Technology transfer at Dutch universities : a performance measure

Technology Transfer at Dutch Universities:

A performance measure Author: Matthijs Pet Student Number: 10453695 Final Submission Date: March 25, 2016 Msc. In Business Administration – Entrepreneurship & Innovation Track University of Amsterdam Supervisor: T. Vinig/R. van der Voort

2

Statement of originality

This document is written by Student Matthijs Pet who declares to

take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is

original and that no sources other than those mentioned in the text and its

references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the

supervision of completion of the work, not for the contents.

3

Table of Contents

ABSTRACT ... 4 1.0 INTRODUCTION ... 5 2.0 LITERATURE REVIEW ... 8

2.1 UNIVERSITY TECHNOLOGY TRANSFER AND VALORISATION ... 8

2.2 THIRD MISSION OF UNIVERSITIES ... 9

2.2 STAKEHOLDERS AND THEIR MOTIVATIONS ... 13

2.2 THE PROCESS OF UNIVERSITY TECHNOLOGY TRANSFER ... 15

2.4 TECHNOLOGY TRANSFER MEASURES ... 18

2.4.1 Efficiency Analyses ... 19 2.4.2 Regression Analyses ... 24 2.4.3 other performance measures ... 26 2.5 FRAMEWORK ... 27 3.0 METHOD ... 29 3.1 SAMPLE ... 29 3.2 DATA COLLECTION ... 30 3.3 CALCULATING TECHNOLOGY TRANSFER PERFORMANCE ... 34 4.0 RESULTS ... 37 5.0 DISCUSSION ... 45 6.0 CONCLUSIONS ... 51 7.0 REFERENCES ... 54

4

Abstract

Next to no work exists on the process of technology transfer in the Netherlands, nonetheless policies at university and governmental level both push for a more entrepreneurial university. Government budgets are shrinking and universities are expected to compete for research contracts and grants, and to be more commercially active in the broadest sense. The main body of scientific publications is based on the United States and almost all quantitative work focuses on efficiency. This thesis applies a novel performance measure for technology transfer at Dutch universities. This work therefore contributes to the existing literature by taking a performance approach and is informative for policy makers and management at university and governmental level. The number of patents, licenses and spin-offs generated by the universities on a yearly basis are taken as the output of the technology transfer process. An estimation of the potential output is made through the use of the total scientific output and the research effort made by the university. The research effort is expressed in terms of investments in capital and labour. Data was gathered on these variables over the time period 2006-2011 from annual reports, technology transfer offices, governmental bodies, ministries and a range of institutes. Comparison of the potential output with the actual output revealed that much is to be gained. Only four out of ten universities perform on a nominal level. A grimmer picture appears when the results are compared with frontrunners from the United States. U.S. universities outperform Dutch universities by a factor of 2,5 to 6,0 when the performance measure suggested is applied.

5

1.0 Introduction

Universities have always fulfilled an important role in western society. What that role constitutes however, has changed over the last few decades. The university has evolved from an institute with a primary focus on freedom and independence of scholarly inquiry to being an important driver in the growth of the regional economy (Audretsch, 2014; Martinelli, Meyer, & von Tunzelmann, 2008). The emergence of the knowledge economy, the unfavourable current research funding conditions due to economic turbulence, and the technology revolution which leads towards an information and communication based society, are all drivers that change demands of higher education systems across the world (Hofer & Potter, 2009). No longer can a university be seen as a centre for education and research alone. Universities in this context are producers and disseminating institutions of knowledge (Guerrero & Urbano, 2012). The valorisation and subsequent dissemination of the knowledge produced by universities is known as university technology transfer (Vinig & Lips, 2015). This fairly recent and increasing emphasis on university technology transfer by universities is seen as the third role of universities. This three-pronged model of education, research and technology transfer has been strongly advocated in scientific literature (Baldini, 2006; Battistella, de Toni, & Pillon, 2015). The main body of research investigating this model has primarily been directed towards the United States where success stories from for instance the Michigan Institute of Technology (MIT) spurred a significant body of scientific work (Agragwal & Henderson, 2002; Shane, 2002). Across the board, universities appear to be not equally successful in commercializing the knowledge produced (Carlsson & Fridh, 2002). Some universities that spend above average have a well below average return on investment.

6 Other universities reap high rewards from a few licenses with many generating no income at all. The Association of University Technology Managers (AUTM) found that the number of patents granted on a yearly basis to U.S. universities has grown from less than 300 to almost 3300 between 1980 and 2005. The revenue generated through these activities has increased from $160 million in 19991 to $1.4 billion in 2005 (Vinig & Lips, 2015). Over the fiscal year of 2004, AUTM reported total revenues for US universities of $2.51 billion based on the income received from licenses and royalties. The total amount spent on research by these institutions over the same fiscal year was $41.24 billion (AUTM, 2004). These numbers are not only large, but inconsistent as well. Universities in the United States appear to spend more than sixteen times the amount that is generated directly by their third mission activities in 2004. Other research strengthens this argument; the third mission of universities mainly results in negative cash flows for the university (Sampat, 2006). Findings indicate that not every university has the same capabilities for transferring technology to society. A university mainly directed at humanities, arts and social sciences produces less knowledge that is directly applicable for business and industry compared to a university that is focused on science, engineering and medicine. (Abreu & Grinevich, 2013; 2014). It is therefore difficult to state that there is conclusive evidence concerning the benefits of the third mission of universities. The body of research covering the Netherlands is next to non-existent. Only one article can be found that investigates the performance of Dutch universities concerning the number of scientific publications in relation to the number of licenses, patents and spin-off companies (Vinig & Lips, 2015). This study shows that institutes like MIT and Stanford University perform almost twice as well in transferring their produced

7 knowledge to society. However, the size and research capacity of the different institutions is ignored. In addition, an assumption is made about the valorisation potential of each university. This assumption is not backed up, neither is it well explained. It can therefore be stated that although the technology transfer process is seen as highly important, and it has been documented before in the literature, there is hardly any evidence regarding the performance of that process in the Netherlands on institutional level. The goal of this thesis is to refine the measurement instrument developed by Vinig and Lips (2015) based on the critique provided above. The remainder of this thesis will be as follows. First, a definition of technology transfer and valorization will be given in the literature review. The reason why these processes are important for universities will be discussed afterwards. This will be followed by a thorough description of the valorization and technology transfer process. The role of the Technology Transfer Offices, and their fit into the overall structure will be discussed as well, taking into account the different factors used for input and output. Existing quantitative research will be discussed as well as the proposed framework. In the method section, the data sample and the method of gathering and processing the data will be discussed. This will be followed by an extensive discussion of the proposed performance measure. The results section informs on the data gathered and the calculations made. It will end with the performance of university technology transfer among Dutch universities as well as three U.S. universities. The discussion will give an overview of the conclusions that can be drawn from the results, including real life implications. An answer to the main question driving this thesis will be given in the conclusion together with suggestions for further research.

