Acknowledge globalisation and innovation trends
Innovation collaboration is increasingly global, for several reasons
The increasing globalisation of knowledge production and innovation activities requires all regions to think beyond their borders. Companies are extending their value chains and markets, recruitment areas and range of innovation partners towards farther reaching locations. While the share of foreign innovation collaborations may be larger in smaller and highly open economies, firms in countries with large domestic markets still seek global partners. There is an increasing share of scientific co-publications with international partners. The share of all publications with an international co-author has tripled from around 7% in 1985 to around 22% in 2007 (Figure 1.1). In terms of patenting, the share of co-patents with inventors in a foreign country has doubled over the last three decades, increasing from 10% in 1980 to 20% in 2008.
Innovation is increasingly multi and inter-disciplinary. Data from “science maps” show the convergence of different scientific fields, such as nanoscience that grew out of the interaction of physics and chemistry. Environmental research is an example of a multidisciplinary field (OECD, 2010b). Innovation is increasingly at the intersection of different technologies and sectors, thus requiring opportunities for such new combinations to arise. For example, many innovations are at the intersection of nanotechnology and biotechnology. Economies of scope can enhance innovation as wider partnerships can create value from diversity, by combining complementary expertise available internationally.
There is a need for greater critical mass in certain fields to compete globally. Knowledge production is characterised by economies of scale, generally requiring international investments and talent. Small regions are often less visible on an international scale. Joining efforts and resources with nearby regions across borders may be necessary to increase the size of the local labour market and the access to innovation resources. Such joint efforts can help the respective regions gain the effective critical mass necessary to become visible internationally, thus attracting foreign firms, investments and personnel. Technology parks and similar initiatives with an international outlook benefit from a wider pool of clients and the cross-border scope also serves international branding efforts. Joint investments and the sharing of resources are increasingly necessary to reach the scale for international excellence. Venture capital (VC) funds work more efficiently when there is a sufficient base of firms in proximity.
Figure 1.1. Scientific publications increasingly involve international collaboration
Source: OECD (2010), Measuring Innovation: A New Perspective, OECD Publishing, Paris, http://dx.doi.org/10.1787/9789264059474-en.
There is a growing need for specialised knowledge as well as both cost and risk-sharing. Firms in regions need to access world-class knowledge and be connected to a wide range of innovation actors. The size of many regions prevents them from offering a full innovation support infrastructure responding to all the specialised needs of regional stakeholders. Innovation advisory services need a degree of specialisation to reach a high level of professionalisation. Moreover, this knowledge specialisation makes innovation processes more risky. Building high-end and targeted research centres, or providing particular S&T equipment, is expensive. Sharing the costs and the risks of such facilities is a way to support future innovations. Accessibility through physical proximity can be an advantage for such joint efforts.
Geographic proximity remains important for the innovation process
The phenomena of agglomeration and clustering (firms, research facilities, skilled workers, etc.) illustrate the persisting relevance of geographic proximity. A broad stream of academic literature has studied the benefits of agglomeration economies in terms of productivity gains.1 According to Rosenthal and Strange (2004), a doubling of the size of urban agglomerations increases productivity between 3% and 8%. Productivity advantages of agglomeration economies have been related to several aspects: i) labour market pooling that gives workers a range of potential employers and the firms access to specialised skills, thus facilitating better labour market matching; ii) variety and specialisation by providers of intermediate goods and services; and iii) knowledge spillovers whereby firms benefit from being near each other because there are areas of special knowledge.2
Innovation activities are highly concentrated in a limited number of regional knowledge hubs. Over 33% of R&D takes place in the top 10% of large OECD regions,3 and 58% of patents are applied for in the top 10% of small OECD regions (OECD, 2013).4 Around one fourth of skilled employment is concentrated in the top 10% of OECD regions.5 The top OECD regions in terms of patenting volume are often responsible for a large share of the national patents, notably in biotechnology and nanotechnology (Figure 1.2).
Figure 1.2. Top patenting regions strong in several technologies
Source: OECD (2010) Measuring Innovation: A New Perspective, OECD Publishing, Paris, http://dx.doi.org/10.1787/9789264059474-en.
