No, no, no. The problem wasn't that the rest of the post didn't show up when you clicked the link.
The problem was that the rest of the post showed up before you clicked the link.
I lost a tag there for a minute, and the problem with this "read more" thing is you have to publish the post first, then you can create the link and republish it. Since the preview function in Blogger will not tell you if you have hosed this up, you can't see it until you post it. Then you gotta be faster than bacon dreamz at finding your missing close span tag. See.
Slightly (OT), the demise of this year's house selling season has been greatly exaggerated.
Cheap-home rush cuts O.C. supply under 6 months.
May 19th, 2008, 12:01 am · 14 Comments · posted by Jon Lansner/O.C. Register columnist
Market watcher Steve Thomas at Re/Max Real Estate Services in Aliso Viejo every two weeks calculates market time, a benchmark of how many months it theoretically takes to sell all the inventory in the local MLS for-sale listings at the current pace of pending deals being made.
By this Thomas logic, it would take 5.82 months for buyers to gobble up all homes listed for sale last Thursday at the current pace of deals vs. 6.08 months two weeks earlier and vs. 8.0 months a year ago. Last Thursdays market time is the lowest in 14 months. A key reason: the latest count deals in escrow of 2,658 is up 166% vs. Jan. 19s wintertime low.
Market time is also shrinking because cheaper houses are in vogue, with market time for homes under a half million down 41% vs. a year ago; and for homes between a half-mil and three-quarters of mil down 38% vs. a year ago. The overall market time is down 27% vs. a year ago.
Says Thomas: it is a matter of time before the data that the media is supplied catches up to the story of today: demand is much stronger than last year.
Me: Where are the "in-the-know", "ahead of the curve" real estate "pro" bloggers on this? The ones who are still trumpeting the "we're going down, we're going down, we're going down!" story?
Is there evidence of adverse self-selection in existing FHA loans? Either by FICO or the source of downpayment? Of course as a government program, some of that is probably intentional. Way back in the way back of time, the perception was that FHA existed to help the "deserving poor" who could't come up with 20% DP but were capable of paying off a loan acording to terms.
"Me: Where are the "in-the-know", "ahead of the curve" real estate "pro" bloggers on this? The ones who are still trumpeting the "we're going down, we're going down, we're going down!" story?"
According to Thomas report: 71.5% of the detached inventory under $500K is bank owned or short sale.
Yeah, that sounds like a healthy market, Seb. I can't imagine anyone taking their home of the market so they don't have to compete with the flood of REOs.
I appreciate all the bottom calls you have been making but I personally won't believe in you bottom calls until you yourself put your own skin in the game. What have you done to take advantage of this once in a lifetime opportunity?
Is there evidence of adverse self-selection in existing FHA loans?
That's an interesting question. In recent years, it seems that the most adverse selection went on in conventional subprime, not FHA.
The connection between FHA and borrower income is actually much more complex than most people think. FHA has never had income limits (except in a couple of targeted programs). What kept it a "low to mod income" program was simply the correlation between borrower income and house price: FHA has always had lower loan limits (until this year).
There's never been any law that says a high-income person who wants to buy a $600K home with a $300K FHA loan can't do that (in a county where the FHA limit goes up to $300K). It's just that they aren't very likely to do it; with a 50% LTV and any kind of decent credit history, you could get a cheaper conventional loan (because there would be no insurance premium of any kind). And, as most high-income people buy more expensive homes and need the higher loan amounts, you just don't get huge numbers of high-income borrowers in FHA.
Unless they have terrible credit or spend so much of their income that they can't save up more than 3% down.
In that sense, it is probable, I think, that higher-income borrowers in FHA are more negatively-selected than low or moderate income borrowers in FHA. So in that sense it's not surprising that the higher-income FHA borrowers would have worse credit profiles.
But all that's ancient history. Now that the FHA limits have been bumped up to jumbo in some markets overnight, you'll have more higher-income borrowers in FHA (because they don't have a downpayment or because they're an underwater refi). This suggests to me that FHA's historical data is now going to be of limited usefulness in predicting performance of its future business. But will that stop them?
Sebastian, That's quite some analysis on that blog. Do you believe it is on par with the quality of CR's work?
Seems to me there is a lot missing. Prices and transaction volumes not just months supply, might be nice.
That aside, I have been suggesting for the past month that we have probably seen the low in transaction volumes. If the FHA bill gets the kabash (it richly deserves) we should see activity really pick up as lenders throw occupants out and aggressively clear their inventories...
Tim said: "I appreciate all the bottom calls you have been making but I personally won't believe in you bottom calls until you yourself put your own skin in the game. What have you done to take advantage of this once in a lifetime opportunity?"
You mean other than being virtually the only one to put myself out there against the daily abuse and ridicule, calling for "no recession" (right so far), a major stock-market low (right so far), new highs in the major indices before the year is over (too soon to know, but increasingly likely), and now a rebound in housing?
Well, my investment in stocks (that I've posted) and my house, I guess.
Fico scores are worthless in my opinion because as a younger person who never took out any student debt I have a 700+ fico score. Even my income does not predict my future behavior when it comes to paying on time. Many people my age have great scores and good incomes but none would hesitate to walkaway from their mortgage if they found themselves underwater.
Look, how did my thread on FHA turn into All Attention Given To Sebastian again?
Would you people just quit? We're trying to have a conversation about the post topic. Go somewhere else to have a pissing match.
Sebastian, it was incredibly rude of you to highjack this thread so fast with such a pissy little comment. I don't want an apology; I want you to go away.
And I thought a FICO was a glossy green leafed bush.
I guess we must be obsessed with credit since no-one pays cash anymore but these actuarial one size fits all predictors are not tamper proof. You can still drown in a river that averages 1 foot in depth. Employ a real live real good credit analyst and you can cut your losses by 95%.
Employ a real person to think? What was I thinking...
"Tanta writes:
This suggests to me that FHA's historical data is now going to be of limited usefulness in predicting performance of its future business. But will that stop them?"
The trouble we have had recently with FICOs is mostly, in my view, that they were relied on to offset extremely high risk characteristics in loans with a lot of "risk layering."
You can get an AMEN from the Shnapster on that one. Your lead to the jump could have been:
"It gets more sagacious from here . . ."
nevermind, I don't want to feed your "massive 'bubble' ego."
nevermind, I don't want to feed your "massive 'bubble' ego."
Don't worry about that. Whenever I do manage to come up with anything like a sagacious point, I manage to distract everyone with typos and broken links.
Frankly, I don't think it's such an especially profound point. But there has been so much nonsense spread about FICOs lately that it seemed worth making (again).
Could the risk based MIP pricing be a pre-cursor to looser FHA underwriting standards?
It's becoming clear FHA is going to be financing a lot of borrowers for at least the next two years. Could looser underwriting be next, justified by the higher MI?
Honestly, I can't see a reason for them to make this change now, except for maybe to factor in FHASecure. Of course, if FHASecure is such a "success" why not keep it on permanently and use it as a 100% purchase program. Their elimination of DAP as a risk factor suggests that's the way it's headed.
Lending money is a serious matter. Too serious to be left in the hands of kids with computers.
FICO scores attempt to judge a persons ability to repay a loan. Repaying a loan also involves a willingness. Perhaps we should have a morality test to match the ability test. Willingness to repay is becoming more important in my opinion.
Is there more? I don't know if this is what BD referred to, but I see a continuation link on both the shrunk and expanded versions, but the latter doesn't change the display. Just the way it is?
"This suggests to me that FHA's historical data is now going to be of limited usefulness in predicting performance of its future business."
I was thinking something similar: FHA may have added some significant complications and uncertainties to its model. I mean any insurance business must face problems of signaling and screening so adding a pool of people who probably pay more attention to (and actively maintain) their FICO scores and also putting more money at risk could introduce a lot more policy risk than realized.
That is, people generally know more about their financial situation than the insurance company and this additional pool of jumbo-loan folks probably are more acutely aware of their financial situation than most. If those same folks know they face large risks then they are more likely to want to buy insurance and any insurance company, FHA included, must either try to minimize the problem that too many people with big risks will buy their product (the problem of adverse selection) or they must price that risk appropriately or both but ...in the case of FHA where are they going to get the data to do that?
It seems to me that this adds more uncertainty to the mix and as we've learned in the CDO/SIV debacle that is generally not a recipe for success.
FICO scores attempt to judge a persons ability to repay a loan.
Actually, you've got the lingo backwards. FICO measures "willingness to repay." DTI, employment history, cash reserves, etc. measure "ability to repay."
Downpayment is a little of both. In the sense it measures the borrower's ability to accumulate savings, it measures ability to repay. In the sense that it functions as "skin in the game," it measures willingness to repay.
