April Prosper.com Collections Update

Yes, I'm alive. I took a week off for my brother's graduation. Unfortunately, no internet where we were staying. D'oh. I'm digging back in. On to the substance:

This is part of my ongoing series monitoring collections efficiency (March's report for comparison). For a reminder on the methodology, I took a snapshot of all of Prosper's loans on April 1 and compared their current status against those same loans on May 2 (operator error, once again). Presumably, loans that don't get further behind have some kind of money extracted in the collections process (not so valid on this run - see below). The statistics are below.

	Total	Got Better	Stayed The Same	Got Worse
Current	14456	226 1.5%	13895 96.1%	335 2.3%
Late	223	56 25.1%	31 13.9%	136 60.9%
1 month late	292	32 10.9%	26 8.9%	234 80.1%
2 months late	235	18 7.6%	11 4.6%	206 87.6%
3 months late	260	7 2.6%	7 2.6%	246 94.6%
4+ months late	946	6 0.6%	930 98.3%	10 1.0%

The Signs Of Collections (SOC) statistics are below. Signs of collections percentage is calculated as the percentage of the loans that either improved or stayed the same.

Month	1 Month	2 Months	3 Months	Total
April '08	19.8%	14.2%	5.2%	39.2%
March '08	24.4%	12.4%	11.3%	48.1%
February '08	13.7%	9.3%	4.7%	27.7%
January '08	20.6%	12.8%	8.0%	41.4%
December '08	23.0%	2.8%	3.8%	29.6%

Since I screwed up and didn't get the data until May 2, this report will be slightly negative. This is because Prosper calculates loan lateness based on the date in a month (and not 30 / 60 / 90 days) and loans that came due between the May 1 and May 2 will not be properly counted. Overall, there has been a mild trend toward improving odds of squeezing money out of a loan that has gone late.

The "4+ months late" category is very hard to quantify percentage wise. The default sales make month over month comparison very difficult, so I'm not going to try.

March Prosper.com Collections Update

This is part of my ongoing series monitoring collections efficiency (February's report for comparison). For a reminder on the methodology, I took a snapshot of all of Prosper's loans on March 3 (yes, I'm a doofus and missed March 1) and compared their current status against those same loans on April 1. Presumably, loans that don't get further behind have some kind of money extracted in the collections process (not so valid on this run - see below). The statistics are below.

	Total	Got Better	Stayed The Same	Got Worse
Payoff in progress	48	48 100.0%	0 0.0%	0 0.0%
Current	13772	214 1.5%	13243 96.1%	315 2.2%
Late	203	54 26.6%	42 20.6%	107 52.7%
1 month late	281	34 12.0%	35 12.4%	212 75.4%
2 months late	273	11 4.0%	23 8.4%	239 87.5%
3 months late	254	4 1.5%	25 9.8%	225 88.5%
4+ months late	740	8 1.0%	723 97.7%	9 1.2%
Defaulted (Delinquency)	1421	0 0.0%	1355 95.3%	66 4.6%

The Signs Of Collections (SOC) statistics are below. Signs of collections percentage is calculated as the percentage of the loans that either improved or stayed the same.

Month	1 Month	2 Months	3 Months
March '08	24.4%	12.4%	11.3%
February '08	13.7%	9.3%	4.7%
January '08	20.6%	12.8%	8.0%
December '08	23.0%	2.8%	3.8%

Since I screwed up and didn't get the data until March 3rd, this report may look overly favorable on collections success. This is because Prosper calculates loan lateness based on the date in a month (and not 30 / 60 / 90 days) and loans that came due between the 1st of April and 3rd of April will not be properly counted. This inflates the numbers for loans that stayed the same, though statistically, this should only shave about 3-4% of the loans in the stayed the same category. Removing that 3-4% would place March on similar footing to January.

