Has LC loan quality dropped?

[viking]

I'd have to look, as the site has changed lately.  But this blog post talks about it: http://www.nickelsteamroller.com/blog/2012/02/featured-filter-revolving-credit-balance/

I don't think that Michael has the Ratio available on his website (at least not today). Looking at the numbers in the blog though, confirms what I suggested above, a Ratio < 0.5 performs worse. In other words, people who claim that they will refinance only a portion of their revolving balance will perform worse that people who ask for more money.
Here are the results from the E-G General Strategy on IR's website (modified to only include Credit Card & Debt Consolidation purposes)

Ratio   Loan #   Loan %   Interest   Loss %     IRR        Std Error
0-0.4      84        10.30%   20.30%   10.60%      8.50%         9.10%
0.5-0.9   274       33.50%   21.10%   3.50%     16.20%          3.20%
1.0-1.1   125       15.30%   21.30%   2.00%     17.10%          5.00%
1.2-1.9   225       27.50%   21.70%   3.90%     16.00%          4.50%
> 2.0      111       13.60%   21.70%   4.40%     15.20%          6.20%

[rawraw]

I'd have to look, as the site has changed lately.  But this blog post talks about it: http://www.nickelsteamroller.com/blog/2012/02/featured-filter-revolving-credit-balance/

I don't think that Michael has the Ratio available on his website (at least not today). Looking at the numbers in the blog though, confirms what I suggested above, a Ratio < 0.5 performs worse. In other words, people who claim that they will refinance only a portion of their revolving balance will perform worse that people who ask for more money.
Here are the results from the E-G General Strategy on IR's website (modified to only include Credit Card & Debt Consolidation purposes)

Ratio   Loan #   Loan %   Interest   Loss %     IRR        Std Error
0-0.4      84        10.30%   20.30%   10.60%      8.50%         9.10%
0.5-0.9   274       33.50%   21.10%   3.50%     16.20%          3.20%
1.0-1.1   125       15.30%   21.30%   2.00%     17.10%          5.00%
1.2-1.9   225       27.50%   21.70%   3.90%     16.00%          4.50%
> 2.0      111       13.60%   21.70%   4.40%     15.20%          6.20%

I found in my analysis the range of .8 to 1.2 produce a larger sample than just 1, while still maintaining an elevated ROI.  Look at that standard error for a ratio  < 0.4, though.  The sample size is smaller than the other samples which you are comparing it to.

[viking]

I'd have to look, as the site has changed lately.  But this blog post talks about it: http://www.nickelsteamroller.com/blog/2012/02/featured-filter-revolving-credit-balance/

I don't think that Michael has the Ratio available on his website (at least not today). Looking at the numbers in the blog though, confirms what I suggested above, a Ratio < 0.5 performs worse. In other words, people who claim that they will refinance only a portion of their revolving balance will perform worse that people who ask for more money.
Here are the results from the E-G General Strategy on IR's website (modified to only include Credit Card & Debt Consolidation purposes)

Ratio   Loan #   Loan %   Interest   Loss %     IRR        Std Error
0-0.4      84        10.30%   20.30%   10.60%      8.50%         9.10%
0.5-0.9   274       33.50%   21.10%   3.50%     16.20%          3.20%
1.0-1.1   125       15.30%   21.30%   2.00%     17.10%          5.00%
1.2-1.9   225       27.50%   21.70%   3.90%     16.00%          4.50%
> 2.0      111       13.60%   21.70%   4.40%     15.20%          6.20%

I found in my analysis the range of .8 to 1.2 produce a larger sample than just 1, while still maintaining an elevated ROI.  Look at that standard error for a ratio  < 0.4, though.  The sample size is smaller than the other samples which you are comparing it to.

I do not understand what you mean by "a larger sample than just 1"?
Yes, the SE is larger for R<0.4 because there are fewer such loans and the spread in returns is probably also higher.
There are more historical loans on NSR than IR because NSR also includes loan types that does not meet today's standards (loan types that can therefore no longer be found on LC). You will therefore, in general, find more loans for many Strategies on NSR than on IR.

