Jobs Numbers – Dramatic Differences

 

Today’s Household Survey paints a catastrophic picture of jobs in America.   Full-time employment down 600,000 in May.   Total employment down 400,000.   Part-time employment up 200,000.   That means we have LOST almost 1.5 million full-time jobs since November 2023.   The unemployment rate — calculated from the Household numbers — is up to 4%.   The only reason it hasn’t skyrocketed is that the labor force participation rate is down.   And, the way the math works, a smaller labor force yields better-looking unemployment numbers.

However, the Establishment Survey shows a completely different version of events.   It shows 270,000 new jobs created.  This continues a 50-year trend of the Establishment Survey (the headline jobs number) exceeding the Household Survey ONLY when there is a Democrat President.   See my post:

https://ricochet.com/1550976/establishment-survey-jobs-report-50-years-of-pro-democrat-bias-revealed/

But the Household Survey squares with how it feels out there.   I’d say it’s much closer to reality.

Published in General
This post was promoted to the Main Feed at the recommendation of Ricochet members. Like this post? Want to comment? Join Ricochet’s community of conservatives and be part of the conversation. Join Ricochet for Free.

There are 37 comments.

Become a member to join the conversation. Or sign in if you're already a member.
  1. Ekosj Member
    Ekosj
    @Ekosj

    In the prepandemic period 2017, 2018, 2019, the employment level  increased about 2.2 Million per year.  In the post pandemic last 12 months the US employment level increased 376,000.   That’s substantially less than half the rate of increase of the 2017-2019 period.   Last month, the employment level went DOWN 408,000.

    https://fred.stlouisfed.org/series/CE16OV

    Worse, the jobs created are mostly part time jobs.   The Household Survey shows the whole truth.   The Household Survey is the data set used to calculate the unemployment rate.   Its bona fides are impeccable.   It breaks down employment into Full Time and Part Tome Jobs.   Since June 2023 we have lost almost 1.5 MILLION Full Time jobs.

    https://fred.stlouisfed.org/series/LNS12500000

    Worst…The BLS Household Survey shows that total employment dropped from 161,866,000 in Nov 2023 to 161,083,000  (edited for typo)  in the most recent report.   That’s a drop of 783,000 jobs.    The only reason the unemployment rate hasn’t skyrocketed is that the Labor Force Participation Rate has declined as well.  And the way the unemployment rate math works, a smaller labor force makes the unemployment rate smaller.

    • #1
  2. colleenb Member
    colleenb
    @colleenb

    Listening to Bloomberg News yesterday they mentioned that all the jobs reports in the last 12 months have been revised downward after the initial numbers came out. Gosh I’m sure the fact that a Democrat is President has nothing to do with that – and, of course, the ‘commentators’ didn’t say anything about that curious phenomenon! And the Powers That Be are so surprised when people don’t trust the elite – or their statistics and facts.

    • #2
  3. Drew in Texas Coolidge
    Drew in Texas
    @Dbroussa

    Ekosj (View Comment):

    In the prepandemic period 2017, 2018, 2019, the employment level increased about 2.2 Million per year. In the post pandemic last 12 months the US employment level increased 376,000. That’s substantially less than half the rate of increase of the 2017-2019 period. Last month, the employment level went DOWN 408,000.

    https://fred.stlouisfed.org/series/CE16OV

    Worse, the jobs created are mostly part time jobs. The Household Survey shows the whole truth. The Household Survey is the data set used to calculate the unemployment rate. Its bona fides are impeccable. It breaks down employment into Full Time and Part Tome Jobs. Since June 2023 we have lost almost 1.5 MILLION Full Time jobs.

    https://fred.stlouisfed.org/series/LNS12500000

    Worst…The BLS Household Survey shows that total employment dropped from 161,866,000 in Nov 2023 to 161,491,000 in the most recent report. That’s a drop of 783,000 jobs. The only reason the unemployment rate hasn’t skyrocketed is that the Labor Force Participation Rate has declined as well. And the way the unemployment rate math works, a smaller labor force makes the unemployment rate smaller.

