Time for the FCRA to be amended/updated.

[Jstic]

The advent of the internet and the information age has steadily increased the importance of credit scores. The current system used to calculate credit scores  has created an environment that can be extremely confusing, and very frustrating for the average consumer. The Fair Credit Reporting Act went a long way toward regulating CRAs and what they could and could not report. What it does not do is regulate the scoring models in a way that would make credit scores more understandable, reliable, and more fairly calculated.

TU currently has at least two scoring models(TU98, TU04), all three CRAs use different formulas for scoring, often times resulting in significant score differentials. One CRA doesn't even allow a consumer to see his/her score unless denied credit or willing to join a specific CU. Different scores are available for credit card applications, auto financing, mortgages. Many, many different scores are available that are rarely if ever used by lenders, creating more confusion and frustration for the average consumer.  CRAs classify different types of credit based on their own scoring model, which can create noticeable differences in scores. I gave an example on these boards a few months ago where one CRA calculated my HELOC as unsecured debt, much like a credit card, and another calculated it as a real estate backed loan, which is what it is. The result was a 40+ differential, just for that one item.

While I am not a big fan of government regulations in general, in this case I believe it could do almost no harm, and a lot of good to change the FCRA to force CRAs to have some semblance of consistency when calculating credit scores. Most consumers don't have the time to come to this forum and study the boards for a month to get educated about all the differences in credit scoring.  It would go a long way toward seeing that scores are more easily understood and calculated with a measure of consistency.  It would also  make for a consumer armed with accurate, reliable information before he/she applies for credit, a job, a home rental, etc.

[RobertEG]

Commerical credit scoring algorithms are federally-protected trade secrets.

I see no way for the govenment to regulate credit scoring algorithms without requiring the disclosure of proprietary trade secrets.  It would require a major, and dangerous, revision of the entire trade secret protection statutes.

[cashnocredit]

---

@RobertEG wrote:

Commerical credit scoring algorithms are federally-protected trade secrets.

I see no way for the govenment to regulate credit scoring algorithms without requiring the disclosure of proprietary trade secrets.  It would require a major, and dangerous, revision of the entire trade secret protection statutes.

---

+1

The only thing govt. regulates re credit scores is to make sure they aren't discriminatory. For instance someone who is 90 y/o might well have a higher risk of being late (pun intended) but the law requires age not be used adversely when scoring seniors. FICO doesn't use age at all. Same for other protected categories.  Score algorithms are evaluated by govt. agencies to make sure they comply with the equal access to credit laws. However, in general the algoritms aren't evaluated, only the large impact of scores across a large sample. The govt. doesn't regulate score accuracy and probably couldn't do so in any realistic manner.

[marty56]

---

@Jstic wrote:

Most consumers don't have the time to come to this forum and study the boards for a month to get educated about all the differences in credit scoring.  It would go a long way toward seeing that scores are more easily understood and calculated with a measure of consistency.  It would also  make for a consumer armed with accurate, reliable information before he/she applies for credit, a job, a home rental, etc.

---

While credit scoring is complex, the road to having a good credit score regardless how it is calculated is simple.  Never pay late, keep CC balances low, don't apply for new credit unless you need it, and finaly never pay late.

[MattH]

As others have noted, credit scores are trade secrets and therefore it would not be legally possible to standardize them. Many large lenders have their own internal models as well; there is intense competition. A lender who can do a better job of predicting credit risk than its peers can gain competitive advantage, while a lender that does a poor job of underwriting can end up going broke.
…
What does need to change is the widespread deceptive marketing. The recent reform that requires Risk-Based-Pricing disclosures is an excellent start, but could be improved in several ways. First, along with telling the applicant Your score was XX which puts you in the YY percentile so your interest rate is higher than our prime rate, the lender should be required to specify exactly which model they use. Second, deceptive marketing of FAKO scores has got to stop. Those selling scores to consumers should only be allowed to sell scores that are actually used by lenders. Third, any person who is preparing to apply for a mortgage or an auto loan should be allowed to purchase the exact same tri-merge report and scores as a mortgage lender or car dealer can purchase, so they can come to the broker's office knowing exactly what the loan officer will see.

