N.J. Attorney General Preserves Disparate Impact Theory of Lending Discrimination for Borrowers

Holland & Knight's Garden State Initiative is a recently established group that supports our New Jersey-based clients and others doing business in the state. For more information about how Holland & Knight can assist with a specific New Jersey-related matter, please reach out to Leonard Bernstein, Richard Smith or Alison Keppel.

The Trump Administration issued an executive order in April 2025 directing federal agencies to no longer rely on the "disparate impact" theory of liability to find financial institutions and others guilty of discrimination under various federal "fair lending" laws, including the Fair Housing Act and Equal Credit Opportunity Act. This was good news for financial institutions and other entities, including those that deal with New Jersey residents. However, New Jersey has responded by taking a different approach.

On December 15, 2025, the New Jersey Department of Law and Public Safety's Division on Civil Rights (DCR) adopted a Final Rule,¹ effective immediately, to "clarify" that the New Jersey Law Against Discrimination (LAD)² prohibits "disparate impact discrimination." The Final Rule indicates that disparate impact discrimination occurs when a person's practice or policy "actually or predictably results in a disproportionately negative effect on members of a protected class," even if the practice or policy is "not discriminatory on [its] face (that is, facially neutral) and [is] not motivated by discriminatory intent," unless 1) it can be shown to be necessary to achieve a "substantial, legitimate, nondiscriminatory interest" (Valid Interest), and 2) there is no less discriminatory alternative that would achieve the same interest."³ A Valid Interest is "substantial" if it is a core interest of the person that has a direct relationship to the person's function, "legitimate" if it is "genuine, not false or pretextual," and "nondiscriminatory" if it does not itself discriminate based on a protected characteristic.⁴

In the press release announcing adoption of the Final Rule, the New Jersey Attorney General (NJAG) noted that "[t]he rules cement critical state-law civil rights protections just as the Trump Administration has moved to reverse key protections against disparate impact discrimination at the federal level." (See the Final Rule, press release and a fact sheet on disparate impact discrimination claim analysis.)

This Holland & Knight alert summarizes the LAD and Final Rule and provides an assessment of the potential impact on financial institutions that make loans to New Jersey borrowers and/or service such loans.

Background: "Fair Lending" in New Jersey

The LAD broadly prohibits any entity to which it applies from discriminating against others because of race, creed, color, national origin, ancestry, age, marital status, affectional or sexual orientation, familial status, disability, liability for service in the U.S. Armed Forces, nationality, sex, gender identity or expression, or source of lawful income used for rental or mortgage payments.⁵ The LAD applies to various types of entities, including, specifically, lending institutions "involved in the making or purchasing of any loan or extension of credit, for whatever purpose, whether secured by residential real estate or not, including but not limited to financial assistance for the purchase, acquisition, construction, rehabilitation, repair or maintenance of any real property or part or portion thereof."⁶ It also requires that its provisions be liberally construed.⁷

Although the LAD makes no mention of "disparate impact discrimination," the press release indicates that the Final Rule "codif[ies] existing case law." That case law, as explained in the preamble to the Final Rule, consists of court decisions going back at least 20 years in which New Jersey courts, including the New Jersey Supreme Court, applied the "disparate impact" test to determine whether claims of discrimination unsupported by proof of intent to discriminate could survive.⁸

Potential Impact on Lenders and Loan Servicers

Because the disparate impact test for discrimination under the LAD has been applied by New Jersey courts for at least 20 years and state fair lending laws are generally not preempted by federal law, the potential impact of the Final Rule on lenders making loans to New Jersey borrowers or secured by New Jersey properties would seem not to be significant. Nevertheless, there are provisions in the Final Rule that may present problems for lending institutions. These provisions – concerning burdens of proof, evidence needed to meet one's burden of proof, reliance on third parties, and the use of artificial intelligence (AI) and other automated decision-making tools – are addressed below.

Burden of Proof

In the housing and housing financial assistance contexts, the Final Rule imposes a two-step burden of proof, which favors persons claiming discrimination (Complainants) over those defending against such claims (Respondents). In other contexts – including employment, public accommodations and contracting – the Final Rule imposes a three-step burden, which is more favorable to Respondents.