I have no problems with FICOs used to approximate the results of a manual review of a credit report. That's all they are: a way of quantifying the relative acceptability of a jillion datapoints on a credit report.
The problem comes in when you give it too much weight in the lending decision. It does not "offset" ability to repay or down payment. That's the mistake we made.
My concern about what FHA is doing is that they're giving too much weight to FICO ranges in setting premiums. That Moody's chart suggests that there isn't that much of a difference in delinquency within the category subprime: once you get down to that range, further distinctions aren't necessarily helpful. So I fear that FHA feels like it is making important risk distinctions, but it isn't really.
This whole thing gets right to the heart of the great contradiction of insurance.
The theory of insurance is to distribute losses so that everybody pays some part of the total community and nobody pays all. However, insurance companies then proceed to slice and dice to apportion 'fairly'... in other words, to return to the original situation where the loss goes to the single loser as much as is possible. If you cut it into fine enough groups, you have no insurance at all - you just have prepaid losses by induviduals with insurance companies taking a percentage for the favor.
In Sebastian's defence, what he did is par for the course online. Your response has the air of someone new to the virtual world.
The only reason Sebastian can "hijack" a thread is because so few pose contrary commentary, leave alone whether it has even a modicum of ration support which Sebastian at least tries to provide.
The fact that he didn't post his comment in a relevant thread is because you hosts don't create threads devoted to free-for-alls. Therefore posters use all of them as such.
OT - Tanta, if you're interested, a related article to your Mortgage Fraud Employee Benefit Post. Seems that Scripps Research Institute is using mortgages to attract top biomedical researchers.
"Scripps issues million-dollar mortgage to lure scientist
By STACEY SINGER
Palm Beach Post Staff Writer
Thursday, May 01, 2008
The Scripps Research Institute has started issuing home loans to sought-after scientists, giving one top recruit a $1 million mortgage for an oceanfront condo that cost $850,000.
Scripps won't say why the mortgage exceeds the condo's purchase price, or whether the loan was derived from a $310 million state grant used to attract the biomedical research powerhouse here in 2003. The terms of that home loan, and three others like it, are a secret that Scripps Chief Operating Officer Douglas Bingham said he will not disclose.
Using the state grant for home loans wasn't contemplated when the Scripps deal was negotiated. But Bingham said that the state's agreement gives the institute "broad authority" to use grant money for employee compensation and benefits.
The million-dollar mortgage recently was filed in the public record. After The Palm Beach Post asked about it, Scripps notified the Scripps Florida Funding Corp., which oversees Scripps' grant.
Tough competition for scientists, coupled with recruits' difficulty in selling their existing homes, prompted Scripps to sweeten offers with housing loans. Bingham said top scientists who bring grants and intellectual property typically have multiple offers, making an attractive recruitment package a must.
Frankly, I don't think it's such an especially profound point.
Well, it certainly shouldn't be. But I'm afraid our sound-bite-oriented culture doesn't like to wrap their heads around issues such as risk-layering, and would rather focus on something like FICO as if it's the only factor to consider in lending.
Lets hear greasy hair Sheila Bair make a nice speech about this topic of interest:
BLOOMBERG
Banks Keep $35 Billion Markdown Off Income Statements
"Banks and securities firms, reeling from record losses resulting from the collapse of the mortgage securities market, are failing to acknowledge in their income statements at least $35 billion of additional writedowns included in their balance sheets, regulatory filings show. Citigroup Inc. subtracted $2 billion from equity for the declining value of home-loan bonds in its..." Banks Keep $35 Billion Markdown Off Income Statements (Update1) - Bloomberg.com
Your response has the air of someone new to the virtual world.
I'm not new, love, I'm different.
I don't care how the rest of the blogosphere does things. Around here, we like a better signal-to-noise ratio.
Sebastian has been pulling that stunt of his for over a year now. I'm tired of it. And I am not obligated to open "free for all threads" so that Seb has some place to pick a fight.
"The problem is not that a 90% LTV loan with a 720 FICO won't outperform, statistically, a 90% loan with a 620 FICO. It will, although it isn't always clear that the difference in performance is all that substantial. The problem is that a 100% loan with a 720 FICO will not necessarily outperform a 90% loan with a 620 FICO. The idea that a high(er) FICO offsets lack of downpayment or high DTI is currently dying a painful death."
I believe I've parted company with you around this place previously with regards to models/scores. FICO basically measures "Willingness to pay" (the W in the SAW) framework, based on bureau-reported tradeline performance. But nothing that I know of says that you can't take LTV, DTI or local economic or market conditions into account when developing an internal model or scoring system.
Rather the opposite; if you were to limit yourself to bureau data, you'd just be mining a claim that Fair Isaac has already worked over extensively. Hell, I was at an OCC conference a few years ago where the panel was point-blank asked if economic conditions should be taken into account and they seemed confused that anybody would have even bothered asking.
That this appears not to have happened in a lot of shops is a mystery. Possibly it was and nobody paid attention because it was more profitable to do business based on the assumption that credit and business cycles were dead. I still remain in the camp that says that these things can and should be modelled, though.
I too would like to click on the comments for a post and see an actual discussion (or not) of the topic du jour, not the usual mix of thoughts. However, I don't believe that humans can be regimented into this, especially not on pseudonymous blogs. Instead, I think what can approximate it is the creation of a convenient outlet for deviant thoughts to be self-marshalled, i.e. a way for commenters to stay on topic and easily post elsewhere when off-topic knowing that other readers will actually read and engage the alternate site.
DailyKos has a feature that allows commenters to create their own diaries, which can be elevated or reduced in visibility. Perhaps something like this would allow posters like Sebastian to include useful commentary without hijacking a thread much to Tanta's disgust (though CR has always manifested a more easy-going approach). Alternatively if the diary infrastructure is cumbersome, perhaps just a link to a forum where regulars can post their own topic and start threads that don't disrupt the main threads discussions.
Of course, this may not fit with the vision of CR, in which case, please ignore my $0.02.
Should the FHA really be making loans like this in the first place? I mean their charter is to help enable the dream of home ownership and all that jazz, correct? How does putting people into 99% LTV loans really serve the public good, or even the borrower? Enabling loans that are more likely to default really promotes home purchasing, not long term ownership.
I still can't get my head around why FHA is underwriting jumbos to people who couldn't get one in the open market. I understand helping 1st time buyers, not subsidizing riskier candidates.
It's clear that FICO is a statistically significant variable.
I don't really know what you mean by that statment.
Look, FICOs just rank-order a universe of people with credit histories. But the slope of the default probability curve here is convex, not linear. There is a signficant difference in default probability between 580 and 620. However, the difference between 780 and 820 is not visible to the naked eye.
For the moment it doesn't matter about those other factors. If you look at the HUD premium chart I linked to, it is making premium pricing distinctions within an LTV group. That is to say, at the same LTV a borrower with a 550 FICO will pay more premium than a borrower with a 580 FICO. I, however, am suggesting that once you get that low, it is hard to say that there's much of a difference. In any case I would like to see some data on that from HUD. The Moody's data suggests that the performance of loans under about 660 is all the same, regardless of how you bucket the under-660 group.
But nothing that I know of says that you can't take LTV, DTI or local economic or market conditions into account when developing an internal model or scoring system.
Nothing I know of says any different.
But we aren't talking about FHA's TOTAL scoring system for loans.
We're talking about how it prices its insurance premium. And the premium matrix that it is rolling out charges a signficantly higher premium for lower-FICO loans at the same LTV. This is not the same calculation it uses for approving loans.
"Where are the "in-the-know", "ahead of the curve" real estate "pro" bloggers on this?"
Sebastian, wouldn't you agree that the table on Lansner's blog shows that the months-of-supply on houses UNDER $750K have fallen, while the months-of-supply of houses over $750K have RISEN?
If we were seeing a sustained turn-around, I would assume that the high end would see it as well (especially since 'everyone' says that the high end won't be hurt by the downturn).
Anther thing to consider...the cutoff (the mid $700K) seems rather close to the cut-off for "jumbo conforming loans", which leads me to believe that without Fannie/Freddy/FHA backing, we wouldn't see this reduction in months-of-supply. In other words, a lot of the reduction in months of supply in those price ranges might be caused by the start of "easy credit", this time backed by the US government instead of sub-prime and alt-a outfits.
here in the DC burbs, Charles County MD in particular we are still growing every month and have 15 months inventory.
i think your story has a key word..."cheap" you see we still have prices mostly stagnant while values are going down. our small county has the highest FC rate in all of MD.
we add 300+ homes a month and "sell" 110 a month... yoy total sales in dollars is down 59%...
