Jon Wierks of First American Data & Analytics on the Future of AVMs

Jon Wierks heads development of valuation analytics for First American Financial Corp., Santa Ana, Calif., including the company’s automated valuation models. He has played a role in developing many of the industry’s most commercially successful AVMs for the past 25 years. He can be reached at jwierks@firstam.com.

MBA NEWSLINK: AVMs have been widely used by lenders for decades, but you think AVM usage will spike in the coming year: why is that?

JON WIERKS: AVMs are very versatile tools, and they align extremely well with a variety of market needs in 2023 and beyond. In a tough market cycle, like the one we’re in, every client is looking for ways to cut costs and accelerate decision making to help compete for the business that’s still out there. As a result, lenders are thinking of new and expanded ways to use AVMs to help.

Home equity lending is one of the leading opportunities at the moment, and lender-quality AVMs are being used, or at least considered, by a growing number of lenders. They are a fraction of the cost of full appraisals, can deliver accurate values in seconds – rather than days or weeks – making them a great choice for loans and lines of credit that fit certain criteria, such as low loan-to-value and high-credit borrowers.

Appraisals and hybrid tools will still be used on larger credit lines and for tight LTV situations, but most lenders could triage a meaningful percentage of their workflow with AVMs and significantly reduce costs. Considering lenders often absorb the cost of originating home equity lines of credit, increasing efficiency in origination is top of mind.

AVMs are also useful in identifying prospects for home equity lending. A lender can run them against its portfolio and instantly identify their best prospects for a HELOC. For example, if a lender wants to know how many homeowners within their bank’s geographic footprint have 60% equity, they can use AVMs to identify them.

The final area where we expect to see growth in AVM usage is in portfolio surveillance. If the real estate market continues to decline, portfolio monitoring will become critical, and lenders will want to understand their aggregated risk and the markets where that risk is concentrated.

NEWSLINK: What do you say to lenders, and skeptics, who still question the accuracy and reliability of AVMs? What will it take to win them over?

WIERKS: As someone who has developed AVMs for more than 25 years, I’ve seen the evolution of AVM technology and their quality. The latest models are the most advanced and accurate yet. For example, our newest AVMs employ an ensemble modeling approach that combines machine learning with modeling techniques, such as hedonic models, tax assessment models and enhanced HPIs.

Every night we retrain our models and run the AVMs against every residential property in the country. So, the models are constantly learning from an incredible amount of the freshest data.

If a lender is curious about accuracy and hit rate performance, comprehensive testing results are available to help analyze AVM performance by geography, price tiers and other metrics.

NEWSLINK: How will AVMs perform in a changing real estate market?

WIERKS: The key to maintaining AVM accuracy in a changing market is how fast the AVM catches the change in price trajectory, as it goes from rising to flat, or flat to declining. It comes down to how fast the AVM receives updated data, particularly the most predictive data. Models that are fed the latest sales-to-list ratios, inventory trends, days on market, and other data points can detect change earlier and more clearly. Just providing the model with average price per square foot, sales or listings alone, for example, doesn’t give the model a clear picture on what’s actually going on in a market.

AVM developers that are also data providers will get their data faster than developers who are buying data. This is another factor that lenders should consider.

NEWSLINK: As someone who has been involved in AVM development since its beginnings how have you seen these models evolve, and what is the “new, new” thing for AVMs in 2023 and beyond?

WIERKS: Currently, most AVMs can estimate the value of a subject property, but not the current condition of the subject property, nor all of the comparable properties used to value it. As any appraiser will tell you, there are external factors that influence value. Noise factors, such as nearby busy roads, airports, or railroads diminish value, while nearby leafy streets or waterfronts, or properties adjacent to golf courses tend to increase value.

In my opinion, capturing property conditions, external factors and geospatial information is what will elevate AVMs to the next level of performance and we’re actively working on integrating more of this data into our models.

(Views expressed in this article do not necessarily reflect policies of the Mortgage Bankers Association, nor do they connote an MBA endorsement of a specific company, product or service. MBA NewsLink welcomes your submissions. Inquiries can be sent to Michael Tucker, Editor, at mtucker@mba.org.)