Prosper loan modeling

Apropos my last post, I just found p2ploananalytics, which does exactly the modeling I was thinking about. Lessons from their data:

  • Default rate is almost zero for the first 6 months, then rises in a roughly linear fashion, more linear for the higher credit grades (graph). This means that for portfolios with many loans younger than 6 months, projecting current lateness rate forward is erroneous.
  • The loss severity is highest for earlier defaults (graph), but since severity and default are both roughly linear, assuming defaults happen at the midpoint (21 months) and with average severity (70% loss or so) should be reasonable.
  • The various features are far from equal. Specifically, these graphs suggest that loan amount is a small factor, DTI is a small factor, monthly income is a large factor, and number of credit lines / length of credit history / number of inquiries / homeownership are huge factors.
  • A different analysis finds that late loans tend to have high interest rates for their credit grades. That is unsurprising - if someone is risky for their credit grade, and lenders can tell, they should have to ask for more. The question is whether the extra interest is enough to compensate for the higher delinquency rate (as several people pointed out on the thread).

One thing I don't understand is "prepayment risk" or "weighted average life". Given that you can just re-lend the money to a different borrower as soon as it is paid back, I don't understand why there is any cost to the lender to having the loan prepaid, other than having to scan the listings again. Your money is always out there earning x%, whether or not there is a prepayment. The only issue I see is that if interest rates drop overall, your borrowers can prepay and you will be lending out at a lower rate - essentially you bear all the interest rate risk. But that doesn't seem to be what the site is talking about, ie "Due to the early payoff (prepayment), the interest generated from a loan is smaller than what is implied by the amortization schedule."

Anyway, I definitely expect Prosper to be a fascinating data source in the years to come. What I'd really love to see is Prosper for startup investment - venture capital, rather than loans.

Another fun thing I've been thinking about is using Prosper for leverage, given that I have great credit - for example, borrowing money there in order to put it in the stock market.

Share this