    How is that a drop of 783K jobs?  I get a 375K decrease.

    • #3
  4. Drew in Texas Coolidge
    Drew in Texas
    @Dbroussa

    I see this in the IT consulting world.  When times are tough financially, projects are delayed or canceled in part to conserve cash, and because getting rid of contractors/consultants is easier than laying off people.  Hiring is very tough right now.  I have been looking for work for almost a year now, and a friend of mine is in the same boat, though she has been completely out of work for that time.  She recently got a “part-time” consulting gig which will help, but it makes it very tough.  Couple that with the explosion of AI in the IT world and people want AI “qualified” people for products that no one has access to.  It reminds me a lot of when I first graduated college in 90 and was looking for work.  People wanted workers with 5 years of experience on the AS/400 (it was released in late 1988).  Couple that with increasing age discrimination in the marketplace and finding a job as a 55-year-old consultant is getting more and more difficult. My son (an 18-year-old recent High School graduate) is also facing difficulty in getting a job this summer.  The pay is nice at places like the local movie theater, or the local grocery store where $15/hr is the norm, but it also means fewer jobs are available.  There are zero entry-level bagger openings for example, and the movie theater is only looking for people who have food service experience because their entry-level jobs are all filled, and have been for a while.  He is supposed to go to college in late August and will have to pay about $6K for his room and board (he has a scholarship for tuition).  He can borrow about $2,750 via a student loan, but the rest is out of pocket.  If he got a job for the summer he could likely make $3k which would pay for one semester of college, what happens after that I don’t know.

    • #4
  5. Ekosj Member
    Ekosj
    @Ekosj

    Drew in Texas (View Comment):

    Ekosj (View Comment):

    In the prepandemic period 2017, 2018, 2019, the employment level increased about 2.2 Million per year. In the post pandemic last 12 months the US employment level increased 376,000. That’s substantially less than half the rate of increase of the 2017-2019 period. Last month, the employment level went DOWN 408,000.

    https://fred.stlouisfed.org/series/CE16OV

    Worse, the jobs created are mostly part time jobs. The Household Survey shows the whole truth. The Household Survey is the data set used to calculate the unemployment rate. Its bona fides are impeccable. It breaks down employment into Full Time and Part Tome Jobs. Since June 2023 we have lost almost 1.5 MILLION Full Time jobs.

    https://fred.stlouisfed.org/series/LNS12500000

    Worst…The BLS Household Survey shows that total employment dropped from 161,866,000 in Nov 2023 to 161,491,000 in the most recent report. That’s a drop of 783,000 jobs. The only reason the unemployment rate hasn’t skyrocketed is that the Labor Force Participation Rate has declined as well. And the way the unemployment rate math works, a smaller labor force makes the unemployment rate smaller.

    How is that a drop of 783K jobs? I get a 375K decrease.

    Ahhhh    Operator error.    The current number should be 161083 not 161491.   I’ll fix that.   Thanks for catching that

    • #5
  6. Muleskinner, Weasel Wrangler Member
    Muleskinner, Weasel Wrangler
    @Muleskinner

    Ekosj:

    However the Establishment Survey shows a completely different version of events. The Establishment Survey shows 270,000 new jobs created. This continues a 50 year trend of the Establishment Survey (the headline jobs number) exceeding the Household Survey ONLY when there is a Democrat President.

    I haven’t considered any impact of the party of the administration, but there has been some discussion about whether the a divergence of the surveys is a recession indicator. But you have to look at the pre-revised data to get the differences, the thinking is that the Household Survey catches Dow cycle changes in the economy before the Payroll Survey. 

    • #6
  7. Ekosj Member
    Ekosj
    @Ekosj

    Muleskinner, Weasel Wrangler (View Comment):

    Ekosj:

    However the Establishment Survey shows a completely different version of events. The Establishment Survey shows 270,000 new jobs created. This continues a 50 year trend of the Establishment Survey (the headline jobs number) exceeding the Household Survey ONLY when there is a Democrat President.

