Do democratically chosen rules lead to more cooperation and, hence, higher efficiency, than imposed rules? To discuss when such a “dividend of democracy” obtains, we review experimental studies in which material incentives remain stacked against cooperation (ie, free-riding incentives prevail) despite adoption of cooperation-improving policies. While many studies find positive dividends of democracy across a broad range of cooperation settings, we also report on studies that find no dividend. We conclude that the existence of a dividend of democracy cannot be considered a stylized fact. We discuss three channels through which democracy can produce such a dividend: selection, signaling, and motivation. The evidence points to the role of “culture” in conditioning the operation of these channels. Accepting a policy in a vote seems to increase the legitimacy of a cooperation-inducing policy in some cultures but not in others.