In Defense of Living Off Dividends | P2P Lending, Peer to Peer Lending, People to People Lending

#1Mikeon 03.27.08 at 9:33 am

So your defense was “yeah, it’s wrong right now, but some day it’ll be right?”

#2RateLadderon 03.27.08 at 9:54 am

Actaully I think it was closer too.

The “accurate” answer has to be a backward looking number. I am defending him for using the only “accurate” number.

“estimating loss” is far far more complicated and it is a conversation worth having.

Calling him a liar (you know where this happened) is absurd.

#3Peer Lendon 03.27.08 at 10:34 am

Hi Kevin,

Taken in -conjunction-, I think both of our responses balance out to an approach at a healthy perspective on this matter, and I agree, in principle, with your statement that “All estimated loss calculations have issues. They all also have merit when taken in context”.

However, I think you’re forgetting that, rigorously, the method you’re defending is itself an *estimation* – one which just happens to be the most rosy and optimistic estimation possible – and, so, many people, apparently, wanted to provide the context which, you agree, allows for such *estimations* to have *merit*.

You can choose to focus on my “fantasyland” remark, and characterize my post as an “attack”, but, I do wish you wouldn’t – and, with all due respect to the author of the original piece, I (and apparently others who knew better) tend to speak honestly when someone who’s dispensing authoritative-sounding financial advice presents something that’s not only misleading, but betrays a gross lack of understanding of the topic upon which they’re offering counsel.

Here’s a brief recapitulation, from my perspective. If you disagree with my take, I’m sure you’ll let me know – and I welcome that as healthy and productive for everyone, too:

On one end of the spectrum, I see you saying “It’s not over til its over” (“You can’t know for sure what your return will be until all loans have run their 36(+) month course, so let’s assume (read: *estimate*) that they all pay and perform perfectly over time – and print that number.”) (My read: Fantasyland)

To me, this language succinctly sums up the calculation (and logic) that the author of that piece was attempting to employ, and which you’re attempting to justify by waving hands to call into question the very act of *estimation* itself – not realizing that you’re trying to defend an estimation, by damning them all.

On my end of the spectrum, I’m saying “It’s not over til it’s over, but if the horse is limping out of the gate, it’s probably safe to assume it won’t place” (“You can’t know what your return will be until all loans have run their course, but you can know what past returns have been under similar conditions, and estimate accordingly, with a not-marginal degree of precision/accuracy.”)

In the context of credit and lending, we both know my perspective is the more rational – for the same reasons that:

a) credit scores exist in the first place,
b) Prosper goes further than a traditional credit score and uses a predictive scoring model like Experian ‘s ScoreX Plus rather than a static snapshot FICO product,
c) Prosper provides *estimated* return guidance on a per-loan and per-portfolio-plan basis, based on *estimated* default probabilities,
d) Prosper incorporates both artificial and market-based “roll-rates” (numeric quantifications of the *probability* that a loan in a certain state, say, 30 late, will either flip current or go 60 days late = ie, “estimates”) into its marketplace statistics and the aforementioned per-loan and per-portfolio guidance, etc.

There are a hundred other reasons as well, but those are the ones that pop into my head first, simply because I know you’ve been instrumental in getting them into place, and therefore know, as I do (and as I have stated publicly, see: Prosper Does Something *Wonderful* For Its Lenders) how valuable they can be, are, and eventually will be (on an ever-increasing basis) to lenders and borrowers alike.

There are a number of other factors I haven’t addressed, and it’s tiresome to do so, and I’d like to go to lunch now (and I know you know about them already – which makes it all the more exhausting!), but if you want to talk about amortization & declining balances (the “annualization” factor, which has the potential to add some heartily negative non-zero integers to the final product), follow-on assumptions re: whether one reinvests, or about levels and timing of reinvestment (which carry further, likely less weighty/fractional, but also quantifiable, risks – which I’ve written about before here, then… I guess I’ll see you after coffee.

Honestly, I was satisfied that lenders who were reading would be introduced to EricsCC & LendingStats, and might learn more about the “context” of the various ways to estimate one’s return – and I would have let the matter stand.

However, for you to make a statement like “Why on earth should he guess at the value of not yet realized losses? When the losses are realized they will be accounted for in his account balance.” Well.. I’m sorry, but that’s fantasyland.

-todd

#4RateLadderon 03.27.08 at 10:51 am

@Todd

Yes I felt the “fantasy land” was unwarranted.

I let this blog post through (and the subsequent comments) so that these issues could be brought up… And I think the issues needed to be aired and discussed.

I think many people blindly follow Eric and LS. Neither of those calculations are public. That blind faith needs to be addressed every now and again. Thanks for your help.

No offense to you, but I had to delete a couple of comments where accusations like “liar” where made. Clearly he is not a liar nor does he live in fantasy land..

My defense is by pointing out the complexity of the problem and that his approach is one perfectly reasonable approach to the problem. An approach that will yield the right result in the end. In fact an approach that will yield an ever closer to the final correct answer over time.

Your blog commentary is always welcome please try to keep the discussion personal attack free.

