*Note – this report was first published as part of the Mortgage Pages research series on 1/30/09

– Credit Scoring Background – Bad Math, Smoke and Mirrors
- Fannie’s & Freddie’s Black Boxes – misused and manipulated for years
- ‘Beat the Computer’. Automated Decisioning – A disaster unfolding
- Buying better credit scores through more debt
- Exotic mortgages sold as a way ‘to improve credit scores’When the Dam Broke
- So, What Next — Will credit scores inhibit a recovery?

Like liquefaction following an earthquake that exposes bad math and engineering sometimes performed decades prior, the quaking of the financial markets has exposed the systemic stranglehold that credit scoring has on all lending. The lender’s, rater’s, and consumer’s universal acceptance and over-reliance upon it was in large part responsible for the ‘lapse’ in lending standards that led to the global credit crisis. On a go-forward basis, credit scoring may inhibit a robust consumer-led credit recovery despite how much money is spent to promote lending.

Finger pointing is rampant these days, and it shouldn’t be too long before some of the blame and inquiry properly flows to credit scoring and its sudden and systemic influence over all lending. Upon this realization, financial institutions will be staring into the abyss. Nothing has changed.

In past mortgage and housing cycles, credit scores were not used as they were during the bubble. Even during the 1991-1994 refi-boom and subsequent housing downturn that lasted until the late 90’s, scores were not used as a primary determinant. It was not until early this decade that credit scores evolved into what they are today to mortgage and housing.

Background

The two most important items found in every consumer loan are obviously a borrower/borrowing entity, and also a credit score. Many loans, such as credit cards, don’t even require collateral and rely solely upon the borrower’s perceived ability to repay derived mostly from a consumer credit score.

Down the road, as the crisis is analyzed to death from those looking to pin the blame on as many names as possible, perp-walks will become common. In this ‘discovery and hang ‘em phase’ it is probable that someone with power will realize that credit scores and their wide acceptance — as a primary metric when analyzing a borrower’s ability to repay — was the paper tiger that made for the start of the lapse in lending standards that has nearly taken down the entire system.

Credit scoring may also be a primary reason for banks unwillingness to lend and will lead to a continued contraction of consumer credit into the future. Because of continued over-reliance upon credit scoring and its aversion to excessive consumer debt, consumers simply don’t qualify for credit easily attainable in the past. It is now a world where those that need the credit can’t get it and those that don’t, can.

Why are the credit raters made to carry all of the blame when credit scorers that provided a primary metric used to make the loans for the securities that the raters rated, getting a free pass? Credit scorers perform their task long before the raters do on a class of loans or bonds. Without credit scoring there would be nothing to rate — or perhaps the loans they rate would have been made using time-tested forms of ability to repay such as income, cash-flow and collateral value

Credit Scoring Background – Bad Math, Smoke and Mirrors

Each of the three primary credit reporting firms, TransUnion, Equifax and Experian, has their own scoring systems. “FICO score,” incidentally, is to “credit score” much like “Xerox” is to “photocopy”. Both are cases in which proprietary corporate creations have evolved, often to the chagrin of the company’s involved, into generic terms at least in the eyes of the general public. Additionally, some banks and credit card issuers utilize their own scores and modeling.

Back in the mid-90′s when credit scoring was still nascent, the bureaus’ formulae weren’t as fine-tuned and it wasn’t uncommon for mortgage underwriters to see a high score on someone with obviously lousy credit (or visa-versa). But such anomalies didn’t much matter to the large lenders and Wall Street investors, whose high-math statisticians were able to “prove” correlations between various credit scores and default rates. “Someone with a 620 score was x% more likely to default than someone with a 680 score”, they said, and y% more likely than someone with a 720 score.

Over time, the perception built that very little mattered other than credit scores. Eventually the banks, Wall Street and the ratings agencies could point to credit scores with their corresponding “proven” default rates and risk-price their loans accordingly. But like so many other things during the bubble years, their black boxes were made of bad math, smoke and mirrors.

