Understanding the revision of home sale statistics

by Dan Krell
© 2011
DanKrell.com

As the housing market slid, the National Association of Realtors® (NAR) was often criticized for producing home sale data that seemed unrealistic. As criticism seemed to peek, NAR announced earlier this year that they were seeking to “re-benchmark” data for counting the number of homes that sold.

According to a December 13th Reuters report (Existing home sales to be revised down from 2007: NAR), the NAR is “revising down” home sales statistics because of double counting, “indicating a much weaker housing market than previously thought.” The news sparked cries of “fraud!’ and “told you so’s” across the blogosphere; while some used the news as a marketing opportunity to tout their data as unwavering.

However, according to the NAR’s press release, “Q&A on Re-Benchmarking of Home Sales” (economistsoutlook.blogs.realtor.org), the main reason for the re-benchmarking is for data drift that occurred during the housing downturn; re-benchmarking is a common aspect of estimating economic data (much like the government’s GDP and employment figure revisions). The re-benchmarking is only for existing home sales and not home prices.

According to Lawrence Yun, NAR Chief Economist, data drift is to blame for the over estimates. The monthly existing home sales data that is reported by NAR is compiled from MLS boards across the country. Data drift was revealed when comparisons were made with other available home sales data.

Data drift is a term that describes the change of non-constant variables used in statistical measurements. The data drift in NAR’s existing home sale data was described as being caused by several factors: an increasing reliance on Realtors®, double listings, and inconsistencies across MLS boards.

Although MLS data typically tracks Realtor® home sales data, there are homes that are also sold by home builders and for-sale-by owners (fsbo) which are not typically reflected in the MLS. Dr. Yun believes that some of the data drift is due to the increasing reliance on Realtors® as the market deteriorated to sell homes they typically did not sell in the past (by fsbo’s and builders).

Additionally, it was realized that MLS home sale data was duplicated in some instances. In some regions, it is not unusual for Realtors® to belong to more than one MLS board. In some of those instances, Realtors® often input the data in two or more MLS’s; thus resulting in a duplicate sales.

As technology and markets advance, local and regional MLS boards found themselves changing to increase the quality of the MLS data, as well as expanding to provide service in outlying areas. Although many MLS boards attempt to adhere to consistent data standards and practices, compiled home sale data is not always consistent across all the MLS boards. Additionally, as MLS coverage grew, it could have been logically assumed that the quantity of home sales reported for the growing MLS boards would increase because of the wider coverage.

Additionally, Dr. Yun stated that the census data used to benchmark the MLS data has also changed; the U.S. Census changed the data it collected by changing survey forms. In re-benchmarking, the NAR expects a revision of existing home sales to account for the increase of MLS entries of new homes as well as homes that sold multiple times within a 12-month period (flips). The re-benchmarking should also account for fsbo variances that were not previously adjusted.

The revisions are expected this week.

More news and articles on “the Blog”
This article is not intended to provide nor should it be relied upon for legal and financial advice. This article was originally published in the Montgomery County Sentinel the week of December 19, 2011. Using this article without permission is a violation of copyright laws. Copyright © 2011 Dan Krell.