In the comments to Doug Mataconis’s post at OTB on the September labor situation report from the Bureau of Labor Statistics, which see here, our old friend Michael Reynolds remarks:
Why are we still responding to the early reports? Have we not been through one revision after another?
which, as it turns out, is a very good question.
Every month the BLS issues a report on the employment situation for the previous month which includes a first revision for the second previous month and a second revision for the third previous month. So, for example, the employment situation report issued in October contains the first estimate for September, a first revision for August, and a second revision for July. Whether you think the revisions are better than the original estimate is in the eye of the beholder.
Here’s a summary of the results for the last twelve months:
Month | First estimate | First revision | Second revision | Direction of first | Direction of second |
September | 142,000 | ||||
August | 173,000 | 136,000 | Down | ||
July | 215,000 | 245,000 | 223,000 | Up | Down |
June | 223,000 | 231,000 | 245,000 | Up | Up |
May | 280,000 | 254,000 | 260,000 | Down | Up |
April | 223,000 | 221,000 | 187,000 | Down | Down |
March | 126,000 | 85,000 | 119,000 | Down | Up |
February | 295,000 | 264,000 | 266,000 | Down | Up |
January | 257,000 | 239,000 | 201,000 | Down | Down |
December | 252,000 | 252,000 | 329,000 | Up | |
November | 321,000 | 353,000 | 423,000 | Up | Up |
October | 214,000 | 243,000 | 261,000 | Up | Up |
The data are noisy, of course, but there are clearly a lot fewer jobs being created now than at the end of last year. My first order approximation suggests that revisions tend to be in the direction of trend but that may be a trick of the eye.
As to why we pay so much attention to data that are being revised, what fun would it be to ignore them? More importantly, revision is the way that science progresses and being subject to revision is actually more important than being measurable at all.
My conclusion is that we should be treating the present employment situation much more seriously than we are and that we need some policy revisions, something unlikely to occur with all of the hoopla about the 2016 presidential election. However, I’m glad to see the data and even happier that the BLS revises their data over time.
I saw that, and it is a good question. But it’s also Michael cleverly dodging the issue. When I go over there I still skim his comments. He wastes no time in quoting local statistics when they bolster his arguments, including that no Republican knows anything about economics or business. He speak with conveniently forked reserved tongue.
Economic data are always choppy and imprecise. Too many measurement problems and variables at work. That’s why you look at trends, or, if you are in the business of owning businesses and have representation in certain sectors, look to performance and reality in things you know about and experience. Manufacturing is on the balls of its ass. Capital spending is anemic. Consumer spending is reasonable, but no doubt driven by interest rate manipulation (home and auto), subsidized industry – eg health care, and creature comforts. (Hospitality). It should come as no surprise that employment trends follow, even if somewhat loosely. It’s not a solid economic framework we have here.
Of course, if we would just give more paid family leave and raise the minimum wage to $100,000/ yr…………
Thanks Dave, for such a concise summary of employment trends. Of course to some such stark data is irrelevant.