very rudimentary look at the mean of firm’s R&D expenditures into basic research as a percentage of total R&D expenditures shows no sign of a decline (see Appendix B). George (2015) claims that the withdrawal of firms’ engagement in basic research is geographically restricted to the US, with no such decline noticeable in Germany and China. This could be a reason why no such decline was found in German firms. Even if this claim holds true, this underlines that the decrease of in-house basic research is at most a phenomenon with globally rather limited implications.
As noted earlier, most findings underpinning the phenomenon were based on
associated with high-impact innovations” (p. 366). A policy shift of firms regarding publications may thus not be unthinkable.
Another argument that the importance of a potential decline is overstated can be derived from the following logic: Basic research has always played a marginal role in firms’ R&D. Therefore, a decline may be by and large inconsequential. The former statement is very appositely argued by Teece (2010):
“Basic research usually ends up in scientific publications, so it is hard – if not
impossible – to secure strong intellectual property protection for scientific knowledge. As a result, it is very difficult to charge for discoveries, even if they have the potential to generate high value for society, so very few firms invest in basic research. Spill-overs (externalities) are simply too large; profiting from discovery is simply too difficult. There is no easy for-profit business model for capturing value from scientific discoveries in a world where science wants to be open and rapid dissemination of scientific knowledge through journals, conferences and professional contacts is almost inevitable: not surprisingly, most basic research is not funded by business firms, but by governments” (Teece, 2010, p. 185).
That the decline may be unsubstantial can be derived from another piece of evidence: arguments about a decrease in intra-firm basic research have been made repeatedly in the past (Griliches, 1986; Tijssen, 2004). E.g., Griliches has already warned in 1986 that “the absolute decline in basic research in industry which occurred in the 1970's may turn out to have been very costly to the economy in terms of foregone growth opportunities” (p. 153). Yet, despite all previous warnings there is no substantive evidence of an economic slow-down or a stall of R&D development caused by too little firm-internal basic research. Between 1986 (after Griliches assertions) and 2018, the GDP in the US has more than quadrupled from 4,590.1 billion USD to 19,390.6 billion USD (nominal; IMF, 2018). Major macroeconomic
consequences and important forgone growth opportunities as predicted by Griliches (1986) are therefore hard to imagine. Arora et al. (2018) provide no compelling reasons why it should be different this time.
industries. This illustrates exemplarily that a contingency perspective should be taken5. The second question is, whether the locus of basic research should be firm-internal. Here, two divergent opinions are noticeable in the literature: On the one hand, several authors
emphasize that basic research should be performed in-house. These authors have claimed that it is very important that firms perform basic research activities in-house as opposed to
(solely) relying on external sources (e.g. Arora et al., 2018; Matthews, 2015; George, 2015). E.g. Cockburn & Henderson (1998) stress the relevance of in-house basic research for absorptive capacity6. On the other hand are authors, for whom the precise locus of basic research knowledge seems to be of secondary importance. These authors consider that the lack of basic research within firm R&D may not be exclusively related to the fact that firms do not partake in it: Rather, these authors take a more global view on the issue and find that the commercialization of basic research knowledge is failing as opposed to (solely) the internal creation of basic research knowledge (e.g. Zahra et al., 2018). This research stream stresses the difficulties around the translation of existing insights, especially by including URI’s and government policies in the consideration. Scholars have thereby termed these challenges as the “valley of death” (Auerswald & Branscomb, 2003, Branscomb & Auerswald, 2002; Markham, Ward, Aiman-Smith & Kingon, 2010). E.g. Brunswicker & Vanhaverbeke (2015) list some of the challenges in the commercialization of basic research findings in university-industry relationships, for instance “…cultural differences, long-term scientific research versus exploitation-oriented research of industrial organizations and incompatible rewards systems—with universities focusing on publishing and firms ‘protecting’ results”.
