1 Introduction

With this study we aim at a comprehensive mapping of the technology activities of private firms and the public research sector (i.e., universities) for Switzerland. We want to identify the collaboration potential or knowledge and technology transfer potential between the private and the public research sector. The well-known concepts of “inert areas” (see Leibenstein 1989), “satisficing behavior” (see Simon 1956), “bounded rationality” and “technological competences and knowledge” (see Nelson and Winter 1982), “absorptive capacity” of a firm (see Cohen and Levinthal 1989), the resource based view of a firm (see Penrose 1995; Wernerfelt 1984; Barney 1991) or technology trajectories (see Dosi 1982) are used in the economic literature to describe the ability of a firm to perceive, process and apply external knowledge and/or to change its innovation behavior in order to further develop their technology base and to develop and commercialize new products. We learnt from these concepts that collaboration among actors with similar technology/knowledge bases are more likely than among partners with a very different knowledge background. Thus, technology proximityFootnote 1 matters and it is desirable that private enterprises know about the technology activities at universities and can make use of such activities.

Technology proximity between the two sectors (private and public research) indicates their collaboration potential. It tells us whether they “speak a similar language”. With the study at hand we look at the potential for KTT (Knowledge and Technology Transfer) based on “technology proximity” and how it is currently used.Footnote 2 To this end we collected information about the technology activities of firms (patent statistics) and the technology activities of universities (technology fields were assigned; see section on data). Furthermore, we used data on technology transfer activities between the two sectors. We combined the three datasets for the purpose of this study. Section two discusses technology orientation and KTT with universities. In section three we discuss the components of an empirical model and formulate the hypotheses. Section four explains the empirical strategy in order to answer the hypotheses. Section five introduces the different sources of data. Section six shows the results and answers the hypotheses and section seven concludes.

2 Technological orientation and knowledge and technology transfer with universities

Technology (knowledge) proximity between firms is an important factor for the probability of collaborations and the likely outcome of a research partnership. Technology proximity is usually measured through the patent activities of firms (see Jaffe 1986; Nooteboom et al. 2007; Cantner and Meder 2007). Although the same argument would be valid for RD (Research and Development) collaborations between firms and universities, this field of research did not get the same attention. Lack of data could be one reason for it. However, technology proximity might play an important role especially for the intensity of transfer contacts with universities. It could be assumed that similar technological orientation of partners facilitates collaboration or even makes a partnership likely. On the other side it is plausible to assume that more radical type of innovations result from collaboration partners with rather different knowledge bases. The empirical investigations so far are not conclusive in this sense. Nooteboom et al. (2007) state that the relationship between the technology orientation of RD partners and explorative research results in terms of patents follows an inverted U-shape. This means that there exists an “optimal” distance in technology orientation of research partners that is different from having identical knowledge bases and different from having very different knowledge bases. However, Cantner and Meder (2007) do not find an inverted U-shape effect of technology proximity when they are looking at the cooperation propensity. It is indicated that RD partners with very different knowledge bases are unlikely to agree on a partnership at all. Both studies measure technology proximity with patents and do not (to a lesser extent) take into account different dimensions of proximity, like geographical, institutional, and social ones.Footnote 3 Following Boschma (2005) and Boschma and ter Wal (2007) or Breschi and Lissoni (2006) it is likely that the effect of the observed technology proximity is influenced by social or institutional factors. This could mean that, e.g., institutional proximity may compensate the lack of technology proximity in order to agree on a partnership. It is likely that firms are closer in institutional proximity with one another than firms with universities. Hence, firms may have also institutional incentives to collaborate. Institutional differences maybe an obstacle for collaborations between firms and universities. Given those institutional peculiarities it is interesting to look at the importance of technology proximity for knowledge and technology transfer activities between firms and universities.

On an industry level we know that there are sectoral patterns of technology activities (see Pavitt 1984; Verspagen et al. 1994; Schmoch et al. 2003; Broekel 2007) leading to differences in importance of public knowledge for firms’ innovative behaviors. Especially for industries with a strong technology base, like the pharmaceutical industry, access to university research became increasingly important (see Powell et al. 1996; in the case of biotechnology).

On a firm level we have got a good understanding about the characteristics of the transfer process but limited knowledge about the importance of technology proximity for the intensity of transfer activities. We know about Switzerland that about 28% of firms with more than 5 employees have transfer contactsFootnote 4 with universities (see Arvanitis et al. 2007). Large firms and firms in the high-tech sector are significantly more likely to have transfer activities compared to smaller firms and firms in any other sector. Informal, personal contacts and KTT through graduates or the education activities of the universities are the most important forms of KTT in Switzerland (see Arvanitis et al. 2007). Similar studies for other countries and regions also emphasis the importance of human capital and more informal transfer forms (see OECD 2002; Blume and Fromm 2000; Lessmann and Rosner 2004; Salter et al. 2000; Arundel et al. 1995). Furthermore, we know that especially through publications, patent/licenses, and spin-offs university knowledge flows into the entrepreneurial world (see Kaufmann and Tödling 2001; for the importance of transfer offices). Access to human capital (see Audretsch and Stephan 1996; Stephan 1996) or problem solving capabilities (tacit knowledge), access to new research or development of new products are among the important motivations for transfer activities (see Schartinger 2000; Hall 2004). Important hindering factors are related to “firm deficiencies” (e.g., firm’s questions being not interesting for science institutions or lack of interest for scientific projects) but also to “deficiencies on part of the science institutions”. Similar results are found for Austria (see Schibany et al. 1999). More concretely, firms observed that it is sometimes difficult to commercialize R&D results coming from transfer activities and they also noted that to some extent the R&D orientation of public research institution is not interesting for firms (see Arvanitis et al. 2007). Despite of the mentioned impediments many transfer projects lead to successful market products. It was found that on average innovation and firm performance are positively related (see Arvanitis et al. 2008).

