Print

[Fred93]

Fred93 has unfortunately updated his blog this evening.

This installment examines the recent performance of Prosper's collections department.

http://prospers.org/blogs/Fred93

[NPX]

Sorry if you've already answered this question, but are you tracking number of loans or number of loan-occurrences in collections?

In other words, if a loan goes to collections, cures, and then goes back to collections, is that 0/1 or 1/2?

[onthefence]

Excellent research.

[ira01]

Thanks Fred.  As always, great work.

One thought that occurred to me, however, is that while "cures" are certainly the gold standard for collections, they are not the only thing (or even the most important thing) that matters.  Ultimately, what lenders really care most about is how much money is being collected from late borrowers.  For example, one of my two Late-4's hadn't made a payment since September 5, when all of a sudden it made a payment (of about 120% of a monthly payment) on March 19 (when it was 5.5 months late).  I was overjoyed!  Obviously, that payment did not "cure" the loan (in fact, it didn't even move it out of the Late-4 category), but it represented $911 in lenders' pockets (after subtracting NSF and Collection Fees).  But that "success" is not reflected in your analysis.  Similarly, one of my two Late-3's made its last on-time payment on September 11, but has since made 4 payments in collections (the first was a double payment, and each of the next 3 were half-payments), totalling 3.5 months of payments.  That is $1,785 in the pockets of lenders, also not reflected in your analysis. 

Thinking back, I believe that I may have never had a loan in collections cure; yet, as noted above, I have received non-trivial collections payments.  Is it possible for you to track the percentage of dollars collected by the CA's versus dollars of loans placed with the CA's?  Wouldn't that be a more useful number?

[Fred93]

Sorry if you've already answered this question, but are you tracking number of loans or number of loan-occurrences in collections?  In other words, if a loan goes to collections, cures, and then goes back to collections, is that 0/1 or 1/2?

Good question.  I'm tracking the numbers provided by Prosper on their collection agency page.  I have no information on the details of how these numbers are computed. 

One could compute a similar sort of number from the export database.  It contains a table with a row for each time a loan's status changes.  In other words there is an entry when a loan goes "1 month late" and an entry when a loan becomes "current".  These are approximately, but not exactly the same as a loan having gone to collections and brought current.  One could process the export database by two algorithms and see which one matched the collections page more closely.

[iLIE]

Can you get the 12 loans that actually made payments to see if they paid up to current?

[Fred93]

One thought that occurred to me, however, is that while "cures" are certainly the gold standard for collections, they are not the only thing (or even the most important thing) that matters.  Ultimately, what lenders really care most about is how much money is being collected from late borrowers.  ...  Is it possible for you to track the percentage of dollars collected by the CA's versus dollars of loans placed with the CA's? 

I agree.  I used to track a percentage provided by prosper called "net collected".  I think this is exactly what you are thinking about.  Unfortunately, on 10/30/07, Prosper suddenly stopped providing this number.  The row is still present in their web page display, but it is filled with dashes instead of numbers.

They do display a number called "Gross collected".  One could track that number vs time, but unfortunately that isn't one of the numbers I've been collecting vs time.  I collect these numbers daily by manually reading the collections page and typing a few numbers into a spreadsheet.  In retrospect it might have been better if I had captured everything by saving an image or something, but when I started I didn't imagine problems such as Prosper dropping the statistic I was tracking.  I simply thought net was better than gross for my purposes, so that's what I tracked.

So, yes I could start tracking the gross collected number, but I have no way to get the historical data.  Gross collected an "brought current" numbers are similar in magnitude, so I don't see any evidence that one would draw a different conclusion.  In my posts on collections back when I was tracking net collected I used to plot both "brought current" (which I think was called "cured" in those days) and "net collected" on the same chart.  If you go back and look at some of those charts you will see that the two curves wiggled at the same time.

So yes in principle, but I think in the end these approaches tell you about the same thing.

[Xenon481]

Running with Ira's thoughts, is it possible that Amsher is actually doing pretty decently at $'s collected / $'s delinquent, but they are purposely not bringing the loans current so that they continue to collect their fees if the borrower makes another payment?

With no reduction in fees vs. time, Amsher has no incentive to actually bring the loans current, just to collect payments.

[Fred93]

Running with Ira's thoughts, is it possible that Amsher is actually doing pretty decently at $'s collected / $'s delinquent, but they are purposely not bringing the loans current so that they continue to collect their fees if the borrower makes another payment?

No.  If such a thing were true, the "gross collected" percentage (of $) would soar above the "loans brought current" percentage, but if you look at the collections agency page, you will see these numbers are similar.

Of course this assumes that the numbers given to us on the collection agency page fairly represent what is happening inside the agency.  I have no visibility into what is happening inside the agency. 

[xraider]

Thanks, Fred.  I sob every time you update your blogs.