After these further considerations, a more comprehensive conclusion can be drawn: This thesis has found some quantitative proof that rising external R&D expenditures cannot be considered as an antecedent to a potential decline in firm-internal basic research
expenditures, since the two are treated as complements by firms. Furthermore, it was shown that contingency factors such as firm size and R&D capacity affect the relationship. Yet, it was also shown that the underlying assumptions of this thesis (that a decline is present and
5 Analogous to this thesis, possible contingencies could also be firm size, R&D capacity or R&D intensity.
6 Here, it is noteworthy that Cockburn & Henderson (1998) and subsequent researchers did not simultaneously investigate the appropriateness of applied research for absorptive capacity. Thus, it may well be that firms can built sufficient absorptive capacity for basic research knowledge by
that the decline is important) are far from undisputed. An alternative literature stream was evinced that could steer research in a more fruitful direction, should other potential drivers similarly be refuted.
Theoretical and Managerial Implications
By refuting one of the major assumed reasons for a firm’s decreasing engagement in basic research, theory can now focus on the other suspected drivers of a (potential) firm-internal decline. How future research may proceed is detailed in the subsequent section. Furthermore, this thesis adds another view to the discussion whether internal and external knowledge sources can be seen as substitutes or complements and specifically advances it in the special context of basic research. The view of Cassiman & Veugelers (2006) in that both mechanisms can be seen as complementary, is confirmed. As noted earlier though, the findings of this thesis are purely descriptive. The significance of several covariates in the regression models of this thesis also reminds scholars of the necessity to consider a contingency perspective while analyzing this research field.
Due to the descriptive nature of this thesis, managerial implications are rather limited. However, the doubts around the actual existence and the negative effects of the phenomenon should be noted by practitioners. Decision-makers in firms should be aware of the benefits and detriments of engaging in internal basic research to make informed choices. Managers are moreover well advised to critically scrutinize claims that the decline is present and “alarming” (Matthews, 2015) or that entire countries will “…continue to lose well-paying jobs, further hollowing out the middle class, and creating even greater income inequality“.
Research Limitations and Future Research
Several factors may impede the reliability and validity of the results. Some of the most important limitations are outlined in the following.
The fact that the time horizon of the panel data set is relatively short and that the used panel is highly unbalanced is an important limitation of this thesis. As described earlier, determining the most appropriate model was challenging, since the Hausman test was not conclusive due to the low within variance in the regression models. Advanced econometrical tests could not be used post-factum due to restricted access to the data base. While random effects models where chosen, this choice may not be completely incontestable (cf.
to be rejected. The difference is, that by using fixed effects models, complementarity could not have been established and the influence of most moderators and controls is insignificant (see Appendix A).
Even though a time lag was introduced to address reversed causality issues to some degree, it is impossible to rule it out entirely. It should be noted that “…in nonexperimental research, it is very difficult to attribute causality to an IV” (Tabachnick & Fidell, 2014, p. 35). Therefore it may be that the relationship proposed in the conceptual model may work in the opposite direction. Similarly to Laursen & Salter’s (2014) arguments, this issue is of secondary importance here: The key insight that internal and external knowledge sourcing act as complements as opposed to substitutes remains unaffected by the direction of causality.
In the domain of measurements, authors have previously noted that surveys are problematic in that the definition and differentiation between concepts such as “basic research” is difficult and confusing for respondents (e.g. Branscomb & Auerswald, 2002). This may be a source of unintended response bias.
Also, as evident by the maximum total R² in the random effects models (0.2978) a large part of the variance still remains unexplained. This could be due to fact that key influencing factors were omitted. Relatedly, another important control variable is missing in the conceptual framework: firm age is frequently mentioned by research as an important control variable (e.g. Arora et al., 2016; Berchicci, 2013; Steinberg et al., 2017).