With the study at hand we will combine the knowledge about KTT and the technology proximity between the actors in order to identify unused transfer potentials and improve the knowledge base for policy making.Footnote 5

3 Empirical model and hypotheses

Whether a certain type of information or knowledge is important for the innovation activities of a firm depends very often on its knowledge base. Cohen and Levinthal (1989) called the ability to make use of knowledge from other institutions or firms, the absorptive capacity of a firm. The absorptive capacity is quite often approximated through R&D activities or the skill-level of the employees. We learnt from broad empirical studies that the absorptive capacity (measured by the skill-level of employees or R&D activities) is an important determinant for KTT activities (see Arvanitis et al. 2007; for Switzerland). Laursen and Salter (2004) investigated for the UK the types of firms that use universities as a source for innovation. They found also that variables related to the absorptive capacity of a firm such as R&D intensity and long-term R&D show a positive correlation with KTT activities. However, the absorptive capacity is measured in a very general way (skill-level, R&D activities). In order to choose co-operation partners we need to know more concretely the technology orientation of a firm, since a high skill-level can be found in banks as well as in pharmaceutical companies—nevertheless there is no reason to assume that they have a higher probability to co-operate in R&D, since their technology base is too different.

Firms are not anymore the sole actors in their innovation processes (see Malerba 2007). Research co-operations or informal contacts with universities, suppliers, or customers essentially modified the innovation behavior of firms. The partner choice or their perception of what might be an interesting research partner is directed by the technology base of a firm, their working routines, or their quest for new application areas for existing knowledge or technology (see Dosi 1988). The technology proximity between partners is one important driver for collaborations (see Cantner and Meder 2007). Only in rare cases firms seek collaborations in order to “radically” change their technology base, like it was the case with the rise of molecular biology (biotechnology) in pharmaceuticals; from an ex-post point of view the (chemical based) pharmaceutical companies enlarged their knowledge base rather than substituted it. The technology base of a firm is defined as cumulated knowledge, learning, or capabilities from past experiences and in this study it is expressed through patent fields (technology fields).Footnote 6

It is understandable that firms try to continue their work in the same technology field applying similar working routines. It is also an empirical fact that firms try to diversify their external linkages, not only between different types of knowledge partners, e.g., suppliers, customers, and universities, but also within one type of partner (e.g., universities) (see Arvanitis et al. 2007). What might drive such a behavior? Firstly, they can create a greater amount of “incoming spillovers” (see Shapiro and Willig 1990; Greenlee and Cassiman 1999) in order to modify their knowledge base and to update their expertise and to enlarge their research networks. Secondly, such contacts make it easier to recruit graduates or researchers and private R&D activities increase the probability that highly skilled workers stay in the geographical area (see Sumell et al. 2009). Thirdly, contacts with different university institutes help firms to “escape” from knowledge lock-in. Fourthly, funding schemes force firms to collaborate with (different) universities (that is the case in SwitzerlandFootnote 7).

Against this background we want to test the following four hypotheses in which we distinguish between smaller and larger firms.

3.1 Hypotheses

H1

Technology proximity between universities and private enterprises increases the probability of transfer activities and makes it more likely to have more than one university link.

This hypothesis emphasizes the importance of a firm’s absorptive capacity (Cohen and Levinthal 1989; Nooteboom et al. 2007; Schmidt 2008). It should be easier for them to assess the relevance of university research in familiar areas. In case that several universities are doing research in those fields, it is likely that firms contact more than one university. This hypothesis emphasizes the exploitative nature of transfer activities rather than their explorative character (see March 1991). To elaborate on this hypothesis we will look at technology fields that are important in both sectors and estimate the probability of transfer activities (see Eq. 1):

$$ \it{intense} _{\it{i}} = \beta_{1} techfield\_hp_{i} + \beta_{2} pat_{i} + \beta_{3} educ_{i} + \beta_{4} foreign_{i} + \beta_{5} size + \beta_{6} dind1 + \cdots + \beta_{30} dind25 + \varepsilon_{i} $$
(1)

Intense i measures on the firm-level the number of transfer contacts with different universities (see Table 1), e.g., if a firm has transfer activities with two universities, ‘intense’ receives the value 2. Techfield_hp i identifies technology fields (patent classes) that are important for both, the university sector as well as the firm sector, so called “high potentials”. ‘Important’ means that a technology field (patent class) is ranked within the 20 most important technology fields sorted by the number of patent class inscriptions for all firms and all universities respectively. Pat i controls for the patent activities of firms (0/1). Educ i controls for the share of tertiary-level vocational educated employees within a firm. Foreign i tells us whether a firm is foreign owned (0/1). Size i measures the number of employees (log of full-time equivalents) and dind1 to dind25 controls (on a two-digit level) for the industry affiliation of a firm (see Table 2). Following H1 it is assumed that intense i is positively correlated with techfield_hp i and also positively correlated with our further proxy for the absorptive capacity educ i . Significant positive signs are also expected for our controls for pat i , foreign i , and size i . For the description of the dependent and independent variables see Tables 1 and 2.