8

2.0 Literature Review

The main body of research published can be categorized along the themes of the mission of universities, the different stakeholders and their motivations, the process of technology transfer and the organizational structures that support the process, regional or international comparisons, the impacts of university research, the tangible outputs of university research and the efficiency of university technology transfer. These topics will be discussed in greater detail below, starting with a definition of university technology transfer and valorisation. 2.1 University Technology Transfer and Valorisation The concept of technology transfer originates from knowledge transfer as can be found within the field of knowledge management. Knowledge transfer is defined as the process through which one unit (eg., group, department, or division) is affected by the experience of another (Argote & Ingram, 2000). It is an intentional and goal oriented interaction between two or more social entities. During the process, the stock of knowledge either remains stable or increases (Autio & Laamanen, 1995). Knowledge within an organization can reside in the individual members, roles and organizational structures, the organization’s standard operating procedure and practices, its culture, and the physical structure of the workplace (Walsh & Ungson, 1991). Simpler said, knowledge is embedded in the members, tools, tasks, and the various sub networks formed by crossing or combining these three elements (Argote & Ingram, 2000). The process of transferring the knowledge is complex, due to its ability to be both tacit and explicit of nature (Jasimuddin, Klein, & Connel, 2005). Technology transfer is considered an active process during which the technology (and the knowledge related to it) is transferred between two distinct entities (Autio & Laamanen, 1995).

9 Valorization encompasses all activities that contribute to ensuring that the outcomes of scientific knowledge add value beyond the scientific domain (Benneworth & Jongbloed, 2010). It encompasses al activities aimed at making scientific research available to social entities outside the university. It is broader than the pure commercialization of knowledge in the sense that it is about the use of the knowledge beyond licenses, patents and spin-offs. It includes the co-production of knowledge with non-academic groups and the training of individuals. The valorisation and subsequent dissemination of the knowledge produced by universities is known as university technology transfer (Vinig & Lips, 2015). 2.2 Third Mission of Universities Over time, the function of the university within society has changed from an institute centred on the individual lecturers to a player within the knowledge market that contributes directly to the economic growth of regions and nations (Audretsch, 2014; Scott, 2006; Etzkowitz, 1998). Scott (2006) describes the transformation of the role of the university within society from the late medieval times to the postmodern era. During all these stages, service to society is the major role of the university. Throughout time, universities have played an important role in providing higher educational services in the realms of teaching, research and a diverse set of other academic services to the church, governments, individuals, an the larger public. During the later Middle Ages (1150-1500), when western universities first arose, the main emphasis was on teaching. The revival of mercantilism, the growth of cities and the urban middle class, as well as the intellectual renaissance during the 12th century gave rise to this new type of western institution. The philosophical goal of the

10 university was the pursuit of truth and learning. The dominant organizational structure of universities know today, with faculties, examinations, curriculums, and the bachelor and master degrees originates from this era. To be more precise, most of modern universities across the globe are modelled after the universities of Paris, Oxford and other European medieval universities (Scott, 2006). Early modern Europe (1500-1800) saw the rise of the nation state and the university as a research institute. The mission of the university shifted from teaching the individual to servicing the government of the nation state and faculty research. External service activities became critically important as well (Etzkowitz, 1998; Audretsch, 2014). The number of universities in Europe doubled during this era, mainly due to the on-going theological and political struggles between Catholicism and Protestantism. The Italian Renaissance University, which flourished between 1475 and 1600, was the prototype for the Humboldtian universities. Favouring vernacular languages, instead of the Medieval Latin, and with a greater emphasize on the individual, free will, and human values. The humanistic Humboldt model, founded by Alexander Humboldt in the 1800s and with a primary emphasis on freedom and independence of scholarly enquiry, eventually disrupted the link between the church and the university (Scott, 2006). This development appeared in the same time frame as the rise of the universities in the United States. Many American universities developed an effective set of institutional mechanisms that enabled the commercialization of research through the Morrill Act. Also known as the Land Grant Act, this law signed by Abraham Lincoln in 1862 granted land to each state that was to be used indefinitely to fund colleges beneficial to agricultural and mechanical development (Audretsch, 2014) The importance of more applied knowledge became more prominent with the Second World War and changes in policies based on research on the endogenous

11 growth of nations pioneered by Robert Solow (Audretsch, 2014). The so-called Solow model, wherein capital and labour were identified as main drivers of economic growth and the standard of living, shaped the role of the university from the 1950’s onwards. Technical change within this model is seen as manna falling from heaven. In this capital and physical labour driven perspective on the economy, the role of the university was in the realms of social and political sciences. With the explicit inclusion of knowledge alongside physical capital and labour within the model of endogenous growth the university was seen again as an important source of economic growth. Vannevar Bush, advisor to president Roosevelt, emphasized in 1945 that the new scientific knowledge originating within the Unites States was crucial for the creation of employment, economic growth and the dominant position of the United States in science and technology. His perspective still has an enormous influence on the federal policies of the United States (Kumar, 2010). The triple helix model, where the university plays a dynamic role in the creation of wealth, is now seen as the leading form for creating knowledge based economic development within an innovative environment (Etzkowitz & Leydesdorff, 2000). The university is seen as a cost effective and creative inventor and transfer agent of both knowledge and technology (Etzkowitz, Webster, Gebhardt, & Terra, 2000). A key role for the university is to a collect talent, therefore acting as an important driver for nations and regions in building capabilities and survival within the knowledge economy (Florida & Choen, 1999). Most aspects of the university contribute to the generation of entrepreneurial capital, both directly or through an orientation enhancing and celebrating freedom of creativity and inquiry, and with promoting awareness of these values beyond the direct influence of the university. This third goal of the university is to promote technology transfer and increasing the number of start ups as well as to

