Yeah, I’ve been harping on this but it has my attention and I’m dismayed that the mass media hasn’t picked up on it …
Last Thursday’s BLS release on Unemployment Claims did it again … revised upward the prior week’s estimate.
Let’s look at the numbers for the past couple of weeks:
In each of the past 5 weeks — maybe longer, I just started tracking then — the so-called “Advance” estimate of weekly unemployment claims eventually got revised upward … by a fairly consistent 1%.
There’s a name for a consistent unidirectional error in forecasts.
It’s called SYSTEMATIC STATISTICAL BIAS.
So, you gotta wonder: why haven’t the crack statisticians at the BLS noticed the bias and started correcting for it?
Either they’re incompetent, or they’re as biased as they’re data.
Why does it matter?
Because the “headline numbers” each week are calculated by subtracting the advance number — which is consistently understated — from the prior week’s final number — which is consistently raised up.
In other words, there is a consistent bias to report bigger drops in unemployment claims than are real … and in 2 of the past 4 weeks, to report drops in weekly unemployment gains when there were actually increases.
No wonder the President thinks the private economy is doing fine.