FACULTAD DE CIENCIAS ECONÓMICAS Y EMPRESARIALES
Doctorado de Economía y Gestión de la Innovación RD 99/2011
Doctoral Dissertation
THE DETERMINANTS OF THE ECONOMIC IMPACT OF R&D AND INNOVATION COLLABORATION BETWEEN
FIRMS AND KNOWLEDGE PROVIDERS
Author
CARLOS VIVAS AUGIER Supervisors
ANDRES BARGE-GIL
Universidad Complutense de Madrid
JOSÉ GUIMÓN
Universidad Autónoma de Madrid
Madrid 2019
ACKNOWLEDMENTS AND DEDICATORY
To all of those who never stop pushing me to finish my thesis (you know who you are), despite sometimes it annoyed me. Several times I hesitated to continue and you told me to keep going because you believed I could do it. I am not sure if I would have done it without your constant encouragement.
Thank you Andres (Barge) for your patience and support all these years. Your contributions, learnings and guidance are invaluable. Thank you for never stop encouraging me despite the long pauses I had to take due to my job. For always finding the time to help me when I had the time to work on my thesis during vacations and weekends. For taking time off your personal time and the time with your family to review and send me your comments to my latest work. I could not have asked for a better supervisor.
Thank you José (Guimón) for always reminding me of all I needed to do and never let a deadline pass me by. For all your help to deal with requirements to continue with my research. For your advice and your contributions. For always being available and taking the time to help me with anything, even administrative paperwork.
TABLE OF CONTENT
CHAPTER I – INTRODUCTION ... 1
A. Motivation and Justification ... 1
CHAPTER II – IMPACT ASSESSMENT LITERATURE REVIEW ... 5
A. Introduction ... 5
B. Methodology ... 6
B.1. Identification of Studies ... 6
B.2. Selection of Studies ... 7
B.3. Quality Assessment ... 8
B.4. Data extraction and Data Repository ... 8
B.5. Overview of Articles Resulting from the Literature Review ... 9
C. What is known about the impact on firms of collaboration? ... 10
C.1. What kinds of firms collaborate the most with knowledge providers? ... 10
C.2. Do firms benefit from collaborating with knowledge providers? ... 13
C.3. Which firms benefits the most from collaboration? ... 17
D. Conclusions and Discussions ... 19
CHAPTER III – DETERMINANTS OF THE ECONOMIC IMPACT ... 22
A. Introduction ... 22
B. Literature Review ... 23
B.1. About the Determinants and the Impact of Collaboration ... 23
B.2. About the Partner Selection and the Impact of Collaboration ... 25
C. Methodology ... 27
D. Econometric Analysis ... 30
D.1. Description of the Dataset ... 30
D.2. Definition of Variables ... 31
E. Econometric Results ... 34
E.1. Average Effect of Cooperation with Universities and KIBS ... 34
E.2. Heterogeneous Effects According to Characteristics of Firms ... 38
F. Discussion, Conclusions and Implications ... 42
CHAPTER IV – DRIVERS FOR COMMERCIALIZING KNOWLEDGE ... 44
A. Introduction ... 44
B. About Technology Institutes ... 45
C. Methodology of the Study ... 46
D. Overview of Studied Cases ... 51
D.1. CEA (France) ... 51
D.2. SINTEF (Norway) ... 51
D.3. SP (Sweden) ... 52
D.4. TECNALIA (Spain) ... 52
D.5. TNO (The Netherlands) ... 52
D.6. VTT (Finland) ... 53
E. Analysis of Drivers for Success ... 54
E.1. Customer Factors ... 54
E.2. Commercial Staff Factor ... 57
E.3. Research Strategy Factor ... 58
E.4 S&T Policy Factor ... 59
E.5. Impact Factor ... 61
E.6. External Factors ... 63
F. Conclusions and Recommendations ... 66
CHAPTER V – CONCLUSIONS AND POLICY IMPLICATIONS ... 69
BIBLIOGRAPHIC REFERENCES ... 73
APPENDIXES ... 83
Appendix I – List of Articles Subcategories ... 83
Appendix II – List of Articles of the Literature Review ... 84
Appendix III – Instrumental Variable Assessment ... 90
Appendix IV – OECD Sectorial Classification ... 92
INDEX OF TABLES
TABLE 1–RESEARCH PLAN AND SUMMARY ... 4
TABLE 2–SYSTEMATIC REVIEW PROCEDURE ... 6
TABLE 3–KEYWORD AND SEARCH STRINGS ... 7
TABLE 4-SUMMARY OF T1STUDIES ... 12
TABLE 5-SUMMARY OF T2STUDIES ... 15
TABLE 6-SUMMARY OF T3STUDIES ... 18
TABLE 7–SUMMARY OF VARIABLES ... 33
TABLE 8–BASELINE AND KEY RESULTS ... 35
TABLE 9-ANALYSIS OF STABILITY OF EFFECTS USING OSTER’S METHOD ... 36
TABLE 10-TOBIT REGRESSION MODELS ... 37
TABLE 11-HETEROGENEOUS EFFECTS ACCORDING TO SIZE AND R&D INTENSITY OF THE FIRM ... 39
TABLE 12-DISTRIBUTION OF FIRMS BY GROUP ... 40
TABLE 13-HETEROGENEOUS EFFECTS FOR THE INTERACTION OF SIZE AND R&DINTENSITY ... 41
TABLE 14-HETEROGENEOUS EFFECTS OF COOPERATION WITH UNIVERSITIES AND KIBS ... 42
TABLE 15-GENERAL CHARACTERISTICS OF TIS &COUNTRY ... 53
INDEX OF FIGURES
FIGURE 1-DISTRIBUTION OF STUDIES ... 9
FIGURE 2-RANKING DISTRIBUTION ... 48
FIGURE 3-KNOWLEDGE SERVICES OF GREATEST INTEREST TO FIRMS ... 56
FIGURE 4-DETERMINANTS OF COLLABORATION WITH FIRMS ... 57
FIGURE 5-COMMERCIAL STAFF FACTOR (AVERAGE DISTRIBUTION) ... 58
FIGURE 6-RESEARCH STRATEGY FACTOR (AVERAGE DISTRIBUTION) ... 59
FIGURE 7-S&TPOLICY FACTORS (REGIONAL LEVEL) ... 60
FIGURE 8-S&TPOLICY FACTORS (NATIONAL LEVEL) ... 61
FIGURE 9-S&TPOLICY FACTORS (EUROPEAN LEVEL) ... 61
FIGURE 10-IMPACT FACTOR ... 63
FIGURE 11-ENVIRONMENTAL FACTORS (PUBLIC POLICIES) ... 65
FIGURE 12-ENVIRONMENTAL FACTORS (SECTORIAL COMPOSITION) ... 66
FIGURE 13-ENVIRONMENTAL FACTORS (COMPETITORS) ... 66
THE DETERMINANTS OF THE ECONOMIC IMPACT OF R&D AND INNOVATION COLLABORATION BETWEEN FIRMS AND KNOWLEDGE PROVIDERS
The role played by external sources of knowledge or knowledge providers like universities, public research organizations, technology institutes and Knowledge Intensive Business Services (KIBS) is key for supporting R&D and innovation activities of firms. As a result of key assets of knowledge providers like research infrastructure, human resources, accumulated knowledge and expertise, knowledge providers are capable of generating new knowledge to develop new applications for industry. Therefore, knowledge providers have a noteworthy role for ensuring the success of the innovation process and for increasing the impact of R&D and innovation on economic growth of society.
