New home sales up 0.7 percent in August; or NOT? Down 3.4 percent from a year ago or down 27.8 percent?

Dennis Norman

Yesterday I did a post about the new home sales report issued by the U.S. Department of Commerce and, just like much of the media and bloggers out there, I reported that new home sales rose in August, albeit a scant 0.7 percent and that the seasonally adjusted annual rate of new home sales was down 3.4 percent from a year ago.

So this morning I wake up early thinking about the figures on the report and a couple of things that always bug me about the new home sales data from the Commerce Department (all normal guys wake up on a Saturday morning thinking about this stuff, right?).  One of the things that bothers me is the way a “sale” is determined.  This is straight from the explanatory notes on the report:

Preliminary new home sales figures are subject to revision due to the survey methodology and definitions used. The survey is primarily based on a sample of houses selected from building permits. Since a “sale” is defined as a deposit taken or sales agreement signed, this can occur prior to a permit being issued. An estimate of these prior sales is included in the sales figure.

Well, if you look at the margin of error for a confidence level of 90 percent on that piece of data you will see it is plus or minus 16.2 percent.  So, for August what the report is really showing is that the seasonally adjusted rate of new home sales could be UP as much as 16.9 percent from July or DOWNas much as 15.5 percent from July.  Hmm,…

The methodology for determining new home sales is quite different than the method used to track existing home sales by the National Association of REALTORS.  NAR counts a closed transaction as a sale making the data much more consistent and accurate.

Another thing that concerns me and I can’t really figure out how they do it is the “seasonal” adjustments.  Mind you, I’m not an economist, and don’t claim to be, and I realize that there are different times of the year when it is normal to have more sales or fewer sales, however it appears to me that the “seasonalness” of prior months still affects the current estimates even after the season has passed.  For example, in August 2009 there were 38,000 actual homes sold (not seasonally adjusted) according to the report, the exact same number as August 2008.  However the seasonally adjusted annual rate for August 2008 was 444,000 homes and for August 2009 is 428,000 homes.  Clearly the prior months have to play into this and since as of August 2008 the number of homes actually sold YTD (not adjusted) was 365,000 and for August 2009 it was 263,000 it would make sense that August 2008′s annual rate would be higher but what I don’t get is if actual sales is almost 28 percent below last year, and the current month is even with a year ago, then why is the current annual rate only 3.4 percent below a year ago?  I’m sure there is some statistical formula to prove this but I’m not so confident about the “seasonally adjusted” numbers.  Before an economist bashes me, let me point out that, as I mentioned above, in August 2008 the projected annual rate of new home sales was 444,000 homes and by December this number was down to 374,000.  Guess what the actual numbers were for the year?  485,000.  My guess is an economist will tell me that the monthly projections of annual rates is not meant to be a prediction of what the actual numbers for the year are going to be but just an illustration of what life would be like if that month’s sales were projected out in an effort to show a trend.  However, I think much of the media, and probably most of the public looks at a prediction of the results for the year.

Before I drone on any further, let me point out some statistics from the report that I think are much more useful and a better indicator of what is going on than most of the numbers being reported.  I’m talked about the actual numbers, not seasonally adjusted (note the much lower margins of errors on the numbers below):

• New homes sold in 2008 through the end of August; 365,000.   YTD new homes sold this year through the end of August; 263,000.  Therefore YTD new home sales is actually down 27.9 percent from last year, not the 3.4 percent being reported.  The margin of error for this statistic is plus or minus 3.1 percent meaning the range could be from 2009 being down 24.8 percent to it being down 31.0 percent which is a much more consistent indication in my opinion than the numbers we discussed above.
• New homes sold in August 2009; 38,000.  New homes sold in August 2008; 38,000.  Therefore no change in new home sales from the same month as last year. 
• New homes sold in August 2009; 38,000.  New homes sold in prior month; 38,000 for no change.
• Now, for what its worth.  As I think anyone in the real estate business would expect, new home sales (non-adjusted) has increased every month this year through August which would be considered normal and would indicate to me that the market is performing “normally” in terms of the “seasonality” of the market.  I think everyone would expect the summer months to be the best and then the numbers start dropping in September and continue to drop until year end.

So what does all this mean?  I think it means that the new home market is functioning but at a greatly reduced pace.  My guess is that at the rate we are going we will end 2009 with about 385,000 – 395,000 new homes sold versus 485,000 homes sold last year and 776,000 sold in 2007.  Clearly we have a way to go before I think we can say we are in a recovery but hopefully we are close to the bottom of the new home market.

Related Posts

1. Reports and headlines say home sales increased 10.1 percent in October, up 23.5 percent from a year ago; I say up 6.6 percent for the month and almost even with last year
2. New Home Sales in US through October; UP 5.1 percent OR DOWN 24.1 percent from a year ago, take your pick
3. New homes started through October down 32.5 percent from last year; New home completions still outpacing sales by over 30 percent
4. New home sales up slightly in August; Prices down 7 percent for the month
5. ‘Actual’ Existing home sales drop 5.2 percent in November; Up 4.2 percent from a year ago