The rise and fall of R&D networks

Abstract

Drawing on a large database of publicly announced R&D alliances, we empirically investigate the evolution of R&D networks and the process of alliance formation in several manufacturing sectors over a 24-year period (1986-2009). Our goal is to empirically evaluate the temporal and sectoral robustness of a large set of network indicators, thus providing a more complete description of R&D networks with respect to the existing literature. We find that most network properties are not only invariant across sectors, but also independent of the scale of aggregation at which they are observed, and we highlight the presence of core-periphery architectures in explaining some properties emphasized in previous empirical studies (e.g. asymmetric degree distributions and small worlds). In addition, we show that many properties of R&D networks are characterized by a rise-and-fall dynamics with a peak in the midnineties. We find that such dynamics is driven by mechanisms of accumulative advantage, structural homophily and multiconnectivity. In particular, the change from the "rise" to the "fall" phase is associated to a structural break in the importance of multiconnectivity.1 Besides, the hypothesis about the path-dependent character of R&D network evolution also underlies another stream of theoretical works, which in the latter years have tried to account for the observed properties of R&D networks and their dynamics (see in particular the works of Cowan among firms; see also Newman, 2003) and the presence of "nested" core-periphery architectures (see Bascompte et al., 2003). In this way, we refine the existing knowledge on R&D networks by detecting new stylized facts about the structural features of those networks and shed further lights on the drivers of the process of alliance formation.Third, building on the above-mentioned structural analysis, we perform a longitudinal analysis of the determinants of R&D alliance formation. In this analysis, our dependent variable is a firm dyad, and the observation unit is every potential pair of firms in the R&D network. We then investgate which combination of attachment rules provides a good description of the empirical evolution of alliances in the sample considered. We focus on the mechanisms of alliance formation which have received more attention in the literature so far, namely: (i) accumulative advantage, (ii) structural homophily (or diversity) and (iii) multiconnectivity (see Powell et al., 2005;Rosenkopf and Padula, 2008). Moreover, we conduct regression analyses for different time periods. In this way we are able to check if different attachment rules may account for different evolutions of the network over the observed sample. Finally, we also consider separately alliances formed among incumbent firms in the network, and alliances where at least one firm is entering the network, in order to check whether drivers of alliance formation are different across incumbents and entrants.Our results show, first, that the evolution of R&D networks has been universal across different sca...

The role of endogenous and exogenous mechanisms in the formation of R&D networks

“…However, neither the network endogeneity nor the exogenous partner selection, taken independently, are able to explain the topology of observed R&D networks. Endogenous mechanisms alone would lead to more and more centralized network structures over time, which we do not observe in reality6. On the other hand, exogenous mechanisms alone would lead to more regular networks topologies, which we do not observe neither.…”

“…However, these clusters are not isolated; previous studies69 have detected the existence of “shortcuts” connecting different clusters, as well as the formation of alliances with new partners not yet belonging to the R&D network. This observation suggests that firms diversify some of their alliances, rather than just establishing collaborations within a specific cluster.…”

“…The firm activity distributions are right skewed and dispersed over several orders of magnitude, as in many other social and technological systems242526. Contrary to most of the R&D network indicators, that display strong variability and dependence on time6, the activity is a stable attribute that can be assigned to firms and effectively estimate their prominence. Indeed, empirical firm activities are robust with respect to the time t at which they are measured: shifting the time window – of any length Δ t – along the 26 years reported in the dataset does not affect the results (see SI).…”

Network Formation with Local Complements and Global Substitutes: The Case of R&D Networks

“…König et al, 2014a;König et al, 2011], 2 providing an explanation for why nestedness has been observed in empirical R&D networks [Tomasello et al, 2016]. Nested split graphs further have a core-periphery structure, which has also been documented in empirical studies on R&D networks [cf.…”

“…Moreover, the corresponding adjacency matrix is stepwise, characterizing a nested split graph, 28 and becomes increasingly sparse with increasing linking costs ζ. The fact that empirical R&D networks are characterized by nestedness has been documented in Tomasello et al [2016]. Further, note that nested split graphs are paramount examples of core-periphery networks.…”