Why the Housing Industry Should Not Depend on AVMs

NOV 15, 2012 2:21pm ET

While automated valuation models are industry tools used in the traditional property valuation process, they lead to flawed results for determining an accurate value for a specific property. AVMs are data based computer programs using real estate information such as tax assessments, comparable sales, property characteristics, MLS boards and price trends to provide an estimate of value for properties. Simply put, these are variables that may yield very dubious results. The industry should do away with these tools for three main reasons: lack of data integrity, poor interpretation of available data and overlooked market influences.

One key attribute lost when relying on AVMs is data integrity. It is the old garbage in/garbage out theory. The data that AVMs draw from is only as good as the databases that house it. And because databases rely on others for the information, you are not collecting pure data. This is obvious simply by reading the exculpatory clauses at the bottom of all your agreements with AVM providers, stating they are not responsible for the data they are regurgitating. They know the data collected is imperfect; those clauses protect them from any liability.                   

Another problem with AVMs is that they require a lot of interpretation. One must piece together all the contents, and from that, weigh in on the degree of reliability to determine its value. It is similar to analyzing a BPO or an appraisal; one has to read or analyze it from the perspective that the automated data that was pulled makes sense given the subject property that one is looking at. Based on that interpretation, one must be able to draw conclusions. If that is not possible, one will be making important business decisions based off of improperly analyzed data.

Also, AVM providers ignore market influences, and market influences have a huge impact on market value. Some would say market influences are too subjective to take into consideration, but value is all about subjective opinion. Take geographic location, for example. A home’s zip code has a significant impact on its value. Someone with a zip code of 90211 living across the street, merely a stone’s throw away from the affluent 90210 has a significant property value effect to the tune of about 10 to 12%. By ignoring market influence, AVMs ignore solid, validated, verifiable value.

That is why working with an agent or appraiser is so important. Their organic knowledge of the market is critical, not only for coming up with accurate property value, but also for selling that property in a timely manner. They effectively counsel clients on the most efficient way to sell the property.

There is a risk that AVMs could pick the wrong sales comps altogether to review because they lack true judgment and evaluation. AVMs simply attempt to follow the purchaser’s instructions based off an algorithm. For instance, if the appraisal order was to find the most recent sale closest to a subject property, an AVM robotically attempts to find that. But the result could very well be the closest property found is an REO sale when you are completing an appraisal for a new purchase money loan. You then just cheated that purchaser out of potential value. On top of that, computers are incapable of inspecting homes, and AVMs do not include pictures of the property. Because of that, in many cases, an AVM’s comparable properties can differ greatly from the subject in terms of updates, quality of construction and marketability of the home.

For a more streamlined, accurate valuation process, the industry should lean on agents or appraisers rather than unreliable, error-fraught AVMs. As the real estate industry looks ahead toward a restored future, we must let go of some of the ineffective tools that could lead to expensive mistakes.


Alice Sorenson is COO of LRES, a national provider of residential and commercial valuations and asset management for the mortgage, banking, credit union and real estate industries.

Comments (0)

Be the first to comment on this post using the section below.

Add Your Comments:
Not Registered?
You must be registered to post a comment. Click here to register.
Already registered? Log in here
Please note you must now log in with your email address and password.
MultimediaSee All »