Abstract (click to expand): Although many important economic decisions are taken by groups (e.g. households and committees) considering costs and benefits over time, we have little empirical evidence on their intertemporal behavior. This paper conducts a field experiment and uses reduced-form and structural methods to study how within-group differences drive group-level behavior. We measure time preferences using effort-based tasks in the field, create groups randomly, and collect data on the time preferences of both groups and their constituent individuals. We find that groups are more present-biased than individuals, and that individuals with higher present bias drive decisions within groups. Further, we find that groups exhibit greater present bias when the difference in discount rates within the group is larger. Finally, we document that present bias in group decisions is reduced when bargaining power in the group is less symmetric, making the group act more like an individual.