As the saying goes, there are lies, damn lies, and statistics, and nowhere can that be more true than the stats that we hear about when politicians are talking about unemployment, notes John Mauldin of Thoughts from the Frontline.

The non-farm payroll numbers showed an increase of just 96,000 jobs in August. This was almost dead on the average for the last six months, which is 97,000—down from 205,000 the previous six months, echoing the pattern of last year and suggesting the numbers for the next two months will be soft as well, just prior to the election.

Almost all of the growth is in the birth/death number, which this month was 87,000 new jobs. For new readers, this is the number that the Bureau of Labor Statistics creates to account for new businesses created that were not part of the employment survey. The BLS surveys established businesses to get the employment data (thus, this survey is called the establishment survey).

Since, almost by definition, they can't survey new businesses, and net new businesses are one of the more important parts of the employment picture, BLS makes an estimate based on historical data. Then over time, they revise the overall numbers. The direction of the revisions is very important, as it tells us much about the underlying employment trends.

The revisions were all negative this time. That suggests to me that this month's birth/death number is probably higher than it should be, and over time it will get revised down as well. Ugh.

Looking at the data, we find that 28,000 jobs were created in the bar and restaurant business. Those are not exactly high-paying jobs. My anecdotal observation is that more than a few college graduates are taking those jobs.

The average earnings data was flat for the month, as the year-over-year number fell to +1.7%, which is below inflation. Workers are falling behind.

Worse, hours worked fell by one-tenth of an hour. That doesn't seem like a lot, but when you add the hours up, that one-tenth is the equivalent of several hundred thousand jobs in terms of actual pay, which is of course a drag on consumer spending.

We are down 4.7 million jobs from the pre-recession peak. At less than 100,000 jobs added per month, it would take four more years just to get back to where we were five years ago. Which would not appreciably reduce the unemployment rate, due to the growth in population.

All that was disappointing, but not anything we should be surprised by. No, the thing that caused me to do a double-take was the drop in the unemployment rate from 8.3% to 8.1%.

How can only 96,000 new jobs cause a drop in the unemployment rate of 0.2%? That is really a rather large number and should have required closer to 250,000 to 300,000 new jobs. In addition, the population actually grew by 213,000.

"The number of unemployed fell by 250,000, and the unemployment rate fell by 0.2 to 8.1%. But those declines were largely the function of labor force withdrawal.

"Flows data confirm this suspicion: 195,000 went from employed to not in labor force, and 226,000 from unemployed to not in labor force, both very high numbers by historical standards. The number classed as not in labor force but wanting a job rose by 437,000 to 7 million, or 2.9% of the population, the highest of the Great Recession/Tepid Recovery cycle." (The Liscio Report)

The U3 unemployment rate has dropped from 9.1% last August to 8.1% this August. How does that square with numbers in the real world? The working-age population has risen by 3.7 million people, and the number of employed people (including part-time) has increased by 2.3 million. How can an increase in the number of people that logically seem unemployed translate into a reduction in the unemployment rate?

To reduce the unemployment rate, the trick is to reduce the number of people in the work force by significantly more than the number of new jobs. Under the rules, 368,000 people were removed this month from the labor force. The labor force is made up of those who are either employed or unemployed, but you are not considered unemployed if you have not looked for a job in the last four weeks, or are in school, or...

Next: The unemployment-to-population ratio

|pagebreak|

Now, here is something Lacy told me today in a private conversation, as we were talking about the rather large drop in the labor force, and it surprised me. He noted that you are not considered to be in the labor force if you have not been considered unemployed during the last year. (I must admit I cannot find that on the BLS Web site. And at 3 AM in the morning, it would be rude to call his room. I will confirm this later today and note that to you next week if I got the details wrong.)

Here is the interesting thing. It used to be that you were in the labor force if you had been looking for work sometime in the last four years, but that was changed to one year in the latter Clinton years.

If we used that older (and to my mind, more reasonable) standard, the unemployment rate would be at least 1% higher, and perhaps a lot more. I call the old standard more reasonable because it counts discouraged workers who would take a job if they thought they could find one.

Look at this graph from the St. Louis Fed FRED database. The labor force participation rate is in red, and the employment-to-population ratio is in blue. The employment-to-population ratio is down to where it was in the late 1970s. The participation rate is down to 63.5%, roughly where it was some 30 years ago.

chart
Click to Enlarge

Note that if you are on disability, you are not considered to be in the labor force. As of April, we have added 5.4 million people to the disability rolls since the beginning of 2009. This is several million above the previous trend. There are now almost 9 million on disability. There are many who drop off each year, but many of them are going from disability to Social Security.

"As the Congressional Budget Office explains: "When opportunities for employment are plentiful, some people who could quality for [disability insurance] benefits find working more attractive...when employment opportunities are scarce, some of these people participate in the DI program instead."

"The explosive growth in disability enrollment also 'helps explain some of the drop in the labor force participation rate,' noted economist Ed Yardeni on his blog." (Investor's Business Daily)

In 1992, there was one person on disability for every 35 workers. It is now about one for every 16 workers. If disability had stayed at the pre-recession growth trend, unemployment would be at least 1% higher, and perhaps as much as 2%.

The bottom line is that true unemployment is closer to 10%, and perhaps significantly more. We just don't know. Underemployment is still in the range of 16%. And that does not count people who have a job for which they are far overqualified and who are making much less money than they would if they could find a job in their chosen field.

I should note to all those people who think I am being overly pessimistic that John Williams at Shadow Stats, who uses the US government methodology from 30 years ago, tells us that U-6 unemployment is around 23%. The difference is in how you create the model.

The feds keep changing the rules, and it should be no surprise that with each new rule the number of people officially counted as unemployed drops. And if you can't find a job, whether you are officially unemployed or not, it's no fun.

Subscribe to Thoughts from the Frontline here...

Related Reading:

Forget Gold: Platinum Is the Play
Is This the Week for QE3?
Keep Bilging and Paddle Like Hell