Types of knowledge and diversity of business-academia collaborations: implications for measurement and policy

The idea that basic research is the main source of innovation was already advanced in the beginning of the twentieth century, mainly by natural scientists and managers of company labs who were comparing large firms, business sectors and national economies by their research and development (R&D) intensities in an attempt to establish the links between R&D activities and economic performance (Fagerberg et al. 2011; Godin 2008). This reasoning then became a key idea in Bush (1945), a still highly influential report. Bush was the first policy advisor who forcefully explained the fundamental role of scientific research in underpinning economic competitiveness and advocated a new line in policy thinking:

Bush emphasised that the US could no longer rely on application of knowledge discovered abroad but must pay increased attention to discovering this knowledge as basic research is the ultimate source of technological progress. ‘For many years the Government has wisely supported research in the agricultural colleges and the benefits have been great. The time has come when such support should be extended to other fields.’ (ibid.)

Then the classic articles on the ‘simple economics of basic scientific research’ by Nelson (1959) and the ‘allocation of resources for invention’ by Arrow (1962) marked a new beginning. Since then various economics schools have also applied and adapted their own analytical tools and methods to examine various aspects of RTDI processes, and a new paradigm, namely the evolutionary economics of innovation has also ‘evolved’.³

These ideas have gradually led to what is known today as the science-push model of innovation. By the second half of the 1960s, the so-called market-pull model contested that reasoning, portraying demand as the driving force of innovation. An extensive debate has evolved between these two approaches, trying to establish which is more accurate in describing innovation processes, and especially identifying the most important information sources for innovation.⁴ Then both became variants of the linear model of innovation when Kline and Rosenberg (1986) suggested the chain-linked model, stressing the non-linear property of innovation processes, the variety of sources of information, as well as the importance of various feedback loops. This latter one has been extended into the networked model of innovation; more recently called the multi-channel interactive learning model (Caraça et al. 2009).

In sum, both the science-push and the networked (interactive) models of innovation emphasise the role of universities and PROs as important information sources for innovation. The main difference between these approaches is how they portray the other actors: the networked model considers various types of knowledge—besides R&D results produced by academic organisations—and thus highlights not only B-A collaborations, but the significance—in many cases the necessity—of further types of co-operations as well, namely those between innovators, on the one hand, and their suppliers, competitors, users, other business partners, as well as professional associations, on the other.⁵

Mainstream economics⁶ depicts actors as rational agents facing known and calculable risks and driven by the aspiration to make optimal decisions. In contrast, evolutionary economics of innovation posits that uncertainty is an inherent feature of innovation processes and optimisation, therefore, is excluded on theoretical grounds. Further, while the availability of information has been a focal question in mainstream economics for decades, a major lesson of the evolutionary account of innovation is that firms’ performance is determined by their accumulated knowledge—both codified and tacit—and skills, as well as learning capabilities. Information can be obtained via normal market transactions, and thus mainstream economics can readily treat information as a special good.⁷ In contrast, knowledge cannot be bought and used instantaneously—and that applies a fortiori to the types of knowledge required for innovation (how to exploit readily available pieces of information in a new way, e.g. by combining information on different subject matters, how to utilise experience and skills accumulated through previous search processes, and how to assemble these various types of knowledge). One must go through a learning process to acquire knowledge and skills, and it is not only time-consuming, but the costs of trial and error need to be incurred as well. Hence, the uncertain, cumulative and path-dependent nature of innovation is reinforced. Cumulativeness, path-dependence and learning lead to heterogeneity both at micro and meso levels (Castellaci 2008; Dosi 1988; Dosi et al. (eds) 1988; Fagerberg et al. (eds) 2005; Hall and Rosenberg (eds) 2010; Malerba 2002; Pavitt 1984; Peneder 2010).

