Risk-Sharing Tests with Network Transaction Costs

In a world with costly transfers, some agents with high transaction costs may not find it feasible to trade. Hence, their consumption will co-vary with their endowment, leading to imperfect risksharing. In this paper, we augment the canonical risk-sharing model to incorporate frictions in the form of transaction costs. In this augmented model, given a particular network structure, risk sharing will happen within networks and not the whole universe of agents. We show that transaction costs and the implied network structure of trade have important implications for the tests of risk sharing. Using this model, we derive a structural risk-sharing test that uses consumption and production data alongside the trade network structure. We implement our method using data from the global trade of three major staple food commodities. Comparing our estimates with the benchmark of frictionless trade, we find some evidence of transaction costs impeding risk-sharing in these commodities.