Qual Quant (2011) 45:1053–1057 DOI 10.1007/s11135-011-9489-y

Networks and geography in the economics of knowledge flows. A comment Nadine Massard

Published online: 2 June 2011 © Springer Science+Business Media B.V. 2011

Abstract This paper proposes a comment on the article by M. A. Maggioni and T. E. Uberti within this issue. It focuses on one main question: can we say that the literature on Networks and Geography has reached the initial objective assigned to it: “Demonstrate that networks and geography are the necessary ingredients for every study of the innovative process at any level of analysis: from individual agents, to institution/organization, from the regional to the national and international level”? The answer to this question is presented following three main steps: (1) How do geographical positioning and network positioning interact to impact the innovative performance? (2) What is the influence of the geography on the networking strategies (cooperation choices)? (3) What is the influence of the network upon geographical strategies (location choices)? The conclusion is that the field described by Mario Maggioni and Erika Uberti is, as their paper shows, fast expanding and already has a history and appreciable results. It still clearly appears however that much still remains to be done if the potentiality offered by the cross-use of spatial and network analysis when dealing with knowledge diffusion and its impact on innovation is to be fully exploited. No doubt that the important advances accomplished during these last years in the collecting of data related to localized relationships will provide the necessary incentives to develop the still greatly lacking micro-econometric analyses. Keywords

Network · Geography · Knowledge flows · Innovation

The last 20 years have been marked by some radical transformations in the real processes for the production and diffusion of knowledge. The main characteristic of these changes is probably the increasing importance of networks in the process of knowledge creation. During the 90s and early 2000s, numerous empirical studies, demonstrating the highly concentrated distribution of innovative activities over space, have resulted in an abundant literature on clusters and the geography of innovation. Lately, this new evolution towards the

N. Massard (B) UMR-CNRS 5824, GATE Lyon Saint-Etienne, University of Saint-Etienne, Saint-Etienne, France e-mail: [email protected]

123

1054

N. Massard

reinforcement of the networked nature of innovation brings forth new issues for the theory of R&D cooperation on the one hand, and for the geography of innovation, on the other hand. Firstly, the literature on network formation emphasizes the role of network effects on collaboration behavior. The traditional literature on R&D collaborations assumes that in bilateral collaborations the benefits result from the individual characteristics of the agents involved. The network formation theory suggests, on the contrary, that in order to assess the benefits of a bilateral relationship, we have to take into account not only the individual partner’s knowledge, but also the knowledge that each partner is able to access through her own network of collaborations. Henceforth, the firm’s position within the network is considered to be a key determinant of cooperation. Very few authors propose empirical studies based upon this theoretical background. In this context, ‘network effects’ refer to the effects analyzed in social network analysis (SNA) such as those due to prior acquaintance or preferential attachment. Secondly, the fact that scientific research is increasingly realized within networks is likely to modify the spatial diffusion of knowledge. Are we witnessing the death of distance and consequently the death of geographically bounded knowledge externalities? Questioning the reality and effectiveness of the role played by geographical proximity for knowledge flows gives rise to many new issues for the literature on the geography of innovation. Is the diffusion of knowledge flows becoming progressively free from spatial constraints and thereby more likely to weaken geographical disparities? Or, on the contrary, are we heading towards a new reinforcement of the agglomeration forces resulting from the remaining spatial constraints and organizational considerations? To answer such questions, a better understanding of how distributed groups of scientists or innovators are formed and collaborate to create and diffuse knowledge, and of the specific role played by geography in these processes, is needed. Such is the context which largely explains the development of the research stream on “Networks and geography in the economics of knowledge flows” over the last 10 years. Contrary to the enthusiasts of “the death of distance”, the review proposed by M. Maggioni and T. Uberty does not oppose the network and the geographical perspective in the economics of innovation. Its main objective is to demonstrate “the need to mix and merge spatial and network analysis”. This review is most informative and highly useful thanks to its detailing of the main theoretical and empirical issues at stake as well as the attention that it pays to the data. It also sheds motivating lights on the open questions for future research. The developments showing the non-neutrality of the methodological choices, especially those concerning the mode of representation of the networks derived from available data, or those concerning the adaptation of the SNA framework to economic questions are most enlightening for any researcher who is new to the field as well as for any reader who is interested in finding a relevant interpretation of the network analyses which are presented to him. Within this comment I will focus on the main question which comes to mind at the end, can we say that the literature on Networks and Geography has reached the initial objective assigned to the paper: “Demonstrate that networks and geography are the necessary ingredients for every study of the innovative process at any level of analysis: from individual agents, to institution/organization, from the regional to the national and international level”? I will present my own answer to this question following three main steps: 1. Geographical positioning versus network positioning and the innovative performance 2. Influence of the geography on the networking strategies (cooperation choices)

