Network Structure and Credit Rating

Abstract

We consider a model in which productive bilateral links are formed between heterogeneous agents who differ in their innate productivity. Local information is complete but an outside planner can observe only network properties. We ask if consistent credit rating -- where agents' ratings are increasing in their productivity -- is possible using network characteristics alone. The key to our results is that the network structure is endogenous since the use of agents' network neighborhood properties in generating ratings also impacts their incentives for link formation. Network structure and credit scores are therefore determined jointly in equilibrium. We show that if the cost of link formation is not too low, there is a pairwise stable equilibrium under credit rating where the network structure is a connected nested split graph (CNSG) with neighborhood size increasing in type. This is also the unique equilibrium if we consider a class of truthtelling equilibria (that is, equilibria in which the network structure separates types, enabling consistent credit rating). Further, among networks that separate types, this specific CNSG constitutes precisely the optimal structure.