Good essay! I am glad to see you taking the time to really work through your idea of what is going on, and where things will go, and appreciate the effort!
Just to keep the conversation up, I agree with your predictions, except I think that construction employment won't go down much, if at all. Skilled, even moderately, blue collar/trades workers are still hard to come by, and I expect that while there might be movement between jobs for many of the construction trades the total jobs will not decline much if at all. I wouldn't be surprised by a slight increase, either, but in general I think that very small oscillations around 0 are more likely than a real loss anything like the tech sector.
One point of disagreement or criticism, you write:
"It is the small net number of job gains or losses that determines the change in real GDP. A huge amount of bottom-up churning (4.5 million new jobs) lies underneath the GDP gain, which comes from the net new jobs (0.2 million)."
That isn't exactly true, and indeed your PSST theory suggests it shouldn't be. The NET change could be 0 while still having a positive (or negative) effect on GDP, as workers leave jobs they are less productive in towards jobs where they are more productive. The pattern changing is what matters, not the total number so much. One could imagine a badly run construction company failing, causing a loss of 100 jobs, then immediately another construction firm hiring all those workers (and buying all the physical capital) and putting them right back to work, creating 100 new jobs. Net 0 job creation, but the organizational capital of the surviving firm being better allows for higher productivity for the same workers, raising GDP.
"Some of the 4.5 million new jobs are actually old jobs with new occupants."
Suppose there was a way to measure how many of the 4.5 million new jobs are actually new jobs and not old jobs with new occupants. Creative distruction % of job churn. A measure of PSST patterns changing. Would this measure be fairly constant? Would it rise during the recovery period after the collapse of an obsolete sector? Would it fall during periods of extensive job hording? If the measure could be applied industry by industry, it would be more informative than one average for the whole economy.
I think asking whether the Fed is in control of GDP is a bit of false strawman. The Fed's job, difficult as it may be, is to flatten out the business cycle with the goal of maximizing long term growth, whether the peaks and valleys be due to labor hoarding, changes in spending and/or saving, price shocks, etc. (Minimizing unemployment is also a Fed goal.) The great difficulty is getting adequate quality information in a timely manner and making the right decisions. One could argue the Fed has gotten better at this but surely there is room for further improvement.
"but I doubt that the college graduating class of ‘23 who are hoping for jobs in the non-profit sector will want to work for Chick-Fil-A. So I think we will see some modest declines in the net increases in employment."
What empirical tests would shed light on what the Fed can control?
If the Fed can influence inflation, does that not logically require that they have non-trivial influence on NGDP?
(My bias: I like Arnold's PSST and think its underrated, but I'm inclined to agree with most macroeconomists that the Fed has a fair amount of control, and I find Sumner's specific views fairly persuasive as well.)
I agree with everything you say (AFAIK Scott Summer does too) , but I don't see the macro policy implication. Which taxation or expenditure decisions of Congress are affected? What financial instruments should the Fed be buying or selling in what amounts based on what kind of information?
really love this "austrian" like article. when it comes to employment i think it obviously depends on the skills needed. often finding replacements cost a fortune so before letting certain employees go companies think twice
A more operational alternative to the labor hoarding explanation of business cycles arises from the stickiness of consumption and sequestered capital. I have publications on this beginning in 2017: the one empirical pub is in Financial History Review providing an explanation of the timing of the Tulipmania boom and bust.
Revisiting Arnold's encouraging mention of JOLTS, the Job Openings and Labor Turnover Survey, I have long considered conventional reporting of unemployment figures misleading.
As an example, last month's Wall Street Journal November 4 headline read:
"Economy adds 261,000 jobs; unemployment rate increases to 3.7%".
No, per Arnold above, 4.5 million were added and 4.3 million were lost;
using Arnold's figure, 261,000 was the net gain, not the number of jobs added.
Here's JOLTS progress from the Bureau of Labor Statistics release today:
"October job openings edge down; hires and separations change little."
"Job openings edged down to 10.3 million on the last business day of October. Hires and total separations changed little at 6.0 million and 5.7 million, respectively."
It seems to me that the 4.5 or 4.7 million (5.7 & 6.0) are far more important than the small difference between them. I'm pleased to see the BLS reporting this as a top line item.
My recollection is that I got the insight from Alchian & Allen describing job changes as massive churn (up and down magnitudes massive, net change not) in "Exchange and Production", my MBA Micro textbook and their successor to the classic "University Economics" . Tyler Cowen and Alex Tabarrok have a nice graph in their textbook, tall columns for gain & loss, with a small difference.
I wonder if PSST is empirically distinguishable from real business cycle theory (RBCT) when considering aggregate data? Of course, they are distinguishable when observing individual markets or industries b/c RBCT is top down and doesn't produce measurable predictions about distinct industries.