Table 1 Dependent variable
Full size table
Table 2 Independent variables
Full size table
Table 3 Composition of the data set—number of observations according to industries
Full size table
Table 4 Technological fields of universities/science institutions
Full size table
Table 5 Regression results “all firms” (dependent variable “intense”)
Full size table
Table 6 Regression results “firms with less than 500 employees” (dependent variable “intense”)
Full size table
Table 7 Regression results “firms with less than 300 employees” (dependent variable “intense”)
Full size table
Table 8 Significant and not significant technological fields—all firms
Full size table
Table 9 Significant and not significant technological fields—firms with less than 500 employees
Full size table
Table 10 Significant and not significant technological fields—firms with less than 300 employees
Full size table

H2

There are no transfer activities in technology fields that are not important for both private enterprise, and universities.

This hypothesis addresses the ‘possible’ case where firms and universities join in order to develop a new technology. “New” means that they do not have research experiences in related technology fields. Collaborations under such conditions are intuitively and empirically (as the results show) very unlikely. However, this hypothesis completes the four possibilities of technology related transfer motivations, i.e., frequent technology activities on both the firm and university side; frequent technology activities on the firm side but not on side of the university; frequent technology activities on the university side but not on the firm side; and infrequent technology activities on both firm and university side.

Following the concept of absorptive capacity and the resource-based view perspective, it is very unlikely to see transfer activities in technology fields where both partners are not experts. Thus, we would assume that technology transfer takes place only occasionally or does not take place in those technology fields at all. In the first case we would expect an insignificant sign of the coefficient and in the latter case a significant negative sign for techfield_lp i is expected (see Eq. 2).

$$ \it{intense}_{\it{i}} = \beta_{1} techfield\_lp_{i} + \beta_{2} pat_{i} + \beta_{3} educ_{i} + \beta_{4} foreign_{i} + \beta_{5} size + \beta_{6} dind1 + \cdots + \beta_{30} dind25 + \varepsilon_{i} $$
(2)

Equation 2 differs from Eq. 1 only in terms of the proxy techfield_lp i . Techfield_lp i identifies technology fields (patent classes) that are not important for both, the university sector as well as the firm sector, so called “low potentials”. ‘Not important’ means that a technology field (patent class) is ranked within the 20 least important technology fields sorted by the number of patent class inscriptions for all firms and all universities respectively.

H3

Firms do not have transfer activities with universities in technology fields that are frequently researched by private enterprises and not frequently researched at universities.

It is very unlikely that firms have transfer activities with universities in technology fields that are unimportant in the academic world and thus not well researched at universities (see Eq. 3). As a consequence we do not expect significant transfer results in those technology fields (techfield_ls i ).

$$ \it{intense} _{\it{i}} = \beta_{1} techfield\_ls_{i} + \beta_{2} pat_{i} + \beta_{3} educ_{i} + \beta_{4} foreign_{i} + \beta_{5} size + \beta_{6} dind1 + \cdots + \beta_{30} dind25 + \varepsilon_{i} $$
(3)

Equation 3 differs from Eq. 2 only in terms of the proxy for techfield_ls i . Techfield_ls i identifies technology fields (patent classes) that are important for the firm sector and not important for the university sector, so called “lone stars”. ‘Not important’ for the university sector means that a technology field is ranked within the 20 least important technology fields sorted by the number of “hits”Footnote 8 for technology fields (see also section on data issues). ‘Important’ for the firm sector means that a technology field (patent class) is ranked within the 20 most important technology fields sorted by the number of patent class inscriptions.

H4

Firms do have transfer relations with different partners’ from universities in spite of technological differences, if they want to essentially modify or change their technology orientation.

With this hypothesis we emphasize a more resource (capability)-based view of a firm (see Penrose 1995; Wernerfelt 1984; Barney 1991; Barney et al. 2001) and the more explorative nature of transfer collaborations from the firm’s point of view (see March 1991). From a resource-based perspective firms are heterogeneous as to their resource endowments and capabilities. Thus, the resource endowment is firm-specific and relatively difficult to transfer or to modify. Teece et al. (1997) mention several reasons for the persistence of firm behavior due to the specificity of resource endowment: firms lack the organizational capacity to develop new competences, some assets are not tradable (e.g., tacit knowledge), and needed inputs have to be bought at relatively high prices that reduce possible rents. In this context, the “sticky” character of the resource endowment makes it difficult to change the knowledge base of a firm even when market conditions urge them to do so. Strategies are necessary to change or modify the resource endowment and thus improve firms’ performance (see Wernerfelt 1984; Kor and Mahoney 2004). KTT with universities is one feasible way to modify the knowledge base of firms within their technology path. This is confirmed by firms’ assessments of the main motives for KTT activities with universities. Firms are motivated, firstly, to get better access to human capital (see Geisler and Rubinstein 1989; Schartinger et al. 2001; Onida and Malerba 1989). Secondly, to have better access to knowledge and technology for improving the firm’s knowledge base (see Lee 2000; Santoro and Chakrabarti 2002; Schmoch 2003; Arvanitis et al. 2007). Thirdly, KTT is used to build up new fields of research (see Onida and Malerba 1989; Lee 2000). To work on this hypothesis we look at technology fields that are frequently researched at universities and not frequently researched in private enterprises. It is assumed that firms with such a type of transfer situation want to modify or change their knowledge base. Those firms are likely to be found in markets with intensive R&D related competition, e.g., high-tech industries. Thus we would expect a positive correlation between number of transfer activities and such technology fields (see Eq. 4). A negative correlation would be against this hypothesis.