"A study of an entire year`s applications..." Maybe they were talking about reverse mortgages in 2005. It is a posibility. A lot of smart old people made out like bandits thanks to FHA.
without hijacking a thread much to Tanta's disgust (though CR has always manifested a more easy-going approach).
How do you know that CR doesn't just quit reading comment threads that degenerate into Sebastian-baiting?
I myself just quit reading threads that degenerate like that. And I miss the opportunity to discuss the post topic with people who care about the post topic, because of course they too have quit the thread at that point.
Noise drives out signal. We are now a few hours into this thread and people are still "responding" to Sebastian even after he went away. (Anyone want to bet how long that will last?)
Anyone who wishes to start the financial world's version of Kos is welcome to do it. I think, for instance, it would be fantastic if you and Seb got together and did that. Everyone should pursue his or her own vision.
"I don't really know what you mean by that statment."
Statistical significance is a matter of math (F Value). It's certainly reasonable for a variable to lose importance as other variables gain.
I read what you wrote. I conclude we're in agreement that other statistically significant variables should be considered. My point is that FICO is (perhaps) one variable in a curvilinear model.
Tanta,
Was I reading this right? You said "the difference between 780 and 820 is not visible" and "the performance of loans under about 660 is all the same." Between these ranges there is presumably increasing risk of loan failure. Does FHA's rule of a minimum FICO of 580 make no sense then?
Regarding the seller contributed down payments I believe the recently settled California lawsuit mandates that FHA continue to accept these. FHA's response has been to come up with some risk related pricing which superficially seems like a good idea. Is it not? http://209.85.173.104/search?q=cache:XdVlAGLwl50J:seattlepi.nwsource.com/local/339241_downpayment12.html+nehemiah+fha+lawsuit&hl=en&ct=clnk&cd=3&gl=us
FHA is not supposed to make money on these insurance premiums but it is also not supposed to lose money. From the article you quoted I gathered that all risk variables would go into FHA's pricing. Isn't FHA required by law to publish these 60 days before implementation?
Using projected economic conditions to predict loan performance would be extremely difficult and outside of the capability of the mortgage finance system. That's the skill billionaire hedge fund operators claim.
John Stumpf of Wells Fargo famously said that he worked in an industry that had busily been inventing new ways of losing money, even though it already had plenty of good old-fashioned ways of doing so. Not so famously, and at the risk of getting folksy, when I was a kid, my Dad always used to harp on using the right tool for the job (he got mad when I used a screwdriver as a chisel, or maybe the other way around). Question to Tanta--are you basically saying that the FICO score is a perfectly good 'tool' that has recently been misused for one purpose (to Dad's irritation), and now is being misused for a new purpose (to Stumpf's observation)? In principle, it would seem to me risk-adjusted and variable premia for insurance is a fair and laudible goal. How to get there is another and more difficult matter--viz., the sex bias in life insurance and annuity tables, which are at least consistent in their discrimination (so that men pay more for life insurance and women pay more for annuities, because insurance companies pay out faster on the life policies to the men and pay out longer on the annuities to the women, given the sex-based difference in life expectancies).
Second, trying to call a national bottom to the residential real estate market is impossible. I have a slightly different question: whether the areas that have fall hard and fast (i.e., Miami, San Diego), will continue to fall harder and faster, or whether they were merely first and the areas that have only slowly started to fall and fallen only moderately (i.e., Seattle and Charlotte) will catch up with them and then stay in the slump longer. My sense, though I can't support it, is that the process will be gentler in those markets where it was slow to start and has been milder, and that it will remain harsher in those markets that have already suffered the most. And, further, that no one can predict at this point which market will start a recovery first, but that any talk of recovery is premature--once bottom is hit, that's where price levels will remain for the foreseeable future, since asset bubbles don't reflate--at least not in the time horizons on which financial markets operate.
Of course the lower income fico scores are better. If you take out a heloc and use it to significantly pay down your cc bills your score will look great (at least temporarily), even though your credit situation has grown far more precarious.
Those lower income people are the ones that have been trying to massage their scores and they know how from radio talk show financial people. Contrarily, if you are more secure financially, you don't consume credit the same way to have a score boost, nor do you obsess over what your score may be, just assuming it must be good.
I sort of resent FICOs for being undemocratic or something. But yeah, the more cogent reason to object is that they aren't the best data. AFAICT, lenders use FICOs because it is easy and it's a number. That subprime plot illustrates that at least in some circumstances it provides practically no utility.
Tanta has taught is that a down payment, saved up over time by the borrower, is a much better indicator of the quality of the loan. It isn't that hard to measure this, so why don't they?
I would disagree that FICOs primarily measure willingness to pay, by the way. I would say they measure meticulousness and organization. Those things are not unrelated, and so the FICO does have utility, but they aren't the same thing. A person who is sloppy enough to have paid some bills late might nevertheless be very committed to paying them. In particular, such a person might be quite determined to stay in her home even if she has to eat rice and beans.
Question to Tanta--are you basically saying that the FICO score is a perfectly good 'tool' that has recently been misused for one purpose (to Dad's irritation), and now is being misused for a new purpose (to Stumpf's observation)?
Let me try to state my concern with an analogy to health/life insurance:
Do insurers charge smokers higher premiums than non-smokers? Yes, they do.
Do they publish premium schedules that distinguish between two or more packs a day, one and 1/2 to two packs, one to 1 1/2 packs, 1/2 to 1 pack, and less than 1/2 pack? Not as far as I know.
Of course, they have no particularly reliable source of such fine data. Then, even if they did, they'd have to have some medical evidence that someone who smokes 20 a day is really at a substantially greater risk than someone who smokes 15 a day.
Now, take a look at the Moody's chart I posted. All of the subprime loans perform worse than the Alt-A or (prime) jumbo loans. No surprise there. But within that subprime category, they aren't showing very much difference between the three FICO buckets. There's obviously a slightly elevated DQ rate with the blue line (below 500) but what's the difference here between the 501-620 and the 621-680?
I'm not sure it's so much misusing FICOs in terms of purpose as it is making unnecessarily fine distinctions just because we have the kind of data that can be bucketed like that. I am reminded of (home) PC users agonizing over whether they should buy 10 gigs of memory or 12 gigs. They'll probably use three.
The main enhancement of the econometric model was to improve the incorporation of borrower credit history information at the loan-level. By using borrower information at the individual loan level, the model is more sensitive and can better differentiate loans that have the same product type, origination year, loan size and initial
LTV, but have different borrower scores. Modeling at the loan level also provided better control for potential bias due to choice based sampling of historical credit scores. The estimation results confirm that credit history is among the most influential factors explaining the claim probability among individual FHA-insured mortgages and is also a significant factor in explaining prepayment behavior.
I am a sloppy payer, but I always pay. Frankly, I don't need credit, so it doesn't matter to me.
And as to the ciggie analygy, I think
I read that people who only smoked half a pack a day were distinctly healthier than those that smoked a pack.
Likewise those curves are at a distinct time in our economic history, so they might look different at a different time.
Like the half pack a day smoker living in the unpolluted country vs.
a very polluted city.
The thing about not looking around to see what's going on in the economy, sometimes it might be hard and subtle, and sometimes it might smack anyone in the eye with obviousness. And sometimes we have people in denial when it's already started. A person up the thread said he asked about external conditions, and was looked at blankly. Clearly, the people at the conference hadn't even thought of it,
rather than rejecting it as too hard or too uncertain.
This is why people are supposed to make a difference; we are supposed to think about stuff. Hah.
Thanks for the link, bacon dreamz. I skimmed Appendix A rather quickly, and I still don't see why they decided on the FICO buckets they chose. I guess I'll have to spend some more time with it.
300 to 499: 1.2376 175 bps
500 to 539: 0.9128 175 bps
540 to 579: 0.6847 175 to 559, 150 over 559
580 to 619: 0.5089 150 to 599, 125 over 599
640 to 659: 0.1092 125 bps
Lawyer Liz: "I am a sloppy payer, but I always pay. Frankly, I don't need credit, so it doesn't matter to me."
I am by nature a sloppy payer who always pays. But I do need credit so I have had to adopt a rigorous routine wherein I pay bills every week and have my computer remind me to make sure payments went through on the appointed day. All to feed my FICO.
I remember when it used to be fine to pay your utilities every other month.
The cigarette analogy?
Not true for a group insurance policy; true for individuals. That's exactly the point: just like banks wanting to make loans to people who don't need money, insurance companies want to sell to people who will not have losses. But why would those people want to buy insurance?
To get into the argument about whether FICO matters or down payment matters or income matters is to buy into the premise that insurance should not be a group sharing of loss, but should be 'fairly' allocated. In other words, those who have losses should have losses, those who don't shouldn't have. Then what is the purpose of insurance?
