Wednesday, July 4, 2007

Are all Prosper loans within a credit grade created equal?

The easiest way to categorize risk is by credit grade. At Prosper you can search and filter loans by credit grade. Many lenders, including me, pay particular attention to higher credit grades while excluding the lowest credit grades completely.

Today we ask the question: Are all loans within a credit grade created equal?

Let's start with some statistics. The following are the average lender rates by credit grade. Note, this is not the actual interest earned by the lender, just the rate the borrower agrees to pay the lender at the time the loan is made.

Credit GradeInterest
AA10.59
A12.65
B14.91
C17.60
D20.78
E24.05
HR24.03

Here are the same statistics for all loans that are current:

Credit GradeInterest
AA9.43%
A11.41%
B13.83%
C16.50%
D19.53%
E22.98%
HR23.06%

Now, lets take a look at the numbers for the loans that have defaulted:

Credit GradeInterestPercent Above Average LoansPercent Above Current Loans
AAnot enough datan/an/a
A12.85%0.2%1.44%
B16.62%1.71%2.79%
C20.24%2.64%3.74%
D22.65%1.87%3.12%
E25.77%1.72%2.79%
HR25.93%1.9%2.87%

So, what do these numbers tell us? Well, if we look at loans in the B through HR categories, the loans that defaulted had a 1.7% through 2.6% higher interest on average than other loans in that credit grade. What this means is that prior to defaulting lenders considered these loans a higher risk and didn't bid down the interest rate as low as they did for other loans.

The difference is even more pronounced when you look at the difference in interest rates between loans that are current versus loans that have defaulted. Lenders put as much as a 3.74% risk premium on loans that ended up defaulting - clearly lenders were seeing something they didn't like in the listings compared to other listings of the same credit grade.

There are two things that we should learn from this data. The first is that it is probably not a good strategy to look for the highest rates in each credit grade. If you consistently seek out the highest interest rates in each credit grade then you are going to have a higher rate of defaults then if you were sticking to loans that are closer to average or below average for those credit grades.

The second lesson that we can learn from this data is that there are other important pieces of information when looking at a loan. It is important to look at the whole picture including number of delinquencies, debt to income levels, income, public records, and revolving credit balance. Basically you want to ask yourself questions like:
  • Does this person have enough resources to pay back this loan?
  • Does their past credit history show they can be trusted with credit?
  • Does their purpose for the loan make sense to me?

What you will find is that some lenders will stick to the higher credit grades to lower their risk, but then they seek out the loans within that credit grade that pay the highest interest rate to the exclusion of all other criteria. Some lenders, for example, might set a standing order that would bid on loans only if they are at a higher than average % for that credit grade. Then they wonder why they are having a higher than average default rate in their portfolio. The answer is that lenders have allowed them to close at higher interest rates because they correctly assessed that they were higher risk loans in spite of their good credit grade.

14 comments:

RateLadder said...

Rate alone is only a proxy for risk.

Apply proper extended credit requirements and go for higher rates (through autofunding for example) and price risk.

Reap the rewards.

Matt said...

Yes, if you excluded everything else, the interest rate does end up being a pretty good indicator of the risk of the loan. The question becomes which is better a B loan at 16% or a D loan at 16%? In many cases I think the D loan will be a better choice since it was bid down to a below average rate for the credit grade while the B is at a above average rate for the credit grade.

Tom said...

Very interesting analysis Matt. How do you recommend using this information since the final interest rate is not known until the close of the auction?

Mike said...

I don't find this surprising. If we assume that lenders are doing a reasonable job pricing loans, then the interest rate is proportional to the default rate (I made such an argument and have stats to go with it). I'd then expect the default group to have a higher average interest rate.

I'll see if I can work up a more detailed thought experiment to illustrate it.

Mike

Matt said...

If a posting is significantly above the credit rating for a loan it usually gets bid down pretty quickly if it is a good loan. If it is not attracting many bids then you should take a closer look at the rest of the stats before bidding.

Matt said...

Mike, you have a really good article on that. It seems as a group Proser lenders are doing pretty well at assessing risk. Let me know if you work out your more detailed experiment. I would be interested to see how the numbers come out.

thisguy said...

Matt, agree 100% on this. About 2 weeks into my Prosper experience I found forums. One of the very first posts I read stuck with me every day I am here on Prosper

0% of 29% = 0%

Point being defaults kill you. Getting paid for 3-6 pay periods out of 36 pay periods, at a rate 3-5% higher than average for a grade is much worse than being paid 2% below the average for the grade for an entire 36 month term. Since I saw that post, I bid down heavily on things I want in, assuming I have correctly identified the correct loans; hence I will pay 19% for a E I truly believe in or 16% for a D, etc. Nice to see the numbers confirming. Usually I don't believe in mob mentality but in this case 'the herd' seems to have gotten it right

thisguy said...

as an aside I just noticed there are 2 of you guys posting to this site, matt and tom, hence I was confused today when I saw a post on forums that was named mateo instead of spyder5 or whatever the other name is :) both with the same signature. You guys need to do an 'about us' link or post. Make that clear that you guys are co-authors of this blog. Good job overall though - I like the analysis.

Matt said...

Thanks for the suggestion, I think we will do an about us page. I am Matt and I go by Mateo for both my prosper profile and on the forums. Tom is my brother and he goes by Spider5 on the forums and for his Prosper account.

Tom said...

Thisguy, very good comments and you're right - we do need to do an about us post. Just haven't got around to it yet. I think we also need to easily identify which blog posts are written by which author (me, Matt, or a guest author). Right now you have to scroll down to the bottom of the post. Maybe add our pictures to the header of the post or something. Of course, that might scare some people away. ;)

Mike said...

matt -

I put my thought experiment up on my blog.

Mike

Matt said...

Mike,

Interesting thought experiment. I am going to sort through some data to see if I can determine what the exact % increase is in defaults per 1% increase in interest rates so you can add that to your thought experiment.

gelos said...

What about the dollar amount of the loan?
I just looked at Prosper's 30 day average rates and within the same credit grade the higher loan amounts have the higher APR.
Would it be possible that the reason loans with a higher than average apr, default at a higher rate is because the loan amounts are higher?
Higher loan amounts of course have higher monthly payments which may be harder to make if things get tight for the borrower.

Matt said...

gelos,

Loans for larger amounts have higher interest rates. I think this is partly due to the inherent increased risk with larger loans. However, I think part of it is also due to the fact that it takes a lot more lenders to bid down a 25,000 loan than it does to bid down a 1,000 loan. So, I think that plays a role as well.