CoreLogic, a data provider, and National Association of Realtors (needs no introduction), have been in a somewhat public spat about the quality of NAR’s data. CoreLogic contends that NAR over-states the number of homes sold in any given period by up to 20%. (The original news item is “Decline in real estate sales greater than stated?” Inman News, Feb. 15.)
A little background: What we think of as “the MLS” is not a national thing; it’s actually a collection of regional MLS associations. The national NAR numbers collect data from the regional MLS associations–but not from all of them. So not all homes get sold on the MLS and not all MLS data feeds the national NAR number. NAR compensates by making adjustments every so often. They have an extended explanation here.
There are some people who believe the NAR purposely inflates its numbers to make it look like there are more homes sold than there actually were. I disagree. I believe they spin the numbers in an absurdly optimistic manner sometimes–and often in a way that is detrimental to consumers–but I don’t think it’s appropriate to criticize their data gathering and reporting on this basis alone. They are a trade group using their own trade data–same as Mortgage Bankers, NAHB, moving van companies, payroll service companies, and whatnot. They attempt to adjust it to the only real, reliable data on the number of sales, which is the long-form Census question in 2000.
CoreLogic has a much bigger sample. CoreLogic has access to the title documents and public records of almost 80% to 85% of the home closings in the U.S., double the coverage of NAR. In addition, because CoreLogic gets its data from public records, whatever biases might be inherent in users of the MLS would not affect CoreLogic data. If asked whose data I trust more, I would say CoreLogic.
But does the raw number matter? For most readers, I would say no. Most readers want to know the trends–the ups and downs, how the sales volume and selling prices interact, whether the months’ supply is going up or down, or high relative to its historic norm. If the NAR data is biased in 2008 and similarly biased in 2010, then you should be able to compare 2008 and 2010 and still come to a valid conclusion. Besides, NAR comes out at much smaller geographies than CoreLogic. (However, this site has much smaller geographies than even NAR!)
The NAR warns against “data drift.” The longer they are from their periodic revisions, the greater the potential for their estimates to be wrong. That means that data from 2010 may not be biased in exactly the same way as data from a few years back. NAR appears to recognize that this would be a problem. They indicated in their explanation that they are looking toward an annual benchmarking exercise.
One senses that the CoreLogic attack on NAR is somewhat unsporting, considering that they are competitors in many ways. However, this is undoubtedly good for consumers that someone is credibly challenging NAR’s dominance as a real estate data provider. NAR gets far too much play in the press. As one of the only providers of data, they get the automatic–and probably undeserved–quotes and authority to analyze the data. If this challenge makes NAR data more complete, or erodes their dominant position as data provider, then consumers and readers will win.