Parsing the September Employment Release

Overall and private nonfarm payroll (NFP) employment surprised on the downside — but the fact that private missed by a smaller amount suggests that the slowdown is a little less pronounced than indicated by the headline number.

First consider private vs. total NFP, as well as private NFP hours:

Figure 1: Nonfarm payroll employment (black), private nonfarm payroll employment (tan), and aggregate weekly hours of private nonfarm payroll employment (pink), all seasonally adjusted, in logs, 2020M02=0. Source: BLS, and author’s calculations.

The slowdown was much less pronounced for private NFP, hinting at (the oft mentioned) problems with seasonal adjustment issues for government employment, particularly education-related.

Furthermore, private hours continued to rise fairly smartly, at 6.2% m/m annualized (and at 3 month annualized changes). (h/t Jason Furman and Powell, who cites the advice of Steven Braun).

Additional insight into the role of the government sector can be seen in a decomposition of net job losses relative to NBER peak month (2020M02):

Figure 2: Employment relative to 2020M02 for manufacturing (blue), accommodations and food services (tan), government (green) and all other nonfarm payroll employment, in thousands, seasonally adjusted. Source: BLS and author’s calculations.

One of the widening bars is government employment. And accommodation and food services fails to shrink, indicating stalled recovery. That latter definitely points to the impact of the delta variant on high contact services consumption.

Finally, wages continue to rise in nominal terms, possibly in real terms (we have September average hourly wages, but not September CPI). In any case, it’s important to remember that the average does not control for composition effects. A bit more information is provided by a disaggregated look at the extremes, i.e., manufacturing (goods producing) vs. high contact services (leisure and hospitality services). In all cases, these are production/non-supervisory numbers.

Figure 3: Real average hourly earnings in private nonfarm payroll employment (black), in manufacturing (red), and in leisure and hospitality (teal), CPI deflated, in logs 2020M02=0. September observations deflate using Cleveland Fed nowcast as of 10/10. All series pertain to production/non-supervisory workers. NBER defined recession dates shaded gray. Sources: BLS via FRED, Cleveland Fed, NBER and author’s calculations.


Clearly, leisure and hospitality services have surged far ahead of goods production, after suffering a big decline. That being said, real hourly earnings in leisure and hospitality are only 2.9% higher than 2014-19 trend. Another perspective on how real wages have evolved is gained by looking at actual levels (rather than normalized) of real wages.

Figure 4: Real average hourly earnings in private nonfarm payroll employment (black), in manufacturing (red), and in leisure and hospitality (teal), 2014-19 log-linear trend (gray), all CPI deflated to 2020$. September observations deflate using Cleveland Fed nowcast as of 10/10. All series pertain to production/non-supervisory workers. NBER defined recession dates shaded gray. Sources: BLS via FRED, Cleveland Fed, NBER and author’s calculations.

More on the implications of the release for assessing overall economic activity here. A comprehensive examination by Furman and Powell here. See also Calculated Risk.



43 thoughts on “Parsing the September Employment Release

  1. pgl

    Furman and Powell was indeed a comprehensive and well thought out discussion that closed with this optimism yet caution:

    “The next many million jobs should be relatively easy to get back given the large number of openings and continued rapid economic growth. More difficult will be the last several million jobs. Jobs may end up short of pre-pandemic projections if people have left the labor force for good (e.g., due to early retirement) or if other changes have permanently reduced labor demand (e.g., employers hiring fewer workers who are higher skill and higher paid). The hope is that all the jobs can be regained, productivity could rise to support wage increases, and that continued high levels of demand could put the economy on a future path that is better than what was expected before the pandemic. Time will tell.”

    A lot of this depends on us finally crushing the virus as well as smart policy choices. Get your vaccine, wear your mask, and let’s get the Build Back Better proposals passed.

  2. Moses Herzog

    I just got a strawberry coolatta from Dunkin Donuts that tasted like skim milk mixed with water and ice cubes. Threw out a $3.70 drink after two swallows from the straw. Uncle Moses is super cheap, and Uncle Moses gets super angry when these things happen. Maybe Dunkin
    Donuts should rename it rancid and disgusting watered down skim milk coolatta. Would at least be truthful. That’s my update on fast food employment for today. Don’t eat at Dunkin’ Donuts for 3 years and see if the idiots can learn how to make a fruit flavored drink. Guess where my money is on the accomplishment of that.

