Author Topic: Cumulative ROI by Vintage Beginning 14Q1  (Read 2353 times)

rawraw

  • Hero Member
  • *****
  • Posts: 2746
    • View Profile
Re: Cumulative ROI by Vintage Beginning 14Q1
« Reply #15 on: January 08, 2018, 07:34:15 PM »
I don't think we can ignore that low grade LC notes performed poorly at the same time low FICO consumer loans were performing badly at lenders at large.  While LC's desire for growth and loosening standards didn't help, these are actual typical of cycles.  Lending standards relax as competition heats up and then losses cause them to tighten, so management's decisions is correlated with the cycle.  Some managers avoid falling for this trap, but they sacrifice growth. 

Fred93

  • Hero Member
  • *****
  • Posts: 2063
    • View Profile
Re: Cumulative ROI by Vintage Beginning 14Q1
« Reply #16 on: January 08, 2018, 07:55:15 PM »
I don't think we can ignore that low grade LC notes performed poorly at the same time low FICO consumer loans were performing badly at lenders at large.  While LC's desire for growth and loosening standards didn't help, these are actual typical of cycles.

I think you imagined that "cycle".  2014-2015-2016 looks pretty damn flat to me.  In fact, the fed data shows 2015 to be a little better.  Its at the trough of the chargeoff curve.  The 2015 loans from Lendingclub were the stinko loans. 

So I'm not so quick to accept excuses.


nonattender

  • Hero Member
  • *****
  • Posts: 710
  • I am not here.
    • View Profile
Re: Cumulative ROI by Vintage Beginning 14Q1
« Reply #17 on: January 08, 2018, 08:54:05 PM »
I don't think we can ignore that low grade LC notes performed poorly at the same time low FICO consumer loans were performing badly at lenders at large.  While LC's desire for growth and loosening standards didn't help, these are actual typical of cycles.  Lending standards relax as competition heats up and then losses cause them to tighten, so management's decisions is correlated with the cycle.  Some managers avoid falling for this trap, but they sacrifice growth.

Meta, it sounds like you may've gotten a job with a bunch of banker/suits.  You seemed sharper before.  They may be dragging you down.
A little nonsense now and then is relished by the wisest men.

rawraw

  • Hero Member
  • *****
  • Posts: 2746
    • View Profile
Re: Cumulative ROI by Vintage Beginning 14Q1
« Reply #18 on: January 09, 2018, 03:52:50 AM »
I don't think we can ignore that low grade LC notes performed poorly at the same time low FICO consumer loans were performing badly at lenders at large.  While LC's desire for growth and loosening standards didn't help, these are actual typical of cycles.

I think you imagined that "cycle".  2014-2015-2016 looks pretty damn flat to me.  In fact, the fed data shows 2015 to be a little better.  Its at the trough of the chargeoff curve.  The 2015 loans from Lendingclub were the stinko loans. 

So I'm not so quick to accept excuses.


I don't know how many times you're going to cite the Fred statistics. I think I've argued with that data source enough, but it never seems to stick. If you want to think every market observer imagined the rise in credit costs across the Subprime space, then that's fine.
 
For others who don't have the history of my critiques, this is what my first google result said :

NEW YORK, March 14 2016 (Fitch) Delinquencies on U.S. subprime auto ABS have eclipsed 2009 recessionary levels and are now at a level not seen in nearly two decades, according to Fitch Ratings. Subprime delinquencies of 60 days or more hit 5.16% for February reporting, marking the highest level observed since October 1996 (5.96%).

@nonattender I'm sorry I won't join the LC bashing club. I know it's popular on this forum, for whatever reason, to act like LC is the sole source of evil. I just suspect it's because most people's exposure to credit markets is just these peer lending products. I never said that they are blameless or that I agreed with what they did. And I don't understand how explaining how credit standards change is somehow banking propaganda. But whatever, this is enough for now lol

hessinger

  • Newbie
  • *
  • Posts: 38
    • View Profile
Re: Cumulative ROI by Vintage Beginning 14Q1
« Reply #19 on: January 09, 2018, 04:46:46 AM »
Instead my theory FWIW is that LC's management of its underwriting criteria has had a much larger impact over this period

I would be curious to see this same style of the graph created on Prosper data to compare not that that is a definite baseline. Anecdotally, I have several LC and Prosper accounts that were running in parallel over the past few years, and while I've seen a slight drop in performance on my prosper accounts, they haven't seen anywhere near the drop I see on my LC accounts. It was certainly around 2016 when I started to see a pretty large divergence in performance.
« Last Edit: January 09, 2018, 04:49:46 AM by hessinger »

Fred93

  • Hero Member
  • *****
  • Posts: 2063
    • View Profile
Re: Cumulative ROI by Vintage Beginning 14Q1
« Reply #20 on: January 09, 2018, 05:42:33 AM »
If you want to think every market observer imagined the rise in credit costs across the Subprime space, then that's fine.
...
NEW YORK, March 14 2016 (Fitch) Delinquencies on U.S. subprime auto ABS have eclipsed 2009 recessionary levels and are now at a level not seen in nearly two decades, according to Fitch Ratings. Subprime delinquencies of 60 days or more hit 5.16% for February reporting, marking the highest level observed since October 1996 (5.96%).

