Research
The labor market is sturdy the way a parked car is sturdy
Sturdy is not the same as strong
The July jobs report that landed this week was dressed for a war. Employers posted 7.27 million openings, a tick above June, and the unemployment rate sat at 4.1 percent. Heather Long of Navy Federal Credit Union called it “low fire, low hire.” That is a polite way of saying companies are not recruiting and not dumping people either — a description that, in a week when Brent crude settled at $94.65 and regular gasoline was about $4.10 a gallon, sounds almost like good news. It is not a boom. It is the absence of a bust.
The openings number comes from the Job Openings and Labor Turnover Survey, which this research catalogue cannot reproduce for 2026. What the Bureau of Labor Statistics payroll survey can show, through July 2026, is the hiring that actually showed up on employers’ books. Seasonally adjusted nonfarm payrolls stood at 158.9 million. They fell by 23,000 in July. The three-month average gain was 20,000 jobs. For the first seven months of 2026, the average monthly gain was 61,000 — matching the round number in this week’s reporting, and a bounce from a 2025 that added only 116,000 jobs across the entire year. That is the official record of “sturdy.”
U.S. payroll growth has downshifted to a crawl
View exact chart values
| month | three_month_avg_change | monthly_change |
|---|---|---|
| 2022-01 | 478 | 190 |
| 2022-02 | 532 | 819 |
| 2022-03 | 500 | 490 |
| 2022-04 | 539 | 308 |
| 2022-05 | 366 | 301 |
| 2022-06 | 348 | 434 |
| 2022-07 | 483 | 714 |
| 2022-08 | 479 | 290 |
| 2022-09 | 408 | 220 |
| 2022-10 | 289 | 357 |
| 2022-11 | 293 | 303 |
| 2022-12 | 253 | 100 |
| 2023-01 | 279 | 434 |
| 2023-02 | 275 | 290 |
| 2023-03 | 264 | 68 |
| 2023-04 | 200 | 241 |
| 2023-05 | 196 | 280 |
| 2023-06 | 249 | 225 |
| 2023-07 | 223 | 163 |
| 2023-08 | 202 | 218 |
| 2023-09 | 179 | 156 |
| 2023-10 | 178 | 159 |
| 2023-11 | 147 | 127 |
| 2023-12 | 147 | 154 |
| 2024-01 | 152 | 175 |
| 2024-02 | 178 | 206 |
| 2024-03 | 203 | 228 |
| 2024-04 | 166 | 64 |
| 2024-05 | 123 | 78 |
| 2024-06 | 76 | 87 |
| 2024-07 | 73 | 53 |
| 2024-08 | 50 | 9 |
| 2024-09 | 72 | 155 |
| 2024-10 | 66 | 33 |
| 2024-11 | 107 | 134 |
| 2024-12 | 135 | 237 |
| 2025-01 | 108 | -48 |
| 2025-02 | 77 | 42 |
| 2025-03 | 20 | 67 |
| 2025-04 | 72 | 108 |
| 2025-05 | 63 | 13 |
| 2025-06 | 34 | -20 |
| 2025-07 | 19 | 64 |
| 2025-08 | -9 | -70 |
| 2025-09 | 23 | 76 |
| 2025-10 | -45 | -140 |
| 2025-11 | -8 | 41 |
| 2025-12 | -39 | -17 |
| 2026-01 | 61 | 160 |
| 2026-02 | -4 | -156 |
| 2026-03 | 73 | 214 |
| 2026-04 | 69 | 148 |
| 2026-05 | 142 | 63 |
| 2026-06 | 77 | 20 |
| 2026-07 | 20 | -23 |
The downshift started years before Hormuz reopened
It is tempting, in a week of missile exchanges, to treat every weak print as a war print. The payroll tape does not cooperate. Average monthly job gains ran about 210,000 in 2023, 122,000 in 2024, and 10,000 in 2025. The 2026 year-to-date average of 61,000 is better than last year’s stall and worse than any pre-pandemic year in this extract. The three-month average, which smooths the month-to-month noise, spent most of 2022 above 250,000, drifted through 2023, cracked below 100,000 in mid-2024, and spent stretches of late 2025 below zero. July’s 20,000 three-month average is not a new shock. It is the latest reading on a long downshift.
