Macro ReviewHow to read
Macro · Labour · Payrolls

Hiring averages 137k a month, above the pace that holds unemployment steady

How fast are employers adding jobs, and is it enough to hold unemployment steady?

BLS Employment Situation · 6 Nov 2026Data to Oct 2026 · updated 8 Nov 2026
3-month average+137kto Oct 2026 · +90k vs breakeven ≈47kabove breakeven
Payroll changeⓘ+165kOct 2026 · +118k vs breakeven ≈47kabove breakeven
Industries adding jobs, 3 monthsⓘ60.4%Oct 2026 · 64.0% · narrower than 2015–19
Net revision, prior two monthsⓘ+84kOct 2026 release · past-year average −18k a month
Health care & social assistance+191ksix months to Oct 2026 · 39% of the jobs added
Employment, year on year+0.56%Oct 2026 · +885k over 12 months

In brief

Is hiring keeping pace with the labour force, and how broad and how firm is it?
  • Payrolls grew by 137k a month over three months (the 39th 1990), above the ≈47k a month that holds unemployment steady.→ Exhibit
  • About 47k jobs a month keep the unemployment rate stable, on today's population and ageing trends.→ Exhibit 1
  • 60.4% of private industries added jobs over three months, against 64.0% in 2015–19.→ Exhibit
  • Education & health and professional & business account for +343k of the +495k jobs added over six months.→ Exhibit
  • This release revised the prior two months by a net +84k; first prints were revised by −18k a month over the past year.→ Exhibit
What would change this view
What would change this viewTriggerTodayDecided by
Hiring falls below breakeventhree-month average < 47k a month (breakeven, Oct 2026 inputs)+137k (Oct 2026)not crossedNov 2026 jobs report (BLS)
Breadth widensthree-month > 62.260.4 (Oct 2026)not crossedNov 2026 jobs report (BLS)
The latest three months are revised the other waynet revision to Aug 2026–Oct 2026 beyond 55k downthis release +84k (Oct 2026)pendingNov 2026 and Dec 2026 jobs reports (BLS)
The reasoning behind each point
  • Pace
    Payrolls grew by 137k a month over three months (the 39th percentile since 1990), above the ≈47k a month that holds unemployment steady.

    Payrolls rose 165k in Oct 2026 (first estimate); six-month average +82k. The three-month average sits at the 39th percentile of all months since 1990 and below the three-month average's 2015–19 mean of +193k. Private payrolls average +126k over three months; government +11k. Employment is +0.56% on the year.

  • Breakeven
    About 47k jobs a month keep the unemployment rate stable, on today's population and ageing trends.

    The scenario range is 42k to 104k a month. Population 16+ is growing 0.48% a year over the last 11 months on BLS's population-consistent series, from Nov 2025 because Oct 2025 was not collected (0.79% over 36 months), and population ageing alone lowers the participation rate by 0.12% a year (age-specific rates held; observed participation −0.16% a year over 24 months, which includes cyclical and other moves). Applied to 159.3 million payroll jobs that is 47k a month; the historical rule averaged 76k in 2019.

  • Breadth
    60.4% of private industries added jobs over three months, against 64.0% in 2015–19.

    Over one month 57.6%. A diffusion index of 50 means as many industries gained as lost; breadth is narrower than in the pre-pandemic expansion.

  • Where
    Education & health and professional & business account for +343k of the +495k jobs added over six months.

    Health care and social assistance alone added +191k. The weakest sector over six months: information & financial (−94k). Information employment is −119k (−4.2%) on the year and financial activities −99k.

  • Revisions
    This release revised the prior two months by a net +84k; first prints were revised by −18k a month over the past year.

    Sep 2026 moved from +29k to +60k and Aug 2026 from +133k to +186k. Over the last 12 months with a third estimate, the final sample-based change differed from the first print by −18k a month on average (7 of 12 revised down); BLS's May 2003–present average is +7k. Annual benchmark revisions come on top.

  • Uncertainty
    The breakeven is a rough guide: out of sample it has missed the year-ahead unemployment rate by 0.7pp, about ±98k jobs a month.

