Religion and the Family: The Case of the Amish

I construct a model of religion as an institution that provides community enforcement of contracts within families. Family altruism implies that family members cannot commit to reporting broken contracts to the community, so the community must monitor contract performance as well as inflicting punishment. The community has less information than family members, and so community monitoring is inefficient. I provide evidence from a study of Amish institutions, including qualitative evidence from sociological accounts and quantitative evidence from a novel dataset covering nearly the entire Amish population of Holmes county, Ohio. I find that 1) Amish households are not unitary, 2) the Amish community helps to support families by inflicting punishments on wayward family members, 3) without the community Amish people have difficulty committing to punishing family members, and 4) Amish community membership strengthens family ties, while otherwise similar religious communities in which there is less need for exchange between family members have rules that weaken family ties. My model has implications for understanding selection into religious practice and the persistence of culture.