1. A theory is only good if it can inform practice. In practice the central banks have a few dials and it is important for them to know what to do with those dials - and this is probably the most important application of macro-economy theories. You want a more bottom-up theory - then how would it inform these top-down decisions?
2. I think there is a similarity between the economic system and forest fires and other https://en.wikipedia.org/wiki/Self-organized_criticality and what you write about the bottom-up process of creative destruction, with the build up of zombie companies like dry wood in forests, is very close to that kind of thinking.
3. You write elsewhere that you wait now for a 'creative destruction event' to clean up the economy - how do you know that we are close to one?
4. The 'herd behaviours' seem to be circumstances where we do need top-down governance.
"assumption that the Fed is in control. Dropping that assumption takes many people out of their comfort zone"
How anybody can believe that the Fed is in control of anything is beyond me. Just take a look at the value of the dollar (their primary objective) over a century...
You have misunderstood Scott Sumner’s market monetarism. First, it is not presently true that when NGDP differs from its trend path, the Fed has messed up, because the Fed has not accepted NGDP trend-level as its proper target. But suppose it did so (as, Sumner thinks, it should). Then the Fed should use its open-market operations to control not NGDP, but *the market forecast of NGDP*. It is irrelevant that NGDP is produced not by the Fed (from the top) but *from the bottom up*; the Fed is fully capable of manipulating the market's forecast of the nominal level, however that is produced.
The Fed has a few big dials - interest rates being the most discussed and watched. But like a thermostat in many (most?) open office buildings, turning the heat up or down causes hot or cold spots, and is particularly variable if there is on open window on that floor.
Interest rates are big influence - but not controlling.
Gov't tax rates, and especially changes, are a big influence.
Gov't spending is a big influence.
For each local region, sales & sales changes of key industries in that region are a big influence.
For all firms in an industry, the price & production plans of competitors are an influence.
Tech changes are an influence.
Public fashion desires are an influence.
House construction & completions are an influence.
No one of these influences is controlling, they all have a variable influence. On inflation, an increase in interest rates to reduce inflation can be met, or meet, an increase in gov't debt based spending that increases inflation such that none can accurately predict what near future CPI inflation will actually be like. Or perhaps a tax cut (of same deficit amount), which allows firms to more rapidly make changes.
Complicated by the reality that measured CPI inflation is not total real inflation (of price changes in all things bought).
The PSST theory is more obviously "true" than Sumner's Fed dial - but for any policy advisor, modelling with such a dial allows much more confidence in giving policy advice.
Inflation too high? Higher rates...
In many firms, the decision to invest in a new project or even a new job is based on the expected Return On Investment. When interest rates are high, fewer new projects have highly positive expected ROIs - so fewer new adjustments are tried.
Macro is not and cannot ever be a "repeatable" experiment type science as is physics. Nor is it a value free engineering science because of too many influences no matter what dial is changed, since so many people use so many different influence dials, often in opposition.
It's too bad that PSST isn't more taught and used for more research into further research and data collection between micro & macro (midi? mini?) aggregations for correlation analysis.
My youngest sister is an occupational therapist. She took a new position 4 weeks ago, but the health care facility still hasn't been able to put her to work because they haven't been able to get her fully put "into the billing system" so that she can do all the paperwork for patients. So they have been paying her salary for a month now without her actually working. I asked her how this happened, and she told me that her manager told her that when they advertised for the position, they didn't expect to be able to fill it until Spring 23 at the earliest, so they just weren't prepared for her to actually be working right now.
Your sister's situation reminded me of my daughter's, and I wonder how "they" account for my daughter's job transition.
She had been working for the last several years part-time from home as a contract "in-your-home" RE loan closer (recently-divorced mom of 4 kids (18, 16, 14, 12 yo.) ... which demand for went gang-busters during the covid years, but more recently proved insufficient for her income needs, given the drastically falling real estate market in the metro-Denver CO area.
So, some months ago, with her income future looking increasingly frail & with the youngest child now older in age, she returned to her pre-marriage and pre-children career of service provider for the developmentally disabled, now (as earlier) as an employee for the local facility, DDRC (ddrcco[dot]com].
So, she went from self-employed contractor status to that of FTE. Do you suppose that's a job gain or a job loss, or net zero?
Good essay! I am glad to see you taking the time to really work through your idea of what is going on, and where things will go, and appreciate the effort!
Just to keep the conversation up, I agree with your predictions, except I think that construction employment won't go down much, if at all. Skilled, even moderately, blue collar/trades workers are still hard to come by, and I expect that while there might be movement between jobs for many of the construction trades the total jobs will not decline much if at all. I wouldn't be surprised by a slight increase, either, but in general I think that very small oscillations around 0 are more likely than a real loss anything like the tech sector.
One point of disagreement or criticism, you write:
"It is the small net number of job gains or losses that determines the change in real GDP. A huge amount of bottom-up churning (4.5 million new jobs) lies underneath the GDP gain, which comes from the net new jobs (0.2 million)."
That isn't exactly true, and indeed your PSST theory suggests it shouldn't be. The NET change could be 0 while still having a positive (or negative) effect on GDP, as workers leave jobs they are less productive in towards jobs where they are more productive. The pattern changing is what matters, not the total number so much. One could imagine a badly run construction company failing, causing a loss of 100 jobs, then immediately another construction firm hiring all those workers (and buying all the physical capital) and putting them right back to work, creating 100 new jobs. Net 0 job creation, but the organizational capital of the surviving firm being better allows for higher productivity for the same workers, raising GDP.
"Some of the 4.5 million new jobs are actually old jobs with new occupants."
Suppose there was a way to measure how many of the 4.5 million new jobs are actually new jobs and not old jobs with new occupants. Creative distruction % of job churn. A measure of PSST patterns changing. Would this measure be fairly constant? Would it rise during the recovery period after the collapse of an obsolete sector? Would it fall during periods of extensive job hording? If the measure could be applied industry by industry, it would be more informative than one average for the whole economy.
I think asking whether the Fed is in control of GDP is a bit of false strawman. The Fed's job, difficult as it may be, is to flatten out the business cycle with the goal of maximizing long term growth, whether the peaks and valleys be due to labor hoarding, changes in spending and/or saving, price shocks, etc. (Minimizing unemployment is also a Fed goal.) The great difficulty is getting adequate quality information in a timely manner and making the right decisions. One could argue the Fed has gotten better at this but surely there is room for further improvement.
"but I doubt that the college graduating class of ‘23 who are hoping for jobs in the non-profit sector will want to work for Chick-Fil-A. So I think we will see some modest declines in the net increases in employment."
...What makes you think they want to hire them?
What empirical tests would shed light on what the Fed can control?
If the Fed can influence inflation, does that not logically require that they have non-trivial influence on NGDP?
(My bias: I like Arnold's PSST and think its underrated, but I'm inclined to agree with most macroeconomists that the Fed has a fair amount of control, and I find Sumner's specific views fairly persuasive as well.)
I also like PSST, and agree that the Fed has a big influence. How much? 20-80%? 40%? 70%?
I agree with everything you say (AFAIK Scott Summer does too) , but I don't see the macro policy implication. Which taxation or expenditure decisions of Congress are affected? What financial instruments should the Fed be buying or selling in what amounts based on what kind of information?
really love this "austrian" like article. when it comes to employment i think it obviously depends on the skills needed. often finding replacements cost a fortune so before letting certain employees go companies think twice
A more operational alternative to the labor hoarding explanation of business cycles arises from the stickiness of consumption and sequestered capital. I have publications on this beginning in 2017: the one empirical pub is in Financial History Review providing an explanation of the timing of the Tulipmania boom and bust.
Revisiting Arnold's encouraging mention of JOLTS, the Job Openings and Labor Turnover Survey, I have long considered conventional reporting of unemployment figures misleading.
As an example, last month's Wall Street Journal November 4 headline read:
"Economy adds 261,000 jobs; unemployment rate increases to 3.7%".
No, per Arnold above, 4.5 million were added and 4.3 million were lost;
using Arnold's figure, 261,000 was the net gain, not the number of jobs added.
Here's JOLTS progress from the Bureau of Labor Statistics release today:
"October job openings edge down; hires and separations change little."
"Job openings edged down to 10.3 million on the last business day of October. Hires and total separations changed little at 6.0 million and 5.7 million, respectively."
It seems to me that the 4.5 or 4.7 million (5.7 & 6.0) are far more important than the small difference between them. I'm pleased to see the BLS reporting this as a top line item.
My recollection is that I got the insight from Alchian & Allen describing job changes as massive churn (up and down magnitudes massive, net change not) in "Exchange and Production", my MBA Micro textbook and their successor to the classic "University Economics" . Tyler Cowen and Alex Tabarrok have a nice graph in their textbook, tall columns for gain & loss, with a small difference.
I wonder if PSST is empirically distinguishable from real business cycle theory (RBCT) when considering aggregate data? Of course, they are distinguishable when observing individual markets or industries b/c RBCT is top down and doesn't produce measurable predictions about distinct industries.
