Characteristics and effectiveness of university spinoff support programmes

Transcription

1 Characteristics and effectiveness of university spinoff support programmes José María Beraza-Garmendia Escuela Universitaria de Estudios Empresariales, University of the Basque Country UPV/EHU, San Sebastián, Spain, and Arturo Rodríguez-Castellanos Facultad de Ciencias Económicas y Empresariales, University of the Basque Country UPV/EHU, Bilbao, Spain Abstract Purpose - This study identifies different models of programmes for supporting the creation of spin-offs in universities, analyses their differentiated characteristics and identifies the determinants of their effectiveness. Design/methodology/approach - The analysis was performed using data collected by way of a survey, targeted at the heads of spin-off support programmes in universities in the United Kingdom and Spain, to which we applied the factorial and cluster analysis techniques, and then a logistic regression analysis to confirm their results. Findings - This analysis enabled us to identify in these universities three types of spin-off support programmes. Among the various types of programmes identified, we found one that appears to be model for effectiveness. We have also found the existence of a certain country effect on the characteristics of the effective model. Finally, we have been able to note the importance given by the literature to university R&D activity and to the existence of a favourable environment to the performance of spin-off programmes. Research limitations/implications This research is limited by the use of the number of spin-offs and the survival rate as performance indicators of support programmes. Future research should consider their effective contributions to economic growth and the extent to which such effects could be related to university-level policies. Practical implications - The typology of the spin-off support programmes identified in this work provides insight that enables us to make some recommendations for improving the least effective models. Originality/value - This study contributes to the understanding of the role of university policy measures on spin-off support programmes effectiveness, and on how these policies are influenced by the environment. Keywords: University entrepreneurship, Knowledge transfer, Academic spin-offs, Technology transfer offices, University spin-off support programmes Article Classification: Research paper Introduction Studies on the mission of the university have placed increasing importance on transferring research results to the market as a source of development and competitiveness (Geuna, 1999; Etzkowitz et al., 2000; Mowery and Sampat, 2001; Shane, 2004). However, this is a complex interactive process, involving a wide variety of agents (Benner and Sandstrom, 2000; Etzkowitz et al., 2000; Okubo and Sjoberg, 2000). Conscious of the difficulties it poses, universities and governments have begun establishing policies for promoting this type of development (European Commission, 2004;2008). In this context, the creation of knowledge-based firms has become particularly important in recent decades (Callan, 2001; European Commission, 2002; Shane, 2004), bringing with it a proliferation of university support programmes for setting-up spin-offs (Golob, 2003; Shane, 2004; Del Palacio et al., 2006; Helm and Mauroner, 2007). Nonetheless, there are considerable differences in the way the various programmes are structured and operate, in terms of their aims, strategies, functions and activities and the organisational structures and services they offer (Roberts and Malone, 1996; Clarysse et al., 2002; Degroof, 2002; European Commission, 2002; Clarysse et al., 2005; Wright et al., 2007; Mustar and Wright, 2010). In addition, many of these programmes are newly-founded and consequently have neither a solid organisational structure nor clearly identified activities (Heirman and Clarysse, 2004; Clarysse et al., 2005). For this reason, different models of 1

2 academic spin-off support programmes have been proposed (Roberts and Malone, 1996; Clarysse et al., 2002; Degroof, 2002; Clarysse et al., 2005). Now, these works are intended to U.S. universities and some European countries like France, Belgium, United Kingdom, the Netherlands, Germany, Italy and Ireland. Therefore, it may be interesting to analyze the reality in other countries, and validate the models identified. For this, the main aim of this work is to identify the different models of spin-off support programmes to be found in British and Spanish universities and to analyse their differentiated characteristics, distinguishing those that seem to be more effective. We believe that this study is of particular interest, since by studying universities from two European countries with a different degree of experience in developing spin-off support programmes we have a varied sample that may identify different types in supporting the creation of spin-offs, so that the results can achieve a degree of generality beyond the analysis for both countries. Moreover, most studies carried out to date on this issue have analysed spin-off creation in a single university or in a small number of benchmark universities (for example, Roberts and Malone, 1996; Clarysse et al., 2002; Degroof, 2002; Serasols et al., 2009). This study between wider sets of universities will therefore provide a more realistic view of the position of the spin-off support programmes in universities. For this purpose, we use data collected by way of a survey, targeted at the people in charge of university programmes for supporting the creation of spin-offs, and a statistic analysis was performed applying univariate and multivariate techniques. Firstly, we have applied the technique of factorial analysis to identify the most significant variables explaining the characteristics of these programmes. Secondly, using the cluster analysis technique, we have classified the universities. Using a one-factor analysis of variance (Anova) and a logistic regression analysis we went on to describe the differentiated characteristics of each of the clusters found, also analyzing the extent to which the results support the hypotheses and lead to identify the characteristics of the most effective programmes. The paper consists of an introduction and five other sections. In the second one, we review the literature on types of university spin-off support programmes, identifying the different models suggested by different authors, and propose the hypothesis of the study. Thirdly, we explain the research methodology used. Fourthly, we perform several multivariate statistical analysis, especially the cluster analysis technique, to identify the different existing models of spin-off support programmes in British and