Saturday, May 23, 2009
The curves are "noisy" (ie they jump up and down a lot) and are not as orderly as the curves on the prosper chart. That's because the volume of loans at Lendingclub is still too low to get really good stats.
Lets slide these curves over to a common origin, so we can visualize how common their shapes are...
Just look at where these curves crossed the 390 day line, (ie 30 days after the 360 day line, because it takes 30 days for a loan to become 1 month late) or visualize where they might cross the 390 day line as they extend, and that tells you what fraction of loans went bad in the first year. This is then an estimate of the annual default rate for Lendingclub loans.
Yipes! What you see is that these various cohorts are defaulting at rates between 9%/year and 17%/year. Conclusion: These are pretty junky loans. It seems unlikely that lenders will do well when such a large fraction of borrowers are not repaying lendingclub loans.
The best discussion among P2P lenders occurs at www.prospers.org. See you there!
Saturday, May 2, 2009
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.
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, and in forum discussions on the old prosper forum, now archived at www.prosperreport.com
For the "big picture" you can just look at that last chart. From the point of view of these performance statistics, Prosper.com loans run about 1150 days. (1 month before the first payment is due, 36 months during which payments are due, and another month before that last payment is declared 1 month late) The numbers aren't final for another 2 months, because it is possible for loans to recover during the next three months before the loan is charged off at the 4 month late point. Therefore about 1250 days is the final point. Look at the curves, and extend them to where you believe they will be at 1250 days.
Looking at the first year of Prosper.com loans, it is pretty clear that they're headed toward something like 42% bad. Loans that originated later (the shorter curves) are doing a bit better. They might end up around 35% bad. Big picture: On the average, these are junk loans. Caveat emptor. Any investor who doesn't realize this will do very poorly loaning money to folks who about 40% of the time don't pay you back!
This month, after a several month hiatus, Prosper has done more "unoriginations". For those of you who haven't followed these statistics for a long time, this requires some explanation. On the Prosper performance web page, they show a number labelled "loans originated". One might assume that this is the number of loans that were originated. It is not. This number goes down over time. Loans are from time to time removed from the set of loans described by these statistics. I track this, and count loans that are removed as having gone bad. (The name "unoriginated" comes from a word I learned in the third grade. In the old Sovient Union, people who were an embarassement to the government became "unpeople". Record of their existance was simply removed.) It is important to understand that these unoriginated loans are not loans that paid off, or are late, or have been charged off. Those categories are explicitly shown. The unoriginated loans are never explicitly shown. They just disappear.
The numbers I chart above are ratios. The numerator is the number of loans that have gone bad, and the denominator is the number of loans originally created (originated). Although Prosper reduces the number "originated" in their display over time, I keep the number constant, the original number of loans made. A ratio with a constant denominator is easy to understand as it evolves over time.
Lets look at an example. Consider the loans Prosper originated in October 2006. Initially there were 743 loans in this cohort. That's the number I use as a denominator. Starting in April '07, Prosper showed that there were only 742 loans originated in Oct '06. History changed. Starting in August '07, Prosper showed that only 740 loans had originated in Oct '06. Starting in March '08, Prosper showed only 736 loans originated in Oct '06. History keeps changing. In July '08, Prosper changed the number back to 740. Some unoriginated loans suddenly un-unoriginated themselves. I couldn't make this stuff up. In May '09 Prosper changed the number to 738. Here and there a few loans seem to come and go from the database. Nevertheless, I hold the denominator constant at 743, the number of loans actually created in October 2006. The latest statistic for Oct'06 is then calculated like this: The numberator is 16 loans in the 1-3 month late category + 281 loans charged off + 5 loans unoriginated = 302 bad loans. The denominator is (of course) 743 total loans. Finally 302/743 = 40.65% of the Oct'06 loans have gone bad so far.
Oct'06 is not the only month to have recent unoriginations.
Nov'06 originally 701 loans, in Apr'09 they showed 698, in May'09 they show 697.
Dec'06 originally 969 loans, in Apr'09 they showed 965, in May'09 they show 964.
Mar'07 originally 1161 loans, in Apr'09 they showed 1160, in May'09 they show 1159.
Apr'07 originally 1101 loans, in Apr'09 they showed 1098, in May'09 they show 1097.
Jun'07 originally 951 loans, in Apr'09 they showed 950, in May'09 they show 947.
Jul'07 originally 935 loans, in Apr'09 they showed 934, in May'09 they show 932.
Sep'07 originally 758 loans, in Apr'09 they showed 758, in May'09 they show 757.
Oct'07 originally 928 loans, in Apr'09 they showed 928, in May'09 they show 927.
I could go on, but I'm sure you get the idea. What causes these loans to be removed from the data? In the early days of Prosper, there were some cases where the loan origination numbers went down like this, and we were told that it was due to loans where identity fraud had been discovered. Prosper decided to treat those loans as if they never existed. There was some merit to their position, because at that time Prosper repaid the lenders (investors, note purchasers) in full for these loans. However, in later years, although I have been in discussion with a large number of lenders, I have never heard of Prosper repaying a lender in full for any loan. Therefore, I doubt that admitted identity theft is the reason now. (Prosper employees, feel free to add a comment to this blog explaining where these loans have gone.) Whatever the reason, this is a detail that adds complexity to the task of understanding loan performance. I wish they wouldn't rewrite history.
A few days ago Prosper ended their self-imposed quiet period, and has started originating loans again. They've changed the credit score limits, and a few other things that seem likely to change the statistics a bit, but it doesn't look like they've done anything to fix the structural problems which I have often discussed. Therefore I expect that as in the past, a large fraction of the loans will go bad.
The best discussion among P2P and Prosper.com lenders always found on prospers.org.