Yves Zenou, Stockholm University

The promotion of cooperation between firms, universities, research centres and public agencies has been at the heart of the Science and Technology Policy in Europe over the last couple of decades. The motivation behind this policy is the belief that cooperation by institutions and firms across the various European countries represents a means by which the technological and economic gap between the US and Europe might be narrowed. Starting from the Sixth Framework Programme (2002-2006), policy actions are to be more focussed on identifying crucial nodes and networking centres of excellence, stimulating R&D in particular in high-tech industries, that would represent the backbone of a truly European Research Area.
Moreover, recent studies (see for instance Hagedoorn 2002) have established that joint ventures seem to have become less popular while non-equity contractual forms of R&D partnerships, such as joint R&D pacts and joint development agreements, have become very important modes of inter-firm collaborations. These contractual agreements cover technology and R&D sharing between two or more companies in combination with joint research or joint development projects.

However, little is known on how efficient these R&D inter-firm collaborations are and how the structure of these networks affects the outcomes (profits) of these collaborations. The purpose of this project is to contribute to a better understanding of these issues by providing both a formal analysis of R&D networks and empirical tests.

Three papers

To understand better these issues, we have written three different papers.
In the first paper (A. Cabrales, A. Calvó-Armengol and Y. Zenou (2009), "Social Interactions and Spillovers: Incentives, Segregation and Topology," Working Paper, which has been revised and resubmitted to the Review of Economic Studies), we use the recent developments of graph theory (Goyal, 2007; Jackson, 2008; De Marti and Zenou, 2009), to take into account both the direct and indirect collaborations between firms in a network and thereby capture the full effect of a network on the participating firms. To be more precise, we analyze a two-stage model in which the organizations first make a socialization effort, and after observing that effort (but before the alliances crystallize) they decide the resources that they are going to dedicate to R&D. A crucial innovation of our study is that the socialization effort is "generic". That is to say, the companies dedicate resources to form alliances, that allow them to benefit from the R&D of other organizations, but they do not focus the resources in forming alliances with a particular company. Nevertheless, the probability (or intensity) of the relation with another company depends on the product of their efforts. We first characterize the equilibria of the model: existence, uniqueness, symmetry. An important question is how a change in the return to the innovation affects the relative effort of research and socialization. We find that the effect is relatively stronger on socialization. This can explain the explosion in agreements of collaboration in R&D in the recent past (Caloghirou, Ioannides, Vonortas, 2004).

Let us turn to the policy implications of our model. Given the external effects, in our model, the investments are sub-optimal and therefore the public agents should be interested in subsidizing it. Furthermore, our theory suggests to public authorities to be cautious and to be aware of the downside effects associated with collaboration. With all their benefits, partnerships have the negative potential to block competition and create various kinds of static and dynamic monopolies (in existing and future markets, respectively). Our model highlights the trade-off between the costs and benefits of R&D partnerships and design an optimal policy that takes into account this trade-off.

In our second paper (Bertrand, O., Nilsson Hakkala, K. and P-J. Norbäck, 2008, "Cross-Border Acquisition or Greenfield Entry: Does it Matter for Affiliate R&D?" Working Paper, Research Institute of Industrial Economics), we investigate how the entry mode of foreign direct investment (FDI) affects the affiliate R&D activities using unique data on Swedish multinational firms over a long period of time (1970 to 1998).
Indeed, an increasing share of foreign direct investments (FDI) is now taking place in the form of cross-border mergers and acquisitions (M&A). M&As accounted for approximately 80% of all FDI transactions over the 1990's. The increase in FDI through foreign acquisitions has raised some concerns for policy-makers. Governments tend to be skeptical towards foreign takeovers of domestic firms, in particular when the acquired domestic firms are endowed with technological assets. For instance, the rumors about a takeover bid of the French dairy producer Danone by the American company PepsiCo provoked an outcry in the French political arena, some politicians swearing to protect this French company from any foreign take-over. A few weeks later, the French government officially proposed to shield ten "strategic" industries, including biotechnologies, secure information systems, casinos and the production of vaccines, from foreign acquisitions. There is a fear that the innovative activity of the acquired firms will be reduced or shut down, thereby depriving the local economy of strategic technologies and technological spillovers. While many countries encourage inflows of greenfield FDI (start-ups), restrictions on foreign acquisitions of domestic firms are common (Mattoo et al., 2004). Greenfield investments are then seen as having a positive impact on host countries by, for instance, developing new research and development (R&D) capacity in the host country and creating technological spillover benefits.

Motivated by these concerns, we investigate empirically whether the choice of entry mode of FDI, that is M&A or greenfield entry, is of importance for affiliate R&D activities. To this end, we use unique data on affiliates of Swedish multinational firms collected by the Research Institute of Industrial Economics. Controlling for parent-, affiliate-, industry and country characteristics, we first show that acquired affiliates are, on average, more likely to do R&D and have a higher level of R&D intensity than affiliates created by greenfield investments. These results persist over time and with the age of affiliates, as well as for different firm types and industries. Our findings thus suggest that discriminating against cross-border acquisitions in order to promote greenfield investments may be counter-productive for a host country aiming at increasing R&D investments through inflows of FDI. This difference in observed R&D is explained by differences in parent, affiliate, industry and country characteristics as well as by different reactions to these characteristics, as predicted by the recent theoretical literature on international mergers and acquisitions (M&As). The results also suggest that M&As are, to a larger extent, motivated by asset-seeking motives than greenfield entry, especially in the 1990s.

In our last paper (Svensson, R., 2007, "Commercialization of Patents and External Financing during the R&D-Phase," Research Policy 36, 1052-1069), we empirically analyze how different factors influence the decision to commercialize patents. To perform this analysis, we use a unique database on 860 Swedish patents granted to small firms and individuals. The possibility to follow the commercialization process of individual patents makes this database unique. The database has information about if and when the patent was commercialized, the outcome of the commercialization as well as which strategies the owners used; for example, whether the patent owner received financial and management support from external financiers during the R&D and commercialization phases, and which commercialization mode was used, etc. It is then possible to examine which factors are crucial for the time it takes until commercialization starts and for the commercialization to become successful. Such an analysis has seemingly never previously been undertaken. First, using a survival model, we test how different factors affect the time it takes until the commercialization of patents starts. Here, we analyze how networks with, and support from, external financiers - private and government - during the R&D-phase influence whether the patent is commercialized. Second, we want to answer the question: Which strategies are crucial for the commercialization to be successful? The outcome is related to the presence of external financiers, whether the inventors have an active role during the commercialization phase and which kind of commercialization mode is chosen.

Since the owners know more about the patents than potential external financiers, problems related to asymmetrical information are present. To overcome these problems when inventors and small technology-based firms need financing, Sweden has for a long time relied on government support rather than private venture capital firms. The empirical results show that the larger is the share of patent-owners' costs covered by government financial support during the R&D phase, the lower is the probability of patents being commercialized. This lower degree of commercialization is likely to depend on: (i) the soft terms of the government loans, where the patent owner can avoid paying back the loan if the patent is never commercialized; and/or (ii) that the government is not able to select promising projects. The first explanation is related to moral hazard and the second one to adverse selection. The policy suggestion is for government to change the design of the loans, to base them on firms rather than projects.