For the third time in recent years, the Supreme Court has agreed to consider the question of whether the disparate impact doctrine is encompassed in the Fair Housing Act. Twice the cases were settled before the court could make a ruling. The issue has long been of great interest to lenders and varied other types of organizations whose activities are or might be covered by the FHA.

The disparate impact doctrine holds that an organization may be found guilty of discrimination if its practices disproportionately disadvantage a protected group (even if the effect is unintentional) and it cannot provide a sound justification for the practice. The case slated to be heard by the court, Texas Department of Housing and Community Affairs v. The Inclusive Communities Project, Inc., hardly involves a straightforward application of the concept.

A fair housing group accused the Texas public housing authority of discrimination because it approved a higher proportion of low-income housing tax credit applications in areas that had very high minority representations than in areas with lower minority representations. The difference in approval rates allegedly contributed to a pattern of housing segregation. The challenged procedures would probably allow minorities to secure more low-income rental units, but fewer of those units would be available to them outside of predominantly minority areas.

While that aspect of the matter may receive some attention, the court's decision about the general applicability of disparate impact to the FHA will likely turn on a different issue. The question is whether the court will find that the Department of Housing and Urban Development's February 2013 rule applying the disparate impact doctrine to the FHA was a reasonable interpretation of the statute.

The court will be urged to give great deference to the enforcing agency's interpretation of the statute, even if it would be inclined to reach a different result. But there are reasons why the court might not defer in this case, including the possibility that the FHA — as interpreted by HUD's rule — is too vague to withstand constitutional scrutiny.

The principal vagueness problem rests not in the application of the disparate impact concept itself (though that concept is murky enough in many settings). Rather, the problem lies in the provision whereby an organization can be held liable even for a justified practice, if an alternative practice is available that equally serves the organization's interests while having "a less discriminatory effect."

This provision requires organizations seeking to comply with the law to make judgments about what a less discriminatory alternative might be. This undertaking is complicated by the near-universal failure to understand a basic statistical pattern.

Concerned about large relative (percentage) racial differences in areas like mortgage rejection rates, regulators have for two decades been encouraging lenders to relax lending criteria that adversely impact higher proportions of minority applicants than white applicants. Relaxing any standard — including a credit score or income requirement — will tend to reduce the size of relative differences between the rates at which minority applicants and white applicants meet the standard and have their applications approved. But relaxing the criteria will tend to increase the size of relative racial differences between the rates at which applicants fail to meet those standards and have their applications rejected. (For graphical illustrations with actual credit score or income data, see my March 4, 2013 letter to the Federal Reserve Board and my Spring 2006 Chance magazine editorial.)

Unaware of the latter part of the picture, regulators have continued to monitor the fairness of lending practices on the basis of relative differences in adverse outcomes — e.g., loan denial rates. Thus, when lenders comply with encouragements to relax certain requirements, they increase the chances that the government will accuse them of discrimination.

This pattern makes it difficult to determine whether an alternative practice would in fact reduce disparate impact, even when one fully understands the statistical issues (as I explained in a 2013 workshop paper). It is even more difficult to do so when one does not know those issues exist, as has generally been the case among both those enforcing the laws and those subject to them.  

The void for vagueness doctrine, which requires that a statute provide sufficient guidance for a person to know what is or is not unlawful, is principally applied to criminal laws. But it can be applied in a civil context as well. The difficulty of determining which alternative is "less discriminatory" could be sufficient to cause the court to find HUD's interpretation unconstitutionally vague. Even if the court will not go that far, however, HUD's longstanding failure to understand the matter may well cause the court to accord less deference to agency expertise than it usually would. 

So far, of course, the court itself does not know that these statistical issues exist. Once it does, it may need to rethink a number of disparate impact issues.

Meanwhile, on Nov. 3, Judge Richard J. Leon of the U.S. District Court for the District of Columbia (in a case discussed here in January) issued a strongly worded opinion finding that HUD's interpretation of the FHA to include the disparate impact doctrine was manifestly unreasonable. The ruling should curb HUD's use of disparate impact until the Supreme Court makes a decision.

James P. Scanlan is a lawyer in Washington. He specializes in the use of statistics in litigation.