In Depth Analysis: CalculatedRisk Newsletter on Real Estate (Ad Free) Read it here.

Thursday, October 25, 2007

More on September New Home Sales

by Calculated Risk on 10/25/2007 10:40:00 AM

For more graphs, please see my earlier post: September New Home Sales

Let's start with revisions. Last month I wrote:

The new homes sales number today [August] will probably be revised down too. Applying the median cumulative revision (4.8%) during this downtrend suggests a final revised Seasonally Adjusted Annual Rate (SAAR) sales number of 757 thousand for August (was reported as 795 thousand SAAR by the Census Bureau). Just something to remember when looking at the data.
Sure enough, sales for August were revised down to 735 thousand. I believe the Census Bureau is doing a good job, but the users of the data need to understand what is happening (during down trends, the Census Bureau overestimates sales).

This makes a mockery of headlines like this from the AP: New Home Sales Rebound in September. Sales did not "rebound", in fact the September report was horrible, and the sales number will almost certainly be revised down.

For an analysis on Census Bureau revisions, see the bottom of this post.

New Home Sales and RecessionsClick on graph for larger image.

This graph shows New Home Sales vs. Recession for the last 35 years. New Home sales were falling prior to every recession, with the exception of the business investment led recession of 2001. This should raise concerns about a possible consumer led recession - possibly starting right now!

The second graph compares annual New Home Sales vs. Not Seasonally Adjusted (NSA) New Home Sales through August.

New Home Sales
Typically, for an average year, about 78% of all new home sales happen before the end of September. Therefore the scale on the right is set to 78% of the left scale.

It now looks like New Home sales will be in the low 800s - the lowest level since 1997 (805K in '97). My forecast was for 830 to 850 thousand units in 2007 and that might be a little too high.