Three Myths Investors Have About P2P Lending

It is amazing how many investors make their investment decisions based on so-called common sense. They “know” that small loans are a better investment or that you want a borrower with a low revolving balance on their credit cards. Because obviously this will lead to better returns. Really?

I prefer to look at the data history and use that as my guide. We can all have a hunch about something but hunches are often wrong. Today I am going to dispel several myths that I see repeated by bloggers and new investors on a regular basis. Here are three investment criteria that common sense suggests would lead to higher returns but that the history of both Lending Club and Prosper prove to be different.

To test my theories here I looked at all loan data on Lendstats from loans originating in 2009 and 2010 at both companies. These loans are all past their peak default period and have an average age of around 21 months.

1. Small Loan Sizes Are Best

This is one of the most common myths I see. Investors filter loans based on the loan size with the thinking that a smaller loan means a more manageable payment and therefore a lower likelihood of default. Now, what makes a small loan is subjective so I just chose the approximate median loan size to analyze roughly equal segments of loans.

Lending Club Loan SizeLending Club ROIProsper Loan SizeProsper ROI
Less than $10,0005.13%Less than $5,0004.22%
$10,000+5.79%$5,000+6.09%

At Lending Club the median loan during the period analyzed is right around $10,000. So, taking all loans of $1,000 – $9,999 generates an estimated ROI across all loan grades of 5.13% and loans of $10,000 or more have an ROI of 5.79%. This is before Lending Club increased the maximum loan size from $25,000 to $35,000 so all these loans are for $25,000 or less.

At Prosper this analysis is a little more complex simply because Prosper limits the loan amount borrowers can apply for based on loan grade. In order to try and compare apples to apples I just looked at loan grades AA and A because these borrowers can borrow any amount up to $25,000 the same as at Lending Club. Here the median loan size was around $5,000 but the same trend is seen. Loans below $5,000 returned an estimated ROI of 4.22% and loans of $5,000 or more returned 6.09% – a marked difference.

2. Choose borrowers with low revolving credit

One would expect that borrowers that have very little revolving credit would be the best kind of borrower to have. But one would be mistaken. First we should define revolving credit – basically we are talking about credit card balances here, specifically the total balance owing on all credit cards. The credit card companies report this data monthly to the credit bureaus and your credit report is updated.

Lending Club Revolving CreditLending Club ROIProsper Revolving CreditProsper ROI
Less than $8,5004.89%Less than $7,0008.61%
$8,500+6.17%$7,000+9.75%

At Lending Club the median revolving credit amount is around $8,500. So we take half the loans that have revolving credit of less than $8,500 and we get an estimated ROI of 4.89%. For $8,500 and above the ROI is 6.17% – a significantly higher number.

At Prosper the median amount for revolving credit is around $7,000. For loans below $7,000 the estimated ROI is 8.61% and for $7,000 and above it is 9.75% a similar large difference.

Again we see data that goes against common sense. Borrowers with more debt perform better than borrowers with little debt. Michael from Nickel Steamroller has just published an excellent analysis of revolving credit at Lending Club and provides much more information about this one filter.

3. Low Debt-to-income ratio is better

Debt-to-income is a ratio that expresses the percentage of a person’s income that goes towards paying debts. For p2p lending purposes both Lending Club and Prosper do not include any mortgage debt in the debt to income ratio. Lending Club has rules around debt to income – if your credit score is below 720 the maximum debt to income ratio you can have is 25%, for scores of 720 or more that increases to 30%. Prosper has no publicly available rules like that but we know they do take it into consideration when doing their underwriting.

One important point I should note about debt to income. Not all borrowers are verified for income so this ratio may not reflect a borrowers real income in all cases. Having said that, the numbers are still useful, particularly reflective of what they don’t show.

Lending Club Debt to IncomeLending Club ROIProsper Debt to IncomeProsper ROI
Less than 13%5.59%Less than 18%9.08%
13%+5.75%18%+9.42%

At Lending Club the median DTI ratio is around 13% and at Prosper it is 18%. This is expected because Prosper allows a higher risk borrower on to their platform than Lending Club (with corresponding higher interest rates). Still you would have expected that a higher debt to income ratio would lead to a lower ROI for investors. But we see in the table above that is not the case. Now, these numbers are close so I am not going to categorically state that a higher DTI ratio is better but we can say that lower DTI doesn’t necessarily lead to higher returns.

There are other examples where common sense is not rewarded. Do some analysis on credit score or even length of employment (only available at Lending Club) and you will be surprised that what you expected to be true is not reflected in the data.