12 ensure that people add tot the emerging entrepreneurial society (Audretsch, 2014). The shift from a teaching and researching institute to one with a more central role in the development of economies arises from the internal development of the university, external influences on academic structures and the prevalence of innovative clustering at the regional level (Etzkowitz, Webster, Gebhardt, & Terra, 2000). A large portion of the leading role of institutions from the US is seen as a consequence of the Bayh-Dole Act, that enabled universities to claim ownership of the intellectual property generated by the universities’ researchers (Carlsson & Fridh, 2002), and the increase of linkages between research and industry (Wright, 2014). Unclear rules and low commercialization rates of university technologies are seen as a major reason for passing the Bayh-Dole Act (Rasmussen, Moen, & Gulbrandsen, 2006). In addition, some form of protection, via licenses and patents, were deemed necessary as an incentive for firms, universities, and investors. Some researchers claim that the rise and maturing of molecular biology and the use of computers within scientific domains were leading in increasing the patenting and licensing activities prior to the Bayh-Dole Act. The act itself merely accelerated and magnified trends that were already in place (Colyvas, et al., 2002). The impact of the Act is found to be not very large on the number or the quality of research. It is still questioned whether patens and exclusive licenses are the best way to maximize social returns of public research funds (Mowery, Nelson, Sampat, & Ziedonis, 1999) (Mowery, Nelson, Sampat, & Ziedonis, 2001). The European Union followed suit, for instance with the abolishment of the so called “professor’s privilege”, transferring the intellectual property of the researcher to the university (Schoen, van Pottelsberghe de la Potterie, & Henkel, 2014) (Rasmussen, Moen, & Gulbrandsen, 2006) or other changes in intellectual property laws to encourage ownership of inventions by the institutions (OECD, 2003). Subsequently,

13 many universities have sought to develop systematic and professional knowledge transfer and intellectual property management systems in the form of technology transfer offices (TTO’s) (Grimaldi, Kenney, Siegel, & Wright, 2011). 2.2 Stakeholders and their motivations Different actors are involved in the collaboration to bring new technologies to the market. There are three main reasons for universities to focus on generating new firms instead of collaborating with existing firms. Firstly, the firms originating at universities will acknowledge the university’s competence, financial situation and long-term mission. Thereby increasing the chances of creating long-term partners. Secondly, it provides a buffer for economic fluctuations. Collaboration with universities tends to only exist during years of economic growth. This would make the university vulnerably to the economic cycle. Thirdly, the establishment of new firms is the most visible output of universities. This increases the likelihood of acquiring public funding compared to the muddier realm of collaborative interaction with existing industry (Rasmussen, Moen, & Gulbrandsen, 2006). Other reasons to engage in technology transfer activities found within the same research are the attraction and retaining of highly talented people, securing future research funding, and for training benefits to students that follow entrepreneurship programmes at the university. The primary motive of the university scientist is the recognition received by the scientific community for the research produced. This emanates from presentations at conferences, publications in journals, and research grants. Other possible motives may be financial gain from the invention, the growth of the social network, and additional funding for the research program (Siegel, Waldman, & Link, 2003). Involvement in commercialization activities may also be an alternative career path. A lot of individuals

14 that set up a company from out a university appear to be frustrated with their situation due to a lack of research funding or promotion possibilities (Rasmussen, Moen, & Gulbrandsen, 2006). Table 1: Stakeholders involved in university technology transfer Stakeholder Actions Primary

Motive(s) Secondary Motive(s) Organizational Culture University Setting research agendas/distribution of research funding Creating long term partners, a financial buffers, most visible outputs Attraction of talented people Future research funding and training benefits Bureaucratic University

Scientist Discovery of new Knowledge Recognition within the scientific community Financial gain and a desire to secure additional research funding Scientific TTO Works with faculty and firms/ entrepreneurs to structure deal Protect and market the university intellectual property Facilitate technological diffusion and secure additional research funding Bureaucratic Firm/

entrepreneurs Commercializes new technology Financial gain Maintain control of proprietary technologies Organic/ Entrepreneurial Source: adopted from Siegel et al. (2003) and Rasmussen et al. (2006) The primary motive of the TTO is to protect and market the intellectual property generated by the university’s researchers. Secondary motives may include the promotion of technological diffusion and increasing the research budget via royalties, licensing fees and sponsored research agreements (Siegel, Waldman, & Link, 2003). The main motive of firms and entrepreneurs is the financial gains to be obtained through exploiting the university’s technology at hand. The secondary motive is to maintain control over the technology with for instance an exclusive worldwide license (Siegel, Waldman, & Link, 2003).

15 2.2 The process of university technology transfer Within the current triple helix system, several absorptive capacity mechanisms are necessary to disseminate the knowledge produced to society (Audretsch, 2014). These mechanisms ensure that the applied research is transferred through spill over mechanisms. Within the current society, science parks, incubators and technology transfer offices perform this task (Bercovitz, Feldman, Feller, & Burton, 2001) (Jensen, Thursby, & Thursby, 2003). Other mechanisms that can be used for technology transfer are publications, consultancy work, exchange programs, joint venture, training, contract research and cooperative R&D agreements (Lee & Win, 2004). TTO’s are offices within universities whose main goal is to guide the process of disclosing the invention made by scientist through patenting and licensing (Siegel, Veugelers, & Wright, 2007). They are responsible for the protection of intellectual property created by the university and the management of the commercialization process (Marman, Phan, Balkin, & Gianiodis, 2005). The process of disclosing the scientific discovery starts with the evaluation of the innovation for patenting by the university scientist and the TTO through invention disclosure by the scientist to the TTO. When the invention is disclosed, the TTO evaluates the commercial potential of the technology and makes the patenting decision. Often, interest shown by an industry partner is sufficient justification for filing the patent (Siegel, Waldman, & Link, 2003). When an initial industry partner is not present, the TTO manager starts with marketing the innovation through the network of the TTO after filing the patent (Siegel, Veugelers, & Wright, 2007). An important decision for the TTO manager is whether to seek global or domestic patent protection. This decision is based on the perceived market value and the resources of the TTO office. Global patents

16 are far more cumbersome and expensive to acquire (Siegel, Waldman, & Link, 2003). The faculty staff is often involved in the marketing phase due to their involvement with the technology, making them a natural partner for the interested firms, and they are in a good position to identify the possible partner. Successful marketing attempts end with the licensing of the innovation to an existing firm or with the emergence of a start-up with or without the help of the university (Siegel, Veugelers, & Wright, 2007). It is important to state that this linear model is not necessarily an accurate representation of the transfer process of all technologies (Siegel, Waldman, & Link, 2003). Interestingly, both Siegel et al. (2003) and Jensen and Thursby (2001) report that many firms will license the technology before the patent is filed, implying that the main roles of the TTO are invention disclosure, managing the stock of available technologies, and the maintenance and re-negotiation of licensing agreements. In the final stage of the process of technology transfer, the negotiation stage, the partners decide the benefits for the university such as an equity stake in the new venture, follow on research agreements, or payments based on the production or revenue of the product containing the new technology (Siegel, Waldman, & Link, 2003). The process as described by Friedman and Silberman (2003), starting with research expenditures and ending with jobs wealth is depicted in figure 1.