The "4+ months late" category is very hard to quantify percentage wise. The default sales make month over month comparison very difficult, so I'm not going to try. There were 62 loans that left the "Default (Delinquency)" which are presumably part of Prosper's collections-by-lawsuit effort. This has created a statistical tracking problem because they are now marked as Repurchased. Repurchased has typically been for loans that were reacquired by Prosper due to fraud. Most statistics reporting sites like Lending Stats do not count repurchased loans for ROI calculation purposes, even though it isn't appropriate in this case.

February Prosper Collections Report

This is part of my ongoing series monitoring collections efficiency (January's report for comparison). For a reminder on the methodology, I took a snapshot of all of Prosper's loans on February 1 and compared their current status against those same loans on March 3 (yes, I'm a doofus and missed March 1). Presumably, loans that don't get further behind have some kind of money extracted in the collections process. The statistics are below.

	Total	Got Better	Stayed The Same	Got Worse
Payoff in progress	28	28 100.0%	0 0.0%	0 0.0%
Current	13271	241 1.8%	12719 95.8%	311 2.3%
Late	228	58 25.4%	36 15.7%	134 58.7%
1 month late	297	22 7.4%	19 6.3%	256 86.1%
2 months late	265	11 4.1%	14 5.2%	240 90.5%
3 months late	229	5 2.1%	6 2.6%	218 95.1%
4+ months late	585	7 1.1%	511 87.3%	67 11.4%

This does not look good. The Signs Of Collections (SOC) statistics are below.

Months Late	February SOC	January SOC	December SOC
1 Month	13.7%	20.6%	23.0%
2 Month	9.3%	12.8%	2.8%
3 Months	4.7%	8.0%	3.8%

This is, by all accounts, a rather large statistical kick in the shins for lenders. There are two things that may distort February's collections numbers, however. First, Prosper transitioned their primary collections company from Penncro to AmSher. I find it unreasonable to expect a perfectly clean transition with no drop off when making a change of this magnitude. Second, I screwed up and didn't get the data until March 3rd. This may overstate the unsuccessful collections efforts.

The "4+ months late" category is very hard to quantify percentage wise. The default sales make month over month comparison very difficult. There were 62 loans that transitioned to "Default (Delinquency)" which are presumably part of Prosper's collections-by-lawsuit effort. We'll see them rattle around further in next month's dataset.

January Prosper Collections Report

This is part of my ongoing series monitoring collections efficiency (December's report for comparison). For a reminder on the methodology, I took a snapshot of all of Prosper's loans on January 1 and compared their current status against those same loans on February 1. Presumably, loans that don't get further behind have some kind of collections activity happen succeed. The statistics are below.

	Total	Got Better	Stayed The Same	Got Worse
Payoff in progress	30	30 100.0%	0 0.0%	0 0.0%
Current	12620	182 1.4%	12122 96.0%	316 2.5%
Late	265	82 30.9%	33 12.4%	150 56.6%
1 month late	304	51 16.7%	12 3.9%	241 79.2%
2 months late	249	16 6.4%	16 6.4%	217 87.1%
3 months late	271	12 4.4%	10 3.6%	249 91.8%
4+ months late	356	11 3.0%	339 95.2%	6 1.6%

This is a rather notable improvement over December, with good signs of collections on "2 months late" and "3 months late" loans.

Months Late	December Signs Of Collection	January Signs Of Collection
1 Month	23.0%	20.6%
2 Month	2.8%	12.8%
3 Months	3.8%	8.0%

If these numbers are taken at face value, this translates into a 63.5% chance of a "1 month late" loan going to "4+ months late" without any collections yielding cash, improving from December's 72% chance. Not good enough, but better.

The "4+ months late" category is very hard to quantify percentage wise. The default sale in December thinned out the herd, improving the percentages, but making month over month comparison very difficult. It is noteworthy that 11 "4+ months late" loans showed signs of improvement, which appears to be a record.

Also of note, 1 4+ month late loan was repurchased in January.

Update: Chris' pickiness is technically correct (see comments) (grumble grumble), so I fixed it. Strikethroughs for deletions, italics for additions.