123

Networks and geography in the economics of knowledge flows

1055

a. Geography and the opportunity to create links b. Geography and the benefit of creating links 3. Influence of the network upon geographical strategies (location choices)

1 Geographical positioning versus network positioning and the innovative performance Being able to access knowledge and to efficiently exploit it internally is the key factor for developing innovative capacities. The question is what matters most within this process: being located nearby the relevant actors, in terms of needed knowledge, or being connected with them through knowledge networks? Most of the studies trying to answer this question limit themselves to the comparison between different measures of distance using networks as an alternative conception of space and comparing the role of relational versus geographic distance between actors. With this aim, the use of spatial econometric tools applied to networks as well as geographical space constitutes an important advance and generally shows the co-existence of these different proximity effects. So far however, very few approaches allow us to really disentangle or organize into a hierarchy the different sources of auto-correlation. Moreover, it remains difficult to know exactly what these interdependencies cover. In particular, it would be interesting to disentangle pure knowledge externalities on the one hand and voluntary strategies to access knowledge on the other hand and to analyze the impact of geography upon the efficiency of these two forms of knowledge diffusion. The main results found in the empirical literature concerning this point appear rather ambiguous. Some authors emphasize the basic role of social networks and show that what we usually name local knowledge externalities are, in reality, the mere consequence of the localization of inter-personal relationships. Hence, they observe that what are usually considered to be spatial effects limiting the diffusion of externalities would mainly rely on network effects explaining the creation of voluntary collaboration between actors. Therefore, it is not easy to isolate the specific effect of the local environment from the different network effects due to the collaborations established by local actors. Other authors, on the contrary, insist upon the specific role of pure knowledge externalities and consider that not only space but also networks are means for pure knowledge externalities to emerge and diffuse. According to this approach, the structural characteristics of networks, as they result from the aggregation of individual cooperation choices, are more or less efficient for transmitting knowledge externalities and therefore for reducing the innovation cost of individuals. To our knowledge however, none of the existing empirical studies crossing network and geography really analyze the role played by the spatial structural configuration of networks for the innovative efficiency of the nodes belonging to this networks. On the whole, it seems clear that, beyond using networks as a new metrics for evaluating distance between regions, there is still a lack of new methodologies which would be capable of plainly exploiting the assets of the network approach in terms of strategic and structural analysis (including spatial strategies and structures) in order to better understand innovative processes. As Mario Maggioni and Erika Uberti state, existing studies either explain the geography of innovation using network analysis or explain knowledge networks using geographical analysis. They have still not however, found the way to really connect the network and the geographical analysis and benefit from both their specificities in order to better understand the innovative process and efficiency.