#5NewHorizon on 03.27.08 at 11:33 am

“Why on earth should he guess at the value of not yet realized losses? ”

I think when one sees the comparison he makes to 3-yr CDs, most will figure he’s ALREADY guessed his not yet realized losses to be $0. I think it’s reasonable to question where that $0 figure came from. Or are you saying that since it’s only a guess, $0 is no more worse a guess than any other figure – say $1? 500 Quatloos says the $1 figure will be closer to the realized losses than $0.

#6RateLadderon 03.27.08 at 11:43 am

My point is that any prediction is a prediction.

Any backward looking number is the right answer at that point in time.

The further into the future you go more the ongoing default process will be accounted for in the account balance provided by prosper and the cash flows into and out of Prosper.

To take it to an extreme. When someone stops lending and their last loan reaches a final state. The right answer to what rate they received will be an annualized calculation similar to the one used by WAB or will be a function of IRR using the cashflows (depending on your druthers). Not Eric’s CC prediction.

Banks use an ROA not an ROI calculation for a reason.

#7Mikeon 03.27.08 at 2:28 pm

@RL:

LendingStats calculation method is public, and anyone can verify how it was calculated for an individual.

Estimated ROI = Avg. Interest Rate – Avg. Fee – (1 – (1 – Late Percent – Default Percent)^(365/Portfolio Age))

#8Peer Lendon 03.27.08 at 6:03 pm

RL: My point is that any prediction is a prediction. Any backward looking number is the right answer at that point in time.—

Your point is so ridiculous as to not even be wrong – or, for that matter, a point.

Your “arguments” wind up in this ridiculous parallel universe where, logically, when you get sick, you decide to wait until you’re dead to go and see a doctor.

RL: I think many people blindly follow Eric and LS. Neither of those calculations are public. That blind faith needs to be addressed every now and again. Thanks for your help.—

I mentioned in my first comment on the Prosper blog, quoted way up at the top of this page (the one you decided to characterize as a personal attack – instead of actually reading) that any lenders who wanted to understand the methodology behind the differing ROI calculations could do so at the sites.

Mike linked to the LendingStats’ ROI methodology above, which you claim is not “public” and taken on “blind faith”. Your assertion is false. At the end of this paragraph, I’ll link to EricsCC’s ROI calculation, but, first, I just wanted to point out how utterly ridiculous it is for you to imply that LendingStats and/or EricsCC are somehow at odds against Prosper, their very raison d’etre, by providing a misleading or even an “opaque” ROI calculation. How.. absurd.

Here are links to explanations of both of their methodologies, both PUBLIC.

EricsCC ROI Methodology
LendingStats ROI Methodology

RL: No offense to you, but I had to delete a couple of comments where accusations like “liar” where made. Clearly he is not a liar nor does he live in fantasy land.—

I don’t know why you think I’d be offended – I made no such comments or accusations, and have no control over the actions of others. However, while
I would not call someone who was merely exhibiting a lack of understanding a “liar” – I might indeed say that some element of their argument was “straight out of fantasyland” – and expect them to be able to handle it, especially when I lay out for them why, and present explanation. If you’ll read what I actually wrote, rather than working off the “impression” you seem to have of what I wrote, you’ll see that I say his *ROI formula* is straight out of fantasyland – so, whether he lives there (or just picked that calculation up while he was passing through?) I have no idea, and so made no personal comments.

Maybe you can shed light – have you seen him around the neighborhood? :)

RL: Your blog commentary is always welcome please try to keep the discussion personal attack free.—

Only if you promise to scold me like a kindergartner for absolutely no reason.

At this point, I’m resigned to think that either you’re being deliberately evasive – or that you… aren’t being deliberately evasive – and… simply don’t understand.

-t

#9dvd on 03.27.08 at 6:44 pm

So, if that is so, I’ve got a deal for you:
I’ve got a great Argentinian bond, it pays 7.8% pa.
I’ll sell it to you for slightly below face value.

#10RateLadderon 03.27.08 at 7:23 pm

I can provide you sql access to the data… show me the sql statment that reproduces my ericss ROI or my LS ROI.

http://www.proprosper.com

It is one thing to say this is what it is supposed to do. It is another thing entirely to say here is exactly how it does it. I like both LS and EricSCC I value their work… I track my ROI value they produce to better undstand it. I don’t think either of them have correct answer at this point in time.

#11RateLadderon 03.27.08 at 8:14 pm

Quicken calcualets ROI in the same manner as WAB. (although it annualizes the value)…

Clearly the calcualtion is not out of fantasy land.

#12Personal Loan Portfolioon 03.27.08 at 8:21 pm

I believe that Living off Dividends is making the mistake of basing the ROI of P2P loans on the same formula he would use to calculate ROI of stocks. Stocks are priced in an open market every day and can be sold realizing their value. P2P loans do not work that way. When I look at my stock portfolio today and say that I have make 10% over the last year (That is fantasyland!) my stocks were priced today by thousands of players who agree with the current value of my stock portfolio and theoretically I could sell my stocks to lock in that 10% gain.

