A couple of days ago Lending Club added this new page to their site that included the graph above. I found this extremely interesting. For years they have told us the average returns of the different loan grades on their statistics page but this is the first time they have provided a range of returns within each grade.
Here is the fine print that explains the top 50% and bottom 50% numbers in the above graph:
3 – Top 50% and Bottom 50% are dollar-weighted averages of individual loan performances for each grade calculated from either the best performing half or the worst performing half, respectively, of all issued loans in each grade as of May 25, 2011. In practice, if a portfolio of Notes was created from the worst performing half of all issued loans in a specific grade, the return would be roughly equivalent to the Bottom 50% of that specific grade as depicted in the chart above. Conversely, if a portfolio of Notes was created from the best performing half of all issued loans in a specific grade, the return would be roughly equivalent to the Top 50% of that specific grade as reported in the chart above.
Obviously the top 50% of loans in all loan grades would include no defaults whatsoever, and the bottom 50% would include all the defaults. As you would expect the average return for the top 50% of loans within each credit grade increases as the interest rate gets higher. But as you can see for the bottom 50% of loans there is no opposite trend. I would have expected bottom half of the G loans to perform badly, I am actually surprised they managed a positive return, but what is fascinating is that E rated loans have the best performance when taking the bottom 50% of loans.
What this chart tells me is that investing in A rated loans is not a recipe for safe returns. Sure there is far less deviation in returns for A rated loans but both C and E rated loans perform better even when taking the bottom 50% of all loans. I may tweak my selection criteria after seeing this chart and load up on more E-rated loans and lessen my exposure to D-rated loans. Based on these numbers any way you slice it E-rated loans are a better bet than D-rated loans.
As always I am interested to hear your thoughts.
There are 2 ways to explain the D bottom 50% number.
1. Just random luck. Each E loan is expected to be more likely to default than each D loan. However over a given sample size, these default rates can’t be predicted exactly and have room for fluctuation. It looks like significantly more D loans have defaulted than predicted. This could just be due to a bunch of D’s being unlucky and losing jobs or something uncontrollable.
2. Lending Club is placing some people in the wrong categories. Some less qualified people are getting D rates that they don’t deserve and should actually be in the higher risk E/F categories.
Peter – since you plan on buying more E’s and fewer D’s, you think that #2 is a likely reason?
(btw, if you don’t think #2 is likely, then you are in #1 territory and you are following the same logical path whereby a roulette wheel comes up red 3 times in a row and you bet on red since you think its more likely)
@Brian, Interesting points. I was assuming #2 but upon deeper investigation I am not sure if that is even true. I did a little study of default rates of D and E grade loans and this is what I discovered:
All loans Grade D – 5.2% default rate
All loans Grade E – 4.1% default rate
Now when you look at loans that are at least a year old you get a different result.
Grade D loans originated before 6/30/10 – 10.6% default rate
Grade E loans originated before 6/30/10 – 12.0% default rate
This is more what you would expect. So maybe I was premature in my suggestion that E rated loans are better. Newer E rated loans certainly seem to be performing better than D rated loans but as time goes on it looks like the reverse is true.
My first instinct was that the bottom 50% stats for D/E were just due to luck, but the more I think about it – I don’t think so anymore.
-This chart is based on tens/hundreds of thousands of loans. sample size is too big for the numbers to be skewed by a few outliers as was my initial instinct.
Looking at those 2 dots, I am convinced that Lending Club either now or in the past has made significant errors in assessing risk at the D/E level. Impossible to tell just based on this graph whether D is too low, E is too high or both, but there’s no way the points should line up as they do.
(based on your 5.2/4.1 vs 10.6/12 numbers, it certainly looks like they tweeked something during the past year and have been giving people D ratings that should have been E/F)
The main question for us investors is this: Has Lending Club recognized this error, found its source, and corrected it? If it hasn’t already, will it be able to and when? (ie, have they found out why they are getting more D-loan defaults than they expected or fewer E loan defaults than expected, and have they adjusted their loan rating criteria based on this for new loans?)
->If they have adjusted their loan rating criteria, then any action shifting one’s portfolio makeup based on this graph is moot as the future loan ratings aren’t directly comparable to the past anymore. (and might even be bad if LC overcorrects on any actions they take)
->If they haven’t adjusted their loan rating criteria, then I completely agree that E is a far better place to be putting your money than D.
They always had this information here: https://www.lendingclub.com/info/statistics.action the bottom left, although not sure what the difference is, as I see the numbers are different.
Oh I see that these averages are ffom May 25th, interesting that thay changed so much in a few days…
@Brian, We can only guess as to exactly what changes Lending Club makes to their risk management, they don’t make those things public. But if you follow along closely you can ascertain a couple of things:
1. They change the credit policy from time to time. If you look at the full data download you will see a bunch of loans that say “does not meet the current credit policy” in their status field. These loans would have presumably been rejected outright or their loan grade would have changed if they were applying for a loan today.
2. They change their interest rates. Lending Club did not make any announcement but some time in the last 2-3 months their rates went up 1.5% – 3%. So a rate of 16% used to be a E1 rated loan and now that rate is a D2 or D3. This tells me that they have realized they were underpricing D-rated loans and they have adjusted accordingly.
So, from this one could surmise that the graph shown at the start of this article, while interesting, will likely change a great deal in coming months.
@Moe, I know from past experience that a few extra defaults (or a sustained time with no defaults) can change around the total NAR for each grade considerably.
