About 75 loans are past 1 month late, and therefore should have gone to the collection agencies. Almost all of these have gone to Penncro, simply because Penncro has the lowest fees, and almost all lenders have apparently chosen an agency based on fees. A few loans have gone to Alliance One. None have gone to NCI.

We now have statistics telling us how the collection agencies are doing, and the results are scandalous. They're collecting almost nothing.

(IMG:http://img.villagephotos.com/p/2006-6/1187065/prospercollections11.JPG)

I believe that Penncro has 54 loans. Thats the number that would make that "1.85% cured" be exactly 1 loan. Penncro has had these loans for an average of 35 days.

54 loans for 35 days, and all they have to show for it is 1 loan cured?

It seems to me that there ought to be many more loans that go "1+ month late" than loans which default. To make that true, loans that go late need to be "cured" most of the time. Well darn it, "most of the time" and "1.85%" just don't match up! Net recovery of "3.88%" doesn't impress me much either.

Something is wrong. There are several possibilities:

1. The Prosper statistics are wrong. Loans are actually being cured, but they don't show up correctly in the statistics. If so, Prosper needs to get off their buns and fix this pronto.

2. The statistics are right. The collection agencies are incompetent. If so, Prosper needs to get off their buns and find some competent collection agencies.

3. The statistics are right. The collection agencies are competent. We're screwed. Loans that go late are all gonna default. If this is true, then the loans are gonna do much worse than the Experian statistics, and we're charging interest rates that are way too low.

I don't know which it is.

There are also intermediate possibilities, such as:

4. The collection agencies are marginally competent, but need serious supervision which they aren't getting. If so, Prosper needs to get off their buns and supervise.

I urge Prosper to talk with us more about what is actually going on with the collection agencies. Getting information has been like pulling teeth. It took a lot of complaints from lenders before they turned on the statistics at all. They tired us out. But the job isn't done! Getting a few numbers was just one step in the battle. I'm not going away because they replaced "0" with "1.85%". We have to get to the root of the problem. Something is very wrong, and prosper must talk with us about it.

I urge lenders to switch their collection agency preference to NCI. Unless a significant number of lenders switch, no loans will ever go to NCI, and we will never learn how well they do. Even if a bunch of you switched today, it would take a couple of months for the first loans to go to NCI, and another month to find out how they perform. That's three months minimum before we know anything. This is like a huge ship that takes months to turn a bit. I have talked about this before, and I've been astonished at the responses I got. Several lenders told me that they wouldn't switch because NCI was too expensive, even though what matters is how much of your money you get back net of collection agency fees, and right now you're getting essentially nothing. Without some loans going to NCI we will be ignorant forever.

I urge lenders to make your feelings known to prosper. To make lenders a success, and to make prosper a success, collections has to work.

pninen,

I wrote a data scraper to dig through the performance data. It's quasi up and running (I run it on my own platform at home and all of the variables are hard coded, no pretty UI yet). The biggest problem I have so far is how to anaylize the data and present it in a meaningful way.

One thing I can do I guess is track all of the loans made in say March, and watch their payment progress (1mo, 2mo, 3mo) on a monthly basis. This would give some sort of an idea on whether or not the payment histories on a given set of loans are getting better or worse.

When I get home, I'll try to whip some stuff up and have it posted in a few hours.

I already use NCI

Prosper sure does leave a lot hanging out there for speculation..

I'm going with the you're screwed on the HR & NC loans..


Prosper is too busy making silly things like community payment & not addressing real concerns.

Such as

Age Verification

Actual listing info verifications..You could have 1 person reading listings that will get funded & comparing what they know against what is posted..I'm not as comfortable as some lenders are with accepting lies from borrowers!!

Interesting points, Pninen and Bamalucky. And I'll look forward to seeing lenderguy's analysis.

One thing we need to know is what the normal collection success rate is for an unsecured loans to mostly high-risk borrowers. I bet it isn't great.

Bamalucky, I agree prosper should delete listings (with no chance of re-listing) for those who obviously lie in the listing. But the people who will uncover the lies are us, the lenders ... not prosper staff. It would take Prosper weeks to verify every sob story here...but lenders themselves have caught some borrowers readily lying.

