Credit constraints are central to development economics theory. However, there is scant direct evidence that supports the existence of such constraints. Traditional tests observe how consumption changes after an unexpected income shock. Such changes can also result from myopic behavior or precautionary savings. This study uses a randomized control trial to explore the effects of enabling savings as a tool to smooth consumption, keeping income constant. The study focuses on community instructors in Mexico. Instructors have to deal with idiosyncratic shocks and shocks related to settling in to new communities. For a group of instructors participating in this study, administrators switched 34 percent of monthly payments to quarterly payments. The switch reduced abandonment of service from 23 to 18 percent. This behavior is consistent with the standard model with credit constraints. It is not consistent with a model without credit constraints or one with myopic individuals.