The literature points to the need for better mechanisms to assess the resulting impact from collaboration between firms and knowledge providers. Therefore, a proper framework for assessing the resulting impact of R&D and innovation collaboration between firms and knowledge providers is most needed. Yet, circumstances such as no single universally accepted measure of innovation exists. These circumstances contribute to the assessment of the resulting impact of collaboration and the efficiency of the linkages between firms and knowledge providers in benefit of innovation remaining a difficult task. The complexity and difficulty to accurately measure the impact of R&D and innovation remains a key challenge in need of a proper solution for the benefit of policy-makers, R&D managers and society.
This research focuses on increasing the available stock of knowledge with regards to assessing the resulting impact of collaboration between firms and knowledge providers by analyzing how the determinants of the impact of firms and the type of knowledge provider firms collaborate with influence the economic impact of R&D and innovation cooperation.
This aim can be split into two general objectives. On the one hand, to analyze if the resulting impact on firms from R&D and innovation cooperation vary depending on the type of knowledge provider they engage with. On the other hand, to determine if characteristics like size and R&D intensity of firms contribute to maximize the impact of R&D and innovation activities executed in collaboration with knowledge providers.
First, a systematic review of the empirical literature was carried out for: (i) developing stylized facts; (ii) evaluating the methodological choices made by researchers; and (iii) identifying gaps in previous research that should be addressed in future research. Regarding the first
ABSTRACT
question, results show that size, R&D activity, technological level of the industry and cost obstacles positively influence the use of knowledge providers. Regarding the second question, the stylized facts are that the impact on technical results, such as new products or process, and on investment is positive. A positive impact is also suggested on economic results although evidence is not that robust. Finally, stylized facts on the third question are still to be developed. Evidence from studies addressing it are not conclusive. Regarding the second objective, the analysis focused on existing biases caused by methodological issues.
The great majority of studies do not account for sample selection and endogeneity.
Second, a quantitative approach (econometric) was adopted to shed light on some of the unanswered questions regarding the economic impacts of firms’ collaboration with knowledge providers. In this study, the evaluation was structured in a casual framework, which required reflecting on the potential ‘unobservables’ that may bias the results.
Cooperation with universities and KIBS shows a positive effect on sales from new-to-market products. This effect is very large in magnitude. Although average effects of cooperation with universities or KIBS are very similar, important differences can be found for different types of firms. Universities achieve their highest effect on small firms without R&D intensity and in large firms with R&D intensity. KIBS achieve their highest impact in SMEs with high R&D intensity and in large firms with low R&D intensity.
Third, a qualitative approach (case study) was employed to explore the determinants of the impact of knowledge providers, in order to shed some light on the drivers and characteristics of knowledge providers contributing to maximize their impact on firms. From the point of view of the determinants of the impact of knowledge providers, the understanding of the needs of industry and the inclusion of the challenges of industry in their research strategy seem to play a very significant role. From the perspective of the determinants of collaboration of knowledge providers, the characteristics that seem to act as drivers for collaboration are proficiency and expertise working with industry, highly specialized abilities and top skills in project management to deliver results in time and budget.
The conclusions of this research have a key implication for policy-makers, R&D managers and academic researchers. These actors should acknowledge the fact that collaboration between firms of certain characteristics with specific types of knowledge providers increases the resulting impact of R&D and innovation. Moreover, these actors should agree that the lack of an evidenced-based framework for suggesting the best ‘matching’ combinations to maximize the impact of the R&D and innovation collaboration between firms and knowledge providers is a relevant asset that requires the attention of all these parties.
CHAPTER I – INTRODUCTION
A. Motivation and Justification
Nowadays, the importance and relevance of innovation and knowledge as drivers of wealth is most acknowledge. The fact that the literature addressing the impact of innovation has increased over the last years and keeps on growing and diversifying is a clear evidence of this. In a knowledge-based economy, knowledge is a key input for production (Foray, 2006), therefore its generation, diffusion and application is crucial for today’s modern societies development and wealth. In the process of production and consumption of knowledge several actors take part (Mokyr, 2002). On the one hand, science aims to increase the available stock of knowledge and its use for the benefit of society. On the other hand, industry seeks ways to employ the available stock of knowledge for production and commercialization purposes. Linkages between science and industry contribute to enhancing the impact of public investments in research and to increasing the knowledge- intensity of society (OECD, 2013; Arnold, 2004).
