Print

[2potatoes]

So as a lender I have found myself avoiding loans where the borrower indicates that they "need" the loan, but up until now I haven't had any rational basis for doing so, other than the general belief that if someone needs me to lend to them that they are less likely to be able to pay me back in the future.

I've just performed an analysis of some old loans and found that I'm not nuts. I haven't seen a similar analysis anywhere else yet, and thought others might be interested in these findings:

Description	Total Loans	Never Recovered	Recovered	Paid
Pre 2007 loans where title contains "need"	652	46.9%	52.8%	51.2%
Pre 2007 loans where title does not contain "need"	5321	37.0%	62.9%	61.4%
2007 loans where title contains "need"	1092	47.7%	52.3%	50.6%
2007 loans where title does not contain "need"	10387	37.3%	62.7%	61.5%
2008 loans where title contains "need"	861	34.8%	48.9%	48.6%
2008 loans where title does not contain "need"	10704	30.3%	55.0%	54.3%

So at first glance it looks like in loans that have completed (up to and including all of 2007) there's nearly a 10% difference in the percentage of loans that are not fully recovered and those that aren't when the title contains the word "need". Not all 2008 loans have completed yet--it would be interesting to look at the correlation between the word "need" and the status of late loans.

For the purpose of this analysis I considered these statuses to be Recovered:
Repurchased, Defaulted (SettledInFull), Defaulted (PaidInFull) and Paid.

I considered these statuses to be Never Recovered:
Charge-Off, Defaulted (Bankruptcy) and Defaulted (Delinquency)

Edit to add that I've started a blog covering the topic and to show the charts a bit more clearly: http://lendingtuber.blogspot.com/

[xraider]

Very interesting... and even harder to tell how Prosper comes up with a 20% default rate...

[312lender]

Great stuff.  I think you could be on to something.  But there needs to be more analysis done.  Your numbers would be more telling if you broke it up by Prosper Score (AA, A, B...).

[frinxor]

Great stuff.  I think you could be on to something.  But there needs to be more analysis done.  Your numbers would be more telling if you broke it up by Prosper Score (AA, A, B...).

this. the word "need" might correlate with a lower Prosper score - which would explain the lower %s

[2potatoes]

Yeah, it does seem like there's a difference across most of the credit scores (with the exception of A's for some reason.)  Here is the data on loans originating from 2005-2007:

Credit Score	Paid With "Need"	Paid Without "Need"	Difference
AA	79.6%	87.0%	7.4%
A	76.3%	76.2%	-0.1%
B	60.5%	69.3%	8.8%
C	57.1%	62.3%	5.2%
D	53.0%	60.0%	7.0%
E	45.4%	49.3%	3.9%
HR	32.3%	37.0%	4.7%

I discuss this a bit more on my blog: http://lendingtuber.blogspot.com/2011/07/need-series-initial-findings.html but I'm starting to think that it's worth deeper exploration to see if something else is going on or if there is some actual correlation there.

Edit: A's are -.1 instead of .1 as no-whammies points out below

[frinxor]

Cool stuff. It makes sense by intuition - nice to see the numbers on it. You should also include the sample sizes for each category.

[pioneer11]

Cool stuff. It makes sense by intuition - nice to see the numbers on it. You should also include the sample sizes for each category.

We used to beat up folks on RML for appearing desperate.

[2potatoes]

Cool stuff. It makes sense by intuition - nice to see the numbers on it. You should also include the sample sizes for each category.

Yes, indeed. I made a correlation matrix for the loans today and it seems that the word "need" is correlated with failure to repay more when it's in the Description than it is when it's in the title. Still have to run the numbers to see if the results are statistically significant...

[cowdog]

If anyone thinks there is any sort of correlation or the numbers mean anything, I have a way you can get rich:

In 630 draws for the MegaMillions lottery, number 48 came up 71 times, but the number 49 only came up 39 times.

http://www.calottery.com/Games/MegaMillions/WinningNumbers/numberfrequency.htm

 

[2potatoes]

I certainly agree that the numbers might not mean anything at all and that it will take more than a simple correlation to draw a conclusion as to whether or not there is any significance in the findings.

However, unlike the lottery, picking loans is not a purely random affair. It seems feasible that the way a borrower thinks about a loan, as expressed in their loan description, can be correlated to the loan outcome.

Besides, if I'm already avoiding requests where the borrower is expressing a need for the loan, I might as well look for numbers to rationalize my behavior. 

[Ray Kremer]

Well it's like the old adage says, the best people to give loans to are the ones that don't need a loan. The idea being the people most desperate for a loan are the ones least likely to be able to handle paying it all back. Add to that another factor: people who are desperate for a loan but not smart enough to avoid sounding desperate in their Prosper write-up. If you're looking for purely qualitative factors in picking loans, it's as good as any I suppose.

[no-whammies]

Credit Score	Paid With "Need"	Paid Without "Need"	Difference
AA	79.6%	87.0%	7.4%
A	76.3%	76.2%	0.1%
B	60.5%	69.3%	8.8%
C	57.1%	62.3%	5.2%
D	53.0%	60.0%	7.0%
E	45.4%	49.3%	3.9%
HR	32.3%	37.0%	4.7%

Fixed.  A's swing the other way.

[no-whammies]

Also, you need the sample size for each.

[JGuide]

I've performed an analysis of ...
            percentage of loans that are not fully recovered ...when the title contains the word "need".

Try the word "Help"

[2potatoes]

At first glance "help" has a 9% difference between paid and unpaid loans -- with 1930 loans having "help" in the title. The word "please" fairs worse with a 16% difference between paid and unpaid loans -- but only 476 loans are titled with "please."  (These are all the Pre-2008 loans.)