Oops! Jobs # Revised Down

Investment Analysis Clubs Macro Economic Trends and Risks
1

The Bureau of Labor Statistics’ preliminary annual benchmark review of employment data suggests that there were 818,000 fewer jobs in March of this year than were initially reported.

The downward adjustments were limited to the private sector, with nearly half in the professional and business services industry (revised down by 358,000, or 1.6%.) Other sectors showing large negative swings included the information industry (down 68,000, or 2.3%), leisure and hospitality (down 150,000, or 0.9%) and manufacturing (down 115,000, or -0.9%).

5 Likes
2

Okay, here’s an strictly non-partisan, but equal opportunity political rant.

How can government be so stupid to rely upon such archiac practices for collecting information of such vital importance to government and business alike?

Employment rates are perhaps THE most closely watched canary in the economic coal mine. Perceptions of where employment trends are headed drive BILLION DOLLAR decisions every day in the Federal Reserve, in the government and through every industry. Yet note the text of the story from August 21 of 2024…

The Bureau of Labor Statistics’ preliminary annual benchmark review of employment data suggests that there were 818,000 fewer jobs in March of this year than were initially reported.

Translated, the BLS is STILL preparing a PRELIMINARY estimate of a statistic involving employment activity from FIVE MONTHS AGO.

Does that make any sense?

Unless American employment trends have shifted sharply towards busking in a subway with an open guitar case, the VAST majority of “employees” are paid via electronic processors. I seriously doubt ANY employer is mailing a paper check to the IRS each weekly / bi-weekly /monthly payroll for the amount of payroll taxes due. All of those payments are taking place electronically and likely settle within 24 hours.

Think about all of the data those payroll tax payments must reflect:

  • SSN or Individual Taxpayer ID Number (ITIN) of the worker
  • Employer ID Number (EIN) of the firm paying the income
  • an indication whether the pay involves hourly rates or salaried work
  • an indication of hours worked (so state / federal governments can watch for OT violations)

From all that data, it should be obvious how to devise a more accurate and nearly instantaneous analysis of “job” changes in the market.

  • every unique combination of SSN/ITIN and employer ID is a POTENTIAL “job”
  • if a combination (SSN/ITIN + EIN) appearing in data for prior week / month STOPS appearing, that COULD be a job elimination
  • if a NEW combination (SSN/ITIN + EIN) not previously seen in prior terms begins appearing, that COULD be a job addition
  • if the SAME SSN/ITIN appears in a “loss” and a “gain” bucket, that could be due to:
    a) that worker QUITTING job A and starting job B
    b) that worker changing from contractor A to employer B for same work
    c) that worker changing from employer A to contractor B for same work
    d) that worker staying in the same job which has been shifted between subsidiaries that triggers a change in employer ID

There should be enough other metadata available in what employers have to submit with payroll tax payments to help filter out scenario (d).

By adjusting time-series criteria to set minimums for consecutive periods where (SSN + EID) must appear to count as an add (so one week of pay that never recurs doesn’t count as an “add”) or minimum consecutive periods where an SSN fails to appear entirely before counting as an eliminated job, a relatively simple report can be generated with virtually no human intervention or tweaking that could trigger accusations of political manipulation.

Since the incoming payroll tax data reflects absolute dollars and hours, summary statistics could extrapolate trends regarding overtime, pay raises (wages and salaries…), etc. It would provide instant insight into labor patterns such as people working multiple part-time jobs. It would provide nearly instant insight into income losses stemming from hurricanes, floods, port shutdowns due to bridge collapses, you name it. Little need to GUESS impacts, the data should be immediately available.

If only there was a base… in which we could put data… allowing information to be extracted with a human-friendly language… that allowed queries to be executed based on the structure of the data… I bet some standard scheme or protocol could be devised to control the transmission of data between a collection of computers, all interconnected together as a network or internet…

Gosh, if only that technology existed, this would be a piece of cake.

(sigh…)

WTH

9 Likes
3

Now you’re getting into my area of professional expertise.

You are correct in that payroll tax payments are made very regularly by employers. Really big employers need to pay the payroll taxes the day after payroll is paid. Most other employers get a couple of days to pay the taxes. The very smallest can pay them monthly.

But there is no employee-level data in those payments. None at all. Not to the IRS. All the IRS sees is the amount of the payment, the employer’s ID number, and the calendar quarter in which the payroll was paid. That’s it.

At the state level, individual states might collect more information. I’m only intimately familiar with my home state, California. Here, the payroll tax payments are due on basically the same schedule as the federal tax payments. And, like the federal payments, contain no employee-level data.

Every quarter, California gets employee level data, but only name, SSN, and wages paid and state taxes withheld during the quarter. No hourly vs. salary. No hours worked. No overtime information.

California also has a new hire reporting system. Within a few days of hiring, the name and SSN of the new employee must be reported to the state. But that’s it. (This reporting is used to help the state with tax and child support collections - so you can’t avoid wage levies by moving employers. The state will catch up to you with a fresh levy within one or maybe two paychecks at most.) This information might help with new hire statistics. But there is no counterpart reporting for those leaving jobs.

The data your are assuming simply doesn’t exist where it is accessible to nationwide statisticians. Hence the need for estimates.

–Peter

7 Likes
4

But there is no employee-level data in those payments.

Well, that explains it. We are idiots. I can think of far more intrusive information the government could demand than seeing my income / hours per paycheck. They’re GOING to know the total income amount by the end of quarter anyway. The VALUE of this data to understanding the economy is VAST and all of the information is right there. It would likely take little effort to alter the report format sent to the IRS and run the jobs weekly rather than quarterly.

Frustrating.

WTH

5 Likes
5

Imagine if we also had a verification system to ensure that all those SSNs and ITINs were of legal status (citizens or otherwise)?
What a concept!

But who would want that?

3 Likes
6

Little effort? I think it would take a major IT project study and a big program to hire and train a bunch of COBOL programmers to modify all the code to handle this. There would then be a big program to build out new data centers to hold a bazillion bytes of the data, archive it forever and keep backups…imagine all the floppy disks this will take!

Mike

3 Likes
7

Some states would SCREAM “STATES’ RIGHTS !!”–even though all the states have ZERO “rights” with regard to federal policy/law–as specifically stated in the US Constitution.

1 Like
8

Rec’d and laughed (sometimes bitterly and sometimes merrily) every post in this thread.

USA employment regulations and laws are obscenely structured to protect the exploitation of illegal immigrant workers, as has been customary ever since WWII. The knock on effects statistically legally and politically grow worse each year.

d fb

4 Likes