Free Markets, Free People

# Counting Unemployment

The most recent release of unemployment data has raised some questions, namely, how can we lose 20,000 jobs in the same month that the unemployment rate declined to 9.7%.  The answer is simple: The unemployment rate is essentially a made-up figure.  And I can give you a much more accurate way to measure the unemployment rate.

First, let’s take a brief look at how the monthly Employment Situation figures are compiled by the Bureau of Labor Statistics.  The BLS combines two surveys to compile the Employment Situation.  The first survey is the Establishment Survey.  That’s a pretty accurate survey, because it consists of asking businesses to provide hard payroll data on the number of existing jobs.   The second is the Household Survey, which is where the train runs off the rails.

For the Household survey, they ask if you are employed.  If the answer is “No”, they then ask if you if you’re actively looking for a job. If the answer is no, then they just simply take you out of the labor force.  They don’t care whether you aren’t looking for work because you know there are no jobs available, or whether you’ve retired and are planning to sail a sloop across the Pacific.  If you aren’t actively looking for work, you aren’t part of the labor force.  So, the official unemployment rate generally understates–sometimes substantially–the real level of unemployment.

Fortunately, there is a better way to calculate the rate of real unemployment, and the BLS web site conveniently provides you with all the data you need to do it.  From here, we only need three items: The Civilian Noninstitutional Population, the Participation Rate, and the number of Employed.

The first thing we need to do is figure out the Labor Force Participation Rate during the most recent period of full employment.  If you take the average monthly labor force participation rate from the 70 months between Jan 04 and Oct 08, you get a participation rate in the labor for of 66% of the Civilian Noninstitutional Population.

Next, you multiply the Civilian Noninstitutional Population by 0.66.  That gives you the size of the normal labor force at full employment.

Next, you take the number of Employed, and calculate the actual rate of unermployment using the following equation:

1-(Employed/Normal Labor Force)=Unemployment Rate.

So, with this method, we can compare the unemployment level of Oct 08, right before the economy cratered, to last month.  When we do so, we get the following results:

OCT 08:
Civilian Noninstitutinal Population:
234,612,000
Participation Rate: 66%
Labor Force:
154,843,920
Employed: 145,543,000
Unemployment Rate: 6.0%

Jan 10:
Civilian Noninstitutinal Population:
236,832,000
Participation Rate: 66%
Labor Force:
156,309,120
Employed: 136,809,000
Unemployment Rate: 12.5%

Note that this calculation for Oct 08 is very close to the official unemployment rate of 6.1%.  But as the economy gets worse the official employment rates show greater and greater variance.  In other words, the official unemployment rate becomes progressively less accurate as the Employment Situation worsens, substantially understating the actual rate of unemployment.  This is, by the way a feature of the BLS’s method, not a bug.  It is no coincidence, as our Soviet friends used to say, that discouraged workers fall out of the labor force calculations.

Now, this measure I’ve explained doesn’t tell us anything about people who are working only part-time, when they’d prefer a full time job, so it doesn’t tell us much about underemployment.  But it does tell us, based on the recent historical labor force participation rate, what the size of the labor force should be.  Once we know that, it becomes very easy to see what the actual rate of unemployment is in real terms, rather than the notional terms provided by the Household Survey.

According the BLS, however, the Civilian noninstitutional population has increased by 2,220,000 people  from 234,612,000 to 236,932,000, while, at the same time, the civilian labor force has shrunk by 2,055,000 people  from 155,012,000 to 153,455,000.  Using the BLS numbers, then, the labor force participation rate is 64.6%.  That kind of demographic change might be expected in a couple of years when the baby Boomers begin retiring in large numbers, but for right now, it seems…counter-intuitive.

In any event, 12.5% unemployment is a far more realistic number than the BLS estimate of 9.7%.

 154844

## 9 Responses to Counting Unemployment

• Bingo!

• And just how do they derive those Civilian Noninstitutinal Population, Participation Rate, Labor Force and Employed numbers?
For regularly employed they could use tax withholding records, but self-employed only report that data quarterly.
Estimates are usually pulled out of the anal cavity.

• Through this in with that whole “deficit of trust” meme.  I know that politicians lie and cheat and steal, but it seems that we’ve been getting a good, hard look at just how much they do it over the past few years.  What DOESN’T the government lie about?  What numbers DOESN’T the government twist, cook, or flat make up?

George Orwell
1984

• It’s been years since I read 1984, so I’m wondering what Orwell would say about calling a 5% spending increase, after a proposed 7% increase, a “spending cut”.

• I don’t think you’ve got a good enough reason not to trust their 64.6% Participation rate.  There is more to reducing that number than retirees.  The participation rate starts at 16 years of age and goes up from there.
Women 16 and Older, Women 20 and Older, and Both 16-19) you’ll see that there has been a very significant drop in the workforce participation of of 16 to 19 year olds, from 48.2 % in Jan of 2000 to 32.4% in Jan of 2010.

• Darn it, it dropped part of my statement, the second paragraph should start,
If you’ll look at the breakdowns for the participation rates (Men 16 yrs and Older, Men 20 yrs and Older, …

• You misunderstand my point.
In normal economic conditions, these people DO participate in the labor force. So my quibble isn’t that the labor force is evaporating according to the BLS definition of the labor force. My quibble is with the definition of the labor force.
The decline in the labor force is essentially people who would have a job if it was available. So, magically removing them from the labor force simply hides real unemployment.
The real question is not haw many people are participating in the labor force right now. It’s how many people normally participate in the labor force.
Theoretically, according to the BLS definition, you could lose 9 million payroll jobs, and 9 million people could become discouraged and drop out of the labor force. In that case, the unemployment rate would remain essentially unchanged. But, of course, that can’t possibly be an accurate measure of the unemployment rate, can it?

• I think this is a very telling calculation and like all truly brilliant ideas, upon hearing it I wonder why nobody seems to have thought of it before, particularly why it hasn’t occurred to the super-geniuses at BLS.  On the one hand it seems perfectly reasonable not to include people not looking for jobs in calculating the unemployment rate, since we wouldn’t want  college students, early retirees, and stay-at-home moms counted as unemployed if they’ve chosen to not participate in the work force.  But surely the BLS knows the limitations of their figures and is capable of coming up with something like this.

I would think, though, that it’s only good for short term comparisons like the one you’ve used it for, periods too short for cultural changes.  If, for example, we entered an era where most families with children could afford for one parent to stay home, and if the culture changed in such a way that this situation was normalized and/or highly prized, we might see a large number of people drop out of the labor force.  Because of this, Dale’s calculation wouldn’t make good comparisons between, say, now and the 1950’s, or now and some future era where we retire at 50 because we have earned enough to support us for the rest of our lives.