Friday, October 5, 2012

The drop in the unemployment rate overstates the improvement in the economy


September private sector job gains, according to the establishment survey, were a bit weaker than expected (104K vs. 130K), and they have only increased at a very modest 1.1% annualized rate over the past six months, which is a lot slower than there 1.7% gain over the past year. According to this number, the economy is growing slowly and even slowing down some. But the unemployment rate fell unexpectedly to 7.8%—which is quite counterintuitive. For conspiracy theorists this is too good to be true, because it also happens to be exactly the rate during the month that Obama assumed the presidency.

I'm not one for conspiracy theories, but I note that the household survey, which is used to calculate the unemployment rate, showed a very large 660K increase in private sector jobs in September, most of which were part-time jobs. As the chart above shows, the household survey is much more volatile, on a month-to-month basis, than the establishment survey, and it has been unusually volatile this year. (There are several reasons for this, including faulty seasonal adjustment factor.) There are times, like now, when you just can't take the results of the household survey at face value; it's very unlikely that the economy suddenly and dramatically improved last month. Whether the numbers are fudged or not, you have to discount the household survey's figures substantially.

I've learned that one good approach to interpreting these two widely varying data series is to split the difference between them over some reasonable period. They don't track each other very closely from month to month, but over time they usually agree. So: since the end of January, 2010, when both surveys showed employment hitting a post-recession low, the two surveys have recorded the addition of approximately the same number of private sector jobs: 4.9 million according to the household survey, and 4.7 million according to the establishment survey. That works out to about 150K per month. That's OK, but it's far below what is needed if the economy is to get back on track. We need about 120K per month just to keep up with population growth. The jobs growth we have had in recent years is consistent with the unusually slow recovery that we've had. Most of the decline in the unemployment rate is due to a large number (5 million or more) of people dropping out of the labor force. You can see that in the chart below.


I end up seeing not much change in the economy as a result of these two reports. The economy is creating a modest number of jobs each month, but it is nothing to get excited about. It's likely that the strong growth in part-time employment that we see of late in the household survey is a by-product of the substantial decline in the number of people receiving unemployment compensation this year. As I've mentioned before, once their unemployment benefits expire, as they have for over 1.2 million this past year, people have an added incentive to find and accept a job, even if it pays little and even if it is part-time. It's good that they are working, but that means the jobs numbers are most likely overstating the effective improvement in the economy.

30 comments:

Gloeschi said...

For once, I must agree with Scott's assessment.
On that topic: how can government employment be 'revised'? The government doesn't know how many people it employs? I know, there are many different agencies etc, but what's the problem? "Well, this guy was half in, half out, we didn't know..."?
Anyway, seems that when Zerohedge says data is fudged, it's a conspiracy, but when cranky old Jack Welsh tweets about it, the market panics.

brodero said...

I am disappointed you chose to look at the data in a political light. Your perogative but it detracts from
your valuable blog. That being said,
The non seasonally adjusted jobless
rate (taking out seasonals) is now 7.57%...lowest since December 2008...
when President Romney ackowledge this data when it becomes very favorable
for him.

Anonymous said...

According to the household survey, 838K full-time jobs were added in September. Part-time jobs, on the other hand, *declined* by 26K.
http://www.bls.gov/news.release/empsit.t09.htm

Scott Grannis said...

I'm not saying the numbers are fudged. Whether they are or not is immaterial. They are so volatile of late that they must be discounted. And the large gains in part-time workers means the number of jobs overstates the actual improvement in employment.

brodero said...

I agree the rise of part-time workers is troubling. The marker is use is if
it rises year over year. Every recession has this number going up year over year first . 8.6 miilion September 2012...last year 9.3 million September 2011.

brodero said...

The BLS makes seasonal errors it doesn't fudge....you are feeding the ignorant

William said...

ECRI WEEKLY LEADING INDICATOR Dips

A measure of future U.S. economic expansion eased modestly last week though the annualized growth rate continued to climb, a research group said on Friday.

The Economic Cycle Research Institute said its Weekly Leading Index edged down to 126.3 in the week ended Sept. 28 from a revised 126.6 the previous week.

The index's annualized growth rate rose to 4.7 percent - its highest level since June 2011 - from 3.8 percent.

William said...

LIPPER FUND FLOW REPORT

Week of 10/03/2012
Equity Fund Outflows -$2.8 Bil;
Taxable Bond Fund Inflows $2.6 Bil
xETFs - Equity Fund Outflows -$2.4 Bil;
Taxable Bond Fund Inflows $2.4 Bil

Week of 09/26/2012
Equity Fund Inflows $1.1 Bil;
Taxable Bond Fund Inflows $4 Bil
xETFs - Equity Fund Outflows -$1.3 Bil;
Taxable Bond Fund Inflows $3 Bil

William said...

