Who gives, who cares?

We derive the testable implications of a network model of altruism on data consisting of the transfers made between agents in an economy over a finitely many time periods. We precisely characterize rationalizable transfer data: that is, we characterize when we can find time invariant preferences and time varying initial incomes such that the transfers are obtained as an equilibrium of the network transfer game with altruistic agents. We show that, despite only observing transfers (and neither preferences nor initial incomes), the model can be refuted and moreover that the transfers reveal the strength of altruistic ties. Our model is computationally straightforward to test and this test is robust to measurement error in that only the direction of the transfers matters but not the actual transfer amount. Finally, we discuss the implications of our results to the economics of informal transfers.

Joint with Rahul Deb (Toronto), and Ludovic Renou (Queen Mary).