The available stock of knowledge allows for the generation of new knowledge and several valuable outputs like new ideas, new technologies and new inventions. When these novelties reach the market and generate economic value, they are called innovations (Fagerberg, 2006). For industry, innovation is a crucial and valuable asset given its strong contribution to competitiveness (Abramo et al., 2009). The industry is constantly seeking and using the available stock of knowledge to create and increase its competitive advantages for satisfying the market and ensuring its survival (Morcillo, 2007). When the industry’s stock of knowledge is incomplete, it turns to cooperation activities with other partners to acquire the missing skills and expertise to increase it and for taking advantage of new opportunities (Arnold, 2004; Geisler, 2001).
The role played by external sources of knowledge or knowledge providers like universities, public research organizations, technology institutes and Knowledge Intensive Business Services (KIBS) is key for supporting R&D and innovation activities of firms. As a result of key assets of knowledge providers like research infrastructure, human resources, accumulated knowledge and expertise, knowledge providers are capable of generating new knowledge to develop new applications for industry (Perkmann and Walsh, 2007; Bozeman, 2002; OECD, 2002; Kline and Rosenberg, 1986). Therefore, knowledge providers have a noteworthy role for ensuring the success of the innovation process (Arnold, 2004) and for increasing the impact of R&D and innovation on economic growth of society (Van Beer et al., 2008).
Chapter I - Introduction
The scientific literature states that firms cooperating with external sources of knowledge are likely to be the ones with a higher innovation propensity (Lööf and Broström, 2006).
However, before this study, there was no systematic review of the literature summarizing the state-of-the-art regarding the impact of the collaboration between firms and knowledge providers. The scientific literature addressing the assessment of the resulting impact of the collaboration between firms and knowledge providers is highly fragmented and its results are not based on stylized facts (Vivas-Augier and Barge-Gil, 2015). This situation might have lead to the misleading conclusion that the evidence and answers to key questions with regards to the resulting impact of the collaboration between firms and knowledge providers in R&D and innovation are already available.
The literature points to the need for better mechanisms to assess the resulting impact from collaboration between firms and knowledge providers (Beise and Stahl, 1999). Therefore, a proper framework for assessing the resulting impact of R&D and innovation collaboration between firms and knowledge providers is most needed (Arnold, 2004). Yet, circumstances such as no single universally accepted measure of innovation exists (Freel and Harrison, 2006). These circumstances contribute to the assessment of the resulting impact of collaboration and the efficiency of the linkages between firms and knowledge providers in benefit of innovation remaining a difficult task. The complexity and difficulty to accurately measure the impact of R&D and innovation remains a key challenge in need of a proper solution for the benefit of policy-makers, R&D managers and society (Aschhoff and Schmidt, 2008; Izushi, 2003; Bozeman, 2000; Shapira et al., 1996; Bozeman and Coker, 1992).
There are several factors that contribute to the complexity and difficulty of assessing the resulting impact of collaboration between firms and knowledge providers in R&D and innovation. Geisler (2001) states that the perception of an outcome can diverge based on the perception of value of the user. This implies that the measurement of the resulting impact is not unique. Izushi (2003) and Shapira et al., (1996) highlight the biases in measuring the resulting impact introduced by the time-lag between the execution of R&D and innovation activities and the resulting impact of such activities. Shapira et al., (1996) highlights the bias introduced by attribution. Attribution states that the intensity of the resulting impact of the collaboration between firms and knowledge providers in R&D and innovation could be a consequence of several factors, hence accurately determining the cause-effect between these factors and impact is not a simple endeavor. Amongst these factors, the characteristics (determinants of the impact) and the profile (partner selection) of firms and knowledge providers engaged in R&D and innovation collaboration influence the resulting
impact. This consideration implies that different ‘matching’ combinations between firms and knowledge providers influence the success and impact of collaboration. Hence, the great importance of an impact assessment framework based on robust evidence that considers them to determine the best ‘matching’ to achieve an impact out of the collaboration between firms and knowledge providers.
This research focuses on increasing the available stock of knowledge with regards to assessing the resulting impact of collaboration between firms and knowledge providers by analyzing how the determinants of the impact of firms and the type of knowledge provider firms collaborate with influence the economic impact of R&D and innovation cooperation.
This aim can be split into two general objectives. On the one hand, to analyze if the resulting impact on firms from R&D and innovation cooperation vary depending on the type of knowledge provider they engage with. On the other hand, to determine if characteristics like size and R&D intensity of firms contribute to maximize the impact of R&D and innovation activities executed in collaboration with knowledge providers.
To achieve such aims, this research is structured in four chapters. Following this introduction, Chapter II reviews the state-of-the-art with regards to assessing the resulting impact of R&D and innovation collaboration between firms and knowledge providers. This will provide insights and key facts about the existing body of knowledge on impact assessment, and unveil how the scientific community has handled the difficulties to properly assess such impact. Chapter III aims to determine how the economic impact from R&D and innovation collaboration between firms and knowledge providers is influenced by the determinants of the impact and the type of knowledge provider firms collaborate with. This evidence is of great relevance for R&D managers and policy-makers to articulate and foster innovation out of the collaboration between firms and knowledge providers. From a different perspective, Chapter IV explores the determinants of the impact of knowledge providers that help achieve the best results out of collaboration with firms. This evidence would help R&D managers and policy-makers to maximize the impact of R&D and innovation on industry from collaboration with knowledge providers. Finally, Chapter V summarizes the main policy implications and key conclusions of this research for incentivizing innovation out of the collaboration between firms and knowledge providers.
Chapter I - Introduction
Table 1 – Research Plan and Summary
General Objective Specific Objective Hypothesis Methodology Publication Strategy
To determine how the resulting economic impact from R&D and innovation collaboration between firms and knowledge providers is influenced by the characteristics of the firm (determinants of the impact) and the type of knowledge provider the firm collaborates with (partner selection).
O1 – To assess the state-of- the-art with regards to
assessing the impact from R&D and innovation collaboration between firms and knowledge providers.
H1 – The determinants of the impact and the influence of the type of knowledge provider over the resulting impact from the collaboration between firms and knowledge provider have receive little attention by the scientific community.
Systematic Literature Review to identity stylized facts about the assessment of the impact of collaboration between firms and knowledge providers.
Published in Journal of Economic Survey (Vivas-Augier and Barge-Gil, 2015).