[cashnocredit]

---

@MattH wrote:
As others have noted, credit scores are trade secrets and therefore it would not be legally possible to standardize them. Many large lenders have their own internal models as well; there is intense competition. A lender who can do a better job of predicting credit risk than its peers can gain competitive advantage, while a lender that does a poor job of underwriting can end up going broke.
…
What does need to change is the widespread deceptive marketing. The recent reform that requires Risk-Based-Pricing disclosures is an excellent start, but could be improved in several ways. First, along with telling the applicant Your score was XX which puts you in the YY percentile so your interest rate is higher than our prime rate, the lender should be required to specify exactly which model they use. Second, deceptive marketing of FAKO scores has got to stop. Those selling scores to consumers should only be allowed to sell scores that are actually used by lenders. Third, any person who is preparing to apply for a mortgage or an auto loan should be allowed to purchase the exact same tri-merge report and scores as a mortgage lender or car dealer can purchase, so they can come to the broker's office knowing exactly what the loan officer will see.

---

There is no way to know what particular score brand, score vintage, a lender will use. Secondly, many CC lenders use a custom score to measure expected profitability and these can vary considerably from risk only scores such as the FICO scores available to us. And then there are acquisition scores that factor in identity as well as first party fraud risks.

However, I agree on deceptive marketing of FAKOs. Places that sell them should disclose they are not used by lenders as well as provide additional information such as the K-S statistic and Gini coefficient if they have done any work "validating" the predictability of their FAKOs.

http://fic.wharton.upenn.edu/fic/11-19-05%20conf%20summary.pdf

[MattH]

Interesting article in your link! I would also note, a lender has different reasons for buying a credit score than does a consumer. The lender who buys my score wants to know, what is the risk I don't pay them back. When I buy a score I want to know, how will lenders see me. Suppose both FICO scores and FOOBAR scores have an r-squared of 80%, that is they predict about 80% of the variation in credit risk of consumers. If I am a lender, either score should at least in theory be equally useful to me if they are equally good statistical predictors of outcomes. In practice, most lenders will stick to what they know (some flavor of FICO score) even if some competing model claims to be as good or better. But if I am a consumer, what I really care about is what score will a loan officer see. If FICO and FOOBAR scores both have r-squared for predicting consumer behavior of 80% and are independently developed, then I would expect the r-squared FOR USING FOOBAR SCORE TO ESTIMATE FICO SCORE to be about 64%. The free FICO score estimator on bankrate.com (it computes an estimated score from information you provide) probably has an r-squared for estimating FICO scores much better than 64 percent! So for a consumer, the free Estimator is more useful than any FAKO score, which why I think regulators should crack down on sellers of FAKO scores.

[marty56]

Just a thought be it seems to my that the FAKO score providers CMS tend to be cheaper then the FICO score providers or at least allow some sort of daily pulls.  If Scorewatch was say $9.95 per month and allowed daily EQ pulls and a month pull of all 3 reports.

I know probably only a CRA  can provide services like that but imagaine if Scorewatch could.  This thread would have cobwebs on it.

I know my Truecredit score are almost worthless but at $9.95 a month and a little research by me on them, it is a hard act to follow.

[cashnocredit]

---

@MattH wrote:
Interesting article in your link! I would also note, a lender has different reasons for buying a credit score than does a consumer. The lender who buys my score wants to know, what is the risk I don't pay them back. When I buy a score I want to know, how will lenders see me. Suppose both FICO scores and FOOBAR scores have an r-squared of 80%, that is they predict about 80% of the variation in credit risk of consumers. If I am a lender, either score should at least in theory be equally useful to me if they are equally good statistical predictors of outcomes. In practice, most lenders will stick to what they know (some flavor of FICO score) even if some competing model claims to be as good or better. But if I am a consumer, what I really care about is what score will a loan officer see. If FICO and FOOBAR scores both have r-squared for predicting consumer behavior of 80% and are independently developed, then I would expect the r-squared FOR USING FOOBAR SCORE TO ESTIMATE FICO SCORE to be about 64%. The free FICO score estimator on bankrate.com (it computes an estimated score from information you provide) probably has an r-squared for estimating FICO scores much better than 64 percent! So for a consumer, the free Estimator is more useful than any FAKO score, which why I think regulators should crack down on sellers of FAKO scores.

---

Indeed, the consumer is interested in the score a lender uses, regardless of how accurate another score may be.

I'm sure you can see the dilemma regulators are in. Prohibiting a consumer from purchasing a credit score that may or may not actually be more predictive of risk because it may be less predictive of a FICO score that currently has 90% market share or more seems against the usual policy of not regulating to the encouragement of monopoly.

I think prominent disclosure that credit scores purchased are not used by lenders generally work but I'll go one further. It should be disclosed by lenders what credit scores they use and whether they can be purchased by consumers. Lenders would be free to use whatever FICO, FAKO, or custom variation they wished but must disclose it.