The return on innovation-related investments declines with distance. There are many studies that document the phenomenon of the spatial decay of knowledge spillovers (Box 1.1). For example, if R&D investments are made in a particular location, the impact on growth is generally observed to be limited to a certain radius around that investment. The same finding is observed with patent citations, whereby the frequency of citing a given patent does fade with distance. One recent analysis shows that the change in the probability of citing a patent in the same region (metropolitan area) is generally twice as high as citing a domestic patent more generally in Europe and three times higher in North America. Citation probability decays after around 200-250 kilometres in Europe, and a shorter 150 kilometres in North America, after which point distance no long plays a role (Thoma, forthcoming). Therefore, even for codified knowledge, such as a patent, proximity still matters. Some form of tacit knowledge, which comes from inter-personal interaction, is clearly still important to transfer knowledge.6
Box 1.1. Place still matters for innovation: Knowledge spillovers and spatial decay
Several scholars have debated the geographic dimension of knowledge spillovers, as measured by different innovation-related variables: i) sectoral concentration of firms; ii) human capital characteristics; iii) R&D activities; iv) patents and patent citations. All of these studies claim that the geography matters for innovation activities and that the concentration is beneficial for regional development and economic growth (see, for example, Carlino et al.  and Ejermo  for a more detailed discussion on this topic).
Studies have shown the importance of proximity as evidenced by the concentration of jobs and firms. Ellison and Glaser (1997) proposed a dartboard approach across industries and regions and a scoring index to demonstrate that firms and employment are spatially concentrated at a higher degree than a random distribution. This index has been subsequently improved and extended by Ellison and Glaser (1999) and Duranton and Overman (2008). Rosenthal and Strange (2005) analyse the impact of agglomeration of human capital on productivity, finding that proximity matters and that the positive effects of knowledge spillovers driven by the spatial concentration of educated workers decline as distance increases. In the same vein, Arzaghi and Henderson (2008) study the networking effects of the advertising agency industry in Manhattan and they find that those spillovers have a very rapid decay with distance (approximately 750 meters).
The productivity gains of joint R&D projects among G-5 countries in the OECD area have been shown to be geographically bounded, as the productivity gains decline with the distance between R&D partners (Keller, 2002). Orlando (2004) finds that both geographic and technological R&D spillovers are significant and important. Buzard and Carlino (2009) look at the concentration patterns of R&D labs in the United States, finding that geographic clustering of labs is significantly different from random spatial patterns. In addition, they also find a strong positive correlation between geographic clustering of R&D labs and knowledge spillovers as proxied by patent citations.
Jaffe et al. (1993) proposed for the first time to consider patent citations as a paper trail for the existence of geographical knowledge spillovers. They find that patent citations are geographically localised even when controlling for a pre-existing concentration of technologically related activities. Thompson (2006) illustrates that patent citations are geographically concentrated both between and within a country. In Agrawal et al. (2008), patent citation and co-ethnicity data are used to study the impact of spatial and social proximity on knowledge flows. The authors find that both geographical and social proximity have an impact and, in particular, that knowledge flows between inventors fade with distance.
More recently, Lychagin et al. (2010) compare different kinds of R&D spillovers depending on geographic, technology and product-market proximity. They find that local spillovers are significant, showing a gradual decay over space. Kerr and Kominers (2010) use patent citations to measure spillovers in geographical areas and relate them to clusters and shapes of firms. Murata et al. (2011) use micro-level and geolocalised data on patent inventors to analyse knowledge spillovers in different technologies. They find that spillovers are localised for most technologies (95%) and diminish with distance. Carlino et al. (2012) detect patterns of local concentration of R&D clusters and find that patent citations occurring in those clusters are significantly more geographically concentrated than patent citations on average. In addition, they show that R&D labs are most significantly clustered at small spatial scales (a quarter of a mile) and that the significance decays rapidly with distance.
In an econometric study covering all regions in 25 EU countries, Rodriguez-Pose and Crescenzi (2006) try to discriminate between the influence of internal factors and external knowledge and institutional flows on regional economic growth. The empirical results highlight neighbourhood effects: not only is R&D investment within the region important for growth, but R&D investment in nearby regions has impacts on a region’s growth. They also indicate the importance of proximity for the transmission of economically productive knowledge, as spillovers show strong distance decay effects. In the EU-25 context, the study found that only the innovative efforts pursued within a three-hour travel radius have a positive and significant impact on regional growth performance.
Source: Agrawal et al. (2008); Arzaghi and Henderson (2008); Buzard and Carlino (2009); Carlino et al. (2012); Duranton and Overman (2008); Ejermo (2009); Ellison and Glaeser (1999); Ellison and Glaeser (1997); Jaffe et al. (1993); Keller (2002); Kerr and Kominers (2010); Lychagin et al. (2010); Murata et al. (2011); Orlando (2004); Rodriguez-Pose and Crescenzi (2006); Rosenthal and Strange (2005); Thompson (2006).
Evidence on collaborative activities for patenting highlights the importance of geographic proximity, as well as other forms of proximity. For example, collaboration for invention activities is almost 50% in the same region (OECD, 2013). When looking specifically at co-inventions between public and private co-applications, around 40% of those collaborations take place within the same region, even in countries with other strong regions and international collaboration networks, such as Germany and the United States (Figure 1.3). Policies that shape collaboration between the public and private sector are more likely to favour same country collaboration. However, it is likely that other forms of proximity are relevant as well (Box 1.2).