O2 – To analyze how the type of knowledge provider firms collaborate with influences the economic impact of
collaboration between firms and knowledge providers.
H2 – The resulting impact on firms from collaborating with knowledge providers vary depending on the type of knowledge provider they engage with.
Econometric (Causal Effects) using Panel Data from PITEC (Data based on Community Innovation Survey (CIS) and R&D Survey of Spain) to analyze the influence of the determinants of the impact and partner selection on the resulting economic impact from collaboration between firms and knowledge providers.
Submitted to Research Policy (Barge-Gil and Vivas-Augier, 2018).
O3 – To explore the
determinants of the impact of firm to achieve the best results out of collaboration with knowledge providers.
H3 – Characteristics like size and R&D intensity of firms contribute to maximize the impact from the collaboration between firms and knowledge providers.
O4 – To explore the
determinants of the impact of knowledge providers to achieve the best results out of
collaboration with firms.
H4 – Knowledge providers that best align their scientific knowledge and assets with the needs of industry contribute best to R&D and innovation activities of firms.
Case study for exploring the drivers and determinants of the impact of the most successful knowledge providers
collaborating with firms.
Published in Innovation:
Organization &
Management Journal (Vivas-Augier, 2016).
CHAPTER II – IMPACT ASSESSMENT LITERATURE REVIEW
A. Introduction
Firms’ direct links with knowledge providers has grown remarkably in the last decades (Amara and Landry, 2005) fostering the interest of academics and policy-makers in this area (Perkmann and Walsh, 2007). The reasons for the growth of such links are related to the evolution of the role of knowledge in society and economy (Carlaw et al., 2006), which is now crucial to firms’ competitiveness (Miotti and Sachwald, 2003). Furthermore, a number of factors make external knowledge highly appealing to firms (Dahlander and Gann, 2010;
Arora et al., 2001): (i) the higher complexity, interdisciplinary and speed of innovation processes raises costs and risks, (ii) the development of markets for knowledge, allowing for division of labour in innovation processes and (iii) the new information and communication technologies easing coordination between organizations. On the side of government and policy-makers, several initiatives have been launched for fostering links between firms and knowledge providers (Martin, 1996; Geroski, 1992). As a consequence, these linkages are being evaluated more systematically to improve political instruments promoting collaboration (Jaffe, 2008; Mowery, 1999). Yet this growing literature is highly fragmented (Lichtenthaler, 2005) and no systematic review has been carried out on their findings and methodological choices.
Although the term ‘knowledge providers’ may include a wide variety of organizations, for this specific work the attention is a subset of organizations interacting with firms based on the provision of knowledge-intensive services. Within these subset of organizations, universities, public research organizations, technology institutes and KIBS (including consultants) can be grouped in. This choice is supported by the results of some empirical works showing that patterns and motives of interactions with these organizations are different from those of interactions with other organizations, such as suppliers, customers or competitors (Narula, 2001; Tether, 2002; Miotti and Sachwald, 2003; Laursen and Salter, 2004). From an academic point of view, theoretical models have been mainly concerned with R&D interactions with competitors and, to a lesser extent, with supplier or competitors. That is, like in other areas of innovation studies (Hong et al., 2012), research on R&D interactions with universities, public research organizations, technology institutes and KIBS have been mainly empirically driven. Hence, it is crucial to review this empirical literature to investigate what it is known on this topic. This work analyses how the scientific community has approached the empirical evaluation of direct linkages between firms and knowledge providers with the triple objective of (i) developing stylized facts, (ii) evaluating the
Chapter II – Impact Assessment Literature Review
methodological choices made by researchers and (iii) highlighting avenues for future research.
The method followed is the systematic literature review procedure (Tranfield et al., 2003), employing specific criteria for inclusion and exclusion of articles in and from the review. A total of 100 articles were finally included (see Appendix II – List of Articles of the Literature Review). These articles’ key information was stored in a data repository specifically designed for recording their characteristics. The articles were classified into three groups according to the research question they addressed: (i) What are the determinants of the use of knowledge providers? (ii) Do knowledge providers have an impact on firms’ results (and how much impact do they generate)? and (iii) What are the determinants of impact? This strategy allowed for developing stylized facts about the benefits achieved by firms using knowledge providers, analysing the approach followed by researchers to deal with this topic and identifying future areas of research.
B. Methodology
The procedure used for systematic literature review was the one described by Tranfield et al., (2003), which is composed of three main stages (see Table 2).
Table 2 – Systematic Review Procedure
Stage I – Planning Stage II – Executing Stage III - Reporting - - Identification of the review
inquiry
- - Preparation of the review plan (protocol)
- - Identification of studies - - Selection of studies - - Quality assessment - - Data extraction and Data
Repository
- - Elaboration of the systematic review report (Synthesis) - - Recommendations
B.1. Identification of Studies
Keywords were selected considering (i) its relevance for finding articles addressing the utilization of knowledge providers and (ii) its precision to avoid the inclusion of non-relevant publications. The chosen keywords were grouped into four categories (see Table 3). The first category was used for grouping keywords referring to impact assessment (C1 – Impact).
The second category collected terminology for firms (C2 – Industry). The third group included
terms to describe a “utilization” condition (C3 – Relationship). The fourth and final group collected keywords addressing the linking activity (C4 – Activity)1.
Like in other studies (Di Stefano et al., 2012) the ISI Web of Knowledge (WoK) was chosen for this research. The search string returned a total of 37,920 publications. The list of publications was then narrowed to articles under the social science category. The total number lessened to 17,450 results. Then, these results were refined by subareas. A total of 32 subareas were included to refine the final search string (see Appendix I – List of Articles Subcategories). The final search returned a total of 15,767 publications.
