Informal Insurance, Caste Networks, and Financial Development: Evidence from the Indian Social Banking Experiment
We develop a simple model of informal interpersonal insurance in caste-based networks to analyze the impact of access to banking on insurance. In the model, individuals are able to share idiosyncratic risk and are subject to limited commitment, which they mitigate through punishment that is mediated by their caste-based network. The model illustrates that, al- though the effect of banking on overall insurance is ambiguous, it should be more beneficial to smaller caste groups. The reason is that smaller caste groups have denser networks that make them better able to deal with limited commitment, which is exacerbated by their access to banking. We empirically test the model predictions by exploiting the fact that places with more caste fragmentation are formed by a larger number of smaller caste groups. Following an empirical strategy in the spirit of Burguess and Pande (2005) and Jayachandran (2006), results indicate that, while access to banking improves overall insurance, it is more beneficial in places with more caste fragmentation. Moreover, there is evidence that banking may crowd out overall insurance in places with caste homogeneity.