For once, I agree with Twist over at Housing Doom. There ought to be more transparency in the data provided through the Arizona Regional Multiple Listing Service. Perhaps then we could put an end to many of the dubious conspiracy theories surrounding the MLS.
Last week, Business Week’s Peter Coy posted an online article based on Twist’s data claiming the MLS manipulates its data every December. To what purpose, I’m still not certain. There was one fatal flaw with the theory – the supporting data set was incorrect. Four of us entered the MLS database, ran the numbers and came up with the exact same information – numbers completely different than Twist’s. The article was subsequently withdrawn in the fact of the actual data.
Today Twist offered a vigorous defense of her conclusions and her data, extrapolated from the market stats provided on the public side of the ARMLS site. In short, she is taking the number of total listings and factoring out closed sales to determine how many listings were canceled, temporarily of the market or expired.
What’s missing, as she readily admits, is the pending sales – homes which no longer would be shown as actively listed but also are not yet a closed sale. Says Twist, “… I believe the result here, while imperfect, is accurate enough for our purposes.”
Well, no. It’s not. At this moment there are more than 4,000 pending sales in ARMLS. Is there anyone who believes leaving roughly 10% of the data out of any equation will lead to a correct conclusion?
(The obvious question is why these homes are shown as Pending when that status places them somewhat in statistical limbo – we’ll come back to this.)
Twist concludes … “Consumers shouldn’t have to turn to their local bubble blogger for extrapolated data. If the reports are public, then the supporting data should be public as well.”
She’s right. Consumers shouldn’t have to turn to their local bubble blogger, especially if the extrapolated data they unfortunately may take as fact is flawed. She’s also right in that the full data should be made public although another question is raised … would it even matter? When your days are spent accusing Realtors are lying about everything in order to make a buck, would you really believe the data even if it was given to you?
I think not. And the proof comes in today’s defense flying in the face of the actual data. Twist quotes “Realtor L” as “L” tries to explain the reasons Twist’s data is wrong. At some stage, “L” ought to have provided Twist the actual numbers. Or maybe “L” did and Twist simply chose not to use them. Finding the numbers is a simple process for an agent – log onto ARMLS, search for canceled/expired/TOM listings, set a timeframe and press Search. Takes about 22 seconds.
Is it worth taking 22 seconds so as not to mislead your readers? And while I clearly lack Twist’s statistical chops, when I’ve worked with a data and have formulas providing clearly impossible results (such as the negative number of so-called “delisted” homes in three separate months), I tend to question the veracity of my formula first. If your statistical analysis concludes absolutely that the sky is green, it doesn’t mean the sky is green.
“L” did raise the specter of “exclusive listings” and in the context of Twist’s post, it seems as if real estate agents are withholding listings and data from the MLS in a deliberate effort to skew the numbers. Nowhere is it mentioned that an exclusive listing is one in which the seller – the seller – either has asked the agent or given the agent permission not to enter the listing in MLS. Why? Usually to save the seller a couple of percent on a co-broke fee … given the frenzy in 2005, the MLS was largely superfluous.
Back to the idea of the “Pending” status. If you’ve ready my blog for any length of time this should be a familiar refrain. But the Multiple Listing Service is a marketing device that allows one brokerage to communicate its listings to other brokerages. It’s no different than the databases Napa Auto Parts uses to share inventory with AutoZone.
If the use of “Pending” and “Active with Contingencies” statuses cause statistical issues, frankly, that’s irrelevant. The data isn’t there for statistical purposes. It’s there to help us do our jobs on behalf of our clients more efficiently.
“Pending” means the house is under contract without contingencies and has not yet closed. “AWC” in its various permutations means the house is under contract with contingences. In either case, we’re communicating to other agents that there is a contract pending on the house and proceed appropriately.
Nothing complicated. Nothing duplicitous. Just a simple inventory system.
Maybe ARMLS should provide data on pending listings as well as canceled/expired/TOM. But my hunch is the data still be would be taken out of context for the sake of proving a misguided conspiracy theory.