Corrente

If you have "no place to go," come here!

BLS Jobs Report Covering July 2012: Looks Can Be Deceiving

The big number for July is 163,000 jobs, seasonally adjusted, added to the economy. This is a good, but not great number. A great number, a number indicative of a strong expansion would be 200,000+. For example, in 1998, job creation seasonally adjusted averaged 227,000 jobs a month; in 1999, it was 254,000. Unfortunately, that 163,000 number isn't real.

(Revisions to the jobs figures, seasonally adjusted, for the last two months were something of a wash. May was increased 10,000 from 77,000 to 87,000 while June was even worse than thought dropping 16,000 from 80,000 to just 64,000.)

As I have been trying to point out recently, seasonally adjusted numbers are normalized. They do not reflect what is going on in the economy now. Rather they are a line drawn through the dips and rises that occur throughout the year (the unadjusted numbers), supposedly to show what the underlying trend is. As a result and especially with regard to this month's signature jobs number, the difference between the adjusted and unadjusted number can be radical.

The official unemployment rate, the U-3, also a seasonally adjusted number, edged up from 8.2% to 8.3%. It is back to where it was in January-February. That the jobs and unemployment numbers went in opposite directions is an artifact of the fact of the seasonal adjustments and that the two numbers come from different surveys.

Diving into this month's numbers, the potential labor force as represented by the non-institutional population over 16 (which is never seasonally adjusted) or NIP increased 199,000 from 243.155 million to 243.354 million. Multiplying this by the employment-population ratio (58.4% seasonally adjusted) gives us 116,000 an estimate of the number of jobs needed to keep up with population growth in July. July's jobs number beat this by 47,000.

Looking first at the Household survey, the actual labor force (the employed and those the BLS defines as unemployed) decreased 150,000, seasonally adjusted (SA), from 155.163 million to 155.013 million. This brings it back essentially to the May level of 155.007

Seasonally unadjusted (NSA), the labor force increased 141,000 from 156.385 million to 156.526 million. Contrast this with the 1.387 million it increased in June and you can see that the summer work season is over. Also if 2012 maintains the trend of the last decade, the July number represents the peak in the labor force for the year.

The labor force participation rate, the ratio of the actual labor force to the potential labor force (NIP) decreased 0.1% to 63.7% (SA) and remained unchanged at 64.3% (NSA).

The employed decreased seasonally adjusted 195,000 from 142.415 million to 142.220 million.

Unadjusted the employed decreased 76,000 from 143.202 million to 143.126 million. As I remarked last month, most employment (NSA) increases occur January to July. There may be an October or November spike which can even be higher than the July number with the December number being within ± 400,000 of the July figure. So on the ground, net hiring for the year is pretty much over. This is an important reason why I increasingly discount the seasonally adjusted, which are also taken to be the official, numbers.

The U-3 unemployed increased seasonally adjusted 45,000 from 12.749 million to 12.794 million.

Unadjusted, the unemployed increased 216,000. This is down from the 913,000 increase last month but is consistent with the influx of summer workers tailing off.

The U-6, the BLS' broader measure of un- and underemployment, increased seasonally adjusted 0.1% to 15.0%.

The seasonally adjusted U-6 rate (15.0%) represents 23.569 million people and reflects the 12.794 million of the U-3 unemployed seasonally adjusted, 8.246 million involuntary part-time workers (up 36,000 from June), and 2.529 million marginally attached to the labor force (up 46,000; these have looked for work in the last year but not in the last month).

The BLS measure of its undercount, those it defines as not in the labor force who want work but have not looked for work in the last month (seasonally unadjusted) decreased 320,000 in July from 7.157 million to 6.837 million (essentially back to the May figure of 6.835 million). You have to wonder where some of these people went given that the labor force (NSA) only increased by 141,000.

In my alternate calculation of the BLS undercount, I compare the current labor force to where we would expect it to be in a solid economic expansion: labor participation rate of 67%. The difference between these two is my measure of the undercount.

.67(243.354 million) = 163.047 million (where the labor force should be)

163.047 million — 155.013 million = 8.034 million (the real BLS undercount)

This is a increase of 283,000 from the June figure of 7.751 million.

With this number we can now go back and calculate where the U-3 and U-6 really are, that is the real unemployment and real disemployment rates.

