Updating DataQuick
(DQ) stats each week, you learn week-to-week blips are not always meaningful. A huge gain or drop is
often reversed a few weeks later. When the change is not quickly reversed, we instead get weeks and weeks
of boring, minimal change, draining the excitement from what looked like a trend change.
However,
this will not be not another rant against the smoothly rolling RE publicity machine of DQ. Instead we note
some anomalies which may finally(!), show somebody is actively manipulating this data.
First,
recall the May 21, 2008 press release included:
+"Home
sales were up 30 percent last month compared with March."
+"(T)he
steepest March-to-April increase in almost two decades for the county"
+"A
total of 1,440 new and resale houses and condos (aka "All Homes") changed hands in Santa Clara
County in April, up from 1,105 in March" (arithmetic shows that is indeed 30.3%)
DQ, perhaps
conveniently, did not publish an "All homes" figure in their March press release, so we can't compare
the press releases to see if the March figure was really 1,105.
What
we do have is March 27 data, at which point the "All Homes" figure was 1,198. This implies the last three days of March
saw a decline of -93 houses.
We also
know April 24-dated figures showed 1,140 sales. This implies the last six days of April saw an increase of +300 houses.
Isn't
this odd? March sees a decline in the last few days of the month to a new, all time low in sales. By
the 24th of April, sales are less than 5% above this figure. Then over the last six days of April an avalanche
of 300 sales (nice round number) comes to the rescue?
This
has got be it. We finally caught DQ fudging their numbers to generate a positive headline!
However, let's
go through the prior 12 months prior of month-end sales changes:
Press
Month Date Qty Release# Diff.
Apr'07 23rd 1,989 2,009
+20
May'07 24th 1,851 2,179 +328
Jun'07 28th 1,891 2,163 +272
Jul'07 23rd 1,898 1,910 +12
Aug'07 27th 1,670 1,908 +238
Sep'07 26th 1,459 1,235 -224
Oct'07 26th 1,134 1,381 +247
Nov'07 26th 1,128 1,317 +189
Dec'07 19th 1,399 1,265 -134
Jan'08 28th 910 869
-41
Feb'08 23rd 816 984
+168
Mar'08 27th 1,198 1,105
-93
Apr'08 25th 1,136 1,440 +304
In short,
nine of 13 months saw sales gains in the last few days of the month. A negative figure was actually unusual. April,
2008's +304 was not even the largest gain over that time span.
OK, this
is not the smoking gun we suspected. What about the flip side?
DQ's May
2-dated sales figure was back down to 1,256. This is decrease of -188 in two days! In
other words 60% of the 300-house avalanche for month-end April dropped back out of the data in the first two days of
May?
Building
a similar table comparing the month-end figure with the first report of each new month, we find:
Press
Month Release# Date Qty Diff.
Apr'07 2,009 10th 1,855 -154
May'07 2,179 8th 2,088 -245
Jun'07 2,163 10th 2,009 -154
Jul'07 1,910 7th 1,812 -98
Aug'07 1,908 7th 1,678 -230
Sep'07 1,235 9th 1,171
-64
Oct'07 1,381 5th 1,262 -119
Nov'07 1,317 7th 1,360
+43
Dec'07 1,265 7th 1,202
-63
Jan'08 869 6th
816 -53
Feb'08 984 10th 1,028
+44
Mar'08 1,105 7th 1,140
+35
Apr'08 1,440 2nd 1,256 -184
Ten of
the thirteen "new" months showed decreases. However, -194 was actually lower than a couple of the other months.
Again,
no smoking gun.
Conclusion:
Sometimes we dig into the numbers and find a revelation. Unfortunately, digging into these numbers does not uncover
DQ stepped out and cheated on last month's data. Based on researching the data, there is not smoking gun.
At least
not this month...
