Rotating savings and credit associations (ROSCAs) - informal groups whose members contribute a fixed sum to a common pool at regular intervals, with the entire pool disbursed to one member each period until every participant has received a payout - appear across societies that share little else. Economists long treated such arrangements as vestigial, the sort of institution that formal banking would inevitably supplant. Their persistence, and even proliferation, amid expanding financial infrastructure has forced a reassessment. One influential account, which I will call the substitution thesis, holds that ROSCAs are simply a response to the absence of accessible banks: where formal credit is scarce or costly, members pool resources to finance lumpy purchases - a sewing machine, a dowry, a roof - that individual saving would delay. On this view the association is a second-best expedient, and its decline should track the spread of formal institutions. The substitution thesis has intuitive appeal and predicts, correctly, that ROSCAs cluster where banking penetration is low. Yet the thesis struggles to explain a recurrent finding: ROSCAs frequently coexist with, and are used by members who also hold, formal accounts. Something beyond mere access must be at work. A second account emphasizes commitment. Saving alone requires resisting the continual temptation to spend; the ROSCA externalizes that discipline, because failing to contribute exposes a member to the censure of neighbors whose own payouts depend on her reliability. The pool, in effect, converts a private struggle of self-control into a public obligation. This commitment account explains why the affluent, who plainly have banking access, sometimes join: they are buying enforcement, not liquidity. I find the commitment account the more powerful, but it too is incomplete. It treats the social tie as mere collateral - a lever for enforcement - when the tie is frequently the point. Fieldwork records members who value the periodic gatherings, the standing they earn by hosting, and the dense web of mutual knowledge that lets the group extend informal insurance beyond the ROSCA's formal terms. To read these features as instrumental is to invert the order of explanation: the association does not borrow social capital to secure savings so much as it uses the discipline of saving to sustain the social capital. None of this implies that ROSCAs are costless or uniformly benign. Their reliance on reputation makes them fragile to migration and to the anonymity of large cities; a defaulting member near the end of a cycle can impose losses that no formal recourse will recover. And the very intimacy that enforces contribution can also entrench exclusion, since those without established standing are rarely admitted. The lesson is not that informal finance is superior to formal finance, but that the two answer partly different needs. A policy that regards ROSCAs merely as a symptom of financial underdevelopment - to be cured by extending banks - will misjudge both what these associations do and why, even amid abundant alternatives, people continue to choose them.