We uncover the unintended benefit of a conditional cash transfer program: overcoming coordination failures. Beneficiaries interact regularly, giving rise to a coordination device which we measure experimentally using a minimum effort game. Those enrolled for over a year exert significantly higher effort, which is not due to either willingness to cooperate or network density. Supplying high effort relies on individual beliefs that others will reciprocate, a belief that holds in equilibrium for intermediate levels of payoff-responsiveness. This is the first study of the extent to which a government intervention can set a precedent toward efficient coordination among communities