Real unemployment: 12.794 million (U-3 unemployment) + 8.034 million (undercount) = 20.828 million (up 328,000 from June)

Real unemployment rate: 20.828 million / 163.047 million = 12.8% (up from 12.6% in June)

Real disemployment: Real unemployment + involuntary part time workers = 20.828 million + 8.246 million = 29.074 million (up 364,000 from 28.710 million in June)

Real disemployment rate: 29.074 million / 163.047 million = 17.8% (up from 17.6% in June)

The U-3 long term unemployed (those unemployed for more than 6 months and who have looked for work in the last month) decreased 185,000 to 5.185 million. As reported by the BLS, this is 40.7% of U-3 unemployed, down from 41.9% in June. (However, there appears to be a 46,000 discrepancy in the A-12 figures from which this is drawn and the U-3 unemployed.)

In unemployment by race, overall unemployment (SA) among whites was unchanged at 7.4% with male over 20 unemployment decreasing 0.1% to 6.9%; female over 20 unemployment increasing 0.2% to 6.8%; and teen unemployment both sexes increasing 0.6% to 21.5%.

African American unemployment decreased 0.3% to 14.1% with African American male over 20 unemployment increasing 0.6% to 14.8%; African American female over 20 unemployment decreasing 1.2% from 12.7% to 11.5%; and African American teen unemployment both sexes decreasing 2.7% to 36.6%. Overall the African American unemployment rate (U-3 seasonally adjusted) remains about twice that of whites.

Turning to the Establishment survey, 163,000 nonfarm jobs (SA), the headline figure were added in July bringing the total to 133.245 million.

However, unadjusted the total number of nonfarm jobs decreased 1.204 million. That tells a different story. But I don't want to scare anyone because this is a pattern that has been repeated for at least the last 3 years and mostly has to do with seasonal education job losses at the local level (and its economic effects are likely mitigated by unemployment benefits or in the case of teachers having their salary paid out over 12 months). What I wanted to point out was that while government lost 1.231 million going from 21.907 million to 20.676 million, the private sector only added 27,000 going from 112.165 million to 112.192 million. That's weak and it undercuts the official 163,000 (SA) number comprised of 172,000 private sector jobs minus a loss of 9,000 government jobs.

I also wanted to point out here (and for this I will just use seasonally adjusted numbers), in the 10 years from July 1990 to July 2000 (both of which were near pre-recession highs, so similar points in the business cycle), the economy added 22.227 million jobs. In the 12 years from July 2000 to July 2012, the economy has added 1.241 million, and is still 4.737 million below the last pre-recession high of December 2007. The startling difference between the 1990s and the period of the 2000s to now reflects slow job growth in the 2000s up to the beginning of the December 2007 recession where we fell off a cliff followed by renewed slow jobs growth. Remember too that jobs numbers do not take into account the growth in the US population over this time.

Average weekly hours for all private workers remained unchanged in July at 34.5 hours. Average hourly earnings increased 2 cents to $23.52 and weekly earnings increased 69 cents to $811.44.

Average weekly hours for production and nonsupervisory employees (blue collar jobs) also remained unchanged at 33.7 hours. Their average hourly earnings also increased by 2 cents to $19.77 and their average weekly earnings increased 67 cents to $666.25.

It continues to look like real wages, that is wages with inflation taken into account, are stagnant or losing ground.

Basically, we can say much the same about the real economy, as reflected in the unadjusted jobs numbers. Government lost a huge number of jobs but that was expected. The private sector, on the other hand, did much more poorly creating only 27,000 (NSA) as opposed to the much better 172,000 (SA) which accounted for the net 163,000 (SA) headline number. That 27,000 is significantly less than the 116,000 jobs needed to stay even with population growth.

And if we look at my calculated levels of real unemployment (12.8%) and disemployment (17.8%), both of which are partly based on seasonally adjusted numbers, the economy got worse last month. Disemployment is now above 29 million.

Sources:

Household data (Employment/unemployment)
Statistical significance: +/ - 400,000
The A tables: http://www.bls.gov/cps/cpsatabs.htm
A 1 for most information and categories
A 2 Unemployment by race
A 8 Part time workers
A 15 U 6 un- and under employment
A 16 Persons not in labor force

Establishment date (jobs)
Statistical significance: +/ - 100,000
The B tables: http://www.bls.gov/ces/cesbtabs.htm
B 1 Total jobs and jobs by industry/type
B 2 Weekly hours, all employees
B 3 Hourly and weekly earnings, all employees
B 6 Weekly hours, blue collar
B 7 Hourly and weekly earnings, blue collar

pdf of jobs report

0
No votes yet

Comments

twig's picture
Submitted by twig on

I am wondering if people who freelance are counted in any way. It would seem like writers, photographers, artists, etc., would not be included. I can't imagine how they could be tracked, since they pick up a job here and there and then eat beans when it's slow.