Spanish universities. Fifthly, we make a descriptive statistical analysis and a logistic regression analysis to characterise each of the models identified and highlight its main differentiating features; also we discuss the results obtained considering the degree which support or not the hypotheses. In the sixth section, we summarises the main conclusions obtained. Theoretical framework Literature review We then proceed to review the main typologies of university spin-off support programmes developed in the literature. These mostly consist of qualitative research based on case studies. Below we show the characteristics and results of these studies. Roberts and Malone (1996) can be considered the pioneers in this area. For them, the two main dimensions of a policy of knowledge commercialization through the creation of spin-offs are selectivity and support, for each of which they distinguish two levels: high and low; they argue that there are only two viable models of support policy: low selectivity/low support and high selectivity/high support. A low selectivity/low support policy means that many spin-offs are created to which little support is given. The choice of the projects eligible for support is left to external agencies (for example, venture-capital funds), which are seen as having greater experience and expertise in choosing possible winners and prevent possible conflicts of interest. This strategy makes sense in settings in which spinning-off is common practise and venture capital is in abundant supply. By contrast, in a high selectivity/high support policy, a small number of spin-offs is created to which a large degree of support is given, giving them large possibilities of success. This strategy is more likely in environments where spinning-off is more unusual and venture capital is scarce. In this context, a university wanting to develop a support policy has no choice but to take the place of the financial market, playing the role of a financial investor, and encouraging a culture of entrepreneurship. Degroof (2002) analyses how an environment unfavourable to entrepreneurship affects the type of spin-offs created in the academic institutions. Based on a study of spin-off creation in five organisations of this type, he distinguishes two models for spinning off in this type of 2

3 environment. One model is that pursued by the specialist research institutions and the other is more typical of universities. The process used by the specialist research institutes has a long incubation period, lasting several years. The new company is only incorporated when it can avail of protected technology, a business plan with strong market potential, a convincing business model to draw on and lastly an entrepreneurial team capable of leading the project with the help of venture-capital organisations, company directors from the industry and other advisers. In contrast, the process pursued by universities does not normally include incubation or assistance in developing the business plan. The spin-offs are founded at a very early stage, when the project is still undefined. As a result, in most cases the business is developed after the spin-off s incorporation, when it is already up and running as a firm. The author goes on to relate the results of the study to Roberts and Malone s selection and support dimensions in the spin-off process and concludes that the process adopted by the specialist research institutes involves pursuing a high selectivity/high support policy, whereas the process identified in the universities involves a low selectivity/low support policy, thus contradicting Roberts and Malone s conclusions. However, the author notes this process is not static, but becoming increasingly more sophisticated as the institutions learn from their experience. This approach sheds lights on one important practical aspect: the difficulty of establishing from the outset a high selectivity/high support policy in an environment that is unfavourable to entrepreneurship. Implementing such a policy requires considerable resources and skills, which may not exist in the universities; moreover, implementing them involves, among other factors, bringing about a considerable cultural and structural change. Even in unfavourable environments, therefore, a university can start out in a position of low selectivity/low support and gradually move towards a position of greater selectivity and support, although it remains to be seen whether all universities are capable of following this path and at what speed. Finally, Degroof (2002) notes that the sort of spin-off process pursued will have implications for the type of firm created. His study shows that in reality different types of companies are created by different types of research institutions, pursuing different processes. The specialist research institutes create venture capital backed firms, after a long process of incubation, whereas in the universities the companies created at an initial stage are mainly lifestyle spin-offs, although as the programmes gain greater experience, more growth-oriented spin-offs may emerge, particularly prospector spin-offs, if the environments are unfavourable to enterprise. For their part, Clarysse et al. (2002), Clarysse et al. (2005) and Wright et al. (2007) analyse the extent to which these institutions differ when it comes to organising and managing spin-off activities, the resources needed to undertake these activities appropriately, and whether the differences in organisation and management lead to different types of spin-off. They identify three possible missions of research institutions in this respect, aimed to the creation of three different categories of spin-offs: self-employment oriented, economically profitable and success-oriented. The mission of a research institution is to stimulate the creation of self-employment oriented spinoffs when the objective is to create employment and enhance development in a depressed region. Alternatively, it can set itself the mission of stimulating the creation of economically profitable spin-offs. Finally, the creation of success-oriented spin-offs arises when the aim is to create businesses capable of generating high returns for investors. They go on to make an in-depth analysis for each of the institutions selected of the different activities into which a proactive spin-off process can be broken down, following Degroof s division by stages (Degroof, 2002), and of the resources needed to perform the spin-off process. They initially distinguish three models of support for the creation of spin-offs: the low or self selective model, the supportive model and the incubator or protective model. In the low selective model the aim is to generate as many spin-offs possible. To this end, the purpose is to stimulate entrepreneurial initiative, relegating to a second place the analysis of the economic or financial potential of the initiatives. This means that the main activities are opportunity search and awareness creation. The firms created may be based not only on technology, but also on skills developed within the university. Hence many companies are created, but only a few have the ambition to grow, and even fewer to actually succeed. In short, the companies created can be classed as lifestyle spin-offs. In the supportive model, the aim is to create companies with economic potential and a will to grow, which will set up in the immediate territorial area of the institution, foster regional development and forge links with the institution that will encourage industry relations. In this model, management of intellectual property and business plan development, and public and private funding to allow the projects to develop in the initial stages are essential. Consequently, fewer 3

