On December 21, the National Association of Realtors (NAR) will release revisions that will lower its estimates for existing home sales dating back to 2007.

For nearly a year, NAR has been in consultation with several government and housing-market entities—including the NAHB, HUD, Fannie Mae, Freddie Mac, the Mortgage Bankers Association, and American CoreLogic—to determine how best to collect and estimate home sales.

Walt Molony, a spokesman for the Realtor group, told Builder on Monday that NAR’s data had become harder to benchmark and report accurately after the U.S. Census Bureau decided that it would not use its decennial long form for the 2010 national census. That long form, says Molony, gathered information about homeowner mobility in previous years that the substitute short form—with its seven basic questions per person—does not touch upon. The shorter form also doesn’t ask about house values.

Molony says NAR also found that using courthouse records as the basis for its sales estimates proved to be deficient because there were “too many holes” in how municipalities tracked sales and kept records. “It took us much longer than we expected” to come up with a viable alternative, he says.

Early this year, NAR’s sales figures drew flak from CoreLogic, a Santa Ana, Calif.–based real estate analytics firm; and the Mortgage Bankers Association. They contended that the Realtors’ estimate for 2010—4.91 million units—was anywhere from 15% to 20% too high. However, Molony points out that CoreLogic, in its discussions with NAR, expressed its own frustration with using court documents to follow sales.

So the Realtors’ “Plan B,” says Molony, will be to use the annual American Community Survey (ACS) as its starting point for calculating and benchmarking existing home sales.

In June 2009, the Missouri Census Data Center did a comparison of ACS vs. the decennial census long form. ACS dates back to 2001, and its national sample began in 2005. ACS is nowhere near as comprehensive as the decennial census. “Data are not yet available for around 1,400 of the nation’s over 3,100 counties and for over 90% of the nation’s places (incorporated and census-designated),” the Missouri Census Data Center wrote. However, the big advantage of ACS is its frequency compared to a national census that gets conducted only once per decade. “The ACS provides us with a new set of refreshed data every year,” the Center stated, and new questions can be added to it annually.

“If you are a person who focuses on the big picture—what is happening nationwide or statewide, or even city or countywide for larger cities and counties—then you are probably going to be pretty happy with the ACS as a replacement for the long form,” stated the Missouri Census Data Center. “But if you need data for smaller areas, for your town of 10,000 people or your local wards, planning areas, or transportation zones, then you may not be as happy with what's new.”

Robert Denk, NAHB’s assistant vice president for forecasting and analysis, thinks the size of ACS’s sample—in 2010, it was 3 million households, of which 1.9 million responded—is more than adequate to come up with statistically viable home-sales numbers. He compares ACS to the Current Population Survey, which is “highly regarded” even though its sample is only 60,000 households. “I think the smartest thing that NAR could have done right now is to attach its sales to the ACS,” says Denk, who attended the meeting on Monday morning where NAR disclosed to its consultation partners its intentions and rationale.

Molony notes that NAR ultimately would like to rely on courthouse records for its sales estimates, which Denk thinks is feasible in the future. “With electronic data storage and transfers, these [records] are bound to get better.” However, Denk notes that smaller, rural areas still tend to have incomplete information, and that data processes even in larger markets aren’t always standardized, which makes benchmarking and market-to-market comparisons all the more challenging. “We see the same problems with permit data, too,” he says.

It remains to be seen whether NAR’s data will gain credibility with the trade group's switch in methodology, and whether other entities that follow house sales and prices will refine their methods in response. At least one competitor, CoreLogic, “is pretty solid with our own methodology,” says spokesperson Alyson Austin. CoreLogic’s Home Price Index is based on tracking sales of the same home addresses over a 30-year period. It covers 6,621 zip codes (or 58% of the U.S. population), 608 core-based statistical areas (86% of the population), and 1,149 counties (84% of the population).

But Austin acknowledges that the housing industry in general can only benefit when organizations such as NAR improve the quality of their data. “It’s important to find a methodology that works,” she observes.

John Caulfield is senior editor for Builder magazine.

Learn more about markets featured in this article: Los Angeles, CA.