Kinship is an essential factor in human social life. Many years of research devoted to develop a better understanding of kinship bear witness to this fact. Important advances were made on conceptual, modeling and empirical grounds. Computational social science---in particular through social network analysis and social simulation---contributed its part to it. Notwithstanding, multiagent simulations of social systems rarely take into account kinship-based social interactions, especially when claiming to be empirical. We combine generative social science's basal argument "grow it!" with the concept that social structure is not reified, but a pattern emerging from interactions between individuals, and introduce a multiagent social simulation that "grows" kinship structures on the basis of socio-demographic and marriage interactions in Pakistan's Rural North-West Frontier Province. The modeling proposed has generalizable demonstrator value in that it is shown how published ethnographic data can be used in building credible and cross-validateable formal models that then can be re-used in modular model building of larger models of society. We conclude with a critical discussion of taking into account endogenized social structure in formal models of conflict.