$$ \text{int} ense_{\text{i}} = \beta_{1} techfield\_np_{i} + \beta_{2} pat_{i} + \beta_{3} educ_{i} + \beta_{4} foreign_{i} + \beta_{5} size + \beta_{6} dind1 + \cdots + \beta_{30} dind25 + \varepsilon_{i} $$
(4)

Equation 4 differs from Eq. 3 only in terms of the proxy for techfield_np i . Techfield_np i identifies technology fields (patent classes) that are not important for the firm sector but important for the university sector, so called “not used potentials”. ‘Not important’ for the firm sector means that a technology field is ranked within the 20 least important technology fields sorted by the number of patent field inscriptions. ‘Important’ for the university sector means that a technology field (patent class) is ranked within the 20 most important technology fields for universities.

3.2 Endogeneity

Since it is possible that the technology profile of a firm is determined by its past transfer activities, there is a potential endogeneity of the technology fields and the intensity of knowledge transfer. However, following Dosi (1982, 1988) we see that firms follow a technology trajectory and external knowledge is only considered if it lies within this technology path. Innovation is an incremental path dependent process. Therefore it is unlikely that the number of transfer contacts with universities radically change the existing knowledge base of firms. Furthermore Cantner and Meder (2007) found that past collaboration experiences do not increase the probability of collaborations with different research partners. This indicates that past collaboration activities did not lead to a radical change in the knowledge base, in such a way that it had caused collaboration contacts with other partners. Thus it is plausible to assume that the knowledge base of a firm changes in a more sluggish way than the number of university contacts (our dependent variable) and endogeneity is not an issue.Footnote 9

4 Empirical strategy

In the following we describe the necessary (preparative) steps and estimation procedures in order to estimate our Eqs. 1, 2, 3 and 4 (see above).

  1. (a)

    Firm side: we sorted the technological fields (class level) according to the number of firms’ patent field inscriptions.

  2. (b)

    University side: we sorted the technology fields (class level) according to the number of technology fields assigned to universities (hits).

  3. (c)

    We compared the 20 most important (frequently researched) technology fields on the part of private enterprises with the 20 most important technology fields on the part of the university sector and looked for similarities and dissimilarities. In the same way we investigated the 20 least important technology fields in both the sectors public universities and private enterprises.Footnote 10

  4. (d)

    As a result we could identify four quadrants (see Figs. 1, 2, 3), i.e.,:

    Fig. 1
    figure 1

    Technological fields and the probability to have technology transfer. All firms. Note High potentials—upper right corner (technological fields frequently found in private enterprises and in universities). Low potentials—lower left corner (technological fields not frequently found in both private enterprises and universities). Not used potentials—lower right corner (frequently found at universities and not frequently found in private enterprises). Lone stars—upper left corner (frequently found in private enterprises and not frequently found at universities). Frequency refers to the 20 most important (according to counts in the respective technological field) or 20 least important technological fields. ‘a47’, ‘b29’, ‘a01’, etc. symbolize the technological field (see Table 11 in the Appendix for the description). Highlighted symbols identify technology fields with significant relations to the intensity of transfer activities. ± indicates if the direction of the observed relationship

    Full size image
    Fig. 2
    figure 2

    Technological fields and the probability to have technology transfer. Firms with less than 500 employees. Note High potentials—upper right corner (technological fields frequently found in private enterprises and in universities). Low potentials—lower left corner (technological fields not frequently found in both private enterprises and universities). Not used potentials—lower right corner (frequently found at universities and not frequently found in private enterprises). Lone stars—upper left corner (frequently found in private enterprises and not frequently found at universities). Frequency refers to the 20 most important (according to counts in the respective technological field) or 20 least important technological fields. ‘a47’, ‘b29’, ‘a01’, etc. symbolize the technological field (see Table 11 in the Appendix for the description). Highlighted symbols identify technology fields with significant relations to the intensity of transfer activities. ± indicates if the direction of the observed relationship

    Full size image
    Fig. 3
    figure 3

    Technological fields and the probability to have technology transfer. Firms with less than 300 employees. Note High potentials—upper right corner (technological fields frequently found in private enterprises and in universities). Low potentials—lower left corner (technological fields not frequently found in both private enterprises and universities). Not used potentials—lower right corner (frequently found at universities and not frequently found in private enterprises). Lone stars—upper left corner (frequently found in private enterprises and not frequently found at universities). Frequency refers to the 20 most important (according to counts in the respective technological field) or 20 least important technological fields.. ‘a47’, ‘b29’, ‘a01’, etc. symbolize the technological field (see Table 11 in the Appendix for the description). Highlighted symbols identify technology fields with significant relations to the intensity of transfer activities. ±indicates if the direction of the observed relationship