The Key Takeaway

The important lesson here is this. Don’t rely on what you think should be correct when investing. There is enough loan history for investors to see what has and hasn’t worked in the past. I think investors should base their investment decisions on data and not on a hunch that may or may not be true.

What do you think? Do you still use criteria that you think should lead to higher returns or do you look closely at the data? As always I am interested to hear your comments.

Subscribe
Notify of
46 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
Michael
Feb. 27, 2012 7:22 pm

Great write up! Another point about myth 1 is smaller loans will always fund faster since they require less money to fund. If you are sorting by percent funded because you think the community has vetted the loan, it might be a “crowd sourcing trap”.

Dan B
Dan B
Feb. 27, 2012 8:10 pm

Regarding the inclusive time period & data for Lending Club, is it accurate to state that the $35k maximum occurred after the period examined & was therefore not included in your analysis of large loan sizes? Would your conclusions still hold true if an adjustment were made to include them? How would investment decisions going forward be affected by this new reality

Bryce M.
Bryce M.
Feb. 27, 2012 8:14 pm

Looking at tables is a good place to start, but better analyses will have multiple contingencies allowed to control for other factors. For example, to expand on your Point #1, suppose that a very large number of the larger loans are also credit card refinancings. Then the result that you conclude of larger loans performing better is spurious, and simply related to the aforementioned fact. If you had made a two dimensional table that controlled for type of loan, you might have seen that larger loans perform worse.

Tables can only go so far, and a modeling framework that allows for multiple, simultaneous controls may reverse some of your points above (it did for me).

Roy S
Roy S
Feb. 27, 2012 8:32 pm

@Bryce, I think you are making Peter’s point. He is saying that there are people who seem to jump to a conclusion about one factor or another, but do not do any further research. Therefore, to the person who thinks that smaller loans will perform better, they aren’t looking at anything else. Obviously, no one filter can predict the best possible scenario, and that is what I believe Peter is saying.

I use Lendstats to develop my own personal investing strategy in hopes of achieving the best results. As more information comes in, I will need to continually adjust the metrics to best account for the changes in the borrower demographics.

Newlyfrugal
Newlyfrugal
Feb. 27, 2012 9:11 pm

I have 301 loans at LC and occasionally invested in smaller loans of $2k to $6k. One loan for $2k required only $61 in monthly payments for 36 months. I thought that I was safe due to the small payments. Wrong. The debtor NEVER made one single payment and LC charged off the loan after several months. Now I don’t even look at small loans ($2k to $6k.) Buyer beware.

Bryce M.
Bryce M.
Feb. 27, 2012 9:16 pm

I dunno, Roy, it seems pretty clear Peter is trying to debunk the idea that smaller loans are better using a simple table. “today i am going to dispell several myths.” Unidimensional analyses can be dangerous.

Smaller loans really do tend to default less once you account for many other confounding factors such as loan type.

Roy S
Roy S
Feb. 27, 2012 9:56 pm

@Bryce, I understand why you would say that. I haven’t taken the time to look specifically at loan size to determine whether smaller loans really do tend to default less as you state. I’m assuming that you’re not only accounting for loan type, but geographic location, occupation, credit score, DTI, inquiries, interest rate on the loan, etc in order to state that smaller loans default less frequently. Also, what do you consider to be a small loan? There are so many aspects to a borrower’s profile that it is difficult to narrow down what factors determine who is most likely to default, especially when so few Notes have been issued. Therefore, I return to my original position in that Peter was trying to point out that such uni-dimensional analysis is flawed. Again, I haven’t looked at the data, but taking Peter’s argument if someone were to have invested in all “small” loans and funded them 100% and another had done the same for large loans, it would hold true that the person who invested in the large loans would have outperformed the person who invested in the small loans. So a person who looks solely at the size of the loan because they “know” that small loans perform better than large loans should be expected to have a lower than average ROI. Obviously, I can’t read his mind, but this was my interpretation of this post.

Roy S
Roy S
Feb. 27, 2012 10:01 pm

I should note that my analysis is based on the assumption that a person who “knows” small loans perform better than large loans would only be taking into account the size of the loan and nothing else. Therefore Peter’s uni-dimensional analysis is fitting.

On another note, at Prosper (where loan size is limited based on the member’s Prosper grade) it wouldn’t surprise me that if Peter had done the analysis for all loan grades instead of AA and A that loan size would actually be a good predictor of ROI and that smaller loans would perform better as the overall ROI for the lower grades are currently higher for the lower grades than for the AA and A grades. So, the person may end with a higher ROI for the wrong reason, and that reason may change in the future.