4 spin-offs are created than in the previous model. Companies created in this model can be classed as prospector spin-offs. The incubator model seeks to create solid companies, with the result that venture capital organisations are involved in them from the outset. Applying this model requires a research group that enjoys international recognition in a given technology. In addition, the technology transfer unit must be capable of incubating the project, facilitating the recruitment of outside managers, attracting international venture capital from the initial stages and forming the base of the company s intellectual property. Companies created along these lines can be classed as venture capital backed spin-offs. In a later work, extending the study sample, the authors identified two further categories, differing from the three previous models: resource deficient organisations and competence deficient organisations. In the resource deficient group they include spin-off programmes which have ambitious goals, but lack the necessary resources to achieve them. The lack of resources is reflected in the fact that the programmes do not have enough financial resources, the staff does not have either the expertise or the necessary contact networks to carry out the activities required, and in a lack of support from the institution s management. Because of these deficiencies, such programmes are positioned as being weak support and are therefore incapable of achieving the aims initially set out. The competence-deficient group includes spin-off programmes that have sufficient resources, but lack the skills needed to perform the necessary activities. This group shows what can happen if a support programme is endowed with a large quantity of resources to establish supportive or incubator models, but lacks the knowledge aimed to integrate these resources in a way that will generate the required skills, which can only be developed over time. Summing up, we draw the following conclusions from the literature: firstly, there is no single model for a spin-off support policy; secondly, the two main dimensions of a policy of support for this type of company are selectivity and support; thirdly, it is very important that the universities, based on environment-specific conditions, know what their objectives are and clearly set out what resources and activities are required to achieve these objectives; and fourthly, establishing a high support and high selectivity policy requires considerable resources to which many universities rarely have access on an individual basis; and even if they can, they may require a certain period of time to develop the necessary competences (skills) to utilise these resources efficiently. Hypotheses The literature review shows that the effectiveness of a university spin-off support programme depends on several factors. Some of these factors reflect the policies followed by the support programmes described in the previous subsection. However, other factors can be studied from the perspective of the resource based view (Wernerfelt, 1984; Barney, 1991; 2001; Grant, 1991). We have considered both perspectives by raising the hypotheses of this study. Thus, according to several authors (Hague and Oakley, 2000; European Commission, 2002; Shane, 2004; Wright et al., 2007), an innovating environment appears to have a positive effect on the characteristics and results of university spin-off support programmes. Spin-offs are more common in places where high technology start-ups are more common because the components necessary to create spin-off companies (experienced managers, customers and suppliers, and so on) tend to be present in those areas. Hence: H 1. An innovation environment promotes the effectiveness of university spin-off programmes. Roberts and Malone (1996) and Brunitz et al. (2008) conclude that a high selectivity/high support policy is more likely in environments where spinning-off is more unusual and venture capital is scarce because in this context a university has no choice but to take the place of the financial market, playing the role of a financial investor, and encouraging a culture of entrepreneurship. Also, Fini et al. (2011) found that universities should limit their support in contexts in which there is a significant contribution offered by ad-hoc regional support mechanisms. Hence: H 2. A policy of high support/high selectivity is more likely in unfavourable environments. But, however, as has already been mentioned, Degroof (2002) shows that, in unfavourable environments, the spin-off process in the universities involves a low selectivity/low support policy, but this process becoming increasingly more sophisticated as the institutions learn from their experience. Therefore, a university can start out in a position of low selectivity/low support and gradually move towards a position of greater selectivity and support. Hence: H 3. Universities, in unfavourable environments, initially be forced to start with low selectivity/low support programmes for the creation of spin-offs. Before knowledge can be successfully commercialized, knowledge needs to be generated. Therefore, the commercialization of research results through spin-offs is partly influenced by the 4

5 quantity and quality of the research at the university. According to Lockett et al. (2003; 2005), O Shea et al. (2005), Powers and McDougall (2005), Rodeiro et al. (2008), Gómez et al. (2008), Vinig and van Rijsbergen (2009) and Van Looy et al. (2011), the university s stock of technology is one of the most widely used inputs with a positive influence on the university spin-off process. Another factor is academic quality (Shane, 2004). According to this author, spin-offs are more likely to be founded to exploit the technology of more prestigious universities than to exploit the technology of less prestigious ones. Nevertheless, Rodeiro et al. (2008) do not find significant associations between the academic quality and the spin-off activity. Hence: H 4. Previous stock of technology promotes the effectiveness of university spin-off programmes. On the other hand, some studies that have analysed institutional aspects consider that a university s spin-off activity reflects its institutional behaviour. Universities with a culture of supporting activities to market their research results will obtain better technology transfer and spin-off creation results (Searle, 2006; Fini et al., 2009; Rasmussen and Borch, 2010; Uyarra, 2010). Roberts (1991) maintains that the university's social rules and expectations are a decisive factor in its technology transfer activity. Likewise, Polt et al. (2001) highlight the fact that many factors that are critical to the success of knowledge transfer (suitable institutional environment, organisational culture, institutional leadership, appropriate incentive system, level and orientation of the research, legal context) cannot be resolved by the intermediation structures on their own. As a result, they will often fail to promote knowledge transfer if these barriers have not been overcome beforehand. Similarly, Solé (2005) argues that a university whose objectives include a commitment to development will create more companies than one in which transfer holds a less prestigious position. Hence: H 5. The university s commitment to promoting an entrepreneurial culture promotes the effectiveness of university spin-off programmes. The relationship between the resources possessed by a Technology Transfer Office (TTO) and the technology transfer performance has been studied thoroughly. Rogers et al. (2000) found that large personnel resources are essential in achieving effective technology transfer. Likewise, Thursby and Kemp (2002) found a positive relation between the number of TTO staff and the licensing activity. Also, O Shea et al. (2005), Gómez et al. (2008), Caldera and Debande (2010), Vinig and van Rijsbergen (2009) and Van Looy et al. (2011) found that the number of TTO staff has a positive influence on the number of spin-off companies formed. Nevertheless, Lockett and Wright (2005) and Rodeiro et al. (2008) do not find significant associations between the number of TTO staff and the spin-off activity. Hence: H 6. The human resources available in the programme foster the effectiveness of university spin-off support programmes. In addition to the quantity of personnel resources, also the quality and experience of the personnel is a predictor of the effectiveness of university spin-off programmes. According to Lockett and Wright (2005), O Shea et al. (2005), Powers and McDougall (2005), Rodeiro et al. (2008), Gómez et al. (2008) and Vinig and van Rijsbergen (2009), the greater the experience that a TTO has in handling technology transfer activities, the more their staff will have developed the skills and abilities necessary to handle the spin-off process. Hence: H 7. The experience of the programme contributes to the effectiveness of university spin-off support programmes. At the same time, some authors believe that one of the characteristics of university spin-off support policies is their level of proactivity (Pirnay, 2001; Degroof, 2002; Wright et al., 2007). Degroof (2002) says that in an unfavourable context for entrepreneurship, universities are obliged to consider ways of increasing the number of ideas which can be spun off. This requires a proactive search for technological opportunities with commercial potential. Along the same lines, Moray and Clarysse (2005) suggest the creation of an Idea Board with a technological orientation that will enable proactive identification of technological ideas or opportunities. However, a proactive attitude can pose problems in terms of efficiency: institutionalising the systematic identification of promising research projects may require mobilising a volume of resources that is not justified by the aims being pursued (Pirnay, 2001). Hence: H 8. The proactivity in searching for and detecting ideas contributes to the effectiveness of university spin-off support programmes. University researchers normally have limited financial resources which are insufficient to meet the financial needs of the initial phases of a spin-off. The university s participation in financing these early phases is therefore critically important (Mustar, 1997; Heirman and Clarysse, 2004; Shane, 2004). Moreover, this participation reflects its commitment to exploiting the research commercially and to stimulating innovation (McCooe, 2004). 5

6 At the same time, in return for the spin-off s use of the university s intellectual property, the university generally receives an initial amount of money and/or royalties; in recent years, however, it is becoming increasingly common for it to take a share in the capital of the spin-off, which either complements or substitutes this initial payment (Bray and Lee, 2000; Matkin, 2001). These authors argue that this practice aligns the interests of the two parties, thus improving relations between the university and the spin-off, and provides greater prestige and legitimacy to the spin-off. The results of a European Commission study suggest that spin-off support programmes that have funds to invest tend to be more productive, in terms of the number of spin-offs created per project supported (European Commission, 2002). Moreover, the researchers/entrepreneurs sometimes have a commercial instinct and can adopt the role of manager/entrepreneurs, although this does not happen in most cases. On many occasions, therefore, the spin-off cannot be set up and developed without participation from external individuals with complementary skills and networks of relations, which the spin-off can call upon when it needs to make various kinds of decision. In general, however, an excessive dependency on the university by the spin-off may be counterproductive. It is important for the new company to have its own identity and to be located outside the university. The company must be run by the people who work in it and not by university staff (Belani, 2004). Hence: H 9. The university s involvement in the spin-offs contributes to the effectiveness of university spin-off support programmes. One of the dimensions of a policy of economic exploitation through the creation of spin-offs is selectivity (Roberts and Malone, 1996; Degroof, 2002; Clarysse et al., 2002, Clarysse et al., 2005; Wright et al., 2007). A university must decide how rigorous it is going to be in selecting research results that might be exploited through a spin-off. It is important to bear in mind the university s motives. If the aim is to