  3. ltr

    October 9, 2021

    Building Back Better and the September Jobs Report
    By Dean Baker

    If President Biden was designing a jobs report to advance his Build Back Better agenda, he could not have done better than the September jobs report. (No, he didn’t manipulate the data.) It shows the need for improving our caring infrastructure to make it possible for more women to work. It also shows the need to improve our infrastructure to limit supply chain disruptions.

    But before getting to these issues, it’s first important to dispel the idea that this was a bad jobs report. The September data showed a 0.4 percentage point drop in the unemployment rate, bringing it to 4.8 percent. Most analysts had predicted a drop of just 0.1 or 0.2 percentage points. We didn’t get the unemployment rate down to 4.8 percent following the Great Recession until January of 2016. And, this decline was due to workers getting jobs, not the unemployed dropping out of the labor market. The number of employed in the household survey increased by 526,000.

    The negative view of the September report is based on the weaker than expected job growth reported in the establishment survey. The increase of 194,000 in payroll jobs was well below the 400,000 to 500,000 job gain most analysts had expected. While that seems like a bad story, a closer look shows otherwise.

    The biggest contributor to the weak job growth story was the loss of 161,000 jobs in state and local government education. There is not an obvious explanation for this job loss (I’ll come back to this issue), but suppose that we didn’t see this sort of job loss in public sector education. Suppose that we instead regained more of the education jobs lost in the pandemic.

    The private sector added a healthy 317,000 jobs in September. If the public sector had added 100,000 jobs for the month, as many had expected, the Bureau of Labor Statistics would have reported job growth of 417,000 in September, well within the generally expected range.

    So, the big question is why are we losing jobs in education instead of adding them? It’s hard to come up with a good story here. Some analysts have suggested the problem is with the seasonal adjustment. On an unadjusted basis, state and local education added 1,033,000 jobs in September. The argument is that the seasonal adjustment is inappropriate for this year, since the pandemic has altered the normal timing of the school year and employment patterns.

    But this argument really doesn’t fit. If we just look at the non-seasonally adjusted data, employment in public education is down 431,000 from the levels of September of 2019, a decline of 4.8 percent. Schools are back to in-class instruction pretty much everywhere. Unless we have seen a sharp rise in student-to- teacher ratios, or a large decline in support staff, this drop in employment from before the pandemic doesn’t make sense.

    Anyhow, we will need to sort out what is going on with employment in public schools, but if we set that issue aside for the moment, the jobs picture in the establishment survey looks pretty good. As noted earlier, the 317,000 private sector jobs added in the month was very much consistent with expectations. But an element of the picture that has not gotten nearly the attention it deserves is the increase in the length of the average workweek.

    The average workweek rose by 0.2 hours. This rise in average weekly hours, coupled with the rise in employment, led to a rise of 0.9 in the index of aggregate hours. This is the largest increase since March.

    A story that is consistent with a rise in hours, coupled with a limited increase in payroll employment, is that employers are having trouble getting the workers they need to meet the demand they are seeing. They adjust to this situation by working their existing workforce more hours….

    1. rjs

      education jobs have been an outlier all year….BLS applies a standard seasonal adjustment to them, which doesn’t account for the effects of the pandemic…hence we get reports that we had a 220,700 job increase in local school districts in July, even as 902,100 school teachers were furloughed for the summer, because the number of layoffs did not meet the norm, simply because local education jobs were lower to begin with…the opposite happened in September, as local school systems added 144,200 fewer jobs and state universities adding 16,600 fewer jobs than is normal for September, shortfalls from the norm which are logged as decreases….a similar dynamic played out in employment in private educational services; while 248,600 more were working in private education in September than in August, it’s reported as an 18,900 job decrease….

      1. Moses Herzog

        Just a bare thin string relation to this question: Have we yet found out the whole deal with women’s employment being slow to recover?? Forgive me if Furman or somebody answered this already

    2. rjs

      just an anecdote: local TV news is reporting on a job fair in Cleveland hosted by 29 area businesses ready to hire…just 25 people showed up, and they closed early..

    1. pgl

      Oh gee – the My Pillow clown?

