I was out on Friday, so I didn’t get a chance to review the GDP revision in the BEA release (I take one day off, and that’s what happens).
Here’s the official announcement:
For 2001-2004, real GDP grew at an average annual rate of 2.8
percent, 0.3 percentage point less than in the previously published
estimates. The average annual rate of growth of real GDP from 2001:IV
to 2005:I is 3.3 percent, 0.2 percentage point less than in the
previously published estimates. Revisions to year-to-year growth rates
were small.Revisions to 2002-2004 estimates
The percent change from the preceding year in real GDP was revised
down for all 3 years: From 1.9 percent to 1.6 percent for 2002, from
3.0 percent to 2.7 percent for 2003, and from 4.4 percent to 4.2
percent for 2004.
Those are pretty hefty downward revisions of 0.3%, 0.3% and 0.2% per year.
Bottom line: The tendency is for the initial read on GDP to be revised upwards in the subsequent 2 releases (Advance, Preliminary and Final), and then ultimately to be revised back downwards (closer to reality) — at keast in this post-recession cycle; I haven’t gone back over the historical numbers far enough to conclude whether this has occurs in this order traditionally.
As we discussed yesterday in Juiced Data, if you massage numbers long enough . . . well you know the rest. But the same applies to charts: Depending upon the scale you select to use, your choice makes the revision look moreor less significant:
Bad:
click for larger graphic
Chart courtesy of ArgMax
>
Not so Bad:
click for larger graphic
Chart courtesy of Northern Trust
(The full data can be found at BEA)
Here’s the tell: Anyone who relies upon the initial read is likely to have a below consensus expectation for growth; Those who rely on the 2 subsequent revisions, will expect above consensus growth.
>
Am I not paying attention, or is the Bush administration not the record-holder for most revisions to GDP and employment figures?
They always seem to trumpet whatever mediocre numbers come out and then more often than not, a quiet revision several weeks later squeaks out on a late Friday afternoon?
FYI,
There’s a scheduled GDP revision (to one quarter’s worth of data, always at 8:30am, usually on the last business day of the month) every month because the Commerce Department has to make assumptions for the data that are not collected in time for the initial release (inventories & int’l trade, mostly). As more data are received, the numbers for the relevant quarter are updated.
There are also larger benchmark revisions that deal with changing statistical methodologies, etc. that happen more rarely (annually, and every four years).
They always seem to trumpet whatever mediocre numbers come out and then more often than not, a quiet revision several weeks later squeaks out on a late Friday afternoon?
Posted by Chinatronic.com