Religions are clubs and theory tries to explain why people want to join them and then stay. Current theory is good at replicating corner solutions: extremists that invest at high cost (time, money and future prospects) to participate in a religion, like it is the case for sects, strict religious orders or extremist militants. But this theory has nothing to say about moderate religion, that is people that participate in religion an hour every Sunday and then occasional additional activities, and thus provide much less sacrifice and yet still .
Michael Makowsky shows that ones you combine agent heterogeneity and repeated decisions, moderate religious groups are viable and may even dominate the landscape. Members of a group contribute time and money, and the wage heterogeneity convexifies the scale of agents. One more area where heterogeneity appears crucial to understanding data, even if this means that solutions are complex and may have to be solved by computer.