The world is complex, and you can look at data sliced up may different ways.  There are many different statistics out there for many different slices and subgroups.  You're quoting "subprime" statistics, and in particular "subprime auto".  There is no question that subprime auto went to heck in 2015/6/7.  I don't deny it.  Its just that we weren't supposed to be in that category.

LC loans aren't supposed to be "subprime", and they aren't "auto" loans.  Different people have different definitions of "subprime".  Bloomberg in the chart below uses "credit score < 620" to represent "subprime".  But LC loans are all credit score >=660.



It is interesting to see in that chart that the subprime auto loans didn't perform like the subprime bank and credit union loans during 2016/7.  I think the reason for this is that during this period auto inventory levels were rising and the auto industry lowered their credit standards.  I can't prove it, but that explanation seems to fit the data.

Similarly, I believe LC and Prosper lowered their standards in the 2015/6 era.  LC "C" and "D" loans weren't supposed to behave like "subprime auto", but in those vintages, they did.

I do blame LC for lowering their credit standards during this time, and being entirely opaque about the fact that they were doing it.  Frankly I think it was a near-death experience for them.  If they had done just a little worse, they'd be out of business now.

To be clear, even the bank subprime loans (blue curve) did fine during this period. 

Having gone thru this period, I now have a deeper appreciation for the fact that the LC credit grades "A B C D E F G" are more plastic than I had imagined.  Just because "D" loans performed a certain way in the past, in various economic conditions, doesn't mean that the new "D" loans (which may be written to completely different credit standards) will bear any relationship to historical "D", etc.  Because of my conservatism I didn't bear the brunt of this.  My returns stayed positive, just came in a lot lower than I expected.  And... gave me quite a worry as they were falling, because I didn't know how far they would fall.  For me, and for many other LC investors, this caused a substantial loss of confidence in the LC team, 'cause we don't know when they will do it again.

I do believe they're recovering from the underwriting screwup of 2015/6.  I think the charts posted by Rob L at the beginning of this thread show that.  Its too soon to be sure, but it looks like at least a modest recovery.  The other way to observe this is to look at my own returns.  They've stopped going down.

rawraw

  • Hero Member
  • *****
  • Posts: 2746
    • View Profile
Re: Cumulative ROI by Vintage Beginning 14Q1
« Reply #21 on: January 09, 2018, 10:54:54 AM »
I think I better understand your point now. This  whole thread is how low grade LendingClub notes didn't perform well. Do you want to compare these notes to prime borrowers instead? Just because they are a few points above 660 doesn't mean they aren't impacted by the same variables. I disagree they shouldn't be subprime. They are low FICO borrowers paying credit card level rates on installment debt, many  at levels above the usury rate of their state. Maybe LC investors didn't appreciate this level of risk, but it's not like we haven't discussed it ad nauseum on the forum leading up to 2016.

Your auto explanation on the differences is likely correct. And yes, banks lend differently than non banks. Do you expect LendingClub to be similar to a bank? I do not.

And yes, I agree that I don't think many people understand of credit standards are not fixed. It is a very important part of the credit dynamic.  That's why I think it's probably best to backtesting certain characteristics and not certain grades, since grade is a made up thing that changes whenever they want it to.

Rob L

  • Hero Member
  • *****
  • Posts: 1982
    • View Profile
Re: Cumulative ROI by Vintage Beginning 14Q1
« Reply #22 on: January 09, 2018, 11:23:16 AM »
This  whole thread is how low grade LendingClub notes didn't perform well.

Personally I already knew (all too well) they have performed poorly.
I'm interested in looking to the future to see if things are getting better, and by how much. It will take time.

Also, I was surprised that, even in the higher grades, there was no risk premium to be found in riskier loans (say C vs B).
That's puzzling.

rawraw

  • Hero Member
  • *****
  • Posts: 2746
    • View Profile
Re: Cumulative ROI by Vintage Beginning 14Q1
« Reply #23 on: January 09, 2018, 12:18:59 PM »
This  whole thread is how low grade LendingClub notes didn't perform well.

Personally I already knew (all too well) they have performed poorly.
I'm interested in looking to the future to see if things are getting better, and by how much. It will take time.

Also, I was surprised that, even in the higher grades, there was no risk premium to be found in riskier loans (say C vs B).
That's puzzling.
Yeah I know, just trying to explain that LC wasn't the only one having weakness. What do you mean by risk premium? If it's the difference in earnings, this tends to disappear in times of stress. I've been replying on my phone so not the easiest to look at the graphs to see what you are referring to

Rob L

  • Hero Member
  • *****
  • Posts: 1982
    • View Profile
Re: Cumulative ROI by Vintage Beginning 14Q1
« Reply #24 on: January 09, 2018, 03:59:51 PM »
This  whole thread is how low grade LendingClub notes didn't perform well.

