Wednesday, October 19, 2016

Debunking a Doomer Dumpster Fire at Daily Kos

 - by New Deal democrat

I generally don't bother with this any more, especially since the site in question with rare exception long ago ceased to be a source of original work, but every now and then some dumpster fire of DOOM, usually derivative of Zero Hedge, shows up at Daily Kos.

Yesterday we got an example, and since there are people who still think there is some credibility there, I thought I would set the record straight.

The piece in question claims that the record low in initial unemployment claims is false, and that about half of all jobs created since 2009 are fake, based on the BLS's "birth-death" model.

Both of these claims are demonstrably false. 

Let's start with the second one.  The monthly jobs report is based on a survey of employers.  Since some new employers start up each month, and some others go out of business, and the BLS has no independent means of verifying how much of each there were in any given month, they use a model to estimate how many businesses were "born" or "died" that month, and plug that into the jobs number.

Here's what the diarist has to say about that:
 [W]hat about Obama's "15 million new private-sector jobs since early 2010"? That's an impressive number, amirite?
  But are those jobs real? Or are they conjured up out of a mathematical model? Allow me to introduce you to the BLS's Birth-Death Model. ...  Based on the raw BLS numbers, the Birth-Death Model accounts for 7,487,000 of those 15 million jobs.
Ummm, no.

Here are the facts.  The BLS isn't stupid.  They know that their birth-death estimates can throw off the monthly job numbers.  So they turn to a source that can give them a much more accurate count.  That source is the number of businesses who file with the states to join the unemployment insurance system.  Every year in March the BLS corrects its monthly reports for the prior year based on the much more complete record of how many employers went into or out of business, as revealed in those fillings with the states. 

For example, here is the announcement just last month of the preliminary revisions for 2016:

In case it isn't clear already, here are the spot on sentences in that announcement:

"Each year the Current Employment Statistics (CES) survey employment estimates are benchmarked to comprehensive counts of employment for the month of March.  These counts are derived from state unemployment insurance (UI) tax records that nearly all employers are required to file."


So, while the birth-death model is undoubtedly having an effect on 2016 reports, it is no longer reflected in any of the job numbers for 2015 and before, which have all been revised based on this comprehensive benchmark.  The diarist is spouting nonsense.

These same comprehensive state records are the basis for debunking the first claim as well.

Here's the graph that the diarist in question uses to make the claim that the 40 year low in initial jobless claims are bogus:

Here's the problem:  that graph doesn't show what percent of INITIAL unemployment is covered by unemployment insurance vs. not, but rather what percent of ALL unemployment, including continuing claims (of, e.g., 10 or 25 or 40 weeks) is covered.  Since many states cut back the number of weeks of unemployment insurance they will pay, more of the unemployment at the longer durations such as 26 weeks is not covered by insurance.

To find out if that affects initial claims, all we have to do is go back to that same comprehensive record of employers who pay into the unemployment insurance system.  Once a quarter, the DOL surveys how many employees are covered in the unemployment insurance system.  That statistic is called "covered employment."  Here it is (in blue below) compared with total employment (red):

Do you see any big drop-off in that number?  Of course not.  But just to give you the accurate percentage, in the below graph I have divided covered employment by all employment.  The result is the percentage of all jobs that are covered by unemployment insurance:

Notice that from 1980 forward, during expansions that percentage has been pretty consistent at about 94% to 95%.

So, finally, let's take the number of initial jobless claims -- which have been making new 40 year lows repeatedly this year -- and divided them by the percentage of covered employment.  The lower the % of covered employment, the higher the "normed" number of initial claims will be.  And here is the result:

Norming for the percent of covered employment dies not change the fact that initial claims are running at 40 year lows.  Again, the diarist is spouting nonsense.

One more thing: he drags out the hoary chestnut that
"   Another way to keep unemployment numbers low is to simply stop counting people," citing to 
\the decline in the Labor Force Participation Rate.   

 As I have constantly pointed out, in that very same survey the Census Bureau asks the people who are not in the Labor Force whether or not they want a job, and every month about 90% of those people say thaey don't.  The Census Bureau helpfully reports, each and every month, the number of people who aren't in the labor force who say they "want a job now," and here it is:

It is about 1.5 million higher than at its previous lows, but on the other hand it is about equal to the number in 1995, not a notably awful year.

Now, every now and then, just a like a blind squirrel, a Doomer will find a useful "nut" of information.  But not this time.

And by the way, probably the best way to determine if the *number of jobs added* in an expansion has been weak or not, is to compare what percent of jobs compared with the working age population has been each month.  Here's what that comparison looks like for prime working age 25-54: 

And here's what it looks like for age 15-64 (although I don't really think high school and college aged kids should be added in):

In terms of the number of jobs added in any given month, this expansion has been weak, but at least as good if not better than the one that preceded it.