Evolutionary economics of innovation does not focus exclusively on R&D. This school identifies various types and forms of knowledge, all relevant for innovation. In particular, the importance of tacit knowledge is stressed, besides codified knowledge. Practical knowledge—acquired, developed, revised and transmitted when performing various tasks—is obviously of crucial importance for the innovation process. Hence, scientific knowledge is far from being the only, or most important, type of knowledge required for a successful introduction of new products, processes, services or organisational and market innovations. As for the sources of knowledge, R&D is clearly among the vital ones (both for codified and tacit knowledge). Besides in-house R&D projects, however, results of other R&D projects are also widely exploited for innovation process: extramural projects conducted in the same or other sectors, at public or private research establishments, home or abroad. Further, a number of other sources of knowledge are also of significance for innovations, such as design, scaling up, testing, tooling-up, trouble shooting and other engineering activities, as well as ideas from suppliers, users and NGOs (including patient groups), inventors’ ideas and practical experiments (Hirsch-Kreinsen et al. (eds) 2005; Klevorick et al. 1995; Lundvall (ed) 1992; Lundvall and Borrás 1999; von Hippel 1988), as well as interactions among engineers, designers, artists and other creative ‘geeks’. In general, all sorts of trial and error processes, learning by doing, using, interacting and comparing contribute to knowledge generation. Further, knowledge embodied in advanced materials and other inputs, as well as in equipment and software is also utilised by innovative firms. All rounds of the Community Innovation Survey clearly and consistently show that firms regard a wide variety of sources of information as highly important to innovation.⁸

In brief, policy implications of evolutionary economics can be derived from two closely related claims. First, the success of firms is largely determined by their abilities to exploit all the above types of knowledge, coming from both R&D efforts and other activities. Second, knowledge generation, diffusion and exploitation takes place in, and is fostered by, networks, clusters and other forms of co-operation and communications. The quality and frequency of these interactions are largely determined by the institutions—the ‘rules of the game’—and other properties of a given innovation system, in which they take place.⁹ STI policies, therefore, should aim at strengthening the respective—sectoral, regional or national—innovation system and improving its performance by tackling systemic failures hampering the production, circulation and utilisation of any type of knowledge required for successful innovation (Dodgson et al. 2011; Edquist 2011; Foray (ed) 2009; Freeman 1994; Lundvall and Borrás 1999; OECD 1998; Smith 2000). Concerning B-A collaborations, deliberate policy efforts are needed to promote its various types, serving knowledge-intensive activities of all firms, regardless whether the aim is a radical innovation, an incremental one or ‘just’ solving an important technical problem.

The business sector is the most important research performer at an aggregate level in the EU27 countries both in terms of its share in GERD and employment, followed by the higher education and the government sectors (Table 1). The share of the private non-profit sector is around 1 % by either measure, and thus it is not analysed here.

The number of researchers (counted as full-time equivalent (FTE)) employed by businesses has increased from 500,377 in 2000 to 763,993 by 2012 in the EU27 countries, and thus remained the largest employers of researchers.¹¹ This pattern is not repeated at a country level: in 2012, businesses were the largest employers of (FTE) researchers in 12 EU countries, while the higher education sector took the lead in 11 EU countries, and the government sector in a single country. The share of business enterprise researchers in the EU27 total was 46.5 % in 2012 and varied between 15.2 % (LV) and 62.3 % (AT) in the national total at a country level. This ratio was above 50 % in 11 EU countries and under 30 % in 8 ones (Fig. 1). Business R&D expenditures (BERD) have increased from €111,181.1 m in 2000 to €145,652.6 m in 2012 (PPS at 2005 prices), that is, by 31 %. The share of GERD performed by the business enterprise sector was 62.4 % in 2012. At a country level, this ratio was ranging between 22.6 % (LV) and 77.2 % (SI) in 2012, with 6 countries above 67 %, 7 relatively close to the EU27 average ratio, that is, between 57 and 67 %, 6 between 40 and 57 %, and another 5 below 40 %¹² (Fig. 2).

Higher education (HE) organisations were the second largest employers with 412,473 FTE researchers in 2000 at the EU27 level and 660,040 in 2012, that is, 40.2 % of the EU27 total. Again, there is a great variety at a national level: the share of HE FTE researchers in the national total was ranging between 24.9 % (HU) and 66.8 % (LV) in 2012. It was close to the EU27 aggregate figure, i.e. stood between 37 and 43 % in 4 countries, below 37 % in 11 countries, in the range of 43–60 % in 5 countries, and above 60 % in 4 countries (Fig. 1). The total EU27 R&D expenditures in the HE sector (HERD) have increased by 51 % in absolute terms: from €36,933.9 m in 2000 to €55,776.0 m in 2012 (PPS at 2005 prices). The share of GERD performed by the HE sector is significantly lower: it fluctuated between 21.2 and 23.9 % in 2000–2012 at the aggregate level of 27 EU countries. The HERD/GERD ratio varied between 8.0 % (BG) and 53.7 % (LT) in 2012 at a country level. In 5 countries, it was in the range of 8–21 %, in another 6 close to the EU27 ratio (between 21 and 27 %), in 11 ones between 27–40 %, and in 2 ones above 50 % (Fig. 2).