A few notes:
1. A theory is only good if it can inform practice. In practice the central banks have a few dials and it is important for them to know what to do with those dials - and this is probably the most important application of macro-economy theories. You want a more bottom-up theory - then how would it inform these top-down decisions?
2. I think there is a similarity between the economic system and forest fires and other https://en.wikipedia.org/wiki/Self-organized_criticality and what you write about the bottom-up process of creative destruction, with the build up of zombie companies like dry wood in forests, is very close to that kind of thinking.
3. You write elsewhere that you wait now for a 'creative destruction event' to clean up the economy - how do you know that we are close to one?
4. The 'herd behaviours' seem to be circumstances where we do need top-down governance.
"assumption that the Fed is in control. Dropping that assumption takes many people out of their comfort zone"
How anybody can believe that the Fed is in control of anything is beyond me. Just take a look at the value of the dollar (their primary objective) over a century...
The Fed, like most central banks, has an inflation "target" of 2%.
You have misunderstood Scott Sumner’s market monetarism. First, it is not presently true that when NGDP differs from its trend path, the Fed has messed up, because the Fed has not accepted NGDP trend-level as its proper target. But suppose it did so (as, Sumner thinks, it should). Then the Fed should use its open-market operations to control not NGDP, but *the market forecast of NGDP*. It is irrelevant that NGDP is produced not by the Fed (from the top) but *from the bottom up*; the Fed is fully capable of manipulating the market's forecast of the nominal level, however that is produced.
The Fed has a few big dials - interest rates being the most discussed and watched. But like a thermostat in many (most?) open office buildings, turning the heat up or down causes hot or cold spots, and is particularly variable if there is on open window on that floor.
Interest rates are big influence - but not controlling.
Gov't tax rates, and especially changes, are a big influence.
Gov't spending is a big influence.
For each local region, sales & sales changes of key industries in that region are a big influence.
For all firms in an industry, the price & production plans of competitors are an influence.
Tech changes are an influence.
Public fashion desires are an influence.
House construction & completions are an influence.
No one of these influences is controlling, they all have a variable influence. On inflation, an increase in interest rates to reduce inflation can be met, or meet, an increase in gov't debt based spending that increases inflation such that none can accurately predict what near future CPI inflation will actually be like. Or perhaps a tax cut (of same deficit amount), which allows firms to more rapidly make changes.
Complicated by the reality that measured CPI inflation is not total real inflation (of price changes in all things bought).
The PSST theory is more obviously "true" than Sumner's Fed dial - but for any policy advisor, modelling with such a dial allows much more confidence in giving policy advice.
Inflation too high? Higher rates...
In many firms, the decision to invest in a new project or even a new job is based on the expected Return On Investment. When interest rates are high, fewer new projects have highly positive expected ROIs - so fewer new adjustments are tried.
Macro is not and cannot ever be a "repeatable" experiment type science as is physics. Nor is it a value free engineering science because of too many influences no matter what dial is changed, since so many people use so many different influence dials, often in opposition.
It's too bad that PSST isn't more taught and used for more research into further research and data collection between micro & macro (midi? mini?) aggregations for correlation analysis.
Arnold, as a blogging style I hugely appreciate your humility: "I think of NGDP as built from the bottom up" -- your thoughtful opinions.
But theories are based on facts: "NGDP is built from the bottom up."
Which is true in reality, so as the theorizer your influence would be greater stating as facts those facts used in the PSST framework.
More college students studying economics learning PSST would be good; don't know how to get that result.
What the Fed really controls is monetization of government debts. Powell seems to want to get the Fed out of that business. Good luck with that.
My youngest sister is an occupational therapist. She took a new position 4 weeks ago, but the health care facility still hasn't been able to put her to work because they haven't been able to get her fully put "into the billing system" so that she can do all the paperwork for patients. So they have been paying her salary for a month now without her actually working. I asked her how this happened, and she told me that her manager told her that when they advertised for the position, they didn't expect to be able to fill it until Spring 23 at the earliest, so they just weren't prepared for her to actually be working right now.
Your sister's situation reminded me of my daughter's, and I wonder how "they" account for my daughter's job transition.
She had been working for the last several years part-time from home as a contract "in-your-home" RE loan closer (recently-divorced mom of 4 kids (18, 16, 14, 12 yo.) ... which demand for went gang-busters during the covid years, but more recently proved insufficient for her income needs, given the drastically falling real estate market in the metro-Denver CO area.
So, some months ago, with her income future looking increasingly frail & with the youngest child now older in age, she returned to her pre-marriage and pre-children career of service provider for the developmentally disabled, now (as earlier) as an employee for the local facility, DDRC (ddrcco[dot]com].
So, she went from self-employed contractor status to that of FTE. Do you suppose that's a job gain or a job loss, or net zero?