17 Figure 1: The process of university technology transfer (Friedman & Silberman, 2003) McAdam et al (2005) investigated the technology licensing process and the business building process within informatics and biosciences faculty and schools. Researchers within their process model are influenced by the discovery of the technology and awareness of disclosure options by the technology transfer office. After assessment of the technology, the decision on the development of the technology is made, followed by the decision of whether a license, spin-off, joint venture or sale of the patent is used to exploit the innovative technology. However, the researchers found that this route was barely followed due to that the approval process and database of research grants was independent of the technology transfer organ. This led to weak positions for the university when negotiating intellectual property agreements with the research partners. A tendency for the inventors of the technology to be over optimistic about the development of the technology and naïve about the marketing aspect was also found. Lowe (2006) created a staged model, based on the transfer of tacit knowledge or consulting and explicit knowledge or licensing. The creation of economic value through spinoff creation follows four stages. The first stage is where the business ideas are generated from research. The second stage is

18 where new venture projects are finalized. In the third stage, the venture is launched and the fourth is where the economic value generated is fortified (Ndonzau, Pirnay, & Surlemont, 2002). The commercialization process and success of universities in the United States have been scrutinized thoroughly in scientific literature (Carlsson & Fridh, 2002). Many researchers attempted to document the success through descriptions of spin-offs, licenses and patents (Henderson, Jaffe, & Trajtenberg, 1998) (Carlsson & Fridh, 2002) (O'Shea, Allen, Chevalier, & Roche, 2005). However, the same cannot be said about European, left alone Dutch, universities. Commercialization should be a voluntary activity for faculty members. It should be stimulated but not made obligatory. The different researchers should be free to publish and use results for further research and commercialization should not replace the more traditional activities of the university (Rasmussen, Moen, & Gulbrandsen, 2006). 2.4 Technology Transfer Measures A broad range of research trying to capture the technology transfer process can be found in the literature. Some only try to explain the differences in the amount of spin off companies generated (Algieri, Aquino, & Succurro, 2013; Aldridge & Audretsch, 2011), others try to capture a broad range of technology transfer activities (Carlsson & Fridh, 2002; Thursby, Jensen, & Thursby, 2001; Rogers, Yin, & Hoffmann, 2000). In this section, different attempts at measuring the technology transfer process will be discussed in order to develop to a process model that can be tested and to provide hypotheses. First, a selection of papers that use efficiency analysis will be discussed, followed by a

19 2.4.1 Efficiency Analyses Anderson et al. (2007) used a three-stage data envelopment analysis (DEA) approach to assess the efficiency of 54 U.S. universities. The input of their model is the research budget of each university. The outputs are the licensing income, the number of executed licenses and options, the number of patents filed and the number of patents issued. They found that only seven universities were efficient based on the input/output measures. Additionally, the researchers tested the effect on the DEA score for the presence of a medical school and whether the university was public or private using a linear regression on the same sample. Both measures were found significant. A study performed on 44 Spanish universities considered four inputs and three outputs in their DEA model. Berbegal-Mirabent et al. (2013) follow Anderson et al. (2007) in using the research budget as an input measure. Other input measures used are the total faculty, the administrative staff, and the administrative expenses. Their input can thus be seen as a combination of monetary spending on research and the amount of employees dedicated to research and education. The outputs used in their model are the number of graduates, the number of papers published and the number of spin-offs created. Their selection of inputs and outputs suggests that they are taking a broader perspective on technology transfer by including all three missions of the university, thereby deviating from the more generally accepted process model of Friedman and Silberman (2003), and Siegel et al. (2003) who perceive the number of papers published as an outcome of research expenditures, but an input for the invention disclosure procedure. Interestingly, the Spanish universities appear to be far superior to their counterparts in the United States, 21 of the 44 universities were operating efficiently. The difference in the proportion of the universities that are operating at efficient levels could be explained through the inclusion of the first and second mission

20 of the universities in the study performed by Berbagal-Mirabent et al. (2013). Berbegal-Mirabent et al. (2013) continue their investigation by clustering the different universities based on a variety of measures including the relative size and specialization of the university, as well as the experience with technology transfer as measured by the number of spin-offs created in the past, and the presence of a high tech sector in the region. Curi et al. (2015) investigated the productivity of French technology transfer offices after the introduction of two government reforms. Again, a DEA is performed. This time, the inputs are the amount of labour of the TTO and the amount of scientific publications produced by each university. The researchers distinguish between the core output, the number of patent applications and ancillary output. The latter include the number of patent extension requested and required and software applications. The choice for the number of patent applications as the main output of the technology transfer process can be ascribed to the function of the TTO’s in France. They do not normally engage in marketing activities. Chapple et al. (2005) focus on the number of licenses or licensing income in their parametric and non-parametric tests. Using both a DEA and stochastic frontier analysis (SFA) utilizing the number of invention disclosures, total research income, number of TTO staff, external legal intellectual property expenditure, the presence of a medical school, the age of the TTO, and the regional GDP and R&D intensity in a total of four models the researchers found that U.K. universities are operating at one fifth of their average efficiency. This means that U.K. universities could increase the number of licenses five fold with the given levels of input. Most of the research described here appears to base their modelling and theoretical argumentation on Siegel et al. (2003) and Thursby & Kemp (2002). Due to

21 their relative importance to the field, these two articles will be discussed in detail below. Both Siegel et al. (2003) and Thursby & Kemp (2002) base their research on data from the AUTM licensing survey spanning from 1991 to 1996 for academic institutions in the United States. Siegel et al. (2003) use a SFA analysis to assess the efficiency of universities in obtaining licenses and revenue from licenses. As core predicting values in their models they use the amount of invention disclosures, the number of staff employed by the TTO and external legal expenditures. They extend their model with environmental and institutional determinants of inefficiency. These include whether the university is public, has a medical school, the age of the TTO, state level R&D and real output growth which is a GDP measure. All core-predicting values are significant at either the 5% or 1% level. State level R&D appears to be only relevant for the number of licensing agreements and the age of the TTO is only statistically significant for the average annual licensing revenue. Thursby & Kemp (2002), using a DEA analysis with multiple outputs focus more on the type and quality of the university as predicting values for the efficiency of universities. They perceive the number of licenses executed, the amount of industry sponsored research, the number of new patent applications, the number of invention disclosures and the amount of royalties received as output of the technology transfer process. As inputs the amount of federal support and the number of employees of the TTO is used. Additionally, they use the number of faculty in biological sciences, engineering, and physical science and their respective quality as input factors. The researchers find that 58 of the 112 inhibit some degree of inefficiency. Interestingly, Siegel et al. (2003) find a mean efficiency of over 75% and 80% for their models. The lower average efficiency score of Thursby and Kemp (2002), besides their choice of instrument, could be explained by their model design. Comparing their outputs to the