Dan B
Dan B
Feb. 27, 2012 10:05 pm

Roy & Bryce………….Perhaps I’m missing something in this quick moving discussion. but isn’t the unspoken factor here the simple reality that in most cases your average $3k borrower doesn’t have the same financial profile as your average $30k borrower?

Bryce M.
Bryce M.
Feb. 27, 2012 10:17 pm

Dan, that could be an additional confound worthy of control (an extra dimension in an n-dimensional table).

Roy, sounds like we are on the same page. I just felt it important to note that someone walking away under the impression that the reality goes against common sense would be wrong.

I have only studied LC performance. Prosper may have other relationships. Lastly, I define small as anything below a certain (undisclosed) cutoff! Hah.

Money Infant
Feb. 28, 2012 6:19 am

I had no idea that P2P lending analysis had become so complex, but this is good information to get someone thinking. I admit I’ve never waded into the P2P lending pool, but I did have the idea that smaller loans would be better. Glad I saw this post first.

Brad
Brad
Feb. 28, 2012 6:43 am

This is very interesting.Considering P2P lending is so young— I am wondering if you know what the default rate in loans given by a bank (unsecured signature loans) –and if loan amounts (small or larger) are also not a factor in the banking industry?

What do you think is the most important factor to look for in a loan? Does it matter what state a borrower is from?

Thanks Brad

Lou lamoureux
Lou lamoureux
Feb. 28, 2012 9:31 am

@Bryce it’s pretty common to read a blog post by someone who’s got a couple hundred bucks into LC and he lists his criteria something like this “I only invest in loans under $x because the payments are manageable and I want borrowers who have been employed 10+ years because that shows they have a good job.” as Peter has stated, they don’t do ANY analysis, they just call it a “common sense” approach.
As Peter says in his Key Takeaway “don’t rely on what you think should be correct when there is so much loan history available” He’s not advocating large loans. He’s writing to the n00bs who don’t do ANY research and he’s trying to educate them.
Lou

Terry
Terry
Feb. 28, 2012 11:58 am

I used the NSR website to tweak out an LC filter, and I was amazed at how many of the tweaks were contrary to what I would have thought was common sense.

KenL
Feb. 28, 2012 2:31 pm

Another thing to note is that in the early days of Prosper (the wild west days) and in the early days of LC small loans did perform better and defaulted less than larger ones in large part because there were a lot more scammers going after p2p loans and of course the scammers went for the gold. However, the performance of larger loans gradually improved primarily because both companies improved flagging, verification and grading methods. These days large loans generally go through a lot more verification than smaller ones and this certainly helps their performance.

Roy S
Roy S
Feb. 28, 2012 6:08 pm

I’ve been seeing a few changes over at Lendstats, but it seems to be more of a rarity to actually hear from Ken on this blog. And with Ken doing better than 99.7% of everyone else, most of us could probably use a little more education from him! (myself included)

I’d be interested to know how much research people put into p2p lending prior to opening an account. When Prosper and LC are very transparent with their loan data, it is difficult to justify not digging into it. And with sites like Lendstats out there that are dedicated to aiding people in slicing and dicing the data, I don’t see why your post, Peter, should even be necessary–unfortunately, it seems it is. In my research I came across both your blog and Lendstats. And Prosper made it easy for me: https://www.prosper.com/tools/3rdParty.aspx. (Actually, when I first looked at this page I believe there were about 6 sites they listed. Of those the only 2 that worked are the only two still listed.) I’m not saying everyone should become an expert prior to opening an account, but a little due diligence goes a long way. It helps you avoid the big mistakes…or at least gives you the opportunity to correct any you may have already made.

Bryce M
Bryce M
Feb. 28, 2012 6:25 pm

Although I can only speak for myself, I spent about 20 hours in one weekend before I opened an account. In that time I had developed my strategy and the tools necessary to maintain the investment. In the month since, I’ve put in maybe another 40 hours checking new hypotheses, adding data outside of LC (state- and county-level Federal Reserve data/BLS), and validating the model on loans not used in my original research (that had since completed in that month).

Of course, I have put in years of time learning how to code and work with data, so my 20 hours might be someone else’s 200.

Dan B
Dan B
Feb. 28, 2012 7:24 pm

Roy S………….I agree that it’s great to hear from Ken, but a clarification/correction is in order. Ken is not doing better than 99.7% of “everyone else”, as you put it. He is doing better than 99.7% of those who who are investing in Prosper post July 2009. A rather important distinction given that Prosper investors constitute but a minority of p2p investors in the US…………..& furthermore, post July 2009 Prosper investors constitute a minority of the total amount of Prosper investors historically. So although I have great respect for Ken, I’m not breaking ground on that statue quite yet 🙂

Dan B
Dan B
Feb. 28, 2012 7:28 pm

Bryce…………..Did I read that wrong or are you saying that you’ve put in 60 hours in just the 2 months you’ve been investing in p2p?

