The bigger problem is the trend. From month to month (February to March 2012), there is slippage in all three numbers:
- New home sales down 7%
- Existing home sales down 3%
- Housing starts down 6%
This lends credence to the idea that the warm weather did pull some buyers forward.
Another issue: why are the housing numbers so...weird? How come builders and Realtors are reporting strong traffic, and generally sales, but the numbers are erratic?
In the case of the Census Bureau, which does the New Home Sales numbers, the problem is a limited sampling size.
For example, the Census Bureau says New Home Sales for March were 7.1 percent below February. But read the fine print: it actually says that there is a 90 percent confidence that the TRUE NUMBER is plus or minus 20.7 percentage points from that 7.1 percent.
What this means is that the TRUE NUMBER is between down 13.6 percent (down 20.7 points) OR up 27.8 percent (up 20.7 points).
Think about this: the real number is somewhere between down 13.6 percent or up 27.8 percent. And that is only with 90 percent confidence!
How can the number be so uncertain? Because the sample size is very small. It's based on a sample of houses selected from building permits.
Can't we get better numbers? Sure we could. But we'd have to pay more for it.
These issues were already old over 20 years ago, when I was CNBC's Real Estate Correspondent.
One day in 1991 or 1992 I trekked to the Census Bureau to interview the poor bureaucrat who kept the statistics. When I asked him about the poor quality of the stats, he knew exactly how much more money the Census Bureau would have to spend to get more accurate statistics.
What was the chances he would get more money? He just shrugged.
I don't know if that fellow is still there, but I can tell you the Census Bureau has not got more money.
Bookmark CNBC Data Pages:
Want updates whenever a Trader Talk blog is filed? Follow me on Twitter: twitter.com/BobPisani.
Questions? Comments? firstname.lastname@example.org