As I pointed out in your Petri Dish thread on this issue, Prosper policy prohibits using age as bidding criteria.

I switched to NCI, after being having the two others as my default for 2.5 months each.

Collection agencies collect between three to ten (the usual high end) percent of the bad paper they handle. That's why junk debt buyers can pay only a few pennies on the dollar for bad paper and still not make a killing. This is not incompetence, by the way. This is the status quo.

Late addendum: You can watch collectors commiserate about this and other topics on this discussion board: http://www.collectionindustry.com/forum/index.cfm

Except here the CAs are (or should be) playing a different role here from what they usually do.

CAs normally get accounts after charge-off, when they are 90 or 120 days past due. And JDBs, I would think, get them anywhere from 180 days to 20+ years delinquent.

Prosper is not (or should not be) simply referring 30-day-late accounts to collection agencies as bad paper, but is (or should be) outsourcing the normal in-house collections function (bringing the account current and restoring the relationship between lender and borrower). Because the account goes to the CAs at the 30-day mark, our statistics should be dramatically different from normal CA results. The correct statistics for comparison would be creditors' internal collection departments, not CA/JDB collection rates.

My only quibble with pninen's option number 3: It is possible that most of the lates occurring on first and second payments were simply fraudulent (or deeply flawed) from the get-go, and that lates on subsequent payments will show a dramatically higher cure rate.

However, it seem sto me that I have heard enough anecdotal reports of loans that have been brought current after going 30+-days late to suggest that at least part of the problem lies in the statistics. Can ericscc.com or wiseclerk.com come up with a list of loans which were 30+ days late at some point but are now current?

Geez, there is a forum for everything.

Age should be verified, not to grant a loan, but to make sure borrower is telling the truth. One borrower got funded saying she was a grandmother. Yet her credit was only 6 months old, no reference to a job she claimed to have had for over 15. Her answers to questions sounded like she was a valley girl. In a few months when the loans really start defaulting, oh boy.

Lots of people are saying they have good results so far in terms of repayment. I would like to know who has the worst performing portfolio in terms of defaults.

Was that one of the very early days loans?

No that is a very recent loan. I pointed it out to a few of the high hitter lenders. They reported the loan after it funded, to make sure Prosper to extra due diligence before funding.

This is why I can never walk away from the computer for more than 5 mins, I miss all the good stuff. Very interesting.

No, I think she's talking about "BestGrandma." If we're talking about the same person, I respectfully disagree w/ MsAva here-- to me, the borrower didn't sound like a Valley Girl, she just sounded (as we say around here) "real country." Not surprising if she's lived her whole life in Beaumont. :)

I didn't want to say which loan since it funded. But that is one loan I want to follow very closely. Other than the Prosper info, the only thing we have to go on is our instinct. Hard when you are not looking someone in the face.

I think I do remember a bit of chatter on that one. Was that the one where the daughter became a lender to bid on the loan?

So many loans, so little memory left.

This question about "BestGrandma" is something I'm a bit surprised about. I bid on her loan and am waiting on it to fund. I don't understand why someone would be surprised that a 52-year-old woman wouldn't have a long credit history. She said her husband died a few years ago and had always encouraged paying cash for stuff. I have a grandma who's 90 who probably doesn't have any credit history at all because she owns her house free-and-clear and has never owned a credit card. She lives in a small country town in the middle of the US where this kind-of thing is common.

Also, there are some people, especially women, who are married to men who do all the financial transactions and the women don't get involved. If the man dies, they have no credit because it was all in his name. In this day-and-age, it is not a good situation to be in, but it does happen.

Just my thoughts...

I am surprised, no work history at all in the credit report after working at the largest company of that type in the country. The job should have shown up. The loan is funded. No need to rehash it.

Being a mortgage broker, there are just some flags we look for. A face-to-face interview would have eliminated those fears.

Yep, that's the one.

I looked back, unless the daughter has another name for lending, I didn't see her bid on the loan.

My job (or past jobs) didn't show up on my credit report until I applied for a refi of my mortgage, even though my credit history goes back 10 years.

I was pissed at the mortgage place for reporting it. My job has no business being on my report.