Hours tell a similar story of a market that cooled without collapsing. Average weekly hours for all private employees were 34.3 in July 2026, a tenth of an hour above the 34.2 that prevailed through much of 2025 and well below the 35.0 peak in January 2021, when the rebound from lockdowns was still stuffing overtime into a short-staffed week. Manufacturing overtime was 3.1 hours, up a couple of tenths from 2025 but below the 3.5 hours common in 2018. Firms have not sent everyone home. They have also not asked the remaining people to work like it is 2021.
The temp-help buffer is still missing
Temporary-help services are the part of the labor market employers can turn on and off without a press release. They did the off part with enthusiasm. Seasonally adjusted temp-help employment peaked at 3.16 million in March 2022. By July 2026 it was 2.51 million — about 656,000 jobs, or 21 percent, below that peak. Indexed to January 2019, total nonfarm payrolls were up about 6 percent in July 2026. Temp-help jobs were down about 15 percent. The series bounced a little in 2026, adding about 11,000 jobs over the year to July. That is a rounding error on a hole that opened when the post-pandemic hiring binge ended.
If “low hire” has a mascot, this is it. Staffing agencies are where firms experiment with demand. When those jobs stay 21 percent below the last peak while headline payrolls grind out 61,000-a-month gains, the experiment is not expansion. It is caution. Nothing in these figures proves the Iran war caused the hole; the decline was well underway in 2023. What they do show is the buffer that used to absorb a shock is already thin before this week’s oil spike asked it to do anything.
Temporary-help jobs never recaptured the 2022 peak
View exact chart values
| month | temp_help_thousands |
|---|---|
| 2019-01 | 2953.4 |
| 2019-02 | 2940.5 |
| 2019-03 | 2935.6 |
| 2019-04 | 2953.8 |
| 2019-05 | 2960.3 |
| 2019-06 | 2964.6 |
| 2019-07 | 2961.6 |
| 2019-08 | 2974 |
| 2019-09 | 2981.9 |
| 2019-10 | 2963.4 |
| 2019-11 | 2951.1 |
| 2019-12 | 2936.6 |
| 2020-01 | 2917.3 |
| 2020-02 | 2888.1 |
| 2020-03 | 2867.9 |
| 2020-04 | 1950.4 |
| 2020-05 | 2030.4 |
| 2020-06 | 2224.1 |
| 2020-07 | 2361 |
| 2020-08 | 2478.1 |
| 2020-09 | 2498.7 |
| 2020-10 | 2595.2 |
| 2020-11 | 2675.6 |
| 2020-12 | 2749.8 |
| 2021-01 | 2804.9 |
| 2021-02 | 2829.9 |
| 2021-03 | 2868.9 |
| 2021-04 | 2816.8 |
| 2021-05 | 2806.6 |
| 2021-06 | 2856.9 |
| 2021-07 | 2889.3 |
| 2021-08 | 2896.3 |
| 2021-09 | 2906.2 |
| 2021-10 | 3023.4 |
| 2021-11 | 3050.6 |
| 2021-12 | 3079.9 |
| 2022-01 | 3077.4 |
| 2022-02 | 3147 |
| 2022-03 | 3161.4 |
| 2022-04 | 3132.6 |
| 2022-05 | 3113 |
| 2022-06 | 3106.9 |
| 2022-07 | 3095.1 |
