CFPB and Other Federal Regulators Consider Regulations to Reduce Algorithmic Bias in Automated Home Ratings | Jenner and Block

Automated assessment models

AVMs are defined by law as a “computerized model[s] used by mortgage originators and secondary market issuers to determine the collateral value of a mortgage secured by a consumer’s principal residence.[3]

According to the CFPB, AVMs are increasingly being used to appraise homes, a trend driven “in part by advances in database and modeling technology and the availability of larger property datasets.”[4] The benefits of better AVM technology and increased data availability are their potential to reduce home appraisal costs and turnaround times. However, like algorithmic systems in general, the use of AVMs also presents risks, including data integrity and accuracy issues.

Additionally, there are concerns that AVMs “may reflect bias in design and function or through the use of biased data[,] [] may introduce potential fair lending risk.[5] Due to the “black box”[6] nature of the algorithms, regulators are concerned that “without proper safeguards, flawed versions of these models could numerically delineate certain neighborhoods and further embed and perpetuate historical disparities in lending, wealth, and home values.”[7] “Overvaluing a home can potentially lead the consumer to take on more debt, increasing the risk to their financial well-being. On the other hand, undervaluing a home can result in a consumer being denied access to credit for which they were otherwise qualified, which could result in a rescinded sale or credit being offered on less favorable.[8]

The proposed rule

On February 23, 2022, the CFPB published a 42-page overview, detailing several possible options for regulation, which provides insight into the agencies’ thinking as to the scope of future regulation.

The proposed rule will be a joint interagency rule, as the CFPB retains enforcement authority over noncustodial institutions, while the Federal Reserve Board of Governors, Comptroller of the Currency, Federal Deposit Insurance Corporation, National Credit Union Administration, and the Federal Housing Finance Agency retain enforcement authority over “insured banks, savings associations, [] credit unions[,] . . . [and] federally regulated subsidiaries owned and controlled by financial institutions.[9]

To address concerns about data integrity, accuracy and reliability, the CFPB is considering two options, one “principled” and the other “prescriptive”. A principles-based approach would require entities to maintain their own “AVM policies, practices, procedures and control systems” to meet the first four quality control standards noted above.[10] The CFPB recognizes that this may be preferable as, in general, stringent requirements may not be able to keep up with changing technology and could be a significant burden on smaller entities. On the other hand, if agencies decide to enact a prescriptive rule, they consider requiring controls related to “fundamental errors” that could produce inaccurate results, “monitoring of availability management, usability, integrity and security of the data used”, a clear separation between the people “who develop, select, validate or monitor an AVM” and the employees involved in the “loan origination and securitization process”, and the ongoing validation of entity MAVs through “random sample testing and examination”.[11]

As part of the same proposed rule, the CFPB and the aforementioned federal regulators are also considering adding a non-discrimination quality check under their authority to “take into account any other such factors.” . . determine[d] be appropriate. »[12] The CFPB recognizes that a stand-alone non-discrimination factor may be unnecessary as non-discrimination may already be encompassed in three of the first four quality controls stipulated by law. Additionally, AVMs are subject to federal non-discrimination laws such as the Equal Credit Opportunity Act (ECOA) and the Fair Housing Act (FHA). However, the CFPB argues that “an independent requirement for institutions to establish policies and procedures to mitigate fair lending risk in their use of AVMs. . . . can help ensure the accuracy, reliability and independence of AVMs for all consumers and users. »[13]

To combat discrimination in lending, federal regulators are considering both a flexible, principles-based approach, similar to the approach described above, and a prescriptive non-discrimination rule. A principles-based approach would provide businesses “the flexibility to design policies, practices, procedures and fair lending control systems appropriate to their business model”[14] and “depending on an institution’s risk exposure, size, business activities, and the extent and complexity of its use of AVMs”.[15] In contrast, a prescriptive rule “would specify[] AVM development methods (e.g., data sources, modeling choices) and AVM use cases” to mitigate the “risks that lending decisions based on AVM outputs will generate illegal mismatches”.[16]

Last month’s announcement was prompted by the CFPB’s requirement to convene a small business review panel before releasing a proposed rule that “could have a significant economic impact on a significant number of small entities.”[17] The plan released by the CFPB sought feedback from small businesses, such as mortgage brokers with annual revenues of $8 million or less, building societies with annual revenues of $41 million or less, $5 million, and secondary market finance companies and others. non-custodial financial intermediation companies whose annual revenue is equal to or less than 41.5 million dollars. For these smaller entities, the plan presents more than forty questions and a first opportunity to influence the rule-making process.[18]

Next steps

As the CFPB diagram shows, much remains up in the air as regulators continue to consider their options. Since the CFPB is subject to enhanced regulatory processes for regulations affecting small entities, we have this first look at the agencies’ thinking about the AVM algorithmic bias. Over the next few months, the CFPB will convene the Small Business Review Panel, release the panel’s report, and work with its federal partners to draft a proposed rule that is subject to the standard notice and comment process.


[1] The CFPB shares enforcement authority over AVMs with the Board of Governors of the Federal Reserve System, the Comptroller of the Currency, the Federal Deposit Insurance Corporation, the National Credit Union Administration, and the Federal Housing Finance Agency.

[2] The Dodd-Frank Wall Street Reform and Consumer Protection Act requires federal regulators to adopt rules ensuring that AVMs meet certain quality control standards designed to: “(1) ensure a high level of confidence in estimates produced by automated valuation models; (2) protect against data manipulation; (3) seek to avoid conflicts of interest; (4) require random sample testing and examination; and (5) take into account any other such factors that the agencies deem appropriate. 12 USC § 3354(a) (2010).

[3] § 3354(d).

[4] End consumer. Prot. Bureau, Overview of Proposals and Alternatives Under Consideration, Small Business Advisory Review Panel For Automated Valuation Model (AVM) Rulemaking, 2 (February 23, 2022) https://files.consumerfinance.gov/f/documents/cfpb_avm_outline-of-proposals_2022-02.pdf.

[5] Identifier. at 24.

[6] Identifier.

[7] Press release, Consumer Fin. Prot. Bureau, Consumer Financial Protection Bureau outlines options to prevent algorithmic bias in home appraisals (February 23, 2022), https://www.consumerfinance.gov/about-us/newsroom/cfpb-outlines-options-to-prevent-algorithmic-bias-in-home-valuations/

[8] End consumer. Prot. Bureau, Overview of Proposals and Alternatives Under Consideration, Small Business Advisory Review Panel For Automated Valuation Model (AVM) Rulemaking, at 24.

[9] Identifier. at 2 o’clock.

[10] Identifier. at 21.

[11] Identifier. at 22.

[12] 12 USC § 3354(a) (2010).

[13] End consumer. Prot. Bureau, Overview of Proposals and Alternatives Under Consideration, Small Business Advisory Review Panel For Automated Valuation Model (AVM) Rulemaking, at 25.

[14] Identifier.

[15] Identifier.

[16] Identifier.

[17] Identifier. at 3.

[18] Identifier. at 29 years old.

Comments are closed.