back technological projects with strong growth potential, it should implement a very rigorous selection process, accepting only the most interesting projects. In this way, it will be in a position to offer them all the attention and resources they need to realise their promising perspectives. If, on the contrary, the aim of the model is to create as many spin-offs as possible, these spin-offs need not be only technological, but can also be based on skills developed in the university. Hence: H 10. The universities pursue different spin-off support policies, generate firms of different characteristics, and have different levels of effectiveness. Fig. 1 shows graphically these hypotheses, which together make a representative model of the conditions for the effectiveness of a university spin-off support programme. Below we present the research methodology used, including the model of the spin-off process used as a starting point for the empirical study, the process employed for gathering and processing the information, and the questions asked in the questionnaire used as the basis for this study. Methodology Starting model for the spin-off process Several authors have used a linear approach to describe the spin-off process and its various stages (Roberts and Malone, 1996; Reitan, 1997; Pirnay, 2001; Degroof, 2002; Golob, 2003; Hindle and Yencken, 2004; Shane, 2004; Vohora et al., 2004; Clarysse et al., 2005). There is, however, no consensus as to the number of stages, what they should be called and at what point the process begins. In designing the empirical study we have used a model of the linear process of valorisation via spin-off, which sets out the various stages a university needs to consider in supporting the creation of spin-offs, with particular emphasis on stages and activities at which the university s direct intervention can be decisively important. These can be grouped analytically into four basic stages: promotion of entrepreneurial culture, search for and detection of ideas, evaluation and valorisation of ideas, and creation of spin-offs, as shown in Fig. 2. Population of European universities, selection of the sample and informationgathering technique The study population comprises European universities in general, and Spanish and British universities in particular, that perform some type of activity involving spin-off creation[1]. The system used to identify the study population and select the sample was as follows: firstly, we searched in Google for a list of European universities, classified by countries; secondly, we visited the websites of each of these universities one by one; thirdly, we tried to identify a body specifically devoted to supporting the creation of spin-offs, if none could be found we looked for the body responsible for technology transfer activities or failing that, the body responsible for 6

7 research/innovation/corporate relations; and finally we identified the persons in charge of these bodies: name, position, telephone number and address. Altogether a total of 74 universities were identified in the UK and 255 in the rest of Europe (not including Spanish universities). We sent these universities a letter of presentation by , inviting them to fill out the online questionnaire. The information-gathering process ran from the beginning of November 2005 to the end of February Replies were received from 25 universities in the United Kingdom and 42 universities elsewhere in Europe, representing a response rate of 34% and 17%, respectively. In the case of Spanish universities, the letter of presentation was sent to practically all universities with a Technology Transfer Office (TTO), the great majority of which were public universities. 35 replies were received, representing a response rate of 58%. Altogether, a total of 389 letters were sent, and 102 replies were received, representing a response rate of 26%. Not all universities that replied to the questionnaire answered all the items. Specifically, eleven universities replied to few items and twenty-four furnished no information on spin-off creation. Thus only 67 of the 102 universities in the sample replied to all or practically all items. We then proceeded to discard universities with a certain number of incomplete replies or those that might be considered atypical, finally obtaining a database comprising a total of sixty-five universities, of which eighteen were in the UK, twenty-three in Spain and twenty-four elsewhere in Europe. The rate of complete replies is therefore 17% for the sample as a whole, 24% for British universities, 38% for Spanish universities and 9% for other European universities. Questions This empirical study used information taken in a survey on a series of variables related to resources and results, activities, organisational structure, the university s relations with the spinoffs, how long the programmes have been in place and the effectiveness achieved. As stated, in selecting the questions used in the research, we used the model of the linear process of valorisation via spin-offs shown in Fig. 1 which distinguishes four basic stages. To these four stages we have added a further section corresponding to general information. The questions included in each of the five sections are described below. General Information The university itself is the most important element in the support programmes, as a source of marketable research results (Polt et al., 2001). For this reason, as well as the identifying data, three additional questions were included in the survey[2]: the type of body performing the spin-off setup support activities; the number of people in that body; and the activities related to this type of company carried out at the university. Promotion of entrepreneurial culture In unfavourable conditions for entrepreneurship, a prerequisite for creating spin-offs is the promotion of an entrepreneurial culture amongst university personnel (Jack and Anderson, 1999; Klofsten, 2000; Pirnay, 2001; Trim, 2003; Henry et al., 2005). Four questions were included in the survey for this reason: the university s level of commitment to promoting entrepreneurial culture; the actions being taken to promote this culture; the number of people that will benefit from these activities and the degree of effectiveness obtained. Search for and detection of ideas Commercially exploitable ideas deriving out of university research do not normally arise spontaneously (Long and McMullan, 1984; Siegel et al., 2003; McDonald et al., 2004; Shane, 2004). Four questions were included in the survey for this reason: the university s level of proactivity in searching for and detecting ideas; the actions carried out to search for and detect ideas; the source of the entrepreneurial ideas and the degree of effectiveness obtained. Evaluation and valorisation of ideas The ideas initially detected need to be assessed to determine whether they meet a series of prerequisites for viable commercial exploitability (Wright et al., 2004). Likewise, the university, as an institution, and the people from whom the idea comes must support the project in order for it finally to become a spin-off (Baxter et al., 2004; Vohora et al., 2004). Nine questions were included in the survey for this reason: the use of a specific methodology for evaluating and exploiting ideas; the use of external personnel in technological assessment; the use of external personnel in market assessment of the idea; the profile of these personnel; who assumes the 7