Table 3 – Keyword and Search Strings
Category Keywords
C1 – IMPACT impact* OR assess* OR evaluat*
C2 – INDUSTRY Firm* OR Enterprise* OR "Private Sector" OR Industr* OR SME* OR Compan*
C3 – RELATIONSHIP Link* OR Relation* OR Cooperat* OR Collaborat* OR External OR Partner* OR Alliance C4 – ACTIVITY Innovat* OR R&D OR research OR transfer* OR support OR consultan*
SEARCH STRING 1 May 25th 2014 37,920 Results
Topic=(impact* OR assess* OR evaluat*) AND Topic=(Innovat* OR R&D OR research OR transfer* or support or consultan*) AND Topic=(Firm* OR Enterprise* OR "Private Sector" OR Industr* OR SME* OR Compan*) AND Topic=(Link* OR Relation* OR Cooperat* OR Collaborat* OR External OR Partner* OR Alliance)
SEARCH STRING 2 May 25th 2014 17,450 Results
Topic=(impact* OR assess* OR evaluat*) AND Topic=(Innovat* OR R&D OR research OR transfer* or support or consultan*) AND Topic=(Firm* OR Enterprise* OR "Private Sector" OR Industr* OR SME* OR Compan*) AND Topic=(Link* OR Relation* OR Cooperat* OR Collaborat* OR External OR Partner* OR Alliance)
Refined by: General Categories=(Social Science) SEARCH STRING 3
May 25th 2014 15,787 Results
Topic=(impact* OR assess* OR evaluat*) AND Topic=(Innovat* OR R&D OR research OR transfer* or support or consultan*) AND Topic=(Firm* OR Enterprise* OR "Private Sector" OR Industr* OR SME* OR Compan*) AND Topic=(Link* OR Relation* OR Cooperat* OR Collaborat* OR External OR Partner* OR Alliance)
Refined by: General Categories=(Social Science AND 32 Sub Areas)
B.2. Selection of Studies
Once all results had been collected, they were imported into citation management software – EndNoteTM X2. Afterwards, the 15,787 articles’ titles and abstracts were reviewed to select those relevant to this study. Eligibility criteria had to be as objective as possible. Hence, the
1 Due to the multiplicity of names and terminology given to some knowledge providers, the inclusion of the typology of knowledge providers, as part of the search string, was considered inappropriate.
Chapter II – Impact Assessment Literature Review
inclusion criteria were designed to include quantitative evidence of the impact on firms of their direct links with knowledge providers.
1. The article must use empirical quantitative methods.
2. The impact must be inflicted upon firms. Hence the unit of analysis is to be the firm itself rather than industry or geographical region.
3. There must be a direct or formal relationship established between the impact-inflicting and the impact-receiving party (analysis of pure spill-overs were excluded)
4. The article must refer to the impact of collaboration and use of knowledge providers.
The knowledge providers were grouped as: Universities (UNI), Research Institutes (RI) and KIBS.
A total of 153 articles remained once the title and abstract review had been done.
B.3. Quality Assessment
The 153 articles from the review were fully read and the final sample was reduced to a total of 100 articles. Some articles were excluded because they did not match any of the four inclusion criteria, although this was not clear from the abstracts. The motives for exclusion were:
• The utilization of knowledge providers was undistinguished from the utilization of other categories of agents beyond the scope of this research (e.g. customers or competitors). As a result, the impact from knowledge providers could not be disentangled.
• No formal relationships existed between firms and knowledge providers. That is to say, pure spill-overs were the focus of the analyses.
• The unit of analysis was not the firm but the knowledge provider, the region or the sector.
• The data analysis was merely descriptive.
B.4. Data extraction and Data Repository
Two types of information from each article were retrieved and stored in the data repository:
(i) General data from the articles (year, journal, geographical scope, industrial scope, data source, etc.) and (ii) data from the empirical research executed (method, sample, dependent variables, independent variables, results, etc.).
B.5. Overview of Articles Resulting from the Literature Review
The sample’s year distribution is from 1997 to 2014 (May). Over 60% of the sample is from articles published between 2008 and 2014. Hence most of the articles included in this research are recent studies, evidencing an increasing interest on this topic. The most commonly exploited data sources were Community Innovation Survey (CIS) and other CIS- style surveys (34 articles, 34%)2.
The utilization of knowledge providers was studied mainly in Europe (60% of the papers) and US (11%). From the industry perspective, manufacturing industries were included in almost every work, while service industries were addressed together with manufacturing in about half of the studies3.
2 Using the Guidelines for Collecting and Interpreting Innovation Data included in the Oslo Manual. The list of CIS-style surveys is made up of Mannheim Innovation Panel (MIP) from Germany, Panel de Innovación Tecnológica (PITEC) from Spain and surveys from Canada, Korea and Taiwan. Other surveys employed are Cambridge Business Research Survey (CBR) and Know Survey.
3 Just two articles dealt with service sector alone. Specific sectors were addressed in 21 papers, focusing mostly in medium or high tech manufacturing industries like Pharmaceutical, Electronic, Chemical, Biotechnology and ICT.
Figure 1 - Distribution of Studies
Chapter II – Impact Assessment Literature Review
C. What is known about the impact on firms of collaboration?
Articles were classified in three different types (T1, T2, T3) according to the research question(s) they addressed: (i) What kind of firms collaborate with knowledge providers? (T1, 36 articles); (ii) Do firms benefit from collaborating with knowledge providers? (T2, 63 articles), and (iii) Which firms benefit the most from collaboration? (T3, 29 articles)4. These three types of studies were not mutually exclusive, as some of them (about 30%) addressed several research questions.
C.1. What kinds of firms collaborate the most with knowledge providers?
About 36% of the articles from the review were classified into T1 (36 articles). Out of the 36 articles from T1, 50% merged different knowledge providers together in their analysis. For example, research institutes and universities. Another 36% analyses the determinants of collaboration of just one knowledge provider each. Finally, just 14% compares the determinants of using different types of knowledge providers. All in all, universities were included in 29 articles (81%), research institutes in 24 articles (67%) and KIBS in 9 articles (25%).
An important methodological issue is the potential existence of sample selection as samples are usually composed of innovators or R&D performers. If this is the case, results should not be generalized to the whole population of firms, unless this selection issue is dealt with.
None of the T1 studies address sample selection, with the exception of Chun and Mum (2012), who finds it to be very significant.
