So, this is what you were cooking up yesterday, when CR mentioned writing; I better do a word count, and then maybe even read it...

Re: "fitting loans into a matrix "

Was this done by phone and with slide rules; don't get me wrong, I image the old ways had less fraud, but do tell...

otiose = superfluous (new one there for me)

I'm not sure you'll play ball here, but when you speak of the old days, I jump forward to rating models and FICO scores and the recent abuse and fraud that was in the system, which was widespread; I like the term collusion. To wit, the recent rating models seem to still be based on these old days your relating too, i.e, these new models are using old histories of a time when debt and creditworthiness were less volatile and less leveraged.
I like to think this new age of attention deficit/lotto mentality skewed the old data and distorted the payment history statistics which are still related to the new models.

To make a long boring post more useless, I think that after 9/11, as rates were dropped to 40+ year lows, everyone that had a mortgage in America did a refi and as millions of loans were scrambled, someone forgot to re-calibrate the ratings models.... what think dude?

Nice Work. Thanks.

Tanta-

Since "quite" means "not really" in British, I imagine (and hope) the UK will adopt "quite prime" lending once this crisis is past.

This could be yet more grist for your explanatory mill come next crash.

BG writes:
Tanta,

...You just shouldn't be able to originate paper where all signs point to "no".

I think I heard a cheer from the operations side

Re: "I think it will be a long time before lending standards ease significantly, and I think subprime will come back first. But I do suspect we've seen the last of Alt-A for a much longer time.
"

I continue to wonder where cash flow will come from, in regard to banks that are writing off multi-billions. Nonetheless, when you see NAR or real estate hype, many suggest a bottom is in, yet with tighter standards and homes still way over-valued in places, I fail to see how banks will be able to support dividend growth in the future and grow sales without this type of fraud backing them. The covered bond scam is just another way to repackage old debt into new debt, so I'm wondering what the driver will be for the economy, and why people think that housing will turn back into a flippers mkt?

Besides, usually when you look at the credit reports of people who have "good credit" but "unstable incomes," you simply find that they've done OK managing very small amounts of personal debt. We have to get away from this idea that successfully managing a credit card account for a year or two with a $10,000 limit qualifies you to buy a $500,000 house.

The key point there is the price of the house. I'm one of those people with a FICO score of 810 and an unstable income from being an independent contractor. I just barely scraped by via tax returns for a loan on a $175k house back in 2002.

For someone self employed, they can own a home with a traditional loan as long as they are excellent money managers. In the years when you do really well, you pay off all credit cards and the car. You also make sensible choices about what you can really afford based on your "lean" years as opposed to your "good" years. You don't EVER make long term financial decisions based on what you can earn in "good" years, you make them on what you can afford to carry in the "lean" years.

It's the people who are lousy money managers that need to be parted from the ability to get no-doc loans.

...You just shouldn't be able to originate paper where all signs point to "no".

unless it got you to the finish line...a CLOSING....who cares what happens after that.

Oh yeah...I never forget one time waiting for my car to be serviced I wandered into the showroom and one of the salespeople came up to me and said "pick one, any one, and I will make SURE you drive it home today...."

The effect was not the one intended ( I am assuming) since it became obvious to me that he was all about moving product No Matter What and that it had absolutely nothing to do with incidentals like whether I, say, needed a car, or could afford a car, or had a Driver's License....

It was about moving product, as much and as fast as possible...and less about putting people in securely into housing.

Rob,

I disagree based on one criteria alone: performance. Backintheday when we sat aaround with the data ; those loans did fine. It isn't a coincidence that things started going sour at the same time the derived FICO trumped the enumerable source material.

Correlation =/= causation. The problem doesn't have much to do with FICO per se (though I'm of the opinion that it's not that hard to do better), but rather with how it is used. FICO only measures historical performance. By definition, it can't deal with income and stability thereof, collateral and the soundness of the market for same, or clever fraud/ identity theft. If a FICO or FICO-like score is used as a proxy for a lot of tedious work looking at bureau printout, it works fine. FICO, or rather, the various FICOs, still rank-order risk quite well. The problem is, you need a person or another, different system to capture market effects, attempted fraud and recent changes in the applicant's status.

Also, higher up, you need somebody paying attention to the trends and applying a sort of meta-analysis. i.e. "2003 vintage looks too good compared to predictions; wonder if it has to do with refinancing bads going into 2004?"