Even Prosper.com provides an estimated (i.e. predicted) return rate based on a basket of similar loans which discounts your estimated ROI when you bid on a loan. Since there is no secondary market for Prosper.com (or lending club for that matter) loans, the stated value of the remaining loan does not reflect the actual value of the loan. Of course, any prediction is a prediction, but basing a prediction only on the present moment looking backwards will cause you to walk into a wall if you are lucky — off a cliff if you are not.

If someone bids on 10 HR loans at an average interest rate of 25% today and all 10 loans make their first payment, will you let the lender post an average return of 25% on the website with an average loan age of 40 days? Not even Prosper.com’s bidding guidance would estimate that return and I do not believe that you should be publishing such claims on the official blog.

Doctor: Sir, I must inform you that you have a most horrible form of cancer.
Patient: What are my chances of surviving, Doc?
Doctor: You have a 100% chance of living…
Patient: That is fantastic!
Doctor: Wait, you did not let me finish. You have a 100% chance of living through the present moment. Next month, you’ll be dead.
Patient: Oh… I liked the 100% answer better.

Insurance, stocks, bonds, and other financial products are all valued based upon predicted values, so predicted value is an acceptable method of valuation in real world finance. Returns to date, especially when there is no way to liquidate a position, are not. Arguing with the methodology of LS or ECC is acceptable. However, denying that predicting future value based upon the past performance of similar financial products is an acceptable valuation method is not. Even Prosper.com’s bidding guidance disagrees with you on this point.

Regardless of what formula will arrive at the most correct valuation, that the current formula being used by Living off Dividends is most likely incorrect. If anyone disagrees with that, I propose you putting your money where your mouth is and buy the remaining portfolio off of him at the advertised ROI for the remaining loan balances. Any takers? If not, what exactly would you pay?

#13Chrisfs on 03.28.08 at 9:14 am

The fact there are are multiple ways of computing ROI speaks to the confusion about what is proper.
I have to say I have long been in RL’s camp regarding the attempted prediction future losses. You may be able to predict the loss rate of a large group of loans, such as all C loans, but to take that prediction and try to apply it to a much smaller subset of those loans (say, 5 C loans in a portfolio,) and say that those 5 will act exactly as the larger group is a big misapplication of statistics. They don’t contain the aspect of the larger group. To expect them to behave exactly like the larger average is to expect every family to have exactly 2.6 kids and every husbands to be exactly 6 ft tall. And yet that’s what the predictions in Ericscc and Lendingstats do. To an extent, that some measuring stick is better than none, it has some usefulness, but when people try to mix apples and oranges take it and say to someone “you’re lying because lendingstats and Ericscc say your return is much smaller” that’s misleading as well.

The Doctor analogy unintentionally points this out.
Doctor: You have cancer
Patient: What are my chances?
Doctor: Going by studies of outcomes of large groups of people, the average survival is that 40% die within 1 year,
Patient: So within a year, I’ll be 40% dead….. what the heck does that mean ?

Doctor:, Oops, it means there is a very wide range, some people die in 1 month, some beat it completely and live for 50 years. I can give you averages, but I can’t tell how long you individually will live.

You can say that 25% of HR loans eventually default (made up number), and I can say, “well my 5 HR loans have been doing great for over a year now, I’m making 27% off of them because they have never been late”, you can’t turn around and say “well you’re a liar because at least one of those is supposed to default” . My return is what it is. You can say “don’t expect the return to be constant” and that’s fine, but don’t insist that you know more than you know.

#14Personal Loan Portfolioon 03.28.08 at 3:40 pm

The Doctor analogy unintentionally points this out.
Doctor: You have cancer.
Patient: What are my chances?
Doctor: Going by studies of outcomes of large groups of people, the average survival is that 40% die within 1 year,
Patient: So within a year, I’ll be 40% dead….. what the heck does that mean ?

Doctor:, Oops, it means there is a very wide range, some people die in 1 month, some beat it completely and live for 50 years. I can give you averages, but I can’t tell how long you individually will live.

Mine was intended to be an exaggeration, but yours is funnier — and more accurate. :) And I must agree with you that this points out how difficult it is to track your P2P lending return accurately. RateLadder’s ROI graph with about five lines for ROI also shows this difficulty.

I wish Prosper would apply their lending guidance to your portfolio to give a projected return, but as you mention it is applying the population stats to a very small sample. It won’t work for someone with a small portfolio like me or WealthBuildingLessons. Although, it would work with better accuracy for the larger lenders.

By the way, I don’t think that he is lying. I just think that he may be disappointed because he will not have that same return at the end of the three year period. Additionally, he may lead other people to lend without considering the risks due to the rosy numbers. But who cares about his return anyway — we are talking about a few grand that is not going to make or break anyone’s retirement.

#15Peer Lendon 04.03.08 at 2:37 pm

A doctor who tells me I’m going to survive against 99.7% odds is a bad doctor or a miraculous healer – but, either way, wants to make sure I pay up on my way out.

-t

Leave a Reply

Your email address will not be published. Required fields are marked *