Bryce M
Bryce M
Feb. 28, 2012 7:36 pm

I pulled my first extract 10 January. 60 hours is more or less right, not counting the time I manage my account.

Dan B
Dan B
Feb. 28, 2012 8:10 pm

Bryce………..wow, that’s a lot of time/work.

Roy S……………I didn’t spend more than an hour or so researching p2p before opening a a $500 Lending Club account in November 2009. Probably spent another 20 or so hours within the first year setting up filters etc. My results for the first 11 months were excellent. Months 12-14 brought me back down to just above average.

Then I had an epiphany on month 15 & have since recovered all the lost ground & then some. Today, after 2 yrs & 3 months I stand at an ROI of 13.5% with 780 LC notes made up of 70% 3 yr. notes. I run 2 other OPM accounts at LC that are a bit under a year old with ROIs over 16%. How much time have I spent researching in order to increase my performance so dramatically?? None, zero, zilch. Moral of the story? Figure out what really matters & don’t waste time on things that don’t.

Fred
Fred
Feb. 28, 2012 9:32 pm

Peter, et al.,

Now that we identified some key factors on ROI, perhaps the one step to do is to formulate all of this into a single equation (i.e., through multiple regression line).

Any takers? Anyone to share their experience building their quantitative P2P ROI model?

Bryce M
Bryce M
Feb. 28, 2012 10:54 pm

@Fred – Yes, my 60 hours culminated in 2 multiple regression models that I use to invest. It is an odd position in that I naturally (I’m in higher ed.) want to share, but am simultaneously concerned about sharing so much that I crowd myself out of what I believe to be the best loans. In 18 months I’ll have a much better idea if I’m full of crap or have succeeded in beating the averages.

Dan B
Dan B
Feb. 28, 2012 11:18 pm

Peter……….I agree with everything you’ve said about Ken. However I understood that this post of yours encompassed all p2p investors………….not just Prosper investors from July 2009 to now. Or am I wrong about that?

If you read Roy’s post you will see that he suggested that Ken was outperforming “99.7% of EVERYONE ELSE”. Clearly that is not accurate. As you yourself just pointed out he is outperforming the other 627 investors in that small universe of 800 notes+ post July ’09 Prosper investors. An important distinction. However we are in agreement in that that the 2,0 time period is the most important category for Prosper investors today & going forward.

Roy S
Roy S
Feb. 28, 2012 11:20 pm

@Dan, You are correct that this is among Prosper lenders, however, you failed to mention how many LC lenders are making greater than 18% ROI after LC’s SEC quiet period. And how many of those have invested a 6 figure amount? I think by all standards, Ken’s strategy is serving him well.

I do find it interesting that your strategy’s ROI matches up well with the average return for an F grade Note. Perhaps I should invest in all HR Notes on Prosper since the average seasoned return for an HR note is 17.71%? Not only would I not have to spend time doing any analysis, but my returns would exceed even your ROI.

@Bryce, I understand your concern about over-sharing and crowding yourself out. I haven’t created any models. I don’t have any practical experience working with regression models, and it’s been too long since my statistics and econometrics courses that I would have a lot to relearn. It sounds like you’ve spent a lot more time on this than most, I hope your work pays off, and I look forward to hearing about your results in the future.

Dan B
Dan B
Feb. 28, 2012 11:45 pm

Roy S………My ROI is over an extended period of time. I don’t care what you invest in. If you can outperform me over a period of 2 or more years I’ll eat a bug or any insect or rodent you’d care to choose. Peter frowns upon betting challenges here on his blog, but I’d be open to any you’d like to propose. But before you risk your, I assume, hard earned money, you should probably talk to Peter. He’s somewhat familiar with my LC strategy & will be able to give you a better idea of what you’ll be up against. But please feel free to ignore his advice. I beg you to challenge me.

Bryce M
Bryce M
Feb. 28, 2012 11:52 pm

Love to hear alternatives people use. I only use the tools I know about.

What’s the gist, Dan? I have a hard time imagining some super simple technique. Nothing is usually so simple / explains everything.

Dan B
Dan B
Feb. 29, 2012 12:09 am

Sorry Bryce but unlike most of the people here I was distracted for most of the past 25 years of my life. So even though I could tell true stories that would be as exciting as most people’s fantasies, those years did nothing to enhance my finances . Therefore I can’t afford to “give” important discoveries away. But if you’d like & if you’re prepared to spend 4 figures you can get my email from Peter & we can discuss privately.

