This paper investigates the welfare impact of income redistribution for public goods in networks. First, the authors measure the impact of income redistribution and show that it affects each consumer only insofar as it affects his neighbourhood. Second, they characterise Pareto-improving income redistributions and relate them to the network structure. Third, in the case of Cobb–Douglas preferences, they establish a new link between two well-known concepts of the comparative statics of income redistribution: the neutrality result and the transfer paradox. Collectively, the authors' findings uncover the importance of the −1 eigenvalue to economic and social policy: it is an indication of how the network structure induces consumers to absorb the impact of income redistribution.