Extension of credit to people in developing markets has been a long time challenge. Banks, of course, look to repayment history to make such determinations but in much of the world, banking relationships and repayment track records are few. But history has demonstrated that extension of credit in developing markets can be effective and profitable. Just look at the Grameen Bank's high micro-loan repayment rates.

To address this repayment data dearth, Lenddo.com developed a lending data set in multiple developing countries, having gone into the lending business just to generate the data it needed to tune its machine learning capability. Lenddo then built its algorithms that examine some 1,000 characteristics in the data drawn from social, mobile, and other sources. This Payments on Fire podcast with Lenddo.com's founder Jeff Stewart takes a look at lending in developing countries, social and mobile data sources, and examines the algorithmic "black box" that is at the heart of the company's approach to making credit decisions in "thin file" markets.