It takes some time for funds to clear in the Prosper system. I believe her daughter signed up to be a lender after I did and I'm still waiting for my funds to clear. I'm guessing it funded before her daughter got a chance.

Well I have read hundreds of credit reports. The section for job history is very important. All three credit agency have it.

It's not always accurate. When I last pulled my reports, I realized my job history was really, really out of date. The last listing was for a job I held between 1997-2000 showing as current. I sent the companies my updated history (including my foray into the dot com bust) and everything was copasetic within a month or so.

As I understand it, employment history on CRs is generally based on what creditors report, which in turn is based on what the borrower puts on a credit application. Since she only had, I think, two trade lines other than the medical collection, it's not totally surprising that neither of them bothered to report the employment data. (IOW..... I think the issue here is not the length of employment, or the size of the company, but the number of creditors who might have reported that data.)

Still, I would have preferred to see vetted pay stubs.

The daughter did not bid; I don't know why. (As a lender on the daughter's loan, I had recommended that she not bid, out of concern that the finances of new businesses are almost always tighter than expected. I don't know if my recommendation was a factor or not.)

The daughter did make an early payment on her loan; that payment has successfully cleared.

I agree, it's the creditors that report. Companies themselves don't report.

In my case, it was the mortgage I refinanced with.
None of my 12 credit cards, original mortgage, or school loans ever reported my job history. So it's very easy to see how someone with little to no credit history would never have their job reported.

I can only find a few possible examples, but I've only been tracking historical loan statuses for just over a month now (since 7/12).

The examples:

Listing #5978
Was 1 month late on 7/12
Was 2 months late on 7/21
Was current on 8/1

Listing #858
Was 2 months late on 7/12
Was current on 7/17

Listing #3188
Was Late on 7/12
Was 1 month late on 7/28
Was current on 7/31

My understanding is that when a loan makes it to collections, you can kiss it goodbye. Seriously, people who won't pay simply won't pay. PsychDoc is correct, this is why people buy bad loans for a few pennies on the dollar, its nearly impossible to collect. This is why I feel for debt collectors, its a thankless and cruel job.

Caladia,

I love Beaumont! They love my NY accent there, they say I talk funny. Super-fun downtown too.

Jeff "Yee-ha, cuz" D.

Jeff! Your goal in life is to nudge me off topic, isn't it?

... must... not... take... bait... :P

I don't know enough about the Collection Agency issue. I never did understand how collection agencies made loans come current again, except maybe in

( A ) Tracking people down when the original lender couldn't -- maybe the Collection Agencies had better resources for that? (I have no idea.)

( B ) Letting people know they owed the money-- somtimes things really do fall through the cracks and people never got the bill, or moved and forgot about that last $11 electric bill in all the chaos of moving, or whatever...

( C ) Annoyed borrowers so much with repeated phone calls, that paying the bill was preferable to the irritation. It seems to me that this would only work with relatively small amounts?

For larger amounts, the idea I had in my head was that the only real recourse would be for the original lender to sue the borrower, which is not something the Collection Agency would handle, right?

(A) Yep. And they do have better resources, databases etc. But I would venture that while these loans may represent a significant number of loans, they represent a fractional share of the total money.

(B) I don't think most CA's will take such small loans. I think these just get written off. The administrative hassle is just too much for de minimus amounts. Depends on the lender, I'm sure. I don't think GMAC is going after many 11 dollar bills, but SBC might.

? Yeah, but this won't work for 90% +. Either they intend to pay or not.

Jeff "hey, did you see the Astro's game?" D.

Hm, why did my "( A ) ( B ) ( C )" thing work in my original post but not when it was quoted? (Note to self: don't use parentheses in combination with letters, even with spaces between.)

Anyway.

Since I don't know enough about the collection agency aspect, I've been tending to assume that once a loan goes to collections, the money's gone. If it gets brought current, that'll be like "found money," 'cause I'd already given up on it. <_<

Ms Ava, I worked for the same employer for nigh on 23 years. It was a fortune 10 company. There is no indication (present or past) about my employment. I suggest that you are somewhat paranoid about that aspect of the loan.

But, in the spirit of Epictetus, I will wager you on that.