8 leadership in promoting the spin-off; the role generally taken in the spin-off by the research group from which the idea originated; the number of exploitable ideas detected over a one-year period; the percentage of these ideas that are positively evaluated and, finally, the percentage of positively evaluated ideas that lead to the creation of a spin-off. Creation of spin-offs The ideas detected and supported lead to the creation of spin-offs, which can be of varying types and have varying degrees of success (Clarysse et al., 2005), and with which the university can have different links (Lockett et al., 2003). Nine questions were included in the survey for this reason: number of spin-offs created over the last five years; type of spin-offs created; average length of time from detection of the marketable idea to creation of the spin-off; most common source of financial resources; university s stake in the spin-off s capital; university's involvement in its management; survival rate of this type of company; percentage of firms that fail before three years and, finally the year in which the university began activities to support spin-off creation. Multivariate Analysis Factorial Analysis: obtaining representative substitute variables We used the database drawn from the information obtained to make a factorial analysis to determine the most significant variables explaining the characteristics of the university spin-off support programmes. Using this technique we proceeded first of all to identify factors. To do this, we used the Principal Components Analysis data reduction method. To determine the number of factors to be extracted, we used the latent root criterion technique. We then calculated the contributions of each variable to the different factors and selected the variables that contributed most to each one, in order to identify the variables that most appropriately described the university spin-off support programmes, and use them in the subsequent cluster analysis. Forty-seven variables were originally used in this study, all the quantitative variables in the survey; the value of the determinant of the correlation matrix is practically zero; likewise, Bartlett's sphericity test rejected the null hypothesis that the matrix of correlations is an identity matrix with a significance level of 1%, factorial analysis is therefore a relevant technique for analysing these variablesand; and by applying the latent root criterion the final solution chosen was that formed by sixteen factors. In order to improve the solution, a Varimax rotation was used. This solution preserves % of total variability. We then selected the variables with the greatest load for each factor, as representative of each of the factors. In this way, we managed to group the original quantitative variables that are intended to represent different aspects corresponding to the different stages of support in the creation of spin-offs, into sixteen variables. Table 1 shows these variables classified according to the different phases of the process of exploitation by spin-off. Cluster analysis: typology of spin-off support programmes in the universities of United Kingdom and Spain Once the most significant variables had been detected, we have classified the British and Spanish universities using the cluster analysis technique, from the sixteen most significant variables identified in the previous sub-section. The distance used was the Euclidean distance squared, following standardisation of the variables by transforming them into Z scores with mean 0 and standard deviation 1. The clusters were formed using hierarchical clustering and Ward s method. There is no standard procedure for determining the final number of clusters, and in this case the clustering coefficient criterion does not give clear results, since it does not experience relevant changes when the number of clusters is varied. We therefore opted for obtaining various different cluster arrangements, from two to four, and, by using the one-factor analysis of variance (Anova), check whether there were significant differences between the clusters obtained. The results obtained in the Anova analysis show that there are a number of variables with significant differences that are practically the same for the three- and four-cluster arrangements and that differences are smallest for the two-cluster arrangement. We have therefore opted for the three-cluster arrangement, since it allows greater differentiation between the programmes, without proving excessive. In this cluster arrangement, eight variables have significant differences. Table 2 shows these variables. 8

9 Characterisation of spin-off support models and discussion of results We now identify the characteristics distinguishing spin-off support models by means of a Anova analysis and a confirmatory multivariate statistical analysis: logistic regression analysis[3], and discuss the results obtained. Characterisation of spin-off support models Having identified the clusters and the variables with significantly different means, we should now explain in more detail the characterisation of the clusters identified, as representatives of different models or types of spin-off creation support, comparing and contrasting them, to highlight possible similarities and differences. For the purposes of this characterisation, we are first going to use all the quantitative variables included in the survey that show significant differences between clusters, as well as one new variable not included: number of patents. The reason for including this one variable is as follows: given the somewhat subjective nature of cluster analysis and in order to ensure the validity and practical relevance of the solution obtained, it is recommendable to incorporate variables that have not been used to form the clusters, but which are known to vary in value from one another, as is the case with the variable chosen; moreover, this variable is relevant in this respect, as we show below. Patent applications by universities are an indicator of university R&D results and their