Fintech Isn’t So Different From Traditional Banking : Trading Off Aggregation of Soft Information for Transaction Processing Efficiency

We examine trade-offs that Prosper, a large online peer-to-peer lending platform, has made over time in its lending processes. Since its 2006 inception, Prosper has (1) changed the way it sets the interest rates on loans from auctions among peer lenders to pre-set pricing based on a proprietary credit-rating model in December 2010; (2) enabled institutional investors to fund whole rather than just fractional loans in April 2013; and (3) reduced the extent of soft information available to peer lenders in September 2013. In each of these changes, Prosper has traded off the aggregation of soft information by peer lenders for transaction processing efficiency, a trade-off similar to that made by large traditional banks. We provide the following evidence regarding the costs and benefits of this trade-off. On the cost side, after the December 2010 change the distribution of loan interest rates shrinks and the ability of interest rates to predict loan default deteriorates, particularly for loans to low credit quality borrowers for whom soft information is more important for credit screening. After the April 2013 change interest rates for low credit quality loans in the fractional loan market are more predictive of default than are interest rates in the whole loan market. On the benefit side, loan volume and funding speed rise sharply after each of these changes. Moreover, the costs appear manageable, as peer lenders' ability to diversify increases and the weighted-average interest rate of a modest-size portfolio of Prosper loans remain just as predictive of default. Placebo and difference-in-differences tests using a peer-to-peer lender that did not change its lending processes during our sample period (Lending Club) indicates that these effects are attributable to the changes in Prosper's processes