In these studies, dependent variables (Yi) are indicators of utilization. In most cases the indicator is a binary variable tracking whether the firm had some kind of link with knowledge providers (84%). Regarding independent variables (Xi), we grouped them into (i) those referring to characteristics of the firm and (ii) those addressing firms’ motives for using knowledge providers. The 98% of the articles assess variables related to characteristics of the firm. Articles studying the motives for using knowledge providers stand for 31%. Each of these groups is described in following sections.
4 These articles belong to two subcategories of articles. The first subcategory – T3A – is made up of articles aimed at studying the characteristic of firms affecting the impact of using knowledge providers (18 articles).
The second subcategory – T3B – was an indirect result from the review of the articles. It is made up of T2 articles analyzing the impact of using knowledge providers in different subsamples of firms according to some specific characteristics (e.g. small vs. large firms), allowing for indirect determination of characteristics of the firms that influence the impact of knowledge providers (11 articles).
C1.1. Main studied characteristics for using knowledge providers
The most frequently studied characteristic is the Size (29 articles). Size is studied employing both continuous (72%) and discrete (28%) variables. Regardless of the indicator type, the results are conclusive: Firm size influences the utilization of knowledge providers in a positive, statistically significant, way.
STYLIZED FACT 1
The size of the firm positively affects the utilization of knowledge providers.
The second most studied characteristic is R&D activity (25 articles). This characteristic is usually studied employing continuous indicators (72%). Regardless of the indicator, results agree that the influence of R&D activity is positive and statistically significant.
STYLIZED FACT 2
Internal R&D Activities positively affects the use of knowledge providers.
The firm’s Industry is the third most studied characteristic (14 articles). The taxonomies most frequently employed are OECD’s (58%) and Pavitt, K. (1984) (29%). The rest of articles use industry dummies or address specific industries. It is possible to state that as the technological level of the firm’ industry increases, so does the utilization of knowledge providers5.
STYLIZED FACT 3
The tech level of the firm’s industry positively affects the use of knowledge providers.
The rest of the characteristics are not studied as much as the previous ones. However, some of them are worth mentioning. Education level of employees, Education level of executives, Export activity and Public subsidies usually show positive coefficients6. Finally, when Foreign firms are distinguished, no statistically significant effect is found.
5 Note that this stylized fact is different from the previous one because there are R&D intensive firms in low tech sectors and vice versa (Barge-Gil et al, 2011).
6 Note that some of the public subsidies require the firm to cooperate in order to be eligible for the aid.
Chapter II – Impact Assessment Literature Review
C.1.2. Main studied obstacles for innovation that motivate using knowledge providers Obstacles for innovation are analyzed in 11 articles. Cost (82%) and Risk (55%) are the most often studied ones. These two obstacles, and specially Cost, positively affect the use of knowledge providers.
STYLIZED FACT 4
Cost of the innovation process positively affects the use of knowledge providers.
Table 4 - Summary of T1 Studies
Article a Country Survey Dependent (Y) Size R&D Activity Industry b Cost c
Adams et al., (2003) US Cooperation (D) +
Asakawa et al., (2010) JP Use (D) NS
Arranz et al., (2008) ES CIS Cooperation (D) NS NS + (O) NS
Arvanitis & Woerter (2009) CH Use (D) & (C) NS NS
Arvanitis el al. (2008) CH Use (D) + +
Arza & Lopez (2011) AR Cooperation (D) +
Becker & Dietz (2004) DE MIP Cooperation (D) + + + (O) +
Belderbos el al. (2004) NL CIS Cooperation (D) + + + (O) +
Bennett et al (2001) UK
Bercovitz & Feldman (2007) CA Use (C) NS + + (E)
Chun & Mun (2012) KR CIS Use (D) + + NS
De Fuentes & Dutrenit (2012) MX Cooperation (C) NS + NS
Eom & Lee (2010) KR CIS Cooperation (D) NS NS NS
Fontana et al., (2006) EU (6) d KNOW Cooperation (C) + + Freel & Harrisson (2006) UK Cooperation (D) + +
Fritsch (2001) DE Cooperation (D) + +
Fritsch & Lukas (2001) DE Cooperation (D) + +
Garcia & De Lucio (2008) ES Cooperation (D) + + + (P)
Janeiro et al., (2013) PT CIS Use (D) - +
Johnson et al., (2007) UK Use (D) NS
Lambrecht & Pirnay (2005) BE Use (D) NS
Laursen & Salter (2004) UK CIS Use (D) + +
Lopez (2008) ES CIS Cooperation (D) + NS +
Okamuro & Nishimura (2013) JP Use (D) NS
Miotti & Sachwald (2003) FR CIS Cooperation (D) + + NS -
Mole et al., (2008) UK Use (D) NS
Mole et al., (2009) UK Use (D) NS
Montoro et al (2006) ES Use (D)
Mora el al., (2005) ES Use (D) +
Robson & Bennet (2000) UK CBR Use (D) + + + (E)
Saez et al., (2002) ES Cooperation (D) + + + (P) NS
Segarra & Arauzo (2008) ES CIS Cooperation (D) + + + (O)
Teirlinck & Spithoven (2012) BE Use (D) NS + NS
Tether (2002) UK CIS Use (D) + + + (O)
Van Beers et al., (2008) EU (2) d CIS Cooperation (D) + + + (P)
Veugelers & Cassiman (2005) BE CIS Cooperation (D) + +
Notes: a The number of articles listed is lower than the total number of articles from T1 studies because some articles did not include any of the independent variables from the table. b Tech Level of the Sector. c Obstacle for Innovation. d More that one country is used in the analysis. (C) Continuous Variable. (D) Discrete Variable.
(E) Specific sectorial classification or other (i.e. Manufacturing and Services). NS Non-Significant. (O) OECD sectorial classification. (P) Pavitt’s sectorial classification.
C.2. Do firms benefit from collaborating with knowledge providers?