commercial orientation. While not all academic spin-offs are based on patented knowledge, a relationship can also be expected between the number of patents and the business creation process in the universities. However, in some cases the number of patent applications by the universities varies greatly from one year to another. For this reason, the variable used in the analysis was the average number of patent applications during the period in the case of British universities, and in the case of Spanish universities. Table 2 shows the means of all the quantitative variables with significant differences obtained in the survey for the clusters of spin-off support programmes. Taking as our reference these variables we go on to describe the profiles of each of the three types of programme identified: Cluster 1 is made up of twenty universities, seven in United Kingdom and thirteen in Spain, with a relatively small level of patent activity, which have spin-off programmes started up recently and have a small number of people devoted to spin-off support; they give relatively minor importance to the evaluation and valorisation of ideas and, conversely, a relatively greater importance to the promotion of entrepreneurial culture; their commitment to promoting an entrepreneurial culture and the importance given to business design/project competitions, entrepreneurship courses and promoting the spin-offs created is limited, with the result that their activities benefit a small number of people and they have little success in promoting an entrepreneurial culture; they are not very proactive in searching for and detecting ideas, they place little importance on monitoring the projects undertaken by research groups and the business design/project competitions, with the result that they have little success in searching for and detecting ideas and, therefore, the number of ideas detected is very limited; they give a positive rating to and spin-off a small percentage of the ideas detected; they generate few ventures, but with a high survival rate, and do not intervene in their running. Cluster 2 is in turn made up of eleven universities, seven in United Kingdom and four in Spain, with a relatively high level of patent activity, experience in spin-off support, and with a large number of people devoted to spin-off; they give relatively greater importance to the evaluation and valorisation of ideas, and to the spin-off support; their commitment to promoting an entrepreneurial culture and the importance given to business design/project competitions, entrepreneurship courses and promoting the spin-offs created is high, with the result that their activities benefit a large number of people and they have quite success in promoting an entrepreneurial culture; they are quite proactive in searching for and detecting ideas, lesser extent the cluster 3; moreover, they place quite importance on monitoring the projects undertaken by research groups and great importance on monitoring business design/project competitions, with the result that they have quite success in searching for and detecting ideas and, therefore, the number of ideas detected is relatively high; they give a positive rating to and spin-off a small percentage of the ideas detected, even exceed the percentage of cluster 1; they generate a large number of spinoffs though their rate of mortality is also high, and they tend to intervene in their running, but not actively. Finally, cluster 3 is formed by ten universities, four in United Kingdom and six in Spain, with a relatively small level of patent activity, which have spin-off programmes started up recently and 9

10 have a small number of people devoted to spin-off support; they give relatively greater importance to the evaluation and valorisation of ideas and, conversely, a relatively minor importance to the promotion of entrepreneurial culture; their commitment to promoting an entrepreneurial culture and the importance given to business design/project competitions, entrepreneurship courses and promoting the spin-offs created is limited, with the result that their activities benefit a small number of people and they have little success in promoting an entrepreneurial culture; they are quite proactive in searching for and detecting ideas, they place importance on monitoring the projects undertaken by research groups and the business design/project competitions, but they have little success in searching for and detecting ideas and, therefore, the number of ideas detected is very limited; they give a positive rating to and spin-off a high percentage of the ideas detected; they generate an intermediate number of spin-offs, with a reduced mortality rate, and do not normally intervene in their running. Secondly, starting from a smaller number of variables with significant differences between the clusters identified, we shall characterise the profiles of these clusters. Specifically, the seven variables used in this characterisation are as follows: The age of the spin-off support activities as an indicator of experience and, consequently, of availability of the necessary skills to carry out the activities required. The number of people in the body, as an indicator of the resources available to the spin-off support programme. The university s commitment to promoting an entrepreneurial culture, as an indicator of integration of entrepreneurial vision into the university s strategy and, consequently, in the day-to-day activity of the organisation. Proactivity in searching for and detecting ideas, as an indicator of capacity to detect possibilities for commercialisation of the knowledge generated in the university from an early stage. Selectivity, as an indicator of the university s rigour in selecting the results of research capable of being commercialised via a spin-off. This variable has not been included in the survey, but has been obtained indirectly from the percentage of ideas detected that are positively rated and percentage of positively rated ideas that are spun off variables. A smaller selectivity value means that a large percentage of positively-rated ideas detected are spun off, and therefore, that the corresponding cluster pursues a low-selectivity policy; inversely, a lower selectivity value means that a small percentage of positively-rated ideas detected are spun off, and therefore, that the corresponding cluster pursues a high-selectivity policy. The effectiveness rate, as an indicator of the results of the support programmes. This variable has not been included in the survey, but has been obtained indirectly from the average number of spin-offs created between 2000 and 2004 and percentage of spin-offs who survives for 3 years. The involvement in the spin-off s management, as an indicator of the university s involvement