When a bank considers a loan, it looks at the borrower’s income, assets, credit history, and plans for the money. On microloan websites, lenders have one other way to evaluate a borrower’s creditworthiness. They can observe the behavior of other lenders.
With a colleague, Peng Liu of Cornell University, I have been studying Prosper.com, the largest of the microlending sites. On Prosper, lending is transparent. Borrowers make requests in public postings and typically rely on multiple lenders. Prosper assigns credit ratings to borrowers, and friends of borrowers can post endorsements. Once the process is under way, lenders can see how other lenders respond to the listing.
We analyzed over 2 ½ years of Prosper data to determine the dynamics of lender behavior. We thought we might see what is known as “irrational herding,” or mimicking. If irrational herding is at work, then a listing that received a strong initial response would attract more and more lenders. As we sifted through the data, we found no evidence this was happening.
Instead, we discovered that lenders did, indeed, respond to the behavior of other lenders, but in a more complex fashion. They modified their actions based on attributes of the loans, a process we call “rational herding.”
The qualities of a loan that lenders valued were precisely the opposite of what one might expect. If two loans were doing equally well after the initial listing, lenders would tend to go with the borrower with the lower credit rating. And lenders tended to avoid borrowers who had received positive friend reviews.
Why would lenders prefer a borrower with a lower credit rating? We believe it is because lenders conclude there must be something very desirable about a listing if a borrower is receiving loans in spite of a bad rating. And if a loan is doing well after getting a positive friend review, lenders tend to discount the loan’s success, attributing it to the review.
We wanted to find out how effective this counterintuitive approach of micro-lenders is, so we tracked the performance of loans, looking at default rates. We found that lenders were, indeed, accurate in their judgments. Loans to borrowers with the lower credit ratings favored by lenders performed better than those given to borrowers with good credit ratings. Similarly, loans to borrowers with positive reviews not favored by lenders performed worse than those to lenders without the reviews.
Microlending has emerged in recent years as an important alternative to traditional credit markets, which are still struggling to recover from the sub-prime meltdown and global financial crisis. The good news here is that while those who make microloans may not have formal training or professional experience, they are savvy and aware. And they are proving to be quite accomplished at judging the quality of the loans they issue.
Juanjuan Zhang is Class of 1948 Career Development Professor; Associate Professor of Marketing
Share your thoughts