Credit Scores and Credit Reports, by Evan Hendricks

Chapter 2

Credit Scoring — Advanced

You can observe a lot by watching.           

- Yogi Berra  

Fair Isaac Corporation's explanation is a good starting point, and for some, it will suffice. But of course, there's more to the system - much more.

Fair Isaac actually has several scoring systems or models. The most widely used is the "classic risk model." Equifax's version of classic is called "Beacon." Formerly called "Empirica," Trans Union's version in 2004 was renamed "TU FICO Classic." Experian's version is known as the "Experian/Fair Isaac Risk Model."

Classic is used by the majority of mortgage lenders. It is also the version you buy when you go to www.myfico.com. More than 70% of the 100 largest financial institutions use FICO scores to make billions of credit decisions each year, including more than 75 percent of mortgage loan originations, according to Fair Isaac.

In addition, there are variations of the model that are tailored to different kinds of credit: auto, bank card, personal finance and installment loans. Some mortgage lenders and credit card issuers have developed their own credit scoring models, which they use alongside of the FICO model when evaluating an application.

Classic was first developed in 1989, but its use increased sharply in 1995 when the two major mortgage underwriters, Fannie Mae and Freddie Mac, endorsed them for use by mortgage lenders. In 2001, Fair Isaac introduced its more advanced "Next Generation," or "NextGen," scoring model. NextGen is supposed to be more precise and more reliable, but it has not yet been widely adopted, in part because Fannie Mae and Freddie Mac have not endorsed it.²⁰

How It Works: 'Variables' and 'Score Cards'

In creating its Classic model, Fair Isaac identified about three hundred possible characteristics on consumer credit reports that might be predictive of future credit risk, according to Craig Watts, Fair Isaac's Consumer Affairs Manager.²¹ These included combinations of data types, such as Factor A divided by Factor B. Fair Isaac score developers then winnowed that list down to a couple dozen of the most predictive characteristics, which were then incorporated into the Classic FICO scoring model. These two-dozen or so characteristics are listed in Chapter 1 and can be found at the MyFico Web site.²²

Fair Isaac's Classic uses 10 "Scorecards" to calculate the credit score. The scorecards find key attributes in the consumer's credit report and then groups the consumer with other consumers who have the same key attributes. Watts said the scorecard approach greatly improves the predict-ability of the scores.

²⁰ Equifax's trademarked version of "NextGen" is called "Pinnacle;" Trans Union's is "Precision" and Experian's is the "Experian/Fair Isaac Advanced Risk Score."
²¹ Craig Watts and Karlene Bowen consistently and helpfully responded to a series of questions from the author.
²² http://www.myfico.com/myfico/CreditCentral/ScoreConsiders.asp

"For one thing, consumers are compared with like-consumers so the model can be sensitive to subtler distinctions between them. Here's how it works. The FICO formula first checks to see which of the 10 consumer groups your credit report best matches. Let's say you have a serious delinquency on your report. You will be put into a group with other consumers who also have at least one serious delinquency on their reports, and your FICO score will be calculated using a scorecard customized to the credit risk features found in that group," he said.

Fair Isaac said the NextGen model is more precise because it breaks the consumer population into 18 groups or scorecards. As with classic FICO scores, the groups are based only on the credit history data from their credit reports.

Impact On Mortgage, Refinance Rates

Although actual use varies greatly among lenders, for some mortgages and refinancing, the cut-off line for your interest rate can come at 20-point intervals. Again, depending on the lenders or refinanciers, the FICO Classic score qualifying you for the best rate can range from 680-760. Remember our first "general rule," the lower the score, the more you pay. Anything below a 620 is considered "sub-prime," which means the borrower is considered "very high risk" and must pay much, much higher interest rates.

Fair Isaac has its roots in risk assessment. The first purpose of the FICO score is to show how likely you are to become at least 90 days late in making payments in the next 24 months based upon patterns in your credit history, compared with patterns of millions of past consumers.

