Deconstructing microfinance with agent-based modeling: an unorthodox approach to risk analysis and management

      Financial networks have been described as complex dynamical systems with “tipping points” and “regime shifts”, the understanding of which, as it was recently argued, could greatly benefit from drawing analogies with seemingly unrelated research areas. Inspired by such ideas, we propose an unorthodox modeling framework for the rigorous analysis of the mechanisms and associated risks that pertain to sustainable functioning of the financial institutions outreaching to the poor, i.e. the so-called microfinance institutions (MFIs). Specifically, we constructed an agent-based model using the elements of the network theory in a manner that mimics the system design of an MFI. We found that in a deteriorating economic environment confounded with adverse selection, the risk of moral hazard may cause a collapse similar to the phase transitions in physical systems. An after-the-fact recovery required the conditions to improve beyond what led to the collapse in the first place. These findings suggest a set of measurable quantities helpful in sustainably balancing an MFI between the requirements to reasonably price loans and operate fully self-financed.