Prosper Referral Program Ending, Better Act Now!

You've got until June 30, 2008 (4 days from today) to register with Prosper and take advantage of the bonus for signing up as a lender.

Nobody knows what will happen to the referral program as of now, but if you want to be sure and get your $25 lender bonus, here is your chance. Act now, you can join Prosper here.

Subscribe to this blog's RSS feed

Its been a topic of conversation among blogs and message boards: are there any institutional lenders using Prosper? Is it even allowed?

Well, the answer is yes, they are allowed and as of now, they are being promoted. Prosper allows asset managers, hedge funds, and other institutions the opportunity to use peer-to-peer lending. To some this may take away some of the steam of people lending to people, but the reality is that for the Prosper marketplace to survive long term, there needs to be more volume.

An interesting aspect of this is that institutions can leverage Prosper's openness of data to create their own lending models and fund loans accordingly. It's not known if there are any such institutions which have already become lenders but it would be interesting to see if these kinds of lenders can actually be attracted.

Prosper.com is now providing chapter and filing date information for bankruptcies filed by prospective borrowers. If a Prosper loan listing appears where a bankruptcy has taken place, a new set of fields will appear with details of the incident.

This is a positive developement because up until not we were always left guessing as to what the public records on the loans were and often times you'd see a question for the borrower asking what it was.

In some cases the chapter and filing date for the borrower is not available in which case you'll just have to ask.

If you want to read a great book that will turn a lot of ecomonic theories on their heads, read Freakonomics. It's a great book. The author, Stephen J Dubner, is apparently a fan of Prosper and makes some interesting observations about what he sees in the marketplace.

The article is called "Adopt and Prosper?", and points out how lenders are often swayed by a good cause.

Prosper has done a lot of work to make data transparent. Increasingly economists and credit analysts have started to study the Prosper model.

Economists Ginger Zhe Jin and Seth Freedman of the University of Maryland have just published a study looking at Prosper since inception to determine average returns.

The most interesting finding is that the probability for a loan to default peaks at around the 10th month and then goes down from there. Prosper's average portfolio as a whole is currently 9.7 months which is the peak of the curve.

The study also showed that the best returns were seen for loans whose interest rate was up to 25%. The implication is that borrowers willing to pay the higher rates are riskier. Maybe an obvious conclusion but there are many folks that come to Prosper seeking only high risk loans and quickly get into problems and become dismayed. Balanced portfolios do better.

The study, titled "Dynamic Learning and Selection: the Early Years of Prosper.com" is a very interesting read for anyone interested in the metrics behind lending.

It looks to me like lender rates have moved up a bit. Certainly not a scientific study but I noticed the rates on my B loans in May seem to have finished higher than recent rates.

Five B loans in May: 13.55%, 12.40%, 15.95%, 13.55% and 13.50%

Seven B loans in April: 12.40%, 12.40%, 12.40%, 14.60%, 12.40%, 19.10%, and 13.80%

Didn't make many A loans but here is the data

April A loans: 10.50%, 10.50%, 12.70%, 15.49%
May A loans: 11.29%, 11.10% (all in the beginning of the month)

Two thoughts:
1. perhaps the WebBank expansion to all states has resulted in the same lender dollars going after more loans.
2. wondering if there might be and end of month effect where lender dollars dry up towards the end of the month and therefore rates are higher.

I'm going to do some more investigation over the weekend to see if I can dig up more concrete data to either support or refute my observations.

And the big news! Those payments on the two 4+ month late loans I wrote about yesterday continue to show payments, now they have been credited to my account: $1.51 and $1.47 I do hope those payments stick! Not that I have any hope of these becoming current after all this time but any payment is like found money.

There is a forum thread over on prospers.org that got me thinking about why lenders have to depend on third party sites for their ROI figures. And I'm not talking about the ROI of other lenders but your personal ROI. Prosper has all the data necessary to provide each lender with a snapshot of their ROI at any point in time. Now we could probably argue endlessly on how to calculate the ROI, how to treat late loans, accrued interest, etc.. But for most lenders a simple ROI based on cash flow would give them a reasonable number that would make sense to most folks. Prosper has all the numbers, they know how much was deposited and what was withdrawn from an account. Borrowers either pay or they don't, none of this accrued interest on an account that hasn't seen a payment in months. Principle balance could be carried at full value until it hit the limbo state of 4+ months late at which it could be estimated at something like 5% of outstanding principle (or zero percent). People understand cash and to some extent cash flow. Why do we have to track ROI ourselves and/or rely on third part sites for this information?