The Importance Of Being Recent

Thus, Fair Isaac scoring models place a great deal of weight on how recently you had a credit problem. As the following chart illustrates, in a proportional sense, a major delinquency in the past year has a 93% negative impact, while a major delinquency between 1-2 years-old has about a 60% negative impact; a major delinquency between 2-3 years-old has a 44% negative impact; a 3-4 year old delinquency has a 33% impact; any delinquency older than 4 years has only a 22% negative impact. (See Fair Isaac slide.)

In fact, under the FICO scoring system, sometimes how recently the event occurred is more important than the event itself. This is true for all derogatory items, from a 30-day late payment to a collection or judgment. A 30-day late payment from last month is going to reduce your score significantly while a judgment from 6 years ago, while still reducing the score somewhat, is insignificant. The amount of money involved in the event is not always significant. While the amounts of delinquent balances are important to a score, the "recency" of the event can be more important. So even if it is only for $25, if you have a collection or public record in the past few months, it does serious damage to your credit score. In other words, a $40 balance on an account that is currently 60 days late, in some cases can do more damage to your credit then a $3,000 collection account that appeared on your credit report four years ago. Delinquencies are also scored according to "severity." Is the delinquency a "first-ever" event or the latest in a series? Is it 30-days or 90-days?

Another little known fact is that if you pay off an old collection account, it "updates" to your credit report and once again becomes a "recent" event that is highly damaging to your credit score. That is, a collection from the year 2000 that is paid off in February 2004 is re-scored in March 2004 by FICO as a very recent, highly damaging entry. Thus, while it's best to pay off very recent collections, strictly from a credit-scoring perspective, it might be better not to pay off an old collection and to let it drop off the credit report at the seven-year limit. Fair Isaac has disagreed sharply that paying an old collection constitutes recent activity on a delinquent account.²³

Unpaid Medical Bills & Parking Tickets

This is why some knowledgeable insiders feel that minor medical collections can unfairly drag down credit scores and undeservedly make people look like greater credit risks than they really are. Moreover, it's not uncommon for medical bills to go unpaid and be sent to collection agencies because of mistakes on the part of health insurers or HMOs. As a consumer, it is important to understand that a minor medical collection, or any collection or default judgment, can seriously damage a credit score. This is significant because researchers are learning that medical bills may account for more than 50 percent of collections.

²³ There appears to be disagreement between Fair Isaac and some credit bureaus on this issue.

Whether it is a service bill or mail order product, it is not uncommon for consumers who feel they did not receive the service or product that they ordered, or who feel they simply are not responsible for a debt, to refuse on principle to pay the debt. Some people who think they were wrongly given a $20 parking ticket, refuse to pay, and move out of the jurisdiction where they were ticketed. They might think that one minor item will not damage their credit that much.

But many collectors and creditors already know that it is damaging, and will use the threat of reporting to the credit bureaus as a means of pressuring consumers into paying debts. For many major forms of credit, like a mortgage, a consumer must resolve all outstanding collections and negative public records before a loan will be approved. Accordingly, creditors view credit reporting as an arm of debt collection - a sort of last resort that will catch up with non-paying consumers sooner or later. This practice "crosses the line" when creditors and collectors threaten to report debts - or actually report debts - that they know or should know are not the responsibility of the consumer.

Heavy Credit Card Use

Another common misconception about credit scoring is that paying off your credit card balances in full every month guarantees a top-notch credit score. If fact, if the balance is high on the day of the month that your credit card company reports to the credit bureaus, that will be the amount reported. The fact that the balance is paid off with each month's billing statement and returned to "zero" might not be reflected on your credit report. Many self-employed or small business people, or those who make many business- or profession-related purchases on their credit cards and then pay off the balances each month, fit this profile.

Say for instance, that by the end of the month, you make $7,000 worth of purchases using a credit card with a credit limit of $8,000. In the middle of the month, at the time of the due date, you pay off the entire balance that appeared on your last statement.

But from a credit reporting perspective, you aren't paying off the entire balance: by the time your payment arrives at the credit card company, your true balance has changed due to incoming charges. While you may pay the full amount shown on your statement, the account never reaches a true zero balance, unless you cease charging for a month. Since the credit card companies choose a particular day of each month to report to the bureaus, whatever the balance on that day is the one that goes on your credit report. Your credit score may remain unchanged if your purchases are consistent month to month, or the score may drop if the balance is higher, or rise if that balance is lower.

