Payroll employment increases by 228,000 in November; unemployment rate unchanged at 4.1%

[progree]

22. 90% chance that the unemployment change is within +/- 300,000 of the stated value

Last edited Sat Dec 9, 2017, 10:37 AM - Edit history (3)

Unemployment rate = Unemployed/Civilian Labor Force:
October unemployment rate is 6,520,000/160,381,000 = 4.065% -> rounds to 4.1%
November unemployment rate is 6,610,000/160,529,000 = 4.118% -> rounds to 4.1%

That they are "statistically significant" doesn't mean they aren't very volatile with a huge margin of error.

On statistical noise, I found this BLS technical note on sampling error -- http://www.bls.gov/news.release/empsit.tn.htm . Based on what it says, there is a 90% probability that the Establishment Survey's non-farm employment increase is within +/- 120,000 of the stated number. And a 10% chance that it is off by more than 120,000.

And in the Household Survey, there is a 90% chance that the monthly unemployment change is +/- 300,000 of the stated number (note this is 2.5 times the Establishment Survey's employment's sampling error). Also, that there is a 90% chance that the unemployment rate is about +/- 0.2% of the stated number.

Given that November's unemployment change was +90,000, that means there is a 90% chance, based on sampling error alone, that the unemployment change is somewhere between -210,000 and +390,000.

The above only covers sampling error. There are also many other sources of error (search the above link for "non-sampling error" )

Here are the monthly changes in the Employed from the Household Survey, in thousands:
http://data.bls.gov/timeseries/LNS12000000
2016: 503 510 258 -273 30 32 456 109 271 -24 146 63
2017: -30 447 472 156 -233 245 345 -74 906 -484 57
January and February data are affected by changes in population controls.

Notice how it bounces around -- not surprising with statistical sampling error alone being +/- 300,000 (for the unemployed -- I don't know what it is for the employed ).


Ignoring monthly changes in January and February, the population controls adjustment months, and leaving out the two outliers: -484,000 and +906,000, it bounces around between about -250,000 and +500,000

This is the non-farm employee count from the Establishment Survey, in thousands
https://data.bls.gov/timeseries/CES0000000001

2016: 126 237 225 153 43 297 291 176 249 124 164 155
2017: 216 232 50 207 145 210 138 208 38 244(P) 228(P)
(P): Preliminary

With the much larger sample size of the Establishment Survey, and again throwing out the two outliers: +38,000 and +297,000, the monthly changes are much smoother, varying between about +40,000 and +290,000