The Dismal Science

It’s Like eBay Meets

Does peer-to-peer lending work?

Back in 2007, it took little more than a steady pulse to get a loan, albeit a subprime one, from credit officers eager to push loans out the door. Now that the real estate bubble has gone bust, a steady job and 20 percent down is scarcely enough to persuade banks to lay out for a mortgage, home repairs, or anything else.

To fill this financing gap, an increasing number of borrowers are turning to “peer to peer” networks that connect individual borrowers directly to lenders, cutting out the banking middleman. These networks have now financed nearly a half a billion dollars in lending. This is still a long way from the $931 billion in loans and leases that Bank of America had on its balance sheet in 2008, but it’s growing rapidly. Peer-to-peer lenders describe themselves as a solution to many of the banking sector’s current weaknesses, from the lack of small-business finance to the evils of payday lending (which now serves as financing of last resort for those shut out of formal banking altogether).

Economists have been studying these peer-to-peer lending programs from the beginning, and their findings are now starting to show up on the Web. They’ve discovered that while the sites may be useful for some high-risk borrowers—those who stood little chance of attracting loans from traditional banking institutions—these credit markets also result in loan decisions tainted by human frailty and bias. It seems that the middleman—with his credit models and balance-sheet analysis for evaluating prospective borrowers—may provide some value after all.

In the old days, if you wanted to opt out of traditional banking, you’d need friends and family willing and able to finance home improvements or business ideas. Similarly, the only alternatives for safekeeping one’s nest egg were trustworthy acquaintances or the proverbial mattress. The peer-to-peer movement argues that Web-based technology may extend this personal banking network into a nationwide community of lenders and borrowers that could obviate the need for conventional banks.

On Prosper, the largest of the peer-to-peer sites, the process is a little like Lenders select from among a catalog of prospective borrowers much as singles “shop” an Internet dating site for potential partners. A typical listing includes a personal narrative making a case for the loan and a picture of the borrower. (These pictures often also include kids, puppies, and other images that might tug on a lender’s sympathies.) In addition to this (unverified) information, Prosper lists hard data from borrowers’ credit reports, including past delinquencies, credit lines, and ranges for income and credit rating. Once a desirable borrower is identified, lenders place bids specifying how much they’re willing to lend to a particular borrower and at what interest rate. So if many lenders perceive that a borrower is a good credit risk, he’s likely to get all the money he needs at low rates. High-risk cases will end up paying higher interest rates and may not attract funding at all.

Why might the general public do a better job at evaluating the chances of default than the trained personnel of banking institutions? Most obviously, person-to-person loans aren’t burdened with an extra layer of potentially corrupting bureaucracy. We now know that loan officers were motivated to push loans out the door without sufficient concern for risk and default, with their banks quickly packaging and selling off bundles of subprime mortgages in what turned out to be a ruinous game of hot-potato lending. Personal lenders may also better pick up on “soft” characteristics of borrowers—a compelling story or convincing picture—that may be a true indication of likely repayment. Loan officers burdened by institutional rules on borrower collateral or credit history may not have the discretion to act on such information.

Yet there are equally compelling weaknesses to the personal banking model. A quick browse of Prosper or any other peer-to-peer site makes it clear that borrowers trade on emotion. The rational, calculating banker may not be swayed by the photo of three adorable kids (who may not even be the borrower’s own kids) designed to influence Prosper lenders. Further, the bank’s job doesn’t end with borrower selection: Banks also watch over borrowers to make sure they’re using the money for productive investments rather than trips to Vegas and put the screws to people with outstanding loans to make sure they get paid back.

Researchers are finding that there is some truth to both views. A team of Harvard and University of Amsterdam economists using detailed data from Prosper—including precise credit scores for all loan applicants—found that Prosper lenders are in fact quite adept at assessing the creditworthiness of prospective borrowers. Though the site keeps borrowers’ credit ratings confidential—they report 40-point ranges rather than actual scores—those with higher ratings nonetheless obtained lower interest rates than those with low ratings. This is partly because borrowers use hard information like repayment history and credit card records, much as a bank officer would. But the researchers also find that a lender’s ability to predict credit score is more accurate than could be explained by accounting for the tangible financial data available on Prosper’s Web site, implying that they also make (accurate) inferences about credit risk from the pictures and stories posted online. This “soft” information is particularly important for higher-risk borrowers with little financial history to guide lenders’ decisions. Overall, the researchers conclude that Prosper’s credit market operates quite efficiently and without a bank pocketing a slice of the proceeds.

Yet others have found that there are significant limits to the rationality of personal lending, which is encumbered by many all-too-human prejudices and idiosyncrasies. For example, my colleague Enrichetta Ravina has documented that there is a massive beauty premium—i.e., cheap loans for pretty women—enjoyed by Prosper borrowers, despite the fact that better-looking people are in fact more likely to default on their loans. Economists Devin Pope and Justin Sydnor find that racial discrimination also taints the online loan market—black borrowers are much less likely to obtain funding and more likely to pay higher interest rates relative to otherwise-similar whites looking for financing.

(The researchers also find that loans to black borrowers are more likely to turn delinquent, to some extent validating the higher interest rates they face. However, given the relatively short history of Prosper lending, it’s still too early to tell whether these late repayments will translate into higher default rates.)

What does all of this suggest for the future of peer-to-peer lending? Small-scale lenders have shown themselves to be savvy in their lending decisions. They are affected to some extent by human miscalculation and bias, but then again, so are bank loan officers. Employees of big banks are human beings, too, and may be similarly seduced by beauty or vulnerable to racism. Add the entertainment value of finding and following your own loan “investment portfolio,” and the peer-to-peer movement seems likely to be more than a passing fad. (Though it’s unlikely that individual lenders will ever have the means or sophistication to work out the terms of a million-dollar Park Avenue penthouse mortgage or a billion-dollar loan for Chrysler.)

Yet before we celebrate the end of banking as we know it, it’s important to remember that the Web sites connecting borrowers to lenders are middlemen themselves. And this presents an additional layer of risk for those making loans through these sites. In fact, if you visit the Web sites of Prosper, you’ll find that it’s currently shuttered to new lending, pending regulatory approval from the Securities and Exchange Commission. (Lending Club, another leading site, just completed a similar “quiet period.”) Among other things, regulators are concerned with what will happen to lenders’ investments if a site goes bankrupt. If, say, Citibank went belly up tomorrow, depositors’ savings would be insured up to $250,000 by the FDIC. Not so for those choosing to invest their savings through a peer-to-peer network. There is the possibility lenders may face delayed repayment if they get repaid at all (though both Prosper and Lending Club now have backup plans in case they go under). So once again, it raises the concern of businesses taking risks with other people’s money. And given the lessons of the past few years, the SEC is probably wise to make sure this latest round of financial innovation doesn’t end up as the financial crisis of the future.