Latest Breaking News

Last edited Sat Jul 4, 2015, 11:36 PM - Edit history (3)

Quite a few not so good (or bad) numbers from the Household Survey. Remember most of the statistics in the Household Survey are very volatile from month to month. So it's important to look at the past 12 months figures (which are included) before deciding how the economy is doing lately.... Looking at the past 12 months smooths out most of the month-to-month volatility.

As an example of the month to month volatility, consider the month-to-month changes in the Household Survey's count of the Employed over the last 2 1/2 years:
http://data.bls.gov/timeseries/LNS12000000
[font face = "courier new"]
Thousands of employed, change from the previous month:
2013: 48 101 -55 291 225 135 250 13 9 -844 1037 181
2014: 535 95 495 -72 144 379 154 50 156 653 71 111
2015: 759 96 34 192 272 -56
January and February numbers are affected by changes in population controls.[/font]

And these are seasonally adjusted numbers! The unadjusted numbers are even more rocky.

This is far more volatile than the much more reliable Establishment Survey that produces the payroll employment:
http://data.bls.gov/timeseries/CES0000000001
[font face = "courier new"]
Thousands of payroll employees, change from the previous month:
2013: 205 314 115 187 219 127 164 256 150 225 317 109
2014: 166 188 225 330 236 286 249 213 250 221 423 329
2015: 201 266 119 187 254 223
The last 2 months are preliminary[/font]


Even worse than the Household Survey's Employed, the Household Survey's Labor Force has even more volatility. Here is the month-to-month changes in the Household Survey's Labor Force over the last 2 1/2 years:
http://data.bls.gov/timeseries/LNS11000000
[font face = "courier new"]
Labor Force, thousands, change from the previous month:
2013: 272 -429 -370 375 161 199 -129 -103 19 -933 689 -257
2014: 439 202 492 -760 209 71 348 -30 -173 398 159 -273
2015: 1051 -178 -96 166 397 -432
January and February numbers are affected by changes in population controls.[/font]

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 payroll employment increase is within +/- 90,000 jobs of the stated number. And a 10% chance that it is off by more than 90,000.

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 3.3 times the payroll employment's sampling error). Also, that there is a 90% chance that the unemployment rate is about +/- 0.2% of the stated number.

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

The individual components that go into these numbers have an even larger sampling error.

Right-wingers love to find the aberrant statistic or two of the month in these wildly volatile data series and make it out to be the story of the Obama administration, rather than what it really is -- just one month's number in a very statistically volatile data series.