Credit cards have long been subject to the Equal Credit Opportunity Act and Regulation B provisions that prohibit discrimination on the basis of race, gender, marital status and other characteristics set out in the law. Recently, there has been a lively debate about whether disparate impact should apply to credit transactions. Ultimately, the U.S. Supreme Court may decide this issue under the Fair Housing Act. But, unlike rules implementing the Fair Housing Act, Regulation B has, for more than 35 years, noted that Congress intended disparate impact to be applicable under the ECOA. Thus, a Supreme Court decision under the Fair Housing Act may not directly preclude action under the ECOA. In the meantime, the reality of a banking examination is likely to intrude upon this legal issue.

Specifically, we may see increased use of disparate impact analysis for credit cards. This is due to two factors. First, the Consumer Financial Protection Bureau has embraced the use of disparate impact in its review of creditor practices. In speeches, bulletins, exam procedures and other documents, the CFPB has made clear that it will examine institutions' policies and practices to determine whether they have a disparate impact on a prohibited basis.

In fact, examination procedures issued by the CFPB this past summer expressly raise the question of whether underwriting or pricing policies/practices use factors that may have a disparate impact on a prohibited basis. Second, recent information suggests that credit card issuers have tightened underwriting standards for credit cards – and made less credit available to some consumers – to better manage risk.

So, do card issuers need to worry?

In October, the CFPB issued a report on the impact of the Credit Card Accountability, Responsibility, and Disclosure Act on the consumer credit card market. The report found that originations, approval rates and the average amount of credit lines for general-purpose credit cards have rebounded somewhat from the 2008 recession. However, the CFPB found there are considerable differences between prime and subprime cardholders, with subprime cardholders regaining less credit than prime cardholders. A similar finding was made in September by the American Bankers Association. The ABA report found that since the recession ended in 2008, lines of credit for "superprime" credit cards had increased in 2010 and 2011 (but declined in 2012), but lines for subprime cards had consistently declined since 2008.

So, what does all this mean? First, card issuers often use credit score cutoffs to manage risk. Studies have shown that credit scores are highly predictive in rank-ordering the risk of payment (or nonpayment). Because risk of default is substantially related to a consumer's credit score, credit card issuers often manage risk by not offering credit to consumers in lower credit score bands or by offering lower credit lines. (Issuers also, of course, charge different rates for credit, to risk-base the price of credit that is extended.)

Second, data show that credit scores are significantly correlated with race. A 2007 Federal Reserve Board report to Congress on credit scoring found that the race and ethnicity of individuals in the lowest credit score ranges varied significantly. The FRB study of a representative sample of credit records found that more than 50% of blacks and 30% of Hispanics were in the lowest two deciles of credit score ranges, compared to 12% for Asians and 16% for non-Hispanic whites.

So what does all this have to do with the ECOA and discrimination? At some point, the CFPB may start pursuing questions about whether an institution's credit scoring models adversely affect subprime cardholders, and whether black and Hispanic applicants are statistically more likely to be subprime cardholders than white applicants. In other words, do existing scoring models have a disparate impact on the ability of black and Hispanic applicants to obtain credit, and does the amount of credit made available to those persons differ based on their credit scores?

As we know, such inquiries are just the beginning of the journey down the road of disparate impact. The three-part legal test for disparate impact is, in theory, well understood. First, does a neutral policy/practice disproportionately and negatively affect persons on a prohibited basis? Second, is there a sufficient justification (a legitimate business need) for the approach? Finally, can the policy/practice be achieved as well by means that are less disparate in their impact?

But the reality of putting that test to work is a bit different. There are numerous questions to consider. For instance, when is it appropriate to begin testing for disparate impact? Moreover, if a creditor reviews its scoring models it needs to consider several issues. As an example, what are the relevant populations to consider, in evaluating any potential disparate impact claim? There are many ways of creating potential pools of consumers within protected classes. It is critical to determine what the relevant pools should be. Second, what does it mean to have a sufficient business justification if a policy/practice has a disparate impact? It is clear that a theoretical justification will not be adequate. Data is good, but certainly not the only way of establishing a legitimate business reason for the policy/practice. Finally, what exactly does it mean for there to be another way of accomplishing the policy/practice with a less disparate impact?

It is likely not too early to start thinking about these issues. Remember: when the CFPB comes knocking, it's probably not to bring you a candygram.

Before rejoining Morrison & Foerster, Leonard N. Chanin served as assistant director of the Office of Regulations of the Consumer Financial Protection Bureau.