How does the payrolls report come out so quickly?
Grumpy Editor is grumpy — and today he’s grumpy about the monthly payrolls report.
Anyone dealing with the federal government knows it takes time to get responses to simple questions while tabulation of accurate official figures often takes months, sometimes years, so Grumpy Editor finds it interesting that the March “employment situation summary” from the Bureau of Labor Statistics was wrapped up immediately after last Thursday, the last day of the month, ready for dissemination within 8½ hours.
Statistics galore were in the lengthy news release embargoed until 8:30 a.m. (EDT) on April 1.
That’s speedy work on highly-followed data…
Eyebrow-raising editors, usually suspicious of figures — from attendance at baseball games to precipitation amounts from the Weather Service — were mum on the instant tabulations that made front pages in many newspapers around the country and led-off radio and television news on April 1.
This is an easy mistake to make, but it is a mistake. Most of us think that the monthly jobs report comes out on the first Friday of the month, but in fact it technically comes out on the third Friday after the end of the reference period. And the reference period, for any given employer, is the pay period that includes the 12th of the month (regardless of the length of that pay period).
Essentially, the U.S. Department of Labor sends out forms asking employers to report on their payrolls every month. The companies report what their payroll figures are for whichever pay period included the 12th of the month. The Bureau of Labor Statistics then collates all that data, and three Fridays after the 12th of the month, it releases a preliminary estimate for how many people had jobs that month, how much they were paid, and so on.
The following month, the BLS revises those numbers, and the month after that there’s a final figure. If you want to have a look at the size of the revisions going all the way back to January 1970, you can do so here. And if you’re feeling particularly nerdy, you can try and work out whether payroll reports which come out on the first of the month end up getting revised more, on average, than reports which come out on say the 7th of the month. (Seriously, if you want to do that, I’ll send you whatever books you want from the large pile which is growing alarmingly on my windowsill, and which can be seen in the background of this photo.)
The point here is that the payrolls report is a snapshot of a point in time; it’s not something you get from adding up daily data from each of the days in the month. And the snapshot is taken on the 12th of the month, not at the end of the month. The markets love the payrolls report because it’s super-fast, rather than because it’s super-accurate: it’s the closest thing we have to an official national data series showing how the economy’s doing right now. Yes, you can raise your eyebrows at how accurate it is — I’ve done so myself.
My take on the jobs report is that once upon a time it was useful, but that nowadays, with a large increase in self-employment and with the margins of error dwarfing the actual numbers reported, it’s becoming largely irrelevant. The only useful thing you can do with it is look at it through squinted eyes and try to discern vague trends.
But insofar as the jobs report is inaccurate or unhelpful, that’s not a function of the fact that it comes out towards the beginning of the month rather than towards the end; it’s much more a function of the rise in self-employment and small businesses, which the BLS has great difficulty measuring. Or at least I think that the date of release is pretty much irrelevant. If you fancy crunching the numbers and telling me for sure either way, as I say, I’d be much obliged.