    I haven’t considered any impact of the party of the administration, but there has been some discussion about whether the a divergence of the surveys is a recession indicator. But you have to look at the pre-revised data to get the differences, the thinking is that the Household Survey catches Dow cycle changes in the economy before the Payroll Survey.

    You should check out my post on the numbers.   Link is in the OP.   You might find it interesting.

    • #7
  8. Steve Fast Member
    Steve Fast
    @SteveFast

    I’ve found  your  posts fascinating, and I did a deep dive last time you posted on this to try to understand why there is a difference between the two surveys. It seems that economists argue all the time about which survey is more accurate at a given moment and why there are significant differences. I gave up trying to understand why because most of them seemed unable to understand why.

    The short answer seems to be that there are major shortcomings in each dataset, so it is not surprising that there are wide differences in them.

    A few of the discussions:

    https://www.bls.gov/web/empsit/ces_cps_trends.htm

    https://www.businessinsider.com/the-differences-between-the-household-and-establishment-employment-surveys-2014-9?op=1

    https://cepr.net/job-growth-is-the-household-or-establishment-survey-right/

    • #8
  9. Mark Camp Member
    Mark Camp
    @MarkCamp

    Steve Fast (View Comment):

    I’ve found your posts fascinating, and I did a deep dive last time you posted on this to try to understand why there is a difference between the two surveys.

    There is no difference in the results of the two surveys. In fact, it is logically impossible for there to be: they measure two different things.

    If you are interested in knowing what those two different things are, please let me know and I will explain tomorrow, and give you a reference so that you can verify the facts for yourself. (It’s too late in the day now.)

    • #9
  10. Steve Fast Member
    Steve Fast
    @SteveFast

    Mark Camp (View Comment):

    Steve Fast (View Comment):

    I’ve found your posts fascinating, and I did a deep dive last time you posted on this to try to understand why there is a difference between the two surveys.

    There is no difference in the results of the two surveys. In fact, it is logically impossible for there to be: they measure two different things.

    If you are interested in knowing what those two different things are, please let me know and I will explain tomorrow, and give you a reference so that you can verify the facts for yourself. (It’s too late in the day now.)

    There’s a lot of discussion of it online, which I have read; so you don’t need to explain it tomorrow.

    • #10
  11. Mark Camp Member
    Mark Camp
    @MarkCamp

    Steve Fast (View Comment):

    Mark Camp (View Comment):

    Steve Fast (View Comment):

    I’ve found your posts fascinating, and I did a deep dive last time you posted on this to try to understand why there is a difference between the two surveys.

    There is no difference in the results of the two surveys. In fact, it is logically impossible for there to be: they measure two different things.

    If you are interested in knowing what those two different things are, please let me know and I will explain tomorrow, and give you a reference so that you can verify the facts for yourself. (It’s too late in the day now.)

    There’s a lot of discussion of it online, which I have read; so you don’t need to explain it tomorrow.

    I was responding to your statement from earlier this evening that you gave up trying to understand.

    But it sounds as though you may have found out since then.

    Anyway, maybe we should confirm tomorrow that we are both talking about the same two metrics. Also, I would like you and other curious readers, if any, to have the link anyway, so that next time it comes up, you can find the authoritative answer and not be dependent on online discussions. (If you are like me, it’s hard to remember the details of stuff like this, or where to find the source documents!)

    • #11
  12. CACrabtree Coolidge
    CACrabtree
    @CACrabtree

    Drew in Texas (View Comment):