| 2022-08 | 3080.8 |
| 2022-09 | 3081.6 |
| 2022-10 | 3070.5 |
| 2022-11 | 3036.7 |
| 2022-12 | 2968.6 |
| 2023-01 | 2967.3 |
| 2023-02 | 2949.3 |
| 2023-03 | 2921 |
| 2023-04 | 2906.8 |
| 2023-05 | 2898.8 |
| 2023-06 | 2867.8 |
| 2023-07 | 2833.9 |
| 2023-08 | 2809.8 |
| 2023-09 | 2787.6 |
| 2023-10 | 2773 |
| 2023-11 | 2734.7 |
| 2023-12 | 2685.8 |
| 2024-01 | 2694.4 |
| 2024-02 | 2676.3 |
| 2024-03 | 2670.3 |
| 2024-04 | 2655.1 |
| 2024-05 | 2641.5 |
| 2024-06 | 2615.6 |
| 2024-07 | 2598.3 |
| 2024-08 | 2582.7 |
| 2024-09 | 2564 |
| 2024-10 | 2547.8 |
| 2024-11 | 2550 |
| 2024-12 | 2541.9 |
| 2025-01 | 2539.3 |
| 2025-02 | 2533 |
| 2025-03 | 2528.3 |
| 2025-04 | 2527.5 |
| 2025-05 | 2513.1 |
| 2025-06 | 2504.6 |
| 2025-07 | 2494.2 |
| 2025-08 | 2483.3 |
| 2025-09 | 2471.6 |
| 2025-10 | 2451.8 |
| 2025-11 | 2465.5 |
| 2025-12 | 2451.4 |
| 2026-01 | 2470.2 |
| 2026-02 | 2471.9 |
| 2026-03 | 2479.9 |
| 2026-04 | 2490.4 |
| 2026-05 | 2490.6 |
| 2026-06 | 2501.6 |
| 2026-07 | 2505 |
Payrolls edged up. The temp-help buffer did not.
View exact chart values
| month | nonfarm_index | temp_help_index |
|---|---|---|
| 2019-01 | 100 | 100 |
| 2019-02 | 100 | 99.56 |
| 2019-03 | 100.16 | 99.4 |
| 2019-04 | 100.35 | 100.01 |
| 2019-05 | 100.37 | 100.23 |
| 2019-06 | 100.52 | 100.38 |
| 2019-07 | 100.58 | 100.28 |
| 2019-08 | 100.74 | 100.7 |
| 2019-09 | 100.87 | 100.96 |
| 2019-10 | 100.93 | 100.34 |
| 2019-11 | 101.07 | 99.92 |
| 2019-12 | 101.16 | 99.43 |
| 2020-01 | 101.31 | 98.78 |
| 2020-02 | 101.49 | 97.79 |
| 2020-03 | 100.56 | 97.11 |
| 2020-04 | 86.92 | 66.04 |
| 2020-05 | 88.66 | 68.75 |
| 2020-06 | 91.74 | 75.31 |
| 2020-07 | 92.8 | 79.94 |
| 2020-08 | 93.84 | 83.91 |
| 2020-09 | 94.48 | 84.6 |
| 2020-10 | 94.94 | 87.87 |
| 2020-11 | 95.12 | 90.59 |
| 2020-12 | 94.99 | 93.11 |
| 2021-01 | 95.2 | 94.97 |
| 2021-02 | 95.55 | 95.82 |
| 2021-03 | 96.12 | 97.14 |
| 2021-04 | 96.35 | 95.37 |
| 2021-05 | 96.67 | 95.03 |
| 2021-06 | 97.17 | 96.73 |
| 2021-07 | 97.8 | 97.83 |
| 2021-08 | 98.17 | 98.07 |
| 2021-09 | 98.47 | 98.4 |
| 2021-10 | 99.01 | 102.37 |
| 2021-11 | 99.45 | 103.29 |
| 2021-12 | 99.84 | 104.28 |
| 2022-01 | 99.96 | 104.2 |
| 2022-02 | 100.51 | 106.56 |
| 2022-03 | 100.84 | 107.04 |
| 2022-04 | 101.04 | 106.07 |
| 2022-05 | 101.24 | 105.4 |
| 2022-06 | 101.53 | 105.2 |
| 2022-07 | 102.01 | 104.8 |
| 2022-08 | 102.2 | 104.31 |
| 2022-09 | 102.35 | 104.34 |
| 2022-10 | 102.58 | 103.96 |
| 2022-11 | 102.79 | 102.82 |
| 2022-12 | 102.85 | 100.51 |
| 2023-01 | 103.14 | 100.47 |
| 2023-02 | 103.34 | 99.86 |
| 2023-03 | 103.38 | 98.9 |
| 2023-04 | 103.54 | 98.42 |
| 2023-05 | 103.73 | 98.15 |
| 2023-06 | 103.88 | 97.1 |
| 2023-07 | 103.99 | 95.95 |
