The Ground Shifted
The Ground Shifted
The Ground Shifted
What the January 2026 Jobs Report Does Not Say
The Bureau of Labor Statistics reported 130,000 new jobs in January 2026. The confidence interval on that number is plus or minus 122,000. That means the true gain could be as low as 8,000. It could be as high as 252,000. The headline was treated as good news. It is noise.
The real story is in the revisions. Every year the BLS re-anchors its survey estimates to hard tax records from the Unemployment Insurance system. This year that process erased 898,000 jobs from the March 2025 employment level. That is a downward revision of 0.5 percent, more than double the average benchmark correction of the last decade. The models had been wrong. They had been wrong by a wide margin and in one direction. They overestimated growth. The birth-death model, which estimates new business formation between survey periods, had been overcounting. It was recalibrated. The statistical machinery itself broke down, and the repair is still underway.
The damage runs through the whole year. Total job growth for 2025 was initially reported at 584,000. After revision it stands at 181,000. That is 69 percent of the year's gains gone. Not lost in layoffs. Lost in arithmetic. The jobs were never there. November and December were revised down by a combined 17,000. The year did not end well and it did not end where anyone thought it ended. The October 2025 federal government shutdown prevented data collection for that month. The missing data added noise to the annual revisions and made the current numbers harder to read.
The national unemployment rate is 4.3 percent. For Black men over twenty it is 7.3 percent, nearly double the national figure. In January alone their labor force participation dropped 0.8 percentage points, from 69.6 to 68.8 percent. Their employment-population ratio fell from 64.9 to 63.8. These are not small moves for a single month. These are people leaving the workforce. The headline does not show this. Table A-2 of the Household Survey does.
Approximately 1.8 million workers have been out of work for 27 weeks or more. That number is up 386,000 from a year ago. The long-term unemployed now make up a quarter of all jobless workers, and the average duration of unemployment has reached 23.9 weeks. That is close to the 27-week line where the government classifies it as long-term. The pool is hardening. The longer someone stays out, the harder it is to get back in. The resume gap widens. The callbacks stop. Skills atrophy. Networks thin out. The person who lost a job in June 2025 and is still looking in January 2026 is not the same candidate they were seven months ago, and the market treats them accordingly. Employers know this and the unemployed know it better.
Then there is the question of hours. There are 4.9 million workers classified as part-time for economic reasons. They want full-time work and cannot get it. That figure is up 410,000 over the year. They are counted as employed. They are not fully employed. Employers are hiring, but they are hiring thin. They are building a just-in-time workforce and the people inside it carry the cost in lost wages and lost benefits and lost stability.
The January gains were not spread evenly. Health care added 82,000 jobs, more than double its 2025 monthly average. Social assistance added 42,000. Construction added 33,000. These three sectors carried the report. Without them the number is thin and the picture is different.
On the other side, the federal government lost 34,000 jobs in January. Since October 2024 it has shed 327,000 positions, a cut of 10.9 percent driven largely by the expiration of deferred resignation offers. Financial activities lost 22,000, with insurance carriers accounting for half of that decline. The sectors that built the old middle class are contracting and the sectors that are growing do not replace what was lost on the same terms.
People with disabilities face an unemployment rate of 8.1 percent against 4.5 percent for those without. Their labor force participation rate is 24.9 percent. For people without disabilities it is 67.8 percent. The gap is wide and it is not closing. These numbers sit in Table A-6 of the report. They do not make the headlines.
A gain of 130,000 jobs sounds like forward motion. Set it against a confidence interval of 122,000 and it is barely a signal. Set it against 898,000 vanished jobs and it is a rounding error. Set it against 4.9 million workers who want more hours and 1.8 million who have been out for half a year or longer and it is not a recovery. It is a holding pattern. The labor market is not collapsing. But it is not recovering either. It is standing still, and the ground beneath it has shifted.