December Collections Report

If a 150" flat panel isn't enough for you, how about some collections info...

This is part of my ongoing series monitoring collections efficiency (November's report for comparison). For a reminder on the methodology, I took a snapshot of the loans on Dec 1 and compared this to the same loans on Jan 1. The statistics are below.

Dec 1 Status	Total	Got Better	Stayed The Same	Got Worse
Paid	1808		1808 100.0%	0 0.0%
Payoff in progress	24	23 95.8%	0 0.0%	1 4.1%
Current	12172	150 1.2%	11647 95.6%	375 3.0%
Late	215	63 29.3%	33 15.3%	119 55.3%
1 month late	316	39 12.3%	34 10.7%	243 76.8%
2 months late	272	1 0.3%	7 2.5%	264 97.0%
3 months late	252	3 1.1%	7 2.7%	242 96.0%
4+ months late	822	2 0.2%	142 17.2%	678 82.4%

As previously noted, loans that improved their state or stayed the same require some kind of payment activity to maintain their position. The table below shows the percentage of loans that show signs of payment versus loan latenesses compared monthly.

Period	1 Month Late	2 Months	3 Months	4+ Months
Dec 2007	23.2%	3.0%	4.0%	0.2%
Nov 2007	20.3%	8.4%	6.1%	0.9%

Overall, the December numbers come out flat. The increase in collections for 1 Month Late loans were balanced by decreases for all other loan categories.

New And Shiny Collections Statistics

I had proposed a new way to look at collections progress back in early December. This method took to snapshots of Prosper's nightly data export, one month apart, and looked at how the lone status changed over time. I took my first snapshot on December 1, but by some fluke of luck, it turns out that I had the Nov 1 download laying around a second computer, so now we get an early look. The underlying assumption is that if a loan's state does not get worse (as in, going from 1 month late to two months late), some payment must've occurred. So, into the data.

	Total	Got Better	The Same	Got Worse
Payoff in progress	28	26 92.8%	0 0.0%	2 7.1%
Current	11697	154 1.3%	11238 96.0%	305 2.6%
Late	252	66 26.1%	40 15.8%	146 57.9%
1 month late	326	42 12.8%	24 7.3%	260 79.7%
2 months late	263	13 4.9%	9 3.4%	241 91.6%
3 months late	231	5 2.1%	9 3.8%	217 93.9%
4+ months late	618	6 0.9%	607 98.2%	5 0.8%

The interesting figures are that between Nov 1 and Dec 1, it looks like 20.3% of "1 Month Late" loans, 8.4% of "2 Month Late" loans, and 6.1% of "3 Month Late" loans had some kind of payment activity that kept things from falling further behind. The blood from a turnip category ("4+ Months Late") had 1% improve, though this is hardly definitive because payments that do not bring the borrower to 3 months late or better are not visible. I am hesitant to infer things from the "Late" category because it is not on a clean month to month boundary.

These statistics imply better collection performance than the official Penncro numbers that Prosper publishes, but a one month snapshot is insufficient to rule out a statistical anomaly.

Some other notes from the statistics include the 5 "4+ Late Loans" got worse by folks entering bankruptcy and 1 "1 Month Late" loan that was repurchased.

New Attempt At Observing Collections Progress

My previous attempt at getting a handle on how collections are proceeding on Prosper following the arrival of the new VP of operations didn't go so well. I did not have a grasp on how successful collections impact the reported loan status. Time to try again.

At the beginning of each month, I'll download the raw Prosper loan data. It'll be processed to determine loans that increased in their lateness (going from 1 month late to 2 months late, for example), and those that showed signs of some monies being collected (held or improved position). The first dataset was picked up on Dec 1, with this being the baseline for all future statistics tracking.

I'm going through the trouble to fill in the gaps with the collection agency statistics. The statistics report how well the collection agencies are performing, but do not discuss borrowers paying directly through Prosper (perhaps because they got a letter) to get the account current. I also find that it has less ambiguity than the statistics that Prosper reports.