    The miss is the root mean square error of the rule's 12-month call on the unemployment rate (0.71pp). Windows ending 2012–2019: RMSE 0.46pp, mean error −0.34pp, correlation 0.47; ending 2023–2026: RMSE 1.08pp, mean error −0.68pp, correlation 0.40. A negative mean error means unemployment fell less than payroll growth implied: participation rose with hiring, and official population growth understated the immigration surge until later January controls caught up. With BLS's population-consistent population the recent-window RMSE is 0.93pp (revised data, not available at the time). A payroll ratio trending 0.25pp a year either way shifts breakeven by 33k.

  • Actual
    Payrolls are adding 137k a month over three months: above breakeven by 90k.

    Sustained for a year, that pace would move the unemployment rate by about −0.65pp (payroll jobs per employed person 0.974, labour force 170.7 million). Over six months payrolls averaged 82k.

The breakeven model · The estimate is wrong if payrolls hold near 47k a month for six months while the unemployment rate's three-month average moves by more than 0.4pp, or if the next population controls revise 16+ population growth by more than 0.4 points a year.

Next: Unemployment — does hiring hold unemployment steady?

BLS, Current Employment Statistics (Table B-1, revisions) and Current Population Survey; Macro Review calculations.labour.payrolls-read

Jobs a month, and the turn in the trend

How many jobs a month, and is the trend turning?

Exhibit 1 · Payroll growth against the pace that holds unemployment steady

Thousands of jobs a month, Oct 2021 to Oct 2026: bars, three-month average; line, twelve-month average; dashed, the breakeven (≈47k); shaded, its scenario range.
About this exhibit
Payroll growth, thousands of jobs a month, monthly, last five years by default: three-month average (bars) and twelve-month average (line), against the breakeven (dashed) and its scenario range (shaded).
Breakeven and actual payroll growth
Thousands of jobs a month on both measures: bars, the three-month average; line, the twelve-month average; dashed, the breakeven; shaded, its scenario range.
Source: BLS, Current Employment Statistics and Current Population Survey; Macro Review calculations · data to Oct 2026labour.breakeven-1

Monthly payroll change and its three-month average, total nonfarm

Thousands of jobs, seasonally adjusted, Oct 2025 to Oct 2026 by default: the one-month change and the three-month average; dashed, the ≈47k breakeven.
About
About this exhibit
Total nonfarm payrolls, thousands of jobs, seasonally adjusted, over the last year by default: the one-month change and the three-month average, with the breakeven dashed; an industry picked in the table below redraws it.
Total nonfarm: monthly change and averages
Thousands of jobs a month, seasonally adjusted; dashed: the breakeven. Select an industry in the table below to chart it.
Source: BLS, Current Employment Statistics (CES0000000001); Macro Review calculations · data to Oct 2026labour.payrolls-1

Monthly change in total nonfarm by sector

Thousands of jobs, seasonally adjusted, monthly, Nov 2023 to Oct 2026 by default: bars, the eight sector groups, adding to the line, total nonfarm.
About this exhibit
Monthly change in total nonfarm payrolls, thousands, seasonally adjusted, last 36 months by default: stacked bars for the sector groups and a line for the total.
Monthly change in total nonfarm by sector
Thousands of jobs, seasonally adjusted; the sector bars add to the total. Hover for values; the industry table below gives the same data.
Source: BLS, Current Employment Statistics; Macro Review calculations · data to Oct 2026labour.payrolls-2

Industries adding jobs, and industries cutting

Which industries are adding jobs, and which are cutting?