      Popular Information first reported on AT&T’s relationship with OAN in February, when OAN repeatedly ran a two-hour movie created by MyPillow CEO Mike Lindell, “Absolute Proof,” that uses discredited conspiracy theories to claim that Trump was the real winner of the 2020 election. “Absolute Proof” aired thirteen times from February 5 to 8 — encompassing 26 hours of OAN’s programming over four days.

      Lindell’s film promoted lies about Dominion, a voting machine company. For example, lawyer Matt DePerno claims that files “were deleted from the Dominion system in Antrim County. We know that for a fact.” Lindell responds, “wow.” Although a clerical error briefly showed a landslide for Biden in the county, it was corrected. There is absolutely no evidence of any vote manipulation in the county and the results were confirmed by a hand recount of the votes.

      While “Absolute Proof” was technically a paid advertisement, OAN promoted it as a news program. In a tweet, OAN billed it as an “exclusive report” about “[g]rowing evidence of election fraud reveal[ing] that the presidency of the United States has been stolen from the American people.”

      1. Moses Herzog

        @ T. Shaw
        First I wanna say this is not any attempt at all to be funny on my part. It is said more than once in the Bible that it is wrong to use the words of God or words inspired by God for false reasons. I would say you’re walking on very dangerous ground.

  4. ltr

    October 11, 2021

    The metamorphosis of growth policy
    By Dani Rodrik

    Development policy has long been divided between two types of approaches. One approach targets poor people directly and seeks to alleviate the poverty of individual households – through income support, health and education interventions, and enhanced access to credit. The other focuses on enhancing economic opportunities and raising overall productivity – through economy-wide macroeconomic and trade policies or legal and regulatory reforms. Call the first social policy and the second growth policy.

    These two types of policies are generally complementary. Aggregate growth may not always help everyone, especially the poor. Consequently, anti-poverty programs will be necessary even when growth policy is doing its job properly. Occasionally, however, social and growth policies have been viewed as substitutes.

    For example, the increased use of randomized policy experiments has allowed analysts to develop causal evidence about social policies – such as cash grants or education and health interventions – in ways that are rarely possible with macroeconomic or economy-wide policies. This, in turn, has led many academics and practitioners to downgrade the practical importance of growth policy relative to social policy.

    That is a mistake, because the real determinants of poverty may lie at some distance from poor households and communities. Economic development requires productive nonfarm jobs. Increasing employment opportunities in cities and encouraging migration from the countryside to urban areas may raise incomes more effectively than helping people become better farmers or providing them with cash grants.

    Indeed, industrialization has been essential to reducing poverty historically. True, the benefits of industrialization-driven economic growth often take time to trickle down. During Britain’s Industrial Revolution, living conditions for urban workers improved very slowly, if at all, for nearly a century until the rise of labor unions and other institutional changes redressed the imbalance of power with employers. But the more recent experience with rapid, export-oriented industrialization in the East Asian tigers and China has compressed this process and produced poverty-reduction miracles alongside growth miracles.

    There are clear signs that we are now entering a new era in which industrialization will no longer be as potent in spreading the benefits of economy-wide productivity gains. Global trends in innovation have significantly reduced the potential of manufacturing industries to absorb low-skill workers. The labor share in value added has fallen rapidly in these branches, particularly for such workers.

    And while globalization has accelerated the transfer of manufacturing from advanced economies to developing economies, global value chains have turned out to be at best a weak vehicle for creating good jobs, because they are a transmission belt for skill- and capital-intensive technologies, and because their business model is based on imported inputs and lack of integration with the local economy. Globally competitive manufacturing industries in developing economies increasingly operate as enclaves, similar to highly capital-intensive, export-oriented extractive industries. They may spur exports and higher incomes for a narrow segment of the economy, but they bypass most workers, and especially the least educated.

    This growth model falls short not only on equity or poverty-reduction grounds; it also fails to promote much growth because higher-productivity activities cannot encompass an increasing share of the economy. Just as resource-rich economies rarely grow for long (outside of terms-of-trade booms), the industrialization model is no longer capable of generating rapid and sustained economic growth.

    What, then, should today’s growth model look like? As always, investments in human capital, infrastructure, and better institutions remain indispensable for long-term economic gains. These are the fundamentals of economic convergence with rich countries. But a growth strategy worth its name must enhance the productivity of the existing workforce, not the workforce that might emerge in the future thanks to such investments….

