LAUSR

Microlending has grown rapidly and now benefits around 250 million people globally, half who would otherwise not have access to credit. Use of social credit systems for microlending risk assessment is most pronounced in Asia, as most Western countries tightly regulate personal information available to lenders. In most of the developing world, geography, social structure, disease, climate and culture have a much stronger influence on credit risk and borrowing than do governmental and corporate systems. In this study, we obtained 784 loan contracts with 3577,912 personal communications and locations. Exploratory analysis found loan default depends on social network structure; graph analysis indicated that those who were likely to default tended to communicate with other likely defaulters. Detailed tests were equivocal, suggesting that social network communication structure provided little additional information to predict default, and may even add noise to the data. Our tests strongly supported the importance of location and proximity to particular sorts of landmarks on the potential for default. Proximity to some landmarks, e.g. city hall, moving companies and train stations, were associated with lower loan default. Others, such as parks, stadiums and bus stations, were correlated with a higher loan default. We restructured our tests based on risk-return versus loan default effect with little change in results.