A large literature has studied how peers affect behavior by exploiting the preexisting social network structure only. What if networks rewire in response to changes in the economic environment, such as a randomized intervention? We exploit a unique panel dataset that contains detailed information on the network of informal financial transactions before and after a field experiment that randomized access to savings accounts in Nepal. First, we show that the intervention affects the structure of the network of informal financial transactions among households. Second, we estimate a panel model of peer effects in expenditure where the network may change endogenously, and we exploit the design of the randomized intervention to instrument for the observed network change. Our results suggest that disregarding the network change would underestimate both total peer effects and the overall impact of the intervention.