So, if you know the date that your credit card company reports to the bureaus, and you want to raise your credit score at least temporarily, then paying off your balance a few days before that date could achieve that goal.

The Fair Isaac slide below charts the impact on credit scores of the balance-to-credit limit ratio:

This is but one example of how someone who in reality handles their credit very responsibly is in fact penalized by the credit scoring system, in part because of the timing or reporting cycles and the current inability of the system to accommodate people who regularly pay off high balances, or at least dramatically reduce them. This also why Fair Isaac advises consumers to keep their card balances low.

This comes under "Factor 2: Amount Owed - Extent of Indebtedness" that we discussed in Chapter 1. The categories within Factor 2 are "Ratio of credit balance to credit limit," "The amount owed on all accounts," and "the amount owed on each type of account." In other words, you might think of yourself as a responsible user of credit, but your credit score is taking a hit in all three of these categories.

Store Cards & Instant Discounts

Credit scores can also drop when a consumer accepts or signs up for a department store charge card and then uses the card to take advantage of an instant discount. Such charge cards usually start with very low credit limits, so any purchases likely will result in a high ratio of "credit balance to credit limit." Next, your score drops because you have opened up a new line of credit and then because of the department store's inquiry to your credit report. Also, the credit scoring system looks less favorably on store charge cards, which give you limited buying potential, than it does on credit cards from major banks, which give you much broader buying potential. Again, your score can take a hit from three different directions.

The department store scenario can have an unforeseen impact when the consumer receives preliminary approval for a mortgage application and then to celebrate, goes shopping. The subsequent drop in credit score forces the mortgage lender to withdraw its approval or make another offer at a less favorable rate.

But, Don't 'Close' Your Unused Credit Cards

In addition to the balance-to-credit limit calculation made on each individual account, the scoring module also makes a separate calculation using all of your revolving accounts. This means closing unused or little-used credit cards can cause your score to drop because it raises, unfavorably, that overall ratio between the balances on all your revolving accounts against the limits on all of your open revolving accounts. It can also have a negative impact because it can shorten the length of time you have credit. Also, the FICO model looks more favorably on major credit cards, like a Visa or MasterCard from a major issuer, or an American Express or Discover Card. So if you feel compelled to close some accounts, close the department store or specialty cards, and the cards with the lowest credit limits.²⁴ Certainly one of the ironies is that the FICO model penalizes you for not having credit cards, but also penalizes you if you use them too much, or in the "wrong" way - as the scoring model defines it.

Inquiries

The impact of inquiries on credit scores has evolved over time, as FICO has made important adjustments (this was mentioned in the previous chapter). Under FICO classic, any mortgage- or auto-related inquiries in the most recent 30 days are ignored (and not scored). Classic will also only count as one inquiry all mortgage-related and auto-related inquiries within any 14-day period that occurred more than 30 days prior or less than one year. Under Classic, then, inquiries would lower your score slightly if they showed you applied for mortgage 32 days and 48 days prior to the score being calculated, because those inquiries were more than 14 days apart, and more than 30 days ago. However, under the NextGen model, the 14-day period is extended to a 45-day period, giving consumers more time to rate-shop without being penalized.

²⁴ Singletary, Michelle, "Play Defense In the Credit Game," Washington Post, Jan. 25, 2004; pg. F1

'Classic' vs. 'NextGen'

A more general and ongoing problem relates to the fact that although Fair Isaac over the years has regularly updated its credit scoring models, many lenders continue to use the older versions. In fact, the latest model, dubbed "Next Generation," was rolled out in 2001. But as Kenneth Harney of the Washington Post reported, it has not been adopted by most lenders and investors²⁵. Fair Isaac recognized the problem but said it could not require any creditor to use a specific version of its software. There were substantial computer system costs to integrating new FICO models into lenders' underwriting programs, "and we are very sensitive to that," Fair Isaac's Craig Watts told the Washington Post.²⁶

Variations between the older and newer FICO models can lead to significant differences in credit scores. In 2002, the National Association of Mortgage Brokers issued a warning to its 13,000 members. Ginny Ferguson, former Secretary of the association and chair of the credit-scoring committee, told the Washington Post that "when lenders use [outdated FICO models] for pricing of loans, we can run into major problems."