I’ve decided to stay out of this whole default numbers discussion this time. It’s all just so way over my head 🙂
Couldn’t help looking a few paragraphs below the chart though & skimmed through the “Who Invests in Lending Club” writeup. Noticed the total number of investors listed at 26,000. Does that number sound low to anyone else?
10k per investor? After the LC-funded stuff gets pulled out, it may be right? BTW, hi everyone. Started investing on LC a month ago. Thanks to Peter, Ken L at lendstats.com, and all the active commentors. I hope to be one myself, lets see if my first loan gets its first payment (due tomorrow!).
@Dan, Very funny. It does sound a little low but with around $188 million in outstanding loans (according to Lendstats) that is only $7,200 per investor. I read an article a few months ago where Lending Club said they were hoping for an average account of over $10,000 by the end of this year. What it tells me more than anything is that their market penetration as far as investors go has a lot of room for growth.
@Larry, Welcome to peer to peer lending and thanks for joining the conversation. I think you might have been thinking about the total loans originated to come up with your $10K per investor number, but a large amount (around $100 million or so) of the total $289 million originated has been paid back. Good luck with your first payment.
Larry – don’t panic when you don’t get paid tomorrow. When the scheduled date hits, the payment goes from “scheduled” to “processing”. It then sits there for 4-6 days normally before it is actually received. It takes several days to process, and the number of days also varies due to weekends and holidays.
Peter and Brian as someone who has been in the subprime lending business for 25 years, I have always felt that Lending Club underestimates the long term default rates on the D, E and F loans. For all loans but especially subprime, defaults are not evenly spread over the course of the loan. The longer the loan is outstanding the higher the chance of default. With Lending Clubs very high growth rate, they simply do no have enough experience to accurately estimate default rates. I assume that they are projecting based on models from other loan platforms. But as Proper demonstrated the internet attracts a higher defaulting customer than brick and mortar certainly for the subprime. So I would be very skeptical of the actual yield being as high as are projected in this chart.
@Steve, I think Lending Club and Prosper would vehemently deny they have subprime borrowers. All borrowers at Lending Club must have a FICO score of over 660 (and there are more restrictions for scores less than 720) and Prosper is is 640 and above. Regardless of what you call them it is clear the riskier the borrower the higher the default rate.
I wouldn’t assume that because they are young companies that are growing fast that they don’t understand the risks. Both companies know that the key to their success is risk management. And they are improving their systems as they go. Many borrowers who made it on to the platform two or three years ago would have no chance today.
Having said all that I would agree that Lending Club overestimates the actual return for investors. I have written about that before and it is something all investors should be aware of. The numbers are skewed high because of the rapid growth and a large chunk of the loans are too young to have any chance of defaulting.
The simple truth is there is very little personal downside in underestimating risk/defaults. Therefore I’d like to propose that Lending Club install a pillory outside their new headquarters. Given their history, Prosper should install 2 or 3 outside theirs, just to be safe.
When unacceptable/ridiculous levels of defaults occur, these p2p companies (who after all pride themselves on transparency), will immediately notify us lenders, thereby affording us the option of swinging by their offices & expressing our, shall we say, displeasure to the people who signed off on those loans. If the offenders wish to avoid the ridicule & punishment of the stocks then they may pay restitution to the lenders in question out of their own pocket.
Now that I think about it I believe a pillory should be placed outside every major bank & subprime lending company & the punishments should be retroactive to before the financial crisis. Considering that next year is an election year I think that this idea can gain a lot of traction with the masses.
Steve S………..So you worked in subprime lending for 25 years huh? I’ll probably be needing an address for you as well 🙂
@Dan, I will give you this, your ideas are always out of the box. Of course, your pillory idea could only work if there was a webcam setup so those of us not in San Francisco can view the whole thing. Although this may be considered cruel and unusual punishment…..
Seriously, though, I believe there is a definite downside for the management at Lending Club and Prosper if they underestimate defaults. Unlike the banking institutions you reference, there will no huge bonuses flowing to the management teams just for “making sales”. The only rewards will come from a sustainable and successful business model which means estimating defaults accurately. Both companies are acutely aware that if we go back to the default rates of the Prosper 1.0 days then peer to peer lending will likely disappear from this country. But that scenario is extremely improbable in my opinion.
Sure I’d agree that there’s going to be consequences in the case of p2p. But remember that just the threat of punishment is often more dissuasive than the punishment itself. Placing a pillory outside the doors of the office will be a daily reminder of the consequences of failure. 🙂
As for cruel & unusual punishment I’m sure that I could round up a boat load of people at prospers.org who would support my proposal & would in fact contend that they were defiled in every way imaginable during Prosper 1.0……………& that this punishment wouldn’t even compare to what they went through.
@Dan, An investment loss versus public humiliation and physical pain….hmmm not sure about that. But from your logic then investors should demand similar punishments to the management at Blockbuster, General Motors, Kodak, AOL (not even mentioning the financial institutions), etc as well as the brokers/advisors that recommended them.
Right on, I’m all for that. I want accountability! When you screw up you pay………..one way or another. Don’t want public humiliation? Then don’t make ridiculous self serving recommendations. Or pay restitution in lieu of the public humiliation………….your choice.
All I have to say is that when this proposal becomes part of the Tea Party platform, you’ll know where it came from.
@Dan, Ha! I will keep a look out for the Tea Party copying your idea….