Ms Tanta,

What is your professional opinion on the amount of discretion used to fudge valuation metrics, i.e, you have appraisers that can fudge and loan officers that can use discretion and banks that can fudge numbers -- so, in the past, did it seem like there was on-going fudging and nudging and massaging and skewing and bias manipulation, or does this latest era of bubble fraud seem to be out of control?

Tanta,

"on--a paid-as-agreed credit account with Fingerhut or K-Mart didn't count as much as a paid-as-agreed or even a mildly delinquent Gold Card. Because of the class biases of the underwriter, not for any logical scoring of credit performance"

I would love a Tanta essay on the class/sex/racial/religious biases of the old approval system. I know there were several cases of blind studies done where equal applicants were treated to different results. It seems like I remember this was major news about 15 years ago.

"The whole rationale for the famous 2/28 ARM was that after two years of good payment history on that loan, the borrower could refinance into a prime loan and thus never have to pay the "exploding" interest rate at reset"

\t
Two years is also the time it takes for a speculator to buy a nascent house and sell it once finished

For those who want an even more detailed look at the early history of subprime, get the recently published book "Chain of Blame," by Paul Muolo and Mathew Padilla.

The trouble with that assumption was that we were busy building a credit industry in which there was plentiful credit--on easy terms--for people with any FICO, any "reputation."
A huge change was made in the late 80s or early 90s and no one held a news conference to announce it -- and people of a certain age weren't aware that things were changing.

I'm 45. When I was in college, and for a few years after I graduated in '85, credit reputation mattered because you couldn't get a credit card without it. As feckless as I was in my 20s, I was turned down for credit cards and store charge cards and even for "overdraft protection" at a bank. A lot of people my age and older have had experiences like this or they knew someone who it happened to.

Then college students started getting credit cards when they were still freshmen or sophomores. Credit became available to risky people. And I think a lot of folks were unaware of this. In the late 90s and early 2000s, they assumed that, if they were approved for credit, that meant they were creditworthy.

Many people were unaware that borrowers, not lenders, had become the credit gatekeepers. Or, in Tanta's words, that "we were busy building a credit industry in which there was plentiful credit--on easy terms--for people with any FICO, any 'reputation.'"

It's easy to borrow too much, at too-risky terms, if you mistakenly believe the lender is making rational decisions.

There were large incentives to game FICO scores when it made difference between getting and not getting a mortgage. If collateral and capability resume their traditional importance then the difference becomes being prime or prime- or subprime plus. Not worth the effort especially if attempts to game can result in reducing the credit scores and being moved to subprime plus to subprime when such gamming attempts are recognized.

LL,

When you are a human being, and you are constantly, if tediously going thru someone's credit history, you get the ability to get a hunch that there's just something wrong there.

I don't believe in hunches, I believe in p-values. People have too many natural cognitive biases, and that doesn't even get into bigotry.

Tanta,

That was a good ol post ( SATURDAY, SEPTEMBER 29, 2007):

1. I have said before that stated income is a way of letting borrowers be underwriters, instead of making lenders be underwriters.
2. What the stated income lenders are doing is getting themselves off the hook by encouraging borrowers to make misrepresentations. That is, they're taking risky loans, but instead of doing so with eyes open and docs on the table, they're putting their customers at risk of prosecution while producing aggregate data that appears to show that there is minimal risk in what they're doing. This practice is not only unsafe and unsound, it's contemptible.
3. All those claims by securities issuers and raters about how we had no idea that gambling was going on in this joint are directly comparable. The tough news for the self-employed "respectable" borrower is that I don't care if you're individually willing to play bagholder: you can't afford to underwrite that collective risk.

People have made millions gaming the FICO.

That's why retrospective analysis doesn't work very well in this type of industry. Retrospective analysis works because nobody at the time knew what they would be measured on. Nobody was actively trying to boost their scores by meeting criterion A, B and C. Once people found out what they were scored on, they started to game the system and FICO scores were inflated. All of the sudden people don't have high FICO scores because they are creditworthy, but because they know how to make it look like it.

Ignorantia juris non excusat or Ignorantia legis neminem excusat

Ignorantia juris non excusat - Wikipedia, the free encyclopedia

Re: What the stated income lenders are doing is getting themselves off the hook by encouraging borrowers to make misrepresentations. That is, they're taking risky loans, but instead of doing so with eyes open and docs on the table, they're putting their customers at risk of prosecution while producing aggregate data that appears to show that there is minimal risk in what they're doing. This practice is not only unsafe and unsound, it's contemptible.