It's not just Classic vs. Next Gen - there can be significant differences within Classic itself. For example, instead of incorporating changes made to Classic by FICO into one version, Equifax released the changes as a newer version. At one point you could sign up for the original "Beacon," or its successor, "Beacon Enhanced," or the third generation "Beacon 96."

²⁵ Kenneth Harney, "Outdated Credit-Scoring System Can Penalize Potential Borrowers," Feb. 23, 2002, pg H01, www.washingtonpost.com/wp-dyn/articles/A53198-2002Feb22.html
²⁶ id.

Currently, Beacon 5.0 is the most commonly used, as Beacon 6.0 has not yet caught on. The FICO models have evolved regularly for various reasons. In the mid-1990s, the FICO model would penalize applicants for having "finance company" loans on their credit histories, or for multiple "inquiries" resulting from a consumer shopping for the best mortgage or auto loan rate. In some cases, the older the version, the lower the credit score.

A potential conflict arises when a lower credit score results in a higher mortgage rate, which in turn generates higher fees for loan officers, and higher rates for the lender when it sells the loan on the secondary market.

Richard Le Febvre, who for years ran AAA American Credit Bureau Inc. in Flagstaff, Ariz., said he often saw mortgage applicants get hurt by older FICO scoring.

A mortgage broker would order a score on an applicant through a small credit bureau like Le Febvre's, which used an updated FICO model. But then the lender to whom the broker intended to sell the loan ran the same applicant through an old model, and the score came out lower. The lender's lower score dictated what the borrower paid, Le Febvre said.²⁷

Sometimes it works the other way. An online lender had received more applicants than it could handle. So the online lender sold the names of those consumers with qualifying FICO scores to a second mortgage lender. However, the online lender used a FICO version that produced higher scores than the FICO version used by the second mortgage lender. Thus, most of the leads turned out to be worthless because they failed to qualify under the second lender's version of FICO. Neither the online lender nor the second lender were aware that there could be such significant discrepancies.

²⁷ id.

TrueCredit Is Not Truly a FICO Score

Further complicating matters is the fact that consumers can buy credit scores from companies other than FICO. At times, these "knock-off" scores can mislead consumers because 1) consumers think that they are actually FICO scores, and 2) they are higher than the FICO score. Accordingly, the consumer thinks he has a high score, but is only a offered a mediocre rate when he applies for credit. This is because the real FICO score used by the lender was much lower than the imitation score obtained by the consumer.

If you find this a bit confusing, well, you're not alone. For instance, at Experian's Web site the credit score available for purchase is your "PLUS Score," a proprietary model developed by Experian and introduced in early 2004. TransUnion sells the TrueCredit score. Neither the Experian nor Trans Union Web sites clearly explain, in an easy-to-find place, that the credit score that you buy from them could differ significantly from the FICO score that your lender is likely to use. They also don't explain that lenders do not use TrueCredit or PLUS scores when making credit decisions.

Equifax's Web site only sells the FICO score, but the problem with this is that there are five industry-specific versions of Classic FICO. The bottom line is that a borrower cannot rely on the score bought at a Web site, even if it is a true FICO, to match the industry-specific score that will be provided to a mortgage company or auto dealer.

Mortgage industry sources say consumers regularly complain that the FICO score actually used by their lender was lower than the scores they bought at any of the Web sites, including the true FICO score available at Equifax.

At www.myFico.com, you can buy your FICO score for each of the three bureaus. If you are just trying to get a general idea of where you stand, the Experian and TransUnion knock-offs are probably good enough. But if you are applying for a mortgage, or an auto loan, or to refinance, a few points can make a big difference. Thus, the prudent thing to do is to make sure you get your actual FICO score, keeping in mind that even that might not be identical to the one that determines your interest rate.