Figure 1.3. Public-private co-patenting collaboration often occurs in the same region
Co-patenting with at least one business and one public applicant over total co-patenting, by location of applicants, 2005-07
Note: A public applicant is a public research organisation or higher education institution.
Source: OECD (2011), OECD Regions at a Glance, OECD Publishing, Paris,
International borders remain important obstacles for the flow of knowledge and other forms of innovation collaboration. Evidence shows that even when regions are physically close and share common areas of technological expertise, there is an additional barrier given the presence of an international border (Box 1.3). In fact, in both North America and Europe, the probability of citing a patent in a neighbouring foreign region is no different than citing one in any foreign region, regardless of distance, showing that the border effects dominate over proximity benefits. Language differences are also significant for patent citations. This implies that there must be a range of associated costs with cross-border collaboration by innovation actors. Therefore efforts to promote regional innovation policy taking into account a cross-border area will need to seek to minimise the costs of the international boundary to better reap the benefits of working together.
Box 1.2. What is meant by the term “proximity” for innovation collaboration?
Geographic proximity is only one of several kinds of proximity that can be relevant for collaboration in innovation. Boschma (2005) has identified five forms of proximity:
- • Cognitive proximity: Actors need cognitive proximity in terms of a shared knowledge base in order to communicate, understand, absorb and process new information successfully. Too little cognitive distance means a lack of sources of novelty. It increases the risk of lock-in or undesirable spillovers to competitors. Too much cognitive distance hampers communication and leads to misunderstanding and limited potentials for interactive learning.
- • Organisational proximity: A certain degree of organisational proximity is needed to control uncertainty and opportunism in knowledge creation within and between organisations. Too little organisational proximity goes along with a lack of control, increasing the danger of opportunism. Too much organisational proximity may be detrimental to interactive learning due to lock-in and a lack of flexibility.
- • Social proximity: Social proximity may stimulate interactive learning due to trust and commitment. Too little social proximity may be harmful for interactive learning and innovation due to a lack of trust and commitment. Too much social proximity may also be detrimental to interactive learning due to lock-in and an underestimated risk of opportunism.
- • Institutional proximity: Institutional proximity is an enabling factor, providing stable conditions for interactive learning to take place effectively. Too much institutional proximity is unfavourable for new ideas and innovations due to institutional lock-in (obstructing awareness of new possibilities) and inertia (impeding the required institutional readjustments). Too little institutional proximity is detrimental to collective action and innovation due to weak formal institutions and a lack of social cohesion and common values.
- • Geographic proximity: This is the spatial or physical distance between economic actors, both in its absolute and relative meaning. Short distances literally bring people together, favour information contacts and facilitate the exchange of tacit knowledge. The larger the distance between agents, the less the intensity of these positive externalities, and the more difficult it becomes to transfer tacit knowledge. This may even be true for the use of, and spread of, codified knowledge. There can also be disadvantages to too much geographic proximity as it can lead to lock-in.
Applying a proximity level analysis, others have documented challenges for creating an integrated cross-border system. Lundquist and Trippl (2013) note three broad concepts of proximity as important for the success of cross-border co-operation among innovation-related actors: physical (geographic), functional and relational proximity. In a study of the cross-border area of Baden (Germany) and Alsace (France), it was a lack of relational proximity (non-tangible dimensions based on degrees of similarity and affinity), and not geographical proximity (accessibility issues) that was the challenge for collaboration (Koschatzky, 2000). According to Maggioni and Uberti (2007), functional distance defined as strong asymmetries in innovation potential and performance limit cross-border knowledge flows between places.
Sources: Boschma, R. (2005), “Proximity and innovation: A critical assessment”, Regional Studies, No. 39, pp. 61-74; Maggioni, M. and E. Uberti (2007), “Inter-regional knowledge flows in Europe: An econometric analysis”, in Frenken, K. (ed.) (2007), Applied Evolutionary Economics and Economic Geography, pp. 230-255, Edward Elgar, Cheltenham; Lundquist, K.-J. and M. Trippl (2013), “Distance, proximity and types of cross-border innovation systems: A conceptual analysis”, Regional Studies, Vol. 47, No. 3, pp. 450-460; Koschatzky, K. (2000), “A river is a river - cross-border networking between Baden and Alsace”, European Planning Studies, Vol. 8, No. 4.
Box 1.3. Quantifying border barriers for innovation:
Evidence from the academic literature
The academic literature includes different attempts to quantify the costs and the barriers with respect to innovation and knowledge spillovers associated with the presence of an international border. These studies make use of different indicators (generally patents and patent citations, scientific publications and R&D expenditures) to statistically assess the importance of the border. They generally consistently find that the border has an impact in terms of a faster spatial decay of science and research spillovers.
