This paper models the formation of R&D networks in an industry where firms are technologically heterogenous, extending previous work by Goyal and Moraga (2001). While remaining competitors in the market side, firms share their R&D efforts on a pairwise base, to an extent that depends on their technological capabilities. First, we consider a four firms’ industry. In the class of symmetric networks, the complete network is the only pairwise stable network, although not necessarily profit or social welfare maximizing. Then, we extend the analysis to asymmetric structures in a three firms’ industry. Only the complete and the partially connected networks are possibly stable, but which network is stable depends on the level of heterogeneity and technological opportunities. The complete and partially connected networks are also the possible welfare and aggregate profit maximizing networks, but social and private incentives do not generally coincide. Finally, we consider the notion of strongly stable networks, where all the possible deviations by coalitions of agents are allowed. It turns out that in the four firms’ case, the complete network is very rarely strongly stable, while in the three firms’ case the partially connected network where two firms in different technological group are linked is, for a large subset of the parameter space, the only strongly stable network.