Well it appears there has been a hiccup in the bad debt sale. This post by Doug Fuller over on the official Prosper blog lays it out, but the simple explanation is that no one wants to pay very much. Take a look at the results from previous bad debt sales:


there seem to be a couple of trends:

1. number of bad loans being sold each time is increasing
2. Rates are decreasing

At the moment there are around 1,100 4+ late loans on the books. Not all of these would qualify for the sale but it is pretty clear that this next sale (whenever it happens) will be the largest yet. A prudent lender would figure on 0% returned from bad loans going forward.

As long as these loans (I've got 6 now that are 4+ late) are sold before the end of the year I'll be able to write off the remaining principle. I find it annoying to have to go look at each of these loans whenever I want to calculate my ROI to subtract the loan value from Prospers figures. Why can't the lending overview screen show us the loan value and principle for each category of late loans? (yes it would look bad)

There is a bigger issue though that has folks over at prospers.org worked up. By delaying the bad debt sale the performance numbers on Prosper look better than they actually would be if the bad loans were sold off. One might argue that we don't know what they are worth until actually sold but Prosper has 8 separate opinions at this point (the bids from the bad debt buyers) and apparently the recent bids make the previous debt sales look generous. Time for Prosper to "mark to market" the bad loans and come clean about what returns actual lenders are able to achieve.

4+ month late loan count today (5/3) according to LS: 1,177

In case you haven't noticed yet, Prosper updated the Portfolio Plan slices as part of last nights upgrade. But a new feature described on the official Prosper blog indicates that the updates are not automatic so you won't be caught by surprise by the changes (good change).

One interesting change is the max DTI is now 40% or less for all slices except for slice #4 of the moderate portfolio which stays at 70% ( And a very odd slice, anyone want to bet on how long that lasts ?)

There were other slight modifications, a new slice thrown into balanced, tightened or added criteria, and almost all slices had the min bid lowered. 5.9% if you are a clean AA and ask for less than $10,000.

So where are the opportunities? It used to be one could underbid the PP slices but Prosper seems to have closed that gap up.

In the AA credit grade it looks like one could underbid the Balanced PP slice 1. This slice bids on clean AA 0-0-0 with 0-1 inquiries and loan amount of $15,000 or more. Maybe stick to the listings with zero inquires and bid 11.49%

Same thing for A credit grade, moderate slice 2 bids 16.15% on 0-0-0 with 0-1 inquires.

The B slices are interesting in that Prosper has added a max revolving credit criteria for two of them, balanced slice 5 and moderate slice 3. Perhaps accept slightly more than the $25,000 limit and lower the number of inquiries. There also seems to be a good amount of overlap in the B grade slices.

The C slices don't seem very interesting unless you are a borrower, 10.45% even with 2 current delinquencies as long as DTI is at 40% or below and inquiries is at zero. $7,499.99 is yours for the asking!

As for the D slice (actually the same criteria in both moderate and aggresive) I won't be bidding but it looks like one might underbid and do ok.

I skipped over the auto-fund only slices (three of them, one each in AA, A and B). Looks like a good deal for an A borrower that has some skid marks on the credit. If your DTI is 40% or below and you only want $4,999.99 or less it is yours for 13.10%. Moderate slice 1 doesn't specify an criteria for current or past DQ's, public records or inquires!

Overall I think the portfolio plans are getting better.

This is another interesting chart using the same data from the post on Sunday. The Y axis is the expected loss and the X axis the lender rate.

Ignoring the credit grades and lines for a moment it sure looks like lender rate is a good predictor of expected losses, as lender rate goes up so does the risk (mostly).

And the AA and A data points make sense and seems to be saying that at the same interest rate an AA will perform better than an A. The AA dots seem to be clustered in the 8%-9% lender rate with an expected loss of 1% and the A dots seem to be in the 9%-11% lender rate with 3%-4% expected loss.

The B dots are strange just like the previous chart, a line moving to the right and down makes no sense.

And then there are the C dots, the trend line seems to fit really well just like the previous chart and it is pretty steep! It seems to be saying that C loans below 15% are pretty good. Just imagine the AA and A trend lines continued up and to the right and intersect the C trend line. (Down and to the right is the good quadrant)

Some thoughts on what all this means for lenders:

1. AA and A can be bid on by the numbers. For 0-0-0, 0-1 inquires and DTI <= 40% a higher loan amount has more risk. A borrowers are more risky than AA.
2. C listings seem to benefit from scrutiny beyond the numbers. Maybe it is spelling and grammar? or the Q&A. wait for listings below $7,500 that are bid below 15% ?
   
3. B listings are the middle ground, there seems to be a sweet spot in the $8K to $10K range and in the 12% to 14% range.

Not sure if any of this can be translated into a usable bidding strategy. And in a few months the performance data may tell a completely different story.