The curse of centrality in weighted networks

In a wide range of economic, social, and political networks, agents allocate a finite resource, such as time, across their differently valued connections with other agents. A growing theoretical literature inves- tigates how these pre-existing weighted networks affect optimal allo- cations. This paper offers, instead, an empirical analysis identifying the effect of elements of the network on actual allocations. For that, we design a novel experiment and analyze the resulting compositional data. By focusing on conflict networks, we find that resources allo- cated against individuals with higher centrality exceed the theoretical benchmarks: the curse of centrality.