Payroll employment by industry, Oct 2026

Level in thousands; changes in thousands of jobs; shading against each series' own 2015–19 distribution (full at 2σ). Select a row to chart it above.
SeriesLevel1m chg3m avg6m avg12m chgYoYShare1m change, 24 monthsAs of
Headline
Total nonfarm159,293+165k+137k+83k+885k+0.6%100.0%Oct 2026
Total private135,961+159k+126k+82k+943k+0.7%85.4%Oct 2026
Goods-producing21,653+15k+27k+22k+201k+0.9%13.6%Oct 2026
Private service-providing114,308+144k+99k+60k+742k+0.7%71.8%Oct 2026
Government23,332+6k+11k±0k−58k−0.2%14.6%Oct 2026
Goods-producing
Mining and logging605−1k±0k−1k−1k−0.2%0.4%Oct 2026
Construction8,367−3k+10k+9k+124k+1.5%5.3%Oct 2026
Manufacturing12,681+19k+18k+14k+78k+0.6%8.0%Oct 2026
Durable goodsmanufacturing7,928+25k+17k+17k+123k+1.6%5.0%Oct 2026
Nondurable goodsmanufacturing4,753−6k±0k−3k−45k−0.9%3.0%Oct 2026
Private services
Trade, transportation and utilities28,818+22k+22k+17k+104k+0.4%18.1%Oct 2026
Wholesale trade6,068.8−9k+1k+3k+14k+0.2%3.8%Oct 2026
Source: BLS, Current Employment Statistics, seasonally adjusted · data to Oct 2026labour.payrolls-3

How broad the gains are

Are the gains broad, or concentrated in a few industries? The private counts the industries adding jobs

Share of private industries adding jobs over one, three and six months

% of private industries adding jobs plus half unchanged (diffusion index), seasonally adjusted, monthly, Oct 2016 to Oct 2026; dashed: the three-month span's 2015–19 average.
About
About this exhibit
BLS private diffusion index (% of the private industries adding jobs, plus half unchanged), seasonally adjusted, monthly, last ten years by default, over 1-, 3- and 6-month spans, with the 50 line and the 3-month span's 2015–19 average.
Private-sector diffusion indexes
Percent of about 250 private industries adding jobs (plus half unchanged), seasonally adjusted. 50 = as many adding as cutting.
Source: BLS, Current Employment Statistics (CES0500000021, -22, -23) · data to Oct 2026labour.payrolls-4

Six-month change in payroll jobs by sector group

Six-month change in employment to Oct 2026, thousands, by sector group (adds to total nonfarm); bars scale with each change, the largest at full width.
Sector group6m changeShare of the move
Education & health+199k
Professional & business+144k
Goods-producing+134k
Trade, transport & utilities+103k
Other services+16k
Government+1k
Leisure & hospitality−8k
Information & financial−94k
Total nonfarm+495k
About this exhibit
Table with bars: the change in employment over the latest six months, thousands, for the sector groups and total nonfarm; the bars scale to the largest gain.
Where the six-month gain came from
Change in employment over the six months to Oct 2026, thousands, by sector group (adds to total nonfarm)
Source: BLS, Current Employment Statistics; Macro Review calculations · data to Oct 2026labour.payrolls-5

How much the first print moves

How much does the first print move?

Revisions to first payroll estimates and their twelve-month mean

Thousands of jobs, last monthly estimate minus first, monthly, Oct 2021 to Oct 2026; line: the twelve-month mean; dashed: the average since 2003.
About this exhibit
Third minus first estimate of each month's payroll change, thousands, monthly, last five years by default, with its twelve-month mean and the average since 2003 (dashed).
Third minus first estimate
Thousands of jobs: each month's revision and its 12-month mean; dashed: the average since 2003
Source: BLS, CES revisions; Macro Review calculations · data to Oct 2026labour.payrolls-7

