Do FICO scores really mean anything? Were lenders who relied on these traditional credit reports misguided?

Just as Moody’s and Standard & Poor’s are on the griddle because of their failure to spot trouble in the mortgage-backed securities market, so is Fair Isaac Corp. (FIC $24) feeling some heat for the inability of its credit score to predict defaults. Although the company maintains that its scores fulfilled expectations, some in the mortgage industry say the scores proved unreliable.

There are now suggestions of upheaval in the credit landscape: the growing presence of new rival VantageScore, rumored price reductions for FICO scores, and a revamp in Fair Isaac’s mystery model, which seems to suggest that the product is indeed flawed. The questions circulating about the company’s iconic product have not helped the stock price, which is off 36% through the past year.

There has long been criticism of FICO scores from those who claim they are unfair and often simply wrong. A 2004 report from the Federation of State Public Interest Research Groups found that 79% of all credit reports contained errors. The ability of consumers to manipulate their scores is the stuff of water cooler conversations, not to mention ads like one I received this morning from an outfit suggesting it could boost my score by as much as 50%. In fact, credit reports can be influenced by such things as getting a credit card company to erase a late payment as payback for being a good customer, opening a new credit card account, or by paying off a loan.

That the scores are susceptible to management argues against their reliability. It’s an important issue, because FICO scores can determine how much someone is paying for a mortgage or for another sort of loan. It became even more significant when subprime mortgage lenders began to rely almost exclusively on credit scores in making the risky loans that ultimately proved disastrous. As they comb through the rubble of the housing industry, some lenders are concluding that FICO scores were not very helpful. Last fall, Glenn Costello at Fitch charged that lenders’ overdependence on FICO scores were a major contributor to the unexpectedly high default rates on vintage 2006 loans.

FICO scores, as almost everyone knows by now, range between 300 and 850, with the median score in America being 723. Anything below 620 is considered subprime. The higher the score, the less one pays for a mortgage. On Fair Isaac’s Web site, potential borrowers discover that if their FICO score is between 620 and 659, their rate on a $300,000 30-year fixed mortgage would be 7.129%, for a monthly payment of $2,022. Those whose scores fall in the next-highest category, 660-699, would pay 6.319%, or $1,861 a month — almost $2,000 a year less. That’s a lot of cheeseburgers.

Most industry analysts would say that mortgage lenders should not have relied exclusively on FICO scores, but during the heady real estate boom many in fact did. The vice president of global scoring for Fair Isaac, Tom Quinn, says the scores are not meant to estimate how much a lender might lose, but rather ranks the risk of default. In other words, his company’s job is to say that one borrower is more likely than another to pay off a loan. “What we are not doing,” he says, “is creating the probability of default for a given score. The model is not designed to give the exact odds.” In fact, though, the company used to promise just that, industry sources say. “Historically, the company was a very aggressive marketer,” one industry insider says. “They may have oversold the product in the past.”

Indeed, VantageScore, which was developed by the three main credit bureaus and introduced in 2006, does claim a predictive capability. CEO Barrett Burns says the VantageScore product, which is currently most often used to prescreen customers for credit cards, will predict the likelihood that a consumer will become 90 days delinquent within a two-year time frame. He says his company’s score does not predict losses. He also says that “there isn’t anyone who isn’t rethinking their model.”

Mr. Quinn maintains that all through the mortgage cycle, those with higher scores proved less likely to default than those with lower scores, even in the subprime space.

Nonetheless, the company has recently launched FICO 08, which incorporates some tweaking of the model. “It is more reflective of the new consumer,” Mr. Quinn says. He maintains that the model update is a normal revamp, but for sure the new product reflects some shortcomings in the program that has been essentially unchanged since 2000. In fact, in a presentation to the Mortgage Bankers Association, Fair Isaac management made the case that FICO 08 had significantly greater predictive capability than the former version, especially for subprime borrowers.

As it turns out, the real villain causing the astounding surge in default rates — and the best predictive agent — has been high loan-to-value ratios. On first-lien pools of mortgages, the LTVs moved steadily higher throughout this decade, according to Fitch research. In 2000, homeowners borrowed on average 77% of the value of their property. This figure rose to 81.3% by 2003 and 86.2% by 2006. During that period, FICO scores actually climbed for borrowers, to 625 on average in 2006 from 595 in 2000. The rise in defaults, in other words, has been in spite of increasing FICO scores.

Heaven knows there were plenty of reasons that the mortgage industry came unglued. Poor judgment across the board comes to mind. But perhaps the most significant failure was that which led to the rise in loan-to-value lending: the assumption that home prices would go up forever. As we survey the investment landscape over time, it seems that expecting trends to continue has always been at the heart of the worst disasters. Oil traders take note.

peek10021@aol.com