[rawraw]

I'd have to look, as the site has changed lately.  But this blog post talks about it: http://www.nickelsteamroller.com/blog/2012/02/featured-filter-revolving-credit-balance/

I don't think that Michael has the Ratio available on his website (at least not today). Looking at the numbers in the blog though, confirms what I suggested above, a Ratio < 0.5 performs worse. In other words, people who claim that they will refinance only a portion of their revolving balance will perform worse that people who ask for more money.
Here are the results from the E-G General Strategy on IR's website (modified to only include Credit Card & Debt Consolidation purposes)

Ratio   Loan #   Loan %   Interest   Loss %     IRR        Std Error
0-0.4      84        10.30%   20.30%   10.60%      8.50%         9.10%
0.5-0.9   274       33.50%   21.10%   3.50%     16.20%          3.20%
1.0-1.1   125       15.30%   21.30%   2.00%     17.10%          5.00%
1.2-1.9   225       27.50%   21.70%   3.90%     16.00%          4.50%
> 2.0      111       13.60%   21.70%   4.40%     15.20%          6.20%

I found in my analysis the range of .8 to 1.2 produce a larger sample than just 1, while still maintaining an elevated ROI.  Look at that standard error for a ratio  < 0.4, though.  The sample size is smaller than the other samples which you are comparing it to.

I do not understand what you mean by "a larger sample than just 1"?
Yes, the SE is larger for R<0.4 because there are fewer such loans and the spread in returns is probably also higher.
There are more historical loans on NSR than IR because NSR also includes loan types that does not meet today's standards (loan types that can therefore no longer be found on LC). You will therefore, in general, find more loans for many Strategies on NSR than on IR.

It means making the ratio 1 instead of a range of .8 to 1.2 eliminates a lot of loans without much of a benefit.

And what do you mean by "loan types that don't meet today's standards?"  Are you talking about that bit of information on the download file?

Yes, the SE is larger for R<0.4 because there are fewer such loans and the spread in returns is probably also higher.

Which means you can't simply compare the IRR of one to the other.  It's a little more nuanced than that.  They may or may not be statistically different.

[viking]

I'd have to look, as the site has changed lately.  But this blog post talks about it: http://www.nickelsteamroller.com/blog/2012/02/featured-filter-revolving-credit-balance/

I don't think that Michael has the Ratio available on his website (at least not today). Looking at the numbers in the blog though, confirms what I suggested above, a Ratio < 0.5 performs worse. In other words, people who claim that they will refinance only a portion of their revolving balance will perform worse that people who ask for more money.
Here are the results from the E-G General Strategy on IR's website (modified to only include Credit Card & Debt Consolidation purposes)

Ratio   Loan #   Loan %   Interest   Loss %     IRR        Std Error
0-0.4      84        10.30%   20.30%   10.60%      8.50%         9.10%
0.5-0.9   274       33.50%   21.10%   3.50%     16.20%          3.20%
1.0-1.1   125       15.30%   21.30%   2.00%     17.10%          5.00%
1.2-1.9   225       27.50%   21.70%   3.90%     16.00%          4.50%
> 2.0      111       13.60%   21.70%   4.40%     15.20%          6.20%

I found in my analysis the range of .8 to 1.2 produce a larger sample than just 1, while still maintaining an elevated ROI.  Look at that standard error for a ratio  < 0.4, though.  The sample size is smaller than the other samples which you are comparing it to.

I do not understand what you mean by "a larger sample than just 1"?
Yes, the SE is larger for R<0.4 because there are fewer such loans and the spread in returns is probably also higher.
There are more historical loans on NSR than IR because NSR also includes loan types that does not meet today's standards (loan types that can therefore no longer be found on LC). You will therefore, in general, find more loans for many Strategies on NSR than on IR.

It means making the ratio 1 instead of a range of .8 to 1.2 eliminates a lot of loans without much of a benefit.

And what do you mean by "loan types that don't meet today's standards?"  Are you talking about that bit of information on the download file?

Yes, the SE is larger for R<0.4 because there are fewer such loans and the spread in returns is probably also higher.

Which means you can't simply compare the IRR of one to the other.  It's a little more nuanced than that.  They may or may not be statistically different.

1. One of the ratio ranges is actually set to 1.0-1.1 on IR. However, you can bin the ranges together in any way you like (e.g. 0.5-1.1; 0.5-1.9; 1.0-1.9). Based on the table above, I do not think that it would make a large difference. 

2. I was talking about those old types of loans that LC no longer issues. They are for completeness, listed in the history files as "Loans that do not meet the current credit policy" and are also included in the Statistics on NSR (but  not on IR).

3. The mean (IRR) of two populations can be compared using a simple t-test, even though the standard errors are different. If you want to be fancy and do an ANOVA test, then you are correct. The standard deviations should be comparable (but do not have to be identical).

[rawraw]

2. I was talking about those old types of loans that LC no longer issues. They are for completeness, listed in the history files as "Loans that do not meet the current credit policy" and are also included in the Statistics on NSR (but  not on IR).

But how exactly is that determined?

