Aug 17, 2022·edited Aug 17, 2022Liked by Joseph Politano
What a difficult paper to write but excellent. I suppose establishment survey is the same as the non farm data but swapping back and fore between them in the post is natural. Yes on a second read it is clearly stated to be so.
Somewhere recently it was said economists have no true theory of inflation. The inherent differences between establushment and household survey suggests the employment data reporting is a problem also. Add to this the fact that the measuring of a recession is up in the air too. It would all suggest a dismal science falling apart with no precise knowledge of its foundational data and so at any given time and even the important times when the various monthly data is released nothing is really known. Is it an assault on economics. I suppose not but it sure makes things confused.
But our host has performed an excellent task highlighting how the consequences are confusing and contradictory with different policy solutions required.
I decided not to edit the typo establushment because it seemed appropriate for the confused state on employment as potentially very embarrassing.
I’m an engineer with new found interest in economics thanks to you! I learn so much just by reading your pieces.
You mentioned the multiple job holdings pre-pandemic was acyclic but it tracks well ( proportional) with the employment levels - won’t that make it cyclic ?
I'm glad you're enjoying the content so much! It means a lot to me.
I probably should have been clearer about the multiple jobholding stuff. The rate of multiple jobholding is acyclical (as in, the percent of workers with two or more jobs doesn't change much during recessions) but the level of multiple jobholding is cyclical (as in, the number of workers with multiple jobs will decline during recessions).
Are there limitations to get self-employed people included in some type of updated establishment survey? I know the self employed population has historically been relatively small. But if the freelance/contract workforce continues to expand through freelance and gig platforms in the coming years, I would imagine the fed would want much less noisy data for this sector.
Could the establishment survey update to force businesses to disclose 1099 payrolls on a monthly basis?
Yes, I really wish they would include 1099s with the establishment survey. I don't know of any efforts to change this though, changes to the surveys are always slow moving and the agencies are always hesitant to change things too much so that it remains easy to compare to historical data.
I think historical comparisons could still be possible even with more 1099 data collection. If they decide to collect the 1099 data on the side, you could calculate a new-and-improved payroll metric, but you can also calculate the metric by the original payroll definition and make backward-looking comparisons.
Okay—I was doing more research into this topic and found a one-off survey the BLS did on electronically mediated workers (i.e., people who worked for gig-style online programs) as part of the 2017 contingent worker supplement. The data is very interesting, but what I find important is that BLS had to manually recode the data because survey respondents simply didn't understand the questions well enough (bank managers and police officers were self-identifying as gig workers!). Goes to show how messy this stuff can be!
Do response rates differ between the HH and establishment surveys? And have they changed over this period? It seems like one source of divergence - maybe on the HH survey the response rates for unemployed people have recently risen relative to employed people.
Response rates are definitely a big issue, they've been declining for years in both surveys and dropped off pretty significantly at the start of the pandemic. I don't know if there's a big gap between the two surveys though, or how that would effect measurements precisely.
I suspect that the PPP program might have had a negative impact on the accuracy of the employment survey. One of the variables affecting repayment of the loan was number of employees, and I wonder if employers getting a monthly call from BLS about their employee count wouldn't develop a tendency to exaggerate.
What a difficult paper to write but excellent. I suppose establishment survey is the same as the non farm data but swapping back and fore between them in the post is natural. Yes on a second read it is clearly stated to be so.
Somewhere recently it was said economists have no true theory of inflation. The inherent differences between establushment and household survey suggests the employment data reporting is a problem also. Add to this the fact that the measuring of a recession is up in the air too. It would all suggest a dismal science falling apart with no precise knowledge of its foundational data and so at any given time and even the important times when the various monthly data is released nothing is really known. Is it an assault on economics. I suppose not but it sure makes things confused.
But our host has performed an excellent task highlighting how the consequences are confusing and contradictory with different policy solutions required.
I decided not to edit the typo establushment because it seemed appropriate for the confused state on employment as potentially very embarrassing.
Excellent analysis as always!
Thank you! I am glad you enjoyed.
First time reading your work. Very insightful! Just subscribed
Thank you! I'm very glad you enjoyed
I’m an engineer with new found interest in economics thanks to you! I learn so much just by reading your pieces.
You mentioned the multiple job holdings pre-pandemic was acyclic but it tracks well ( proportional) with the employment levels - won’t that make it cyclic ?
I'm glad you're enjoying the content so much! It means a lot to me.
I probably should have been clearer about the multiple jobholding stuff. The rate of multiple jobholding is acyclical (as in, the percent of workers with two or more jobs doesn't change much during recessions) but the level of multiple jobholding is cyclical (as in, the number of workers with multiple jobs will decline during recessions).
Are there limitations to get self-employed people included in some type of updated establishment survey? I know the self employed population has historically been relatively small. But if the freelance/contract workforce continues to expand through freelance and gig platforms in the coming years, I would imagine the fed would want much less noisy data for this sector.
Could the establishment survey update to force businesses to disclose 1099 payrolls on a monthly basis?
Yes, I really wish they would include 1099s with the establishment survey. I don't know of any efforts to change this though, changes to the surveys are always slow moving and the agencies are always hesitant to change things too much so that it remains easy to compare to historical data.
Thanks for the reply! That makes a lot of sense.
I think historical comparisons could still be possible even with more 1099 data collection. If they decide to collect the 1099 data on the side, you could calculate a new-and-improved payroll metric, but you can also calculate the metric by the original payroll definition and make backward-looking comparisons.
Okay—I was doing more research into this topic and found a one-off survey the BLS did on electronically mediated workers (i.e., people who worked for gig-style online programs) as part of the 2017 contingent worker supplement. The data is very interesting, but what I find important is that BLS had to manually recode the data because survey respondents simply didn't understand the questions well enough (bank managers and police officers were self-identifying as gig workers!). Goes to show how messy this stuff can be!
https://www.bls.gov/cps/electronically-mediated-employment.htm
Do response rates differ between the HH and establishment surveys? And have they changed over this period? It seems like one source of divergence - maybe on the HH survey the response rates for unemployed people have recently risen relative to employed people.
Response rates are definitely a big issue, they've been declining for years in both surveys and dropped off pretty significantly at the start of the pandemic. I don't know if there's a big gap between the two surveys though, or how that would effect measurements precisely.
I suspect that the PPP program might have had a negative impact on the accuracy of the employment survey. One of the variables affecting repayment of the loan was number of employees, and I wonder if employers getting a monthly call from BLS about their employee count wouldn't develop a tendency to exaggerate.