Breakeven inputs

Every number the estimate uses, dated and sourced

Inputs to the breakeven payroll estimate, dated and with BLS codes

Values as published to Oct 2026, in thousands or % (population-consistent where stated); series identifiers are BLS codes
InputValueMonthSource
Payroll employment (CES total nonfarm, SA)159,293kOct 2026CES0000000001
Population 16+ (CPS, population-consistent)275,682kOct 2026LNU00000000
Population 16+ 11 months earlier (12-month window; Oct 2025 not collected)274,483kNov 2025BLS experimental series
Population 16+ 24 months earlier272,403kOct 2024BLS experimental series
Population 16+ 36 months earlier269,252kOct 2023BLS experimental series
Participation rate (population-consistent)61.9%Oct 2026LNS11300000
Participation rate 24 months earlier (population-consistent)62.1%Oct 2024BLS experimental series
Participation rate 16–1936.7%Oct 2026LNS11300012
Participation rate 20–2471.9%Oct 2026LNS11300036
Participation rate 25–5483.7%Oct 2026LNS11300060
Participation rate 55+37.2%Oct 2026LNS11324230
Population share 16–196.3%Oct 2026LNU00000012
Population share 20–248.0%Oct 2026LNU00000036
Population share 25–5447.3%Oct 2026LNU00000060
Population share 55+38.4%Oct 2026LNU00024230
Labour force (CPS, SA)170,683kOct 2026LNS11000000
Household employment (CPS, SA)163,615kOct 2026LNS12000000
Payroll jobs per employed person0.974Oct 2026calculated by the site: CES ÷ CPS
Unemployment rate4.1%Oct 2026LNS14000000
About this exhibit
Table of every input to the breakeven estimate: payroll employment, population 16+ now and 12, 24 and 36 months earlier, participation rates overall and by age, population shares by age, labour force, household employment, payroll jobs per employed person and the unemployment rate, each dated with its BLS code.
Model inputs
Values as published (population-consistent where stated); series identifiers are BLS codes
Source: BLS, Current Employment Statistics, Current Population Survey and experimental series.labour.breakeven-4

Revisions in full

Every estimate of the last two years, and BLS's annual revision statistics

Each month's payroll change as first published and as revised

Over-the-month change as first published and at each revision, thousands, seasonally adjusted, Nov 2024 to Oct 2026
MonthEstimate of the change, thousandsRevisionToday's series
1st2nd3rd2nd − 1st3rd − 2nd3rd − 1st
Oct 2026+165–––––+165
Sep 2026+29+60–+31––+60
Aug 2026+162+133+186−29+53+24+186
Jul 2026−23+21−10+44−31+13−10
Jun 2026+57+20+31−37+11−26+31
May 2026+172+129+63−43−66−109+63
Apr 2026+115+179+148+64−31+33+148
Mar 2026+178+185+214+7+29+36+214
Feb 2026−92−133−156−41−23−64−156
Jan 2026+130+126+160−4+34+30+160
Dec 2025+50+48−17−2−65−67−17
Nov 2025+64+56+41−8−15−23+41
Oct 2025n.p.−105 B−173n.p.−68n.p.−140
Sep 2025+119n.p.+108n.p.n.p.−11+76
Aug 2025+22−4−26−26−22−48−70
Jul 2025+73+79+72+6−7−1+64
Jun 2025+147+14−13−133−27−160−20
May 2025+139+144+19+5−125−120+13
Apr 2025+177+147+158−30+11−19+108
Mar 2025+228+185+120−43−65−108+67
Feb 2025+151+117+102−34−15−49+42
Jan 2025+143+125+111−18−14−32−48
Dec 2024+256+307+323+51+16+67+237
Nov 2024+227+212+261−15+49+34+134

n.p. = not published. A: data unavailable due to the 2025 lapse in appropriations; B: first published estimate for October 2025. “Today's series” is the change in the current estimates, after the February 2026 benchmark and seasonal-factor revisions, so it can differ from the third estimate for months before the benchmark.

About this exhibit
Each month's payroll change as first, second and third published, and the revisions between them, thousands, seasonally adjusted, last 24 months.
Revisions triangle, total nonfarm
Over-the-month change as first, second and third published, thousands, seasonally adjusted (last 24 months)
Source: BLS, Nonfarm payroll employment: revisions between over-the-month estimates, 1979–present.labour.payrolls-6

Mean and mean absolute revisions by year

Thousands, seasonally adjusted, as BLS publishes them
YearMean revisionMean absolute
2nd − 1st3rd − 1st2nd − 1st3rd − 1st
2025−28−583158
2024−16−203348
2023−2−303751
2022+18−62128
2021+76+159100181
2020−2−6079130
2019+3+81934
2018+8+123034
2017+2−52031
2016+3−31819
2015−7−42527
2014+21+372240
1979–2003−3+144861
May 2003–now+7+73351
Source: BLS, CES revisions.labour.payrolls-8

Appendix

Survey, revisions, aggregation and the 2025 shutdown44 checks, all pass
About this page

The establishment survey counts payroll jobs by industry each month and revises each month twice before holding it until the annual benchmark. This page reads the pace of hiring on one-, three- and six-month windows, where the jobs are, how broad the gains are across industries, and how much the first print tends to move.