    Full size image

    • ‘high potentials’: technology fields frequently found in private enterprises and at universities;

    • ‘low potentials’: technology fields not frequently found in private enterprises and at universities;

    • ‘not used potentials’: technology fields frequently found at universities and not frequently found in private enterprises;

    • ‘lone stars’: technology fields frequently found in private enterprises and not frequently found at universities;

  5. (e)

    High potentials, low potentials, not used potentials, and lone stars could be identified for three different size-related groups of private enterprises, i.e., all firms, firms with less than 500 employees, and for firms with less than 300 employees.Footnote 11

  6. (f)

    We estimated our Eqs. (1, 2, 3, 4; see above) in order to identify if the technological orientation of a firm has an impact on the propensity and intensity (diversification) of transfer activities with universities.Footnote 12 The number of transfer contacts with different universities/research institutions is the dependent variable (see Table 1). In case a firm does not have transfer activities we assigned a zero. This means that we inflated zeros which suggests a zero inflated estimator for count data. Using STATA software we applied the “zinb” (zero inflated negative binomial) procedure with heteroscedasticity robust standard errors. All estimations passed the “voung test” for the zero inflated negative binomial estimators. The first stage was estimated with two instruments, i.e., “quest” and “info” (see list of independent variables; Table 2). “Quest” (difficulties to get information about research activities at universities) and “info” (firms’ R&D questions are not interesting for universities) are chosen, since they are significant impediments for technology transfer activities with universities. Firms confronted with such obstacles are unlikely to have transfer activities at all and as a consequence we would not observe the number of linkages for those firms.Footnote 13

    In addition to the number of patent field inscriptions on a class level we controlled for patent activities (pat) of firms. Furthermore we control for the education level of the employees (educ), foreign ownership (foreign), firm size (size), and sector affiliation of the firm [25 industry dummies (two-digit)].

  1. (g)

    We added the information of significant technology fields to our quadrants by highlighting the respective technological classification (see Figs. 1, 2, 3).

Table 11 Patent classes found for Swiss firms
Full size table

5 Data

For this study we used three data sources. Firstly, and in co-operation with NetBreeze,Footnote 14 we assigned technology fields to R&D active Swiss firms and Swiss universities based on patent statistics (class level).Footnote 15

On the firm side, we used the information on “esp@cenet (patent application and granted patents around the world—www.espacenet.com). We assigned technology fields according to the patent classificationFootnote 16 (class level) to single firms. Thus we only assigned technology fields to firms with patent activities (920 firms). R&D active firms without patent activities or non R&D active firms had no technological assignment. We did not assign the patent fields manually; instead we used a software program developed by NetBreeze.Footnote 17 Technology fields were assigned on the subclass level. Information on the subclass level was aggregated on the class level, and the section level following the official patent classification. The estimations were made on the class level. On the section level we have 8 different sections, and on the class level we found patent inscription of Swiss firms on 109 different classes.Footnote 18 It is possible that one patent is assigned to different patent classes (technology fields). We screened 5,693 firms (Swiss Innovation Panel; 18 manufacturing industries, construction, and selected services) and found 34,048 patents (1904–2008Footnote 19; see Table 3). The 34,048 patents were assigned to 68,533 patent fields.Footnote 20

On the university side, we also assigned technology fields to science institutions of Swiss universities (ETH Zurich (including research institutes), EPF Lausanne, cantonal universities, and universities of applied sciences). Since patent activities at Swiss universities are still rather decent and they do not represent their comprehensive research activities,Footnote 21 we used the information on their websites about their research activities (websites were scanned in spring 2008). More concretely, we used a classifier that allow us to assign patent classifications to universities’ research activities based on “terms” that could be found on the respective websites. The parameters of the classifier were developed and trained based on 150,000 patent description (see Lang 2008).Footnote 22 The results of the technology field assignment (on the section level) to universities are found in Table 4. In Table 4 we see the number of websites searched for the universities respectively. Furthermore it shows the number of “hits” [classified websites (documents)] and how the “hits” could be allocated to patent sections. Thus “hits” are an absolute measure, indicating how often a patent classification could have been assigned to websites within the server domain of a university. In sum 62,742 websites have been searched and 40,193 websites could have been classified (hits).Footnote 23

Secondly, we collected data in the course of a survey among Swiss enterprises about their transfer activities with universities. The survey was conducted in 2005 and mainly covers the time period 2002–2005.Footnote 24 From this survey we used the information about the intensity of transfer activities, the industry affiliation of firms, firm size, patent activities, education level of the employees, and whether a firm is foreign-owned. The survey was based on a (with respect to firm size) disproportionately stratified random sample of firms with at least 5 employees covering all relevant industries of the manufacturing sector, the construction sector and selected service industries (excluding industries with an expected very low propensity of KTT activities such as hotels/catering, retail trade, real estate/leasing, personal services). Answers were received from 2,582 firms, i.e., 45.4% of the firms in the underlying sample. The response rates do not vary much across industries and size classes with a few exceptions (over-representation of wood processing, energy industry and machinery, under-representation of clothing/leather industry). The non-response analysis (based on a follow-up survey of a sample of the non-respondents) did not indicate any serious selectivity bias with respect to the incidence of transfer activities with universities/science institutions. In a further step we matched the information from the survey with the patent information on the firm-level and received a combined data set of 2,132 observations.

