Is FICO still relevant to the mortgage industry?
I can see it now: no matter how this question gets answered, it’s certain to elicit some response, which in fact, is the reason for the article.
As an industry that migrated with some reservation but eventually went "all-in" with the belief that FICO was a true indicator of performance, it’s a question that deserves discussion, particularly given the issues we’re facing.
Let me start by saying that I don’t believe it’s a simple yes or no answer. No, I’m not running for office and attempting to answer this like a typical politician; I’m conflicted because there are some strong arguments from both sides of the aisle.
It’s been 12 years since Fannie and Freddie began requiring FICO scores on every loan they purchased. Interestingly, the subprime industry took much longer to completely wrap itself around the use of credit scoring. Some of the early subprime pioneers, guys like Vince DiMare at Equity Secured Investments, were skeptical of FICO, and for good reason. As he mentioned to me on numerous occasions, “why do I need a score to tell me what is an acceptable level of risk, when I already know how to underwrite these loans.” No truer words have ever been spoken.
Having worked for GMAC Residential Funding Corporation in the late 90’s, I saw enough performance reports on billions of dollars in loans to become convinced that FICO was a reliable indicator. But even given the volume of data, RFC took years until it moved away from a traditional subprime underwriting methodology to one that was FICO based. I remember a three-year span in which the company had two underwriting manuals for subprime, one for the traditional method and one for the FICO approach. RFC was reluctant to pull the trigger on FICO because even with all the performance reports to support its use, the old-line risk guys weren’t completely bought in.
The erosion in loan performance we’re seeing is not a fault of a poor scoring model, but an industry that forgot it was not an absolute.
Even when RFC had two underwriting manuals, they were still very similar in structure. While the traditional method didn’t pay attention to credit score, both manuals still understood the importance of how all the other credit factors fit into the picture – down payment, performing trade lines, etc. Whether FICO was part of the picture or not, the loan still needed to make sense at every other level and that’s where the industry went askew. It may be the most overused term in the business, but common sense underwriting meant putting borrowers into loans they could afford. FICO wasn’t needed to tell us that.
However, when all of the other credit factors were held constant, FICO was dead-nuts on, and that’s perhaps the most critical part of this discussion. The failure was on the part of the industry taking FICO as gospel, and forgetting that it was still necessary to underwite the file as if we really were mortgage bankers.
Ironically, Equicredit, the former subrime division of Bank of America, experienced six years ago what the rest of the industry is now going through. When I opened my subprime company in 2000, they were the most FICO driven company around. Putting all of their eggs into the FICO basket and ignoring the fundamentals of underwriting meant their loan performance tanked. What’s interesting is that their performance stunk at a time when property values were rising and interest rates were dropping. If this strategy fails under optimal circumstances, what makes anybody think it will work under abysmal conditions like we’re seeing today?
Earlier I wrote about a 2007 Alt-A pool of loans
from Washington Mutual that is performing horribly – 15 percent foreclosure rate with 8 months of seasoning. I was taken aback by how a pool of 705 loans could deteriorate so quickly.
But I was reminded by readers that I had forgotten the very things I wrote about in Greed, Fraud & Ignorance: A Subprime Insider’s Look at the Mortgage Collapse
, the same things I’ve written about here. Somehow, I diluted myself into thinking that scores in this range couldn’t perform this poorly (and so quickly). But the loan characteristics were indicative of just how far we’ve fallen as an industry. Most of the loans were stated income (90%), CLTVs north of 90%, likely closer to 100% but we can’t tell for certain, and mostly pay option ARMs with of course, super-low teaser rates. How many were investor loans and were they really stated income loans or something akin to NINAs? I don’t know but I’ve got a strong suspicion this played into it.
Was it Fair Isaac’s responsibility to modify the scoring model when the industry began to lose control of its faculties? I don’t think so. Yes, Fair Isaac touted the system to the fullest, make no mistake, but I don’t believe it was designed to ever be the sole determining factor for a mortgage loan. If loans that were not considered a good risk in 2000 were acceptable 4 years later, how is FICO to account for operator error.
To be certain, the industry found ways to game the system. Between credit repair companies and a multitude of techniques (from dropping borrowers from the application in order to write the co-borrower with the higher credit score under a stated income loan) to make deals look better than they were, there was no shortage of methods to cheat the process. When an industry forgets the basic tenets under which it operates and the notion of effective risk management becomes non-existent, it seems only logical that FICO would fail us.
The other issue to consider is whether FICO should have considered market volatility. I know that some people will disagree, but much like I believe the rating agencies should have modeled their systems to account for drastic shifts in the market, so should FICO. The powers to be at Fair Isaac will argue that market volatility was never part of their equation (which is true), but should it be? FICO has largely existed in a stable market and has never had to operate within the largest real estate bubble in our history. If you buy into the notion that a credit score of “x”, whatever it might be, will not perform as well in an environment where homes prices are rapidly deteriorating and the economy is heading towards what may be the worst downturn in modern history, then maybe it’s underlying methodology needs to be examined and overhauled. FICO 08 will be new and improved but from what I’ve read it doesn’t seem as though changes are being implemented to factor in for the changing economic landscape. By the way, I’m still not throughly convinced of my own argument here, as modeling has never been forte.
So is FICO still relevant to the mortgage industry? I think it all depends on whether the industry and the securitization process function as they are supposed to. If the rating agencies have a vested interest in the loans they rate (e.g. they rate them with some level of competency so investors all over the world don’t buy the bonds believing they’re something that they’re not) and the rest of us remember the basic fundamentals we learned in Underwriting 101, then yes, FICO is still relevant. If not . . . well, maybe it’s time to fire up the snow cone machine. After all summer is just around the corner.
Note: Richard Bitner is the author of Greed, Fraud and Ignorance: A Subprime Insider's Look at the Mortgage Collapse. As a 14-year veteran of the mortgage industry, he spent five years as the President of Kellner Mortgage Investments, a subprime mortgage company. In addition, he was a Director for GMAC Residential Funding and the National Training Manager for GE Capital Mortgage Insurance (Genworth Financial).