Independent Research: The CFA-NCRA Study

Perhaps the most comprehensive research on credit scores and credit reports to date, found significant differences in credit scores for the same consumer among the Big Three CRAs. The research was published jointly in December 2002 by the Consumer Federation of America (CFA) and the National Credit Reporting Association (NCRA), which represents the smaller, independent credit bureaus and resellers.

In one phase, the groups set out to learn if there were significant differences between Equifax, Experian and Trans Union. To do this, they compared the credit reports and credit scores from each of the three major bureaus pertaining to 1,500 individuals. The study focused on a 620 score, as that was the line below which consumers paid dramatically higher-priced "sub-prime" loans. Researchers felt that wide disparities between credit scores reflected major differences in the underlying data of Equifax, Experian and Trans Union.²⁸

Specifically, "Phase 1" of the study found that among the three credit reports actually pulled for 1,500 consumers:

• 29% of the credit reports had a range of 50 points or more between the highest and lowest scores.


• 11% had a middle score between 575 and 630 and had a range of 30 points or more between the highest and lowest scores. This meant that one CRA would have a consumer pegged below "sub-prime" while another would have him above "sub-prime."
• 16% had high scores above 620 and low scores below 620.

²⁸ Consumer Federation of America and National Credit Reporting Association, Credit Score Accuracy and Implications for Consumers, December 2002

Overall, the NCRA-CFA study found that

• 21%, or one in five files, could be considered "extremely consistent," with a range of fewer than 20 points between the highest and lowest scores.


• 31%, or one in three files, had a range of 50 points or greater between scores.


• 5%, or one in 20 files, had a range of 100 points or greater between scores.

In Phase 2, researches did not pull full credit reports. Instead, they did an automated comparison of the credit scores among all three bureaus for 502,623 individuals. The resulting differences were remarkably similar to Phase 1.

• 24%, (compared to 21% in Phase One) could be considered extremely consistent, with a range of 20 points or fewer between the highest and lowest scores.


• 43% had a 21-49 point difference.


• 29%, (compared to 31% in Phase One) had a range of 50 points or greater between scores.


• 4%, (compared to 5% in Phase One) had a range of 100 points or greater between scores.

Considering that mortgage rates can change at 20-point, credit-score intervals, the NCRA-CFA study revealed the importance of knowing your credit score before you apply for credit. The study also tested differences between Equifax's and Trans Union's older and newer FICO scoring models, and found the differences to be negligible. The Phase I comparison discovered another unexplained occurrence: in 10% of the cases, instead of there being three credit reports for each individual (one each for Equifax, Experian and Trans Union), a fourth, or fifth, or even sixth credit report would be returned, most of which had credit scores.

These reports were not duplicates. In some cases, the "extra" credit reports clearly were reporting the credit activity of an entirely separate person, as none of the accounts matched with those on the three primary reports. But it was very common for the additional report to contain a mixture of credit data, some of which belonged to the applicant and some of which clearly did not. In still other cases, applicants had split files that appeared to be the result of applying for credit under variations of their name. It was unclear to researchers exactly how various lenders and their systems would interpret these additional credit reports and scores.

Credit Scores Rule

What is clear is that credit scores are a key factor in credit granting, be it automated or manual decision-making, especially in the all-important home mortgage loan business. To understand why, it is necessary to understand how the mortgage market works. Once a mortgage is approved, mortgage brokers and other "originators" of home loans generally look to sell them on the "secondary market." To do so most profitably, they want to sell these mortgages quickly, and that requires the financial guarantees of Freddie Mac's and Fannie Mae's automated underwriting process (AU).²⁹

As the CFA study put it, "In this market, where record volumes of loans are being originated, there is a tremendous incentive to deal only with the loans that will be approved the fastest — the loans that pass the credit score / automated underwriting test."

²⁹ The FICO score is one of 15 factors set by Fannie's and Freddie's automated underwriting systems.

It Was Different Before

Of course, it did not start out this way. Freddie Mac, in a 1995 letter to mortgage lenders, described FICO scores as only "one of the selection factors in our quality control sampling procedures." This, of course, was before Automated Underwriting took off.