I will paypal you $50 if that loan defaults. Your choice on how much you will Paypal me if the loan does not default.

How's that?

Aw, now, y'all, don't be jinxing the loans with bets on 'em! Bad mojo! :rolleyes:

Look at the bottom of the screen where you compose a post. There is a checkbox for "Use emoticons". Check it as appropriate to decide whether you want characters or smilies to appear.

No no, I mean, I didn't WANT those letters to appear as symbols or emoticons. I wanted them to be letters bracketed by parentheses.

EDIT: Oh, I see what you're saying. I can tell it whether it should read my letters/brackets as emoticons or not. Thanks Hollow! :)

Hollow,
I just voiced my thoughts. I didn't bid so my punishment will be lost of interest. No bet. This is not a, "I know more than you contest". I'm just watching for future portfolio picks.

Follow up:

I crunched some numbers this morning. I haven't gotten around to getting webspace yet (or even gotten around to get an imaging host) cause I'm tired and want to go to bed as soon as I finish this. If anybody wants the actual file with my data and graph, you have to PM me your email address. I'll have them out sometime tonight (late tonight for you east coasters).

Disclaimer first: I believe Prosper Andrew is correct when he says that we don't have sufficient data on which to guage late payments and cures. We have data, don't get me wrong, but we don't have enough to draw any valid statistical conclusions. I'm not a stats expert, so I really wouldn't know how much me would need, but what we have is not enough.

Methodology: I looked at loans originating March 1 - March 31. Beginning on June 1, I took payment snaphots on June 1, July 1, and August 1. I didn't begin in May because there was very little data.

I group the results in order of progression -- like one would like to see loans that were one month late in June, two months in July, and three months in August. Then, one might want to see loans that were 1 month late in July and two months late in August, and finally, loans that because one month late in August. They are grouped against % current for the month.

What I can say is that NC's are performing worse than HR. It's also not apparent that the collection agencies are functioning at all -- the % current either stays the same or gets worse as time goes on.

For a discussion topic, here are the % currents for August, as reported August 1st, for loans originating 3/1-3/31.

ALL AA A B C D E HR NC
AugCur 87.04% 96.55% 100.00% 100.00% 94.29% 90.48% 90.00% 65.31% 57.14%

You forgot #5.


pninen is panicking over a statistically insignificant sample of unrepresentative data. Rather than allow a statistically significant sample of data to be collected, he wants to ruin the sample by introducing another variable.

54 loans for 35 days looks like a significant sample to me.

Do you have a methodology on which you base your claim of insignificance? I kinda doubt that you do.

I do NOT claim that one can be precise at this point. I've never said anything like that. However, the numbers are coming back WAY OFF from any reasonable values. You're asleep at the switch and headed for oblivion.

Frankly, I believe that the statistics they're giving us are still faulty, and that we must prosper to dig in and get the real story. Anything else is gross mismanagement.

However, if you wish to keep sleeping, pardon me for making noise.

Be careful. Are you sure that data is commesurate with our situation? You refer to "bad paper", but that usually means loans that have not been paid for some significant time. Prosper uses the collection agencies to work on loans that are ONLY ONE MONTH late. Heck, you and I have both been 1 month late on things in the past. Normally when you're one month late you just get another bill in the mail that's a different color that says hey we didn't get your payment.

Now of course prosper is different for many reasons. One reason it is different is that payments are automated, so we don't have the "forgot to write a check" issue or "decided not to write a check" issue, which are the cause of most 1 month late situatons. In our case, the bank account was empty, either because the individual was low on money, or because it was a "fake" bank account (meaning it was not the individual's main bank account, but was something they set up just for prosper, to deautomate payments), or the account has been closed, etc.

I do believe that loans that go unpaid for several months are trash, or "junk". We agree on that. I am not yet ready to believe that loans that go unpaid for 1 month are "junk". That just seems nuts to me.

In fact, epetroel posted earlier in this thread THREE examples of loans that had gone >1 month late and then recovered. The collection agency statistics are inconsistent with three recovered accounts. They only show 1. So what happened to the other 2? My guess is that epetroel's methodology is crude (he doesn't sample very often) so there are many more recoveries that aren't in his data. Why don't they show up in the statistics?