    I see this in the IT consulting world. When times are tough financially, projects are delayed or canceled in part to conserve cash, and because getting rid of contractors/consultants is easier than laying off people. Hiring is very tough right now. I have been looking for work for almost a year now, and a friend of mine is in the same boat, though she has been completely out of work for that time. She recently got a “part-time” consulting gig which will help, but it makes it very tough. Couple that with the explosion of AI in the IT world and people want AI “qualified” people for products that no one has access to. It reminds me a lot of when I first graduated college in 90 and was looking for work. People wanted workers with 5 years of experience on the AS/400 (it was released in late 1988). Couple that with increasing age discrimination in the marketplace and finding a job as a 55-year-old consultant is getting more and more difficult. My son (an 18-year-old recent High School graduate) is also facing difficulty in getting a job this summer. The pay is nice at places like the local movie theater, or the local grocery store where $15/hr is the norm, but it also means fewer jobs are available. There are zero entry-level bagger openings for example, and the movie theater is only looking for people who have food service experience because their entry-level jobs are all filled, and have been for a while. He is supposed to go to college in late August and will have to pay about $6K for his room and board (he has a scholarship for tuition). He can borrow about $2,750 via a student loan, but the rest is out of pocket. If he got a job for the summer he could likely make $3k which would pay for one semester of college, what happens after that I don’t know.

    Yeah, I remember those AS/400 ads.  I always figured they were written by HR weenies who didn’t know crap about IT.

    I retired in 2004 and up to that point virtually all our consultants were Indians on H1B visas.  Today I tell kids who are looking at IT careers to concentrate on network security or Data Base Administration.  

    Otherwise, I tell them to bone up on math and head down to the union hall.  The trades are paying really well.

    • #12
  13. Steve Fast Member
    Steve Fast
    @SteveFast

    Mark Camp (View Comment):

    Steve Fast (View Comment):

    Mark Camp (View Comment):

    Steve Fast (View Comment):

    I’ve found your posts fascinating, and I did a deep dive last time you posted on this to try to understand why there is a difference between the two surveys.

    There is no difference in the results of the two surveys. In fact, it is logically impossible for there to be: they measure two different things.

    If you are interested in knowing what those two different things are, please let me know and I will explain tomorrow, and give you a reference so that you can verify the facts for yourself. (It’s too late in the day now.)

    There’s a lot of discussion of it online, which I have read; so you don’t need to explain it tomorrow.

    I was responding to your statement from earlier this evening that you gave up trying to understand.

    But it sounds as though you may have found out since then.

    Anyway, maybe we should confirm tomorrow that we are both talking about the same two metrics. Also, I would like you and other curious readers, if any, to have the link anyway, so that next time it comes up, you can find the authoritative answer and not be dependent on online discussions. (If you are like me, it’s hard to remember the details of stuff like this, or where to find the source documents!)

    Economists themselves argue in many articles about the differences (see the links I posted), but maybe you will be able to summarize it succinctly in one comment.

    • #13
  14. Mark Camp Member
    Mark Camp
    @MarkCamp

    Steve Fast (View Comment):

    Mark Camp (View Comment):

    Steve Fast (View Comment):

    Mark Camp (View Comment):

    Steve Fast (View Comment):

    I’ve found your posts fascinating, and I did a deep dive last time you posted on this to try to understand why there is a difference between the two surveys.

    There is no difference in the results of the two surveys. In fact, it is logically impossible for there to be: they measure two different things.

    If you are interested in knowing what those two different things are, please let me know and I will explain tomorrow, and give you a reference so that you can verify the facts for yourself. (It’s too late in the day now.)

    There’s a lot of discussion of it online, which I have read; so you don’t need to explain it tomorrow.

    I was responding to your statement from earlier this evening that you gave up trying to understand.

    But it sounds as though you may have found out since then.

    Anyway, maybe we should confirm tomorrow that we are both talking about the same two metrics. Also, I would like you and other curious readers, if any, to have the link anyway, so that next time it comes up, you can find the authoritative answer and not be dependent on online discussions. (If you are like me, it’s hard to remember the details of stuff like this, or where to find the source documents!)

    Economists themselves argue in many articles about the differences (see the links I posted), but maybe you will be able to summarize it succinctly in one comment.

    We are talking about two completely different things.  I’m answering this question that you implicitly raised and answered, which no economists would ever argue about:

    ”Do the Bureau of Labor Statistics household and employer surveys attempt to measure the same thing?”