GrandmotherOfShnapsLiz wrote:A person up the thread said he asked about external conditions, and was looked at blankly.
Actually, it was BG who wrote that he asked, at a conference, point-bank: "if economic conditions should be taken into account".
I think the problem is, 'economic conditions' are a matter of great speculation, whereas something like a FICO score is a matter of fact. That said, some have already issued special criteria applicable to loans in specific markets which are thought to be weak. How's that for taking 'economic conditions' into consideration?
Well, the correlations certainly seem to support more than one bucket but if they're going to dice things up like that wouldn't an LTV range tell them more? I'm assuming FHA would have a fair amount of data on probability of default for LTV's between, say, 80% - 90% so, given the convex shape of the FICO curve, adding a FICO > 580 would presumably tell them more than breaking down fees by FICO alone.
but if they're going to dice things up like that wouldn't an LTV range tell them more?
The premiums do vary by LTV. Within each LTV band, they then vary by FICO.
I just typed in the first (less than 90%) premiums for an example. If you look at page 26 of the HUD document you can see the whole table. It is too large to retype in the comments and I didn't take the time this morning to try clipping it out of the text to post it.
"I would disagree that FICOs primarily measure willingness to pay, by the way. I would say they measure meticulousness and organization."
Misconception. While any significant level of 30+ DPD will hit your FICO (or I should say FICOs, since there are multiple Fair Isaac models), the big damage comes in the form of defaults, which is not a matter of sloppy payments. In my experience you can maintain a high-600 level FICO with some 30+ late pays, and late pays that don't rise to that level will barely register (since they probably won't be reported to the bureaus in the first place.)
So truly low FICOs are indicative of either an unwillingness or inability to pay. The weight is usually put on the former because A) the model doesn't have access to your income and B) it is assumed that the debts previous lenders made to you had at least some underwriting and were within your ability to service. B) might be questionable to some extent at the moment, but still appears to hold true for the vast majority.
From the article: But is there really a significant difference in default probabilities in these two FICO buckets? Enough to warrant 25 bps in premium? I would really like to see more information on how HUD calculated all this.
CALCULATIONS?!! We don' need your stinkin calculations!
Isn't this over reliance on FICO really about the autopilot, CYA society we live in? "No tolerance policies" No intelligence either - but no need to make a decision either ("I had no choice - it's school policy." or "Whocoodanode? The FICO was fine.")
I mean, if I was in the business, I'd consider it a valuable tool - but only that, and one among many.
I'm 58 years old, and things weren't always this way. I mean that this kind of crap existed - but it wasn't so all-pervasive.
The result is an economy - and quite possibly a society and a nation - teetering on the brink of absolute doom because it is so much more important to make the wrong decision for the "right" reasons than to exercise judgement or even ...GASP!... intelligence, when called on for a decision.
When FICO scores become the economic equivalent of the pictures on the register at McPlasticfood, something is seriously wrong.
"Isn't this over reliance on FICO really about the autopilot, CYA society we live in?"
Depends on what you mean by "over reliance". FICO's nothing more or less than an accumulation of data on old loans and the knowledge of how they turned out, that is, that "judgement" of which you speak. At the risk of stepping on some neuroscience I'm unaware of, it boils down to a collection of Xs and Ys not unlike that developed by an expert as she develops her expertise over time. The main differences are A) the computer never forgets a case and B) the computer ignores whatever you don't tell it to look at.
B) is the crix of the problem being talked about on this thread. DTI and LTV were not and are not available to Fair Isaac. They didn't have access to current data on either borrower incomes or the condition of the marketplace for the collateral, if any.
That is where over reliance may have come into the picture. A human underwriter would have significantly less cumulative experience looking at bureau data than is represented in one of the FICO models, no matter how experienced. But the classic FICO models know naught of LTV, DTI ratios, housing bubbles or employment rates. If you underwrote a loan based only on FICO, it would be the equivalent of trying to predict the outcome of a given baseball game using only sabermetrics. It sure helps, but it ignores a number of salient facts.
I've just read the FHA paper & don't understand why they chucked out DPA funds as a determinant factor in risk-pricing. Am aware HUD tried to rein in the DAP's via a rule change last Oct, but failed this past March (?)when a judge held HUD had violated technical procedures in rule-changing.
Don't know the upshot to that-- if HUD plans to try again to get rid of the DAP's-- this time following proper process-- or whether HUD just threw in the towel.
I do believe DAPS negate affordable housing initiatives, & as everybody's now turning to FHA to fill the gap from sub-primes, it's absurd for FHA to put taxpayers on the hook for a whole new batch of these $0 down loans, sold via inflated "appraisals" to cover the cost of the seller's "gift".
In this blog for realtors,this guy's pimping the Genesis program, a DAP he says will "gift" 25K, which "IS NOT tied to a percentage of the sales price" (his emphasis). "My company (*Prosperity Mortgage)can actually go up to 40K for a GIFT!!! This is huge!!" he says. One of his brethren puts it bluntly "FHA's the way to go these days!".
He's pretty vague on the small detail of obtaining an appraisal inflated by 25-40K to cover the "gift"...and dead silent on what happens to a low-income buyer, with no equity, on a home now over-priced by 40K in a down market.
As to taxpayers left holding the loan-- well, screw 'em.
With all that guff in its paper about "promoting sustainable homeownership", I'm disgusted the FHA did this. Are they craven or just raving mad?
Not too expensive to do, and comforting to believe they work, but they don't."
Alas, no. It's verifiably not true, as I've never seen a loan portfolio where FICO (or again, some FICO model) didn't rank-order for risk. So it works, as long as you take "rank-orders for credit risk" as the job at hand. It does not, however, account for income-based metrics or market risk. This means that it will not give you good predictions for the risk of a given loan or portfolio without additional work to take those factors (and others) into account. 600600600, as the absolute risk of those 600s depends on collateral, income and economic conditions.
Uneasyone,
I think you came very close to hitting the nail on the head regarding FICO scores. Todays corporate management fears a SKILLED workforce so all work must be de-skilled. No need for due diligence, no need to know your borrowers, just press the color coded button that dispenses the happy mortgage.
[The result is an economy - and quite possibly a society and a nation - teetering on the brink of absolute doom because it is so much more important to make the wrong decision for the "right" reasons than to exercise judgement or even ...GASP!... intelligence, when called on for a decision.]
How are you supposed to get economies of scale when you have people exercising judgment?
That's why the case-by-case evaluation of defaults is such a problem for the servicers. They had a volume-based business built on automation.
Now, they have to do manual labor... and they're not liking it very much.
It gets Nerdier from here . . .
Tanta, the rest of the post doesn't show up when you click that. OTOH, at least you got the dots right...
No, no, no. The problem wasn't that the rest of the post didn't show up when you clicked the link.
The problem was that the rest of the post showed up before you clicked the link.
I lost a tag there for a minute, and the problem with this "read more" thing is you have to publish the post first, then you can create the link and republish it. Since the preview function in Blogger will not tell you if you have hosed this up, you can't see it until you post it. Then you gotta be faster than bacon dreamz at finding your missing close span tag. See.
Slightly (OT), the demise of this year's house selling season has been greatly exaggerated.
Cheap-home rush cuts O.C. supply under 6 months.
May 19th, 2008, 12:01 am · 14 Comments · posted by Jon Lansner/O.C. Register columnist
Market watcher Steve Thomas at Re/Max Real Estate Services in Aliso Viejo every two weeks calculates market time, a benchmark of how many months it theoretically takes to sell all the inventory in the local MLS for-sale listings at the current pace of pending deals being made.
By this Thomas logic, it would take 5.82 months for buyers to gobble up all homes listed for sale last Thursday at the current pace of deals vs. 6.08 months two weeks earlier and vs. 8.0 months a year ago. Last Thursdays market time is the lowest in 14 months. A key reason: the latest count deals in escrow of 2,658 is up 166% vs. Jan. 19s wintertime low.
Market time is also shrinking because cheaper houses are in vogue, with market time for homes under a half million down 41% vs. a year ago; and for homes between a half-mil and three-quarters of mil down 38% vs. a year ago. The overall market time is down 27% vs. a year ago.
Says Thomas: it is a matter of time before the data that the media is supplied catches up to the story of today: demand is much stronger than last year.
Cheap-home rush cuts O.C. supply under 6 months. - Lansner on Real Estate : The Orange County Register
Me: Where are the "in-the-know", "ahead of the curve" real estate "pro" bloggers on this? The ones who are still trumpeting the "we're going down, we're going down, we're going down!" story?
Sebastia
got it. i love the charts, btw. very colorful!
i love the charts, btw. very colorful!
