Since October 2021, the prudential bank regulators in concert with the Department of Justice have been pursuing the “Anti-Redlining Initiative” announced by Attorney General Merrick Garland on October 21, 2021. During that time, I’ve been engaged by various banks to help them respond to threats by regulators to refer them to the DOJ for alleged redlining. Undoubtedly, the regulators and the DOJ have embarked on an aggressive campaign to root out the odious practice of redlining, which the Attorney General declared is “pervasive” today. However, I’ve observed some aspects of the anti-redlining crusade that I think are troubling and undermine the integrity of the campaign. I am speaking about the transition to the Census 2020 demographics from the Census 2010 demographics. This transition occurred at the beginning of 2022 and creates potential confusion regarding reported mortgage lending for 2022.

Let me begin by explaining the problems created by the application of the new Census 2020 data, which was initiated effective January 1, 2022, only 2 months after the AG announced the Anti-Redlining Initiative. The implementation of the Census 2020 tracts resulted in an increase from 75,883 tracts in 2021 to a total of 85,395 tracts in 2022. This change has enormous implications that undermine the ability to confidently derive conclusions about potential redlining based on 2022 HMDA LAR data.

• The creation of almost 10,000 new census tracts affects far more than only 10,000 tracts since all new tracts are created from old tracts. This means that potentially 20,000 or more of the 2021 census tracts were split into smaller tracts. In one day, from December 31, 2021, to January 1, 2022, thousands of new tracts were created, and thousands of old tracts eliminated.
• In addition to the tens of thousands of new tracts created and old tracts eliminated, thousands of other tracts had their minority classification changed because of population shifts in those tracts. So, the implications for Majority-Minority tracts and their various subtypes (Majority Hispanic tracts, etc.) are that thousands of more census tracts were affected by the new Census demographic data. Between the old tracts, new tracts, and the tracts with new minority classifications we believe that as many as 40% of the US census tracts changed their minority status between 2021 and 2022.
• In one situation we observed a bank had a Reasonably Expected Market Area (“REMA”) of 815 tracts that changed to 920 tracts in 2022. The change was far more profound than the net increase of 105 census tracts. What we saw was that 104 census tracts were eliminated, and 209 new tracts added to the REMA. In addition, dozens of other tracts had their various minority classifications changed. In that market nearly 45% of the census tracts were affected by the new Census 2020 data implementation.
• A substantial volume of mortgage applications reported in the 2022 HMDA LARs filed by HMDA reporters were initiated during 2021 and many of those applications were in different census tracts at the date of the application compared to the tracts to which they were eventually geocoded at the date of final action during 2022. A bank that processed an application received during 2021 had no way of knowing the ultimate census tract minority status because the assigned census tract is determined at the time of “final action”, not at the time of application. It’s entirely possible that the location of many mortgages initiated during 2021 (and early 2022 before the FFIEC announced the new tracts and their minority classifications) changed census tracts.

We have seen a situation in which more than a third of a bank’s 2022 HMDA LAR applications were received during 2021 and early 2022 before the FFIEC announced the new tracts. Nearly 20% of those records were eventually assigned a census tract that was different from the tract at the time of the application. This can have major implications for redlining analysis. With the very identity of tracts as well as their minority status changed on many applications received during 2021 and early 2022, it would be almost impossible for a bank committed to improving its 2022 “penetration rate” in the majority -minority tracts to know with certainty whether their loans between the late part of 2021 and the first quarter of 2022 were in majority-minority census tracts.

Banks when reviewing their 2022 HMDA LAR should be mindful of these problems, especially if a regulator observes statistically significant low minority tract penetration rates in the bank’s REMA. This “performance context” information exposes census tract volatility and casts a cloud of uncertainty over a strict numbers approach to hot issues such as “redlining”. To determine your exposure to redlining risk related to census tract volatility we suggest that a bank review its 2022 HMDA LAR:

1. Review your REMA to identify how it was affected by the new census tracts. How many new tracts were created? How many tracts were deleted? How many tracts changed minority classification? The answers to these questions will give an indication of how exposed your institution is to the risk of higher-than-normal census tract turnover.
2. Determine how many applications were received during 2021 and the first quarter of 2022 (before the FFIEC announced the new Census 2020 tracts and the related demographics). The answer to this question will provide you with insight into your exposure to mortgage applications processed during the transition period.
3. Review any applications discovered in Step 2 and determine how many were in a census tract that changed between the application date and the date of final action. This will help you to identify specific applications that may have been affected and to quantify your risk.

Remember that no one, not even the regulators, knew the identities nor the minority classification of the 2020 census tracts until the end of the first quarter of 2022 when the FFIEC published the data retroactive to January 1, 2022. If your lending volume in majority-minority tracts for 2022 is being questioned the answers to the above questions may save you from a referral to the DOJ.

[View source.]