OK, I'm confused here.
Some one said that we lost 22,000 jobs
There were 480,000 new unemployment initial claims.
The unemployment numbers went down?
Why can't I see the math behind this miracle?
You can...you just don't know where to look.
First, you're talking about 3 completely different dat sets.
There are the Unemployment Insurances Initial Claims. This is weekly, comes from the state UI offices to the US Dept of Labor. These only measure new claims for UI...not employment, and not all unemployed. Released every Thursday:
UI report
Next is Employment. This comes from the Current Employment Statistics, a monthly survey of 140,000 businesses (410,000 worksites) conducted by the Bureau of Labor Statistics. Coverage period is the Pay Period that includes the 12th of the month. It only covers non-farm payroll jobs, so it excludes farm workers, the self employed, and domestic workers. It's a measure of jobs, not people...if a person holds two jobs, they're counted twice. But it's a fairly large sample, and the margin of error is not too great so it's the most accurate measure of employment, despite the limits of who is covered. Released on the first Friday of every month. Link:
Technical notes to the CES
And last is Unemployment. This comes from the Current Population Survey, a monthly survey of 60,000 households conducted by the Census Bureau on the behalf of the Bureau of Labor Statistics. This measures the Labor Force. It measures People, not jobs. The population is the civilian non-institutional population (those 16 and older not in prison, the military, or long-term physical or mental care facility). Coverage period is the week that contains the 12th of the month. People are asked if they worked during the reference week. If yes, they're employed. If no, they're asked if they looked for work in the 4 week up to the reference week. If yes, they're Unemployed. Employed + Unemployed = The Labor Force....all those actually participating in the Labor market. The rest of the population is "Not in the Labor Force." Some are "Marginally Attached," meaning they had looked sometime in the previous 12 months but aren't looking now for whatever reason but want to and are able to work. That's kind of subjective...they say they want to work, but they're not actually trying to. The preference, since 1941, has been to measure those actually participating, not those who would theoretically participate. Link:
CPS technical documentation
So...UI claims went up and down.
Employment (from the CES) went down 20,000, but the margin of error is close to 100,000, so the change is not statistically significant.
However, Employment from the CPS went UP. Remember, slightly different reference period, different sample, different concepts (For someone with two jobs, the CES counts twice while the CPS counts once. If a person loses one job, the CES will show a loss, while the CPS will show no change). And Unemployment went DOWN. So the Unemployment rate went down, even though though the official employment number went down.