In a market with frayed nerves, employment deserves a calm reading. The number of people working matters; so does the purchasing power of their pay. October’s claims evidence points to stability rather than a sudden deterioration. But assessing the monetary side requires particular care because the 2025 shutdown left gaps in the inflation comparison base.
For the October–December 2026 reporting months, AlphaTraderNews will use the headline Personal Consumption Expenditures (PCE) price index as its principal deflator for monetary employment analysis. The purpose is a consistent, explicitly identified purchasing-power measure across these three reports. This is our analytical choice, not a change to the official BLS real-earnings series.
Employment: stability beneath the noise
The Department of Labor’s November 5 release verifies the following comparison. These are seasonally adjusted (SA) weekly claims, not BLS monthly unemployment estimates. Figures are in thousands.
| Measure | September 26 | October 24 | Change |
|---|---|---|---|
| Continuing claims | 1,714 | 1,718 | +4 (+0.23%) |
| Initial claims, four-week average | 200.50 | 197.00 | −3.50 (−1.75%) |
| Continuing claims, four-week average | 1,710.50 | 1,709.50 | −1.00 (−0.06%) |
Continuing claims were essentially flat across the month; the initial-claims average improved. The latest observation strengthens the low-layoff picture: initial claims fell to 189,000 in the week ending October 31, with a 194,750 four-week average. Continuing claims nevertheless rose 23,000 in the latest available week. Stability across a month does not mean every week improved.
On the same revised vintage, continuing claims were 1,900,000 on December 27, 2025. Against that last-2025-week baseline, the decline was 186,000 by September 26 and 182,000 by October 24. These are point-to-point YTD changes, not YTD averages. Claims cover insured unemployment; they cannot alone establish the national unemployment rate.
Where demand remains
The more revealing vacancy comparison uses not seasonally adjusted (NSA) September data throughout: BLS JOLTS openings divided by CPS unemployment. September is the reference month, not a September 26 weekly reading.
| Sector | Openings, NSA | Unemployed, NSA | Ratio |
|---|---|---|---|
| Total nonfarm / all unemployed | 7,531 | 6,859 | 1.10 |
| Total private | 6,687 | 5,225 | 1.28 |
| Manufacturing | 594 | 423 | 1.40 |
| Durable goods | 407 | 256 | 1.59 |
| Financial activities | 365 | 213 | 1.71 |
| Professional and business services | 1,358 | 753 | 1.80 |
Calculations: AlphaTraderNews, from preliminary September JOLTS and CPS Table A-14. Industry unemployment classifies people by their previous industry; the total-private denominator is nonagricultural private wage-and-salary workers. These cross-survey ratios indicate hiring demand relative to an available-worker proxy. They are not a literal matching system: skills, location and movement between industries matter.
Declaring AI “dead” overlooks evidence of demand
A selloff does not establish that a technology’s adoption has stopped. CompTIA’s September analysis recorded 349,821 active postings requiring AI skills, up 32,327 from August, and 13,636 postings in dedicated AI/machine-learning job titles. Yet its estimate of technology-sector employment fell 10,350. Specialist demand and job losses can coexist.
Logistics and life sciences also need skilled people. BLS’s 2025–35 projections show employment growth of 17.6% for logisticians and 12.6% for medical scientists excluding epidemiologists. These are long-term projections, not current vacancies or proof that every role is AI-driven. The defensible conclusion is that technological change reallocates demand; broad declarations of technological death miss that distinction.
Why use PCE for three months?
The shutdown prevented collection of October 2025 CPI survey prices. Consequently, a conventional October 2026-over-October 2025 all-items CPI comparison lacks its normal published survey base. A reconstructed October index must be identified as an estimate, not presented as an official BLS observation. November and December comparisons also warrant attention to the shutdown’s collection and imputation effects.
PCE supplies a continuous published series with which to maintain our monetary employment analysis. It measures a broader consumption basket, including spending on households’ behalf, whereas CPI reflects a different population, weighting system and scope. PCE therefore answers a related purchasing-power question, rather than reproducing CPI.
There is an essential qualification: PCE was not insulated from the missing CPI prices. BEA says it replaced missing October CPI inputs using the geometric mean of September and November seasonally adjusted indexes, and derived unadjusted estimates using October 2024 seasonal factors. The advantage is a documented, continuous estimation framework—not an unaffected alternative dataset.
We will apply PCE consistently to October, November and December employment reporting months, identify its vintage and adjustment basis, and review the approach after December data become available. Official CPI-based real earnings will remain separately identified wherever available.
What changes in the calculation
Real change = [(1 + nominal change) ÷ (1 + price change) − 1] × 100, with input changes expressed as decimals. Weekly earnings equal hourly earnings multiplied by weekly hours. Aggregate payroll purchasing power additionally depends on employment; a higher hourly wage alone cannot establish either result.
| Assumption or calculation | Result |
|---|---|
| Hourly earnings assumption | +0.20% MoM |
| Weekly hours assumption | Unchanged |
| Headline PCE price assumption | +0.13% MoM |
| Calculated real weekly earnings | +0.070% MoM |
| Employment assumption | Unchanged |
| Calculated aggregate real payroll | +0.070% MoM |
This scenario supports a modest monthly improvement, conditional on hours and headcount. It does not establish positive Y/Y purchasing power. At an assumed 3.62% Y/Y PCE inflation rate, nominal weekly earnings must rise more than 3.62% Y/Y to deliver a real gain. A hypothetical 3.80% nominal gain yields +0.174% real growth; against an assumed 3.969% inflation rate it yields −0.163%. Those are sensitivity calculations, not official October CPI or PCE forecasts.
Our reporting will compare MoM, three-month, YTD and Y/Y changes using compatible earnings and deflator series. NSA comparisons will remain separate from SA headlines. YTD means December 2025 to the reporting month; annualized YTD uses [(current/December)12/m − 1] × 100, where m is elapsed months. Three-month annualization uses the fourth power of the three-month ratio. Annualized rates are extrapolations, not Y/Y observations.
The market test
Low layoffs and persistent vacancies argue against treating every nervous headline as evidence of collapse. But markets need the full wage bill: employment × weekly hours × hourly pay, adjusted for prices. Durable gains support consumption and company revenues; spending maintained through falling savings offers a thinner cushion. See our September income and spending analysis.
For NQ and QQQ, resilient purchasing power can support earnings while sticky inflation can keep yields restrictive. The three-month PCE approach makes our measurement explicit. It is a framework for testing resilience with numbers, not a guarantee of a positive result.
Sources
- Department of Labor: November 5 weekly claims release and historical table
- BLS JOLTS: September NSA job openings, Table 8
- BLS CPS: September NSA unemployment by industry, Table A-14
- BLS: shutdown effects on CPI collection and estimation
- BEA: October–November 2025 technical notes on missing price inputs
- CompTIA: September technology hiring and AI-skills postings
- BLS: 2025–35 occupational employment projections