At an aggregate level the government sector was the No. 3 employer with 166,791 FTE researchers in 2000, and 200,045 in 2012, that is, less than one third of the HE figures. The share of this sector was 12.2 % of the EU27 total in 2012, but the variation at the country level is significant in this case, too: the weight of the government sector is ranging between 3.0 % (UK) and 47.3 % (BG). This share is below 7 % in 7 countries, between 11 and 12 % (that is, very close to the EU27 aggregate) in 4 countries, between 16 and 21 % in 11 countries, and above 39 % in 2 countries (Fig. 1). The share of GERD performed by the government sector was in line with its share in employment, that is, 12.9 % in 2012 at the aggregate EU27 level. At the country level, this share varied from 2.2 % (DK) to 47.6 % (RO) in 2012: it was below 10 % in 10 countries, between 10 and 15 % (i.e. close to the EU27 ratio) in 6 countries, between 15 and 30 % in 8 countries, and close to 50 % in RO (Fig. 2).

As a ‘broad brush’ observation, the more advanced an economy is, the higher the weight of businesses in employing researchers and performing R&D (Figs. 1 and 2). Of course, more detailed analyses would be needed to draw sound theoretical and policy conclusions, taking into account the context of each country, e.g. historical legacies, organisational and institutional factors, as well as recent sweeping changes in the case of the so-called new EU member states. As for the former, the UK clearly shows that context does matter: the share of researchers employed by businesses is lower than expected (35.8 % in 2012), significantly ‘outweighed’ by the higher education sector (59.6 %), but the ‘anticipated’ ratio can be observed as for the share of these two sectors in performing R&D: 63.4 vs. 26.5 %.

As for the latter, businesses in Hungary and Slovenia have achieved a high share in employing researchers, due to a fairly radical restructuring: in Hungary from 27.1 % in 2000 to 55.5 in 2012, and from 31.8 to 53.1 % in Slovenia. These changes were less profound in the Czech Republic: 39.9 vs. 46.6 %. As for performing R&D, the most noteworthy changes occurred in Bulgaria, where the share of businesses increased from 21.4 % in 2000 to 60.5 % in 2012; and in Estonia: from 22.5 to 57.4 %. Again, a detailed analysis would be required to identify if genuine structural shifts or reclassification of research performing organisations have caused these drastic changes. Yet, even this short overview is sufficient to indicate that country differences do matter even when one considers a group of countries characterised by broadly similar features.

BERD is mainly financed by businesses’ own resources: this share was fluctuating in a narrow range of 81.3–83.2 % in 2000–2011. From a different angle, the bulk of business R&D funds is devoted to business R&D activities: 94.8–95.7 % in the same period. It is worth stressing, though, that in some countries businesses fund research activities both at HE institutes and in the government sector (publicly financed R&D institutes, or PROs) to a noteworthy extent.

While at the EU27 level 6.3–6.8 % of HERD was financed by businesses in 2000–2012, at a country level, one can find much more variation both in terms of the ratio of business sources and dynamics (Fig. 3). The share of business sources in funding HERD was around or above 10 % in 6 countries, around 7–8 % in 4 countries, 3–5 % in 8 countries and less than 3 % in 6 ones in 2012. In some countries, this share decreased significantly, e.g. from 30.8 % in 2000 to 16.0 % in 2012 (BG), or from 27.1 % to 5.4 (LV). Overall, this share grew in 10 countries by 2012, among these by around 4 percentage points in Hungary and Slovenia, and by 2.3 percentage points in Germany from an already high level, and declined in 11 countries (missing data for 3 countries).