And I'm not a statistic boss like Bryce, but I'm unsure if a simple T-Test will suffice -- since T-Test assume equal variance among those being compared . We'd probably need to use Welch's T-Test.

[AmCap]

You all need / deserve a strong glass of egg nog :-)

[brycemason]

I agree on the egg-nog bit.

It's hard to have a clear discussion on the web, as the way one defines the outcomes, makes their data set, and conducts their method has much to do with the results. Often people don't lay that out clearly enough to have a fruitful discussion, or readers gloss over key words that have important meaning.

We should have an in-person conference or workshop! Then we could dig in and have more time to learn from each other.

[viking]

2. I was talking about those old types of loans that LC no longer issues. They are for completeness, listed in the history files as "Loans that do not meet the current credit policy" and are also included in the Statistics on NSR (but  not on IR).

But how exactly is that determined?

That was determined by Lending Club. They have changed their underwriting rules but still keep those old notes that no longer meet their current criteria in the history file.

p.s. Merry Christmas, egg-no or not 

[Lovinglifestyle]

I don't know if my observation is correct.  It seems that the quality of LC loans has dropped lately.  I have noticed more loans with "Last Major Derogatory" coming on their pipeline than before.  And I have not been able to find loans that fit my criteria as quickly as I did in the past.  I was able to find at least 1 or 2 loans using the same criteria just a couple of months ago but now I have to go for days without being able to put any money to work.  (And it seems that there are more 60-month C loans coming on the platform as well.)

Can someone shed some light on this?

"PRE-APPROVED OFFER IN THE MAIL" referenced by this applicant could be related.  (Loan 2825704 if the link doesn't work). 

Summary:  some applicants get annoyed by questions because they don't realize where the questions come from.  They can become annoyed enough to withdraw the application (my guess).

https://www.lendingclub.com/browse/loanDetailFull.action?loan_id=2825704&previous=

[investforfreedom]

I don't know if my observation is correct.  It seems that the quality of LC loans has dropped lately.  I have noticed more loans with "Last Major Derogatory" coming on their pipeline than before.  And I have not been able to find loans that fit my criteria as quickly as I did in the past.  I was able to find at least 1 or 2 loans using the same criteria just a couple of months ago but now I have to go for days without being able to put any money to work.  (And it seems that there are more 60-month C loans coming on the platform as well.)

Can someone shed some light on this?

"PRE-APPROVED OFFER IN THE MAIL" referenced by this applicant could be related.  (Loan 2825704 if the link doesn't work). 

Summary:  some applicants get annoyed by questions because they don't realize where the questions come from.  They can become annoyed enough to withdraw the application (my guess).

https://www.lendingclub.com/browse/loanDetailFull.action?loan_id=2825704&previous=

It is interesting that you caught this one.  It seems that this applicant might have received a solicitation from LC, only to be turned off by lenders asking him/her questions on the fly.  What could he/she have expected otherwise?  This individual was asking people nearly $19K.  Don't borrow if you can't even have the courtesy and patience of answering a few questions for $19K.  You would have to jump through some hoops even going on Wheel of Fortune, wouldn't you?  At any rate, I wouldn't have pitched in even if he/she hadn't withdrawn.  He/she has a public record, even though it is dated back nearly 9 years.  Gross income is about $4K per month, but monthly payment is $530 and borrowing 2x revolving credit balance?  It sounds like a bad risk anyway.

In any event, I am seeing so many "major derogatory" showing up in the D-G loans--even C loans as well.  I don't know if it is because they were unaccounted for by LC before or whether LC has relaxed its standards to allow applicants with higher risks.

[Lovinglifestyle]

I don't know if my observation is correct.  It seems that the quality of LC loans has dropped lately.  I have noticed more loans with "Last Major Derogatory" coming on their pipeline than before.  And I have not been able to find loans that fit my criteria as quickly as I did in the past.  I was able to find at least 1 or 2 loans using the same criteria just a couple of months ago but now I have to go for days without being able to put any money to work.  (And it seems that there are more 60-month C loans coming on the platform as well.)

Can someone shed some light on this?

"PRE-APPROVED OFFER IN THE MAIL" referenced by this applicant could be related.  (Loan 2825704 if the link doesn't work). 

Summary:  some applicants get annoyed by questions because they don't realize where the questions come from.  They can become annoyed enough to withdraw the application (my guess).

https://www.lendingclub.com/browse/loanDetailFull.action?loan_id=2825704&previous=

It is interesting that you caught this one.  It seems that this applicant might have received a solicitation from LC, only to be turned off by lenders asking him/her questions on the fly.  What could he/she have expected otherwise?  This individual was asking people nearly $19K.  Don't borrow if you can't even have the courtesy and patience of answering a few questions for $19K.  You would have to jump through some hoops even going on Wheel of Fortune, wouldn't you?  At any rate, I wouldn't have pitched in even if he/she hadn't withdrawn.  He/she has a public record, even though it is dated back nearly 9 years.  Gross income is about $4K per month, but monthly payment is $530 and borrowing 2x revolving credit balance?  It sounds like a bad risk anyway.

