23,000 fewer jobs: A counterintuitive signal from the US job market

If you only look at one number, the US July jobs report might even seem optimistic.

The unemployment rate fell to 4.1% from 4.2% in June. Many people believe that a decline in the unemployment rate usually indicates increased job opportunities and an improved labor market. However, the same report released by the U.S. Bureau of Labor Statistics (BLS) on August 7 showed that non-farm payrolls decreased by 23,000 in July, while the market had widely expected an increase of approximately 80,000.

(Image caption) More empty workstations and sparse employees are appearing in open-plan offices in the United States, reflecting a net decrease of 23,000 non-farm payroll jobs in July, indicating that companies are becoming significantly more cautious in their hiring practices.

The figures for the previous two months have also been revised.

The number of new nonfarm jobs in May was revised down from 129,000 to 63,000, and in June from 57,000 to 20,000, a total of 103,000 fewer jobs than previously reported. In July, job growth turned negative. These three months combined present a far more complex picture than a single 4.1% unemployment rate.

In July, the U.S. labor force decreased by 264,000, with the labor force participation rate falling to 61.4%; the number of employed people also decreased by 87,000, with the employment-to-population ratio falling to 58.9%. Since January of this year, the labor force participation rate has declined by 0.7 percentage points, and the employment-to-population ratio has declined by 0.5 percentage points.

If we only focus on the 4.1%, a very important part of this report is easily missed: fewer and fewer people are remaining in the labor market.

Those who are not included in the unemployment rate

To understand this report, it is necessary to first understand a statistical premise of the US unemployment rate.

A person without a job is only counted as unemployed if they are actively seeking work and meet the relevant survey criteria in the near future. Once they stop looking for work, they may be classified as "out-of-labor force" and no longer included in the denominator of the unemployment rate. Therefore, it is entirely possible for an economy to experience both job losses and a decline in the unemployment rate simultaneously.

In July, the U.S. labor force decreased by 264,000 to 169,094 million from 169.358 million; the number of unemployed decreased by 178,000 to 6.916 million. Meanwhile, the number of people not in the labor force increased by 381,000 to 106.189 million.

Within this large group, there is still a segment of people who do not fully desire to leave the workforce. BLS statistics show that in July, approximately 5.9 million people who are not part of the labor force said they still wanted to work. Of these, about 1.8 million were classified as having "marginal ties" to the labor market, and about 476,000 were discouraged job seekers—those who are not currently actively looking for work but had previously wanted to work and stopped looking because they felt they could not find suitable opportunities.

Of course, there are many reasons for labor force withdrawal. Retirement, family care, education, health conditions, and other personal choices can all cause a person to leave the labor market temporarily or permanently, so the 264,000 decrease in the labor force in July cannot be entirely attributed to job seekers who failed to find employment.

However, when the labor force participation rate, the employment-to-population ratio, and employment growth all weaken simultaneously, these changes deserve to be observed together. For young people looking for their first job, mid-level white-collar workers preparing to change careers, and those who have just left their previous jobs and are trying to re-enter the job market, changes in the employment environment often precede their daily perceptions of the macro unemployment rate.

(Image caption) Street view and electronic data show that, echoing the recent high in May 2025, U.S. financial activity employment has decreased by approximately 121,000 jobs.

Put May, June and July together

The loss of 23,000 jobs in July is not enough to declare a structural shift in the U.S. labor market.

Monthly employment data is subject to sampling errors, seasonal adjustments, and subsequent revisions. In July, the private sector actually added approximately 30,000 jobs, while government employment decreased by about 53,000, including a decrease of about 50,000 in local government education positions, which was one of the important factors contributing to the overall negative employment situation that month.

Therefore, the net decrease of 23,000 jobs cannot be simply interpreted as a sudden, widespread layoff by American companies. The BLS's official description of July's non-farm payrolls remains "minimal change."

If you put May, June, and July together, the situation becomes clearer.

The figure for May, initially reported as an increase of 129,000 jobs, has now fallen to only 63,000; June's figure was revised down from 57,000 to 20,000; and July's figure decreased by 23,000. The BLS also noted that the average monthly job creation over the previous 12 months was only 34,000.

While there hasn't been a widespread and dramatic wave of layoffs in the US, hiring has become increasingly cautious. Recent JOLTS data also shows a similar trend: job vacancies, hiring, resignations, and layoffs haven't seen any out-of-control surges; the labor market seems to be gradually losing its former vigorous liquidity over a longer period.

The jobs are still available, and the company is still hiring, but the door is a bit narrower than it was a few years ago.

The number of new jobs the United States needs is also changing.