The share of business sources in funding HERD is higher than the aggregate EU27 figure in 10 countries, of which 5 are new member states and 1 is a less developed Southern European country. The relatively high ratio of business funding in these countries might be attributed to the low amount of HERD in absolute terms: a few projects commissioned by firms, with relatively low budgets by international standards, can lead to a high weight of business funding in HERD.

The share of business sources in funding Government Intramural Expenditure on R&D (GOVERD) was 5.7–8.9 % at an aggregate EU27 level in 2000–2012. As for the member states, this ratio was in the range of 1.1 % (PT [2011]) and 17.3 % (RO) in 2012. It was above 10 % in 8 countries, 7–9 % in 4 countries, 4–6 % in 8 countries and 1–3 % in 4 ones in 2012 (or 2011) (Fig. 4). This ratio increased in 9 countries (by 7 percentage points in DE, 3–4 percentage points in 3 countries, around 2 percentage points in 2 countries, around 1 percentage point in 3 countries and just 0.4 point in 1 country) and decreased in 13 cases (by 6–12 percentage points in 4 countries, by 3–5 points in another 4 and by 1–2.5 points in the remaining 3 countries).

The share of GOVERD financed by businesses is higher in 10 member states than the EU27 figure, and 6 of these are new members. The low volume of GOVERD in these countries, most likely, is an important factor in explaining the high value of this ratio.

The quality of co-operation among the NIS players can be characterised by firms’ assessments as to the importance of sources of information for their innovation activities. In all countries participating in CIS2008¹³ and CIS2010, the largest share of firms regards their own enterprise or enterprise group as a highly important source of information for innovation, and other firms—suppliers, customers, competitors and commercial labs—are also highly appreciated by a large part of firms. Thus Fig. 5 only presents these business-type sources of information. The other sources—which can be called ‘scientific’ ones in a somewhat simplified way—are depicted on Fig. 6. These are ‘highly important sources of information’ for a significantly lower share of innovative firms. In most countries conferences, trade fairs and exhibitions ranked first in this group, scientific journals and trade/technical publications came second, followed by universities and public research institutes. Universities were among the top 3 in seven countries in 2008–2010: they came second in Estonia, Finland and Hungary, while third in Belgium, the Czech Republic, Poland and Spain. PROs were ranked No. 2 in Spain, while in all other countries No. 5, except Poland (No. 3) and the Czech Republic (No. 4).

Data on innovation co-operation partners are only available at the EU27 level for 2002–2004 and 2008–2010. In both periods, 25.5 % of innovative enterprises reported being ‘engaged in any type of co-operation’. Overall, a larger share of innovative firms have co-operated with business partners (other enterprises in their group, suppliers, clients, competitors and commercial labs) than with higher education institutes (HEIs) or publicly financed research organisations (PROs) (Table 2). Suppliers of equipment, materials, components or software were mentioned by the highest share of innovative firms as co-operation partners in both periods (16.5 and 15.2 %, respectively). HEIs have become partners for a higher share of firms by 2008–2010 (8.8 vs. 10.8 %), and thus ‘overtaken’ three types of business partners (out of five), including other enterprises within the enterprise group. PROs have remained the least frequently mentioned co-operation partners, but 2008–2010 saw a slight increase.

There are significant differences among EU members in this respect, too, and thus Fig. 7 presents country-level data. Almost in all countries the highest share of innovative firms report co-operation with suppliers, with the exception of Finland and the UK (where customers are the top co-operation partners), and Germany (HEIs). It is noteworthy that 23–35 % of innovative firms co-operate with suppliers in 15 countries, and 16 % of firms do so in another 2 countries, while the aggregate EU27 figure is 15.2 %. Similarly, 21–30 % of innovative firms co-operate with clients or customers in 14 countries, and 13–15 % of firms do so in another 3 countries, while the aggregate EU27 figure is 12.6 %. As for competitors or other enterprises in the sector, 8–31 % of innovative firms in 14 countries co-operate with them, as opposed to the ratio of 6.7 % for the EU27 countries. Finally, 12–26 % of innovative firms in 16 countries co-operate with other enterprises within the enterprise group, which is well above the EU27 figure (9.3 %). In short, innovation co-operation with ‘business’ partners are much more widespread in a large number of countries than suggested by the aggregate EU27 data.

