December 6, 2006
Even fishier
(c) copyright View from Silicon Valley,
2006. All rights reserved.
Tracking weekly housing statistics, you get used to being surprised on Saturday
morning. After several weeks of declines in the summer, Santa Clara County sales prices stabilized.
Given flat
prices, an objective observer expects y-o-y percentage price increases to steadily decline. However, published Santa
Clara County y-o-y price gain figures have stabilized, or even improved slightly, over the last few weeks. How is this
mathematically possible?
The relevant 2005 snapshot is:
November 26, 2005:
Median
Price Nov.09'05 from'04
All homes
$660,000 17.9%
Total resale houses $713,000 18.8%
Total condominiums $470,000
17.5%
Total new homes $660,000 1.4%
Then the current week's data is:
December 02, 2006:
Median Price
Nov07'06 from'05
All homes $675,000
2.7%
Total resale houses $732,000 3.1%
Total condominiums $500,000 5.3%
Total
new homes $612,500 11.4%
Ignoring the 2-day difference, an accurate 2006 table
would read:
December 02, 2006:
Median Price Nov07'06 from'05
All
homes $675,000 2.2%
Total resale
houses $732,000 2.7%
Total condominiums $500,000 6.4%
Total new homes
$612,500 -7.2%
"All Homes," "Total Resale Homes" and "Total
New Homes" are all wrong but in the "right" direction. (i.e. they all claim higher figures than proven by DataQuick's
own data.) "Total New Homes" is massively wrong, published at +11.4% when DataQuick's own data shows -7.2%.
The
"Total Condominiums" y-o-y price gain in Santa Clara County is actually under-reported. Checking further, we find errors
in reported y-o-y condo price changes in San Mateo County (+0.9% reported vs. data showing +1.2%) and Santa Cruz County
( -7.1% reported vs. data showing -7.5%).
When the y-o-y condo
number is positive, it's reported as less positive. When the y-o-y condo number is negative, it's reported as less negative.
Could this be a coincidence? Or an honest error? Perhaps.
On the other hand, if you're thinking of buying
or selling a single-family home, you might be re-assured to see condo prices are not inflating or deflating the market.
If there was a conspiracy to minimize the effect of condo sales on overall numbers, deflating
y-o-y sales percentages is an obvious place to start. (FYI, condo sales are 23% (Santa Clara County),
16% (San Mateo) and 18% (Santa Cruz) of last weeks' totals.)
Except for Santa Clara County "Total new homes," the changes
are small. How much impact can such mis-reporting really have?
Pressing on, the last couple weeks' data showed a narrowing the y-o-y decline in unit sales volume.
Admittedly,
there are several possible exaplantions. It could be:
1) a sign of impending local real estate strength,
2)
the reversal of 27-month trend,
3) an additional layer of DQ and/or SJMN incompetence
4) another example of View
from Silicon Valley making a mountain out of a mole hill or
5) another step in a data reporting conspiracy.
Y-o-y sales volume has been so negative for so long we readily admit it's possible the numbers
are actually becoming less negative. Unfortunately, as usual, we prefer to check the actual numbers and subject
our readers accordingly:
Santa Clara County: Oct.25
Nov'07 actual
Median Price Volume from'05 from'05 Nov7vs'05
All
homes 1,905 -28.8%
-26.8% -23.9%
Total resale houses 1,193 -27.4%
-25.9% -25.2%
Total condominiums 442 -28.9%
-27.9% -26.7%
Total new homes 270 -34.5% -28.8% -11.8%
San Mateo County Oct.20
Nov.03 actual
Median Price Volume from'05 from'05 Nov7vs'05
All
homes 558 -19.2%
-16.3% -20.1%
Total resale houses 455 -17.1%
-15.4% -18.9%
Total condominiums 91 -24.1%
-22.9% -26.6%
Total new homes 12 -44.4%
9.1% -7.7%
Santa Cruz County Oct.20 Nov.07
actual
Median Price Volume from'05 from'05 Nov7vs'05
All
homes 230 -43.4%
-15.8% -15.8%
Total resale houses 173 -39.8%
-18.0% -18.0%
Total condominiums 41 -41.5%
-4.7% -4.7%
Total new homes 16 -61.5% -15.8% -15.8%
Make no mistake, y-o-y volume is still
negative by double-digit percentages!
For reasons we cannot explain, Santa Clara County reported worse, San Mateo County reported better and
Santa Cruz County reported the same as the actual results.
Alert readers may also notice 2005 DQ reports tended to publish 15- to 18-day-old data.
Even with sales volume down, 2006 DQ data is running 22 and even 31 days behind. In a rising market, fresher data helps
lead to higher prices. In a falling market, older data helps hold up prices. Is this apparent shift to older data
just another coincidence?
At risk of going still another factoid too far, there has also been an increase during 2006 in
the numbers of weeks where the "as of" date failed to change from one week's publication to the next:
Number of weeks using same "as of" date:
2005 2006
Santa Clara County 1 6
San Mateo County 2 3
Santa Cruz County 2 6
These figures understate the change since 2006 "as of"
dates are still early November. Hopefully, we will see the next two months' worth of data sooner than mid-February.
Are these arrors a dramatic problem? No. Are they another sign the reliability, and possibly the
accuracy, of these figures is degrading? Almost certainly yes.
We want accurate, and timely, RE sales numbers.
Is this too much to ask?
Conclusion:
Do y-o-y unit volume errors on top of y-o-y price errors on top of publishing
"older" data on top of more weeks where the "as of" date is unchanged represent a smoking gun? Do they "prove,"
once and for all, real estate industry-generated stats are manipulated? Maybe, but not quite.
These anomalies
are, however, clear evidence statistics from the RE industry must be carefully scrutinized before being accepted as actual
facts.
* * * * *
The above is not intended as advice to buy, sell or hold any stock, bond, real estate nor
any other financial product or service. Invest at your own risk.