The Most Commonly Used Four-Letter Word in the Mortgage Business: FICODave SullivanFair Isaac Company, credit score, FICOAlmost every mortgage professional will tell you-as four letter words go, FICO is a pretty bad
one. What does FICO mean anyway? FICO is a score that was developed by Fair Isaac
Company, a California-based company specializing in the construction of statistical scoring
models. This score is a single number ranging from 350 to 900, and is calculated using past
credit history data. Fair Isaac developed this scoring model using millions of actual consumer
credit data files to develop a complex and secret mathematical algorithm. And since FICO is the
developer and owner of the model, it is proprietary. Consequently, it is impossible to know
exactly all the credit variables considered and how they are weighed by FICO.
Each Bureau calls its FICO score something different (Equifax-Beacon, Trans Union-Empirica,
Experian-Fair Isaac), but the score it represents is the same no matter what bureau is used. The
risk is defined in the number of accounts that default based on the score. Some example ranges
++Scores below 601 yield 8 good loans for each bad one
++Scores from 700-729 yield 129 good loans for each bad one
++Scores at or above 800 yield 1,292 good loans for each bad one
Overall, the scores that are generated are a good predictor of a bad loan. They should be used to
alert the lender that more investigation is needed, rather than using it to deny a loan. After all,
you do pay big bucks to verify, update, correct, and supplement your borrowers' credit.
The FICO score supplied is obtained from the bureaus and based solely on the "raw credit" as
delivered by Trans Union, Experian and Equifax. What if your borrowers have a number of
derogatory items on their credit that were either reported in error or have been recently settled? It
means you will most likely receive a pretty clean RMCR, but the score will be unjustly low
because it will continue to reflect the original inaccurate raw credit. The bottom line is, without
the rights to the scoring matrix, we have no control over the final score.
What we do know is that the model contains 33 variables that were found in combination to be
most predictive of an individual's future ability to repay a loan. These variables are grouped by
FICO into five categories:
++Previous credit performance
++Current level of debt
++Amount of time that the credit has been in use
++Pursuit of new credit
++Types of credit available
FICO's model evaluates how these variables interrelate and, supposedly, how the interaction will
likely affect mortgage performance. By evaluating the historical performance of loans with
borrowers' credit characteristics, FICO claims that they can tell us something about the future
performance of loans with similar credit characteristics (i.e. the risk of extending credit).
Accompanying every score are four "reason codes." These codes identify 4 of the 33 variables
that have the most influence on lowering the credit score. This gives us some insight into the
credit score. Examples of reason codes are:
++Too few accounts paid as agreed
++Too many inquiries in the last 12 months
++Proportion of balances to credit limit is too high on revolving accounts
(Note: three out of these four reason codes don't even apply to the mortgage industry.)
How can I improve my borrower's credit score now?
The best way to keep the score high is to pay bills on time, avoid collections (i.e. medical
collection) and bankruptcies. If the score is already low, the following is a quick checklist of
things that can be done to improve it.
++Pay down any credit card balance to below 50% of the high credit
++Close all accounts that have ever been late
++Close all but two to three revolving accounts
++Limit the number of inquires made on the file
++Pay all collection accounts
++Write a dispute letter to the repositories on any account that might not be accurate. Changing
the repository file can take time, but it's the only way to raise the score.
The ultimate end result of FICO's modeling is, as I mentioned earlier, a score of 350-900. A
score of zero (0) indicates either that no information was found, or that there was simply not
enough credit history available to generate a score.
Now that you know as much about FICO as I do, you can answer the most popular question: Is
this FICO modeling tool working for us, and is it applicable to our industry? For more
information on FICO, visit www.equifax.com, www.tuc.com, or www.experian.com on the
Dave Sullivan is President of AIR Credit Reporting Midwest. He can be reached at (877) 247-1003 ext: 220.