Friday, November 2, 2007

Update On Prosper's Collection Effectiveness

As I mentioned in a previous blog post, one way to gauge the effectiveness of collections is to look at how delinquent age. If a loan is one month late, in a month it will either be brought current or it will be two months late. From this, we can get near real-time updates on whether Prosper's attempts to improve collections are working.

And on that note, we now have Nov 1 statistics to add to the table.

Month Initially Late
Initially Late
(1 month late)
1 Month Later
(2 months late)
2 Months Later
(3 months late)
Repair Rate
January 07
95
83
82
13.68%
February 07
101
90
90
10.89%
March 07
108
91
89
17.59%
April 07
130
119
119
8.46%
May 07
193
178
171
11.40%
June 07
203
195
186
8.37%
July 07
224
194
187
16.52%
August 07
250
232
227
9.20%
Sept 07
259
229
234
9.65%
October 07
313
264
?
?
November 07
324
?
?
?
The uptick in defaults in the Sept 07 family (1 month late on Sept 1, 2 months late on Nov 1, 3 months late on Nov 1) is peculiar and may be because October has 31 days (a good question for Prosper staff). The Oct 07 family, however, is noteworthy because the repair rate is already at 15.7%. While this is not outside the statistical spread (March and July were also pretty good), it's better than average (11.4%). A decrease would've been a contrary indicator, so this is a reasonable sign. December's numbers should be enlightening.

I'm transitioning my Collections-O-Meter (TM) from "skeptical" to "cautiously hopeful".

2 comments:

Lend2 said...

The uptick in Sept 07 could also be the result of a loan payment that moved a borrower from the 4 Month bucket to the 3 Month bucket. Sadly, it doesn't happen often. But I've seen it happen.

Mike said...

Lend2 -

Very true. I'm putting together a larger list of possible reasons for the deviation.