6 Results

The main results are presented in Tables 5, 6, 7 and Figs. 1, 2, 3.Footnote 25 Only if the coefficient of the technology field is significant the estimation results are presented in Tables 5, 6, and 7. The overlap of technology fields between private enterprises and universities is considerable. Depending on the firm-size between 12 and 14 (out of 20) technology fields are considered to be important for private enterprises and universities. Furthermore, it was found that the technological activities of universities and the technological orientation of firms are an important factor for knowledge and technology transfer, especially for smaller firms. This fact is mostly neglected in related studies. Tables 8, 9, and 10 show the technological fields with a significant impact on firms’ transfer intensity and the sectors with a relative great patent share in the technology field respectively. We present the results for “all firms”, for “firms with less than 500 employees”, and for “firms with less than 300 employees”.Footnote 26 For “all firms” we see that 10 technology fields are significant (see Table 8), for firms with less than 500 employees we see that 12 technology fields are significant (see Table 9), and for firms with less than 300 employees we see that 7 technology fields are significant (see Table 10).

Combining our findings about the overlap of technology fields with the econometric estimations enables us to answer our hypothesis (see Tables 5, 6, 7; Figs. 1, 2, 3).

With hypothesis 1 (H1) we refer to “high potentials”. Looking at the category “all firms” we see that especially R&D activities in the following fields are found in private enterprises as well as universities (high potentials; see Fig. 1):

  • human necessities, i.e., agriculture (a01), medical or veterinary sciences or hygiene (a61)

  • performing operations/transporting, i.e., physical or chemical processes (b01), hand tools, workshop equipment, manipulators (b23), vehicles in general (b60)

  • chemistry, i.e., organic chemistry (c07), organic macromolecular compounds (c08), biochemistry, microbiology (c12)

  • physics, i.e., measuring (counting), testing (g01), computing, calculating, counting (g06)

  • electricity, i.e., basic electric elements (h01), and electric communication technique (h04)

Comparing these results with the results from the econometric analysis (see Table 5; Fig. 1) we see that private enterprises patenting in the field a01 have a significant greater propensity to conduct technology transfer activities with different universities (greater intensity), while firms that emphasize c12 have a relatively low transfer propensity. Especially firms in the machinery industry and chemical industry as well as metal products were filing patents in a01. C12 is mainly researched by firms in the chemical industry. All other fields are not significant.

Constraining our sample to firms with less than 500 employees leads to some important changes (see Table 6; Fig. 2); c12 switches to the category not used potentials and e04 (building—layered materials, layered products in general) is new among the high potentials. Furthermore, firms with less than 500 employees have a greater probability to have intensive transfer activities in three out of 12 high potentials (h04, c08, a01). This indicates that “smaller” firms (<500) make more intensive use of academic research in these technological areas (high potentials). This shows that the concept of “absorptive capacity” is a necessary but clearly not a sufficient condition for transfer activities if we assume that larger firms have a greater absorptive capacity (on average) compared to smaller ones. Smaller firms are likely to have a greater need (lack of internal scope of resources) to collaborate with universities and thus lower absorptive capacity might be compensated through greater need. Firms in the electronic/instruments industry, the electronic industry, and informatics/R&D industry are mainly filing patents in h04. Firms in the chemical industry, machinery and electronic have the greatest number of patent field inscriptions in c08. Like in the category “all firms”, the chemical industry, metal products and machinery are dominant in a01.

If we further constrain our firm sample to firms with less than 300 employees (see Table 7; Fig. 3) we not only find e04 and again c12 among the high potentials, but newly also g02 (optics, making optical elements or apparatus); three out of fourteen technological fields show a significant positive impact on the intensity of transfer activities (a01, c08, g02). These are relatively few compared to firms with less than 500 employees but more than “all firms”. Thus our result that smaller firms have relatively more transfer contacts within the “high potentials” still holds. However, it should be noticed that there is a slight shift in significance; g02 (optics) is only significant in the category “<300”, while h04 (electric communication technique) is only significant in the category “<500”. Only a01 remains significant in all three size categories. Machinery and chemical industry are amongst the dominant industries in c08 and in g02 mainly firms in electronics and machinery industry are filing patents.

With hypothesis 2 (H2) we refer to “low potentials”. Starting again with the category “all firms” we see few patent field inscriptions on both sides of private enterprises and universities, in the following fields (see Fig. 1):

  • Human necessities, i.e., headwear (a42)

  • performing operations/transporting, i.e., generating or transmitting mechanical vibrations (b06)

  • chemistry, i.e., manufacturing of fertilizers (c05), explosives, matches (c06), sugar industry—polysaccharides (c13), skins, hides, pelts, leader (c14), combinatorial technology (c40)

  • Textiles, paper, i.e., robes, cables other than electric (d07)

  • Mechanical engineering, i.e., storing or distributing gases or liquids (f17), steam generation—physical or chemical apparatus (f22)

  • Physics, i.e., instrument details (g12)

Taking into account the econometric analysis (see Table 5; Fig. 1) we see that firms active in these technological fields refrain from transfer activities with universities by trend; for three classifications we observe a negative sign (b06, c05, c40), one is positive (c13) and the rest is insignificant. Firms in the electronic and machinery industry are frequently filing patents in b06 and c05. In c40 and c13 we have only one (firm) observation respectively. This does not provide us with a strong result.

