The Loan Servicing Edition

Are Properties Ever Unsuitable for an AVM?

Are Properties Ever Unsuitable for an AVM? 

A White Paper about Automated Valuation Modeling  | The Loan Servicing Edition
By Robert L. Walker, CMB, CMT, MBA

This second white paper entitled, “Are Properties Ever Unsuitable for an AVM?” focuses on the reality that, while traditional AVM cascades can provide a value on a property, that value will be unreliable for business use in many cases. The problem most users have is that they can’t easily identify the individual AVM property values they should or should not use; therefore, they wind up delaying decisions while they seek and pay for additional property valuation data in the form of a BPO, opinion of value, or a variety of hybrid and traditional appraisal products. The inherent delays and additional costs associated with this process increase loan origination and servicing expenses, as well as causing production and servicing delays, which adversely impact customer service levels and increase risk.

When an AVM Isn't a Suitable Valuation Tool

For many years, equity lenders and loan servicers, among others, have used AVMs and AVM cascades to value properties for driving lending decisions or value-based business decisions. Clearly, AVMs have served the community well by saving a great deal of time and money compared to other forms of valuations like Broker Price Opinions, opinions of value, or appraisals. The strategy of the AVM cascade has also served the industry well by generating geographic coverage and reasonable levels of valuation accuracy using Macro level data. The more recent the available sales or listing data is in an MSA, the higher probability that the model can impute a value for a property of similar characteristics. At the most basic level, an AVM cascade is the selection of different AVM products that have been tested and have been shown to be reasonably accurate at the county level. For example, AVMs A, B and C may each return values for properties in Orange County, California, but AVM product C, on average, seems to get closer to the true value of a property more often than products A or B. Therefore, Product C will typically be run first, followed by AVM product A and AVM product B in a cascade if Product C doesn’t deliver a value or “hit.” The resulting AVM, be it C in first position, or products A or B in subsequent positions within the cascade, is then delivered to the Servicer.

Wouldn’t it be nice if home values were always consistent and that data at the Macro level always translated to its accuracy at the property level? Unfortunately, it does not.

Understanding the Nature of Non-suitable Properties.

At Veros, we asked the question, “what if we could identify individual properties that are unsuitable for an AVM, and tell our clients which properties those are, and then only charge them for those AVMs completed for AVM suitable properties?”

Get the White Paper

If you'd like to test the AI-powered AVM suitability tool or discuss how VeroPRECISION can help your organization, contact us today.

Toll Free:  866.458.3767 Option 2 
Email us: communications@veros.com