AVM Compliance for Home Equity Lenders
In recent years, lenders have increased their use of AVMs in the home equity loan origination process. This makes sense as AVMs are a fast, cost-effective way to value a property. However, regulatory scrutiny of AVMs has also increased in recent months and years. With regulators focusing more on AVM accuracy and bias, lenders should ensure they are aware of regulatory developments, and are taking the correct steps to ensure compliance.
Regulatory Background
In June 2023, six federal agencies requested comment from the public on a newly-proposed rule on AVMs used to determine the value of collateral in loan originations. The proposed rule requires lenders to “adopt policies, practices, procedures, and control systems to ensure that AVMs adhere to quality control standards designed to ensure the credibility and integrity of valuations”.
The agencies’ stated goal is to increase confidence in the use of AVMs through steps such as avoiding conflicts of interest, requiring random sample testing and reviews, and promoting compliance with applicable nondiscrimination laws.
The challenge for home equity lenders relying on AVMs is that the proposed rule does not provide specific guidance on how lenders can ensure compliance. In the absence of such guidance, we recommend that lenders take the following steps to protect themselves.
AVM Selection
Selecting the correct AVM is the first step in ensuring an accurate and unbiased valuation. Lenders should ensure they are using AVMs with the following characteristics:
Cascades
To increase accuracy, lenders should not rely on one single AVM model. Instead, they should select an AVM that cascades through multiple models to choose the most accurate one based on a variety of factors including confidence scores, geography, property type and price tier.
County-Level Model Selection
AVM model performance varies significantly by county, so lenders should not rely on one single model across their geographic footprint. In the most recent testing performed by AVMetrics, a leading third-party AVM testing company, they identified 19 different AVMs that performed best in at least one county in the US. Therefore, lenders should choose an AVM cascade that is customized at the county level.
Quarterly Updates
AVMetrics’ testing also uncovered that AVM performance changes frequently over time as market conditions change and AVM providers update their models. Therefore, lenders should choose an AVM cascade that is tested and updated on a quarterly basis to avoid relying on outdated testing results that will decrease accuracy.
The Interagency Appraisal and Evaluation Guidelines also indicate that lenders should not rely solely on the results of an AVM without also considering information on the property’s actual physical condition, including evidence of a property inspection and photos of the property. Therefore, lenders should strongly consider obtaining a Property Condition Report (PCR) in addition to an AVM to ascertain the actual physical condition of the property.
AVM Testing for Accuracy
After selecting the correct AVM to use, lenders should also ensure that this AVM is tested regularly by an independent third-party such as AVMetrics. This allows lenders to avoid any appearance of a conflict of interest, and to provide regulators with proof of independent testing for accuracy.
It is also important for lenders to work with third parties using robust AVM testing methodology. Testing should include AVM values and benchmark sales in large quantities, at least hundreds of thousands if not millions of properties. Foreclosure sales and other bad benchmarks should be removed. Finally, the AVM’s value should be tested before the corresponding benchmark sale is made public. If the AVM value is tested after the sales price is public, the model will have already incorporated that data, thereby invalidating the test.
AVM Testing for Bias
When it comes to racial bias, AVMs do have several inherent benefits as compared to appraisals. They are blind to borrower and neighborhood demographic characteristics. They also do not rely on human assessment to develop a conclusion of value. However, the concern is that bias could be present in the underlying data fueling AVM models, thereby biasing the AVM results as well.
This is a more challenging area to address than accuracy, and AVM builders and testing providers are still working to identify the best way to combat potential AVM bias. At this point, there is not a consistent industry standard on how to approach this. Therefore, lenders should focus on selecting and testing the most accurate AVM, and continuing conversations with their AVM and testing providers on any emerging developments on the bias testing front.
Summary
While home equity lenders have good reason to use AVMs in their loan origination process, they also need to be aware of increased regulatory scrutiny on AVMs. To ensure compliance, lenders should ensure they are selecting a high-quality AVM cascade that varies by county and is regularly tested by an independent third-party.
CSS offers its lender partners the Advantage AVM Cascade, which is tested on a quarterly basis by AVMetrics. We also offer PCRs to supplement AVMs with a property inspection. For more information, please contact sales@visitcss.com.