Looking at firms with less than 500 employees we have quite similar results (see Table 6; Fig. 2). The technological fields are identical—only significant signs switch to some extent; a46 (brushware), b06, and c40 are significant negative and c13 and c14 are significant positive. Again, we have very few observations (in brackets) in c13 (1) and c40 (1), but also in c14 (3). Again, machinery (b06) and electronics (a46, b06) are frequently filing patents in these technology fields.

The main results still holds if we restrain our sample to firms with less than 300 employees (see Table 7; Fig. 3). Only one new technological field (b04—centrifugal apparatus and machines for carrying-out physical or chemical processes) can be observed. Also machinery and electronics remain important industries in terms of filing patents in significant technology fields (a46, f17). In sum it is obvious that we do not observe—like expected—transfer activities in “low potentials”.

With hypothesis 3 (H3) we refer to “lone stars”. Starting again with the results for “all firms” we see that private enterprises emphasis in their patent activities a number of technology fields that are not emphasized or less emphasized by universities, like follows (see Fig. 1):

  • Human necessities, i.e., furniture, domestic articles and appliances, coffee mills, spice mills, suction cleaners in general (a47)

  • Performing operations/transporting, i.e., working of plastics (b29), conveying, packing, storing, handling thin or filamentary material (b65), hoisting, lifting, hauling (b66)

  • Chemistry, i.e., dyes, paints, polishes, natural resins, adhesives (c09)

  • Mechanical engineering, i.e., engineering elements/units, measures for producing and maintaining effective functioning of machines or installations, thermal insulation (f16)

  • Physics, i.e., horology (g04)

  • Electricity, i.e., generation, conversion, or distribution of electric power (h02)

Considering the econometric estimations (see Table 5; Fig. 1) we see that private enterprises patenting in these technology fields do not have a tendency for or against technological transfer activities in general; none of these technological fields are significant. This result is quite intuitive if it is considered that universities do few or no research in these technology fields. Thus, firms might have problems to find adequate partners. Which industries are predominantly active in those technological fields? Machinery, other industries, electronics, chemistry, and the watch industry are mainly filing patents in those fields.

Focusing on firms with less than 500 employees the results change slightly (see Table 6; Fig. 2); only one technological field (b65) has a significant negative sign. All other are insignificant and thus confirming the results for “all firms”. Furthermore, b22 (casting, powder metallurgy) is substituting c09. The list of important industries for “lone stars” remains identical to “all firms”.

For firms with less than 300 employees we found fewer technological fields (see Table 7; Fig. 3). h02, g04, and b66 can not be found anymore among this group and b05 (spraying or atomizing in general, applying liquids or other fluent materials to surfaces) is new. The composition of important industries for these technological fields does not change.

With hypothesis 4 (H4) we refer to “not used potentials”. This group and the group of “high potentials” are of special interest for policy makers. Here, universities show considerable research activities but firms seem to be less interested in such research or do not have the absorptive capacity. Referring to “all firms” the following technological fields are classified as “not used potentials” (see Fig. 1):

  • Human necessities, i.e., sports, games, and amusements (a63)

  • Fixed constructions, i.e., building—layered materials, layered products in general (e04)

  • Physics, i.e., optics—making optical elements or apparatus (g02), controlling, regulating (g05), educating, cryptography, display, advertising, seals (g09), information storage (g11)

  • Electricity, i.e., basic electronic circuitry (h03), electric techniques not otherwise provided for (h05)

As expected we have predominantly significant positive or not significant results for “not used potentials” (see Table 5; Fig. 1). This indicates that firms’ do not have comprehensive research activities in these fields but try to build in-house capabilities through transfer activities with universities. This supports the explorative character of these transfer contacts (see March 1991). In case of not significant results or negative significant results firms do not have the absorptive capacity to make use of public research activities or they simply do not want (e.g., because of security reasons) to have transfer activities in such technology fields. Referring to all firms we see only one technology field with a significant negative sign (g11). That means, although universities have considerable research activities in g11, private enterprises do not tend to have transfer activities; secrecy, different (time) priorities, or problems for commercializing results could be reasons for it. Firms from the machinery, metal products, electronics, and electrical engineering business are most frequently filing patents in these technological fields.

Looking at firms with less than 500 employees we see very similar results (see Table 6; Fig. 2). Only e05 is substituted by c12 (biochemistry, beer, spirits, wine, microbiology) and g11 is no longer significant. All other variables remain to be significant positive or not significant. Also in terms of active industries, we do not see considerable differences. Machinery, electrical engineering and electronics are still very important industries. In addition, chemistry and construction (in case of c12, and a63) gain some importance as well.

In the category “firms with less than 300 employees” we still get similar results compared to “less than 500 employees” and “all firms” (see Table 7; Fig. 3). The technological fields are significant positive (h05) or not significant. However, we find fewer technological fields (without c12, e04 and g02) and a63 is no longer significant positive. Again machinery, electronics and electrical engineering are mainly filing patents in these technological fields. Other industries and construction gain some importance.

