A Network Structure of ROSCAs (Rotating Savings and Credit Associations): ERGMs (Exponential Random Graph Models) Applied to a Leaders' Network in Rural Uzbekistan
This paper examines the structural effects of Rotating Savings and Credit Associations (ROSCAs) networks in rural Uzbekistan.
It applies exponential random graph models to a village leaders’ network and uses Markov Chain Monte Carlo (MCMC) algorithms to estimate parameters corresponding to structural effects.
Participation in multiple ROSCAs is a common phenomenon in Uzbekistan. The many layers of interaction resulting from this type of participation have given rise to complex network structures that extend over the entire village. The paper examines whether actors choose their ROSCA partners solely on the basis of personal status, attributes or affiliation or, whether they also consider potential network structures resulting from their additional membership. Specifically the paper focuses on a transitive triad structure of the network that facilitates monitoring and enforcement. Results imply that:
- ROSCAs contribute to the formation of the transitive triad structure within the leaders’ network, but work against the degree-based core-periphery structure;
- Propensity of the network towards forming transitive triads contributes to the network’s stability.