Tracking the Truth on Foreclosures

RealtyTrac's data is oft-cited by the media, but some question its accuracy
by Andrew Galvin
The Orange County Register

If ever there were a public relations success story, RealtyTrac is it. The Irvine-based firm's monthly news releases, chock-full of state-by-state foreclosure counts, are devoured by a national media ravenous for data on what many consider a developing crisis. But questions are being raised about whether the firm's oft-cited numbers overstate the real dimensions of the foreclosure problem. And that could create a problem for the company's credibility.
For example, last year, RealtyTrac's data showed Colorado had the nation's highest foreclosure rate. That didn't sit well with state officials, who decided to do their own count of foreclosures and came up with a figure much smaller than RealtyTrac's. Then, in July, RealtyTrac reported 12,602 foreclosure actions in Georgia, giving the state the nation's second-highest foreclosure rate. When the Atlanta Journal-Constitution looked into the numbers, the newspaper found that RealtyTrac had counted more than 2,000 properties twice and sometimes more. RealtyTrac acknowledges it isn't perfect but says its data offers comprehensiveness and context that other providers don't.

Why the discrepancies?
The main reason is that RealtyTrac counts every step in the foreclosure process. So if a home goes into default on its mortgage, is scheduled for auction and then repossessed by a bank, RealtyTrac counts that home three times. RealtyTrac counted 54,747 "foreclosure actions" in Colorado last year. That number wasn't useful because it didn't reflect how many homeowners were actually in danger of losing their homes, said Ryan McMaken, spokesman for the Colorado Division of Housing. "We couldn't really use those numbers for having serious discussions," he said. So McMaken put an intern to work calling all of the state's 64 counties to get a count of how many homes entered the foreclosure process last year. The number he came up with: 28,435.
This summer, partly in response to criticism, RealtyTrac began sorting its numbers to compile a separate count of properties in foreclosure, in addition to total foreclosure actions. RealtyTrac's "unique property" count, published quarterly, found 19,411 properties in foreclosure in Colorado in the first half of this year. That's within a few dozen of the 19,460 counted by McMaken. "I think they're getting a lot closer now," McMaken said, adding that "we might not have to collect our own numbers" anymore.
In the Georgia situation, RealtyTrac admitted it erred. It revised its July count for the state to 8,461 foreclosure actions, down from its initial count of 12,602. "The reporting error resulted from a combination of overlapping data coverage in some areas of Georgia and an anomaly in the formatting of some of the foreclosure records in those overlapping areas," the company said.
RealtyTrac could probably mute much of the criticism of its data if it simply published its unique properties count every month in addition to its total filings count. That's something the company is considering doing next year, said Rick Sharga, RealtyTrac's vice president of marketing.
Does it matter how the data are counted? Jack Kyser, chief economist with the Los Angeles County Economic Development Corp., argues that it does. Figures that overstate problems in the housing market "become sort of a self-fulfilling prophecy in that people are afraid to go out and look for a home," Kyser said. Moreover, inflated data on foreclosures could prompt politicians to push through ill-considered mortgage reforms. "You do something that's good, but it's the law of unintended consequences," Kyser said.
Other factors that could cause RealtyTrac's counts to be higher than others: the company doesn't filter out duplicate filings if two or more loans on the same property go into default, and its monthly reports are based on the date that foreclosure actions enter its database, rather than the recording dates, Sharga said. "We're not perfect; we don't claim to be," Sharga said. "When we do find a mistake, we fix it … and try not to replicate that."