I was more impressed by the title of that slide. "Especially in prime"? Looks like "Only in prime" to me.
I'm confident that the FICO score has some efficacy in predicting credit worthiness but I think the extent of it's usefulness is exaggerated.
Is there evidence of adverse self-selection in existing FHA loans? Either by FICO or the source of downpayment? Of course as a government program, some of that is probably intentional. Way back in the way back of time, the perception was that FHA existed to help the "deserving poor" who could't come up with 20% DP but were capable of paying off a loan acording to terms.
"Me: Where are the "in-the-know", "ahead of the curve" real estate "pro" bloggers on this? The ones who are still trumpeting the "we're going down, we're going down, we're going down!" story?"
According to Thomas report: 71.5% of the detached inventory under $500K is bank owned or short sale.
Yeah, that sounds like a healthy market, Seb. I can't imagine anyone taking their home of the market so they don't have to compete with the flood of REOs.
Sebastian,
I appreciate all the bottom calls you have been making but I personally won't believe in you bottom calls until you yourself put your own skin in the game. What have you done to take advantage of this once in a lifetime opportunity?
Is there evidence of adverse self-selection in existing FHA loans?
That's an interesting question. In recent years, it seems that the most adverse selection went on in conventional subprime, not FHA.
The connection between FHA and borrower income is actually much more complex than most people think. FHA has never had income limits (except in a couple of targeted programs). What kept it a "low to mod income" program was simply the correlation between borrower income and house price: FHA has always had lower loan limits (until this year).
There's never been any law that says a high-income person who wants to buy a $600K home with a $300K FHA loan can't do that (in a county where the FHA limit goes up to $300K). It's just that they aren't very likely to do it; with a 50% LTV and any kind of decent credit history, you could get a cheaper conventional loan (because there would be no insurance premium of any kind). And, as most high-income people buy more expensive homes and need the higher loan amounts, you just don't get huge numbers of high-income borrowers in FHA.
Unless they have terrible credit or spend so much of their income that they can't save up more than 3% down.
In that sense, it is probable, I think, that higher-income borrowers in FHA are more negatively-selected than low or moderate income borrowers in FHA. So in that sense it's not surprising that the higher-income FHA borrowers would have worse credit profiles.
But all that's ancient history. Now that the FHA limits have been bumped up to jumbo in some markets overnight, you'll have more higher-income borrowers in FHA (because they don't have a downpayment or because they're an underwater refi). This suggests to me that FHA's historical data is now going to be of limited usefulness in predicting performance of its future business. But will that stop them?
Sebastian, That's quite some analysis on that blog. Do you believe it is on par with the quality of CR's work?
Seems to me there is a lot missing. Prices and transaction volumes not just months supply, might be nice.
That aside, I have been suggesting for the past month that we have probably seen the low in transaction volumes. If the FHA bill gets the kabash (it richly deserves) we should see activity really pick up as lenders throw occupants out and aggressively clear their inventories...
Tim said: "I appreciate all the bottom calls you have been making but I personally won't believe in you bottom calls until you yourself put your own skin in the game. What have you done to take advantage of this once in a lifetime opportunity?"
You mean other than being virtually the only one to put myself out there against the daily abuse and ridicule, calling for "no recession" (right so far), a major stock-market low (right so far), new highs in the major indices before the year is over (too soon to know, but increasingly likely), and now a rebound in housing?
Well, my investment in stocks (that I've posted) and my house, I guess.
Sebastia
Fico scores are worthless in my opinion because as a younger person who never took out any student debt I have a 700+ fico score. Even my income does not predict my future behavior when it comes to paying on time. Many people my age have great scores and good incomes but none would hesitate to walkaway from their mortgage if they found themselves underwater.
Sebastian what are your major holdings?
Look, how did my thread on FHA turn into All Attention Given To Sebastian again?
Would you people just quit? We're trying to have a conversation about the post topic. Go somewhere else to have a pissing match.
Sebastian, it was incredibly rude of you to highjack this thread so fast with such a pissy little comment. I don't want an apology; I want you to go away.
And I thought a FICO was a glossy green leafed bush.
I guess we must be obsessed with credit since no-one pays cash anymore but these actuarial one size fits all predictors are not tamper proof. You can still drown in a river that averages 1 foot in depth. Employ a real live real good credit analyst and you can cut your losses by 95%.
Employ a real person to think? What was I thinking...
"Tanta writes:
This suggests to me that FHA's historical data is now going to be of limited usefulness in predicting performance of its future business. But will that stop them?"
My guess is "No". But hey, what could go wrong?
and here i sit, a real credit analyst and at the brink of unemployment..
The trouble we have had recently with FICOs is mostly, in my view, that they were relied on to offset extremely high risk characteristics in loans with a lot of "risk layering."
You can get an AMEN from the Shnapster on that one. Your lead to the jump could have been:
"It gets more sagacious from here . . ."
nevermind, I don't want to feed your "massive 'bubble' ego."
barely said: "Sebastian, That's quite some analysis on that blog. Do you believe it is on par with the quality of CR's work?"
Well, if the analysis only presents one side, doesn't that call its quality into question?
Or does real estate only go down now, when before it only went up?
Sebastia
Tanta said: "I want you to go away."
Very well.
Sebastia
nevermind, I don't want to feed your "massive 'bubble' ego."
Don't worry about that. Whenever I do manage to come up with anything like a sagacious point, I manage to distract everyone with typos and broken links.
Frankly, I don't think it's such an especially profound point. But there has been so much nonsense spread about FICOs lately that it seemed worth making (again).
Tanta,
Could the risk based MIP pricing be a pre-cursor to looser FHA underwriting standards?
It's becoming clear FHA is going to be financing a lot of borrowers for at least the next two years. Could looser underwriting be next, justified by the higher MI?
Honestly, I can't see a reason for them to make this change now, except for maybe to factor in FHASecure. Of course, if FHASecure is such a "success" why not keep it on permanently and use it as a 100% purchase program. Their elimination of DAP as a risk factor suggests that's the way it's headed.
Lending money is a serious matter. Too serious to be left in the hands of kids with computers.
FICO scores attempt to judge a persons ability to repay a loan. Repaying a loan also involves a willingness. Perhaps we should have a morality test to match the ability test. Willingness to repay is becoming more important in my opinion.
I would really like to see the work on this one.
It gets Nerdier from here . . .
Is there more? I don't know if this is what BD referred to, but I see a continuation link on both the shrunk and expanded versions, but the latter doesn't change the display. Just the way it is?
"This suggests to me that FHA's historical data is now going to be of limited usefulness in predicting performance of its future business."
I was thinking something similar: FHA may have added some significant complications and uncertainties to its model. I mean any insurance business must face problems of signaling and screening so adding a pool of people who probably pay more attention to (and actively maintain) their FICO scores and also putting more money at risk could introduce a lot more policy risk than realized.
That is, people generally know more about their financial situation than the insurance company and this additional pool of jumbo-loan folks probably are more acutely aware of their financial situation than most. If those same folks know they face large risks then they are more likely to want to buy insurance and any insurance company, FHA included, must either try to minimize the problem that too many people with big risks will buy their product (the problem of adverse selection) or they must price that risk appropriately or both but ...in the case of FHA where are they going to get the data to do that?
It seems to me that this adds more uncertainty to the mix and as we've learned in the CDO/SIV debacle that is generally not a recipe for success.
Sorry, Econoclast. That's fixed, if you refresh.
FICO scores attempt to judge a persons ability to repay a loan.
Actually, you've got the lingo backwards. FICO measures "willingness to repay." DTI, employment history, cash reserves, etc. measure "ability to repay."
Downpayment is a little of both. In the sense it measures the borrower's ability to accumulate savings, it measures ability to repay. In the sense that it functions as "skin in the game," it measures willingness to repay.
I have no problems with FICOs used to approximate the results of a manual review of a credit report. That's all they are: a way of quantifying the relative acceptability of a jillion datapoints on a credit report.
The problem comes in when you give it too much weight in the lending decision. It does not "offset" ability to repay or down payment. That's the mistake we made.
My concern about what FHA is doing is that they're giving too much weight to FICO ranges in setting premiums. That Moody's chart suggests that there isn't that much of a difference in delinquency within the category subprime: once you get down to that range, further distinctions aren't necessarily helpful. So I fear that FHA feels like it is making important risk distinctions, but it isn't really.
This whole thing gets right to the heart of the great contradiction of insurance.
The theory of insurance is to distribute losses so that everybody pays some part of the total community and nobody pays all. However, insurance companies then proceed to slice and dice to apportion 'fairly'... in other words, to return to the original situation where the loss goes to the single loser as much as is possible. If you cut it into fine enough groups, you have no insurance at all - you just have prepaid losses by induviduals with insurance companies taking a percentage for the favor.
