View from Silicon Valley
Is DQ playing with the stats?
Home
Santa Clara Co. median (updated Aug16)
San Mateo Co. median
Santa Cruz Co. median
Santa Clara Co. stats (updated Jul15)
SEMI B:B to Apr'08 (updated Jul28)
SIA Data '04 -Jun'08 (updated Aug15)
Wafer area vs.SIA$ 4Q07(updated Jun21)
VC Funding -4Q07 (updated Apr27)
SV Stats (Updated!)
Links
About Us

July 7, 2006
 
Is DQ playing with the stats?
 
(c) copyright, View from Silicon Valley, 2006.  All rights reserved.
 
 
 
If you troll around the various housing blogs, you come across periodic bursts of opinion that local realtors, national realtors and even statistical agencies are conspiring to distort the statistics.  Reporting agencies allegedly omit some sales or fail to report others.  The MLS de-lists and re-lists houses to make the days-on-market look shorter.  The county allows various transactions not to be reported at the request of buyers or sellers or their agents.
 
And on and on.
 
We have a pre-typed request we occasionally insert into blogs hoping to learn if there is any actual evidence:
 
"I've seen a couple times where somebody thought DataQuick was unbiased?
 
Besides the obvious point they get their money FROM the RE biz, does anyone have any actual evidence?
 
 
Please advise?"
 
Unfortunately, nobody ever responded back to the web site nor in the blogs where such messages were placed.
 
This is the point where we write off the real estate conspiracies.  We put them in the same category as the Plunge Protection Team or the gold cartel or the central bank cartels.  Someday evidence may emerge that these conspiracies existed, but until then we need to focus facts which are already known.
 
Then we noticed an "interesting" anomaly in the weekly Santa Clara County DQ numbers:
 
On June 10, data was "as of" May 24, 2006  (17 days delayed).
On June 17, data was "as of" May 26, 2006  (22 days delayed, only 2 days later?).
On June 24, data was "as of" June 5, 2006.  (19 days delayed).
On June 30, data was "as of" June 5, 2006.  (26 days delayed).
 
Conspiracy theorists might quickly surmise DQ is trying to manipulate the information readily available to the public by delaying their month-end /quarter-end /half-year-end data.  How many real e
 
 
Maybe it's just a coincidence Santa Clara County's y-o-y volume numbers cut this way looked better?
 
            All Homes  Resales  Condos
On June 10:  -28.6%     -32.5%  -33.1%  (May 24)
On June 17:  -24.3%     -28.1%  -30.1%  (May 26)
On June 24:  -31.8%     -35.4%  -40.5%  (June 5)
On June 30:  -23.8%     -27.2%  -34.5%  (June 5)
 
Right away, you see two different data sets for June 5, 2006.  We could go wild & claim still more manipulation but a simpler explanation is probably true-- DQ (and/or the newspaper) forgot to change the date last week.  This stands out as possible in no small part because we've been known to make this error on occasion ourselves.
 
Even so, the June 10 report showed the weakest y-o-y volume YTD in 2006 (or 2005 or 2004 for that matter), but the June 17 numbers showed improvement.  Since many buyers close right at month-end, it's reasonable to imagine they study the stats as month-end approaches.  This June 17 report may have appeared to give buyers encouragement by implying May y-o-y volume was improving.
 
However, the June 24 report shows a fresh YTD y-o-y low. (Also true for y-o-2y and y-0-3y.)  Hmmm...
 
In the seven weekly reports before June 10, the "as of" date increased between seven and 10 days every Saturday.  Then eight weeks back we find:
 
On April 15, data was "as of" March 28, 2006  (18 days delayed).
On April 22, data was "as of" March 30, 2006 (23 days delayed, only 2 days later?).
On April 29, data was "as of" April 11 (18 days delayed).
 
             All Homes  Resales  Condos
On April 15:   -5.6%     -10.6%  -16.4%  (March 28)
On April 22:   -9.7%     -16.0%  -18.1%  (March 30)
On April 29:  -12.3%     -19.7%  -18.3%  (April 11)
 
Again, the decline in the numbers during the "short" week was less than what showed up the following week when a more "normal" number of days was added.
 
San Mateo numbers show a similar phenomenon.
 
On June 10, data was "as of" May 18, 2006  (23 days delayed).
On June 17, data was "as of" May 23, 2006  (25 days delayed, only 5 days later?).
On June 24, data was "as of" May 31 2006.  (24 days delayed).
On June 30, data was "as of" June 6, 2006.  (24 days delayed).
 
Again, maybe it's just a coincidence San Mateo County y-o-y volume numbers cut this way looked better?
 
            All Homes  Resales  Condos
On June 10:    -8.9%    -6.7%    -8.9%  (May 18)
On June 17:    -9.0%    -5.2%   -24.5%  (May 23)
On June 24:   -22.6%   -22.8%   -26.0%  (May 31)
On June 30:   -20.4%   -19.0%   -31.4%  (June 6)
 
San Mateo County numbers are less consistent than Santa Clara's but the addition of five days to June 17 data was only the second time in the last 10 weeks so few days had been added between Saturdays.  Just like Santa Clara County, the figures declined substantially the following week when a more "normal" number of days were added.
 
Santa Crux County numbers seemed a bit "cleaner":
 
On June 10, data was "as of" May 18, 2006  (23 days delayed).
On June 17, data was "as of" May 26, 2006  (22 days delayed, eight days added).
On June 24, data was "as of" May 31, 2006.  (24 days delayed).
On June 30, data was "as of" June 9, 2006.  (21 days delayed).
 
Maybe it's just a third coincidence Santa Cruz County y-o-y volume numbers with a more "normal" number of days, look better?
 
For ten consecutive weeks, the weekly report consistently added between six and eleven days of data.
 
            All Homes  Resales  Condos
On June 10:   -23.1%    -25.5%  -48.3%  (May 18)
On June 17:   -20.9%    -24.1%  -24.5%  (May 26)
On June 24:   -35.3%    -38.4%  -40.0%  (May 31)
On June 30:   -16.4%    -24.9%  -14.3%  (June 9)
 
Like Sun Mateo County, Santa Cruz data is spottier than Santa Clara County.  Even so, Santa Cruz County makes three-for-three showing improvement in mid-May.  Like the other two counties, the third week then shows substantial incremental weakness in the y-o-y numbers.
 
Does this "prove" DataQuick manipulates the numbers?  Not really.
 
It does suggest DQ might be manipulating the amount (or number of "days") of data released.  Similar to how the SEC is "proving" companies manipulated stock options because they always picked a really good date to make an award, there is some evidence of manipulation in DQ's "as of" dates.
 
On the other hand, there are several logical reasons for DQ's data to look this way:
1) DQ is running a slap-dash operation and data comes out whenever it comes out, or
2) Y-o-y numbers are crashing hard and any random variations in the "as of" data serve to exaggerate the weakness when "missing" data is added back in.
 
Slicing the data in different ways cannot hide:
3) Y-o-y volume was the weakest in two years "as of" mid-May, 2006.
4) Data "as of" end-May and early-June, was even weaker than mid-May.
 
If anybody has harder evidence of manipulation, we would love to see it!
 
Conclusion:
Just like Wall Street reporting on the stock market, RE-related data tends to be produced by people with a vested interest in keeping real estate prices going up.  If you just listen to the "pros," it's always a good time to buy on Wall Street and, apparently, also in real estate.
 
Figures don't lie, but liars can figure.
 
Looking at the current trends, it may soon become impossible to disguise the true state of the Silicon Valley real estate market.
 
* * * * *
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.