22 process as suggested by Friedman and Silberman (2003), and used by Siegel et al. (2003) the amount of invention disclosures could be an input of the process instead of an output measure. In other words, Thursby and Kemp (2002) appear to treat the university as a whole as their unit of analysis compared to the TTO as unit of analysis in Siegel et al. (2003). Thursby and Kemp (2002) continue their analysis with a regression of the faculty size and quality and whether the university is public and has a medical school on the obtained efficiency scores. The respective faculty size and quality are included due to possible correlation and relationship problems between the two variables and efficiency. As a result of their regression, they find that the number of disclosures is not an explanatory variable for the efficiency of universities in their technology transfer process. Various explanations are offered, the informal way of handling invention disclosures used by faculty as proposed by McAdam et al. (2005). However, the number of licenses appears to have the greatest impact. This can be seen as an argument for the model design as employed by Siegel et al. (2003). Siegel et al. (2003) found constant returns to scale, whereas Chapple et al. (2005) found decreasing returns to scale considering the size of the TTO. Indicating institutional differences between the U.S. and U.K. Preliminary conclusions that can be made from the discussion of various efficiency models above include that there is no agreement on the variables to include in the measurement. The main disagreement is on the outputs to include in the analysis. This may stem from differences perceived in the process of technology transfer. Contract research for instance, expressed as the income from research conducted for third parties, is included as an input for the efficiency model by Berbagal-Mirabent et al. (2013) and perceived as an output measure by Thursby & Kemp (2002). For an overview of the selected inputs and outputs, see table 2. Another conclusion that an be

23 drawn is that the causality of the different variables is perceived different across the sample of research. Normally, regression analyses are used to indicate the causality and size of the effect of different variables in a model. Because of this, a selection of research using a regression analysis to investigate the third mission of universities will be discussed next. Table 2: Selection of literature using an efficiency analysis Paper Thursby & Kemp 2002 Siegel et al. 2003 Anderson et al. 2007 Berbagal-Mirabent et al. 20131 Curi et al. 2015 Chapple et al. 2005 Method DEA/ regression SFA DEA/ Regression DEA/ cluster DEA DEA/ SFA Dependent/output # Spin-offs D D Licensing income D S D D/S Number of licenses D S D D/S Patent applications D D D Stock of patents D Contract research D Invention disclosures D Measure/input TTO level TTO Budget TTO Size D D D/S TTO age S D/S External legal expenses S D/S University level Papers published D Invention disclosures S D/S Research Specialization D/R C Quality D/R Research capacity D Contract research D D D/S Presence medical school R S R D/S Public private R S R Experience in TT C Regional Factors High Tech Region C D/S State investment in R&D S D: variable used in DEA S: variable used in SFA R: variable used in regression analysis 1 _{Other outputs are omitted due to not being relevant for the third mission of universities.}

24 2.4.2 Regression Analyses Thursby, Jensen & thursby (2001) investigated the objective, characteristics and outcomes of university licensing using a regression analysis on 62 research universities in the United States. AUTM data for the years 1994-1996 is used for data collection. The amount of licenses executed by TTO’s is explained by the amount of invention disclosures, the presence of a medical school and the number of TTO employees. An interesting observation is that if the TTO’s priority is not the execution of licenses, TTO managers appear to aim at attracting additional research opportunities through sponsored research based on the invention disclosed. The number of disclosures explains the number of patents and the more developed the technology the higher the royalty received. One of the most comprehensive investigations of the role of technology transfer offices is performed by Carlsson & Fridh (2002). They studied the organizations and the place of the TTO within the university, the process of technology transfer and characteristics of the TTO itself. Their statistical analysis, on the process of technology transfer includes the same sample as used by Thursby & Kemp (2002) and Siegel et al. (2003), AUTM data for U.S. universities between 1991 and 1996. As performance measurement of the process the researchers, based on a survey held at U.S. universities, propose the number of patent applications the number of patents issued, number of licenses, the license income and the amount of start-ups generated by the TTO. Performance in this sense would be measured at different steps within the commercialization process. The patent application performance measure is based on the age of the TTO the total expenditures on research, and the amount of employees working at the TTO. The license measure, count and income, is dependent on the stock of patents. The amount of spin-offs generated is dependent on the age of the TTO, the

25 total research expenditures and the amount of employees working at the TTO. From all these predictors, the amount of staff employed by the TTO was not a significant predictor for the amount of spin-offs generated. All other predictors were found significant. For all models, except the one explaining spin-off formation, the explanatory value is high. The amount of invention disclosures for instance, can be explained by the age and number of employees of the TTO, and the total research expenditures for a total of 83%. One of the more interesting conclusions provided is that, as a rule, half of the invention disclosures result in patent applications; half of the applications finally result in patents, only a third of the patents are being licensed and a fraction of 10-20% of the licenses yield a significant income. Algieri et al. (2013) investigated the determinants of spin-off creation in Italy with a focus on the role played by the TTO’s. By using a regression analysis on a total of 20 universities, they found that budget available to the TTO, the number of employees working in the TTO, the percentage of faculty dedicated to research, as well as the regional focus on high tech influenced the creation of spin-off companies by the university. Di Gregorio & Shane (2003) focused specifically on the dynamics of spin-off creation by universities. Their findings indicate that the prestige of the institution is an accurate predictor. The quality of the researchers as well as that a higher prestige is a sign of higher trustworthiness, increasing the chances of attracting capital, may be possible explanations for this. The researchers observe a tendency of inventors to create a spin-off if the amount of royalties received from licensing activities is deemed too low. The inventor appears to make a trade-off decision between the profits obtained from licensing and creating a spin-off. Universities that provide start-up capital in the form of an equity stake generate more start-ups as well. Except for on, all predictors

26 that were found significant as an explanation of spin-off formation are of a financial nature, the ease of attracting capital and the perceived benefits for the inventor. The findings from the discussed regression research is summarized in figure 2. Factors with the most impact appear to be the experience of the TTO, the dedication to research in financial and employment terms, and the applicability of the research performed. Figure 2: commercialization process and their significant predictors 2.4.3 other performance measures Trune & Goslin (1998) compared the license income with the costs of the technology transfer offices, the patent fees, legal expense and new research grants as an estimation of the performance of the technology transfer process. Again, the annual AUTM survey was used as the data source. Within this research, only the year 1995 was used. They found that medical inventions reaped the most financial benefits. Universities without a medical school were profitable in only 27% of the cases; the presence of a medical school doubled this number. Vinig & Lips (2015) used a performance measure based on the amount of licenses, spinoffs, and patents produced by the TTO and compared that number with the amount of papers published as a measure of the stock of knowledge. Their results show that the technical universities are top performers.