"After reviewing a number of alternatives, we determined that, within the manual underwriting process, one of the easiest and most readily available tools to assist you in managing the challenging credit-risk environment is the use of either FICO bureau or MDS bankruptcy scores. Using these scores can help you better assess and manage the quality of your loan originations, reduce servicing costs and sustain profitability," Freddie Mac continued.

The 1995 Freddie Mac letter virtually closed out what now seems like a quaint era in which human judgment was required, and numbers weren't everything. The letter continued:

"We want to emphasize that there is no single FICO bureau or MDS bankruptcy score that means an individual borrower will default. However, these scores can help you identify loans that may require a closer look by your underwriter. If your underwriter is able to establish the borrower's willingness to repay as agreed, then we encourage you to consider this is in your investment-quality decision, regardless of what the credit score alone might suggest. Remember that you are still responsible for underwriting the credit reputation, as well as the file as a whole, to make your investment-quality decision."

In the years following this letter, advances in information technology, coupled with Freddie Mac's and Fannie Mae's satisfaction with FICO scores, allowed automated underwriting to take off. This coincided with falling interest rates and resulted in tremendous growth in both home sales and mortgage refinancing, benefiting millions of consumers and the economy as a whole.

In the meantime, the FICO score had emerged as the most important, and at times, the only determinant in the credit granting process.

'Reason' Codes

The FICO models most commonly in use offer several dozen "Reason Codes" as to why your FICO score is less than perfect. (See listing on pages 43-44.)

As you look over these codes, keep in mind that some of these codes should not be acted upon, as they do not represent "stand-alone" categories, but are scored against the broader credit history. In fact, under the newer scoring models, there is one code for "too few bank revolving accounts" and another code for "too many bank or national revolving accounts."

Also, under the new models, the number of scoring factors or reason codes has expanded.

In addition, the three major credit bureaus' scoring models generally can produce special messages after a credit report is analyzed. Below are some examples of so-called "Scoreability criteria" from TU's old Empirica model:

• Not Scored — Insufficient credit; when a credit files does not contain a tradeline opened for at least six months and a tradeline updated within the last six months;


• Not Scored — Deceased; when the subject's Social Security number matches the Social Security Administration's deceased SSN file.


• Alert — When a credit file contains one or more of the following previous bankruptcy, derogatory public record, collection activity, or an MOP of seven or higher.

Credit Bureaus	Unless	Equifax	TransUnion	Experian
Reason Statement	Other	BeaconSM	EmpiricaSM	FICO-IISM
Amount owed on accounts is too high	01	-	-	A/
Level of Delinquency on accounts	02	-	-	B/
Too few bank revolving accounts	03	-	-	C/
Proportion of loan balances to loan amounts is too high	03	-	-	I/33
Too many bank or national revolving accounts	04	-	-	D/
Lack of recent installment loan information	04	-	-	Y/32
Too many accounts with balances	05	-	-	E/
Too many consumer finance company accounts	06	-	-	F/
Account payment history is too new to rate	07	-	-	G/
Too many inquires last 12 months	08	-	-	H/
Too many accounts recently opened	09	-	-	J/
Proportion of balance to credit limits is too high on bank revolving and other revolving accounts	10	-	-	K/
Amount owed on revolving accounts is too high	11	-	-	L/
Length of time revolving accounts have been established	12	-	-	M/
Time since delinquency too recent or unknown	13	-	-	N/
Length of time accounts have been established	14	-	-	O/

Credit Bureaus	Unless	Equifax	TransUnion	Experian
Reason Statement	Other	BeaconSM	EmpiricaSM	FICO-IISM
Lack of recent bank revolving information	15	-	-	P/
Lack of recent revolving account information	16	-	-	Q/1
No recent non-mortgage balance information	17	-	-	R/
Number of accounts with delinquency	18	-	-	S/
Too few accounts currently paid as agreed	19	-	-	T/
Date of last inquiry too recent	19	-	-	-
Time since derogatory public record or collection too short	20	-	-	V/
Amount past due on accounts	21	-	-	W/1
Serious delinquency, derogatory public record, or collection	22	-	-	X/*
Number of bank or national revolving accts with balances	23	-	-	-
No recent revolving balances	24	-	-	U/1
Length of time installment loans have been established	25	-	-	I/0
Number of revolving accounts	26	-	-	I/0
Number of bank or other revolving accounts	26	-	-	I/0

Secret Data?