Personally I already knew (all too well) they have performed poorly.
I'm interested in looking to the future to see if things are getting better, and by how much. It will take time.

Also, I was surprised that, even in the higher grades, there was no risk premium to be found in riskier loans (say C vs B).
That's puzzling.
Yeah I know, just trying to explain that LC wasn't the only one having weakness. What do you mean by risk premium? If it's the difference in earnings, this tends to disappear in times of stress. I've been replying on my phone so not the easiest to look at the graphs to see what you are referring to

My definition of risk premium is that a basket of riskier loans (loans with more returns variance mathematically) should pay the lender a higher rate of return than a less risky basket of loans. Otherwise nobody would (or should) own the more risky basket. Why you may ask? Because the spread of loan returns within a fixed confidence interval (say 95%) for any single basket widens as risk (returns variance) increases (assuming the number of loans in the basket is constant). Diversification can take you only so far. One must own a lot more loans in a risky basket (say E grade) to achieve the mean return of all E grade loans +/- 1.5% with 95% confidence than A grade loans. The lender must be compensated for the fact he/she may simply be unlucky; it's unavoidable.

Take a look at this old thread for more:
https://forum.lendacademy.com/index.php/topic,2637.msg23080.html#msg23080

rawraw

  • Hero Member
  • *****
  • Posts: 2746
    • View Profile
Re: Cumulative ROI by Vintage Beginning 14Q1
« Reply #25 on: January 09, 2018, 04:31:01 PM »
Makes sense. I don't think there is much benefit to buying low grade notes, which is why I've historically held ~60% of my portfolio in ABC.  I never quantified it, I just noticed the very small difference in returns from people on this forum and myself despite my weighted average interest rate often being much lower. 

Rob L

  • Hero Member
  • *****
  • Posts: 1982
    • View Profile
Re: Cumulative ROI by Vintage Beginning 14Q1
« Reply #26 on: January 12, 2018, 05:42:17 PM »
Instead my theory FWIW is that LC's management of its underwriting criteria has had a much larger impact over this period

I would be curious to see this same style of the graph created on Prosper data to compare not that that is a definite baseline. Anecdotally, I have several LC and Prosper accounts that were running in parallel over the past few years, and while I've seen a slight drop in performance on my prosper accounts, they haven't seen anywhere near the drop I see on my LC accounts. It was certainly around 2016 when I started to see a pretty large divergence in performance.

Unfortunately the Inskit data for Prosper doesn't go out as far as we'd like. The comparison is very rough.
Anyway, here's a few comparisons FWIW:







It appears that LC investment in less risky loans outperformed Prosper but investments in Prosper more risky loans outperformed LC.
That leads me to believe @hessinger was invested in more risky loans and Prosper held up much better.
I got it backwards, investing in LC D & E Grades while investing in Prosper A & B Grades.
Now I invest on no Grades.




Rob L

  • Hero Member
  • *****
  • Posts: 1982
    • View Profile
Re: Cumulative ROI by Vintage Beginning 14Q1
« Reply #27 on: February 03, 2018, 11:37:02 AM »
This morning I took a look at the Insikt site to attempt to get a more timely measurement of recent loan performance by vintage. So, rather than Cumulative ROI, I looked at 30+ day delinquency rate as a percentage of outstanding balance, C grade, 36 month loans only. Data as of 12/31/2017 (very recent).

See: https://www.insikt.com/#/invest/mycro/vintage/delinquency30d?po=Partner&or=LendingClub&fr=Quarterly&tY=2017&tM=10&fY=2016&fM=4&fi=creditRating&sGO=C&lt=36

Unfortunately, the news is not good. Don't shoot the messenger (again).

Vintage     3 MOB      6 MOB
------        -------       -------
16Q2         0.64%      1.84%
16Q3         0.64%      1.93%
16Q4         0.51%      1.51%
17Q1         0.39%      1.50%
17Q2         0.51%      2.29%
17Q3         0.73%      n/a

At 6 months on books (MOB) 17Q2 has the all time highest 30+ day delinquency rate percentage on record. The same may be said for 17Q3 at 3 MOB.
Guess it's too early to draw any long term conclusions from this, but these last two vintages are going in the wrong direction.

Meanwhile, interest rates seem to be heading up at the moment; everywhere but LC that is.

Fred93

  • Hero Member
  • *****
  • Posts: 2063
    • View Profile
Re: Cumulative ROI by Vintage Beginning 14Q1
« Reply #28 on: February 03, 2018, 12:27:24 PM »
I often see that 30 day delinquency and 60 day delinquency both show me what looks like a trend, but going in opposite directions!  Extrapolation from either of these in the early months is just very noisy.