so it appears just the opposite of the theory

the days around the very end of the month, AND beginning of the next month are worst?

The end of the month days I understand since they are cash strapped and waiting for their next fix, the paycheck... but the early (1,2,3) days of the next month? Maybe all their obligations are > than their paycheck and its all 'gone' right away

Wait, isn't that how they got here to Prosper in the first place? ;)

Ya, it's just kinda random. Weird that the 3rd is over twice as bad as the 16th. Makes no sense.

*could be* just random statistics, who knows

but for some reason the middle of the month seems to do better 'on average' than end/beginning of month. Not sure if its statistically significant though *staring at headlights* :o

Another way of looking at it. Don't know what conclusions you can draw from it though...
(IMG:http://i181.photobucket.com/albums/x202/cubbiesnextyr/chart2.jpg)

Well I suppose if you get rid of the most egregious highs and lows under the assumption that there's a lot of randomness here which is going to make some nasty anomolies (pninen is probably rolling his eyes at this), then there seems to be a "U" shape with the lowest point at mid-month.

It occurs to me, tho', that many borrowers are paid once every two weeks. So this would make is tougher to discern any day-of-month patterns.

But WTFDIK.

I think it's pretty meaningless.

So it looks like to me that you can get a slightly better chance of the generic borrower being on time if the loan funds on the 15th to 20th. All are <10% late those days.
Nice info!

Isn't the funding date just a function of when Pre-Funding verification was completed?

Exactly, you have very little control over it. Even if you bid only on listings ending around the 10th of the month, you have no idea how long it'll take to actually become a loan.

Fun to look at, but this graph is meaningless.

That's my impression so it tends to be random. However if you started looking at listings and bidding towards the 1st week of the month you could try and "aim" for that window. I have 24 loans right now and while they are all current right now (relatively new ones), I'm still trying to figure out ways to reduce future risk. Out of the 24, 2 are from days 1-9, 2 from 10-14, 10 from 15-20, 1 from 21-25 and 9 from days 26-31.
I don't know what it means yet, but heck, I like numbers! It'll be interesting for me to see if my picked loans will match the overall summaries.

To fit that window you could really bid on listings that ended on the 3rd day of the month through the approx 10th day of the month, since average verification is 1 - 7 days.

All that for 2% downside risk protection? You could find a great loan on the 27th, and not bid on it due to this graph?

Nah, its random stats more than anything methinks

Perhaps at the edge it means something for those borrowers at the edge of the cliff, but hopefully you won't be funding them

Bottom line the way I read it - every day is a dangerous day to lend money :) :rolleyes: :D

I would be that day of month payment would only make a difference in the higher risk credit grades. Specifically HRs. It might have some influence over Es & Ds too.

I had a friends a long time ago that used to run a pizza shop. First of the month they would get slammed with business since the poor got their government check. After that, the money pretty much ran out.

I think many more people are paid twice a month than either once a month or every two weeks. I know that every job I've ever had has paid twice a month. Typically, paydays are at or close to the beginning of the month and the middle of the month. However, rent and mortgages are typically due on the first of the month (although mortgages typically have a 15 day grace period), and those are much higher priority debts for a strapped borrower to pay than Prosper. Thus, the beginning of the month paycheck gets used for that, leaving the middle of the month paycheck with far less demands on it. Thus, it makes a certain amount of intuitive sense to me that Prosper loans due on the 15th and shortly thereafter would be more likely to be timely paid. That said, the data may also just be random noise.

I once was curious about this.

I did a poll of borrowers and, so far, these are the results:

Borrowers are paid:

I thought biweekly pay is pretty much the standard these days. I've never even heard of twice-a-month pay, but apparently is fairly common according to the above post.

Most salaried people I know get paid semi-monthly. Most hourly people I know get paid every other week (bi-weekly). So two months a year they get an extra paycheck.

How strange -- I don't think I've ever heard of anyone I know getting paid biweekly, and as I mentioned above, every job I've ever had (both hourly and salaried) has been semi-monthly. I wonder if each system is more prevalent in certain areas of the country.

From an employer's standpoint I would think that biweekly pay is a pain in the rear, since it complicates budgeting and financial analysis (a different two months of each year have three paydays instead of two). It also requires the time and expense of preparing 26 payrolls per year instead of 24.

I've been paid bi-weekly - it was always a stupid thrill during those 2 months a year we got paid 3 times. :lol:

Biweekly here. Love those 3 pay months. :D

Great thread. Valuable information and very telling

I've worked one hourly/commission job and two salaried jobs, and all were paid biweekly, hence my thinking that was "standard." In any event, I never considered a certain month to be a "3 paycheck month" since the paychecks are still always evenly distributed every two weeks. What difference does it make if three paychecks happen to fall within arbitrarily defined month boundaries?