Important aspects of the system remain a secret. Sources have told this author there are raw data in credit reports - regularly scored by FICO scoring models - that are never seen by consumers because these data do not appear on their credit reports. These raw data show up on "machine readable" credit reports used by lenders and their scoring models, but not on the more familiar reports disclosed to consumers.

For example, one major credit bureau has one code to denote the most recent negative item; a second code marking the previous most recent negative item, and a third code denoting the most recent, worst negative that is more than 24 months old. If consumers are to become more educated about their credit reports and credit scores, shouldn't this kind of information be disclosed as well?

Generation Gap

In late 2001, Fair Isaac announced the rollout of its "Next Generation" scoring models with some fanfare. Fair Isaac's Karlene Bowen predicted before the National Credit Reporting Association that the new "NextGen" model bring more precision and would rapidly be adopted by banks and mortgage lenders in 2002.³¹

Bowen said the key to the "Next Generation" score was that it used complex statistical models to "see through" credit file data to better identify loan applicants who represented the highest risks of delinquency or foreclosure. Based on new analyses of tens of millions of consumer credit files, the Next Generation scores "reward" some people -- moving them into the heretofore rarefied "800" and higher score category. But it also pushed other people below the "600" level that often triggers higher interest rates and fees.³¹

The ranges changed as well. While "800" might be rare under FICO classic because the top score was "850," under NextGen, the top score is now "950" - and the default rate classes are probably adjusted as well. Fair Isaac's Craig Watts indicated for many consumers, the change would not be that great. "NextGen scores correlate to classic FICO scores, so a score of 620 from either model indicates the same likelihood the consumer will become seriously delinquent within 24 months," he said.

³⁰ Kenneth Harney, "Higher Credit Scores On the Horizon," Realty Times, November 12, 2001
³¹ Id.

But lenders for the most part have not adopted Next Generation. Fair Isaac attributed their reticence to the cost of integrating new software. But another important factor is that Fannie and Freddie have not endorsed it. Until they do, don't expect many others to. It took six years for Fannie and Freddie to endorse FICO classic after its unveiling in 1989. There's no telling if and when the two influential mortgage organizations will recognize "NextGen" in the same way.

This means that there will continue to be a pronounced lack of uniformity in the market, which, at a minimum, will generate confusion for consumers trying to understand their actual status in the credit-scoring hierarchy.

Where Do We Stand?

There appears to be a discrepancy in how Fair Isaac and Experian rate the creditworthiness of the U.S. population. Fair Isaac said as of early 2004, this was how U.S. consumers' FICO scores were distributed nationally:

• 800 or higher - 11 percent


• 750-799 - 28 percent


• 700-749 - 19 percent


• 650-699 - 16 percent


• 600-649 - 12 percent


• 550-599 - 8 percent


• 500-549 - 5 percent


• 499 and under - 1 percent³²

If Fair Isaac's estimates are accurate, then nearly 60 percent of U.S. consumers are qualifying for top-notch credit, with credit scores of 700 or above.

However, in March 2004, Experian rated the nation according to its PLUS score, which uses a range similar to the FICO model. However, the PLUS Score was not used by lenders, sources said.

³² Fair Isaac Corp., "National Distribution of FICO Scores;" (Slide)

Still, Experian claimed its survey used the most up-to-date information, and found that the average PLUS credit score was 678. Moreover, it gave the following regional breakdown:

• New England - 699


• West North Central - 695


• Mid-Atlantic - 690


• East North Central - 684


• East South Central - 677


• South Atlantic - 675


• Pacific - 674


• Mountain - 670


• West South Central - 656³³

It was worth noting that New Englanders had the most credit card and installment debt (per capita, $13,566)³⁴, had the most credit cards, had the highest minimum monthly payments ($575) and the most credit usage, according to Experian. But they had the fewest late payments.

The "West South Central," on the other hand, had the lowest per capita debt ($9,297), but the most late payments.

Who Knows?

One key question is whether the credit-scoring system so lacks in transparency, fairness or reliability that a stronger public policy is needed to protect consumers.

