I sent the following email to the author. I received an auto-response that she is out of the office through January 27.
-------------------------
Hi Eve,
I saw your Peer-to-Peer lending article on ContraCostaTimes.com. I was particularly interested in what you wrote about Prosper.com, with which I am very familiar, having been a lender there since March 2007. Your statement that the "default rate on the dollar volume of the loans stands at 4.62 percent" is highly misleading, even if perhaps technically accurate.
I assume that what you meant by that figure (which was no doubt given to you by Prosper's marketing department) is that 4.62% of Prosper's $111 million in originated loans has technically "defaulted" pursuant to Prosper's narrow definition of "default." There are several problems with this "analysis," but here are the two largest:
First, the numerator is much too small, because of Prosper's definition of "default." Prosper only counts a loan as defaulted once it sells it to a junk debt buyer (for pennies on the dollar). Although Prosper is supposed to sell loans to the JDB when the loan is more than 4 months late, Prosper currently only conducts such sales quarterly, meaning that there are many bad loans that qualify to be "defaulted," but which are actually 5, 6, 7 or more months late. This was an even bigger problem about 6 months ago, because Prosper was holding the junk debt sales even less frequently and as a result there were a number of loans that were as much as 9 or 10 months late. Because extremely few loans that are even three months late ever cure (indeed, relatively few two-month late loans cure), all of these loans are defaults in everything but name. There has been a lot of speculation that Prosper could have intentionally delayed selling the very late loans to the JDB's in order to manipulate its default statistics. Prosper claims that infrequent sales are a by-product of the need to accumulate a sufficient volume of bad loans for sale. Whether intentional or not, however, the infrequent junk debt sales certainly has the effect (whether intentional or not) of artificially lowering Prosper's default statistics.
Second, the denominator is much too large. Prosper divides defaults by the number or dollar value of ALL loans originated (currently about $111 million). However, it is impossible for a loan to even qualify for default until it is 5 months old (the first payment is not due until a month after origination, so a loan cannot be four months late until it is at least five months old) -- even if the borrower never makes even the very first payment on the loan. Because Prosper is a relatively new company (it opened to the public in February 2006), which for its first year was growing rapidly, a very significant portion of its $111 million in originated loans is too young to have defaulted yet even if no payment was ever made. Roughly $30 million of Prosper's $111 million in originated loans is less than 5 months old. Thus, the denominator should be perhaps $81 million, not $111 million. And most borrowers do not fail to make their first payment (because that looks too suspicious, and also because the borrower usually still has money from the loan available to make the first several payments), so even eliminating only the last 5 months of loan originations still misleadingly understates the default rate.
Because Prosper's only loan product is a 3-year loan, and Prosper has only been open for business for about 2 years, we are still a year away from Prosper's first loans fully maturing. Thus, it is impossible to state exactly what the default rate will prove to be. However, a number of estimates by very smart lenders using different methodologies have estimated that it will likely be about 20%. That is a far cry from 4.62%.
There are many other problems with Prosper that should be investigated and reported on by the media (I left a comment on the Contra Costa Times website describing a number of them). Unfortunately, virtually all media coverage to date has consisted of "puff-pieces" that publish Prosper's misleading statistics without any critical analysis. If you are interested in looking deeper into Prosper, there is a huge wealth of information available to you:
www.prospers.org is the largest Prosper forum, where many highly seasoned lenders (a number of whom date back to the beginning of Prosper) discuss various aspects of Prosper. You will learn a lot simply by perusing the posts, and if you post a question, you are certain to receive many very knowledgeable responses. There are a number of people who hang out at prospers.org who can answer any question you might have, including slicing and dicing Prosper's data almost any way imaginable.
www.lendingstats,.com and
www.ericscc.com use Prosper's own data to provide an enormous amount of easily used statistical data about Prosper, including a list of every Prosper loan and every Prosper lender, both filterable and sortable by various criteria.
If you have any questions, I would be happy to answer them for you.
Sincerely,
Ira xxxxxxxxxxx (ira01 on Prosper)