27 All three technical universities of the Netherlands and the medical school of Rotterdam were found to be performing well. When the same method is applied to a selection of universities in the United States, the same pattern can be observed. Both engineering universities outperform the other universities. As the stock of knowledge, measured by the amount of papers published, is the only measure of input and the most practically focused institutes outperform the others, this pattern could be explained by the practical nature of the research conducted by engineering and medical institutions. This is consistent with the argumentation provided by Abreu & Grinevich (2013; 2014), universities aimed at social sciences and arts produce a higher percentile of knowledge that is not directly applicable. 2.5 Framework The focus of this research is to develop a performance measurement of the technology transfer process and apply this to the Dutch public universities. The most simplistic form of depicting performance is by means of an input/output model. In this model, the output of the valorization process is the number of patents, licenses and spinoffs. The input of the process consists of the number of papers, as a result of the number of researchers active and the budget available to the researchers. The process of commercializing research is treated as a whole. This means that only the question if something is at hand within the valorization process that results in significant differences from the rest of the pools is answered. What is at exactly causing the differences, is beyond the scope of this study. The reason for this is that a performance measure is to be developed, instead of investigating what lies underneath the performance differences.

28 Figure 3: Proposed model of University Technology Transfer Performance The term valorization potential is used as the three underlying factors, total papers published, number of researchers and research budget, do not always lead to a successful output. Only a fraction of all effort delivered by the university consists of innovative research that results in a potential technology that can be commercialized. It is assumed that these three factors represent a good proxy for the potential of the university’s technology transfer. The majority of the literature available on the topic of university technology transfer is based on TTO’s and universities in the United States, with lesser attention to Europe or Asia. The reported successes of some of these TTOs and their antecedents cannot be automatically transferred to the TTOs in the Netherlands. The total publication rate has been established as a properly working input for the calculation of the performance of the commercial output in one other paper (Vinig & Lips, 2015), the effort directly aimed at the research behind the research output, however, was not addressed in this research. It can safely be assumed that spending more on time and money on a certain topic will lead to better and faster results. Unintended discoveries, inventions that are commercially viable but were not the focus of the research at hand, can be found easily throughout the course of history. It is difficult, if not impossible, to account for coincidences, but safe to state that the more time and money is spent on a

29 topic, the more coincidental discoveries will take place. Therefore, the research budget and the number of researchers are included in this thesis to address this gap.

3.0 Method

3.1 Sample Historically, Dutch universities rank high on research output in international rankings. Rankings such as SJR, Sense and the CWTS Leiden Ranking all show Dutch universities to be in the well above average segment of both quality and number of publications. All universities in the sample established their TTO within the last decade. Several changes within this decade can be observed concerning the structure, the sharing of resources, and the facilitation of cooperation within the sample. The founding of the Innovation Exchange Amsterdam is an example of this. The IXA is the expert interface between Amsterdam-based academic institutions and parties interested in their research findings and knowledge. It is the collaborative entity formed by the TTOs of the Amsterdam University Medical Center (AMC), the University of Amsterdam (UvA), the Amsterdam University of Applied Sciences (HvA), Vrije Universiteit (VU) and the Vrije Universiteit Medical Center. The selection of universities and their respective TTOs is based on the valorisation-ranking research done by Elsevier/Scienceworks (van Leeuwen & Kolle, 2013). The ranking is based on a metric that includes the ability to be entrepreneurial in financial terms, to communicate and to collaborate. As the United States is often seen as a benchmark in performance of technology transfer, where MIT is often seen as the top performer, a few universities from the United States are included in the analysis. An overview of the universities included is given in table 3.

30 3.2 Data Collection The data was collected using data from existing research, annual reports and reports of overarching organizations, (semi-)governmental bodies, and telephone calls or e-mails to the respective contacts necessary. Availability of data, or the person handling the data, constrained the collection heavily. Although some sources delivered, an overall picture complete enough for analysis of other years outside the range of 2006-2011 could not be obtained. It proved to be impossible to extend the data significantly considering the commercial output as provided in the article by Vinig and Lips (2015) within the timeframe available for this thesis. Many sources indicated that it was impossible to disclose information prior to 2006 or more recent than 2011 and the use of data from the NWO to make a more complete estimation of the research budget made it impossible to use data more recent than 2011. Therefore, the data ranges from 2006-2011. Table 3: Overview of the sample

Research Universities Technical Universities Academic Medical

Centers (MC) Leiden University (LEI) Radboud University (RU) Utrecht University (UU) University of Groningen (RUG) University of Amsterdam (UVA) Vrije Universiteit (VU) Wageningen University (WU) Erasmus University (EUR) Delft University of Technology (TUD) University of Twente (UT) Eindhoven University of Technology (TUE) Leiden University Utrecht University University of Groningen Vrije Universiteit Erasmus University University of Amsterdam Universities from the U.S.A. to be included for comparison: Publically Funded Universities Privately Funded Universities Mississippi State University (MSU) New York University (NYU) Massachusetts Institute of Technology (MIT)

31 Concerning the universities from the United States, the data was collected through requests send by e-mail to the TTOs, information found on the websites and in the annual reports of the universities. The TTOs provided data regarding the number of licenses, patents and spin-offs. In some cases, the data was directly retrievable from the websites, in other cases they were provided after contacting the TTO in question. Data on the research budget was extracted from the annual reports by taking the research expenses. Data on the number of researchers was not available. In order to compare the Dutch with their U.S. counterparts, the wages paid to employees was used to calculate indices. The wages paid to employees by the Technical university of Eindhoven in 2006 was used as the base for calculating the indices. The data on the total number of scientific papers published was collected from the Association of universities in the Netherlands (VSNU). VSNU provided a year by year overview of all scientific publications including and excluding publications within the health sector. Such numbers for the universities from the United States were not directly available nor were they disclosed in the annual reports or after approaching the universities by e-mail. The CWTS Leiden Ranking however, uses data on total publications. Therefore, these numbers were used in the calculations. The number of researchers is accounted for by using the amount of Full Time Equivalents dedicated to research per university per year. This data is provided by VSNU as well. The amount of FTEs was obtained including and excluding the health sector. Again, U.S. universities did not disclose comparable data. The annual reports did disclose information on total wages paid, these numbers were compared with the total wages paid by the university that served as base year for the Dutch Sample. The Dutch system of financing universities is complex and far from transparent. In short, the main financing routes are threefold. A lump sum provided by the

