by Calculated Risk on 9/04/2017 11:13:00 AM

Last week, in my employment preview, I noted "My sense (mostly based on history) is that job gains will be below consensus in August." Sure enough.

Here is a table of revisions for August since 2005.  Note that most of the revisions have been up.   This doesn't mean that the August 2017 revision will be up, but it does seem likely.   I'm not sure why the BLS has underestimated job growth in August (possibly because of the timing of seasonal teacher hiring and the end of the summer jobs).

August Employment Report (000s)
Year	Initial	Revised	Revision
2005	169	196	+27
2006	128	183	55
2007	-4	-20	-16
2008	-84	-267	-183
2009	-216	-213	3
2010	-54	-36	18
2011	0	110	110
2012	96	177	81
2013	169	261	92
2014	142	230	88
2015	173	157	-16
2016	151	176	25
2017	156	---	---

Note: In 2008, the BLS significantly under reported job losses. That wasn't surprising since the initial models the BLS used missed turning points (something I wrote about in 2007). The BLS has since improved this model.