in the spin-offs. Fig. 3 shows the profiles of the three types of programmes identified according to these variables with significant differences indicated above. Type 1 has little experience in spin-off support, has few resources for this task and not enjoys a lot of commitment from the university, is little proactive in searching for and detecting ideas and follows a high selectivity policy; as a result, it has little effectiveness in the creation of spin-offs, and do not intervene in their running. Type 2 has a lot of experience in spin-off support, has plenty of resources for this task and enjoys great commitment from the university; it is not very proactive in searching for and detecting ideas and follow a policy of intermediate selectivity; as a result, it has a great effectiveness in the creation of spin-offs, and intervenes in their running. Lastly, type 3 has little experience in spin-off support, enjoys few resources for this task and less commitment from the university, is very proactive in the search and detection of ideas, but follows a low selectivity policy; as a result, it has an intermediate effectiveness in the creation of spin-offs, and intervenes in their running, but not actively. Therefore, type 2 is seems to be the most effective. Most significant features differentiating spin-off support models Once identified the characteristics distinguishing spin-off support models, we proceeded to determine whether there were any statistically significant differences between type 2 and the other and which of these variables best quantified the differences between these two groups of 10

11 universities by means of logistic regression analysis. For the purposes, we are going to use the seven variables included in the above characterisation except the effectiveness rate. Table 3 presents the main statistics of the logistic regression analysis. Cox and Snell s R 2 and Nagelkerke s R 2 statistics show that the overall fit of the model improves in each step, since the value increases. The value of the likelihood ratio (-2 log of the likelihood) falls, which also demonstrates that the overall fit of the model improves with each step. Similarly, the Hosmer- Lemeshow chi-square statistic makes it possible to accept the null hypothesis that there are no differences between the observed probabilities and those forecast by the model, which allows us to confirm the goodness-of-fit of the final model. Finally, Press s Q statistic makes it possible to conclude, with a significance level of 1%, that the success prediction is better than that obtained randomly. Once it was determined that the logistic regression model and the precision in the classification are statistically significant, the results were interpreted by examining the model to determine the relative importance of each independent variable in the differentiation of the groups. For this purpose, we analyzed the coefficients of the model and their significance. The significance of each coefficient was assessed using the Wald statistic. In addition, in order to interpret the logistic regression model obtained, we studied the sign of the coefficient assigned to each independent variable. The positive sign of the coefficient indicates that the increase in the corresponding variable raises the probability that a university is allocated to the group with Code 1. On the contrary, the negative sign of the coefficient indicates that the increase in the corresponding variable reduces the probability that a university is assigned to the group with Code 1 or, contrarily, that it raises the probability that a university is allocated to the group with Code 0. The variables included in the logistic regression model, their respective coefficients, and their levels of significance are presented in Table 4. The logistic regression model obtained reveals that the variables number of people who are part of the body and university s commitment to the promotion of entrepreneurial culture can be seen to be positively associated with the dependent variable. Given that this is dichotomous, type 2 with Code 1 and types 1 and 3 with Code 0, this implies that these variables occur to a greater extent in type 2. On the contrary, the variable year in which the university began activities to support spin-off is negatively associated with the dependent variable, which means that this variable occur to a greater extent in types 1 and 3. These results confirm those obtained in the Anova analysis, with the exception of those corresponding to the variable number of people who are part of the body, which is not significant in the logistic regression analysis. Consequently, in relation to types 1 and 3, type 2 is characterized as having more resources available to the spin-off support programme, a greater integration of entrepreneurial vision into the university s strategy and a greater experience and, consequently, availability of the necessary skills to support spin-offs. Discussion of results Type 2 is found to be the most effective. This type has the resources, experience and university s commitment to support spin-offs. It seems to opt for a policy of intermediate proactivity in searching for and detecting ideas and selectivity of the identified ideas that lead ultimately to the creation of spin-offs. Moreover, it appears more supportive of the spin-offs, since it is the cluster with the most mean time between detection of the idea and spinning-off, since it is the cluster more involved in their management. This results in the creation of a relatively large number of companies, but with a not very high survival rate. Thus, type 2 appear to be a model, implementing a policy that comes close to the supportive model proposed by Clarysse et al. (2002), Clarysse et al. (2005) and Wright et al. (2007). In contrast, type 1 show that it is difficult to establish from the outset a policy of high proactivity/high selectivity/high support, since following this policy requires considerable resources and skills that are not always available. Although universities of this type have opted from the beginning for a high selectivity policy, their low proactivity and reduced involvement in the spin-off s management limits the effectiveness in the creation of spin-offs. Finally, type 3 programme appears to show that some spin-off programmes are resource-deficient or competence-deficient. Among other factors, the lack of resources is reflected in the fact that the support of the university administration is lacking; and the lack of competences is reflected in the absence of the skills needed to carry out the activities required. Although universities of type 3 appear to be very proactive and pursue a policy of very low selectivity, have an intermediate effectiveness in the creation of spin-offs. 11