    (No economists would have reason to argue about it, since any economist, or you or I, or anyone else can look up the answer on the web.  It is at

    https://www.bls.gov/bls/empsitquickguide.htm )

    • #14
  15. Steve Fast Member
    Steve Fast
    @SteveFast

    Mark Camp (View Comment):

    Steve Fast (View Comment):

    Mark Camp (View Comment):

    Steve Fast (View Comment):

    Mark Camp (View Comment):

    Steve Fast (View Comment):

    I’ve found your posts fascinating, and I did a deep dive last time you posted on this to try to understand why there is a difference between the two surveys.

    There is no difference in the results of the two surveys. In fact, it is logically impossible for there to be: they measure two different things.

    If you are interested in knowing what those two different things are, please let me know and I will explain tomorrow, and give you a reference so that you can verify the facts for yourself. (It’s too late in the day now.)

    There’s a lot of discussion of it online, which I have read; so you don’t need to explain it tomorrow.

    I was responding to your statement from earlier this evening that you gave up trying to understand.

    But it sounds as though you may have found out since then.

    Anyway, maybe we should confirm tomorrow that we are both talking about the same two metrics. Also, I would like you and other curious readers, if any, to have the link anyway, so that next time it comes up, you can find the authoritative answer and not be dependent on online discussions. (If you are like me, it’s hard to remember the details of stuff like this, or where to find the source documents!)

    Economists themselves argue in many articles about the differences (see the links I posted), but maybe you will be able to summarize it succinctly in one comment.

    We are talking about two completely different things. I’m answering this question that you implicitly raised and answered, which no economists would ever argue about:

    ”Do the Bureau of Labor Statistics household and employer surveys attempt to measure the same thing?”

    (No economists would have reason to argue about it, since any economist, or you or I, or anyone else can look up the answer on the web. It is at

    https://www.bls.gov/bls/empsitquickguide.htm )

    There’s a lot more to it than the explainer that the BLS put on its web site. Economists debate why the two measures don’t move in tandem, what it means when the two measures diverge, when is one more reliable than the other, etc., etc.

    Then Ekosj adds an interesting question about why the two measures seem to move differently based on who is in the White House.

    • #15
  16. The Reticulator Member
    The Reticulator
    @TheReticulator

    Steve Fast (View Comment):

    There’s a lot more to it than the explainer that the BLS put on its web site. Economists debate why the two measures don’t move in tandem, what it means when the two measures diverge, when is one more reliable than the other, etc., etc.

    I think Mark would say that comparing the reliability of the two doesn’t make any sense, because they don’t necessarily have the same purpose, though I would argue that the two purposes do have some relation to each other.  

    But it’s like asking whether taking my bicycle to the Bedford post office is more reliable than taking our Subaru.  It doesn’t make any sense without more information.   The bicycle is more reliable at giving me some cardio exercise, while the Subaru is more reliable at getting it done quickly and getting me back into my Lazy-Boy.  Reliable is not a standalone word.  One needs to specify, “reliable for what purpose.”  (I have the same beef with words like responsible and accountability.  They are not standalone words.  One needs to know “responsible to whom,” and “accountable to whom.”)  

    Then Ekosj adds an interesting question about why the two measures seem to move differently based on who is in the White House.

     

    • #16
  17. Steve Fast Member
    Steve Fast
    @SteveFast

    The Reticulator (View Comment):

    Steve Fast (View Comment):

    There’s a lot more to it than the explainer that the BLS put on its web site. Economists debate why the two measures don’t move in tandem, what it means when the two measures diverge, when is one more reliable than the other, etc., etc.

    I think Mark would say that comparing the reliability of the two doesn’t make any sense, because they don’t necessarily have the same purpose, though I would argue that the two purposes do have some relation to each other.

    But it’s like asking whether taking my bicycle to the Bedford post office is more reliable than taking our Subaru. It doesn’t make any sense without more information. The bicycle is more reliable at giving me some cardio exercise, while the Subaru is more reliable at getting it done quickly and getting me back into my Lazy-Boy. Reliable is not a standalone word. One needs to specify, “reliable for what purpose.” (I have the same beef with words like responsible and accountability. They are not standalone words. One needs to know “responsible to whom,” and “accountable to whom.”)