It is also interesting to note that there is no clear division between the more and the less advanced member states (or the ones belonging to various groups defined using the so-called Summary Innovation Index). For example, Lithuania, Slovakia and Slovenia are next to Finland, Sweden and Denmark on Fig. 7, while Bulgaria and Romania are in the same group as Germany, Spain and the UK.¹⁴ In other words, the higher occurrence of innovation co-operation does not necessarily mean a better innovation—and ultimately economic—performance. Clearly, there are many other factors influencing innovation performance—and much more determining economic one. As for the former, the quality of co-operation is among those factors. Thus, when analysing B-A co-operation, it is also important to note which co-operation method is the most valuable one for firms.

In most EU countries, co-operation with suppliers, customers and other enterprises within the enterprise group is mentioned by a relatively large portion of firms as the most valuable method (Fig. 8). Yet, in eight countries, co-operation with higher education institutes are among the top three methods: HEIs were ranked first in Germany (6.6 % of the innovative firms mentioned this method as the most valuable for innovation, and only 4.2 % perceived suppliers as the most valuable innovation co-operation partners), second in Hungary (8.5 %), while third in Austria (8.0 %), Belgium (3.9 %), the Czech Republic (4.2 %), Romania (1.7 %), Slovenia (21.3 %) and Spain (3.6 %).¹⁵ PROs are assessed far less favourably: besides Spain, where they are ranked No. 2 (4.3 %), nowhere else they are among the top three.

Finally, Figs. 9 and 10 zoom into innovation co-operation with HEIs and PROs, respectively. Finland is way ahead of other countries in both cases, and although there are no data as to how Finnish firms assess the various types of innovation co-operation partners, it is highly likely that they find co-operation with both HEIs and PROs useful, otherwise they would be engaged in these B-A collaborations to a lesser extent. It is also worth noting that a high share of innovative Finnish firms tends to co-operate: Finland is the only country where any of the seven types of innovation co-operation partners is mentioned at least by 22 % of innovative firms (and on top of that, five types are mentioned by around or well above 30 %) (Fig. 7).

It has been a recurring theme of various reports and policy documents that the intensity, frequency and quality of B-A co-operation in Hungary have been significantly below the desired level (Arnold et al. 2007; Borsi 2005; Havas 2004, 2009, 2011; Havas and Nyiri (eds) 2007; Inzelt 2004; Inzelt et al. 2009; OECD 2008). The SME development strategy of the Ministry for Economy and Transport has stressed that knowledge diffusion between publicly financed research institutes and businesses has been insufficient; directors of PROs have not considered businesses’ interests when defining research themes or assessing researchers’ performance; and researchers have hardly moved between PROs and businesses (GKM 2008, p. 34).

Thus, several Hungarian STI policy measures have been devised with the aim of promoting B-A co-operation, either by making this type of collaboration compulsory, or giving priority to joint project proposals of firms and universities or PROs.¹⁶ These measures seem to have had some positive impacts: various types of statistical data as well as evaluation reports indicate that the frequency of B-A collaboration has increased to a noteworthy extent since 2000. While only 4–5 % of the Hungarian HERD had been financed by firms in 2000–2001, this ratio jumped to 11–13 % in 2002–2006, further increased in 2007–2010, reaching its peak at 15.5 % in 2009, and then dropped to 10–11 % in 2011–2013. As for the composition of GOVERD, the share of business funding started at a high level of 11–13 % in 2000–2001, halved in 2002–2004, exceeded 10 % again in 2005 and was in the range of 9.7–14.3 % in 2007–2013 (Table 3).

Recalling the data presented in Section 3.2, these Hungarian ratios are fairly high compared to the aggregate EU27 figures. As already mentioned with regard to other new EU members, this high ratio of business funding might stem from a few projects commissioned by firms, with a relatively small budget by international standards, given the low level of the Hungarian HERD and GOVERD in absolute terms (€97–243 m, and €106–248 m a year in 2001–2013, respectively, at current prices).

Hungarian universities are the second most important information sources for innovation when the so-called ‘scientific’ sources are considered, moreover, by far with the highest appreciation compared to their counterparts in the other EU countries. As for PROs, they are the least important information sources, just as in most other EU members (Fig. 6).