Schaeffer's Investors Intelligence

______% Bullish___% Bearish
10/03-----46.8-------25.5
09/26-----51---------24.5
09/19-----54.2-------24.5

Bill said...

Brodero,

Don't you think the U6 14.7% number is troubling? That number basically stayed the same. I don't believe in conspiracies either but I'm not sure we should jump for joy about this report.

Public Library said...

On the margin, employment is improving. That is all one needs to know. Rate back to where it was when the GOP handed over reigns. Leave it to the pundits to bicker over whether it should have recovered in 2 or 4 years.

Live prediction markets over at intrade have this horse race heading clearly in one direction…

steve said...

"Your perogative but it detracts from
your valuable blog."
just a pet peeve of mine but the word is PRErogative nit perogative. common mistake

brodero said...

Thanks I butchered a lot in writing today...guess I need that first drink...

Bill said...

Public,

Intratrade had the Supreme Court striking down Obamcare at over 75%
How'd that turn out?

Public Library said...

Bill,

The SC decision was a bi-nominal trade with hardly a scant of inside information about what the Justices were thinking. The election on the other hand, is covered incessantly night and day with poll after poll taken. Slightly different market me thinks…

Bill said...

Public,

Have you seen the Rasmussen polls today on the swing states? I think it's foolish to think the odds of Obama winning are anything close to 70% based on the small spread in these polls, both at the national and swing state level. Obama might win or might even have a good chance to win, but I seriously doubt it's as high as Intratrade's number. As far as the Supreme Court is concerned, most insiders believed it would be struck down (I spoke to some). Robert's decision was a big surprise.

Public Library said...
This comment has been removed by the author.
William said...
This comment has been removed by the author.
William said...

@Public Library -

The INTRADE prediction was as high as 78% just a few days ago and has fallen sharply to 68%.

Jim said...

"when President Romney ackowledge"

Your political sarcasm "in light" of a discussion is ironic. Learn to spell.

brodero said...

Understood Jim...spelling bad..."Will
President Romney acknowledge"

brodero said...

Jack Welch...

http://www.bloomberg.com/news/2012-10-05/welch-conspiracy-theory-on-jobs-data-not-tied-to-reality.html

brodero said...

Gallup 30 day rolling average (which is not seasonally adjusted so you use a year over year comparison)
Unemployment Level
Sept 12,2011 9.0%
Sept. 12,2012 7.9% change -1.1%

Non seasonally adjusted BLS household survey unemployment rate
September 2011 8.77%
September 2012 7.57% change -1.2%

brodero said...

The non seasonally adjusted BLS jobless rate usually drops for October before it starts going up in the later part of the year now how the BLS seasonally adjusts probably depends on the X-12-ARIMA program.
Somebody want to warn Jack Welch ahead of time?

brodero said...

If they used the same seasonal factor used for September 2004
the unemployment rate would have been 7.67% not 7.8%...

Seasonal factor for September 2012 .997495
September 2004 .998897

Anonymous said...

The problem with analyzing seasonal adjustments for something complex like the jobs report (same thing goes for durable goods orders and a lot of other reports) is that the entire report doesn't have a single seasonal adjustment factor; rather, each individual sub-category has its own seasonal factor. Health care jobs will have their own seasonal factor, retail jobs will have their own seasonal factor, government (think teachers) will have their own seasonal factor, and so on. Same goes with the household survey (mostly via age groups, ethnic/racial groups, male and female, etc.). Not sure you can make any conclusions about the overall seasonal factor for the jobs report by looking at the report as a whole without looking at the sub-components.

brodero said...

Don't tell Jack Welch..
Gallup Poll on US Employment
http://www.gallup.com/poll/125639/Gallup-Daily-Workforce.aspx

Set a new low at 7.6%

Unknown said...

Scott: Been tracking you and Wesbury among others for some time. Please clarify something related to this post. Your Household Survey (red line) seems to show a current reading of about 121k. The September report was +873k month over month and the trailing 12 month average was 221k/month. So, how does your "seasonally adjusted" figure reconcile with those numbers. Seems like your chart should have a larger reading for Sep 2012. What am I missing.

Unknown said...
This comment has been removed by the author.
Scott Grannis said...

John: you aren't reading the chart correctly. The y-axes represent the total number of jobs in millions of jobs. Not the monthly change in jobs in thousands.