Advancements in advanced, statistical-based valuation technology have made automated valuation tools significantly more cost effective than traditional property evaluation.

However, the Consumer Finance Protection Bureau and other industry regulators have begun a push to significantly limit, and in some cases completely eliminate, the use of automated valuation tools in property valuation efforts.

Unfortunately for both servicers and originators, this push comes at a time when enhanced features are being added to automated solutions that allow these valuation tools to approach the accuracy of the property valuation gold standard—namely the traditional appraisal.

Recently, DataQuick examined 942 appraisals completed since the beginning of 2013. We recorded the ensuing sale and actual sale price for the property after the appraisal was completed and compared the sale price to three separate valuations generated at the time of the appraisal:

  1. The appraised value itself.
  2. An estimated property value from CMV—one of DataQuick’s AVMs.
  3. An estimated property value from Collateral Validation—a DataQuick solution that calculates a consensus value for a property based on multiple sources, such as DataQuick and third-party AVMs, property listings, tax-assessed valuation models and mark to market valuation estimates.

Our comparisons found that without filtering the results, appraisals still deliver a much higher level of accuracy than the automated valuation tools. Interior appraisals are clearly the most accurate valuation, supporting the regulators’ case for traditional property valuations.
However, what regulators are not taking into account are enhancements to automated tools—specifically advanced filtering capabilities—that now allow risk managers to set a threshold on what property valuations will be accepted in the analysis.

When eliminating some of the higher-risk properties through CMV’s confidence score filter, we found a significant compression between the accuracy measures of each of the three valuation tools as indicated by the green bars in the graph.

Though this method reduces the number of appraisals in the analysis (from the original 942 to 760), it statistically validates the use of automated valuation tools for lower-risk transactions. The case is strongest for CV where 82% of the values were within plus or minus 10% of the actual sale price—just three points below the accuracy metric for interior appraisals.

The property segmentation available through the use of enhanced analytic tools, such as the CMV confidence score, allows servicers and originators alike to deploy property valuation tools that deliver comparable accuracy to a full appraisal at a fraction of the cost.

There will always be a need for the intelligence garnered from appraisals, yet servicers should consider implementing automated valuation technology and the improved efficiencies that come with it into their property valuation process.

Meanwhile, regulators should take a long look before eliminating automated property valuation tools that can be utilized with the same level of confidence as traditional property valuations in many circumstances.

Randy Wussler is vice president of product management and marketing for San Diego-based DataQuick.