Because Congress dedicated much of 2003 to a major updating of the Fair Credit Reporting Act, and because the FTC and federal banking agencies continued working into 2005 to establish new enforcement rules and guidance, it was not seen as likely that Congress would have the appetite for revisiting credit-scoring issues anytime soon.

³³ www.nationalscore.com/USScore.aspx, click on "View All Graphs."
³⁴ Excluding real estate/mortgage loans

Thus, despite any problems with the current credit-scoring and credit-reporting systems, it appears that they are here to stay. Many of the changes passed by Congress, coupled with the financial industry's response to them, should help make the system more transparent and fair to consumers. But as much of this book should make clear, there's a long way to go. A primary goal of this book is to narrow the "knowledge gap" so consumers can gain an in-depth understanding of how these all-important systems work and what consumers can do to improve their lot.

Does Credit Scoring Really Work?

In the 2002 NCRA-CFA study, which was one of few known independent research efforts on credit scoring validity, the authors noted the lack of non-industry assessment of credit-scoring methods.

Despite the gatekeeper role that these scoring systems play regarding access to credit, housing, insurance, utilities, and employment, as well as pricing for those essentials, exactly how the formulas perform the transformation from credit report to credit score is a closely guarded secret. For consumers, regulators, and even industry participants who rely on the computations in their decision-making, the scoring models largely remain a "black box." No scholarly reviews of this extremely powerful market force have been permitted, and apart from reviews by federal banking regulators to protect against discrimination, no government regulator has insisted that they be examined to ensure that they are adequate and fair.

Stephen L. Ross and John Yinger, both professors of economics at Univ. of Connecticut and Syracuse Univ., respectively, made a similar observation in their 2002 book,

The Color of Credit:

No existing credit-scoring scheme, let alone a fully automated underwriting system, has been subjected to the scrutiny of disinterested scholars. There is a significant literature (reviewed in Thomas, 2000) on the technical dimensions of credit scoring, that is, on the best method for devising a credit score. As Thomas points out, however, "comparisons [a-cross methods] by academics are often limited as some of the most significant data like the credit bureau reports are too sensitive or too expensive to be passed to them by users." As a result, the accuracy of credit-scoring schemes remains an open question.³⁵

Fair Isaac said it's never really been a factory for "white papers," a reference to the lengthy, seldom read research papers that some technology companies are known to churn out.

³⁵ Stephen L. Ross & John Yinger, The Color Of Credit (MIT 2003); the passage refers to Lyn C. Thomas, "A Survey of Credit and Behavioral Scoring: Forecasting Financial Risk of Lending to Consumers," International Journal of Forecasting, 16(2) (April-June): 149-172.

But Fair Isaac is confident that its system helps lenders improve their decision-making, and that it's an overall plus for consumers and for the economy.

Most important, Fair Isaac said that its scoring model has been tested over and over in the market by nearly every creditor in the country, and has passed with flying colors, as demonstrated by the company's chart on page 49.

One dissenting voice was Golden West Financial's Herb Sandler, who with his wife, ran one of the most profitable mortgage lenders in America. Sandler told Forbes Magazine that Golden West, based in Oakland, California, doesn't use or trust FICO scores. He said the models were too dependent on borrowing histories accumulated during a relatively benign economy with strong housing prices. Golden West's bad loan rate was just 0.5% of assets compared with 0.7% for the thrift industry, Forbes reported March 1, 2004.

"Obviously we're doing something right," Sandler said.

Other lenders said that Golden West has a special situation, as it specialized in short-term adjustable rate mortgage loans, and had more face-to-face contact with its clientele.

One thing that Fair Isaac is crystal clear on: All of its models rely exclusively on information in consumers' credit reports when calculating credit scores. That is why to gain a full understanding of how the credit scoring system works, one needs to examine the source: the credit reporting system. With the exception of the next chapter, which explores the role of resellers and their position in the system, the balance of this book is devoted to an examination of the credit reporting system and how credit reports are used, and at times, misused.

Continue to Next Chapter  

© 2005 Evan Hendricks and Privacy Times, Inc. All rights reserved.