| 2023-08 | 104.13 | 95.14 |
| 2023-09 | 104.24 | 94.39 |
| 2023-10 | 104.34 | 93.89 |
| 2023-11 | 104.43 | 92.59 |
| 2023-12 | 104.53 | 90.94 |
| 2024-01 | 104.65 | 91.23 |
| 2024-02 | 104.78 | 90.62 |
| 2024-03 | 104.94 | 90.41 |
| 2024-04 | 104.98 | 89.9 |
| 2024-05 | 105.03 | 89.44 |
| 2024-06 | 105.09 | 88.56 |
| 2024-07 | 105.12 | 87.98 |
| 2024-08 | 105.13 | 87.45 |
| 2024-09 | 105.23 | 86.82 |
| 2024-10 | 105.25 | 86.27 |
| 2024-11 | 105.34 | 86.34 |
| 2024-12 | 105.5 | 86.07 |
| 2025-01 | 105.47 | 85.98 |
| 2025-02 | 105.5 | 85.77 |
| 2025-03 | 105.54 | 85.61 |
| 2025-04 | 105.61 | 85.58 |
| 2025-05 | 105.62 | 85.09 |
| 2025-06 | 105.61 | 84.8 |
| 2025-07 | 105.65 | 84.45 |
| 2025-08 | 105.61 | 84.08 |
| 2025-09 | 105.66 | 83.69 |
| 2025-10 | 105.56 | 83.02 |
| 2025-11 | 105.59 | 83.48 |
| 2025-12 | 105.58 | 83 |
| 2026-01 | 105.69 | 83.64 |
| 2026-02 | 105.58 | 83.7 |
| 2026-03 | 105.72 | 83.97 |
| 2026-04 | 105.82 | 84.32 |
| 2026-05 | 105.86 | 84.33 |
| 2026-06 | 105.88 | 84.7 |
| 2026-07 | 105.86 | 84.82 |
One sector is doing the hiring. Several are not.
Over the 12 months to July 2026, total nonfarm payrolls rose by 316,000 jobs, or 0.2 percent. Private education and health services added 550,000. That single supersector more than accounted for the entire net gain. Professional and business services added 115,000, leisure and hospitality 83,000, construction 82,000, and other services 49,000. On the other side of the ledger, government shed 315,000 jobs, financial activities 114,000, information 81,000, trade, transportation and utilities 36,000, manufacturing 14,000, and mining and logging 3,000. Oil and gas extraction, a small slice of mining, was down 2,600 jobs even as crude prices climbed. The war is a fuel shock. It is not, on this evidence, an oil-field hiring boom.
The government decline is not an abstraction about “the administrative state.” Federal employment, seasonally adjusted, fell from 2.94 million in July 2025 to 2.68 million in July 2026, a drop of 252,000 jobs, or 8.6 percent. Most of that break arrived in a single month: the series went from 2.91 million in September 2025 to 2.75 million in October. Census, postal and other federal categories sit inside that total, so the drop should not be read as a pure DOGE headcount. It should be read as a labor market in which the federal payroll stopped being a source of job growth and became a source of job loss — while hospitals and clinics kept the national total from going nowhere. Construction’s modest 82,000-job gain over the year is also a reminder that “housing is weak” and “construction employment is falling” are not the same sentence. The sector added jobs. It did not add them like 2022.
