Automated Underwriting: Is it a Fair Lending Concern?
In April, the Consumer Financial Protection Bureau (CFPB) and several other agencies issued a joint statement that warned about potential bias in automated systems. Additionally, NAFCU posted an article about the Federal Reserve’s concern on fair lending risks associated with using artificial intelligence (AI). Of note, Michael Barr, the Federal Reserve Vice Chair for Supervision, stated that AI techniques “carry risks of violating fair lending laws and perpetuating the very disparities that they have the potential to address.”
However, recently, NAFCU has heard that several credit unions have been under scrutiny from examiners for potential fair lending violations when determining if an automated system should be applied to an applicant at all. Specifically, NCUA examiners have applied regulatory scrutiny when a credit union has a policy of putting applications from applicants in certain age groups through a manual decision process rather than through an automated underwriting system (AUS). For example, an AUS that is set up to not provide an automatic decision for anyone below the age of 21 or over the age 75. Apparently, examiners are providing this scrutiny even when the ultimate result (i.e., whether the application is approved or denied) is the same. According to examiners, the fact that they were treated differently (i.e., put through a different loan approval process) based on their age violates fair lending rules. Credit unions may want to check their AUS to determine if age parameters are set or perhaps even a default feature, as applying such parameters could increase a credit union’s fair lending and compliance risk.
With this in mind, let’s review how the regulations address potential age discrimination:
The general rule in section 1002.6(a) of Regulation B concerning the use of information to evaluate applications provides that “a creditor may consider any information obtained, so long as the information is not used to discriminate against an applicant on a prohibited basis Section 1002.2(z) defines the term “prohibited basis” to include age, among other characteristics, such as race, religion, national origin, sex and more. Section 1002.6(b)(2) provides specific rules concerning use of age information, and states:
“Except as permitted in this paragraph, a creditor shall not take into account an applicant’s age (provided the applicant has the capacity to enter into a binding contract).” (Emphasis added).
Despite this general prohibition, the regulation then goes on to discuss some situations in which age may factor-in to a decision:
ii. “In an empirically derived, demonstrably and statistically sound, credit scoring system, a creditor may use an applicant’s age as a predictive variable, provided that the age of an elderly applicant is not assigned a negative factor or value.
iii. In a judgmental system of evaluating creditworthiness, a creditor may consider an applicant’s age. . . only for the purpose of determining a pertinent element of creditworthiness.”
In any system of evaluating creditworthiness, a creditor may consider the age of an elderly applicant when such age is used to favor the elderly applicant in extending credit.” The commentary to this section provides more details when a credit scoring system considers age as part of its scoring system. Specifically, it provides examples and scenarios when the use of different scorecards based on the age of an applicant may or may not comply with the Act. The first scenario described is not deemed to score age because the system uses a card covering a wide age range that encompasses elderly applicants. In the second scenario described below, the example is deemed to score age and the creditor must ensure that the system does not assign a negative factor or value to the age of elderly applicants.
“In such a system, one card may cover a narrow age range (for example, applicants in their twenties or younger) who are evaluated under attributes predictive for that age group. A second card may cover all other applicants, who are evaluated under the attributes predictive for that broader class.”
“[I]f a system segments the population by age into multiple scorecards, and includes elderly applicants in a narrower age range, the credit scoring system does score age.”
In addition to the situations described above when a credit union may consider age, section 1026.51(b) of Regulation Z provides rules specifically affecting young credit card applicants. The commentary to this section states:
“Relation to Regulation B. In considering an application or credit line increase on the credit card account of a consumer who is less than 21 years old, card issuers must comply with the applicable rules in Regulation B. A card issuer does not violate Regulation B by complying with the requirements in this section.”
The rules included in this section require “financial information indicating the consumer has an independent ability to make the required minimum periodic payments on the proposed extension of credit in connection with the account,” or has a creditworthy co-signer, guarantor, or joint applicant with the financial information which indicates the ability to make the required minimum payments.
Finally, NCUA has published a Fair Lending Guide which provides additional resources on this topic.
Editor’s Note: Shortly after this blog was written, but before it published, the NCUA officially notified credit unions that it views age parameters in Automated Underwriting Systems to be a violation of the Equal Credit Opportunity Act and Regulation B. Additionally, NCUA referenced a February 2022 Letter to Credit Unions which stated:
“A credit union cannot disqualify a loan applicant for automatic loan approval based on the applicant’s age, provided the applicant is of legal age to enter into a binding contract, even if the credit union subsequently approves the application following a manual review of the application file. Credit unions using automated underwriting systems should ensure the system’s settings comply with ECOA’s requirements and do not result in age discrimination.”
Compliance Blog Mentions: