The Federal Housing Finance Agency (FHFA) is gathering public input on modernization proposals that could bring significant changes to the real estate appraisal process for loans backed by Freddie Mac and Fannie Mae.
Accelerated in part because of the pandemic, the organization that regulates the mortgage financing giants has announced that it’s weighing possible changes that could increase the accuracy and speed of the valuation process. The move could open up new opportunities, and create unintended risks, for real estate investment, origination, and more.
Proposals include wider use of appraisal waivers that rely on automated valuation models (AVMs) that draw from existing property data rather than a human appraiser. Or “hybrid appraisals” in which a third-party may collect property information on a home and send it to a qualified appraiser who would then determine value without actually seeing the property in person.
Reimagining Real Estate Appraisals
As the FHFA points out in its Request for Information on Appraisal-Related Policies, Practices, and Processes, the US response to COVID-19 demonstrated how a more flexible appraisal process can assist the flow of liquidity to the housing market.
Just because real estate appraisers were generally deemed part of the essential workforce under COVER-19 restrictions, didn’t mean borrowers and appraisers were willing to allow or perform full interior property inspections.
With mortgage rates still quite low, a lack of appraisers—especially in rural and high-volume areas—are putting pressure on turnaround times that won’t subside with rising vaccinations. Appraisals, which can often run $500 to $800 or more in some areas, have been growing costlier and more time-consuming for years. It’s safe to assume that the situation will only get worse.
With this in mind, here’s our perspective of some of the questions and major themes that stand out in the FHFA’s RFI.
Opportunities for Process Improvements Through Non-Traditional Valuation Services
It’s clear the FHFA recognizes the need to address industry-identified issues of appraiser capacity, turn-times, and training—especially in those aforementioned rural and high-volume areas. It’s also clear that the pandemic accelerated existing and increasingly costly trends that point to the need for new valuation solutions.
While the current disparity in supply and demand might be great for appraisers, it’s potentially disastrous for transacting parties seeking a clear and accurate read on property valuation, quickly and cost-effectively. The ranks of qualified appraisers are thinning out, while market pressures continue to increase at the pace of technological change.
Current appraisal models are unlikely to catch up to demand. And there are myriad opportunities for process improvements that allow non-traditional valuation services to augment traditional appraisals.
For example, remote inspection technologies can facilitate equivalent valuations in shorter turn times, with less risk. In our view, a remote inspection from objective sources paired with AVMs and other tools is the future, regardless of whether stand-alone or hybrid scenarios are adopted.
Geospatial property analytics, for instance, leverage high-resolution aerial imagery and computer vision to extract structured data from overhead visuals. This includes details such as the type of roof construction and its condition, the presence of solar panels, whether there are overhanging trees, the overall size of the property, and whether the property has a pool, yard debris, and more. These same technologies can be used to analyze both the subject property as well as identify the most comparable properties to be evaluated as part of the valuation process.
This can minimize or remove the need for appraisers to climb roofs—instead, they can access current, objective data from overhead imagery to improve the resulting appraisal. By offering appraisers, lenders, investors, underwriters, and others access to current and accurate property condition data, these technologies can help users identify when there is a compelling need for in-person inspections. This would allow appraisers to focus on using their expertise and judgment to appraise more atypical homes.
Performance Implications of Alternative Appraisal Solutions v. Traditional Appraisals
Technologies already exist that enable users to remotely and instantly gather geospatial property condition data on tens of millions of homes across the US—delivered with the speed and coverage of traditional property record data. This includes access to neighborhood, census tract, zip code, or CBSA data, both current and aggregated over time to unearth trends that might otherwise be extremely hard, if not impossible, to do with any traditional appraisal model.
To be sure, the expertise of professional appraisers is extraordinary and valuable—but also subjective. While it is the matter of price and scale that present the greatest challenges to those who must rely on real estate appraisals in financial decision-making, it is not the only consideration.
By way of comparison, property condition data gathered and analyzed using modern AI technologies can dramatically enhance the utility and accuracy of AVMs. In one recent evaluation, geospatial analytics-based solutions showed a 7.7% improvement on PPE10 predictions of on-market valuations.
Controlling for Bias to Remove Racially Discriminatory Practices
Reports of racial discrimination in appraising home values have been gaining a lot of national attention. That includes a recent instance reported by Bloomberg in which a black couple saw the appraisal of their home increase by $500,000 after they took down photos of themselves in the house and had a white friend pose on their behalf during the appraiser’s visit.
Unfortunately, this isn’t a matter of anecdotal aberrations. It’s reflective of systemic biases throughout society. In its RFI, the FHFA is wise to ask for input on discrimination in current collateral valuation practices, as well any concerns that alternative or automated solutions could have their own discriminatory impact.
As Bloomberg reports, even the Appraisal Institute, an international professional association representing real estate appraisers, has come to acknowledge that this kind of bias does exist. To its credit, the organization is ramping up training on unconscious racial bias, updating ethics standards, and more.
But machine learning enables the reduction or even the elimination of racial discrimination in the valuation process as a policy decision that is not reliant on attempts to produce a shift in human biases.
That is not to say discrimination, racial and otherwise, is absent in AI-based solutions. The algorithms these technologies use are only as good as the scale and quality of the dataset they draw from, and the expertise of the data scientists that train them.
The difference is that these technologies also offer a controllable and measurable avenue for monitoring and correcting for biases progressively over time—as a matter of policy, not just positive intention. As a priority in modernization, that by itself could be a game-changer.