Miserably Unemployed
Looking for AI chat?
Pulse

Negative real wages. Hidden slack. The wrong jobs

Wage growth is at 3.4% while CPI inflation is at 3.8%. In what world is a negative real wage a sign of a healthy labor market? A "tight" 4.3% unemployment rate doesn't mean much when the long-term unemployed numbers increased by over 500k this year, 4.8 million are underemployed, and another 6.2 million have stopped looking for work.

We have an automated, high-inflation corporate state that is systematically liquidating high-wage private knowledge roles, leaving workers to choose between a bloated local bureaucracy, a low-wage hospitality gig, or long-term erasure.

Real wage growth
Hidden slack (MATL + PTER)
U-6
Knowledge-work net (12mo)
Movement 1

Prices outran wages

If your paycheck grew 3% last year but everything you buy got 4% more expensive, you got poorer even though the number on your paycheck got bigger. The left chart is the percentage change side by side; the right chart shows the same thing in dollar terms, where both lines start at the same point and the gap that opens between them is how much purchasing power you lost.

Sources: BLS Average Hourly Earnings (CES0500000002, SA) and CPI-U (CUUR0000SA0, NSA). Year-over-year percent changes computed at render time. Real wage growth uses the Fisher identity (1+w)/(1+p) − 1, not the linear approximation w − p.

Movement 2

What the unemployment rate leaves out

The unemployment rate you hear on the news only counts people who are actively applying for jobs right now. It leaves out the person who gave up looking last month, the person who checks indeed.com once a quarter but didn't this month, and the person who wants a full-time job but can only get part-time hours. The government publishes four wider versions of the number (U-3 is the headline, then U-4, U-5, U-6) that each add one more of these left-out groups. The gap between U-6 and U-3 is how many people the headline doesn't see. When that gap grows, the labor market is looser than the headline says it is, even if the headline holds steady.

Sources: BLS Table A-15. U-3 LNS14000000, U-4 LNS13327707, U-5 LNS13327708, U-6 LNS13327709 (all SA, percent). The four measures are strictly nested (U-3 ⊂ U-4 ⊂ U-5 ⊂ U-6) so the population counts in the table below can be read without double-counting.

Movement 3

Where the jobs went

Total US employment went up over the last twelve months, but the growth wasn't spread evenly. Almost every net new job came from three sectors: government, education and health, and leisure and hospitality (restaurants, hotels). The sectors that historically hold the high-paying knowledge jobs (Information, and Professional and business services) added almost nothing or shrank. The total number gets bigger; the composition gets worse.

Source: BLS Current Employment Statistics by NAICS supersector (CES<supersector>01, SA, thousands). The bar list ranks each supersector by its net 12-month change in employees. CES is an establishment-payroll survey and excludes self- employed and gig workers, so a knowledge worker leaving Information for freelance work shows up here as a job loss in Information even though they're still earning.

Movement 4

Is this good for the economy?

When the news says "the economy added 200,000 jobs", it's counting heads, not paychecks. A $40,000-a-year waitstaff job and a $140,000-a-year software job count the same. But if the country trades a software job for a waitstaff job, total wages paid in the economy go down even though the jobs number stays flat. The bar below shows the real dollar swing per industry: how much more (or less) the country paid that sector in the most recent quarter compared with a year earlier, annualised. It tells the real story: not just how many jobs we have, but how much they pay.

BLS QCEW total quarterly wages by NAICS supersector (administrative UI-tax data, covers private + all government). For each sector: latest published quarter minus the same quarter one year earlier, multiplied by 4 to annualise. Includes every supersector with a clean QCEW counterpart; "Government" in this view is QCEW Public administration (NAICS 1028), a narrower slice than the CES Government total used in Movement 3.

Methodology

Numbers cited in the hero (as of latest BLS release)

The hero prose is editorial and the exact numbers cited (3.4% wage growth, 3.8% CPI, 500k long-term unemployed, 4.8M underemployed, 6.2M out of the labor force) are approximate as of the article's framing. Mapped to BLS series:

  • 3.4% wage growth CES0500000003 total private average hourly earnings YoY% (production and nonsupervisory employees). Differs from ECI (Employment Cost Index, quarterly, includes total compensation and controls for employment-mix shifts) by roughly 0.5pp.
  • 3.8% CPI CUUR0000SA0 (CPI-U, all urban consumers, all items, NSA) YoY%. Distinct from Core CPI (excludes food + energy, typically lower) and PCE (Bureau of Economic Analysis, the Fed's preferred gauge, typically 0.3 to 0.6 pp below CPI-U).
  • 500k LTU LNS13008636 (unemployed 27 weeks and over, level in thousands), year-over-year change. Direction correct; historical context: this series peaked near 7.0M in 2010, so current ~1.6M is low by post-GFC standards even as YoY change is sharply up.
  • 4.8M underemployed LNS12032194 (employed part-time for economic reasons, SA, thousands). The cited number is editorial; the live tile shows whatever the latest BLS print is.
  • 6.2M out of labor force No single BLS series gives 6.2M directly. Closest official: marginally attached (LNU05026642, ~1.5M, NSA) or not-in-labor-force (~100M+). The cited figure is editorial framing for the broader "want work but not searching" population.