There's another context that rarely appears in monthly news headlines, but it's changing how we read U.S. employment data.

Federal Reserve researchers, in their analysis of U.S. labor force growth this year, pointed out that slower population growth, changes in net immigration, and a decline in labor force participation due to an aging population could all push U.S. labor force growth closer to very low levels. As the labor force itself no longer increases as rapidly as in the past, the number of new jobs needed to maintain a stable unemployment rate will also decrease.

This is important.

Ten years ago, the health of the US labor market was often measured by the number of new jobs added each month, ranging from tens of thousands to hundreds of thousands. In the future, if population and labor force growth slows significantly over a prolonged period, with job growth near zero, or even occasionally turning negative, it doesn't necessarily mean the economy will immediately enter a recession.

Assessing the labor market therefore requires more patience. Whether hiring continues to decline, whether job vacancies are shrinking, whether wages can maintain growth, how the labor force participation rate is changing, and in which directions are job opportunities shifting across different industries—these indicators are more explanatory than any single monthly figure.

(Image caption) The job search resource center is sparsely populated, and job seekers are faced with limited job information, which symbolizes a decrease of 264,000 in the labor force and a drop in the labor force participation rate to 61.4%.

121,000 jobs in the financial industry

There's another set of figures worth noting in the industry structure of July.

The retail sector lost 19,000 jobs, local government and education lost 50,000, while healthcare still added 22,000. The financial sector lost 14,000 jobs, including approximately 9,000 in credit intermediation and related services and approximately 7,000 in insurance and related services.

Since its recent peak in May 2025, U.S. financial activity employment has decreased by approximately 121,000 jobs.

The financial industry is worth observing in the long term, and the reasons are not mysterious. Banks, insurance companies, asset management, research, compliance, customer service, risk analysis, and back-office operations all involve a large amount of work that relies on information processing, text, data, processes, and professional judgment. These are precisely the areas where generative AI is first entering enterprise workflows.

But BLS’s figures don’t tell us exactly how many of those 121,000 finance jobs are related to AI.

There is currently insufficient evidence to directly attribute the decline in financial industry employment to AI. Interest rate environment, mergers and acquisitions, cost control, business adjustments, financial cycles, and digital transformation itself can all affect the size of the workforce in this industry. Explaining all declines as "AI taking jobs away" is not only inconsistent with existing evidence but also obscures other changes taking place within the financial industry itself.

However, some companies have begun to provide some noteworthy clues.

Just a week before the July jobs report was released, fintech company Chime confirmed it would lay off about 10% of its workforce, affecting approximately 150 positions. The company cited a leaner organization and efficiency gains from AI as factors in the restructuring. While this is still just one company and cannot represent the entire financial industry, it at least demonstrates that some companies have begun to openly discuss the relationship between AI and staffing.

There will likely be more cases like this in the future.

AI has already entered businesses, and its impact on employment is still unfolding.

Regarding AI and jobs, the most reliable approach at present is still to put evidence before conclusions.

In March of this year, Federal Reserve researchers analyzed corporate AI adoption and Lightcast job posting data, finding no significant negative impact of corporate AI investments on subsequent job postings. The researchers also cautioned that existing data is insufficient to support the claim that AI has led to a deterioration in the overall labor market.

The regional business survey of the New York Federal Reserve District provides another perspective.

Among service sector companies already using AI, approximately 12% reported reducing hiring due to AI in the past period; among companies planning to further utilize AI, nearly a quarter expected to reduce hiring in the future as a result. At that time, the proportion of companies directly laying off existing employees due to AI was still very low.

This data may not be representative of the entire United States, but it raises a question worth following: the earliest impact of AI on employment may not necessarily manifest as mass layoffs.

A company originally planned to add five people but only added three; after someone left, the company decided not to replace them; a team of ten people can now handle tasks that previously required more people after changes in tools and processes. These adjustments, if they occur sporadically, are difficult to identify in macro data; however, if they continue for several years, they can gradually change the hiring scale and growth rate of certain professions.

Various business surveys compiled by the Federal Reserve in April of this year also show that AI has rapidly entered the daily operations of American companies. According to the Census Bureau's business survey, approximately 18% of businesses will have adopted AI by the end of 2025; another survey, weighted by the number of employees in a company, estimates that approximately 78% of American workers' companies already use some form of AI, and approximately 54% of workers' companies use large-scale language models.

These sets of figures use different statistical methods and samples, so they cannot be directly compared side-by-side. What they all convey is simple: AI is no longer just an experimental tool for a few tech companies, but is participating in the daily work of more and more enterprises.