    Dani Rodrik is a professor of International Political Economy at Harvard University’s John F. Kennedy School of Government.

    1. ltr

      Importantly, the need for a metamorphosis of growth policy has been anticipated in China and a suitable transformation of economic structures is already policy and being worked on. Dani Rodrik has written a critically incisive essay.

      1. macroduck

        Taiwan has, of course, an even better record than China at economic metamorphosis. Taiwan’s GDP per capita in 2020 was $28,306 vs just $10,500 in China. Taiwan is nearly 3 time better than China at providing for its people. All the more reason for Taiwan to remain free and independent of China.

        In fact, Taiwan’s supervisor performance suggests Taiwan should rule China.

    2. ltr

      Developing countries retain significant potential to increase agricultural productivity and to diversify from traditional to cash or export crops. But even with more productive agriculture – and in fact as a result of it – young workers will continue to leave the countryside and flock into urban areas. They will be employed not in factories but in informal, micro enterprises in low productivity services with poor expansion prospects….

      — Dani Rodrik

      [ The anticipated development problem is focused on here, a problem that countryside or rural infrastructure building should resolve given the Chinese understanding of problem.

      Again, a terrific essay. ]

  5. joseph

    Regarding employment, on twitter today:

    “Congratulations to Pennsylvania for winning the Nobel Prize for Control Groups.”

    1. ltr

      October 11, 2021

      The Nobel in economics goes to three who find experiments in real life.
      By Jeanna Smialek

      David Card has made a career of studying unintended experiments to examine economic questions — like whether raising the minimum wage causes people to lose jobs.

      Joshua D. Angrist and Guido W. Imbens have developed research tools that help economists to use real-life situations to test big theories, like how additional education affects earnings.

      On Monday, their work earned them the 2021 Nobel Memorial Prize in Economic Sciences….

    2. T. Shaw

      More congratulations!

      According to FRED, PA’s unemployment rate was 6.4%, which compares favorably to CA – 7.5% and NY – 7.4%. However, FL and SC were 5.0% and 4.2% respectively. . ; ,

          1. T. Shaw

            Good one, Mac.

            My mistake! CA and NY are definitely achieving superior performance over the cited others.

  6. ltr

    October 11, 2021

    The City of London Is Hiding the World’s Stolen Money
    By Nicholas Shaxson

    In 1969, two years after the Cayman Islands, a British territory, passed its first law to allow secretive offshore trusts, an official government report struck an ominous note. A tide of glossy propositions from private developers, it warned, was washing through the islands. Cayman was fast becoming a state captured by shady finance.

    Those were the pungent beginnings of a modern system brought to light by the Pandora Papers, an enormous data leak coordinated by the International Consortium of Investigative Journalists. The papers exposed a smorgasbord of secretive and questionable financial dealings by more than 330 politicians and public officials from over 90 countries and territories — and over 130 billionaires from Russia, the United States and elsewhere. On display was a dizzying array of chicanery and wealth hoarding, often by the very people who should crack down on it.

    The revelations, published on Oct. 3, are global in scope. But if there is one country at the system’s heart, it is Britain. Taken together with its partly controlled territories overseas, Britain is instrumental in the worldwide concealment of cash and assets. It is, as a member of the ruling Conservative Party said last week, “the money laundering capital of the world.” And the City of London, its gilded financial center, is at the system’s core.

    For Britain, whose bloated financial sector exacerbates widespread economic problems, that’s bad enough. For the world, at the mercy of an economic system rigged for the rich, it’s even worse.

    The offshore ecosystem is, by design, fiendishly complicated. Many intricate and opaque instruments — including offshore trusts, tax loopholes and shell companies — plus banking secrecy and negligent financial regulation shroud the wealthy’s assets in murky legal mists. Central to it all are tax havens, such as the Cook Islands, British Virgin Islands and Jersey (one of the Channel Islands), which can operate like smugglers’ coves. The wealthy and nefarious take their money there to protect it, but also to escape from rules, laws and taxes they don’t like.

    The wealth held in tax havens is staggering: Estimates range from $6 trillion to $36 trillion. And some tax havens are closer to home than many would imagine. The United States, with its shady Delaware shell companies and South Dakota trusts, has long been a big part of the secrecy system. A cluster of European countries, including Luxembourg, Ireland and Switzerland, offer another menu of escape routes. Asia, of course, has Hong Kong and Singapore.

