Last Week It was a tale of two worlds last week. In the rates markets, more and more participants have begun to accept as fact that the inflationary pressures being experienced in all corners of the e...
An Automated Valuation Model, or more commonly known as an AVM, is a program that automatically analyzes various data points to produce an estimate on the current value of a home or property. An AVM includes:
Property identifying information including address and/or assessor’s parcel number and owner name
Prior sales of the property
Living area/square footage
Number of bedrooms and bathrooms
Public record information collected from state and county records, such as deed transfers
Recent sales and listing of similar properties in the local market area
Area pricing trends
I recently sat down with Chuck Mureddu, SVP, Chief Valuation and Compliance Officer, for a Q&A about what you can expect when using an AVM.
Yes. A “no-hit” means that the AVM was not able to produce a value or range of value. Many lenders will try running another specific AVM model or cascade should this occur.
There are several factors that contribute to a “no-hit” including, but not limited to:
The property could not be found (in the property database)*
Multiple properties were found/unit numbers missing
Insufficient quantity and quality of data available for AVM to calculate a value
AVM was able to generate a value, but the confidence score did not meet the criteria
*Address issues could be resolved by modifying or correcting the address entered or by providing additional information such as the property tax ID or parcel number.
The data used by AVM providers are dynamic and subject to change. In this instance, the model may have had enough data last year to return a value, but now the data may be limited within the shifting timeframe to calculate a value. Another reason could be the confidence score may have dropped and did not meet the criteria this year.
To troubleshoot this issue, double check the address to make sure it is identical to the address that was input the previous year or by providing the property tax ID or parcel number.
The data used by AVM providers are dynamic, so there may be more recent data. Many AVMs analyze the nation’s housing stock one property at a time, weekly or bi-weekly. In the period of several months, properties could have sold or listed that are influencing the surrounding values.
The AVM property characteristics can only be displayed if available from the public records data sources. AVMs get their property DNA (characteristic data) predominantly from about one of three public record providers. The underlying tax assessor data, or recorder data, typically have missing fields or values. If the first AVM presented is missing a property characteristic chances are other AVM models will be similar.
AVMs may use hundreds of recent sales in the calculation of the estimated value, but will only display those that best match the subject property. If recent sales are not presented, then it is because the AVM did not find any that match the subject property with similar property characteristics. A possible reason is the property is in a non-disclosure state i.e., Texas does not require that the transfer tax or sales price be provided on the recorded documents. In some cases, AVM providers have relationships with MLS data sources. Using data from this source helps the valuation but there may be a licensing restriction on the use of the data stating it cannot be shown on a report. This kind of data licensing is rare.
Each institution has individual risk tolerances. In some cases, your institution may be comfortable lending on a property as there may be other compensating factors to support your lending decision.
Valuation accuracy will vary depending on a wide range of factors. AVMs are less accurate when valuing properties that are unusual, larger than average, rural, historic, or have particular distinctive features. Each model has a scoring range and will fail when it falls below a predefined score by the AVM model. All AVMs rely on the accuracy, comprehensiveness, and timeliness of the data they use. The most significant factor is the number of recent sales in the area and how similar the property is to surrounding properties.
If your institution determines that you would like place-specific requirements on score minimums, there are models available.
The AVM providers deliver the owner information available in their databases. Some third-party AVMs i.e., Freddie Mac’s HVE, by design do not provide owner names.
If you're looking for a vendor for valuation services, click here to check out our offering!
As the VP of Marketing & Sales Administration, Kymberly is responsible for overseeing all Marketing efforts of SWBC Lending Solutions. In addition, Kymberly works with internal and external customers to support the SWBC Lending Solutions sales team. Kymberly joined SWBC in 2008.
When it comes to understanding the auto lending industry, I’m a numbers guy. Data weaves the story that allows me to kee...
Fannie Mae and Freddie Mac are currently collaborating on a game-changing initiative to redesign the uniform appraisal d...
Put your Comment Below.