Regulators released the final version of the new rule covering the use of automated valuation models or AVMs, which requires that lenders using AVMs ensure their models are reliable, secure, and nondiscriminatory. The finalized version does not contain significant changes from the proposed rule released in June 2023. This rule comes on the heels of a recent rule about reconsiderations of value, a CFPB fair lending suit against an appraiser and lender, and regulatory guidance about the use of artificial intelligence in loan decisioning, indicating that these areas continue to be high on regulators’ lists of concerns.
AVMs are computerized models used to determine the value of a property. The new rule will apply when lenders use AVMs to assess the value of a consumer’s principal dwelling for the purpose of making a credit decision. It will also apply to ‘covered securitization determinations,’ which will include GSEs using AVMs in determining whether to waive an appraisal as well as determinations regarding structuring, preparing disclosures for, or marketing mortgage-backed securitizations.
When using AVMs for this purpose, the rule requires the implementation of quality control standards designed to accomplish five goals:
- Ensure a high level of confidence in the estimates produced;
- Protect against the manipulation of data;
- Seek to avoid conflicts of interest;
- Require random sample testing and reviews; and
- Comply with applicable nondiscrimination laws.
While the new rules state that the quality control standards should be appropriately tailored to the size, complexity, and risk profile of the institution as well as the transactions for which it uses AVMs, it does not contain a small creditor exemption. The 190 pages of the rule also provide limited guidance on what these goals should look like in practice or how they can be adjusted for smaller banks. The rule’s 12-month implementation period is also a fairly short time for banks to develop robust programs for AVM quality control.
It appears that the last two items – sample testing and nondiscrimination – are the most likely to present challenges to small banks. The guidance does not set standards for sample testing and reviews, which will likely lead to some confusion among banks regarding what would be an adequate sample size, what is required for a review, and what standards should be met for an AVM to “pass” the testing. The fifth goal, ensuring nondiscrimination, presents overlapping concerns in that smaller banks will be lending in smaller volumes and smaller geographic areas, which will make it more difficult to identify fair lending issues in AVMs.
The agencies did recognize in the guidance that a standard setting organization (SSO) and third-party testing would likely be useful in this context but declined to take further action toward encouraging the development of such an organization.
What the rule does make clear is that regulators continue to be concerned about third party oversight, computerization, and fair lending, particularly as it affects access to housing. Taken together with other recent guidance and regulatory actions, it certainly seems that this will continue to be an area of regulator focus to which banks should pay specific attention as part of their compliance management.