| « Prosper.com - 06/15/08 late loan stats update | Prosper.com - a mishandled loan » |
Prosper.com - 06/01/08 late loan stats update
Here's the June update to to my late loan statistics charts.
These charts show statistics for the performance of all prosper.com loans. Each curve represents the set of loans that were created in one calendar month. The vertical axis is the fraction of those loans that have "gone bad", in other words are 1 month late or worse (up to and including default). The horizontal axis is the observation date. All data comes from Prosper.com's performance web page.
A larger, more readable version of that chart can be found here
Here's a chart of the same data in which each curve has been slid to the left to a common origin. The horizontal axis is now days since loan origination month.
Explanation of methodology can be found in my prior postings in this blog .
You can plainly see that Prosper loans are going bad at about 20% per year. Just look at the 360 day point on that last chart. As I've reported before, you will see many references in various newspaper and magazine articles that the Prosper loan default rate is 3% per year. These statements are just completely wrong. It is amazing how often this incorrect statement has been reprinted. Prosper's own data (as seen above) shows the correct number is about 20%/year.
The most recent repeat of the lie can be found in Lendoza's blog
http://www.lendoza.com/2008/05/28/marketplace-protection-collections-a-closer-look/ where he wrote:
we will assume that Prosper?s numbers are accurate?.in which case there is a default rate of 3%
He obviously hasn't actually looked at the data. He must have simply copied that from one of the newspaper articles. Prosper's data does NOT show "a default rate of 3%". The only polite word for that is "wrong".
Zcommodore recently showed us another way to look at this same data.
http://www.prospers.org/blogs/zcommodore/2008/05/22/loan_aging_a_study_of_loans_over_time
You can read from his graphs a default rate of a little less than 2%/month. That's about the same as 20%/year I've repeatedly described.
An important factor in prosper.com loan results is how well Prosper collects the payments due on these loans. Please see my prior writings on that subject, including:
Written 05/06/07: Collections is broken
Written 05/04/08: Collections is not improving
Prosper has recently advised us that the most recent auction of old bad loans has failed, and as a result they no longer intend to auction off old bad loans. Failed. Dead. Nobody wants to buy the stuff! As a result of this, Prosper has told us that they intend to stop selling such loans, and instead apply "post charge-off collection techniques" while us lenders still own the loans.
This could be a very positive development. The old auction/sale process was very bad for lenders, as no one ever tried very hard to collect before the loans were sold, and the sale triggered very bad tax consequences for lenders. As I wrote about a year ago in Collections is broken I think this is an appropriate thing to do ... IF Prosper takes this responisbility seriously.
My concern, and the concern of many other lenders, is that Prosper will not take it seriously, as they have never taken their responsiblity to lenders with any seriousness. I recently wrote about how slowly Prosper is proceeding with the legal test against 66 deadbeat borrowers. Will they proceed more gusto against 1000 or 2000? Or does "post charge-off collection techniques" mean letting old loans gather dust on a back shelf?
PS: The best discussion among Prosper.com lenders can be found at http://prospers.org/
Trackback address for this post
7 comments
In case Prosper reads this, with respect to those "post charge-off collection techniques," WHAT ARE THEY? Who's going to do it? What training will they have? What would Prosper consider an acceptable recovery? What will Prosper do if/when bad debt goes bk? What will Prosper charge us for this "service?"
Anyway, regarding, "You can directly read from his graphs a default rate of a little less than 2%/month."
Maybe I'm just picking nits, and I'm not disputing the 2%/mo figure. But still, I'm not so sure that's a "direct read" from zcomm's data?
If, say, a loan goes late 5 times a year, then if I understand correctly, the loan shows up in zcomm's data those 5 times. I have no problem with that. But I'm not seeing how the default rate can be "directly read" from data which count the number of transitions into late-ness.
Thanks for the note about the broken link. I fixed it.
You're right. I shouldn't have used the word "directly" when describing z's chart. You have to know some other things to draw this conclusion, so I guess that perhaps the right word is "indirectly". .
Great site.
First of all, I appreciate you visiting the site and lending some feedback.
The default figure I cited was indeed taken based on what Chris Larsen had indicated both in the press and on the Prosper blog. I felt I needed a default figure to bring some perspective to the article, which was not even about the default rate, but instead focused on Prosper's collection efforts & the subsequent concerns of lenders. Needless to say, it was a mistake on my part for including it as it didn't really add anything to the article.
That said, it was not at all my intention to have anyone take the 3% figure as gospel or even to go into any depth about the default rate in that article, which is why I prefaced the quote you cited with this:
"There is plenty of debate about whether or not the default numbers produced by Prosper, Zopa & other firms are legitimately counted and take into effect all variables"
And followed the 3% figure with this, which included a link to one of your prior posts on the topic:
"(we will discuss this issue at a later date but see here and here for explanations as to why real-world Prosper defaults are likely higher than published)"
This of course was the sole reason for my linking to your blog in the first place. I believe you're right on the mark as far as this data is concerned. Based on what I've seen here and elsewhere, the 3% number is only accurate for Prosper's own internal definition of default, whereas based on the real world lender definition, the percentage is much higher. I simply did not intend to derail the original focal point of the article with a discussion of the accuracies or inaccuracies of Prosper's default rate; hence the external links to this blog and others citing similar data.
Anyway, I've deleted the figure from the article entirely from that post so as to avoid any confusion. I hope this clarifies things a bit.
Thanks again for reading.
-Blair MacGregor
Lendoza.com
Comment feed for this post
Comments are not allowed from anonymous visitors.