    Then Ekosj adds an interesting question about why the two measures seem to move differently based on who is in the White House.

    If you’re asking the question, “How many jobs were added to the US economy during period P?” each one has something to say about that. So they are definitely related. But each one defines “job” differently, and each one samples a different population to calculate an estimate of that number, not to mention a number of other differences in the sampling method. Each provides incomplete information about the number of jobs added.

    Under certain economic conditions, one may be more accurate than the other. Differences in the two can tell something about the types of jobs being added (e.g. part-time vs. full-time) and the sectors experiencing growth or retraction. And that can tell something about the direction of the economy.

    Assuming that they are calculated in an unbiased manner (which Ekosj’s analysis calls into question), it is useful to have two distinct measures of the same variable.

     

    • #17
  18. Mark Camp Member
    Mark Camp
    @MarkCamp

    Steve Fast (View Comment):

    There’s a lot more to it than the explainer that the BLS put on its web site. Economists debate why the two measures don’t move in tandem, what it means when the two measures diverge, when is one more reliable than the other, etc., etc.

    Then Ekosj adds an interesting question about why the two measures seem to move differently based on who is in the White House.

    If the number of jobs and unemployment both go up because more people are working multiple jobs and more people are looking for a job, it is a question for economists to debate why those two things are happening.

    They won’t be debating about the what the jobs survey (establishment survey or “CES survey”—see the document I referenced) and the unemployment survey (“household survey” aka “CPS survey”) are trying to measure: jobs, and unemployment, respectively.)

    Here’s an analogy that may help you to understand the problem with Ekosj’s question.

    Suppose an inspector in a bread factory  accurately measures the mass and volume of the loaves of bread coming off the line using tests M and V, respectively.

    One week, the mass starts going up the volume starts going down. Customers are complaining.

    The engineers have a meeting to diagnose the problem.

    Do they debate your question: “Are test M and test V measuring the same thing?”

    No. As engineers, they know the difference between mass and volume, and they know that test M measures mass and test V measures volume.

    The question they DO need to find the answer to, is WHY the two variables are changing in the direction that they are.

    In the analogy, tests M and V are the establishment and household surveys. The engineers are the economists. The nonsense-question is “Why do the two tests disagree?” It is nonsense because it has an implicit assumption  that is false: “The tests measure the same variable”. The engineers (the economists) don’t argue about what the two tests measure, because they already know what the tests measure, by reading the manufacturing documentation (the BLS website I linked to).

    • #18
  19. Front Seat Cat Member
    Front Seat Cat
    @FrontSeatCat

    Just talk to anyone in any field during the course of a week. I recently had an emergency room visit due to a very bad stomach bug. This was a Wed. morning and the place was almost full. I heard a doctor say we may need to go to triage.  I was wheeled out of my space for a quick 5 minute scan of my abdomen and gallbladder. My space was occupied when I was wheeled back so I had to be left in a hallway space set up for the overflow! I heard medical staff say “no one wants to work and we are short-staffed” – this was echoed among several individuals each trying to do the job of 3.

    They never rushed their jobs or got huffy. They were patient and kind to all – real pros.  That was one place. I saw and heard the same sentiments at stores and am seeing a lot of older retirees filling jobs once held by youth.  Usually one or two registers open only, whether retail or post office and long lines. Seeing lots more check yourself out set-ups.

    Everything changed since Covid and there is no excuse that it has not resolved itself.  But you can clearly see the trends of the WEF’s goals coming about.  It is in bold print in Klaus Schwab’s book from 2016 ‘The Fourth Industrial Revolution’ where we are today for job elimination and many other major changes in society – and it is scary.

    • #19
  20. Steve Fast Member
    Steve Fast
    @SteveFast

    Mark Camp (View Comment):
    Here’s an analogy that may help you to understand the problem with Ekosj’s question.