Dan B
Dan B
Feb. 29, 2012 5:29 am

Like I said, I have nothing but respect for Ken. However what has really impressed me are his overall numbers since day one. I think that’s a far greater accomplishment.

Charlie H
Charlie H
Feb. 29, 2012 9:10 am

The fact that Ken performance has not reverted to the mean is the real accomplishment.

KenL
Feb. 29, 2012 9:42 am

Take it easy guys, lol. Too many compliments may go to my head.
I’m certainly am happy with my returns and I often fantasize what my returns would have been if I had started lendstats earlier.

, one problem with comparing LC portfolio performances (if someone were to offer this service) is that there is no history of sold notes. So if someone sold off all of their 30+ day late loans at 90% discount their performance would look a lot better than it actually is. One advantage of not having any trading info, as with Prosper lenders at lendstats, is you see the performance of all the loans that any given lender made on the Prosper platform regardless of if the loans were sold later or not. In other words, a lender can’t clean his portfolio up to show it off to the public. With Prosper the public sees all loans, sold or not.

Bilgefisher
Feb. 29, 2012 11:23 am

One of the main reasons I invest at prosper is the amount of information available. Like the rest, I did quite a bit of research on loan types to pick my criteria. I had false assumptions on my first few investments. Via lendstats, I learned I was rather lucky with my initial picks. I would rather be purposefully lucky through solid picks than randomly lucky through assumptions.

Jason

Roy S
Roy S
Feb. 29, 2012 5:56 pm

I think the big takeaway here is that with all the information Prosper and LC provide, knowing how to parse the numbers can lead to impressive returns…unless you’re a genius like Dan who has his own simple, super-secret strategy (say that 10 times fast!) that requires no research whatsoever 😉

@Dan, I see some impressive returns, however, my personality doesn’t permit me to take on too much risk. I have ventured into the higher risk Notes, but I will always carry some lower risk AA and A Notes. I am hoping the strategies I pursue with the higher risk Notes will allow me to achieve some nice returns, but no matter what the data say I am somewhat risk averse and so will always adjust my strategy to fit my investing personality. So I don’t believe I will outperform you simply based on my risk tolerances. I will acquiesce to your superior investment strategy.

Dan B
Dan B
Feb. 29, 2012 6:50 pm

Roy S…….It may surprise you to hear that almost 50% of my LC notes are in the B & C grades. But you are a wise man to decline. Trust me, it wouldn’t have been pleasant for you. You would have been bringing a knife to a gun fight…………….& not even a sharp knife at that. 🙂

Ken L………..You never know Ken, maybe some people are just trying to fatten you up for the slaughter. 🙂 Incidentally, although LC NAR numbers can be off due to the scenario you suggest, ROI numbers don’t face the same problem. Or at least my results don’t. It may also surprise you to know that even after selling over 300 LC notes at a small premium & another 150 or so at 1-3% discount in the past 2 years, my NAR & ROI numbers at LC have never differed by more than 1.75%! My ROI numbers have in fact been higher than my NAR.

jim
jim
Mar. 9, 2012 3:49 pm

IMHO

Coming from 20 years in collections. I am only investing in smaller loans.

I’ve seen too many loans of all types on the back never get paid due to size, or larger loans leading to higher debtor justified BK filings. The common sense for me works. I regret every large loan I help fund that went bad, I cant say the same for the small ones. If a bad loan of smaller size is keeping a credit score down too low you can be assured the points to raise the score will be addressed in the most cost effective matter, hence paying smaller debts off firsts. If the economy goes south in anyway in the next 3 years, the larger loans are even more risky at any grade. Its only the bad loans mostly driving overall returns, not the good ones, no matter what size or rate.

Dan B
Dan B
Mar. 10, 2012 3:39 am

Jim…………Always interested hearing from a professional in the collections business. Am I correct in understanding that you believe the “average” consumer to be savvy or strategic enough to make decisions of paying off smaller loans first in order to raise their credit scores? I’m sure some are, but I’m talking the “average”. Have you seen evidence through your work that this behavior is prevalent?

jim marshall
jim marshall
Mar. 12, 2012 9:22 pm

Dan,
I see it every day. All consumer credit counselor give this advice, so do we when collecting debts, as well as mortgage brokers. Don’t underestimate the sub-prime borrowers. Many of them seek guidance after they get in trouble, or are trying to move up/ repair credit scores at some point because they get more educated.

Peter, does this sound familiar?
“Past performance does not guarantee future results.”

I will sleep well at night with my micro loans on auto-pilot and maintaining 15% returns post Prosper 1.0 debacle.