Heck, it the debt can't be discharged, why should the lender be a gatekeeper at all?

Practically speaking, I understand that it is still impossible to squeeze blood from a stone. Not that I've tried, myself, but still....

The lender always has a reason to maximize profits by minimizing charge-offs as compared to interest income. That holds whether the debt is technically dischargeable or not, or secured or not.

We are in the era of management by EIEIO ( Everyone Is Expendable In Organizations). If people with "unstable" incomes should be renters, this begs the question of who precisely has a stable income anymore?

"One of the main reasons we are in a mortgage credit crunch is that two possible models of "recovery" lending--subprime and Alt-A--got used up blowing the bubble. I think it will be a long time before lending standards ease significantly, and I think subprime will come back first."

This a key point. During recovery lenders have always made "subprime" or substandard (as in Alt-A) loans to purchasers of REO as "loans to facilitate." This time lenders made LTF loans too, only this time the loans facilitated the creation of an REO, not the liquidation of one.

The buyers of last resort that normally fuel the recovery have already been tapped. And are tapped out.

Good to have you back Tanta.

@Tanta
I have to disagree with the following:

A lot of folks see the failure of Alt-A as a failure of FICO scores. I don't see it that way. FICO scoring is just an automated and much more consistent way of measuring past credit history than sitting around with a ten-page credit report counting up late payments and calculating balance-to-limit ratios and subtracting for collection accounts and all that tedious stuff underwriters used to do with a pencil and legal pad. I have seen no compelling evidence that FICO scoring is any less reliable than the old-fashioned way of "scoring" credit history.

Rob Dawg said it pretty well, so I am not going to repeat what he said. I will clarify one of his statements though:

It isn't a coincidence that things started going sour at the same time the derived FICO trumped the enumerable source material.

more specifically:

FICO trumped the enumerable source material

Means that the work derived solely from the source is considered of greater value and accuracy than the original data points.

In reality, every derivation always adds noise and inaccuracy into the 'derived' data.

On a side note: doing the long hand work including reserve calculations etc is no longer so tedious now that computers are available.. just stop using them to play games for a bit.

A FICO score can't show:
Debt to Equity ratios,
Debt service costs to disposable income,
ability to handle debts greater than a credit card payment,
why the debt was incurred (which could show poor planning and poor reserves)
While some of that is recorded on a FICO, not all of it is.
FICO relies too heavily on credit card debt and records, which is contrary to identifying someone who can handle mortgage payments reliably. If you have a moderate amount of spare cash at the end of the month, why do you 'need' a credit card? The fact that you have the spare cash would indicate that you might be managing your expenditures so that you don't run out at the end of the month.

In closing off, I am going to explain my scenario:
*Liquid assets approaching 7 digits (I don't count anything less than $1)
*Easily able to pay cash for items over $1000.-- I don't have use the credit card to delay the payment until more cash comes in.
*Have bill payment records in excess of 10 years showing reliable payments.
*Salary over $100k.
*Have had a corporate credit card for over 10 years - any monthly bill is paid immediately and not allowed to 'ride'.
The clinch:
I tried pulling my credit scores, all of the credit agencies go .. 'huh?'.. basically they don't know me because I don't have debt and don't carry debt.

Personally, not being able to pull credit scores on myself doesn't bother me too much because I don't need it.. except for a mortgage. Considering that identity theft is centered around credit scores and my name is almost as common as 'Dave Smith', I am not bothered by the inability to pull a FICO... except again, it is one of the primary determinations for a mortgage, of which FICOs are not that great a predictor of the reliability of that payment. You need to compare apples to apples and oranges to oranges.

With credit cards, you are looking at carrying a debt amounting to between 0 and 30k on depreciating assets that don't have an ability to shelter income from tax.. in other words, non-producing bad debt.

With mortgages, you are looking at carrying a debt amounting to over 300k, sometimes millions on an asset that is a decent inflation hedge and the interest has the ability to shelter income from tax.. in other words a producing, potentially good debt.