Of course its probably that kind of thinking that gets some of our friendly borrowers in trouble. "I have an extra paycheck this month! Woohoo, more beer, cigarettes and lottery tickets for me!" :lol:

The mentality is easy to understand, even if flawed;

Let's say you make 12,000 a year, so 1K per month.
But you're paid every two weeks, or 26 cycles, so 461.54 per payday.
Your bills are calendarized (you pay them each month)
Your mortgage is 400.
In most (10)months, you get two paychecks totaling 923.08, of which 400 needs to go toward your mortgage. In the other two months, you get three paychecks totaling 1384.62. You just won the lottery!

It the same as people who get excited about a tax refund. You just let the government use your money interest free for a year! Woohoo! You didn't win a prize...you just got back your own money that was held hostage!

What happened to the charts from Pninen?

you're gonna have to wait a few days to get the next installment...

At this rate, the loans will 'season' right at about month 42 out of 36 months... :P

The 07/01/2007 update.
(IMG:http://img.villagephotos.com/p/2006-6/1187065/prosperlate2007-07-01.gif)
For details on methodology see my posts early in this thread.

Thanks for the updated numbers pninen. Your effort is truly appreciated.

It looks like April 07 has the best start of any month. Not that it says much.

Kind of like hearing the traffic fatalities for a holiday weekend were only 15 when they were expected to be 23...15 people still lost their lives.

well you haven't had to add that 24% row yet lol

Thanks for the update. Nice to see that nothing has changed.

Maybe if something more happened with collections???

Seriously, what happened with May 06? Is it because the rate was capped at 24%? (wasn't it?)

After looking at some of the data I've accumulated on this, it is my belief that there are a couple of reasons why May 06 is better than other months before and after it.

The biggest reason appears to be that E and HR borrowers with funded loans was a lot lower percentagewise relative to the total number of loans funded that month. I think the lower maximum interest rate contributed to this but only in part since June 06 had a similar low limit as well.

Incidentally, the E and HR borrowers who did fund that month are going late at a much lower rate as well relative to nearby months. You can see those graphs at http://prospers.org/wiki/Data/LatesByMonth. However, C and D borrowers are going late similar to other months. Go figure.

For some reason, C and D borrowers had the largest percentages of loans that month relative to the total and higher than they had before or since as well. I'm wondering if there was a "flight to quality" after the initial euphoria from the earliest adopters as well.

I would post a graph showing this data but I don't have the time right now.

Yep. April'07 is the first month where all loans came from listings created after Prosper's Feb'07 changes. The big change was raising the minimum credit score required, thus cutting off the bottom of the HR range. This should have the effect of reducing the default rate for the whole-Prosper portfolio, and it looks like is indeed having that effect.

Its a race between Oct '06 and Mar '06 to see who will get there first. Oct '06 is coming up fast.

Thank you, pninen, for keeping these charts updated!

Do my eyes deceive me, or are we starting to see a knee in the 1+mo lates for March-April, 2006 loans? So does that mean the 6-month season effect occurs at 14 months? Maybe it is time to be just slightly less pessimistic. ;)

Well I don't know about a "knee", but those first few curves are indeed bending more than one would expect from a constant-default-rate model. (I've tried curve fits using curves from a constant-default-rate model, and they just don't fit very well. In fact I posted some pictures of that in this thread a few months back.)

To me that means that these loans are going bad at a rate that is declining during the first year. That is a cause for some optimism about existing lender portfolios. It implies that the straightforward extrapolation of defaults may be a bit too pessimistic.

I'm not sure about why the lines are bending but one thing to keep in mind is that as time passes, an increasing number of loans are being fully paid as well. After subtracting loans that have already defaulted and loans that have already been paid, you have a much smaller number of loans left that can go late or default in the future.

This has been my theory for a few months

However I was hoping the slower ascension would (1) start earlier and (2) from a lower data point

that said, if you strip out HRs I am sure both would hold true

That is depressing to consider. It would mean that the default rates for a portfolio in which repayments are continually reinvested is significantly higher than the fine numbers shown in pninen's charts. On a positive note, however, March 07 is only the second month not to break 4% at the end of it's third month.

Anyone have the stats on how many loans are paid off or defaulted after a year? We have a couple months of loans over 1 year old now to pull data from.