As for how many jobs it will ultimately add or remove, and which jobs will be redefined, time will tell.

(Image caption) A corporate meeting room discusses AI analytics dashboards and staffing, showing that generative AI has entered daily workflows and is beginning to influence hiring and organizational restructuring decisions.

As capital increases, how many more employees does the company need?

Businesses are investing in AI much faster than the labor market is changing.

Data centers, advanced chips, cloud computing, model training, enterprise software, and automation systems are attracting significant capital investment. For company managers, a very real question arises: after the next round of funding is invested in AI infrastructure and software, how many more employees will the company need in the next phase?

The relationship between AI investment and employment is not necessarily singular. Some companies increase staff due to improved efficiency, revenue growth, and new product launches, while others maintain a leaner workforce while continuing to grow revenue and output. More often than not, the first change may be in the job content itself, with the same position taking on new tasks and existing workflows being rearranged.

These changes are difficult to fully account for in a single quarter.

AI capital expenditures can increase by billions of dollars within months, while the formation, decline, or redefinition of a profession often takes years to leave a clear mark on official statistics. This time lag between capital and labor is likely to become one of the most important windows for understanding the AI economy in the coming years.

Wages have not yet shown typical signs of recession.

There's another figure in July that can't be ignored.

The average hourly wage in U.S. private non-farm companies was $37.62, up 3.2% year-over-year, with the average weekly working hours remaining at 34.3 hours.

This data complicates the July jobs report. Hiring is slowing, labor force participation is declining, some industries continue to contract, but wages are still rising, and recent JOLTS data does not indicate a sudden and significant out-of-control surge in layoffs.

Therefore, it is still too early to describe the US labor market as an "employment crisis".

A more realistic description is that the cooling down is uneven. Some industries are still short of workers, hiring in some white-collar sectors has begun to contract, some companies are still expanding, while others are investing capital in technology and automation. This asynchronous change also presents the Federal Reserve with a more complex policy environment.

If employment deteriorates rapidly while inflation declines significantly, monetary policy can find its direction more easily. However, when hiring slows down, wages continue to rise, and businesses also have to absorb tariffs, energy, financing, and AI investment costs, the policy choices are not so clean.

Following the release of the data on August 7, the market immediately readjusted its expectations for the Federal Reserve's future policy path. For financial markets, a jobs report affects interest rates, bonds, stocks, and the dollar; for ordinary people, it ultimately boils down to some very specific questions—whether jobs are easy to find, whether wages can increase, and what a job that still exists today will look like in a few years.

(Image caption) A comparison of large-scale AI data center facilities with surrounding industrial areas reveals the time lag between massive capital investment in AI infrastructure and a gradual cooling of the labor market.

The next table is more important than the next heading.

The loss of 23,000 jobs in July may, in retrospect, appear as a small dip in the statistical curve years from now. It could also be an early sign of a longer period of change. There isn't enough data yet to determine which scenario it falls into.

What the media needs to do most at this time is to put data that is easily reported separately on the same timeline.

GFM.News will continue to monitor employment changes in the technology, finance, information, and professional services sectors, while also recording corporate hiring and job vacancies, large companies' AI capital expenditures, publicly disclosed AI-related layoffs and organizational restructuring, as well as changes in productivity, wages, and corporate revenue.

If a consistent and clear combination emerges in the coming years—increased investment in AI, continued growth in corporate revenue and productivity, and a prolonged slowdown in hiring in some AI-intensive industries, with a gradual decline in their share of employment and wages—then we will have more data to study another, deeper question: how the new wealth created by technology will ultimately be redistributed among capital, businesses, and workers.

This question cannot be answered in a monthly employment report; it requires data from several years or even longer, and continuous observation of business investment, productivity, hiring, wages, and job changes on the same table.

The first clue left in July 2026 is the simultaneous occurrence of a 4.1% unemployment rate and a decrease of 264,000 in the labor force. For those accustomed to using the unemployment rate to judge the state of the economy, this is enough to remind us that familiar indicators remain important, but we need to read them more carefully.

Some people are still working, and AI is being integrated into the tools they use every day; some jobs are being redefined, and new jobs are being created. Meanwhile, some people have temporarily left the labor market. Looking back a few years from now, what will be most memorable about July 2026 may not be the 23,000 jobs lost, but rather how the relationship between capital, technology, and people has changed during this period.

Disclaimer

This article is based on publicly available information for news research and institutional analysis, and does not constitute any investment, transaction, legal, or policy advice. Employment, economic, and corporate data may change as official revisions or subsequent disclosures occur; readers should refer to the latest publicly available information.