Breakeven payroll growth

B=Ep[(1+gN)(1+gp)(1+gr)−1]B = E_p \left[(1+g_N)(1+g_p)(1+g_r) - 1\right]
Δu12≈−ΔP12−12 Br⋅LF×100\Delta u_{12} \approx -\frac{\Delta P_{12} - 12\,B}{r \cdot LF} \times 100
SymbolMeaningFromAssumed or estimated
BBpayroll jobs a month that hold the unemployment rate constant––
EpE_ppayroll employment, total nonfarmBLS CES–
gNg_Nmonthly growth of the population 16+ (population-consistent)BLS CPS–
gpg_pmonthly growth of the participation rate from ageing aloneBLS CPS, age-specific rates–
grg_rmonthly growth of payroll jobs per employed personBLS CES and CPS–
rr, LFLFpayroll jobs per employed person and the labour force, for the implied change in unemploymentBLS CES and CPS–

Definitions and caveats

  • Survey
    Jobs, not people.

    The Current Employment Statistics survey samples about 119,000 businesses and government agencies representing about 622,000 worksites (BLS Technical Note). Employment counts jobs, not people: a person with two payroll jobs is counted twice. The reference period is the pay period that includes the 12th of the month.

  • Revisions
    Three estimates, then the benchmark.

    Each month is estimated three times: the first preliminary estimate, a second with the next release, and a final sample-based estimate with the one after. The series is then held until the annual benchmark to March unemployment-insurance tax counts (published with January data each February), which revises the not-seasonally-adjusted level back 21 months and seasonally adjusted data back five years. The revisions triangle shows the three sample-based estimates as first published; benchmark revisions are not in it.

  • Aggregation
    Components add to totals.

    Seasonally adjusted totals are sums of seasonally adjusted components (identities checked above); a few detailed series are adjusted independently and do not add. Supersector levels are published in whole thousands, detail to one decimal.

  • Diffusion
    Breadth across 250 industries.

    BLS's diffusion index: the percent of about 250 private industries with rising employment plus half of those unchanged, over 1-, 3- and 6-month spans (seasonally adjusted).

  • Shutdown
    October 2025.

    The October–November 2025 lapse in appropriations delayed the October 2025 establishment survey; October was first published with the November release, so it has no first estimate and September 2025 no second estimate (BLS footnote A). The employment series itself is complete.

  • Norms
    Own history, 2015–19.

    Shading compares each change with its own distribution over 2015-01 to 2019-12; percentiles run from January 1990. Sector bars are the monthly change of each group, seasonally adjusted, in thousands; they add to the change in total nonfarm exactly.

Source: BLS Handbook of Methods, Current Employment Statistics; BLS CES revisions.labour.payrolls-9

Implied and actual year-ahead change in unemployment, breakeven rule

Percentage points, Jan 2011 to Oct 2026, dated at the start of each twelve-month window: actual; implied by the history rule; implied with population-consistent inputs.
About
About this exhibit
Implied against actual change in the unemployment rate over the next 12 months, percentage points, dated at the start of each window: actual; implied by the history rule; implied with population-consistent (revised) inputs.
Implied and actual change in unemployment over the next 12 months
Percentage points, dated at the start of each 12-month window
Source: BLS, Current Employment Statistics and Current Population Survey; Macro Review calculations.labour.breakeven-2

Test statistics

12-month changes in the unemployment rate, percentage points; windows grouped by end month
Windows endingnRMSEMean errorCorr.Mean |actual|
Jan 2012 – Dec 2019960.46−0.340.470.66
Jan 2023 – Aug 2026431.08−0.680.400.26
Recent windows, population-consistent inputsrevised data, not available at the time430.93−0.430.49–

Mean |actual| is the average size of the 12-month change being called; an RMSE above it means the rule adds little in that period. Errors come from cyclical participation, the household–payroll divergence and population revisions.

