¾ÅÉ«ÉçÇø

The paper develops a formal model of coalition-building ("network" formation) among research units that seek competitive funding from a supra-regional program, while also drawing support from their respective regional funding agencies. This approach enables one to ask whether there are stable (equilibrium) outcomes in the interactions among the several funding entities, and to investigate what those outcomes would imply for the evolving distribution of scientific performance within the entire region and its national sub-regions. This analysis is motivated by the absence of frameworks of analysis applicable to problems of design of public R&D funding arrangements in the European Union, and in other regional systems were independent programs of "federal and state" support for research co-exists.   First, a model is developed to analyze how collaborations are formed under different sets of funding rules of an international funding institution, starting with a fixed finite population of research units and an associated distribution of reputed quality, or scientific reputation.. Collaborations are formed in the expectation of attracting supra-national funding, following a specific ordering procedure; this gives rise to a repeated non-cooperative game of coalition (or collaboration) formation with the distribution of payoffs within the collaboration following to a fixed rule. Non-cooperative games of coalition formation developed by Bloch (1995), and Ray and Vohra (1999), provides a useful framework single-period framework. Following Keely (1999), this type of game is applied to a multi-period setting in which a distribution of coalitions is tracked, along with the levels of funding received. The latter are determined according to a rule comparing the distribution of reputations within and across collaborations.