32 government for education and research. This stream forms the basis for the activities performed by the universities. A large portion of the salaries for teachers and researchers, education and research itself, diverse laboratories, libraries, buildings, supporting employees and staff are financed directly through this subsidy. Although the subsidy consists of a portion for research and a portion for education, nowhere within the annual reports or the government budget can be found how this amount is divided. It is treated as a lump sum to be divided by the universities themselves and no transparency is given about the division. This subsidy accounts for roughly 60% of the annual budget of universities. The second stream consists of subsidies provided by the Dutch Organisation for Scientific Research (NWO) and the Royal Dutch Academy of Sciences (KNAW). The subsidies provided by both organisations and more often than not obtained by competition consist mostly of project or researcher bound subsidies. They are directly spent on research. The third stream of income for universities consists of other income. This includes the contract education and research and more specific subsidies from ministries and transnational sources, such as the European Union (VSNU, 2015). The second and the third stream taken together accounts for roughly 30% of the annual budget. The last major source of income consists of tuition fees paid and is not used for funding research, but streams directly towards education. As it is impossible to extract the budget available for research directly from the different sources available, a proxy or estimation of the variable to be measured is used. An interdepartmental policy research over the year 2013 provided an overview of the subsidies within the first stream per university (Ministry of Finance, 2014).

33 Figure 4: Financing the Dutch universities Source: Onderwijs in Cijfers (2015) To make an estimation for the research budget per university, the following method is adopted. Comparing the sum found in the interdepartmental policy research with the lump sum given in the annual reports of the same year of the different universities provides a ratio that can be used for other years. The education implementation service (DUO) provided an overview of the government spending based on the annual reports of the universities. The government spending was multiplied with the ratio calculated over 2013 to give an estimation of the part budgeted for research. The NWO provided data on the subsidies given to the universities up to and including the fiscal year of 2011. After 2011, the rules for handing out subsidies changed and the numbers after 2011 are not comparable. This subsidy is added to the number calculated. For instance, the University of Leiden states in their 2013 annual report that it received a total of 291,5, the governmental research from 2013 states a subsidy for research of 148,5. Thus, roughly 50% of the total subsidy is used for research and the other half for education. Assuming that no significant changes occurred in the timeframe examined, this 50% is used to approximate the research budget between

34 2006-2011 per year for the University of Leiden. The subsidies handed out by the NWO are added to give a number as complete as possible for the total research budget. The same is done for all other universities. 3.3 Calculating technology transfer performance The original model was developed under the assumption that research output provides a good estimation of the potential for the university’s technology transfer. The values 1, 2 and 3% of the total journal publications (TJP) was used to represent the potential for technology transfer or the potential valorization projects (PVP in %). The performance score (PER) is subsequently calculated by dividing the number of actual valorization projects (AVP) by the PVP. This gives the formula: PER= AVP/PVP Where AVP= P(patents) + S (Spin-offs) + L (Licenses) PVP = TJP X VP/100 Using this formula results in a normalized performance measure. A value of 1 in indicates that all potential for technology transfer is used. A value lower than 1 indicates a lower than expected performance, whereas a value larger than one indicates a performance above expectation. In this formula, the potential of total valorization projects appears to be more or less arbitrary, because of the use of 1,2 or 3% of all total journal applications. Adding the variables research budget and number of researchers will give a better indication of the number of possible projects. As stated above, it is assumed that the more time and effort is spent on research within a specific field, the more economically viable technologies will be invented.

35 The status of a scientist is directly influenced by the amount of the publications and the quality of the journals published in. The pressure originating from this has been coined ‘publish or perish’ and is seen as a source for the decay of scientific research (Colquhoun, 2011). A performance measure or performance index including other variables that are aimed more at the research itself, instead of the direct results, will therefore lead to a more balanced view of the performance of technology transfer. However, it is assumed that when more researchers and a higher budget are available within a particular field of study that the knowledge that is available within the field at single university is both broader and deeper of nature and will lead to more applicable or economic viable discoveries. The illustrate this assumption, one may think of research within the field of medicine. If one researcher looks at the influence of carbon monoxide in the case of a sepsis and specializes in the area of the lungs, other researchers my focus on different organs. The more research effort is made (i.e. time and money) the sooner applicable results arise that may lead to a treatment consisting of administering low dosages of carbon monoxide when certain medical conditions are present. An additional consideration is that not all research is equally expensive. Taking the same example as described above, the research includes, among other things, sophisticated measuring instruments, sophisticated specialized software, animals to test on and highly skilled researchers. Taking an example of a social study, the influence of leadership styles on the well-being of employees for instance, will require far lower levels of investments. The key to success here is access to leaders and employees that are willing to be tested on their leadership style and well-being. The main other resource consists of statistical software that is broadly available. An economic viable application is harder to spot as well, except maybe for business ideas in the region of

36 coaching or counselling which are easily copied. Thus, these ideas are extremely hard to On the other hand, research towards alternative fuel sources for automotive transportation requires high amounts of knowledge, testing, resources, etc. New ideas for batteries or storage of hydrogen in fuel cells for instance, require the assembly of multiple test-versions and intensive research. The results of this kind of research, however, is easily protected through intellectual property laws and easily commercialized through sale of the patent, a spin-off or a license. Indicating a high investment/high reward logic. Taking a resource based perspective, a higher investment in time and capital leads to a resource in the form of knowledge that is more valuable, rarer, harder to imitate, and harder to substitute with different knowledge. A high investment in time and a low investment in capital indicates a high likelihood that a resource will be discovered that is easily imitated and not easily protected. A low investment in both time and capital indicates no discovery of new technologies. A low investment in time and a high investment in capital indicates the set-up, or initial work, in a new direction of research. This will lead to publications, but no direct applicable work. A high investment in time and capital will thus increase the possibility for protection in terms of intellectual property and a higher chance of the presence of business opportunities. The 1,2 or 3% used to determine the potential valorization projects could therefore be substituted by a term consisting of time and money, or the research effort. Converting the research budget and the employment date to index-numbers (IFTE for the indices concerning FTEs and IBUD for the indices concerning the budget) yields a percentage that is more informed. This ratio is applied to the total journal publications and gives the number of potential projects.

37 This leads to the formula: PER= AVP/PVP Where AVP= P + S + L PVP = TJP * (IBUD/IFTE) % This formula will be applied to the sample consisting of Dutch universities and a selection of universities from the United States and will provide an overview of the performance that can be compared among categories.

4.0 Results

The total number of scientific publications per university can be found in table 4. The data was provided by VSNU and consisted of the number of scientific publications per university and the number of publications per university without publications in the health sector. It is assumed that all publications within the field of health research can be accredited to the faculty of medicine which in all cases is part of the academic medical centre of the respective university. The data is shown split between the university and the respective medical centre to show the relative importance of research within the health sector. All other data gathered could only be obtained on university level. All calculations involving the number of scientific publications therefore are the sum of the number of publications by the university and the respective medical centre.