Source: Macro Review calculations.labour.breakeven-3

Model and central estimate

The model's arithmetic with every input dated; the inputs are in the Detail fold
About
B = E_p × [(1 + g_N)(1 + g_p)(1 + g_r) − 1]
  E_p   payroll employment                159,293k  (Oct 2026)
  g_N   population 16+, last 12 months      0.48% a year  (0.0396% a month)
  g_p   participation, ageing effect         −0.12% a year  (-0.0103% a month)
  g_r   payroll jobs per employed person     held flat (sensitivity ±33k)
  B  =  47k jobs a month
  scenario range 42k to 104k · ±98k (1 out-of-sample RMSE)
  history rule, same month: 53k
  • Model
    An identity, not a forecast.

    B = E_p × [(1 + g_N)(1 + g_p)(1 + g_r) − 1]. E_p: payroll employment. g_N: compound monthly growth of the civilian noninstitutional population 16+. g_p: monthly growth of the participation rate; central = the ageing effect, Σ p_a·Δs_a ÷ Σ p_a·s_a with age-specific rates p_a (16–19, 20–24, 25–54, 55+) held at today's values and population shares s_a moving as over the last 24 months. g_r: trend in payroll jobs per employed person (central: zero). Holding the unemployment rate constant requires household employment to grow with the labour force, N·p; payroll jobs follow through r.

Source: BLS; Macro Review calculations.labour.breakeven-5

Sensitivity

Breakeven, thousands of jobs a month: population growth window × participation assumption. Central case outlined.
Population growthageing only (24m)−0.12%/yrflat±0.00%/yrobserved trend (24m)−0.16%/yr
Last 12 months0.48% a year47k63k42k
Last 24 months0.60% a year63k79k58k
Last 36 months0.79% a year88k104k83k

Payroll ratio trending ±0.25pp a year: ±33k. At its own 24-month trend (−0.38% a year) breakeven would be -3k.

Source: This model.labour.breakeven-6

Definitions and caveats

  • Inputs
    Population-consistent.

    Central inputs use BLS's population-consistent series (experimental series through December 2025, which accounts for the January 2026 population controls, and official estimates from January 2026), so population growth is not distorted by the January step. Trailing windows end in the latest month.

  • History
    Official series, January breaks removed.

    The historical line uses official series throughout with a documented rule: January changes (when new population controls enter) are replaced by the mean of the other monthly changes in the window; October 2025 (not collected) is skipped the same way. Population window 12 months; ageing effect over 24 months.

  • Scenarios
    Population × participation.

    Scenario grid: population growth over 12, 24 and 36 months (the slowdown in net international migration shows up in the shorter windows) × participation: ageing effect only, flat, or the observed population-consistent trend over 24 months (which includes cyclical moves). CBO's demographic outlook could not be retrieved from this environment (cbo.gov refused access), so no external immigration scenario is used.

  • Test
    Out of sample, revised data.

    Out of sample: at each month t the history rule uses only data to t; the change in payrolls over the next 12 months then implies Δu = −(ΔP − 12·B_t) ÷ (r_t·LF_t) × 100, compared with the actual change. Test periods refer to the end of each window (2012–2019 and 2023–2026), so no window starts in 2020–22, the pandemic and its recovery. The chart dates each pair by the start of its window. The test uses today's revised data, not the data available at the time.

  • Band
    Two measures of uncertainty.

    Uncertainty band: the scenario range, and ±1 out-of-sample RMSE of the 12-month unemployment-rate change converted to thousands of jobs a month at today's labour force.

  • Limits
    What the identity leaves out.

    Population growth is the input most exposed to revision: official CPS population lagged the 2022–24 immigration surge until the January 2024 and 2025 controls. An accounting identity, not a behavioural model: it takes participation and population trends as given, ignores cyclical participation (which rises when hiring is strong) and treats the payroll–household ratio as stable. Population estimates are revised each year.

  • Data gap
    Oct 2025.

    not collected; skipped in growth windows (history rule) and in the out-of-sample pairs that need it

Source: This model documentation.labour.breakeven-7

Before you read

Macro Review is a personal research site, still being built.

Nothing on it is investment advice, a recommendation or a solicitation.

Check every figure for yourself. Each names its publisher — confirm it there before you rely on it.

Figures can be revised, delayed or wrong.

Cones, fans, scenarios, fair values and “what followed” tables are conditional on the model each page names: they describe, they do not predict.

No warranty is given; use is at your own risk.