Photo by sol

Job growth was far stronger than most people, including me, had expected. With the upward revisions to the prior two months’ data, we don’t have to worry about weak job growth pushing unemployment higher in the immediate future. Still, weakening wage growth and a drop in employment rates provide real grounds for concern.

Here are my top five items from the August report:

+ Wage growth continues to slow;

+ Disadvantaged groups, like Black workers and young workers, seem to be doing better;

+ Prime-age workers and college grads are seeing a weaker labor market;

+ There is little evidence of an AI productivity boom;

+ Women continue to get most of the new jobs.

Slowing Wage Growth Is Bad News for Workers

The average hourly wage increased at just a 3.1 percent annual rate year-over-year. Comparing the average for the last three months (June-August) with the prior three (March-May), wages have grown at just a 2.7 percent annual rate. This is down from a rate of over 4.0 percent through most of 2023 and 2024.

Some analysts have pointed out that the slowing is somewhat less pronounced using the Employment Cost Index for total compensation, which includes health care benefits. That does show a slightly better picture, but still a clear slowing from over 4.0 percent at the start of 2024 to 3.3 percent in the year from the second quarter of 2025 to the second quarter of 2026.

It’s also worth noting that, insofar as wages are being diverted to health care spending, that is still likely to increase stress on workers’ budgets. It doesn’t help them that their employers are paying more for their health care.

The basic point here is straightforward: Wage growth should not be slowing in a strong labor market. That is especially true when workers are seeing higher rates of inflation.

Disadvantaged Workers Seem to be Doing Better

The increase in unemployment rates in 2025 among different disadvantaged groups almost seemed designed to illustrate discrimination in the labor market. The unemployment rate rose from a year-round average of 6.0 percent for Black workers in 2024 to a peak of 8.2 percent in November of 2025. By comparison, the unemployment rate for whites edged up only slightly, from 3.6 percent to 3.9 percent.

For young workers (ages 20-24), the unemployment rate went from an average of 7.4 percent in 2024 to a peak of 9.2 percent in September of 2025. For prime-age workers, it went from 3.3 percent to 3.7 percent in September of 2025.

That story has changed hugely in 2026. The unemployment rate for Black workers was down to 6.0 percent in August, the same as its 2024 average. For young workers it was at 7.1 percent, slightly below the 2024 average.

Still, the story on the employment side of the equation is less encouraging. The employment-to-population ratio for Black workers was 57.8 percent in August, down from an average of 59.2 percent in 2024. For Black women, the drop in EPOPs was even larger, from 59.5 percent in 2024 to 57.2 percent in August. For young workers, there is less of a drop, with the EPOP going from 66.3 percent in 2024 to 59.4 percent in August.

Weaker Employment Seems to be Hitting Prime-Age Workers

If the news looks somewhat better for more disadvantaged workers, it is looking worse for prime-age workers. There is little story with unemployment, but there is some deterioration in EPOPs, especially for men. The EPOP for prime-age workers overall stood at 80.4 percent, down 0.3 p.p. from the level a year ago and the year-round average in 2024. However, for men the EPOP was down 0.7 p.p. from its year-ago level and 0.5 p.p. from its 2024 average. For women, there was a rise 0.2 p.p. from a year ago, and the August level is the same as the 2024 average.

There also is an unusual story with college grads faring relatively worse in the labor market than those with less education. The unemployment rate for college grads over age 25 was 2.7 percent in August, up 0.4 p.p. from 2.3 percent in 2024. By contrast, for those with just a high school degree it was unchanged at 4.4 percent.

On the EPOP side, the drops are comparable for the two groups. For college grads it fell 1.8 p.p. from 71.0 percent in 2024 to 69.2 percent in August. For high school grads the drop was 1.9 p.p., from 54.5 in 2024 to 53.6 percent in August, although it had been 54.3 percent in July. In any case, more highly educated workers have not been especially well-protected from labor market weakness, as has normally been the case. It’s also worth noting that the demographics of more people reaching retirement age is a bigger factor for high school grads than college grads. (A higher percentage of college grads are younger.)

AI Keeps Forgetting to Take All the Jobs

It’s hard to find much evidence of AI-related job loss in the establishment data. I had been noting the drop in employment in the insurance industry as one case where there may be evidence of AI having an impact. It lost another 6.3K jobs in August. Insurance is now down 75K (2.5 percent) from its year-ago level. Computing infrastructure and web hosting lost 7.7K jobs in August, putting employment 26.1K (5.4 percent) below the year-ago level.

It’s possible that AI is contributing to these job losses, but this is not exactly a crash, even in these relatively narrow sectors. By comparison, manufacturing lost 1,479,000 jobs, 8.6 percent of employment in the sector, from December 2000 to December 2001.

There may be some temptation to blame the weakness of the labor market for college grads on AI. While AI may be a factor, it seems that cutbacks in the federal workforce and grants, along with cuts in higher education, are more important contributors to weakness in the demand for more educated workers

Women Keep Getting Most of the New Jobs

In the establishment data, women accounted for 158K of the 162K new jobs. This has been a regular pattern throughout the last four years of the recovery. Part of the story is the reversal of job loss during the pandemic, which disproportionately hit women in sectors like hotels and restaurants. But the bigger story is the huge growth in the health  care and social assistance sectors, which disproportionately employ women.

It may seem odd that more than 95 percent of the people employers hired in a month would be women, but it is important to remember that these are net figures. In August, more than 5 million workers left their jobs, either voluntarily or involuntarily (mostly voluntarily). Another 5 million plus workers were hired, with the net figure being a gain of 162K. Women end up being the overwhelming majority because they were less likely to have left their jobs and/or more likely to be hired, but with a denominator of more than 5 million, the sort of discrepancy in job gains reported for August is less implausible.

Overall Picture: August Was Good, but Doesn’t Change the Story of a Weaker Labor Market

There is no basis for a concern that we are about to fall off a cliff, at least until the collapse of the AI bubble, but the slowing of wage growth and drop in EPOPs, even for prime-age workers, are not features of a strong labor market. As I noted before, in a context of slow labor force growth and declining real wages, workers’ consumption cannot drive the economy. The consumption growth we are seeing is based on income from capital and stock gains. That is not a balanced economy.

This first appeared on Dean Baker’s Beat the Press blog.