The two-step process is as follows:

1. A Complainant must show that the challenged policy or practice has a disparate impact on members of a Protected Class. (For example, a lender's policy of not extending loans for homes under a certain value, e.g., $75,000, can disproportionately exclude minority applicants, who may live in neighborhoods within the lender's trade area that have lower property values.)
2. If the Complainant meets this burden, Respondent must then show that 1) the challenged policy or practice is necessary to achieve a Valid Interest, and 2) there is not a less discriminatory alternative means of achieving that interest.⁹ (In the above example, the lender may be able to establish that its minimum loan amount policy is necessary to make a profit, arguably a Valid Interest, by demonstrating that it loses money whenever it makes a loan less than $75,000. Under the two-step process, however, the lender would then be faced with a difficult task, proving a negative, i.e., that there is no less discriminatory alternative means of achieving that same Valid Interest.)¹⁰

The three-step process differs only in that, once a Complainant meets its initial burden and the Respondent follows by meeting its burden to show that its challenged practice or policy is necessary to achieve a Valid Interest, the burden shifts again, back to the Complainant to show that there is a less discriminatory alternative means of achieving that interest.¹¹

Evidence Required to Meet Burden of Proof

The Final Rule indicates that to meet its burden of proof, 1) a party must provide empirical evidence (i.e., that is not hypothetical or speculative) and may offer anecdotal evidence as well to support its allegations, and 2) the opposing party may rebut whether the party with the burden of proof has met its burden. The Final Rule also provides a non-exclusive list of numerous types of evidence that, depending on the facts of the case, may be relevant to establish or rebut the existence of disparate impact, including national/state/local statistics, demographic or census data, local agency data, survey data and "other relevant data."¹²

The Final Rule also indicates that the determination regarding whether a Respondent's policy or practice is necessary to achieve a Valid Interest and whether there is a less discriminatory alternative means of achieving that interest "requires a case-specific, fact-based inquiry."¹³ The logical consequence of this is slower and more protracted claims resolutions, greater expense and an increased likelihood of Respondents suffering reputational harm (even if eventually found not to have engaged in disparate impact discrimination).

Reliance on Third Parties

The Final Rule indicates that if a Respondent's policy or practice results in a disparate impact on members of a Protected Class, the Respondent cannot escape liability under the LAD simply by showing that its policy or practice relied on conduct, standards, products, procedures or systems of an outside person or vendor. Rather, it must take reasonable steps to ensure that the outside person's or vendor's conduct, standards, products, procedures or systems are consistent with the LAD and the LAD Regulation.¹⁴

Automated Decision-Making Tools

While the Final Rule addresses the potentially discriminatory use of automated decision-making tools, it does so only in the specific context of employment.¹⁵

However, in a comment submitted in response to the proposed rule, the American Civil Liberties Union of New Jersey (ACLU-NJ) requested DCR to "explore ways to compel covered entities to disclose information about their use of automated decision-making tools," and DCR responded that "these rules apply to the use of automated decision-making tools by covered entities in all contexts covered by the LAD, not just in the employment context," relying in part on guidance it issued in January 2025 on algorithmic discrimination (Guidance).¹⁶ Hence, it appears that this aspect of the Final Rule applies in the lending context.

The Guidance defined "algorithmic discrimination" to mean "discrimination that results from the use of automated decision-making tools" (primarily ones that use algorithms) and indicated that "the LAD applies to discrimination stemming from the use of automated decision-making tools in the same way it has long applied to other forms of discriminatory conduct."¹⁷ Specifically with regard to disparate impact discrimination, the Guidance stated:

Algorithmic discrimination constitutes disparate impact discrimination when an automated decision-making tool makes recommendations or contributes to decisions that disproportionately harm members of an LAD-protected class unless use of the tool serves a substantial, legitimate, nondiscriminatory interest. Even then, the use of the tool is prohibited if there is a less discriminatory alternative. In evaluating whether there are less discriminatory alternatives, whether the covered entity tested its automated decision-making tool for bias or evaluated alternatives may be considered as relevant evidence.¹⁸

The Guidance also indicates that "[i]t is critical that employers, housing providers, places of public accommodation, and other covered entities … carefully consider and evaluate the design and testing of automated decision-making tools before they are deployed, and carefully analyze and evaluate those tools on an ongoing basis after they are deployed," and that doing so "is necessary to decrease the risk of discriminatory outcomes and thereby decrease the risk of possible liability under the LAD."¹⁹

Despite that the Guidance admits that it is only "guidance" and further declares that it "does not impose any new or additional requirements that are not included in the LAD, does not establish any rights or obligations for any person, and will not be enforced by DCR as a substitute for enforcement of the LAD,"²⁰ lending institutions would be well advised to comply as best as possible with the Guidance and the aspect of the Final Rule that addresses the potentially discriminatory use of automated decision-making tools in the employment context.