It's clear that FICO is a statistically significant variable.
"they were relied on to offset extremely high risk characteristics in loans with a lot of "risk layering.""
Yes...So what is the statistical significance of the other variables such as DTI, CLTV, and negative equity?
Tanta,
In Sebastian's defence, what he did is par for the course online. Your response has the air of someone new to the virtual world.
The only reason Sebastian can "hijack" a thread is because so few pose contrary commentary, leave alone whether it has even a modicum of ration support which Sebastian at least tries to provide.
The fact that he didn't post his comment in a relevant thread is because you hosts don't create threads devoted to free-for-alls. Therefore posters use all of them as such.
OT - Tanta, if you're interested, a related article to your Mortgage Fraud Employee Benefit Post. Seems that Scripps Research Institute is using mortgages to attract top biomedical researchers.
Local News: West Palm Beach, Palm Beach County, Martin & St. Lucie Counties | The Palm Beach Post
"Scripps issues million-dollar mortgage to lure scientist
By STACEY SINGER
Palm Beach Post Staff Writer
Thursday, May 01, 2008
The Scripps Research Institute has started issuing home loans to sought-after scientists, giving one top recruit a $1 million mortgage for an oceanfront condo that cost $850,000.
Scripps won't say why the mortgage exceeds the condo's purchase price, or whether the loan was derived from a $310 million state grant used to attract the biomedical research powerhouse here in 2003. The terms of that home loan, and three others like it, are a secret that Scripps Chief Operating Officer Douglas Bingham said he will not disclose.
Using the state grant for home loans wasn't contemplated when the Scripps deal was negotiated. But Bingham said that the state's agreement gives the institute "broad authority" to use grant money for employee compensation and benefits.
The million-dollar mortgage recently was filed in the public record. After The Palm Beach Post asked about it, Scripps notified the Scripps Florida Funding Corp., which oversees Scripps' grant.
Tough competition for scientists, coupled with recruits' difficulty in selling their existing homes, prompted Scripps to sweeten offers with housing loans. Bingham said top scientists who bring grants and intellectual property typically have multiple offers, making an attractive recruitment package a must.
..."
Frankly, I don't think it's such an especially profound point.
Well, it certainly shouldn't be. But I'm afraid our sound-bite-oriented culture doesn't like to wrap their heads around issues such as risk-layering, and would rather focus on something like FICO as if it's the only factor to consider in lending.
Lets hear greasy hair Sheila Bair make a nice speech about this topic of interest:
BLOOMBERG
Banks Keep $35 Billion Markdown Off Income Statements
"Banks and securities firms, reeling from record losses resulting from the collapse of the mortgage securities market, are failing to acknowledge in their income statements at least $35 billion of additional writedowns included in their balance sheets, regulatory filings show. Citigroup Inc. subtracted $2 billion from equity for the declining value of home-loan bonds in its..."
Banks Keep $35 Billion Markdown Off Income Statements (Update1) - Bloomberg.com
Your response has the air of someone new to the virtual world.
I'm not new, love, I'm different.
I don't care how the rest of the blogosphere does things. Around here, we like a better signal-to-noise ratio.
Sebastian has been pulling that stunt of his for over a year now. I'm tired of it. And I am not obligated to open "free for all threads" so that Seb has some place to pick a fight.
He needs to get his own blog.
Tanta,
"The problem is not that a 90% LTV loan with a 720 FICO won't outperform, statistically, a 90% loan with a 620 FICO. It will, although it isn't always clear that the difference in performance is all that substantial. The problem is that a 100% loan with a 720 FICO will not necessarily outperform a 90% loan with a 620 FICO. The idea that a high(er) FICO offsets lack of downpayment or high DTI is currently dying a painful death."
I believe I've parted company with you around this place previously with regards to models/scores. FICO basically measures "Willingness to pay" (the W in the SAW) framework, based on bureau-reported tradeline performance. But nothing that I know of says that you can't take LTV, DTI or local economic or market conditions into account when developing an internal model or scoring system.
Rather the opposite; if you were to limit yourself to bureau data, you'd just be mining a claim that Fair Isaac has already worked over extensively. Hell, I was at an OCC conference a few years ago where the panel was point-blank asked if economic conditions should be taken into account and they seemed confused that anybody would have even bothered asking.
That this appears not to have happened in a lot of shops is a mystery. Possibly it was and nobody paid attention because it was more profitable to do business based on the assumption that credit and business cycles were dead. I still remain in the camp that says that these things can and should be modelled, though.
CR and Tanta,
I too would like to click on the comments for a post and see an actual discussion (or not) of the topic du jour, not the usual mix of thoughts. However, I don't believe that humans can be regimented into this, especially not on pseudonymous blogs. Instead, I think what can approximate it is the creation of a convenient outlet for deviant thoughts to be self-marshalled, i.e. a way for commenters to stay on topic and easily post elsewhere when off-topic knowing that other readers will actually read and engage the alternate site.
DailyKos has a feature that allows commenters to create their own diaries, which can be elevated or reduced in visibility. Perhaps something like this would allow posters like Sebastian to include useful commentary without hijacking a thread much to Tanta's disgust (though CR has always manifested a more easy-going approach). Alternatively if the diary infrastructure is cumbersome, perhaps just a link to a forum where regulars can post their own topic and start threads that don't disrupt the main threads discussions.
Of course, this may not fit with the vision of CR, in which case, please ignore my $0.02.
Should the FHA really be making loans like this in the first place? I mean their charter is to help enable the dream of home ownership and all that jazz, correct? How does putting people into 99% LTV loans really serve the public good, or even the borrower? Enabling loans that are more likely to default really promotes home purchasing, not long term ownership.
Looks that way to me at least.
I still can't get my head around why FHA is underwriting jumbos to people who couldn't get one in the open market. I understand helping 1st time buyers, not subsidizing riskier candidates.
It's clear that FICO is a statistically significant variable.
I don't really know what you mean by that statment.
Look, FICOs just rank-order a universe of people with credit histories. But the slope of the default probability curve here is convex, not linear. There is a signficant difference in default probability between 580 and 620. However, the difference between 780 and 820 is not visible to the naked eye.
For the moment it doesn't matter about those other factors. If you look at the HUD premium chart I linked to, it is making premium pricing distinctions within an LTV group. That is to say, at the same LTV a borrower with a 550 FICO will pay more premium than a borrower with a 580 FICO. I, however, am suggesting that once you get that low, it is hard to say that there's much of a difference. In any case I would like to see some data on that from HUD. The Moody's data suggests that the performance of loans under about 660 is all the same, regardless of how you bucket the under-660 group.
But nothing that I know of says that you can't take LTV, DTI or local economic or market conditions into account when developing an internal model or scoring system.
Nothing I know of says any different.
But we aren't talking about FHA's TOTAL scoring system for loans.
We're talking about how it prices its insurance premium. And the premium matrix that it is rolling out charges a signficantly higher premium for lower-FICO loans at the same LTV. This is not the same calculation it uses for approving loans.
"Where are the "in-the-know", "ahead of the curve" real estate "pro" bloggers on this?"
Sebastian, wouldn't you agree that the table on Lansner's blog shows that the months-of-supply on houses UNDER $750K have fallen, while the months-of-supply of houses over $750K have RISEN?
If we were seeing a sustained turn-around, I would assume that the high end would see it as well (especially since 'everyone' says that the high end won't be hurt by the downturn).
Anther thing to consider...the cutoff (the mid $700K) seems rather close to the cut-off for "jumbo conforming loans", which leads me to believe that without Fannie/Freddy/FHA backing, we wouldn't see this reduction in months-of-supply. In other words, a lot of the reduction in months of supply in those price ranges might be caused by the start of "easy credit", this time backed by the US government instead of sub-prime and alt-a outfits.
sebastion,
here in the DC burbs, Charles County MD in particular we are still growing every month and have 15 months inventory.
i think your story has a key word..."cheap" you see we still have prices mostly stagnant while values are going down. our small county has the highest FC rate in all of MD.
we add 300+ homes a month and "sell" 110 a month... yoy total sales in dollars is down 59%...
so i think cheap is the cause of the movement...
Curious ,
as Grandad often said . . .
"Yeah , and people in Hell want
ice water"
"A study of an entire year`s applications..." Maybe they were talking about reverse mortgages in 2005. It is a posibility. A lot of smart old people made out like bandits thanks to FHA.
without hijacking a thread much to Tanta's disgust (though CR has always manifested a more easy-going approach).
How do you know that CR doesn't just quit reading comment threads that degenerate into Sebastian-baiting?