123

1056

N. Massard

In this perspective, developing analyses at a spatially aggregated level induces many caveats. For regions for example, it is certainly interesting to be able to assess the respective role of their geographical versus relational proximity to other regions. It would be however, much more interesting to know where the central actors of the knowledge networks are located (that is to say those actors who are centrally positioned within the global knowledge networks as measured by the different centrality indicators designed by the SNA) and how their networking strategies impact the geography of innovation. What role do they play within regions? How do they participate in the internal and/or external connectivity of regions and act as intermediaries between different actors of the innovation process whether they be local actors or not? Very few existing empirical methodologies allow us to estimate these interwoven phenomena and their conception is not easy due to the important endogeneity problems that estimation would face when trying to really assess the interaction between the geographical and network determinants of innovative activities. As is mentioned by Mario Maggioni and Erika Uberti, the development of temporal methods on the one hand and of methods based on the modeling of individual rationality on the other hand constitute important steps towards a better understanding of knowledge flows through the fundamental interaction of networks and geography. One can only note that much still remains to be done to better define the microeconomic foundations of the Network and Geography perspective, firstly concerning the influence of geography upon the networking strategies of individuals and secondly concerning the influence of networks upon the location strategies of individuals.

2 Influence of geography upon networking strategies In our opinion, this point covers two main elements. a. Geography and the benefit of creating links Many studies analyze the determinants of cooperation choices and notably try to determine the relative weight given to the different forms of proximity between actors in their choice to cooperate with each other or not. The underlying determinants of the individual rationality which govern these choices however, are rarely deeply justified. If there is a cost/benefit justification it is mostly limited to the statement that, whatever its nature, proximity reduces the cost of establishing cooperation and enhances the benefit gained for facilitating knowledge flows. The existence of negative effects of proximity on cost and benefit of cooperation is rarely considered an whether or not it is possible to organize a hierarchy between the different forms of proximity (geographical, cognitive, organizational, institutional, social…) and to establish complementarity or substitution relations among them is not mentioned. If these studies are to be used in order to estimate the endogeneity of some results concerning the impact of cooperation upon innovative performances, individual cooperation rationality linked to proximity effects need to be studied in more detail. b. Geography and the opportunity to create links This second point is generally totally neglected within empirical studies whereas it is an important element of the game theory approach of network formation. Indeed, in order to be able to calculate the expected gain of establishing a cooperative link, it is necessary to be aware of the mere existence of potential partners. Selection biases, which are due only to the fact that each agent has a limited knowledge of the whole set of potential partners, are rarely

123

Networks and geography in the economics of knowledge flows

1057

estimated. From this point of view however, one may continue to wonder about the main methods of information diffusion and especially about the influence of geography upon the capacity of agents to be aware of potential partners. Are ICTs sufficient today for actors to find their potential partners? Is the cost of this process really reducing with the development of ICT? One might doubt this when faced with the number of incentives and intermediaries that the public authorities devote to finding partners. Geography can also play a role in this process by reducing actors’ mobility and therefore their capacity to become acquainted. This last point raises the question of the articulation between knowledge flows which circulate among geographically localized actors who cooperate within networks and knowledge flows which circulate among moving actors who create links through their own movements. At this point we should consider the influence of networks upon the geographical positioning of actors that is to say upon their location choices.

3 Influence of the network on location choices In order to complete the presentation of the elements which are likely to create endogeneity within econometric estimations concerning network and space, we should mention the role of collaborative networks in the location choices of innovative activities. It is rather paradoxical today to consider only the factors due to the geographically bounded local environment as determinants of location choices. A more global rationale of the accessibility to knowledge offered by a territory is now needed. This implies measuring the capacity of a territory to facilitate the setting up of beneficial cooperation and the insertion of actors within global networks. Thus, whereas numerous studies have shown a tendency to choose partners who are close, geographically speaking, is it also possible today to find location strategies which consider cooperation choices as determinants of location choices? To conclude, the field described by Mario Maggioni and Erika Uberti is, as the paper shows, fast expanding and already has a history and appreciable results. It still clearly appears however as a fragmented research field which have not yet attained the initial objective of proving the interwoven nature of networks and geography as determinants of innovative processes. Much still remains to be done if the potentiality offered by the cross-use of spatial and network analysis when dealing with knowledge diffusion and its impact on innovation is to be fully exploited. No doubt that the important advances accomplished during these last years in the collecting of data related to localized relationships will provide the necessary incentives to develop the still greatly lacking micro-econometric analyses.

123