In this paper we investigate whether access to savings accounts affects risk attitudes and intertemporal choices. We exploit a field experiment that randomized access to savings accounts among a largely unbanked population of Nepalese villagers. One year after the accounts were introduced, we administered lottery-choice and intertemporal-choice tasks to the treatment and control groups. We find that the treatment is more willing to take risks in the lottery-choice task and is more responsive to changes in experimental interest rates in the intertemporalchoice task. The results on time discounting are less conclusive, but suggest that the treatment group is more willing to delay receiving money. These results suggest that access to formal savings devices has a positive feedback loop for poor families by increasing their willingness to take risks and to delay gratification.