Improving Matching Between Interest Rates and Borrower Riskiness in Microfinance Loans Using Sampling Methods

Increasing default rates coupled with high information asymmetry have the potential to discourage future lenders on Zidisha, an online microfinance platform, from offering money to borrowers at low interest rates. The goal of this analysis is to develop a solution to this emerging problem by improving the quality of information available to the lender about potential borrowers’ riskiness prior to determining the interest rate at which he/she will offer to lend money. To do so, sampling methods will be used to build a posterior predictive distribution of loan outcomes as a function of the interest rate offered. The risk/reward preferences of prospective lenders will then be elicited and combined with predicted loan outcomes to propose the lender-specific optimal interest rate for a given loan. This is done in accordance with the goal to provide borrowers with a low cost of capital in the present while not leaving so many lenders dissatisfied with unprofitable loans that there will be a shortage of capital available for loan in the future. ∗Jack graduated Duke University in May, 2015, with a B.S. in Economics with distinction and a B.S. in Statistics. He currently attends The Ohio State University where he is pursuing a Master’s degree in Economics, after which he will work for McKinsey & Co. in New York City. He can be reached at [email protected].