You can get that from the performance data. Here is the performance of all the loans before July 1, 2006:

https://www.prosper.com/lend/performance.as...0&sn=&tg=0&vb=0

Paid is 17.54%

Defaulted 10.52%

Late 1 month + 8.24%

I keep telling 'em prepayments are bad, bad, bad for lenders. (Except those lenders that are planning to take their money and not reinvest).

Thanks pjz! I shoulda thought of that myself.

So after 1 year, about 1/3 of loans are paid off or late/defaulted. That leaves 2/3 left to become late, so if the rate of becoming late is fairly constant you'd expect the percentage becoming late every month to be 2/3 what it was a year prior.

The months on the graph nearing or passing the 1 year mark seem to have slowed down more than just to 2/3 speed. That suggests that we saw a higher rate of default in the first year than we'll see in the 2nd year of the loans' lives.

ETA: This is probably due to several things including that many of the HRs and Es have already defaulted/lated. So a greater portion of the loans left are high grade borrowers.

I've been a prosper lender for over 12 months now. I started in May 2006. I target borrowers who have no record of lates, zero defaults, 7 years minimum credit history (whenever possible). Most of my loans have 20% DTI, but I fund loans with higher DTI about 30 to 40% of the time. Here's what the loan distribution looks like in my portfolio:

AA - 38%
A - 25%
B -21%
C - 10%
D - 4%
E - 1%
HR - 1%
NC - 0%

I'm not a sophisticated investor, but for what it's worth, after one year :huh: I have more lates then ever before. I've been lucky in that I haven't had a default yet, but it looks like they're coming. Here's the score right now...

214: Current

1: Payoff in progress

2: Late (<15d)

2: Late

2: 1 month late

1: 2 months late

1: 3 months late

1: 4+ months late

That is not good for the type of portfolio you describe. I could not check the type of loans you funded since I do not know your Prosper bidding name. Were they a lot of business loans?

It makes life so much easier if everyone would use the same login name on both the forums and Prosper.

why is it "not good"

what is "good"

Granted 84% of loans are AA to B

But out of 214 loans (I will exclude the <15 late because those seem to pop in and then back out a lot) the lender has

after 12 months, 7 loans of some sort of lateness

7 / 214 = 3.3% late/default rate after 1 year

Is that good? Is it bad? if its bad why is it bad?

Worst case if you believe that borrowers who pay on time the first 12 months, will not continue that behavior in months 13-36, and the rate of late will not decrease its about a 10%ish (not exactly) late/default rate over a 3 year period

Worst than the Experian data leads us to believe for those grades but we all know Experian <20% is 'wrong' for Prosper lending.

So worst case 90% paid back/10% did not

And with assumption that people who pay through months 1-12 won't suddenly turn into deadbeats in months 13-36, BUT assuming some proportion of those will STILL go late/default simply due to 'life happening' to some of them (just the odds, some will lose jobs etc) I'd say probably on target for a 6-7% default rate in total over 3 years

Great? No
Good? Dunno
Bad? Really?

Seems quite expected based on how Prosper works. Not sure what ROI that would return this poster, because I don't know what rates he/she lent at. Probably with AAs to Bs ranging in 9 - 14% range, and with 5-8% defaulting by end of life its going to be a Xish return? I will let X be determined by someone who wants to do more math than me at the moment. Say 200 loans go to term, 14 fail by end of year 3. Average interest rate is 11%. Do the math, what does that generate. Probably better than a CD. I'd argue by buying 200+ loans you are simply buying the 'market' (i.e .like an index fund)

I had a guess that the success of the may 06 loans were based on CP's made by GLs, but maybe it is a reaction of the early lenders after 2-3 months when some loans came up late.

thisguy:

"7 / 214 = 3.3% late/default rate after 1 year....

target for a 6-7% default rate in total over 3 years"

The problem is we do not know the average loan age. A fair guess would be 6 months. That gives a late/default rate of 6.6% after a year and 20% for 3 years. I would call that "bad" for a portfolio that is a mostly low DTI with 84% AA-B and with 38% AA.

As you noted another big unknown is what interest rate is this person getting. For instance the AAs could be AFs at 13% or non-AFs at 8%.

Sorry I forgot to mention a couple of things. My Prosper name is "Georgia_Dad". My average rate of return is 11.6%.

Why don't you call yourself Georgia_Dad on this forum, or change your Prosper name to uniden? Isn't Uniden the manufacturer of cordless phones?

The 7/15/2007 update...
(IMG:http://img.villagephotos.com/p/2006-6/1187065/prosperlate-2007-07-15.gif)
For an explanation of methodology, see my posts near the beginning of the thread.