In a recent article in National Mortgage Professional entitled "Manufacturing Fair Lending”, former chief of the Housing and Civil Enforcement Section at the Department of Justice under Attorney General William Barr, Paul Hancock exposes the misuse of statistical analysis by the DOJ in concert with the federal bank regulators to concoct redlining allegations against mortgage lenders. This is a phenomenon about which I have written several articles in the past year and my impression is that it’s gotten out of hand.

Ever since Attorney General Merrick Garland announced the “Anti-Redlining Initiative” in October 2021 a record number of referrals have been made to the DOJ by bank regulators alleging potential redlining. The DOJ concomitantly has proclaimed numerous settlements with lenders that have been accused of redlining. A common thread in all these cases is the use of statistical analysis to demonstrate disparities in lending volume in majority-minority census tracts.

Faced with the prospect of time-consuming and costly litigation with the US government and the adverse publicity associated with allegations of discrimination, many banks have agreed to expensive settlement agreements with the DOJ. As quoted in the “Manufacturing Fair Lending” article, “Lenders’ inability or unwillingness to litigate against the government, given the expense, inflates perceptions of rampant redlining and the effectiveness of regulators’ “combatting’ efforts.”

We at GeoDataVision have observed abuses of statistical analysis in several situations in which lenders have been threatened with redlining referrals to the DOJ. One of the abuses involves “Simpson’s Paradox” which explains contradictory results when analysis of minority subgroups results in different outcomes compared to the aggregated data (see: Figures Don't Lie but do Regulators Figure)

Another glaring example is the abuse of the “Reasonably Expected Market Area” (aka “REMA”) concept. REMAs are not to be found in any federal statute nor any federal banking regulation. Nevertheless, the REMA concept has been in practice since about 2009. The determination of what constitutes a REMA is strictly in the purview of the regulators who have announced a radically different and expansive approach beginning in 2021.

Prior to the Anti-redlining Initiative examiner manuals provided guidance on how examiners should circumscribe a REMA.  Factors examiners were told they should consider were among other things: the CRA assessment area(s), the geographic dispersion of loans and loan applications, the geographic reach of marketing campaigns, the location of LPO’s, etc.

But in 2022 examiners adopted a radically different approach to REMA’s announcing they would consist of entire Metropolitan Divisions or Metropolitan Statistical Areas. This radical change in defining REMA’s has profound implications for anti-redlining enforcement because many community banks cannot realistically serve or compete in entire MSAs. An immediate consequence of adopting an unrealistic REMA is unreliable if not entirely misleading statistical analysis.

We’ve seen the adverse consequences of unrealistic and expansive REMAs for community banks in several situations. In one case, when one community bank acquired another small community bank the bank REMA resulted in a market 6 times larger than any market served by both banks and it was applied retroactively during the examination period to a time before the acquisition! 

In another situation, the bank’s REMA was defined as an entire MSA even though the bank had branches in only half the MSA. The bank had historically served the counties in the MSA that were largely rural and suburban in nature. But the regulator insisted on expanding the market to include an urban area in which the bank had no branches. This resulted in a distorted picture of the market served by the bank and an accusation of redlining based on that distorted market.

It's time for a public debate about regulators abusing statistical analysis to derive unrealistic and unreliable accusations. It’s time to restore fairness to fair lending regulation.

[View source.]