Fannie’s & Freddie’s Black-Boxes

Fannie and Freddie’s automated underwriting systems (AUS) and risk-based pricing models, that approved trillions of dollars in loans during the bubble years, most heavily weighted credit scores when decisioning and pricing mortgages. The GSE’s black boxes were thought to be so bullet-proof that until recently everyone assumed that every loan that came out of the GSE’s was ‘Prime’. The GSE’s with their ‘implicit, now ‘effective’, government guarantee were securitized and sold across the world as debt that rivaled US Treasuries and Ginnie Mae securities.

These black-box AUS were misused and manipulated for years. And for years I have been astounded at the quality of loans that the GSE’s would buy at ‘Prime’ prices because of their AUS output that were obviously Subprime loans. Over time, I am confident that defaults across the GSE universe will reach serious double-digit percentages.

Recent, hasty roll-out of foreclosure moratoriums and mortgage modification initiatives may kick the can down the road masking the problem a while longer, but just like we have seen time and again with similar programs…ultimately they will do more harm than good and the problem comes back at a greater pace.

‘Beat the Computer’. Automated Decisioning – A Disaster Unfolding

Everybody bought into it. The GSE’s AUS were engineered to generate loan approvals with reduced documentation requirements when credit scores were sufficiently high. Loans with AUS approvals were instantly salable.

As things evolved, income and asset documentation were no longer necessary for Fannie, Freddie, or, well, anybody. Credit scores were king. Alt-A loan applications – stated income, stated assets, “No Doc” etc. – skyrocketed. Since investors believed they could price out the risk for people with low credit scores as well, the subprime universe exploded literally overnight.

The GSE’s black box mortgage underwriting and decisioning became so detached from human involvement and automated that fraud was easy and rampant. Prior to the bubble years, the mortgage underwriting process consisted of the underwriter carefully reviewing all documentation and making a list of ‘conditions’ that the borrower needed to fulfill in order to make the loan salable to its target investor.

But by 2004, GSE underwriting became a game of ‘beat the computer’. In the GSE’s automated world the underwriter enters the borrower information into the system and an instant decision is given. But during the bubble years if the loan was not approved — or if approved with conditions that the borrower was unable to fulfill — anyone up or down the lender food chain with an AUS log-in and an interest in getting the loan approved could play with the inputs in order to get the decision and list of conditions most easy to fulfill. Things work mostly the same today but access to the systems is more closely monitored.

During the bubble years — when most of the loans that the GSE presently own were originated — If the system denied the loan because it was a borderline cash-out with a high LTV, simply raising the borrower’s asset level by $50k would easily change the declination decision to an approval.

An emailed or faxed document showing $50k in retirement was easy to obtain and would satisfy the underwriter’s condition. Some loans were re-run with new inputs dozens of times in order to get a Fannie/Freddie approval with only the conditions that the borrower or loan officer knew they could fulfill.

GSE loan pricing also became so detached from reality that based upon the AUS approval a borrower with a 580 credit score would get the exact same rate a borrower with a 700 credit score for a similarly structured deal. Appropriate risk was not being priced because it was believed that the almighty GSE credit score reliant AUS systems would never approve the loan, or assign it a lower grade status requiring pricing adjustments, if the loan was indeed risky.

The GSE’s were not the only firms highly reliant on black-box AUS. By 2005, Countrywide, Indymac and Chase also had proprietary system used for non-GSE loans.

In the past couple of years, the GSEs and other lenders have done much work ‘tightening up’ their systems. They have also added lender-specific loan guideline overlays and identified “declining markets” regions in order to get ahead of risk. But still, credit scores remain the primary determinant for AUS approvals and to the extent of documentation requirements.

“Past performance does not guarantee future results” — Buying a better credit score

But what everybody forgot was the oldest caveat in the financial world: “Past performance does not guarantee future results.”

As housing prices soared to historic levels, many home owners used their homes as an ATM machine, cashing out periodically via new first and second mortgages to pay off their credit cards and buy SUVs, plasma TVs, electronic gadgets, and (most notably) second homes and investment properties — which further helped fuel the upward spiral in real estate.

People sufficiently savvy to reap the tax benefits by paying off their installment and revolving debts with cash out from their home actually saw their credit scores improve. This was despite the fact that many of them were living well beyond their means. Once their installment and revolving debt was paid off, they went back to using it. Within a year or two following the funding of the original 50% debt-to-income ratio mortgage, the borrowers total debt ratio was 70-80% and future ability to repay completely dependent upon house price appreciation and home equity extraction. House price appreciation and equity extraction was a second job for millions from 2003-2007.

