NAR Addresses Errors in Home Sale Statistics

"The National Association of Realtors® Existing-Home Sales Series is the premier measurement of the residential real estate market." This statement (still) appears in the research section of the NAR web site. Given events of the past week, a more modest self-evaluation might be in order.

On December 21, 2011 NAR released two documents that address some significant problems with its home sale statistics. One is a PowerPoint presentation titled, "Re-benchmarking of Existing Home Sales". Despite the somewhat imposing title, it provides a brief and understandable overview of the issue. The other, "Description of Methodology to Benchmark Existing Home Sales, 2011", is a considerably more detailed description of NAR's data gathering and estimating procedures. It is not for the faint of heart. Both are available on the NAR web site.

The apparent extent of NAR's reporting errors is significant. Existing home sale numbers were revised for the past four years. The yearly discrepancies ranged from 11% - 16%. The four-year average of 5,157,000 existing homes sold was revised downward to 4,420,000 - a 14% change.

"How," one might wonder, "can this be? It's just counting isn't it?"

Well, no, it isn't just counting. It's considerably more complicated than that. And who knew?

The NAR methodology report points out this: "The actual level of monthly home sales for the entire country is unknown. NAR provides existing home sales estimates based on benchmarks and sample data... there is currently no comprehensive, current listing of monthly sales for the entire country based on actual records." The report goes on to state that "Most economic data are based on benchmarks and samples." Here is how benchmarks worked for NAR:

"A representative sample of approximately 200 MLS's from around the country provide NAR with sales and price data on a monthly basis.

* The Monthly EHS [existing home sales] was last benchmarked for 1999 based on the 2000 Census.
* Each month, beginning in January 2000, NAR tracked the percentage change in sales in the MLS data from the same period one year ago.
* The percent change from the MLS data was applied to the benchmarked data to estimate monthly sales."

Let me try to represent this method with a greatly over-simplified example. I know that MLS data doesn't represent the entire number of homes that have been sold in any given period. So, I decide to use as my benchmark the sales data taken from the long form questionnaire in the 2000 census. I figure that is the best way to capture the actual number of sales. (Note, as we will observe later, that county records are not uniformly reliable.)

Suppose, then, that the census data indicates that my market area sold 1,000 homes in July; whereas the MLS data showed 900 sales. Then I will figure that MLS data represents 90% of actual sales. Now, next year - when there is no census data available - I will use MLS data, and extrapolate accordingly, to estimate actual sales. If MLS sales increased to 990, I would estimate actual sales at 1,100.

In recent years, it has been observed that there were growing discrepancies between NAR's estimates and those of other information providers, most notably CoreLogic, whereas they had been pretty close in the past. Investigation into the causes of these differences led NAR to acknowledge that they had overlooked some important changes in the market. These changes meant that the MLS data should be treated differently than it was in the year 2000. Some of the more notable changes were these:

1. In recent years there has been a lower percentage of for-sale-by-owner (FSBO) sales than there were during the benchmark year. So, when NAR statisticians were extrapolating from MLS data to the number of actual sales, they have recently been adding more FSBO sales than actually occurred.
2. NAR reports were for sales of existing homes, not newly constructed ones. During recent years, however, more new-home builders have been putting their inventory into MLS. Thus new homes sales have been influencing the supposed count of existing home sales.
3. Finally, in a slow market it is an increasingly common practice to list homes in more than one MLS. When the home is sold, then, it shows as a "sold" in each MLS in which it was listed. Hence, when accumulated MLS data is used, the same sale is counted more than once, inflating the result.

It is to NAR's credit that they have engaged in a thorough re-examination of their statistical procedures. The lengthy report that has been issued explains the new methods that will be used to estimate existing home sales. Like visiting a sausage factory, though, you might like the product better if you don't become too well acquainted with how it is made. But, it may be the best we can do.

And why not just count sales as recorded by the county? That, unfortunately, is easier said than done. As the report makes clear, "States and counties across the country record home sales transactions in a non-standardized manner... Further, in non-disclosure states, some critical sales information is not publicly available." The bottom line: "...courthouse records do not provide adequate information in the form needed at this time."

National home sales data is extremely important to those who shape and influence housing policy. Both political and business decisions may rely on such reports. To most of us in the business, though, we are reminded of old 1960s theme: you don't need a weatherman to tell you which way the wind is blowing.

Written by Bob Hunt for Real Estate Today at www.retradio.com Copyright © 2011 Realty Times All Rights Reserved.