In the empirical estimations we controlled for the education-level of firms (educ), firm size (size), whether firms have patent activities (pat), and whether firms are foreign owned (foreign). The education-level, firm size, and patent activities are significantly positive correlated with transfer intensity in all estimations (see Tables 5, 6, 7). Foreign owned firms show lower transfer intensity compared to domestic firms for firms with less than 300 employees (see Table 7). In all other estimations being foreign-owned is insignificant. These results clearly indicate that the “absorptive capacity” plays an important role for transfer intensity. Furthermore it is important to control for patent activities of a firm. Otherwise technology field variables would merely indicate whether a firm has patent activities or not.

7 Conclusions

This study tries to map the technology activities of private enterprises and the technology activities of universities in Switzerland in order to detect collaboration potential or knowledge and technology transfer potential between private enterprises and universities. This way we can improve the knowledge base for policy making. For this study we used two data sources. Firstly, and in co-operation with NetBreeze,Footnote 27 we assigned technology fields to R&D active Swiss firms and Swiss universities. Secondly, we collected data in the course of a survey among Swiss enterprises about their transfer activities with universities. We received answers from 2,582 firms, i.e., 45.4% of the firms in the underlying sample.

Looking at the technology proximity between private enterprises and universities we can identify four areas. Firstly, “high potentials” (technology fields frequently found in private enterprises and at universities). Secondly “low potentials” (technology fields not frequently found in private enterprises and at universities). Thirdly, “not used potentials” (technology fields frequently found at universities and not frequently found in private enterprises). Fourthly “lone stars” (technology fields frequently found in private enterprises and not frequently found at universities).

We saw that great technology proximity between universities and private enterprises increases the probability of transfer activities and makes it more likely to have more than one university link. This was observed in several technology fields, like a01 (agriculture), c08 (organic macromolecular compounds), g02 (optics), and h04 (electric communication technique) and especially in smaller firms (less than 500 employees or less than 300 employees). These findings are very much in line with the concept of absorptive capacity and/or a resource based motivation for transfer activities and these alliances have a more knowledge ‘exploiting’ character.

It was also found that there are not transfer activities in technology fields that are not important (not frequently researched) for both private enterprises, and universities. Here, we mainly observed—independent of the size class—not significant or negative significant relationships between the respective technology fields and the probability to have comprehensive transfer contacts. This result is quite coherent, if we think that both sides do not emphasize research in these fields and thus do not accumulate considerable knowledge.

Furthermore, it became obvious that firms do not have transfer activities with universities in technology fields that are frequently researched by private enterprises and not frequently researched at universities. We did not observe significant transfer activities in those fields (one exception) independent of the size class. It is understandable that private enterprises refrain from transfer activities if they feel to have “better” knowledge compared to potential partners at universities.

It was also found that firms want to change or essentially modify their technology orientation with different partners form universities. These findings refer to technology fields in the category “not used potentials”. As expected we saw predominantly significant positive or not significant transfer relationships in those fields. The significant positive technology fields also indicate that private enterprises recognize the relevance of transfer activities to change or essentially modify their knowledge base. This type of transfer contacts emphasizes the more explorative style of the knowledge seeking process from a firm’s perspective. This incorporates a “technology-push” effect from universities to the private sector contributing to the competitiveness of the private transfer partner.

Since we know that transfer activities support the innovativeness and productivity of firms, it is useful to develop policy measures to ease transfer—especially for smaller firms—by taking into account the different functions of private enterprises and universities in the society. From a policy point of view all four fields can be of interest. However, more detailed investigations of the transfer potential of specific technology fields are necessary in order to identify a suboptimal level of transfer activities. The rather rough identification of “high potentials”, “low potentials”, “lone stars”, and “not used potentials” only allows for some general policy remarks. Thus, it seems to be obvious that a lack of transfer activities in some fields of “high potentials” poses a communication/information challenge to transfer policy makers. Firms may not be well informed about research activities in related fields at universities or research goals, time schedules, or the research questions are too different and thus firms refrain from transfer activities. Secrecy may be a further problem, especially in very market related research. A lack of transfer activities in “low potentials” is quite understandable, since there seems to be neither an academic research interest nor any commercial interest. Thus, “low potentials” are of no or low policy interest at least in the short-term. However, “low potentials” can pose long-term strategic challenges, in case the government aims at strengthening the capabilities in such technology fields, if, for instance, promising long-term benefits are likely for the society (e.g., combinatorial technologies). “Lone stars” may have problems to find adequate national academic partners for their research activities, which would pose an information challenge to policy makers or a research strategy challenge to universities, if such technology fields are of academic interest as well. It should be obvious that universities should not take over research goals from the private sector if there is not any academic research interest. “Not used potential” indicates a lack of absorptive capacity or a lack of commercial potential. “Not used potentials” may signalize a communication problem between universities and the private sector or it may indicate that research in those fields is simply of less commercial value at least in the short-term. However, the transfer potential of technology fields should be investigated in more detailed empirical analyses. Furthermore, the study suggests monitoring the technological orientation of firms and universities and comparing their profiles regularly. This would provide policy makers with a sound basis for technology oriented policy initiatives.