Unfortunately, firm age was not available in the used data set and both confidentiality issues and the restricted scope of this thesis have made the use of an alternative source for this missing piece of information and a subsequent merge with the present data set unfeasible.
research should be firm-internal. In that regard, it would also be interesting to examine absorptive capacity from a different perspective: How can firms build an absorptive capacity for basic research? Is it sufficient to leverage competencies in relevant applied research fields? Or is it necessary to perform basic research internally in order to utilize external sources (Cockburn & Henderson, 1998; Cohen & Levinthal, 1990; Zahra & George, 2002)?
References
Andries, P., & Thorwarth, S. (2014). Should Firms Outsource their Basic Research? The Impact of Firm Size on In‐House versus Outsourced R & D Productivity. Creativity and Innovation
Management, 23(3), 303-317.
Arora, A., Fosfuri, A., & Gambardella, A. (2001). Markets for technology and their implications for corporate strategy. Industrial and corporate change, 10(2), 419-451.
Arora, A., Belenzon, S., & Rios, L. A. (2014). Make, buy, organize: The interplay between research, external knowledge, and firm structure. Strategic Management Journal, 35(3), 317-337.
Arora, A., Cohen, W. M., & Walsh, J. P. (2016). The acquisition and commercialization of invention in American manufacturing: Incidence and impact. Research Policy, 45(6), 1113-1128.
Arora, A., Belenzon, S., & Patacconi, A. (2018). The decline of science in corporate R&D. Strategic
Management Journal, 39(1), 3-32.
Auerswald, P. E., & Branscomb, L. M. (2003). Valleys of death and Darwinian seas: Financing the invention to innovation transition in the United States. The Journal of Technology Transfer, 28(3-4), 227-239.
Berchicci, L. (2013). Towards an open R&D system: Internal R&D investment, external knowledge acquisition and innovative performance. Research Policy, 42(1), 117-127.
Branscomb, L. M., & Auerswald, P. E. (2002). Between Invention and Innovation. Retrieved from https://www.nist.gov/sites/default/files/documents/2017/05/09/gcr02-841.pdf
Brunswicker, S., & Vanhaverbeke, W. (2015). Open innovation in small and medium‐sized enterprises (SMEs): External knowledge sourcing strategies and internal organizational facilitators. Journal of Small Business Management, 53(4), 1241-1263.
Caloghirou, Y., Kastelli, I., & Tsakanikas, A. (2004). Internal capabilities and external knowledge sources: complements or substitutes for innovative performance?. Technovation, 24(1), 29-39. Carayannopoulos, S., & Auster, E. R. (2010). External knowledge sourcing in biotechnology through acquisition versus alliance: A KBV approach. Research Policy, 39(2), 254-267.
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.
Chen, J. R., Kan, K., & Tung, I. H. (2016). Scientific linkages and firm productivity: Panel data evidence from Taiwanese electronics firms. Research Policy, 45(7), 1449-1459.
Chesbrough, H. (2003). Open innovation: The new imperative for creating and profiting from
technology. Boston, MA: Harvard Business Press.
Chesbrough, H. (2006). Open innovation: A new paradigm for understanding industrial innovation. In H. Chesbrough, W. Vanhaverbeke, & J. West (Eds.), Open Innovation: Researching a New Paradigm. 0–19. Oxford: OUP Oxford.
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.
Cohen, W. M., Nelson, R. R., & Walsh, J. P. (2002). Links and impacts: the influence of public research on industrial R&D. Management science, 48(1), 1-23.
Cohen, J., Cohen, P., West, S. G., & Aiken, L. S. (2003). Applied multiple regression/correlation
analysis for the behavioral sciences. Mahwah: Lawrence Erlbaum Associates.
Czarnitzki, D., & Thorwarth, S. (2012). Productivity effects of basic research in low-tech and high-tech industries. Research Policy, 41(9), 1555-1564.
Dahlander, L., & Gann, D. M. (2010). How open is innovation?. Research policy, 39(6), 699-709. Enkel, E., Gassmann, O., & Chesbrough, H. (2009). Open R&D and open innovation: exploring the phenomenon. R&d Management, 39(4), 311-316.