Indeed, over the past half-dozen years or so, consumers learned they could literally ‘buy’ improved credit scores through taking on more mortgage debt. Subprime loans – especially the Pay Option ARMs which had artificially low interest rates and exceptionally low payments for two years or less – were openly sold as a way to generate a strong mortgage payment history. Once established, this would improve a borrower’s credit score allowing them to refi into an A-paper loan at better rates before the payments increased.

When the Dam Broke

Once the real estate market peaked, however, borrowers were suddenly unable to refinance out of their financial predicaments. They were over-leveraged, both with respect to LTV/CLTV parameters (the value of their home was/is worth less than what they owe, sometimes substantially) and also vis-‘a-vis their debt-to-income ratios. Many even overstated their income on purpose in order to get in the game making for an incurable situation.

So we have what we have today – a snowballing decline in values and nothing less than absolute chaos and carnage in the mortgage and housing markets. The defaults and foreclosures may have started with the low-credit score subprime borrowers, but they have made their way into the higher grades such as Alt-A, Jumbo Prime and Prime borrowers.

So, What Next — Will Credit Scores Inhibit a Recovery?

To what extent is credit scoring are at all predictive? Have the credit scoring firms adequately tweaked and re-weighted the multiple criteria that go into one’s credit score? And if so, have they done so accurately? At this stage when we are learning more each week especially on the psychology of it all, how could they know? By overlooking the role that the credit reporting agencies played in all of this, perhaps even new vintage loans made in the last year during a period of unmatched house price depreciation are also destined to fail in large numbers.

The large banks and investors remain slow on the uptake. Although income, occupancy and LTV carry the most weight in “absolute” terms, loan approvals and especially loan pricing is still very much credit score driven.

The GSE’s now have multiple credit score/pricing tranches, up from one or two back in 2007. In addition, many banks have applied overlays to the underwriting guidelines prohibiting borrowers with certain scores from getting a loan even if the investor will accept lower scores. This will prohibit many with great income and a perfect record — other than having a few credit cards proactively shut down by the bank — from buying a new home for years, further delaying the recovery.

In the past year, I have witnessed perfect credit risks denied a loan solely because of credit score. In the new-era housing market with values down to affordable levels in some areas around the nation, one could argue that many with stable jobs, good cash flow and reserves — that recently walked away from an exotic loan 50% underwater — are much better credit risks that the scores would imply because they are finally de-levered. Being backward looking will not help to solve the housing crisis.

How can the scorers model credit risk accurately enough to reflect the fundamental shift that occurred during the bubble when a home became the ‘largest investment’ of one’s life vs. ‘a place to life’? With prices off 50% at the median in the bubble states and underwriting guidelines back in check for the most part, the home has again become a place to life. Arguably new vintage borrowers will act much differently with respect to real estate as an asset class than bubble-years borrowers.

How does the shift in behavior with respect to negative-equity, or the fact that being in default or foreclosure does not carry the stigma that I call ‘The Scarlet F’ that it once did, undermine the credit scorers attempt to be predictive?

All of these questions and more make for serious doubts about a consumer-led recovery, an end to the housing crisis and the quality of present vintage lending. By default, credit scores will continue to influence the lender’s contraction or future expansion. But what if the computers are still wrong? They do not have a good track record over the past decade. The future of lending depends upon getting back to basics or lending will continue to contract and brand new problems will arise that nobody can predict.

Best Regards,
Mark Hanson

This document is for your private information only. In publishing research, Mark Hanson and M Hanson Advisors are not soliciting any action based upon it. Mark Hanson and M Hanson Advisors publications contain material based upon publicly available information, obtained from sources that we consider reliable. However, Mark Hanson and M Hanson Advisors does not represent that it is accurate and it should not be relied on as such. Opinions expressed are current opinions as of the date appearing on Mark Hanson and M Hanson Advisors publications only. Mark Hanson and M Hanson Advisors are not liable for any loss or damage resulting from the use of its product. Mark Hanson and M Hanson Advisors are Limited Liability Corp registered in CA.

Print Friendly

Filed on: September, 1 2009 by Mark

Comments are closed.