I myself just quit reading threads that degenerate like that. And I miss the opportunity to discuss the post topic with people who care about the post topic, because of course they too have quit the thread at that point.
Noise drives out signal. We are now a few hours into this thread and people are still "responding" to Sebastian even after he went away. (Anyone want to bet how long that will last?)
Anyone who wishes to start the financial world's version of Kos is welcome to do it. I think, for instance, it would be fantastic if you and Seb got together and did that. Everyone should pursue his or her own vision.
"I don't really know what you mean by that statment."
Statistical significance is a matter of math (F Value). It's certainly reasonable for a variable to lose importance as other variables gain.
I read what you wrote. I conclude we're in agreement that other statistically significant variables should be considered. My point is that FICO is (perhaps) one variable in a curvilinear model.
Tanta,
Was I reading this right? You said "the difference between 780 and 820 is not visible" and "the performance of loans under about 660 is all the same." Between these ranges there is presumably increasing risk of loan failure. Does FHA's rule of a minimum FICO of 580 make no sense then?
Regarding the seller contributed down payments I believe the recently settled California lawsuit mandates that FHA continue to accept these. FHA's response has been to come up with some risk related pricing which superficially seems like a good idea. Is it not?
http://209.85.173.104/search?q=cache:XdVlAGLwl50J:seattlepi.nwsource.com/local/339241_downpayment12.html+nehemiah+fha+lawsuit&hl=en&ct=clnk&cd=3&gl=us
FHA is not supposed to make money on these insurance premiums but it is also not supposed to lose money. From the article you quoted I gathered that all risk variables would go into FHA's pricing. Isn't FHA required by law to publish these 60 days before implementation?
Using projected economic conditions to predict loan performance would be extremely difficult and outside of the capability of the mortgage finance system. That's the skill billionaire hedge fund operators claim.
Two very different comments--
John Stumpf of Wells Fargo famously said that he worked in an industry that had busily been inventing new ways of losing money, even though it already had plenty of good old-fashioned ways of doing so. Not so famously, and at the risk of getting folksy, when I was a kid, my Dad always used to harp on using the right tool for the job (he got mad when I used a screwdriver as a chisel, or maybe the other way around). Question to Tanta--are you basically saying that the FICO score is a perfectly good 'tool' that has recently been misused for one purpose (to Dad's irritation), and now is being misused for a new purpose (to Stumpf's observation)? In principle, it would seem to me risk-adjusted and variable premia for insurance is a fair and laudible goal. How to get there is another and more difficult matter--viz., the sex bias in life insurance and annuity tables, which are at least consistent in their discrimination (so that men pay more for life insurance and women pay more for annuities, because insurance companies pay out faster on the life policies to the men and pay out longer on the annuities to the women, given the sex-based difference in life expectancies).
Second, trying to call a national bottom to the residential real estate market is impossible. I have a slightly different question: whether the areas that have fall hard and fast (i.e., Miami, San Diego), will continue to fall harder and faster, or whether they were merely first and the areas that have only slowly started to fall and fallen only moderately (i.e., Seattle and Charlotte) will catch up with them and then stay in the slump longer. My sense, though I can't support it, is that the process will be gentler in those markets where it was slow to start and has been milder, and that it will remain harsher in those markets that have already suffered the most. And, further, that no one can predict at this point which market will start a recovery first, but that any talk of recovery is premature--once bottom is hit, that's where price levels will remain for the foreseeable future, since asset bubbles don't reflate--at least not in the time horizons on which financial markets operate.
Of course the lower income fico scores are better. If you take out a heloc and use it to significantly pay down your cc bills your score will look great (at least temporarily), even though your credit situation has grown far more precarious.
Those lower income people are the ones that have been trying to massage their scores and they know how from radio talk show financial people. Contrarily, if you are more secure financially, you don't consume credit the same way to have a score boost, nor do you obsess over what your score may be, just assuming it must be good.
Does FHA's rule of a minimum FICO of 580 make no sense then?
When did FHA ever have a minimum FICO of 580?
If you look at the premium chart in the document I linked to (page 26), you will see that they are pricing premium all the way down to a FICO of 300.
http://portal.hud.gov/pls/portal/docs/PAGE/FHA/IMAGE_LIBRARY/5171-N-02%20RBP%20FINNTC%20TO%20PUBLISH%20%205-13-08%20.PDF
I sort of resent FICOs for being undemocratic or something.
But yeah, the more cogent reason to object is that they aren't the best data. AFAICT, lenders use FICOs because it is easy and it's a number. That subprime plot illustrates that at least in some circumstances it provides practically no utility.
Tanta has taught is that a down payment, saved up over time by the borrower, is a much better indicator of the quality of the loan. It isn't that hard to measure this, so why don't they?
I would disagree that FICOs primarily measure willingness to pay, by the way. I would say they measure meticulousness and organization. Those things are not unrelated, and so the FICO does have utility, but they aren't the same thing. A person who is sloppy enough to have paid some bills late might nevertheless be very committed to paying them. In particular, such a person might be quite determined to stay in her home even if she has to eat rice and beans.
Fair Isaac white paper --> Fair Isaac
Question to Tanta--are you basically saying that the FICO score is a perfectly good 'tool' that has recently been misused for one purpose (to Dad's irritation), and now is being misused for a new purpose (to Stumpf's observation)?
Let me try to state my concern with an analogy to health/life insurance:
Do insurers charge smokers higher premiums than non-smokers? Yes, they do.
Do they publish premium schedules that distinguish between two or more packs a day, one and 1/2 to two packs, one to 1 1/2 packs, 1/2 to 1 pack, and less than 1/2 pack? Not as far as I know.
Of course, they have no particularly reliable source of such fine data. Then, even if they did, they'd have to have some medical evidence that someone who smokes 20 a day is really at a substantially greater risk than someone who smokes 15 a day.
Now, take a look at the Moody's chart I posted. All of the subprime loans perform worse than the Alt-A or (prime) jumbo loans. No surprise there. But within that subprime category, they aren't showing very much difference between the three FICO buckets. There's obviously a slightly elevated DQ rate with the blue line (below 500) but what's the difference here between the 501-620 and the 621-680?
I'm not sure it's so much misusing FICOs in terms of purpose as it is making unnecessarily fine distinctions just because we have the kind of data that can be bucketed like that. I am reminded of (home) PC users agonizing over whether they should buy 10 gigs of memory or 12 gigs. They'll probably use three.
from the 2007 MMI Fund actuarial review:
The main enhancement of the econometric model was to improve the incorporation of borrower credit history information at the loan-level. By using borrower information at the individual loan level, the model is more sensitive and can better differentiate loans that have the same product type, origination year, loan size and initial
LTV, but have different borrower scores. Modeling at the loan level also provided better control for potential bias due to choice based sampling of historical credit scores. The estimation results confirm that credit history is among the most influential factors explaining the claim probability among individual FHA-insured mortgages and is also a significant factor in explaining prepayment behavior.
HUD FY2004 Actuarial Review of the Mutual Mortgage Insurance Fund
I am a sloppy payer, but I always pay. Frankly, I don't need credit, so it doesn't matter to me.
And as to the ciggie analygy, I think
I read that people who only smoked half a pack a day were distinctly healthier than those that smoked a pack.
Likewise those curves are at a distinct time in our economic history, so they might look different at a different time.
Like the half pack a day smoker living in the unpolluted country vs.
a very polluted city.
The thing about not looking around to see what's going on in the economy, sometimes it might be hard and subtle, and sometimes it might smack anyone in the eye with obviousness. And sometimes we have people in denial when it's already started. A person up the thread said he asked about external conditions, and was looked at blankly. Clearly, the people at the conference hadn't even thought of it,
rather than rejecting it as too hard or too uncertain.
This is why people are supposed to make a difference; we are supposed to think about stuff. Hah.
here are the regression coefficients for the FICO buckets (for 30yr fixed mortgages) for the conditional claim rates model estimation:
300 to 499: 1.2376
500 to 539: 0.9128
540 to 579: 0.6847
580 to 619: 0.5089
640 to 659: 0.1092
http://www.hud.gov/offices/hsg/comp/rpts/actr/2007appa.pdf
Thanks for the link, bacon dreamz. I skimmed Appendix A rather quickly, and I still don't see why they decided on the FICO buckets they chose. I guess I'll have to spend some more time with it.
OK, and here are the premiums:
For loans with LTV < = 90%:
300 to 499: 1.2376 175 bps
500 to 539: 0.9128 175 bps
540 to 579: 0.6847 175 to 559, 150 over 559
580 to 619: 0.5089 150 to 599, 125 over 599
640 to 659: 0.1092 125 bps
Lawyer Liz: "I am a sloppy payer, but I always pay. Frankly, I don't need credit, so it doesn't matter to me."