They don't get unemployment (usually) and in some cases are probably paid under the table, so the earnings aren't always reported. There must be millions of freelancers around, whether by choice or not. Do you happen to know if they are included in the part-time category, or anywhere, for that matter?

Submitted by ubetchaiam on

"Average weekly hours for all private workers remained unchanged in July at 34.5 hours. Average hourly earnings increased 2 cents to $23.52 and weekly earnings increased 69 cents to $811.44."

"Average weekly hours for production and nonsupervisory employees (blue collar jobs) also remained unchanged at 33.7 hours. Their average hourly earnings also increased by 2 cents to $19.77 and their average weekly earnings increased 67 cents to $666.25."

Given that most job creation is in the service sector that doesn't pay well, what jobs are the above hourly and weekly earnings reflecting?

Submitted by Hugh on

Freelancers would be, I think, considered self-employed. They would be tracked in the Household survey (employed, unemployed) but not in the Establishment survey as to type of job.

The B-3 table contains information about average hours and wages by industry:
http://www.bls.gov/webapps/legacy/cesbta...

As you might guess the lowest paying industries are retail sales and hospitality and leisure.

Some of this information is also contained in the B-3 table on page 32 of the pdf version of this month's report.

Submitted by ubetchaiam on

I live in the San Diego area which has not been as affected as other areas and still has a good amount of construction going on and the wages -average- aren't anywhere near what the table says.
And $16.31 for an average retail trade hourly wage? ROFL.

But thanks for the reference; with these kind of numbers ,it's no wonder they don't want to adjust the monies associated with what's considered the 'poverty level'.

Submitted by Hugh on

Freelancers would be, I think, considered self-employed. They would be tracked in the Household survey (employed, unemployed) but not in the Establishment survey as to type of job.

The B-3 table contains information about average hours and wages by industry:
http://www.bls.gov/webapps/legacy/cesbta...

As you might guess the lowest paying industries are retail sales and hospitality and leisure.

Some of this information is also contained in the B-3 table on page 32 of the pdf version of this month's report.

Submitted by Hugh on

The self-employed are not counted in the Establishment survey. They are counted in the Household survey.

The household survey provides information on the labor force, employment, and unemployment that appears in the "A" tables, marked HOUSEHOLD DATA. It is a sample survey of about 60,000 eligible households conducted by the U.S. Census Bureau for the U.S. Bureau of Labor Statistics (BLS).

This survey is conducted usually in the business week in which the 12th of the month occurs. The treatment of a self-employed person would be no different from any other worker.

So if a self-employed person worked more than 34 hours that week, they were full time. If they worked 1-34 hours, they were part time. If they wanted to work fulltime but couldn't find a fulltime job or work was slack, they would be a part timer for economic reasons. If they worked 1-34 hours and this was their choice or due to child rearing responsibilities, Social Security limitations, etc., they would be classed as part time for non-economic reasons. If they had no work for that week but had looked for work in the previous month, they are classified as unemployed. If they had no work in the reference week and had not looked for work in the previous month, they would be defined out of the labor force.

The A-8 table contains information on the number of unincorporated self-employed. Incorporated self-employed are folded into the category wage and salary workers.

http://www.bls.gov/webapps/legacy/cpsata...

mtngun's picture
Submitted by mtngun on

Great breakdown of the data, Hugh. I always look forward to your take on the job numbers.

shargash's picture
Submitted by shargash on

on the BLS numbers? I'm used to (well-done) surveys having error bars of +/- 2% or so. If the BLS has error bars anything like that, the 163,000 number is a joke -- 163,000, plus or mins 3 million or so. If they don't have error bars like that, how to they get it so precise? 163,000 is 1/10 of one percent of the labor force.

Submitted by Hugh on

I added a source note at the end of the post. For the jobs number, the statistical threshold for significance is 100,000. However as I pointed out in the post, 163,000 is a normalized number reflecting seasonal adjustment. It does not mean that 163,000 jobs were created in July. Indeed 1.2 million jobs were seasonally lost in July. 27,000, a not significant number of jobs were created in the private sector. The 163,000 is just a way of drawing a line through these big swings. It doesn't result in a straight line, but does flatten the hills and valleys. It's defensible if the economy and Americans are doing OK, but it is deceptive, as now, when they're not. It's more informative and more honest to look at the unadjusted numbers and doing year over year, or longer, comparisons to look at longer term trends.

shargash's picture
Submitted by shargash on

I did, however, major in physics in college. IMO, the idea that the BLS is accurate within 0.000625 (100k out of 160m) is ludicrous. But then that's one of the problems I have with economics in general: economists pretend they know things with a precision they do not have.