The National Association of Realtors® recently called on FICO, developer of the credit scoring model of the same name, to revise some of its computations.  At issue is the alleged negative impact on consumers when banks and credit card companies unilaterally reduce or cancel credit lines, even when consumers are in good standing on those accounts.  An NAR spokesman said that FICO's software is "unable to differentiate between innocent victims and people whose behavior genuinely merits reductions," and called the FICO model "archaic."

Craig Watts, FICO's Public Affairs Director, told us that FICO's scoring models are not static.  "They are constantly evolving, but each change reflects data over time, and users do not adopt new models immediately because of the cost and labor involved."  Its last widely publicized change was in the spring of 2008.  At that time FICO predicted that its new scoring system would help lenders reduce the default rates on consumer loans between 5 and 15 percent while actually raising the scores of some consumers. 

Utilization was not among the areas involved in those changes.  The most sweeping tweak was a response to a perfectly legal but potentially devious credit repair scheme called "piggy backing."  An account holder can name another party as an "authorized user" on a credit card and it has long been a strategy for parents to put a child's name on an account which automatically transferred the parent's credit history on that account to the child, giving him a leg up in establishing his own credit.  Wives often access their husband's accounts as authorized users as well, rather than becoming co-borrowers or joint owners of the account.

However, FICO found that credit repair agencies were soliciting persons with high credit scores to sponsor total strangers as authorized users; in effect, paying these sponsors to rent out their credit rating.  The users gained a credit history but not the ability to actually use the card or any access to the sponsor's personal information. After a few months the piggy backer was removed from the card and the sponsor's credit could be rented out to another credit repair customer.  There were stories of sponsors who made as much as $10,000 per month through such programs.  Privacy and credit laws made it difficult for a lender to get personal information about the authorized user so FICO, initially changed its scoring model to eliminate authorized user accounts from scoring, doing away with the automatic boost in the credit score of the piggy-backing consumer. Watts said that this change not only caused problems for honest authorized users but also for the banks so FICO subsequently made an additional proprietary change that protected scores from the dishonest users.   

FICO also changed the way it treated delinquencies.  An individual who had been penalized for a serious delinquency on a single major account while maintaining his other accounts in good standing was likely to see his score increase under the new rules.  However, one who demonstrated a pattern of late payments on multiple accounts probably saw his score go down.  At the time the change was announced FICO said this change could alter an individual's credit score by 20 to 25 points - in one direction or the other.

FICO scores are calculated by weighting several factors from an individual's credit report.  The payment history, which is the main focus of a credit report, accounts for only 35 percent of a FICO score. Given almost equal weight (30 percent) is the utilization of credit.  This is the percentage of credit an individual has used relative to the amount that is available.  A consumer with $10,000 in open lines of credit but only $2000 in debt will score higher in this category than one who has used $9,000 of the available credit.

The length of time an individual has had credit counts for 15 percent of the total score.  A consumer who opened his first department store charge account in 1985 will score higher than one who first got credit in 2009.

The last two categories, each of which count for 10 percent of the score, is a measure of the diversity of the types of credit in an individual's portfolio, and what Watts called "the individuals interest in new credit.

But back to the NAR's complaint.  The Association wants FICO to either totally disregard the utilization rate for consumers who had a reduction or tabulate the score as if the reduction had not taken place.

Watts said that FICO had discussed the issue with NAR and pointed me to a statement from Joanne Gaskin, Fair Isaac's director of mortgage scoring solutions as quoted by Ken Harney in the Washington Post.  "Gaskin, said the FICO model attaches such importance to consumers' available credit and utilization rates - they account for 30 percent of the score - because they are highly accurate predictors of future credit problems."  Gaskin said that "'Research conducted by Fair Isaac last year found that consumers who utilize 70 percent of their available credit "have a future bad rate 20 to 50 times greater than consumers with lower utilizations". Ignoring this key indicator, the study said, would "decrease [the score's] predictive power."

FICO also conducted a study late last year on the subject of credit line reductions.  Their analysis shows that approximately 14 percent of the U.S. consumer credit population experienced a reduction in total revolving credit between April 2009 and October 2009.  Four percent had a risk trigger such as a late payment while 10 percent did not.  FICO found that among the latter group, the affected consumers already tended to be very low-risk with a median FICO score of 757 and low balances and credit utilization ratios.  As a result, small reductions in their total revolving credit lines had a minimal effect on their FICO scores.  The average reduction was $4,800; about 12 percent of the average revolving credit available to this population, and the reduction resulted in average change in utilization from 24 percent to 27 percent between April and October 2009.

FICO said it observed very limited impact on scores for the no-risk triggers group.  Their median FICO scores in fact increased slightly (from 755 to 757) during the period.