Better Markets filed a Comment Letter to the six federal banking regulators in response to their proposal to implement quality control standards for automated valuation models (AVMs) that are used to determine residential mortgage loan collateral values.
Why It Matters. The Agencies’ proposed rule for AVM quality control standards is long overdue and necessary to ensure that all Americans are protected and treated equitably in the pursuit of the American Dream of homeownership. Consistent with the Dodd-Frank Act’s direction, the proposal states that institutions using AVMs must ensure a high level of confidence in the estimates produced by AVMs. In addition, institutions must protect against the manipulation of data, seek to avoid conflicts of interest, and conduct random sample testing and reviews. Finally, institutions must comply with applicable nondiscrimination laws. Since 2010, supervisory guidance on the use of AVMs has been in and this rule would finally make quality control standards enforceable by regulatory agencies.
What We Said. While the proposal is a step in the right direction, it must be strengthened to correct the serious lack of specificity on minimum acceptable standards before it is finalized. Allowing individual lenders to determine appropriate levels for compliance with quality control standards and expecting multiple regulatory agencies to coordinate and assess compliance is impractical, irresponsible, and dangerous. The American people deserve to receive the same level of protection during the mortgage lending process whether they are obtaining a loan from a large mortgage originator or a smaller bank.
Bottom Line. Regulators must remain vigilant to detect discriminatory lending practices to ensure that all borrowers are treated fairly. As AVMs become more complex and employ more data, bias and discrimination becomes more difficult to detect. Furthermore, AVMs by their very nature, rely on historical data. Numerous studies have shown that historical data on home prices, one of the key inputs to an AVM, are influenced by racial and other types of bias. Regulatory agencies, therefore, must stay aware of advancements in modeling practices and technology to have the appropriate skills and resources to assess AVMs accurately.”
Read the letter here or click the button below.