Fabrizio, K. (2006). The use of university research in firm innovation. In H. Chesbrough, W.
Vanhaverbeke, & J. West (Eds.), Open Innovation: Researching a New Paradigm. 134-160. Oxford: OUP Oxford.
Faems, D., Van Looy, B., & Debackere, K. (2005). Interorganizational collaboration and innovation: Toward a portfolio approach. Journal of product innovation management, 22(3), 238-250.
George, B. (2015, December 12). Dow-DuPont Raises Even More Concerns America Is Abandoning Corporate Research. Retrieved from http://fortune.com/2015/12/12/dow-dupont-corporate-research-america/
Gittelman, M., & Kogut, B. (2003). Does good science lead to valuable knowledge? Biotechnology firms and the evolutionary logic of citation patterns. Management Science, 49(4), 366-382.
Grant, R. M. (1996). Toward a knowledge‐based theory of the firm. Strategic management journal,
17(S2), 109-122.
Grigoriou, K., & Rothaermel, F. T. (2017). Organizing for knowledge generation: Internal knowledge networks and the contingent effect of external knowledge sourcing. Strategic Management
Journal, 38(2), 395-414.
Griliches, Z. (1986). Productivity, research-and-development, and basic research at the firm level in the 1970s. American Economic Review, 76(1), 141-154.
Grimpe, C., & Kaiser, U. (2010). Balancing internal and external knowledge acquisition: the gains and pains from R&D outsourcing. Journal of management studies, 47(8), 1483-1509.
Hagedoorn, J. (2002). Inter-firm R&D partnerships: an overview of major trends and patterns since 1960. Research policy, 31(4), 477-492.
IMF (International Monetary Fund). (2018, April). World Economic Outlook Database, April 2018 [Dataset]. Retrieved January 19, 2019, from
https://www.imf.org/external/pubs/ft/weo/2018/01/weodata/weorept.aspx?pr.x=98&pr.y=12&sy=198 0&ey=2023&scsm=1&ssd=1&sort=country&ds=.&br=1&c=111&s=NGDP_RPCH,NGDPD,NGDP DPC,PCPIPCH,LUR,GGXONLB_NGDP,GGXWDG_NGDP,BCA_NGDPD&grp=0&a=
Katz, R., & Allen, T. J. (1982). Investigating the Not Invented Here (NIH) syndrome: A look at the performance, tenure, and communication patterns of 50 R & D Project Groups. R&d Management,
12(1), 7-20.
Kang, K. H., & Kang, J. (2009). How do firms source external knowledge for innovation? Analysing effects of different knowledge sourcing methods. International Journal of Innovation Management,
13(01), 1-17.
Lakhani, K. R., Jeppesen, L. B., Lohse, P. A., & Panetta, J. A. (2007). The value of openess in
scientific problem solving (pp. 07-50). Boston, MA: Division of Research, Harvard Business School.
Laursen, K., & Salter, A. (2004). Searching high and low: what types of firms use universities as a source of innovation?. Research policy, 33(8), 1201-1215.
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. Laursen, K., & Salter, A. J. (2014). The paradox of openness: Appropriability, external search and collaboration. Research Policy, 43(5), 867-878.
Mansfield, E. (1991). Academic research and industrial innovation. Research policy, 20(1), 1-12. Markham, S. K., Ward, S. J., Aiman‐Smith, L., & Kingon, A. I. (2010). The valley of death as context for role theory in product innovation. Journal of Product Innovation Management, 27(3), 402-417.
Matthews, C. (2015, December 21). The Death of American Research and Development. Retrieved from http://fortune.com/2015/12/21/death-american-research-and-development/
Mundlak, Y. (1978). On the pooling of time series and cross section data. Econometrica: journal of
the Econometric Society, 69-85.
Noseleit, F., & de Faria, P. (2013). Complementarities of internal R&D and alliances with different partner types. Journal of Business Research, 66(10), 2000-2006.