Examining the benefits for firms using knowledge providers is the most frequently studied topic in the literature. A total of 63 articles address this topic. Likewise, in articles from T1, most studies address the aggregated impact of several types of knowledge providers in their analysis (55%). In most cases the combined impact of utilizing universities and research institutes is studied (48%). Only in a few articles the impact of these two knowledge providers is assessed together with the utilization of KIBS (9%). In the articles studying the impact of knowledge providers individually (42%), the impact from universities is the most frequently evaluated one (29%). The individual impact of using research institutes (11%) or KIBS (12%) is hardly studied. All in all, universities are studied in 57 articles (88%), research institutes in 45 articles (69%) and KIBS in 13 articles (20%).
In T2 studies, a very important issue to deal with is endogeneity. The decision to establish a link with a knowledge provider is potentially endogenous. Hence, it is likely to be correlated with unobserved factors or ‘unobservables’ that also influence firms’ results (i.e. managerial ability). If this is the case, and endogeneity is not taken into account, results will be biased.
Despite the great importance of endogeneity issues in recent economic and managerial literature, they have seldom been addressed in T2 studies. Only five studies (Arvanitis et al., 2008; Eom et al., 2010, Cummings and Fisher, 2012, Yasar et al., 2012 and Robin and Schubert, 2013) have addressed endogeneity. The five of them used instrumental variable methods, although only Yasar et al., (2012) and Robin and Schubert (2013) provide a
Chapter II – Impact Assessment Literature Review
discussion on instruments’ validity7. In these studies, the main independent variable captures the link between the firm and the knowledge providers. Again, discrete variables are usually employed (86%).
Finally, dependent variables (Yi) are indicators of impact. Up-to 32 different impacts were assessed and 141 impact indicators analysed in the sample of articles from T2. To simplify the analysis of the vast number of assessed resulting impacts, they were grouped into three categories, following Barge-Gil and Modrego (2011): Technical Impacts (including new products, new processes or patents for example), Economic Impacts (including sales, profits or productivity for example) and Investment Impacts (including R&D or capital investments for example).
C.2.1. Main studied impacts out of the use of knowledge providers.
The most frequently assessed impacts are Technical Impacts (39 articles). These impacts are mainly studied using discrete indicators (84%). The impacts evaluated are mainly:
Product Innovation (35%), Patenting (30%) and Process Innovation (18%). The results are mostly positive and statistically significant (68%) with only 7% of the studies finding negative results8. The results are predominantly significant for Product Innovation and Patenting.
STYLIZED FACT 5
Utilizing knowledge providers positively affects Technical Results.
The second most frequently studied impacts in the literature are Economic Impacts (35 articles). The economic impacts are usually analyzed through continuous indicators (80%).
For Innovation sales (53% of articles) the results are mostly positive (68%) and usually statistically significant (53%). When the impact is found to be negative (26%), it is significant in 21% of the sample (2 articles)9. In the case of Sales (22% of articles), the results are always positive (100%), but only 38% of them are statistically significant. For Added value (17% of
7 In addition, Nieto and Santamaria (2010) test for endogeneity and do not reject exogeneity on the collaboration, Frenz and Ieto-Gilles (2009) and Harris et al., (2013) acknowledge the problem and the difficulty to find valid instruments so that they decide to address it by using lags of collaboration variables and matching procedures, respectively. Finally, Fabrizio (2009) uses a fixed effects regression to control for time-invariant unobserved heteroegeneity as a robustness check.
8 These three studies refer to New Patents in the pharmaceutical industry, Early-termination of R&D projects financed under the NIST program in USA and New Products and Processes in manufacturing SMEs from UK, respectively. In all three articles the knowledge providers are Universities.
9 Both studies take place in Taiwan and analyze low- and medium-technology sectors from the same sample.
articles), the results are again positive (100%), with 50% of them being statistically significant.
Evidence on the association between using knowledge providers and economic impacts seems to be positive but results are still not conclusive.
The third most commonly studied impact of using knowledge providers is Investment Impacts (7 articles). These impacts are assessed using only continuous indicators. The addressed impact is R&D expenditure. The use of knowledge providers has a positive (86%)10, usually statistically significant impact (71%).
STYLIZED FACT 6
Using knowledge providers has a positive association with Investment Impacts.
Table 5 - Summary of T2 Studies
Article a Country Survey Independent (X) Tech. b Eco. c Inv. d
Adams et al., (2003) US Cooperation (D) +
Almeida et al., (2011) N/A Use (C) +
Amara & Landry (2005) CA CIS Cooperation (D) +
Arranz & Arroyabe (2005) ES CIS Cooperation (D) + NS
Arvanitis et al., (2008) CH Use (D) + +
Arvanitis & Woerter (2009) CH Use (D) + +
Arza & Lopez (2011) AR Cooperation (D) +
Asakawa et al., (2010) JP Cooperation (D) NS
Aschhoff & Schmidt (2008) DE MIP Cooperation (D) NS
Barge-Gil (2010) ES PITEC Cooperation (D) NS
Becker & Dietz (2004) DE MIP Cooperation (D) + +
Belderbos et al., (2004) NL CIS Cooperation (D) +
Belderbos et al., (2006) NL CIS Cooperation (D) NS
Cohen et al., (2002) US CARNEGIE Cooperation (D) +
Cumming & Fischer (2012) CA Use (C) + +
Eom & Lee (2010) KR CIS Cooperation (D) + NS
Fabrizio (2009) N/A Cooperation (D) -
Fariaa et al., (2010) PT CIS Use (D) NS
Fey (2005) EU (2) e Cooperation (D) +
Freel & Harrison (2006) UK Cooperation (D) -
10 The only exception to this statement (negative although non-significant) in a work focused only on biotechnology firms.
