If you watched the 2012 Olympics, you already know something about valuation cascades, even if you didn’t realize it. A number of events involve multi-layered judging criteria, including diving, gymnastics or dressage. The realm of automated valuation model (AVM) cascades is similar, though obviously more complex, in execution. After careful review of performance by weighting various metrics, an AVM “cascade” is the identification of the best performer for the first, second and third (valuation) positions … or Gold, Silver and Bronze if you will.
In last month’s column, we discussed the Interagency Guidelines which were updated in 2010 and include new language governing the use of AVMs. We covered the specifics of the Guidelines and their requirement for AVM users to either possess the in-house expertise to evaluate AVM accuracy or to find a credible external source that provides it. In this column, we will briefly examine what goes into a proper cascade and important factors to consider.
Regulators are understandably concerned that while AVMs are useful in a number of lending transactions, they need to be continually tested for accuracy. The most reliable cascades will incorporate rules around variables, including geography (usually by county), confidence score, forecasted standard deviation and others. Testing multiple AVMs side by side and focusing on key metrics, allows the tester to appropriately position the AVMs in a quality cascade.
Most cascade designs incorporate three AVMs for a given rule set (i.e. a county). If all industry AVMs are tested and the top three selections are determined to provide (in ranked order) the most accurate result for a required rule set, and none of the three return an acceptable report, there is little to be gained in going further. At that point, it is best to move to your next trusted valuation approach.
AVM users need to ensure that AVMs are periodically tested. Practical guidance dictates this should be done on a quarterly basis, but may vary on the size and scope of the user’s purpose. Regulators expect that, as performance results or industry expectations change accordingly, the rule sets should shift to provide the AVM(s) best suited to provide the most reliable valuations. A list of the AVMs tested, associated findings and supporting documentation, as well as the design of the AVM cascade need to be documented and accessible for any regulator discussions. An experienced tester will thoroughly examine the entire array of AVMs in the market, not just those created by a generous vendor or that emphasizes ones the vendor prefers to sell.
Cascade creation is theoretically a straightforward process, but that does not mean it is simple. Properly tested and crafted AVM cascades provide a reliable, objective and efficient way to maximize your AVM usage. But always keep in mind, you—the AVM user—are responsible for ensuring the AVMs utilized comply with applicable laws and regulations and are consistent with supervisory guidance.
David Rasmussen is senior vice president of operations at Veros Real Estate Solutions. For more information, call (714) 415-6300 or visit Veros.com.