OECD (Organisation for Economic Co-operation and Development). (2002). Frascati manual 2002:
Proposed standard practice for surveys on research and experimental development. OECD.
OECD (Organisation for Economic Co-operation and Development). (2015). HIGHLIGHTS OF THE
OECD SCIENCE, TECHNOLOGY AND INDUSTRY SCOREBOARD 2015. Retrieved January 19,
2019 from http://www.oecd.org/sti/Science-brief-scoreboard.pdf
Oxford Dictionary. (2018a). research | Definition of research in English by Oxford Dictionaries. Retrieved December 20, 2018, from https://en.oxforddictionaries.com/definition/research Oxford Dictionary. (2018b). science | Definition of science in English by Oxford Dictionaries. Retrieved December 20, 2018, from https://en.oxforddictionaries.com/definition/science
Pisano, G. P. (2010). The evolution of science-based business: innovating how we innovate.
Industrial and corporate change, 19(2), 465-482.
Rafferty, M. (2008). The Bayh–Dole Act and university research and development. Research Policy,
37(1), 29-40.
ResearchGate. (2016). Regression with interaction: Shall we look at the main effects? [Forum post]. Retrieved January 19, 2019, from
https://www.researchgate.net/post/Regression_with_interaction_Shall_we_look_at_the_main_effects ResearchGate. (2018). When the moderator is not significant, but the interaction is significant? [Forum post]. Retrieved January 19, 2019, from
https://www.researchgate.net/post/When_the_moderator_is_not_significant_but_the_interaction_is_si gnificant
Rosenberg, N. (1990). Why do firms do basic research (with their own money)?. Research Policy,
19(2), 165-174.
Schmid, T., Achleitner, A. K., Ampenberger, M., & Kaserer, C. (2014). Family firms and R&D behavior–New evidence from a large-scale survey. Research Policy, 43(1), 233-244.
Schmidheiny, K. (2018). Panel Data: Fixed and Random Effects. Retrieved January 19, 2019, from https://www.schmidheiny.name/teaching/panel2up.pdf
Simeth, M., & Cincera, M. (2016). Corporate science, innovation, and firm value. Management
Science, 62(7), 1970-1981.
Steinberg, P. J., Procher, V. D., & Urbig, D. (2017). Too much or too little of R&D offshoring: the impact of captive offshoring and contract offshoring on innovation performance. Research Policy,
46(10), 1810-1823.
Tabachnick, B. G., & Fidell, L. S. (2014). Using multivariate statistics. Harlow: Pearson.
Teece, D. J. (1998). Capturing value from knowledge assets: The new economy, markets for know-how, and intangible assets. California management review, 40(3), 55-79.
Teece, D. J. (2010). Business models, business strategy and innovation. Long range planning, 43(2-3), 172-194.
The Economist. (2007, March 01). Out of the dusty labs: The rise and fall of corporate R&D. Retrieved from https://www.economist.com/briefing/2007/03/01/out-of-the-dusty-labs
Thursby, J., & Thursby, M. (2006). Where Is the New Science in Corporate R&d?. Science,
314(5805), 1547-1548.
Tijssen, R. J. (2004). Is the commercialisation of scientific research affecting the production of public knowledge?: Global trends in the output of corporate research articles. Research Policy, 33(5), 709-733.
Trajtenberg, M., Henderson, R., & Jaffe, A. (1992). Ivory tower versus corporate lab: An empirical
study of basic research and appropriability (No. w4146). National Bureau of Economic Research.
Vega-Jurado, J., Gutiérrez-Gracia, A., & Fernández-de-Lucio, I. (2009). Does external knowledge sourcing matter for innovation? Evidence from the Spanish manufacturing industry. Industrial and
corporate change, 18(4), 637-670.
Veugelers, R., & Cassiman, B. (2005). R&D cooperation between firms and universities. Some empirical evidence from Belgian manufacturing. International Journal of Industrial Organization,
23(5-6), 355-379.