A significantly higher share of innovative Hungarian firms indicate co-operation with universities and PROs than the EU27 average: 21.4 vs. 10.8 %, and 10.2 vs. 6.1 %¹⁷ (Table 2 and Fig. 7). At a country-level comparison, concerning the frequency of innovation co-operation between businesses and higher education institutes, Hungary was ranked four in 2008–2010—up from her fifth position in 2006–2008—and universities had a higher appreciation only by Slovene firms (21.2 % of them perceived this one as the most valuable co-operation method), that is, Hungarian universities are ranked second in this respect (Fig. 9). Taking the frequency of innovation co-operation between business and PROs, Hungary was ranked sixth (up her sixteenth position in 2006–2008). Again, Slovene firms gave the highest appreciation to PROs (Fig. 10).

Finally, two evaluation reports touch upon B-A collaboration in Hungary. The first one considers the impacts of a single measure, called KKK, Co-operative Research Centres, and concludes that B-A co-operation has contributed to improved competitiveness of firms and also led to the set-up of several spin-off firms (Netwin and Laser Consult 2005). The second one, evaluating the use of the Research and Technological Innovation Fund—the most important domestic fund to support RTDI activities—in 2004–2009 has also asserted that due to various schemes, financed by the Fund, B-A co-operation has strengthened (Ernst & Young and GKI 2010, pp. 7, 89–90).

Interviews conducted in four sectors—automotive industry, pharmaceuticals, telecom equipment manufacturing and software development¹⁸—have confirmed that companies and public R&D units (HEIs and PROs) are driven by fundamentally different incentives and goals to be involved in R&D and innovation activities. Hence, there are inherent hindrances to B-A collaboration. In brief, companies are interested in a relatively wide array of R&D activities (from day-to-day problem solving to long-term strategic research, some of which may require producing advanced scientific and technological knowledge, or even path-breaking new theoretical results), but those should lead to business results (e.g. enhanced productivity, larger market shares, entry to new markets, increased profits). Projects are regularly monitored and assessed, and when necessary, a given project could be substantially reshaped (e.g. in terms of the number of participants, R&D methods applied, budget), or even stopped. Thus, tight project management (meeting deadlines and ‘respecting’ budget constraints) and keeping commercially sensible information secret are of vital importance. In contrast, researchers working for universities and PROs are not simply interested, but even forced to disclose their results as quickly and as widely as possible, given the evaluation criteria applied in the academic world. Further, they are usually less accustomed to tight project management, but noticeable changes have occurred in recent years, due to tighter control exercised by both the domestic and foreign funding agencies.

These systemic hindrances to B-A collaborations—different goals and incentives for academic researchers and businesses—are not a unique feature of the Hungarian innovation system. Several ‘profound differences in the “scientific” and “industrial” cultures’—fairly similar ones to those observed in Hungary—have been highlighted in a presentation by the General Secretary of the European Council of Academies of Applied Sciences, Technologies and Engineering (Lukasik 2013).

Based on the interviews conducted with Hungarian firms, at least three fundamentally different types of business-academia collaboration can be identified.¹⁹ No doubt, other types of co-operations might also be found, and a more detailed, more refined classification could also be devised. This tentative taxonomy considers two major aspects: whether there is any ownership link between the partners, and the main objectives of co-operation.

This tentative taxonomy can—and should—be developed into a more detailed and better-substantiated typology. Depending on the objectives of further analyses, the following aspects can be used when refining it: the objectives, organisational form and duration of co-operation; types of participants (domestic vs. foreign universities and firms); major characteristics of the business participants (size, ownership, specific sectoral/technological/strategic features, etc.)

Even this tentative taxonomy is sufficient to stress that heterogeneous firms are faced with different needs, posses distinctive capabilities, set specific goals and thus pursue different RTDI strategies. Hence, different forms and types of B-A co-operations can be observed, with specific goals and activities. STI policies, however, tend to neglect this diversity, and not only in Hungary. For example, major EU policy documents tend to mention only type 3) B-A collaboration, while type 2) ones seem to be equally relevant in improving firms’ innovation performance and hence competitiveness (see, e.g. EC 2013a, 2013b).