    You’re avoiding the actual question he raised.

    It’s not a nonsense question to ask why the two measures differ. They are measuring very similar things, and a lot of economists discuss why they differ.

    • #20
  21. kedavis Coolidge
    kedavis
    @kedavis

    Front Seat Cat (View Comment):
    Usually one or two registers open only, whether retail or post office and long lines. Seeing lots more check yourself out set-ups.

    Last I heard, many places including Walmart are scaling back and even eliminating self-checkout.

    • #21
  22. Ekosj Member
    Ekosj
    @Ekosj

    The simple answer about the differences between the reports is that …

    The household survey employment measure includes categories of workers that are not covered by the payroll survey:

    • the self-employed
    • workers in private households
    • agricultural workers
    • unpaid workers in family businesses
    • workers on leave without pay during the reference period

    Because the household survey count more kinds of workers/jobs one would expect – a priori – that the household numbers would usually be higher.

    But  the payroll survey double counts multiple jobholders.  As it happens, multiple jobholders are at an all time high.  So maybe that’s it.   Maybe it’s multiple jobholders.    

    But you have two things pulling in opposite directions.   You’d think they’d more or less net out.    But they don’t.    That’s what I discovered by looking at 50 years of data side by side.   And it’s especially obvious recently.   I computed a ten-year average difference between the two surveys and compared the recent differences to that average.   (This is the Mann temperature technique)   And sure enough … you get a hockey stick.    Today’s differences are off the charts.

     

    • #22
  23. Ekosj Member
    Ekosj
    @Ekosj

    duplicate deleted

    • #23
  24. Ekosj Member
    Ekosj
    @Ekosj

    Might have found an answer…

    The fact that the “blowout” establishment survey is—as Bloomberg puts it—”Out of Sync With Recent Weaker Economic Data”—may be partly due to the establishment survey’s reliance on the so-called birth-death model. This model is used to estimate how many new jobs were created by new businesses—i..e, “births”—that are missed by the actual survey results. The BLS says it must use “non-sampling methods” to add in these newly created jobs. “Non-sampling methods” means the numbers made up by number crunchers. They don’t show up in any survey. In May, the establishment survey assumed the creation of 231,000 jobs. That’s a sizable number in a report that tells us 271,000 new jobs were created. Wong concludes this very large addition of hypothetical jobs to the establishment total simply doesn’t reflect the real world right now. She writes: 

    …BLS’ model for estimating business births and deaths – which added 231,000 jobs to the nonfarm-payrolls print in May – is lagging the reality of surging establishment closures and falling business formation. We think the underlying pace of current job gains is likely less than 100,000 per month.

    • #24
  25. The Reticulator Member
    The Reticulator
    @TheReticulator

    Ekosj (View Comment):

    Might have found an answer…

    The fact that the “blowout” establishment survey is—as Bloomberg puts it—”Out of Sync With Recent Weaker Economic Data”—may be partly due to the establishment survey’s reliance on the so-called birth-death model. This model is used to estimate how many new jobs were created by new businesses—i..e, “births”—that are missed by the actual survey results. The BLS says it must use “non-sampling methods” to add in these newly created jobs. “Non-sampling methods” means the numbers made up by number crunchers. They don’t show up in any survey. In May, the establishment survey assumed the creation of 231,000 jobs. That’s a sizable number in a report that tells us 271,000 new jobs were created. Wong concludes this very large addition of hypothetical jobs to the establishment total simply doesn’t reflect the real world right now. She writes:

    …BLS’ model for estimating business births and deaths – which added 231,000 jobs to the nonfarm-payrolls print in May – is lagging the reality of surging establishment closures and falling business formation. We think the underlying pace of current job gains is likely less than 100,000 per month.

    When your data don’t match reality,  one of the two needs to adjust. 

