June Jobs Report

[pinqy]

Everyone should remember that due to graduations, June has far more entrants into the civilian labor force, than your typical month.

But the official numbers are seasonally adjusted to account for that kind of noise.

You must be logged in to read this quote.

BLS doesn't kick anyone out of the civilian labor force...the categories are self reported.  If someone reports that they are not working and haven't looked in more than 4 weeks they (and proportionately the number they represent in the sample) are counted as Not in the Labor Force.  It's all based on the survey reponse, not some arbitrary categorizing as you seem to imply.

And as a matter of fact, the Labor Force has increased every month this year.

[pinqy]

Where to begin?

First, the Bureau of Labor Statistics used TWO different surveys with respect to employment.
One is the “Household” and the other the “Establishment” (employer).

True, but irrelevant to the discussion.  When talking about the Labor Force, we're only talking about the Current Population Survey (household) not the Current Employment Statistics (establlishment).

You must be logged in to read this quote.

The officially reported headline rate is the seasonally adjusted rate.  It's better for looking at month to month changes.  Looking annually, the average negates the seasonal effects anyway, so using the seasonally adjusted numbers as an annual average would be a distortion.  So while all publically released data is "official," the officially used number is the seasonally adjusted number.

You must be logged in to read this quote.

Now you’re referring to the birth/death model used in the CES, which is irrelevant to your claims about the CPS.  No such model is used for the CPS.  In compiling the CPS data, of course there are weights used in the aggregation and any individual household could represent anywhere from a couple of hundred to a couple of thousand other households, depending on the sampling area.  That’s standard, tested, statistical practice.  When using a sample of 60,000 households to represent around 250 million people, there are of course statistical “adjustments” but that’s not the same as the deliberate adjustments to gain a desired result you seem to be implying.

You must be logged in to read this quote.

Actually, the changes in population controls are made by the Census Bureau, not BLS (Census also does the collection for the CPS). 

You must be logged in to read this quote.

It wasn’t a “Presumption,” but a necessary change based on better data.  Deriving the national figures from the sample requires assumptions about the population.  Accurate population data cannot be done every month, but on an  annual basis a better population estimate can be made.  It’s not like it’s done in stealth mode or for any reason but to improve the data.

Keep in mind, by necessity, all the released numbers are estimates.  That’s all they can be.  Of course they’re not perfect, but they’re better than anything else.

But in any case I note you have failed to defend your claim about “adjusting” people out of the labor force.  Why is that?

[pinqy]

Assumptions and Methods Used in Preparing Employment Projections

The above include a number of “assumptions” made by the BLS in developing their adjustments to data.

Those are for Employment Projections, not the monthly Employment Situation.  Employment Projections is a 10 year projection of occupational employment (and Occupational Employment is yet another survey, unrelated to the CES).  Those assmptions and methods are not used for the CES or the CPS.

Projections are only 10 year projections specifically to avoid any appearance of trying to fit the estimates to predictions. 

[pinqy]

There is no way the labor force increased in January.  If it did, there would have had to have been close to a million jobs created that month to move the unemployment rate down from 9.4% to 9%.

Every month since January.  Looking at Dec-Jan labor force stats is tricky because January is when Census updates the population controls so there are all kinds of adjustments made to look at Dec-Jan changes. The press release for the Jan data explains what the adjustments and effects were.

[pinqy]

Jobs data slightly improved last month:

NEW YORK (CNNMoney) -- The job market got two optimistic signs Thursday as private sector employers added 157,000 positions in June and fewer people filed new claims for unemployment benefits, according to two reports.The 4-week moving average, which aims to smooth out volatility, came in at 424,750. That's a decrease of 3,000 from the previous week's average.

The jobs report is the ADP report.  Their sample is soley ADP clients.  Now that's a big chunk of employment, but hardly representative of all (and excludes all government jobs).  The BLS report tomorrow will give a more accurate jobs picture (ADP is rarely close to the BLS numbers).

[pinqy]

Sure, they're going to trend the same, but as a predictor of month to month changes, and monthly levels, ADP is often off by a bit and it does not seem to me to be particularly reliable as to what the BLS report will say.

[pinqy]

Like I said...the ADP report is not a good indictaor of the BLS report.  Crappy report, no real change (the UE rate change is not statistically significantly different from zero).

[pinqy]

A rather detailed (and hence lengthy) explanation of the machinations of the adjustment bureau.

Two things of note about that article:
1: the author doesn't give any suggestion on how to deal with the problem of business birth/death without a birth/death model.  
2. He fails to note that every year rebenchmarking is done based on a full count of the state records, so that any error from the birth death model is then corrected.