Chapter II – Impact Assessment Literature Review
Frenz & Letto (2009) UK CIS Use (C) +
George et al., (2002) US Cooperation (D) NS NS NS
Grimaldi & Von Tunzelmann (2003) UK Use (C) +
Guan et al., (2005) CN Cooperation (D) NS
Hall et al., (2003) US Cooperation (D) - -
Han et al., (2013) KR Use (D) +
Harris et al., (2011) UK CIS Cooperation (D) +
Harris et al., (2013) UK CIS Use (D) +
Heirman & Clarysse (2007) BE Cooperation (D) +
Huang & Yu (2011) TW Cooperation (D) +
Hseuh et al., (2010) TW Cooperation (D) NS
Howells et al., (2012) UK Use (D) +
Inauen & Schenker (2011) EU (3) e CIS & MIP Cooperation (D) + +
Inge et al., (2010) BE Cooperation (D) +
Keizer et al., (2002) NL Use (D) +
Kim (2012) US Use (D) NS
Kim & Park (2008) KR CIS Cooperation (D) +
Knudsen (2007) EU (7) e Cooperation (D) +
Lambrecht & Pirnay (2005) BE Use (D) NS NS
Lin et al., (2009) TW Use (C) & (D) + NS
Loof & Brostom (2008) SE CIS Cooperation (D) +
Mas-Tur & Soriano (2014)f ES Use (D) +
Mackun & MacPherson (1997) US Use (D) +
MacPherson (1997) US Use (D) + +
MacPherson (2002) US Use (D) + +
Miotti & Sachwald (2003) FR CIS Cooperation (D) + NS
Mole et al., (2008) UK Use (D) NS +
Mole et al., (2009) UK Use (D) NS +
Mothe & Nguyen-Thi (2013) LU CIS Use (D)
Nieto & Santamaria (2007) ES Use (D) +
Nunez et al., (2010) ES PETRI Use (C) +
Nunez et al., (2012) ES PETRI Cooperation (D) NS
Robin & Schubert (2013) DE CIS Use (D) +
Su et al., (2009) TW Use (C) +
Teirlinck et al., (2010) BE CIS Use (D) NS
Tsai & Wang (2009) TW CIS Use (D) -
Tsai & Hsieh (2009) TW CIS Use (D) -
Un et al., (2010) ES Cooperation (D) +
Vega et al., (2009) ES PITEC Cooperation (D) +
Vega et al., (2008) ES CIS Use (C) +
Yam et al., (2011) CN Use (D) + +
Yasar et al., (2012) CN Use (D) NS +
Zeng et al., (2010) CN Cooperation (D) +
Notes: a The number of articles listed is lower than the total number of articles from T2 studies because some articles did not include any of the dependent variables from the table. b Technical Impact. c Economic Impact.
d Investment Impact. e More than one country is used in the analysis. f Addressing Intangible Impacts. (C) Continuous Variable. (D) Discrete Variable. NS Non-Significant.
C.3. Which firms benefits the most from collaboration?
One recent trend in impact evaluation literature is in the analysis of heterogeneous effects.
That is, the consideration that some firms may benefit more than others from using knowledge providers. Articles classified as T3 focus on analysing firms’ characteristics influencing the intensity of the impact inflicted by knowledge providers. Only 29% of the sample was classified as T3. Therefore, the articles aiming to explore the determinants of the impact are the least frequent ones in the literature.
In contrast to the previous two categories, in T3 most articles study knowledge providers independently (62%). In these articles, the determinants of the impact of firms using universities are the most studied ones (31%). The 21% study the determinants of the impact of firms using KIBS. The remaining articles from T3 study a combination of different types of knowledge providers. Out of these articles, the determinants of the impact of firms using either universities or research institutes are the most studied ones (38%).
In T3 studies, selection issues may be very important in these types of studies. Frequently, the sample is composed only of firms with links with knowledge providers. If this is the case, results should not be extended to the whole population of firms unless this selection issue is dealt with, something that none of T3 studies using selected samples does.
C.3.1. Main studied determinants of the impact of knowledge providers
The Size and R&D activity of the firm are the most frequently studied determinants of the impact of knowledge providers (12 articles each). For Size, different impact indicators were evaluated in several articles so that the total number of evaluations was 46. Results were usually statistically non-significant (70%). Size is found to be positive and significant in 23%
of the evaluations while negative and significant in 7% of the evaluations. It is worth noting that samples are usually selected (i.e. only SMEs) and, when using discrete indicators, the definition of size groups differs considerably across studies. R&D activity was evaluated on
Chapter II – Impact Assessment Literature Review
41 occasions. In 51% of the evaluations the results are statistically non-significant. In 29%
of the evaluations the results are positive and significant. Negative and significant results are found on 17% of the occasions.
The third most studied determinant is the Industry of the firm (7 articles). It was evaluated on 25 occasions (due to industry comparisons). Regarding the technological level of the industry, in 52% of the evaluations, the results were positive and statistically significant. Non- significant results accounted for the 36% of the evaluations. Significant and negative results were found in the 12% of the evaluations.
Some more determinants were studied. In most cases they were studied in a single article, so it is not possible to extract any conclusion or assessment out of them. Also the number of characteristics of the firms assessed is considerably lower in T3 studies than the number of characteristics studied in articles from T1. Therefore, the study of the determinants of the impact is in need of further research in order to develop stylized facts and to investigate the role played by other firms’ characteristics different from size, R&D and industry.
Table 6 - Summary of T3 Studies
Article a Country Survey Dependent (Y) b Size R&D Activity Industry c Barge-Gil & Modrego (2011) ES CIS Tech, Eco & Inv (D) NS
Belderbos et al., (2006) NL CIS Eco (C)
Bennett et al., (2001) UK Success (D)
Bishop et al., (2011) UK CIS Tech & Eco (D) + +
De Fuentes & Dutrenit (2012) MX Tech (C) +
Guan et al., (2005) CN Tech (C) +
Harris et al., (2011) UK CIS Eco (C) NS
Huang & Yu (2011) TW Tech & Eco (C) +
Knockaert et al., (2014) BE Tech & Intangible (D)
Liu et al., (2010) CH Tech (D) +
Loof & Brostrom (2008) SE CIS Eco (C) & Tech (D) +
MacPherson (1997) US Eco (C) + (E)
MacPherson (2002) US Tech & Eco (C) +
Mole et al., (2008) UK Eco (C) + NS
Mole et al., (2009) UK Eco & Inv (C) NS -
Mothe & Nguyen-Thi (2013) LU CIS Intangible (D) +
Nieto & Santamaria (2010) ES Tech (D) -
Nunez et al., (2010) ES PETRI Tech (D) +
Robson & Bennett (2010) UK CBR Success (D) NS