There are many possible reasons.

1. The statistics are just trash.
2. The statistics are not updated frequently.
3. Many of the loans didn't actually go to the CAs at 1 month as Prosper has said they would.
4. {self-censored for civility}

So there are multiple pieces of evidence to support the scandal.

It would be great to have some data from some other lending situation on how loans transition from being 1 month late to either payment or default. However, I am skeptical that your collection agency experience is the right data.

Epetroel's data is missing most non-group lates, so, all else being equal, the number of lates brought current is probably about twice what his records show, even in the absence of sampling issues.

The data is sampled (for the most part) twice a day, so I don't think anything would have slipped through unnoticed.

However, it is only for one month (from 7/12 to today) as I was not tracking historical loan statuses before that point. Also, as traveler505 mentioned, it does only include group lates.

In this case I think prosper has NO resources for this function, and the entire collections department function is outsourced to these folks called collection agencies.

Judging only by what prosper has told us (and there's a whole page in the help section telling us what they do when a loan is late), it would appear that the collection agency makes the FIRST phone call.

If indeed prosper has farmed out the entire collections function, then I would think that the CA is going to have to be the one to handle suits. Dunno what you mean by "larger amounts". Prosper loans go up to $25,000 which seems like enough to be worthy of a suit to me. In fact I believe a case can be made for suits on much smaller loans, even if the cost of the suit exceeds the amount of the loan. Gotta establish boundaries.

At the socal dinner, Chris Larsen said there would be suits. He didn't say how this function would be managed. There simply have to be suits. We can't allow prosper to become known as free money.

Pardon me for doubting you. Wow! That is excellent.

This info will be very helpful as you accumulate more data.

What?

There cannot be suits filed by Prosper or its CAs, except perhaps for losses resulting from fraud.

Unless something has changed, the loans are sold at Day 120 to JDBs. That doesn't allow enough time for litigation before sale, and, after sale, Prosper loses its rights to sue, as it no longer owns the note. The JDBs might sue, but that won't get the lenders squat.

Perhaps this is going to change, and Prosper is going to keep the loans and pursue more aggressive collection methods. (Personally, I think this would be in the financial interest of the lenders, though Prosper staff would have to endure our "why haven't you collected yet" complaints for a longer period of time.) And that could help explain why one newspaper recently reported that lenders would only get paid for defaulted loans if the JDB collects. But that's not what it says on the website.

Agreed and noted.

If I were Prosper, I would be concerned about some lenders taking collections into their own hands. I know that it technically violates the lending agreement with Prosper, but some people might think that the risk is worth the reward.

If that happens, Prosper has the potential to implode. And no one (except perhpas bamalucky) wants that to happen.

It would be a good investment on Prosper's part to:

1) step up collections efforts to the full extent

2) COMMUNICATE that they recognize the issue and are taking the steps they can to make things right.

FB

I do not have complete data either. In my data I find 16 loans that flipped from late (or several month late) to current

listing / date flipped to current
858 07-18
1405 08-11
3188 08-01
5978 08-02
6716 08-22
10972 08-01
11035 08-04
11776 08-09
12444 08-08
14342 07-18
14915 07-26
15935 08-05
17270 07-29
20218 08-12
20795 08-12
20864 08-19

One of the next functions I plan to add at wiseclerk.com is a payment history for loans

Claus

I don't really think you can gauge collection success with the average time in collections being only 35 days. Remember, a collection agency must follow applicable laws/regulations. When they first obtain an account they send notice to the debtor at which time they must give the debtor 30 days to dispute the validity or challenge the debt. During this 30 day period, no collection activity actually takes place. If the debtor actually does request validation or challenges the debt within the 30 day period, the collection agency must refrain from further collection attempts until the original creditor (Prosper) supplies the information/validation of the debt. Some debtors will use the 30 day period as, perhaps, a stall tactic while hopefully putting the money together to bring the account current. In some cases, the initial collection letter or phone call might just be enough to encourage reality and the individual just might pay up. In other cases, obviously not. I just think that statistics based on 35 days in collection can't be viewed as the big picture in terms of success.