The Census Bureau Admits That New Homes Sales Data Is Flawed
Posted on December 3, 2007
Filed under Real Estate Sales
Read the complete post
Thanks for visiting The Mortgage Reports. To stay absolutely current on mortgage markets and important guideline changes, be sure to take my free daily email alerts.
October's New Homes Sales report showed a modest month-over-month improvement from September. But that doesn't mean that the market has bottomed. In fact, it doesn't mean anything at all.
The monthly New Homes Sales figures are highly suspect and the Census Bureau knows it. It doesn't try to hide that fact, either.
On it's Web site, singled out, above the fold, plain as day, begging for attention, the Census Bureau asks:
How does the Census Bureau handle cancelled sales contracts in the published estimates of New Home Sales?
And then it serves up a 309-word statement, summarized below (in bullets):
-
A new housing unit is considered sold when a contract is signed and/or earnest money is exchanged
-
There is no follow up to verify if the sale was closed, or canceled
-
If cancellations are high, New Homes Sales data will be overestimated
It's an admission of guilt, like Colonel Jessup ordering the Code Red.
If we consider that builders are reporting cancellation rates in the 35-45% range, we can conceivably take the 728,000 figure reported and chop it down to 440,000 without feeling bad about it.
But back to the disclaimer. It also says the following (paraphrased):
A housing unit will never be counted twice so if a previously canceled unit is later sold again, the second sale is ignored. Therefore, in an improving market, the New Home Sales will be underestimated.
In other words, what's overestimated today will be underestimated tomorrow. File it under "one more reason to ignore the headlines and look deeper into the data".
There's always more to the story than the headline.
Dan Green is an active loan officer. Email or call 513-443-2020. Dan is on Twitter at @mortgagereports.

I use Scribe to improve my blog SEO