Health care carried the jobs market. Government subtracted.
View exact chart values
| industry | jul2025_thousands | jul2026_thousands | change_thousands | pct_change |
|---|---|---|---|---|
| Government | 23585 | 23270 | -315 | -1.34 |
| Financial activities | 9204 | 9090 | -114 | -1.24 |
| Information | 2861 | 2780 | -81 | -2.83 |
| Trade, transportation, utilities | 28744 | 28708 | -36 | -0.13 |
| Manufacturing | 12625 | 12611 | -14 | -0.11 |
| Mining and logging | 610 | 607 | -3 | -0.49 |
| Other services | 5989 | 6038 | 49 | 0.82 |
| Construction | 8261 | 8343 | 82 | 0.99 |
| Leisure and hospitality | 16848 | 16931 | 83 | 0.49 |
| Professional and business services | 22403 | 22518 | 115 | 0.51 |
| Private education and health | 27412 | 27962 | 550 | 2.01 |
Federal payrolls fell sharply after mid-2025
View exact chart values
| month | federal_thousands |
|---|---|
| 2019-01 | 2808 |
| 2019-02 | 2810 |
| 2019-03 | 2811 |
| 2019-04 | 2827 |
| 2019-05 | 2829 |
| 2019-06 | 2833 |
| 2019-07 | 2829 |
| 2019-08 | 2854 |
| 2019-09 | 2852 |
| 2019-10 | 2837 |
| 2019-11 | 2842 |
| 2019-12 | 2837 |
| 2020-01 | 2855 |
| 2020-02 | 2857 |
| 2020-03 | 2876 |
| 2020-04 | 2875 |
| 2020-05 | 2876 |
| 2020-06 | 2883 |
| 2020-07 | 2908 |
| 2020-08 | 3156 |
| 2020-09 | 3123 |
| 2020-10 | 2981 |
| 2020-11 | 2892 |
| 2020-12 | 2897 |
| 2021-01 | 2884 |
| 2021-02 | 2883 |
| 2021-03 | 2885 |
| 2021-04 | 2894 |
| 2021-05 | 2886 |
| 2021-06 | 2887 |
| 2021-07 | 2892 |
| 2021-08 | 2888 |
| 2021-09 | 2887 |
| 2021-10 | 2878 |
| 2021-11 | 2889 |
| 2021-12 | 2882 |
| 2022-01 | 2879 |
| 2022-02 | 2874 |
| 2022-03 | 2867 |
| 2022-04 | 2866 |
| 2022-05 | 2862 |
| 2022-06 | 2853 |
| 2022-07 | 2862 |
| 2022-08 | 2861 |
| 2022-09 | 2865 |
| 2022-10 | 2870 |
| 2022-11 | 2877 |
| 2022-12 | 2874 |
| 2023-01 | 2880 |
| 2023-02 | 2887 |
| 2023-03 | 2893 |
| 2023-04 | 2904 |
| 2023-05 | 2915 |
| 2023-06 | 2925 |
| 2023-07 | 2930 |
| 2023-08 | 2946 |
| 2023-09 | 2951 |
| 2023-10 | 2956 |
| 2023-11 | 2960 |
| 2023-12 | 2964 |
| 2024-01 | 2971 |
| 2024-02 | 2980 |
| 2024-03 | 2989 |
| 2024-04 | 2992 |
| 2024-05 | 3001 |
| 2024-06 | 3005 |
| 2024-07 | 3007 |
| 2024-08 | 3011 |
| 2024-09 | 3012 |
| 2024-10 | 3013 |
| 2024-11 | 3009 |
| 2024-12 | 3009 |
| 2025-01 | 3010 |
| 2025-02 | 2997 |
| 2025-03 | 2988 |
| 2025-04 | 2976 |
| 2025-05 | 2952 |
| 2025-06 | 2944 |
| 2025-07 | 2935 |
| 2025-08 | 2916 |
| 2025-09 | 2914 |
| 2025-10 | 2748 |
| 2025-11 | 2733 |
| 2025-12 | 2722 |
| 2026-01 | 2685 |
| 2026-02 | 2689 |
| 2026-03 | 2682 |
| 2026-04 | 2681 |
| 2026-05 | 2685 |
| 2026-06 | 2686 |
| 2026-07 | 2683 |
Paychecks rose. The buying power barely did.
