> “amount borrowed is very low.... There is a certain amount of overhead in processing a loan.” this is the crux of the business challenge here, and it’s commo

> “amount borrowed is very low.... There is a certain amount of overhead in processing a loan.” this is the crux of the business challenge here, and it’s common for long-tail businesses like payday lending. fixed and marginal costs are high relative to marginal revenue and to loan amounts. but the social and ethical issue is rooted in bias and discrimination mediated through the credit bureaus who have a poor track record with low-information and/or highly-dynamic credit landscapes. they apply a static model that works fine for the affluent while shutting out the poor. payday lenders transfer default risk to borrowers through high fees rather than relying on imperfect creditworthiness models applied to borrowers that often have little credit history or bad credit from difficult life circumstances. in that situation, is it possible to create credit models that better predict default risk based on information available in our interconnected age? some startups tried using social media data for this, but none seem to have emerged victorious.