Prices (Movement 1)

The YoY% chart computes the per-period year-over-year percent change in CES total-private average hourly earnings (CES0500000002, $/hour, SA) and in CPI-U (CUUR0000SA0, index, NSA), each at render time from the level series. The "in real dollars" companion chart deflates nominal hourly earnings to the earliest period in the window using the CPI-U ratio real = nominal × CPI_base / CPI_period, anchoring both lines at the baseline value so the divergence reads as cumulative purchasing power lost. The hero's "real wage growth" tile uses the Fisher identity (1 + w) / (1 + p) − 1, not the linear approximation w − p; the two agree to within 0.01pp at current magnitudes but the exact form is what's displayed.

Slack (Movement 2)

U-3 through U-6 are BLS's official alternative measures of labor underutilisation (Table A-15), strictly nested by definition (U-1 ⊂ U-2 ⊂ U-3 ⊂ U-4 ⊂ U-5 ⊂ U-6). The "hidden slack" hero tile is the BLS-standard gap between U-6 and U-3 in headcount terms: marginally attached (LNU05026642) + part-time for economic reasons (LNS12032194). The two are disjoint (MATL is people OUT of the labor force who want a job; PTER is people IN the labor force working fewer hours than they want), so they sum cleanly. We deliberately do NOT add long-term unemployed or discouraged workers: long-term unemployed (LNS13008636) is already inside U-3 (they are unemployed and actively looking for 27+ weeks); discouraged workers (LNU05026645) are a strict subset of marginally attached (the subset whose reason for not looking is job-market related).

Composition (Movement 3)

CES (Current Employment Statistics, ~119,000 establishments monthly) is an establishment-based payroll survey. It covers nonfarm wage-and-salary employees and excludes self-employed, unpaid family workers, agricultural workers, household workers, and military. Knowledge workers leaving an Information payroll for independent contracting show as a job loss in Information but are still earning; this caveat attenuates (does not invalidate) the "knowledge-work liquidation" reading. CES values are also a NET concept: a +180k Government number reflects gross hires minus gross separations; the underlying churn is much larger and visible in JOLTS by industry. The diverging bar list's colour buckets (knowledge / government / services / goods) are editorial framings, not industry-standard groupings. Information and Professional & business services in particular are heterogeneous NAICS sectors.

Wage bill (Movement 4)

BLS QCEW (Quarterly Census of Employment and Wages), an administrative dataset covering ~96% of US employment via state UI tax records. QCEW publishes total wages paid per NAICS supersector per quarter at the US national level (own_code 0 Total Covered, agglvl_code 13 US Total by NAICS Sector, industry_code 1011-1028 for the eleven CES-aligned supersectors). For each sector we take (latest quarter total wages - same quarter one year earlier) × 4 to get an annualised wage-bill swing in dollars; the diverging bar plots that swing per sector, the headline above sums them into a net.

Notes. QCEW is unadjusted (no seasonal smoothing) and publishes with a ~5-6 month lag, so the freshest quarter on display here typically trails the CES composition view in Movement 3 by two quarters. "Government" in this section refers specifically to QCEW Public administration (NAICS 1028), a narrower concept than the CES ces-government series (which classifies state/local government workers in education, healthcare, etc. under those industries instead). The CES and QCEW totals differ by 1-3% in any given period because of definitional scope differences, but track each other tightly.

Series IDs

  • CUUR0000SA0 CPI-U, all urban consumers, all items, NSA index (1982-84=100)
  • CES0500000002 Total private average hourly earnings, $/hour, SA
  • LNS14000000 U-3 unemployment rate, SA, percent
  • LNS13327707 U-4 (U-3 + discouraged), SA, percent
  • LNS13327708 U-5 (U-4 + other marginally attached), SA, percent
  • LNS13327709 U-6 (U-5 + part-time for economic reasons), SA, percent
  • LNS13008636 Unemployed 27 weeks and over (long-term unemployed), SA, thousands
  • LNU05026645 Discouraged workers (strict subset of marginally attached), NSA, thousands
  • LNU05026642 Total marginally attached to the labor force, NSA, thousands
  • LNS12032194 Employed part-time for economic reasons (PTER), SA, thousands
  • CES0000000001 Total nonfarm employment, SA, thousands
  • CES<supersector>01 CES supersector employment, SA, thousands (Information 5000000, Financial 5500000, PBS 6000000, Education & health 6500000, Leisure & hospitality 7000000, Government 9000000, Mining & logging 1021000, Construction 2000000, Manufacturing 3000000, Trade-transport-utilities 4000000, Other services 8000000)
  • QCEW US000 own=0 agglvl=13 industry=1011..1028 Total quarterly wages by NAICS supersector at the US national level, sourced from data.bls.gov/cew/data/api/<YYYY>/<Q>/area/US000.csv

BLS Employment Situation (the monthly jobs report, carries CES, U-1..U-6, and the alternative measures) releases first Friday of each month at 8:30 ET; CPI releases mid-month. Both are subject to substantial revision: CES annual benchmark revisions can move headline employment by ±150k jobs; CPI is largely revision-free except for seasonal-factor updates each February. All series are mixed SA / NSA (CPI-U we use is NSA; CES AHE is SA); year-over-year comparisons are insensitive to SA choice since the seasonal cancels. JOLTS response rates have declined from ~70% in 2014 to ~33% in 2024 (BLS, Issues in JOLTS Methodology, Aug 2024), so JOLTS-derived series carry wider standard errors than headline reads.