Print

[Fred93]

Fred (yawn) wrote in the old blog tonite.  
http://fred93blog.blogspot.com/2010/08/prospercom-072010-late-loan-stats.html

The meat...


Nominate for lobby.

[Urbi_et_Orbi]

45.3%?! Good grief!

Nominate for lobby.

[mothandrust]

Look at the the loans originated November 2009 and compare them November 2006-2008's.

Eyeballing the graph I'd say that 6.5%-7% of the November 2009 loans are late, compared with (9 months later) 12% on the 11/2008 loans, and 14% on the 11/2007 loans.  11/2006 was in the "wild west days" (pre-extended credit) and just under 16% were late after 9 months.

This supports the claim that the 3.0 (unsecured) loans are going bad less fast than the 1.0 (secured) loans.

[bluestar148]

An object in motion............2nd for Lobby.

[xraider]

I'm not surprised that with higher credit scores, more careful lenders, generally lower loan amounts and all that the new loans are doing better than the older loans.  Sad to see that there are problems at all with the newer loans, and will be interesting to see how those loans play out over time.

Thanks, Fred!

[ladeeda]

Glad to see you working with this data again, Fred... I just want to confirm, though, that these charts are on a unit basis, correct?

Also, how do you deal with charge-offs and pay-offs for the denominator?

[Fred93]

I just want to confirm, though, that these charts are on a unit basis, correct?

Numerator = # of loans originated in the month which at the observation date are >30 days late or charged off
Denominator = # of loans originated in the month

Also, how do you deal with charge-offs and pay-offs for the denominator?

No special attention is given to early payoffs.  They are by definition not late, so they do not enter the numerator.  They are kept in the denominator.  This produces a ratio where you can see what fraction of ALL LOANS have gone bad.

Charged off loans are counted in the numerator of course, and are kept in the denominator.   I keep all loans that were originated in the denominator, producing a ratio where you can see what fraction of ALL LOANS have gone bad.

[kenL]

Interesting, the third chart on your blog seems to show a trend that the later 3.0 loans are performing worse than the earlier 3.0 loans. For example the August 2009 loans have a low slope which means they are performing great and the Febuary 2010 loans have the highest slope of the 3.0 loans indicating they are performing the worst. Feb 2010 already has a higher % of 1+months lates than Aug 09!!  Does this mean Prosper is getting lazy with collections or with borrower screening again? Or is this poor performance caused by the Prosper update fiasco?

[onthefence]

Does this mean Prosper is getting lazy with collections or with borrower screening again?

Hard to say.  It could be that Prosper is getting desperate to get loans through the door so they are letting stuff slide that they shouldn't.

Or is this poor performance caused by the Prosper update fiasco?

If I were to make a guess, I would say that when prosper first opened up for business again, there were a lot of lenders who had finally had a chance to see their portfolios crash & burn.  I would suspect that very few lenders were putting any money into it (loan volume was low) and competition for good loans was low, so fewer questionable loans were made.

As Prosper stayed open for business longer, more dumb money followed in, making foolish choices, relying on Prosper's new grading system and generally causing more competition for good loans.  Thus more poor loans were made.

[ira01]

This supports the claim that the 3.0 (unsecured) loans are going bad less fast than the 1.0 (secured) loans.

Less fast, but still too fast.  About 6.5% of the September 09 loans are in the crapper already, despite the oldest of them being a mere 11 months old.  That doesn't bode well.  And even worse, almost 6% of the November 09 loans are in the crapper, despite being only 9 months old. 

[kenL]

About 6.5% of the September 09 loans are in the crapper already, despite the oldest of them being a mere 11 months old.  That doesn't bode well.  And even worse, almost 6% of the November 09 loans are in the crapper, despite being only 9 months old.  

September and November 09 are both suffering from the cashcowboy effect. im-sharky accounted for about 10% of September defaults and about 30% of November defaults. He did most of his lending in those 2 months.

Edit:
As far as 1+ month lates go, im-sharky has about 10k in 1+month late loans in September out of about 80k total, that's about 12.5% of the total, and about 30k in 1+month late loans in November out of about 150k, that's about 20% of the total.

[Urbi_et_Orbi]

Prosper has really gone out of their way to hide the link to the performance page.  Probably not long until they will claim, like they did with their forum, that since so few people access them, the performance data is not needed.

So, as a public service, here is a link to the performance data:

http://www.prosper.com/invest/performance.aspx

[God-Father]

Wow - it is hidden.  You can find it pretty quickly on the site map, but without that, it is just a matter of luck.