    • #25
  26. Muleskinner, Weasel Wrangler Member
    Muleskinner, Weasel Wrangler
    @Muleskinner

    The Reticulator (View Comment):

    Ekosj (View Comment):

    Might have found an answer…

    The fact that the “blowout” establishment survey is—as Bloomberg puts it—”Out of Sync With Recent Weaker Economic Data”—may be partly due to the establishment survey’s reliance on the so-called birth-death model. This model is used to estimate how many new jobs were created by new businesses—i..e, “births”—that are missed by the actual survey results. The BLS says it must use “non-sampling methods” to add in these newly created jobs. “Non-sampling methods” means the numbers made up by number crunchers. They don’t show up in any survey. In May, the establishment survey assumed the creation of 231,000 jobs. That’s a sizable number in a report that tells us 271,000 new jobs were created. Wong concludes this very large addition of hypothetical jobs to the establishment total simply doesn’t reflect the real world right now. She writes:

    …BLS’ model for estimating business births and deaths – which added 231,000 jobs to the nonfarm-payrolls print in May – is lagging the reality of surging establishment closures and falling business formation. We think the underlying pace of current job gains is likely less than 100,000 per month.

    When your data don’t match reality, one of the two needs to adjust.

    Which is why you have to look at each release to determine if there is information to be gleaned from divergence of the two survey results. The BLS is always tinkering with the Birth-Death model, and once they know the trend in business creation, the model will reflect that trend. Forecasting is easy, it’s identifying the turning points that is hard, if even possible. 

    • #26
  27. Mark Camp Member
    Mark Camp
    @MarkCamp

    Steve Fast (View Comment):

    Mark Camp (View Comment):
    Here’s an analogy that may help you to understand the problem with Ekosj’s question.

    You’re avoiding the actual question he raised.

    It’s not a nonsense question to ask why the two measures differ. They are measuring very similar things, and a lot of economists discuss why they differ.

    It appears that you and I have been trying from the beginning to have a rational debate using mutually exclusive systems of logic, which is impossible. I appreciate your contribution to the effort, but I think we should abandon it.

    • #27
  28. Steve Fast Member
    Steve Fast
    @SteveFast

    @ekosj, Issues and Insights published an editorial recently that has some overlap with your post. They are debunking the claim that Biden created 15.6 million jobs and show that instead he created 2.4 million, which is below the rate of population growth. (I contend that government cannot do anything but destroy existing jobs or prevent the creation of future jobs, but I digress.)

    They chalk part of the difference up to the fact that people are holding multiple jobs due to Bidenflation. Since the Establishment survey counts jobs and the Household survey counts people with a job, the Establishment survey overstates job growth. None of this is earth-shattering analysis, but it’s nice to see someone thinking along the same lines as you.

    And they include this nice graph:

    • #28
  29. Mark Camp Member
    Mark Camp
    @MarkCamp

    Steve Fast (View Comment):

     

     

    Since the Establishment survey counts jobs…the Establishment survey overstates job growth.

    I suggest that you edit this sentence so that it says in plain English what you mean.  As written it is not logical.

    The number of things is different from a change in the number of things. The sentence assumes they are the same.

    Also, the statement asserts that ‘If a measurement of a system counts the number of some set of things, then it overcounts the number’. That assertion  obviously has no basis in reality.

    • #29
  30. Ekosj Member
    Ekosj
    @Ekosj

    Mark Camp (View Comment):

    Steve Fast (View Comment):

     

     

    Since the Establishment survey counts jobs…the Establishment survey overstates job growth.

    I suggest that you edit this sentence so that it says in plain English what you mean. As written it is not logical.

    The number of things is different from a change in the number of things. The sentence assumes they are the same.

    Also, the statement asserts that ‘If a measurement of a system counts the number of some set of things, then it overcounts the number’. That assertion obviously has no basis in reality.

    Not at all illogical.

    my water meter on my home purports to measure how much water we use.   Yet last summer it reported that we used 3X the amount of water we ordinarily do.   Maybe we had a water leak and the meter accurately reported the additional flow.   
    Or…maybe the meter was malfunctioning.

    The meter was malfunctioning.

    • #30
Become a member to join the conversation. Or sign in if you're already a member.