    But the British network is surely the biggest. The Tax Justice Network’s Financial Secrecy Index, a ranking of tax havens, shows that Britain and its “spider’s web” of offshore satellites would rank first. Over two-thirds of the 956 companies that the Pandora Papers link to public officials were set up in the British Virgin Islands.

    Central to the process is the City of London….

    1. ltr

      The leading comment:

      The City of London Is Hiding the World’s Stolen Money


      Londoner here. Almost every building in the city is owned by some foreign entity nobody knows anything about. Streets are depopulated from St. James to Holland Park. What was 20 years ago a living, breathing city has been reduced to an investment asset made of bricks and wood, plunged in darkness night after night. Real life has been pushed to the ever expanding margins. It’s madness. The agonizing death of a city that has gorged itself on money. From what I hear, Paris, New York and others aren’t too far behind. This is what happens when the mode of measuring (money) becomes the goal.

  7. macroduck

    Furman and Powell refer to the JOLT report, noting that both openings and quits are high, typically signs of a strong labor market. This looks to me like a “Yes, but…” situation.

    Strong, yes, but also a market in which the cost of working is higher than in the past. A high cost of working is consistent with high quits and high opening. A higher cost of working would also mean that a given number of openings will be associated with a slower pace of job creation than in the past. That was true before the pandemic, by the way, and is clearly even truer today.

    What must happen when the cost of working has gone up? Either the benefit of working must also rise or the cost of working must fall back again or the equilibrium level of employment must fall. Raising the benefit of working and lowering the cost of working both amount to increased employment costs.

    Child care benefits, sick and quarantine leave, safe work places, reasonable work schedules and rules would lower the cost of working.

  8. Moses Herzog

    Can you say “Oops!!!” ?? Well, about 10 years worth of “oops”.

    He only upset, women, Native Americans, trans, homosexuals, Blacks, the CEO of the monopoly company he works for. Uhm. Did I forget anyone he offended?? Let me know. Wow, I mean I am not SJW. But…… when are these guys gonna learn we are living in a different world, it is not 1985 anymore kids. That has it’s puses and minuses. In this case, I would say good riddance.

    1. Moses Herzog

      It’s interesting to note, Amy Trask was CEO of the Oakland Raiders for 16. NOw as far as I know she was generally considered to be a capable and competent in her work. Always struck me as sharp when speaking. Now, she was no Bobby Beathard. But niether was she a Matt Millen. She was competent and very sharp of mind from my distant impressions of her out here in fly-over country.

      If the Raiders leadership is sharp (an open question at this point) they’ll re-hire Amy Trask and make her head of a “search committee” to find a good long-term replacement for Jon Dunderhead Gruden. I wager she would find the right fit.

    2. pgl

      The first clue was how he talked about working at Hooters. That anyone would not know he was beyond being a jerk would suggest that person was not paying attention.

      1. Moses Herzog

        @ pgl
        It’s cool to see your perception of sexist men has “improved” since your years long personal adoration of Andrew Cuomo. Months after the nursing home deaths story broke you still had him as Peter Parker-like integrity. Was it the cleavage in the adverts that turned the tide for you on Gruden?? That’s an encouraging sign you’ll be played for less of a dunce in the future.

      1. Moses Herzog

        Is that hard to figure out?!?!?!?~~ that the original investigation was on Snyder. That’s about two sentences in on each and every story that copy/pasted it off NYT (Your braindead personage is still reading “TPM” I assume??~~ when you can’t get Menzie to play “apologetics” for you on it???, which probably explains why you caught that fact with your horrendous reading comprehension. But doesn’t explain why in your nutjob mind why you want to give your F**cked-up personal psycho characteristics to others on this blog. You’re telling us what?? Snyder’s actions excuse Gruden’s?!?!?!? This must be why you still love Cuomo. Were you on Andrew Cuomo’s staff?? Were you reaching for something in a dark room in Cuomo’s state mansion “saving multiple rape victims”) when the story broke?? Wow man. Hope they had the straight jacket on you by early afternoon.

        1. pgl

          I have never seen so much word salad gibberish in my life. How drunk were you when you wrote that nonsense?

Comments are closed.