38 Table 4: Number of scientific publications (TJP) for Dutch universities and the estimations for the U.S. Universities 2006 2007 2008 2009 2010 2011 LEI 2 978 3 445 3 528 3 178 3 426 3 389 LEI (MC) 2739 1559 1667 1806 1855 2108 UU 4 494 5 029 4 355 4 824 4 990 5 012 UU (MC) 2637 2473 2533 2743 3125 3258 RUG 3 199 3 525 3 826 3 489 4 004 3 963 RUG (MC) 1604 1788 1812 1907 2224 2149 EUR 1 763 1 896 1 610 1 997 1 988 1 901 EUR (MC) 2695 2803 3003 3115 3196 3368 UVA 4 788 4 788 4 361 4 421 4 608 4 942 UVA (MC) 2667 2667 3447 3563 3839 4162 VU 3 741 4 050 3 755 3 990 3 762 3 914 VU (MC) 2254 2109 2506 2266 2877 3052 RU 3 059 3 245 2 694 3 057 3 030 3 062 RU (MC) 2231 2317 2598 2542 2779 3122 TUD 6 688 6 653 6 946 6 934 6 486 5 840 TUE 3 335 3 468 3 453 3 593 3 784 3 318 UT 2 928 3 076 3 651 2 955 3 146 3 098 WU 2 829 3 135 4 075 2 988 3 240 3 240 Total 56 629 58 026 59 820 59 368 62 359 62 898 MSU 624 654 677 716 726 735 NYU 3 080 3 251 3 496 3 754 3 881 4 008 MIT 2 685 2 809 2 976 3 124 3188 3 253 Total 6 389 6 714 7 149 7 594 7 795 7 996 The commercialization output is displayed in table 5. Data is taken from the research performed by Vinig and Lips (2015) and where possible data was added by studying annual reports and by contact with the respective TTOs. Not all universities and TTOs were willing or able to provide the information requested. For example, the Innovation Exchange Amsterdam was not willing to give data on request regarding the UVA, the AMC, the Vu and VUMC. Therefore, only the total number of patents, licenses and spin-offs over the years 2006-2010, as found by Vinig and Lips (2015), could be presented. A special case is the university of Wageningen. They were more than happy to provide data, nonetheless only data on the amount of patents granted was available. For

39 completeness, the WU is shown in all the tables but will not be considered in the remainder of the analysis. It is worth mentioning here that the U.S. universities were not only able to respond quicker, but were able and willing to provide more statistics over the time period requested as well. Table 5: Commercial Output expressed in the number of patents (P), Licenses (L), spin-offs (S) and total (T) 2006 2007 2008 2009 2010 2011 P L S P L S P L S P L S P L S P L S LEI + MC 15 - 1 18 13 0 20 23 1 29 22 - 22 6 2 21 2 2 UU + MC - - 3 22 - 3 14 7 - 8 - - 25 - 2 15 - 5 RUG + MC 4 11 1 9 5 6 11 4 2 13 10 5 8 8 1 9 6 2 EUR +MC - - - 11 3 10 12 4 6 11 10 3 15 14 1 UVA + MC 60 patents - 29 license agreements - 13 spin-offs - - - VU + MC 10 17 7 20 5 6 23 5 8 10 7 9 16 - - 19 13 5 RU + MC - - - 9 8 11 9 8 11 9 8 11 9 8 11 TUD 96 - 17 99 - 9 41 - 11 46 16 11 28 9 17 48 9 14 TUE 13 4 20 15 5 19 16 10 17 10 16 15 10 7 12 23 12 5 UT 24 - 1 20 - 1 14 - 3 30 - 4 12 3 6 13 6 7 WU - - - 11 - - 8 - - 6 - - 2 - - 5 - - MSU 11 12 2 4 10 3 7 1 4 5 4 0 8 12 1 3 12 5 NYU - - - 767 patents - 445 License agreements- 70 spin-offs MIT 145 97 23 168 85 24 140 68 20 153 67 18 174 61 17 153 79 26 The symbol - indicates that no data was available For the UvA and its MC only a total over the period 2006-2010 could be obtained Statistics concerning the NYU could only be obtained for 2007-2011 Table 6 displays the number of full time equivalents that is spent on research by each university. The annual reports and data provided by VSNU were used a sources. Statistics on the amount of FTEs deployed was not available for the U.S. universities. Therefore, the wages paid are used and compared with the Dutch university that serves as the basis for the index calculations. Table 7 displays the calculated part of the research budget provided through the subsidy received directly from the government, also known as the first stream of income. The research part in percentages over 2013 calculated in table 7 was used to approximate the part of the first stream of income meant for research. Table 8 displays the results of this computation. Data provided by the NWO on the subsidy that is part of

40 the second stream of income was added to this number, as displayed in table 9. The total estimations calculated in table 9 were taken as well as the total research effort displayed in table 6 and converted to indices in table 10. The lowest number in the year 2006 for each category was taken as base year (100). Table 11 shows the potential valorization projects, the actual valorization projects and the performance, calculated according to the formula given in the methods section. Table 6: The amount of Full-Time Equivalents spent on research per university and faculty and staff salaries for U.S. Universities and TUE 2006 2007 2008 2009 2010 2011 FTE _{LEI + MC 1009,22 1044,60 1986,19 2076,49 2090,67 2177,72 UU + MC 2165,95 2125,50 2356,03 2418,62 2454,96 2530,62 RUG + MC 1472,59 1387,80 1406,60 1479,26 1700,78 1717,16 EUR +MC 1274,50 1059,20 1361,79 1408,18 1387,24 1395,88 UVA + MC 1868,35 1868,35 1145,10 1227,10 1242,05 1285,36 VU + MC 1532,30 1634,90 1565,56 1585,33 1718,27 1743,37 RU + MC 1736,40 1702,10 1736,86 1904,93 1983,54 2257,76 TUD 1581,36 1605,00 1635,45 1598,66 1530,69 1497,99 TUE 902,80 916,00 904,90 1080,70 1098,20 1121,80 UT 976,00 953,00 970,29 985,14 1024,27 1068,29 WU 754,03 778,00 725,14 808,40 818,00 941,60} Wages MSU 223,329 241,86 260,30 270,82 281,22 264,17 _{NYU 462,92 490,525 488,222 503,583 519,58 546,725 MIT 815,45 836,69 896,15 967,38 967,19 1006,46 TUE 118,90 125,20 133,20 141,60 149,70 148,40}