Average hourly earnings for all private employees were $37.62 in July 2026, up $1.78 from January 2025. In 1982–84 dollars — the Bureau’s real-earnings series — the same measure was $11.30, up six cents from January 2025 and 38 cents from January 2019. One month, October 2025, is missing from the real series in this snapshot; neighboring months sit at $11.32 and $11.38. Nominal raises are real in the sense that they appear on the stub. They are not real in the sense that they are outrunning the price level in any dramatic way. That matters in a week when gasoline is a dollar above last year and the 10-year Treasury yield is 4.79 percent. A $4 gallon does not wait for a six-cent real raise.
None of this is a forecast of recession, and none of it assigns blame to Hormuz, the Federal Reserve, or a particular White House hiring freeze. Payrolls are still higher than a year ago. Hours have not collapsed. Temporary help has stopped falling. The honest reading of the official books is narrower and less cinematic: the U.S. job machine downshifted hard in 2024 and 2025, 2026 has been a low-gear crawl rather than a stall, and almost all of the remaining forward motion is in health care. “Sturdy” is accurate if the alternative is a 2020-style collapse. It is a stretch if the test is whether the typical worker is being hired, getting hours, or getting a raise that keeps up with the pump. The August employment report, due after this edition, will update the tape. The July books already say the war is walking into a labor market that had already taken its foot off the gas.
Sources and methods
This is retrospective enrichment using a pinned BLS Current Employment Statistics snapshot (release_id bls-ce-snapshot-2026-08-24, snapshot_created_utc 2026-08-24T10:20:27Z). The historical edition date is 2026-09-02. The research extract was pulled on 2026-09-11. Later revisions may be present in this snapshot relative to what was published on edition day.
Observation window: monthly seasonally adjusted national series, January 2015 through July 2026 for total nonfarm payrolls, the three-month average change, private weekly hours, manufacturing overtime, temporary-help employment, and average hourly earnings; January 2019 through July 2026 for supersectors, federal employment, and oil and gas extraction. Periods are calendar months fully contained in the request window.
Units: payroll series are in thousands of jobs. Hours are average weekly hours. Overtime is average weekly overtime hours of all manufacturing employees. Average hourly earnings are in current dollars and in 1982–84 dollars. Indexes use January 2019 = 100. Year-to-date and calendar-year average monthly changes are simple means of month-to-month first differences in the seasonally adjusted total nonfarm level; the 2015 average uses 11 months because January 2015 has no prior month in the extract. Supersector 12-month changes compare July 2025 with July 2026.
Nulls are not treated as zero. The real hourly-earnings series has one missing month (October 2025) in this snapshot; that month is omitted from real-earnings comparisons. No duplicate observations were found on the extracted series. Seasonally adjusted and not-seasonally-adjusted series were not mixed. Industry totals are not household averages. Supersector changes are not a complete decomposition of the total nonfarm change because of rounding and because private education and health is the private supersector, not the combined public-and-private education category.
The catalogue’s JOLTS product only contains periods from 2000 through 2003 in this pin, so this article does not treat the reported 7.27 million July openings as an official extract. That figure is attributed to contemporaneous reporting. This analysis does not estimate causal effects of the Iran war, tariffs, or federal hiring policy on payrolls. The August 2026 employment report is outside the snapshot.
Research completed 2026-09-11, for the September 2, 2026 news edition. This is retrospective analysis, not a reconstruction of information available that day.