Existing theories of cooperation with repeated interactions under imperfect monitoring cannot adequately explain observed trust in the presence of weak enforcement institutions. We introduce the concept of Self-Commitment Institution (SCI) and embed it in an otherwise standard framework to highlight a novel mechanism. When SCI is enforced, agents voluntarily undertake a perfectly observable and costly action, which mitigates agents’ inclinations toward opportunistic behavior. We characterize social norms both with and without SCI and provide necessary and sufficient conditions under which the former are socially desirable. We find that (i) when agents are sufficiently patient, players who conform to social norms with SCI are more willing to cooperate, even after realization of a negative event: (ii) inter-generational cooperation is easier to sustain with (without) SCI in the presence of weak (strong) monitoring institutions; (iii) for any level of monitoring, the larger the value of future exchanges, the lower is the value of SCI. In addition, we investigate the role of memory and the impact of growth, yielding the following additional findings: (iv) bounded memory does not preclude the enforceability of trust under SCI; (v) productive SCI sustains Pareto superior equilibria for any quality of external enforcement. Our results are broadly consistent with patterns of cooperation and observable practices in ongoing organizations. Joint with F. Lanzia.