In the wake of the Great Recession and an unprecedented housing bust, automated valuation model users want something more that the traditional rearview mirror approach provided by early versions of the technology.
According to industry experts, current market data is key to analyzing today’s home values. As a result, AVM providers are incorporating asking prices and recently sold data into their models to pinpoint market conditions, rather than waiting for public records that may not be filed for months after a deal closes. In addition, many are going a step further, adding forecast models that predict future price trends to their analytics.
Michael Simonsen, CEO of Mountain View, Calif.-based Altos Research, said real-time market data about home price listings and pending transactions gives lenders critical insight that they can’t get from models that rely exclusively on closed and recorded sales information.
“What we do is track the active market, the asking prices and the changes in asking prices,” he said.
The traditional approach makes it difficult to ascertain what the market is doing at the current moment, Simonsen said. In addition, active data from real estate listings, especially the multiple listing service from local Realtor associations and other groups, provides modelers at Altos with a larger sample size, more detail about property characteristics, and more information about changes in supply and demand. Nationally, the company tracks a couple of million property listings weekly, he said.
As a result, AVM and broker price opinion providers increasingly want to incorporate real time data from Altos’s home price index to supplement valuation estimates based on closed sales or visual inspections. In addition to culling more timely data about the market, this approach greatly increases the number of “comparables” used to estimate a home’s value.
“The smart AVM providers are using the active market data because you get environments where you have a small sample size,” Simonsen said. “It’s really hard to do the math where only two properties have sold.”
A traditional BPO might include as few as three comparable sales, whereas Altos’s database may include dozens or more listed properties and their current asking price in that local market.
The listing data also includes information about changes in the listing price, he noted, which helps clients catch potential errors in a BPO or AVM estimate. For instance, if a BPO values a property at $250,000, but the listing history shows that the home was on the market with an asking price of $220,000 for 120 days and didn’t sell, then the BPO is too high.
Simonsen believes that incorporating the real-time data into home price analytics helps lenders make better underwriting decisions based on current loan-to-value ratios.
“I think it’s really encouraging for housing and mortgage markets that everybody is better informed,” he continued. “It’s important for liquidity as well as for lending standards.”
