mothandrust's blog

Prosper lenders who rely on Social Collateral to screen listings continue to outperform those who lend "by the numbers". I had blogged on 3/18/08 about how selecting loans with 0 current DQ’s and 2+ bidding friends who bid $300+ in aggregate would produce a portfolio of 48 loans, all current. But how would a portfolio look constructed with that Social Collateral screen...28 weeks later...?

Loans originated between 2/12/07 (the start of the Extended Data) and 6/1/08 (to allow new loans time to make payments) would give you 75 loans: 66 current, 7 paid off, 1 charged-off, and 1 late.

My personal portfolio does not look anything like that—I have half the loans but three times the lates!

Then I had the epiphany that I should be selecting borrowers rather than selecting listings.

The aggregate 0 DQ Prosper loans over the same period show 595 charge-offs (out of 10963 loans originated) and 371 lates, so that’s 5.4% charged-off and 3.4% late, compared with 1.3% for charge-offs and lates in the Social Collateral portfolio.

Why does Social Collateral work? The principle is the same as considering insider buying of common stocks. If you notice that several directors are all buying stock on the open market, it tells you that that they might know some good news is forthcoming. On the other hand, if you see the CEO and the CFO dumping large numbers of shares, you might not want to be buying what could be the next Enron.

Let’s look at the 2 bad loans out of 75 from this screen and see what secondary screening might have caught them. 

The losers share the following behavioral characteristics:

-The Borrower hides his City

-The Endorsers fail to state the rate(s) and amount(s) bid

-The Endorsers fail to state their relationship to the borrower and its duration

Certainly there are some conventional credit factors one might bring up--high utilization, large loan amounts, etc. but here we are only discussing social factors, since each lender has his own endemic numerology.

This first loan from the Cashfunders group is not even a true multiple BE loan, but is only one of the 75 because Prosper’s performance utility tracks only the amount bid and not the amount winning:

http://www.prosper.com/lend/listing.aspx?listingID=296216 

The borrower has only one real-world "friend" (the relationship is not stated), Kenkobiz, who is on the loan for $100. Cashfunders, who had no previous real-world relationship with the borrower bid $8800 right from the start, but these were all bid off when the rate went 21%.—only $77.69 remained on the loan at the end. Katenvironment’s $125 bids never returned.

There is no social mojo that a stranger’s bid brings to a listing, and I suspect that many of these Cashfunders loans will go the way of the vetted group listings of Prosper’s yesteryear.

This next listing does fit the Social Collateral screening criteria and shows that this criteria by itself is not sufficient to weed out terrible listings. Doesn’t your gut tell you not to lend this borrower "working capitol" to flip a house? Heads you get 10% on your investment, tails you lose it all:

http://www.prosper.com/lend/listing.aspx?listingID=243243 

Recently I have been urging borrowers whose listings feature Bidding Endorsements to take the Transparent Borrower’s Pledge: https://connect.prosper.com/t/3885.aspx although most refuse to take it. Pledge-takers are required to disclose any early prepayment intentions, display their City, put their endorsements in the accepted format, and answer publicly all Questions put to them in the Q&A.

The TBP is substantially what the defunct Sincere Borrower’s Pledge required, only there is no requirement from borrowers to contact late borrowers, and it is open only to borrowers with 0 current DQ’s.

I would have to warn you about combining the TBP with the Social Collateral screen; if you do this, you will find, as I have, that no listing has come forth in the last 3 months that qualifies!