Sunday, June 25, 2006

Appropriate levels of diversification for lenders

I frequently hear people discussing the inadequate diversification of various lenders' Prosper portfolios.

Here's the thing -- other than for tax/mental bucketing/liquidity reasons, who the heck should care if your Prosper portfolio is inadequately diversified?

Leaving aside the issue of what "inadequate" diversification means, rational investors should only care about their total portfolio profiles. Insofar as my Prosper portfolio is horribly unbalanced, but its risk/return/correlation profile makes my entire portfolio higher return, lower risk, and less correlated with bad states of the world, say, I'm still better off. A separate question would be whether an even better total portfolio outcome would be possible with a different Prosper allocation strategy.

I could see Prosper trying to keep lenders diversified so people don't start causing mega negative publicity when their Prosper only portfolio goes underwater due to one default, but other than that...

[Inspired by an exception to pninen's typically well thought out (though equally frequently overly harsh :) ) set of posts...]

Read more!

How I Look at Current Delinquencies, from traveler505

A post from traveler505 on the forums recently caught my eye:

A BK followed by current delinquencies is absolutely a red flag, unless the "now delinq" can be shown to be a reporting error. It is not uncommon for an account that was included in BK to be reporting as "now delinq", so I don't always believe the numbers in the scorecard.

When there are current delinquencies but no BK, I look for evidence that the borrower did what I call a "paperless bankruptcy". In a "paperless bankruptcy," a debtor who could and probably should have eliminated all of her debt in BK, simply decides to stop paying all or most her existing creditors, and then waits for the accounts to get old enough that she can begin applying for new sub-prime credit and gradually rebuild her credit rating by handling the new accounts (and any of the old accounts that survived the "paperless bankruptcy") responsibly.

[Edited to clarify: I think the initial decision is usually to stop paying and hope for the best. The decision to start rebuilding usually comes later.]

A "paperless bankruptcy" is probably harder on the debtor than BK (since they will be dealing with continued collection activity, and the rebuilding process will be slower), and a bit better than a BK for the old creditors (since some of the debts will likely get paid, or be converted into enforceable judgments, rather than all of them being discharged). It is possible that the choice of "paperless bankruptcy" signifies a desire to pay the old creditors once the debtor's financial house is in order, though I suspect that the repayment of old creditors rarely actually occurs, for a variety of reasons.

However, when a borrower who declared BK 2 years ago and a borrower who did a "paperless bankruptcy" at the same time both arrive at Prosper, their scorecard will look very different, even though their actual payment history is similar. The BK will show one "public record" but no "now delinq" on his scorecard, while the "paperless bankruptcy" will show many "now delinq" -- one for each of the old creditors that she stopped paying two years ago. All else being equal (including progress on credit rebulding), I regard these two borrowers are nearly equal risks, the only difference being that the BK borrower is prohibited from filing BK again for a period of time, and that the old creditors may successfully garnish the wages of the "paperless bankruptcy" borrower, negatively affecting her ability to pay new debts (and suggesting that I should look more closely at DTI). (My observation is that two-year-old chargeoffs/collections are fairly unlikely to result in garnishments or executions against bank accounts, but others may have a better set of data.)

This is why I complain so vociferously about the lack of chronological information for the "now delinq" number, and the failure to differentiate between open and closed accounts in the "number of accounts" line. If there are open credit lines, and no account has become delinquent in the past 2 or 3 years, I suspect "paperless bankruptcy". If accounts have joined the "now delinq" field recently, I see a red flag and move on.

I do have an answer to the question of how [] helps, but I'll save that for another time; I've been long-winded enough. [In case anyone is wondering, I've never seen "paperless bankruptcy" advocated as a strategy there or elsewhere; I offer it here purely as a sociological observation of debtor behavior, including my own past behavior.]

Read more!