Okubo and Zitt (2004) study the intra-European S&T co-authorship collaborative network. They also focus on frontier areas and find that EU regions bordering foreign countries are more open towards academic co-authorship with cross-border regions than their national average. However, they also find that the level of preference for other regions within the same country is higher. This phenomenon is accentuated in large European countries. This shows both the importance of geographical proximity (since cross-border regions tends to have more privileged collaboration than other regions with the neighbouring country) but also the great importance of national borders.
Peri (2005) uses patent and patent citation data to estimate knowledge flows across the borders of 147 sub-national regions over the period 1975-96. The author finds that, on average, only 20% of the knowledge spillovers flow over the regional borders and only 9% flow across national borders.
LeSage et al. (2007) try to understand whether knowledge, measured by patent citations, flows more easily within countries than across international borders and to what extent physical distance between inventors is affecting knowledge flows. The authors control for technological proximity between regions and use an econometric model assessing that, overall, knowledge tends to flow more easily within, rather than between, regions across countries. The analysis also shows that language barriers have an even bigger impact than borders.
Greunz (2003) builds a model relying on a knowledge production function measured per R&D expenditure data in order to investigate inter-regional knowledge spillovers across 153 European sub-national regions. The analysis shows that, even when controlling for geographical and technological distance, inter-regional R&D spillovers take place, but to a lesser extent between cross-border regions.
Thoma (forthcoming) finds that in both North America and Europe, the probability of citing a patent in a neighbouring foreign region is no different than citing one in any foreign region, regardless of distance, showing that the border effects dominate over proximity benefits. There is evidence of an increase of the border effect in Europe from early 1990s to 2004, above and beyond distance and language use. The increase of the border effect in North America from the early 1990s to 2002 appears to be even stronger; however, it cannot be compared to Europe, because the evolution of border effect depends also on number of patents invented domestically in each nation.
Sources: Peri, G. (2005), “Determinants of knowledge flows and their effect on innovation”, The Review of Economics and Statistics, Vol. 87, No. 2, pp. 308-322; LeSage, J., M.M. Fischer and T. Scherngell (2007), “Knowledge spillovers across Europe: Evidence from a Poisson spatial interaction model with special effects”, Papers in Regional Science, Vol. 86, No. 3, pp. 393-421; Greunz, L. (2003), “Geographically and technologically mediated knowledge spillovers between European regions”, The Annals of Regional Science, Vol. 37, No. 4, pp. 657-680; Okubo, Y. and M. Zitt (2004), “Searching for research integration across Europe: A closer look at international and inter-regional collaboration in France”, Science and Public Policy, Vol. 31, No. 3, pp. 213-226; Thoma, G. (forthcoming), OECD Regional Development Working Papers, forthcoming, http://dx.doi.org/10.1787/20737009.
Regional strategies need to consider cross-border neighbours as well as wider global networks
A strong cross-border regional innovation system can better take advantage of global networks. The literature on regional innovation systems highlights the relationships among different types of actors co-located in the same place (Cooke et al., 1997). The so-called “triple helix” refers to the close interaction of: i) firms; ii) universities; and iii) the public sector in promoting a strong innovation system (Leydesdorff and Etzkowitz, 1996). If some important knowledge generation or innovation partners are interacting with farther global partners, and they are actively connecting to other actors locally, that global knowledge can be better diffused locally (Benneworth and Dassen, 2011). The terms “local buzz” and “global pipelines” have been used to illustrate the importance of having both strong local and global connections (Bathelt et al., 2004). A regional innovation system on a cross-border basis overcomes obstacles associated with an international boundary for a more integrated system. It therefore can access the two national innovation systems and reach a broader range of global actors (Figure 1.4).
Figure 1.4. Stylised depiction of cross-border regional innovation system integration
Notes: NIS = national innovation system, RIS = regional innovation system, see Annex 1.A1 for the characteristics associated with each stage.
Source: Lundquist, K. and M. Trippl (2013), “Distance, proximity and types of cross-border innovation systems: A conceptual analysis”, Regional Studies, Vol. 47, No. 3, pp. 450-460.
Collaborations driven by physical proximity and collaborations driven by global excellence are not mutually exclusive. Innovation system actors operate on different spatial scales. Cross-border clusters of firms may jointly seek opportunities for collaboration with markets further away. Universities can promote together mobility schemes for staff and students. Irish and Northern Ireland universities, for example, are active in establishing common platforms of collaboration with leading academic institutions in the United States. Cross-border efforts can make the area more attractive for global actors to interact with the region.