Wednesday, May 03, 2006

Why removing higher risk borrowers from Prosper would be a bad decision, and why my personal portfolio has little room for Zopa

Someone recently proposed that we remove higher risk borrowers from Prosper altogether. I disagree (see Excel file at end of post as well).

1)Consider further segmentation of the HR bucket

2)Eventually, raise rate caps

3)But first, increase overall lender education. Encourage social grouping or something similar for lenders, which hopefully will reduce the degree of ill-considered lending (quasi-gambling, in many cases -- not that there's anything wrong w/gambling, unless you consider it investing, and aren't a professional -- which seems to be the case for many a lender).
I suspect this may have to be a non-Prosper sponsored entity (such as the forums I recently opened to all lenders who made a successful bid in response to recent forum events), given the understandable and necessary desire (business-wise, anyway) for to regulate free speech and err on the side of censoring impolite and hurtful, even if legal and efficiency-increasing behavior. There are routine and enormously risky/seemingly ridiculous patterns of behavior on the part of novice lenders that will come back to bite Prosper in the butt unless they are tempered.
The rate cap rationale seems to have failed (that is, force people to only bid on HR's that have considerable group-based social enforcement, or else expose more information to lenders by implying a negative rate of return if one were to bid blindly on the average HR loan) -- new lenders still seem to bid blindly on many HR listings as long as they put out the max rate -- even if said rate implies a negative return for such lenders, before even considering the effects of adverse selection.
I've even considered greasemonkey scripts, or something similar, for loan listings (see below).

Quoting dave from Zopa on Prosper
There are, without question, good HR borrowers - but finding them, without accidently lending to a 'bad' HR borrower is incredibly hard. And you only need 1 defaulting borrower to take down all your returns from many fully paying back borrowers.

A single HR borrower defaulting doesn't much affect the returns on the rest, assuming rates are high enough and diversification adequate, any more than if a single hi-quality borrower defaults (which eventually will occur, Zopa's record notwithstanding -- see below). In fact, if anything, a single high-grade borrower defaulting has a far worse effect on an equally sized pool, given the much thinner margins offered to high grade lenders (even if the likelihood of such an event happening is that much less).

Dave's claim seems silly to me -- given a portfolio of 10 identical $1000 loans, w/1 default at 18 months, using a fairly hacky model I threw together in Excel:

1)at rates of 35.75%
Original funding: $10000
Total $'s returned assuming no defaults
Total $'s received:

2)at rates of 23.75%
Original funding: $10000
Total $'s returned assuming no defaults
Total $'s received:

See excel file here above for details and approximations. Please DON'T HOTLINK!

5)Social and economic benefits for lenders:
Comments from dave of Zopa on Prosper:
I think norcal_cct is right - you can get good returns without HR borrowers, and the risk : return trade off is more favourable. Lending to sub prime borrowers is a very, very different ball game from lending to prime borrowers. Many financial institutions have been burnt badly by trying to play that game - and we didn't want our lenders to get torched!

The higher risk marketplace -- D/E/HR -- offers a much different risk/reward/correlation payoff than the higher grades. Furthermore, the higher grades are dominated, Sharpe wise, by alternative investments. (Even w/respect to correlations, there are ABS' and even more retaily products that give exposure to the same risk factors.)
The Zopa model would be basically useless to me personally, though useful to certain sets of investors who have capacity constraints or financial sophistication needed to pursue alternatives.
In addition, the payoffs accruing from high-grade lending are not as bad as, but close to, those of selling deep out of the money put options -- lots of high return, low risk payouts for long periods of time, but w/massive negative events every once in a while that easily crush all of the previous performance -- and on a Sharpe basis, look far worse than a large basket of (non-adversely-selected) higher risk loans. (In bad states of the world, HR's may also default much more, and are more suceptible to bad events -- but the mathematical multiplier on default rates is going to be much higher for higher quality loans. Furthermore, people will be far less emotionally and mentally prepared for such defaults in the high grade bucket, vs the higher risk buckets.

6)Social and economic efficiency for borrowers:
From the perspective of borrowers, there are plenty of better alternatives to P2P borrowing in the vast majority of cases if one is high-grade. However, higher risk borrowers have few, or no non-ruinous lending sources to turn to, even after proper risk adjustment.

7)As an example, one of the frequent claims (which may be correct in some cases) made by borrowers after extended credit info was made available was that the summary misstated the actual report. Perhaps a segment of customer service dedicated to vetting such claims would be of help (though there are still issues of things misreported on the reports themselves), but this is an not very scaleable solution.

More generally, even better and useful/open source in spirit, would be a eBay-style Q&A segment on each listing. Some sort of greasemonkey script for each loan page (assuming one uses Mozilla Firefox as a browser, rather than IE), distributed among lenders and made by lenders, might be a good alternative, though it will only increase the gap between tech-savvy/overall better informed lenders, and newer, less sophisticated lenders. Until Prosper better integrates forums w/lenders able to view lender-only info for borrowers, adds Q&A, etc, I continue to support my privately run forum open only to registered lenders as a stopgap solution.


At 4:34 PM, Blogger ProsperCouple said...

I agree with you. One thing that I am thankful Prosper is doing now that it wasn't before was allowing a better "view" into the person's credit history! I think this tool will be the life saver of lender and GL alike!

Keep up the good work!:)


At 4:28 PM, Anonymous Anonymous said...


At 8:21 PM, Blogger cellardoor said...

Basically, once one is a registered lender at prosper:

1.Log in
2.Go to listing in question
3.Under Credit Information, click the dropdown tab (the + sign in the grey box) to view tidbits like:

Now delinquent: 0
Delinquencies in last 7y: 0
Public records in last 10y: 1
First recorded credit line: Feb-1989
Total credit lines: 26
Inquiries in last 6 months: 3


Post a Comment

<< Home