P to P lending

June 27, 2006

An interesting new internet business is “Peer-to-Peer Micro-Lending” where borrowers can get personal loans (i.e. without offering collateral) and lenders can bid to lend them the money. E.g. Prosper.Com,


  1. A borrower wants a $2,000 loan. He states a maximum interest rate that he’ll pay (say 15%).
  2. “Lender A” might offer to lend $200 at that rate, “Lender B” may offer to lend $100, and so on.
  3. If the borrower stated a low rate, he’ll find that lenders aren’t willing to lend. On the other hand, if the requested amount (e.g. $2,000) is reached, other lenders may still come in and out-bid. So, a lender may come in an offer to lend $200 @ 14%.

The borrowers participate because they’re getting terms that banks etc. will not give them. It definitely makes sense from their perspective, particularly at the riskier end of the spectrum.

Lenders participate because they expect a better return than they can get on other investments. Why would anyone want to lend money to strangers who are unable to raise money elsewhere? The reasoning is along the lines of Milken’s junk-bonds pitch: while the average default rate is higher, the average interest rate more than compensates; and, if you spread your loans — giving a few $100’s to each of several borrowers — you are only taking on an average credit risk for that group of borrowers, rather than the risk in trusting a single borrower.

The folk at Prosper.com publish average industry default rates for various risk categories (each category is a range of credit-score). Every borrowers risk-category is made public to potential lenders. If a loan goes into default, Prosper will send it to collections. From what I can tell, the site charges 1% to service the loan. So, the math works something like this: the last 3 “D-rated” borrowers are paying about 19%. Prosper’s stats say that the average default rate in this group is about 6%. Prosper keeps 1%. So, the net expected interest to the lender is 12% [=19% – 6% – 1%].

The big assumption is that the default rates will be near the Prosper estimate. There are good reasons they need not be so. The published default rates are averages. A credit score is a good predictor of average default rate, only in the context of a particular lender’s recovery procedures. Some credit card companies are better at keeping defaults down (within the same credit-score band) than others. Will Proper be about average in it’s attempts to collect from defaulters?

Too risky for my blood!

(Update Feb 2, 2007: One thing I wonder about is the default rates that Proper.com lists. Presumably, they’re showing average default rates from the general population.  Are these default rates annual rates, or rates over the lifetime of a loan? I’ve never thought about it before, but recent news-articles speak of 20% foreclosure rate among the lowest-quality mortgages. I assume people will default faster on their credit-cards and on Prosper loans than on their home, so I’d assume that the highest default rates should be over 30% if one considers a long enough period: say 5 years.)