I am by nature a sloppy payer who always pays. But I do need credit so I have had to adopt a rigorous routine wherein I pay bills every week and have my computer remind me to make sure payments went through on the appointed day. All to feed my FICO.
I remember when it used to be fine to pay your utilities every other month.
I guess I'll have to spend some more time with it.
ok, have fun. i'm going to sit here and munch on my sandwich now.
The cigarette analogy?
Not true for a group insurance policy; true for individuals. That's exactly the point: just like banks wanting to make loans to people who don't need money, insurance companies want to sell to people who will not have losses. But why would those people want to buy insurance?
To get into the argument about whether FICO matters or down payment matters or income matters is to buy into the premise that insurance should not be a group sharing of loss, but should be 'fairly' allocated. In other words, those who have losses should have losses, those who don't shouldn't have. Then what is the purpose of insurance?
GrandmotherOfShnapsLiz wrote:A person up the thread said he asked about external conditions, and was looked at blankly.
Actually, it was BG who wrote that he asked, at a conference, point-bank: "if economic conditions should be taken into account".
I think the problem is, 'economic conditions' are a matter of great speculation, whereas something like a FICO score is a matter of fact. That said, some have already issued special criteria applicable to loans in specific markets which are thought to be weak. How's that for taking 'economic conditions' into consideration?
Well, the correlations certainly seem to support more than one bucket but if they're going to dice things up like that wouldn't an LTV range tell them more? I'm assuming FHA would have a fair amount of data on probability of default for LTV's between, say, 80% - 90% so, given the convex shape of the FICO curve, adding a FICO > 580 would presumably tell them more than breaking down fees by FICO alone.
the correlations certainly seem to support more than one bucket
How about a black-box multivariate dealy? They can hire some out-of-work quant guys to build it and 'modernize' the FHA in more ways than one.
I kind of like this premium adjustment matrix:
Downpayment Fund Source:
DAP: +225 bps
Own pocket: n/a
Tipjar: -50 bps
but if they're going to dice things up like that wouldn't an LTV range tell them more?
The premiums do vary by LTV. Within each LTV band, they then vary by FICO.
I just typed in the first (less than 90%) premiums for an example. If you look at page 26 of the HUD document you can see the whole table. It is too large to retype in the comments and I didn't take the time this morning to try clipping it out of the text to post it.
Hey, speaking of tipjars - did that get chucked during the site redo?
Hey, speaking of tipjars - did that get chucked during the site redo?
It's about time somebody noticed.
I heard it was lost during the move...CR didn't label all of the boxes so we just have to wait until he finishes unpacking...
Emma Anne,
"I would disagree that FICOs primarily measure willingness to pay, by the way. I would say they measure meticulousness and organization."
Misconception. While any significant level of 30+ DPD will hit your FICO (or I should say FICOs, since there are multiple Fair Isaac models), the big damage comes in the form of defaults, which is not a matter of sloppy payments. In my experience you can maintain a high-600 level FICO with some 30+ late pays, and late pays that don't rise to that level will barely register (since they probably won't be reported to the bureaus in the first place.)
So truly low FICOs are indicative of either an unwillingness or inability to pay. The weight is usually put on the former because A) the model doesn't have access to your income and B) it is assumed that the debts previous lenders made to you had at least some underwriting and were within your ability to service. B) might be questionable to some extent at the moment, but still appears to hold true for the vast majority.
"It's about time somebody noticed."
I though y'all were making money hand over fist, what with the redesign and all.
Tanta:
Here is the piece (or so I think) that Harney was referring to:
http://edocket.access.gpo.gov/2008/pdf/E8-10625.pdf
Is a discussion of the tip jar related to the main subject? If not, shouldn't it "just go away"?
From the article: But is there really a significant difference in default probabilities in these two FICO buckets? Enough to warrant 25 bps in premium? I would really like to see more information on how HUD calculated all this.
CALCULATIONS?!! We don' need your stinkin calculations!
Isn't this over reliance on FICO really about the autopilot, CYA society we live in? "No tolerance policies" No intelligence either - but no need to make a decision either ("I had no choice - it's school policy." or "Whocoodanode? The FICO was fine.")
I mean, if I was in the business, I'd consider it a valuable tool - but only that, and one among many.
I'm 58 years old, and things weren't always this way. I mean that this kind of crap existed - but it wasn't so all-pervasive.
The result is an economy - and quite possibly a society and a nation - teetering on the brink of absolute doom because it is so much more important to make the wrong decision for the "right" reasons than to exercise judgement or even ...GASP!... intelligence, when called on for a decision.
When FICO scores become the economic equivalent of the pictures on the register at McPlasticfood, something is seriously wrong.
"Isn't this over reliance on FICO really about the autopilot, CYA society we live in?"
Depends on what you mean by "over reliance". FICO's nothing more or less than an accumulation of data on old loans and the knowledge of how they turned out, that is, that "judgement" of which you speak. At the risk of stepping on some neuroscience I'm unaware of, it boils down to a collection of Xs and Ys not unlike that developed by an expert as she develops her expertise over time. The main differences are A) the computer never forgets a case and B) the computer ignores whatever you don't tell it to look at.
B) is the crix of the problem being talked about on this thread. DTI and LTV were not and are not available to Fair Isaac. They didn't have access to current data on either borrower incomes or the condition of the marketplace for the collateral, if any.
That is where over reliance may have come into the picture. A human underwriter would have significantly less cumulative experience looking at bureau data than is represented in one of the FICO models, no matter how experienced. But the classic FICO models know naught of LTV, DTI ratios, housing bubbles or employment rates. If you underwrote a loan based only on FICO, it would be the equivalent of trying to predict the outcome of a given baseball game using only sabermetrics. It sure helps, but it ignores a number of salient facts.
Thread hijack:
I think fico is comparable to polygraph.
Not too expensive to do, and comforting to believe they work, but they don't.
I've just read the FHA paper & don't understand why they chucked out DPA funds as a determinant factor in risk-pricing. Am aware HUD tried to rein in the DAP's via a rule change last Oct, but failed this past March (?)when a judge held HUD had violated technical procedures in rule-changing.
Don't know the upshot to that-- if HUD plans to try again to get rid of the DAP's-- this time following proper process-- or whether HUD just threw in the towel.
I do believe DAPS negate affordable housing initiatives, & as everybody's now turning to FHA to fill the gap from sub-primes, it's absurd for FHA to put taxpayers on the hook for a whole new batch of these $0 down loans, sold via inflated "appraisals" to cover the cost of the seller's "gift".
If this site's any indication, RE agents are busy spreadin' the news: Real Estate Blog - Spend Less Time Playing Tour Guide.
In this blog for realtors,this guy's pimping the Genesis program, a DAP he says will "gift" 25K, which "IS NOT tied to a percentage of the sales price" (his emphasis). "My company (*Prosperity Mortgage)can actually go up to 40K for a GIFT!!! This is huge!!" he says. One of his brethren puts it bluntly "FHA's the way to go these days!".
He's pretty vague on the small detail of obtaining an appraisal inflated by 25-40K to cover the "gift"...and dead silent on what happens to a low-income buyer, with no equity, on a home now over-priced by 40K in a down market.
As to taxpayers left holding the loan-- well, screw 'em.
With all that guff in its paper about "promoting sustainable homeownership", I'm disgusted the FHA did this. Are they craven or just raving mad?
Billy Hill,
"I think fico is comparable to polygraph.
Not too expensive to do, and comforting to believe they work, but they don't."
Alas, no. It's verifiably not true, as I've never seen a loan portfolio where FICO (or again, some FICO model) didn't rank-order for risk. So it works, as long as you take "rank-orders for credit risk" as the job at hand. It does not, however, account for income-based metrics or market risk. This means that it will not give you good predictions for the risk of a given loan or portfolio without additional work to take those factors (and others) into account. 600600600, as the absolute risk of those 600s depends on collateral, income and economic conditions.
Uneasyone,
I think you came very close to hitting the nail on the head regarding FICO scores. Todays corporate management fears a SKILLED workforce so all work must be de-skilled. No need for due diligence, no need to know your borrowers, just press the color coded button that dispenses the happy mortgage.
[The result is an economy - and quite possibly a society and a nation - teetering on the brink of absolute doom because it is so much more important to make the wrong decision for the "right" reasons than to exercise judgement or even ...GASP!... intelligence, when called on for a decision.]
How are you supposed to get economies of scale when you have people exercising judgment?
That's why the case-by-case evaluation of defaults is such a problem for the servicers. They had a volume-based business built on automation.
Now, they have to do manual labor... and they're not liking it very much.