People who know how to manage money well, learn to avoid the bad debt and use the 'good' debt to their advantage. Because FICO's overemphasis on credit card debt, it does not pick up the subtle but important distinction.

ucodegen,

You have a lot there, much of which I don't disagree with. However, you made one statement which is basically incorrect.

one of the primary determinations for a mortgage, of which FICOs are not that great a predictor of the reliability of that payment

FICO scores (and there are multiple models and sub-populations which they were specified off of) generally do rank-order credit risk quite well, including mortgage risk. Note that I said "rank order". That means that a 700 will, on average, pay better than a 600. That's regardless of market factors, the general economy, the local economy or anything else. What is less stable is the actual level of defaults and losses that you will see associated with any given score-band. That varies a great deal based on all those exogenous factors.

As an aside, I'm fairly surprised that you lack a score. Even a very limited installment loan history is usually enough to generate a number. The real problems are with "thin" files where the depth of history doesn't give a lot of confidence in the number and with the "under-banked population".

@BG
You have a lot there, much of which I don't disagree with. However, you made one statement which is basically incorrect.

FICO scores (and there are multiple models and sub-populations which they were specified off of) generally do rank-order credit risk quite well, including mortgage risk. Note that I said "rank order". That means that a 700 will, on average, pay better than a 600. link

I'm not going to quote the whole reference.. hopefully the link works.

I strongly disagree. It is down to the nature of the debt as well as the nature of how the FICO score is generated and can be 'gamed' by credit fixing.

A FICO score can be 'fixed' by establishing credit through credit cards, and a payment history through the cards. It is not established through paying bills on time (unless you use the credit card to pay off the bill and then pay off the credit card - basically debt shuffling.). This is contrary to someone who has been able to manage debt well, or completely avoid it. FICO scores show the ability to shuffle debt.. and I'll explain.

FICO worked while the housing ATM existed. This is because the eventual source of funds to cover credit debt tended to be a refi. This has been workable for close to 10 years. The problem now is that the housing ATM is closed and the debt really has to be paid off, not shuffled to the eventual HELOC or refi. Just because house prices went up for nearly 10 years does not mean they will continue. Just because the FICO worked under the house ATM does not mean it is a good predictor of loan success. Analyze why it failed this time..

As an aside, I'm fairly surprised that you lack a score. Even a very limited installment loan history is usually enough to generate a number.

The problem is that it never really gets to the 'installment' state because I can pay it off easily without resorting to carrying it. This is fundamentally the argument I am making. You have to be in a financial state/state of mind that you would use debt that is not beneficial to long term financial success to be able to get a FICO score. This is contrary to someone being able to carry a mortgage. It only relates to how well they can 'juggle the balls'. When real global financial stress occurs.. FICO fails badly - note the 2x increase in prime defaults we are now seeing. The skillful jugglers are starting to fail. What really matters is the underlying financials, which FICO misses. A mortgage loan should really be evaluated much like a corporate bond.. which it is really much like. Both are secured by the underlying assets of the home or corporation. Both generally have first order of succession in bankruptcy. Wage pays the mortgage and corp income pays the bond.

FWIW I'm British and understood "not quite quite" as meaning "definitely not good enough, but we don't want to say so." I wouldn't say it, but Dorothy L. Sayers or some of her characters definitely would. It expresses a deliberate vagueness or an affected disfluency, used to avoid taking possibly-impolite explicit notice of obvious facts.

Tanta writes beautifully.

ucodegen,

We'll have to agree to disagree. I've got data on loan performance in '07 and up-to-date performance on others that show FICO still strongly rank-ordering default risk. (Not mortgage, but much of it secured, FWIW) YMMV.

I do know that two does not make a trend, but when a 'standard' for evaluating breaks, it is worth figuring out why.

It is like in statistics. You can have a correlation that for some reason just breaks periodically. In my opinion, this is the FICO ratings. You can accept that or step back and figure out why it broke. In statistics, this could be that you are actually using dependent variables instead of independent, and the independent variables have not be quantified in the correlation or not all of them are present. FICO is derived from behavior driven by fundamentals. If you look at the fundamentals, you see the real reason the correlation breaks. FICO misses those that would be good mortgage credit risks because they run little debt and misses those that will default because they were using the home ATM. FICO sort of works, until it breaks. When it breaks, it is spectacular because it is missing several of the independent variables and the values for them just changed.

This gets us back to Rob Dawg's comment:
It isn't a coincidence that things started going sour at the same time the derived FICO trumped the enumerable source material.

The enumerable source material is debt equity ratios, wages and how the person has really been managing debt.