The Bottom Line

Because ascertaining whether a proposed practice or policy will have a disparate impact on members of a Protected Class is a challenging task at best, many financial institutions no doubt viewed the Trump Administration's executive order as a favorable development. The Final Rule, however, effectively deprives New Jersey lenders of any potential benefit from actions taken by federal agencies in compliance with that order. It also codifies (for the first time in New Jersey) "disparate impact discrimination" as unlawful discrimination under the LAD and imposes on home mortgage lenders and servicers operating in New Jersey a more difficult burden of proof in a disparate impact case than is imposed on them under existing federal law.

It is therefore imperative for institutions lending to New Jersey residents or making loans secured by New Jersey properties to ensure, to the extent possible and practicable, that they are not engaging in discrimination of any kind – overt, disparate treatment or disparate impact – against members of any of the numerous Protected Classes listed in the LAD.

Notes

¹ 57 N.J. Reg. 12(2) (Dec. 15, 2025), amending N.J.A.C. 13:16-1.1 et seq.

² N.J.S.A. 10:5-1.1 et seq.

³ N.J.A.C. 13:16-2.1(a), (b).

⁴ N.J.A.C. 13:16-1.3.

⁵ N.J.S.A. 10:5-4, -12.

⁶ N.J.S.A. 10:5-12(i); see also, N.J.A.C. 13:16-1.3 (def'n of "covered entity").

[⁷ N.J.S.A. 10:5-3; N.J.A.C. 13:16-1.2(a) ("[T]he remedial provisions of the [LAD] will be given a broad construction and its exceptions construed narrowly.").

⁸ See, e.g., Gerety v. Atlantic City Hilton Casino Resort, 184 N.J. 391 (2005), in which the New Jersey Supreme Court applied the disparate impact test to determine whether the employment leave policy of a pregnant woman's employer violated the LAD. The court found that the challenged leave policy did not have a disparate impact on women since it applied equally to men (who could also experience serious medical issues requiring them not to work for extended periods of time). See also, In re Adoption of 2003 Low Income Housing Tax Credit, 369 N.J. Super. 2 (App. Div. 2004), in which the Appellate Division applied the disparate impact test in a housing context.

⁹ N.J.A.C. 13:16-2.3(a), (b).

¹⁰ Addressing a comment in this regard from the lending industry, the DCR pointed to a provision in the Final Rule that allows a Respondent to meet this burden by identifying "what alternative practice or policy options it considered and how and why it decided to select the practice or policy it chose."

¹¹ N.J.A.C. 13:16-2.2(a)-(c).

¹² N.J.A.C. 13:16-2.4.

¹³ N.J.A.C. 13:16-2.4(c), (d).

¹⁴ N.J.A.C. 13:16-2.4(e). The American Civil Liberties Union of New Jersey commented that taking such reasonable steps "could be read to create an unintended defense or safe harbor from algorithmic discrimination." The DCR disagreed, stating that in such a circumstance "[a] respondent may still be liable if unlawful discrimination results from [its] use of an outside person's or vendor's conduct." In addition, to provide "additional clarity," the DCR removed from the Final Rule the following example set forth in the proposed rule: "[I]f an employer uses an employee selection algorithm that results in a disparate impact based on gender or race, the fact that the employer did not create the algorithm itself, but instead used an algorithm created by a technology company will not shield the employer from liability, unless the employer took reasonable steps to ensure that the algorithm would not result in a disparate impact based on a protected characteristic and was otherwise consistent with the [LAD] and this [Final Rule] before using it." See Final Rule at pp. 41-43.

¹⁵ N.J.A.C. 13:16-3.2(c).

¹⁶ Final Rule at p. 41; the Guidance is available online.

¹⁷ Guidance at p. 9.

¹⁸ Guidance at p. 11.

¹⁹ Guidance at p. 13 (italics added). In a footnote, the Guidance offers a "non-exhaustive list of steps covered entities can take to identify and mitigate algorithmic discrimination from their use of automated decision-making tools." The list includes "implementing quality control measures for any data used in designing, training, and deploying the tool; conducting impact assessments; having pre-and post-deployment bias audits performed by independent parties; providing notice of their use of an automated decision-making tool; involving people impacted by their use of a tool in the development of the tool; and red-teaming their automated decision-making tools – or purposely attacking the tools to search for flaws."

²⁰ Guidance at p. 1, fn. 1.

Information contained in this alert is for the general education and knowledge of our readers. It is not designed to be, and should not be used as, the sole source of information when analyzing and resolving a legal problem, and it should not be substituted for legal advice, which relies on a specific factual analysis. Moreover, the laws of each jurisdiction are different and are constantly changing. This information is not intended to create, and receipt of it does not constitute, an attorney-client relationship. If you have specific questions regarding a particular fact situation, we urge you to consult the authors of this publication, your Holland & Knight representative or other competent legal counsel.