In any event, I am seeing so many "major derogatory" showing up in the D-G loans--even C loans as well.  I don't know if it is because they were unaccounted for by LC before or whether LC has relaxed its standards to allow applicants with higher risks.

According to LC, the "major derogatory" attribute definition as provided by TransUnion is "months since most recent 90-day or worse rating".   I think this is helpful information which was simply not previously included.

[rawraw]

I don't know if my observation is correct.  It seems that the quality of LC loans has dropped lately.  I have noticed more loans with "Last Major Derogatory" coming on their pipeline than before.  And I have not been able to find loans that fit my criteria as quickly as I did in the past.  I was able to find at least 1 or 2 loans using the same criteria just a couple of months ago but now I have to go for days without being able to put any money to work.  (And it seems that there are more 60-month C loans coming on the platform as well.)

Can someone shed some light on this?

"PRE-APPROVED OFFER IN THE MAIL" referenced by this applicant could be related.  (Loan 2825704 if the link doesn't work). 

Summary:  some applicants get annoyed by questions because they don't realize where the questions come from.  They can become annoyed enough to withdraw the application (my guess).

https://www.lendingclub.com/browse/loanDetailFull.action?loan_id=2825704&previous=

As I thought he would, Lending Club CEO Renaud Laplanche replied to yesterday’s post with exactly the numbers I asked for. The loss rate for loans where Lending Club has verified the borrower’s income is 2.8%; the loss rate for loans where Lending Club hasn’t verified the borrower’s income is lower, at 2.7%. So Ron Lieber’s worries do indeed seem to be misplaced.*

Laplanche explains why Lending Club doesn’t verify income on all its loans:

The main reason why we do not perform income verification on 100% of the loans is to avoid adverse selection: the borrowers who have perfect credit history and do not exhibit any particular risk factors (who fall into the 40% we do not verify) are also those who have the least tolerance for a cumbersome income verification process, and are most likely to abandon that process and seek funding elsewhere. They are, however, the exact kind of borrowers we want to retain.

http://blogs.reuters.com/felix-salmon/2011/02/07/lending-clubs-loss-rate-numbers/

[faeriering]

@ rawraw & viking,  sorry for causing such a stir.

I'm still using lendstats to do much of my analysis because I've been lazy and haven't wanted to build a spreadsheet and pull out my stats book.  So I'm probably making choices a little less off of analytical results (ie I'm not doing a t-test or the like to see if the change that I'm seeing is just noise in the data or a real measureable difference).  Many of my loans are debt consolidation, but not all.  I do try to shoot for a ratio of .8-1.2, I tend to just do the math in my head instead of running it through a spreadsheet when I'm evaluating the loans for this ratio.  Lendstats does have some data on this ratio and there is a bit of a difference in ROI from the loans that are close to one and those that are further off, but that is with the rest of my filters in place.

Subjectively, I know Ken L really liked loans that were close to 1 on the ratio of Loan/revolving debt as long as the fit the rest of his loan criteria as well.  One of the challenges I have to this, is that not all the debt shows up in revolving debt.  If the person has several lines of equity against their home, these wouldn't show up here . . . of course their interest rate on those loans should be <6% so why they would want to consolidate to a higher rate . . . doesn't make much sense to me, but I wouldn't have gotten into the situation that a lot of these borrowers are in anyways, at least I hope I don't.

[rawraw]

@ rawraw & viking,  sorry for causing such a stir.


Subjectively, I know Ken L really liked loans that were close to 1 on the ratio of Loan/revolving debt as long as the fit the rest of his loan criteria as well.  One of the challenges I have to this, is that not all the debt shows up in revolving debt.  If the person has several lines of equity against their home, these wouldn't show up here . . . of course their interest rate on those loans should be <6% so why they would want to consolidate to a higher rate . . . doesn't make much sense to me, but I wouldn't have gotten into the situation that a lot of these borrowers are in anyways, at least I hope I don't.

But that's the key for me.  I have verification of CC debt.  So my favorite loan is a CC consolidation.  I  have to "trust" that their statements about other debts is accurate.  And generally, in this interest rate environment, their debts are lower than a LC interest rate.  I avoid people who refinance to higher interest rates.