Proponents of automated valuation models and hybrid, data provider-assisted appraisals hope to dovetail with efforts to address potential bias in human appraisals. News reports of appraisal bias, along with a Freddie Mac study that found appraisals were more likely to fall short of the contracted sales price in minority neighborhoods, join academic research in suggesting there may be widespread bias in appraisals.

Questions remain over how extensive the problem is, and what role appraisers play, versus lenders, regulators or real estate agents. Appraisers have said they merely reflect a market that is already unequal.

Despite the unknowns, regulators have promised to take action. The Department of Housing and Urban Development is spearheading an inter-agency task force to make recommendations on how to address appraisal bias, and although the industry awaits their determination, industry stakeholders have said they were impressed with the level of engagement and the seriousness of HUD’s effort.

Tech firms in the appraisal business — like HouseCanary — hope to be included in a strategy to address bias, if not hailed as its outright solution. Some observers have said that technology cannot diversify the appraisal profession, or remove the years of redlining that devalued minority neighborhoods.

“We need more imaginative, thoughtful, dynamic workforce development patterns, not just technology and automation as a response to the challenge,” Cy Richardson, a senior vice president at the National Urban League, said last month at an appraiser listening session hosted by the Appraisal Subcommittee.

Whether automated valuations are a solution to appraisal bias remains to be seen. It does, however, present a much cheaper and faster alternative. While a typical appraisal could cost $400 to $500 and take several weeks, HouseCanary says it can perform a “condition-informed evaluation” within one to four days, for $100.

HousingWire had a chance to speak with HouseCanary CEO Jeremy Sicklick about how technology could change appraisals — or not. Read the interview below.

This interview has been edited for length and clarity.

Georgia Kromrei: How do you disentangle the question of bias from the legacy of discriminatory practices that devalued minority neighborhoods?

Jeremy Sicklick: A lot of the bias is built in to how comparables are selected. Whether it’s an inappropriate comparable due to gross living area, or pulling properties that are dissimilar by age. Freddie Mac found that in more minority areas the comps were on the lower end of the range. In white areas they were on the higher end of the range. In more minority areas, comps were consistently closer to the subject property. In white areas they were further away.

We select comparables based on a mathematical similarity score, on a 1 to 100 scale. What we’re looking for is most similar properties in terms of size, condition, distance etc. … but what we’ve found in comparing predominantly minority versus predominantly white, there was no bias in how we select comps. You have to get the comp selection right to ensure there’s no bias.

GK: But even if you get comp selection right, if everything is done according to the proper process — how can you take out the historical bias and discrimination?

JS: We try to figure out what a property is worth when it trades. That’s what we’re focused on. When buyer and seller meet and agree to a price — that’s what we’re trying to find. In our models we do not leverage information which has an inherent bias like household race, ethnicity, income that would violate fair lending laws.

Imagine I have two properties that are exact matches, they’re each 2,000 square feet, two bedrooms, two bathrooms, constructed in the 1980s, one in neighborhood A worth $200,000, and one in neighborhood B, a $300,000 property.

Value is made up of two things: land value and structural value. The structural value is the same. The land value is different. That is a locational difference. And in real estate everything is location, location, location.

But you can at least understand there’s a locational difference, and put in place incentives to close it up, like the Opportunity Zones program. That’s at least one way the government has tried to address locational differences in the past.

GK: The FHFA raised some concerns about appraisals performed with third-party data providers, specifically about diffusing the responsibility in valuations. Whose responsibility is it if something goes wrong?

JS: A lot of FHFA’s pushback was related to the inspection itself. In our experience, the inspection by a trained inspector or Realtor can be trusted. They’re already doing inspections on evaluations that are heavily used by Wall Street today that we are a part of.

(HouseCanary provided valuations for Blackstone-founded Invitation Homes, the largest owner of single-family homes in the country.)

Once you train an inspector to go in and get property details and capture that in an app, it can be a very trusted process. There’s nothing about an appraiser that’s better than someone you’ve literally trained for a few days. We see real life examples of the inspection being done well.

GK: What about this question of whether borrowers might be more likely to dispute a value they don’t like if it is not a traditional human appraisal. “I want a real appraiser,” they might say. Could you see that happening?

JS: There could be that pushback. One thing about where modern tech like us comes in — the appraisal is not just a number in a range. We provide 30 or 40 page reports that have all the comps available in that area.

You get a lot of really detailed information around not only the value, but the context, that helps explain why that is the value. So, of course, if you don’t like the value, you can try to get a different one. But you really have to start to wonder, if that’s not the value, then were those not the comparables? Why is your property trading higher than any other property in the neighborhood? There are layers of context, and it gets harder and harder to change that, because we use 500 comparables and 50 that were very close, whereas a historical appraiser would use three comps.

GK: What do you hope to see come out of the PAVE task force?

JS: I do think that technology, including automated valuation, can be part of the solution to address racial bias in appraisal. Where the automated valuation tech is used in a responsible way, we believe it can be part of the solution to addressing appraisal bias.

Modern technology should be used where we have enough comps and we understand what’s happening in the market enough to have real confidence in that property’s value in a tight range.

Where appraisals are still used, we believe modern technology could be used as an independent third-party check where there has been appraisal bias.