Dirk's analysis is excellent! But the NAR September numbers are really for sales in July and August (they are counted at closing, not when the agreement is signed) - so counting the Saturdays in September doesn't make sense to me.
I have expressed my problems with the reporting of the SAAR number so now I'll ask a question, why is that the number the media reports?
If I were a newspaper editor (and I was) I would start my stories with the actual data (which is what I did) and get to numbers which took some explaining at the end, for people who were looking for deeper analysis.
I'm guessing they are trying to make things simple, but in attempting to simplify, they are actually misinforming people. It's really quite bizarre. Especially from savvier outlets like Bloomberg.
CR, for year to year comparisons I only use the non-seasonally adjusted data. It makes no sense to do otherwise.
Also, last year Katrina hit at the end of August, and suppressed home sales somewhat. This leads me to believe that Sept 2005 home sales are slightly seasonally underadjusted, and thus that comparing Sept 2005 actuals to Sept 2006 actuals gives, if anything, a slightly rosier picture than warranted of the true situation.
Good catch MOM
But as per CR "But the NAR September numbers are really for sales in July and August" so Katrina effects (the natural disaster causing Low stat) will be in effect on the next month's report, yes? In any case, the regional breakdown should show an improvement in the South y/y as the 2005 sales for Sept should be a dismal benchmark.
We think the worst is yet to come in Santa Clara County. Thru the September report, Santa Clara County y-o-y volume was negative for the 22nd straight month (since December, 2004) and 25th out of the last 26 months.
In the last 12 months, new permits issued represent a 4.3-month supply, regardless of other houses for sale. Since the "bottom" in late-2002, total permits issued represent a 16.6-month supply.
During that same time, jobs & population are essentially flat.
There were fewer business days this year vs. last year. The Lawler Economic & Housing Consulting Report correctly predicted that the seasonally adjusted home sales number would show a smaller yoy drop than the not seasonally adjusted numbers. The NAR uses the X-12 ARIMA model that includes trading day and holiday adjustments.
The LEHC prediction for seasonally adjusted sales was 6.185 million (they track local markets and do a bottoms up projection), which was the best of all predictions I saw.
I'd be most happy to send out October 19th daily that had the LEHC projection.
Dirk's analysis is excellent! But the NAR September numbers are really for sales in July and August (they are counted at closing, not when the agreement is signed) - so counting the Saturdays in September doesn't make sense to me.
Thanks to Dirk.
Best Regards.
It is easy. Just like I've been saying. All we need is instances of unfavorable formulaic adjustments and I'll shut up.
I have expressed my problems with the reporting of the SAAR number so now I'll ask a question, why is that the number the media reports?
If I were a newspaper editor (and I was) I would start my stories with the actual data (which is what I did) and get to numbers which took some explaining at the end, for people who were looking for deeper analysis.
I'm guessing they are trying to make things simple, but in attempting to simplify, they are actually misinforming people. It's really quite bizarre. Especially from savvier outlets like Bloomberg.
CR-
Do you have a spreadsheet you could post with the New Home Sales and prices by month? I would like to do analysis of the dollar volume of sales.
PW, for New Home data, you can get it at the Census Bureau (pdf):
New Home Sales
New Home Prices
Remember the last few months will be revised. I have little confidence in prices until they have been revised a few times.
Best Wishes and Have fun!
Thanks for the additional info, CR! You rock.
--Bill B.
CR, for year to year comparisons I only use the non-seasonally adjusted data. It makes no sense to do otherwise.
Also, last year Katrina hit at the end of August, and suppressed home sales somewhat. This leads me to believe that Sept 2005 home sales are slightly seasonally underadjusted, and thus that comparing Sept 2005 actuals to Sept 2006 actuals gives, if anything, a slightly rosier picture than warranted of the true situation.
Good catch MOM
But as per CR "But the NAR September numbers are really for sales in July and August" so Katrina effects (the natural disaster causing Low stat) will be in effect on the next month's report, yes? In any case, the regional breakdown should show an improvement in the South y/y as the 2005 sales for Sept should be a dismal benchmark.
We think the worst is yet to come in Santa Clara County. Thru the September report, Santa Clara County y-o-y volume was negative for the 22nd straight month (since December, 2004) and 25th out of the last 26 months.
For more data, please visit:
http://www.viewfromsiliconvalley.com/id273.html
We also track housing permits issued at:
this
In the last 12 months, new permits issued represent a 4.3-month supply, regardless of other houses for sale. Since the "bottom" in late-2002, total permits issued represent a 16.6-month supply.
During that same time, jobs & population are essentially flat.
For more Silicon Valley-related news, please visit:
http://www.viewfromsiliconvalley.com
Thanks!
There were fewer business days this year vs. last year. The Lawler Economic & Housing Consulting Report correctly predicted that the seasonally adjusted home sales number would show a smaller yoy drop than the not seasonally adjusted numbers. The NAR uses the X-12 ARIMA model that includes trading day and holiday adjustments.
The LEHC prediction for seasonally adjusted sales was 6.185 million (they track local markets and do a bottoms up projection), which was the best of all predictions I saw.
I'd be most happy to send out October 19th daily that had the LEHC projection.
Any one have a source where they show the original estimate vs final revised numbers... as a monthly series going back say ten years?
Now THAT would be interesting to analyze.