How Unemployment AND Employment Are Rising... At The Same Time

The Bureau of Labor Statistics (BLS) recently released its monthly jobs report, indicating an addition of 178,000 jobs and a modest 0.1% decrease in unemployment. However, this data contradicts other economic indicators, such as consumer sentiment hitting its lowest level since data collection began, and a Conference Board survey showing the highest number of consumers ever finding jobs hard to get. Indeed’s data also pointed to an extremely defensive hiring posture among companies. Further evidence contradicting the BLS report includes over 60,000 job cuts announced in March alone, low hiring shown by the Fed's internal data, an increase in weekly unemployment claims to 219,000, and the BLS's own report noting a shrinking labor force by 396,000. These discrepancies—rising unemployment claims, job cuts, and a shrinking labor force alongside an official jobs figure increase—have eroded trust in these statistics, especially with frequent and substantial revisions. The "total non-farm payroll count" from the BLS is the widely cited figure that influences bond yields, Federal Reserve decisions, and political rhetoric. However, this number increasingly misrepresents what most people understand as "jobs." When the BLS reports 178,000 jobs added, many interpret this as 178,000 previously unemployed Americans gaining new employment. In reality, the calculation is far more complex and less reassuring. The BLS compiles this data through two separate surveys: the establishment survey and the household survey. The establishment survey collects data from individual businesses on their payrolls, while the household survey questions hundreds of thousands of households about their employment status. Historically, these two surveys largely aligned, but over the last decade, their relationship has significantly diverged. Research from the San Francisco Fed shows the gap between the two surveys is at its widest recorded level, accelerating since 2020. The establishment survey, which typically generates the headline number, consistently paints a more optimistic picture than the household survey. For example, last month, the payroll survey reported 178,000 new jobs, while the household survey indicated a loss of 64,000 jobs. While the BLS acknowledges these issues in its technical notes, publishing the household numbers, revisions, and methodological footnotes, this detailed information is rarely consumed by the public or market movers. Instead, headlines based on the establishment survey drive market reactions, political narratives, and business and worker decisions, despite this survey's structural limitations in reflecting the modern American economy. The disconnect between the reported numbers and reality stems from three main problems. First, "payrolls are not people." The establishment survey counts payrolls, not unique individuals. In today’s economy, a single person can hold multiple jobs, meaning an increase in payrolls doesn't necessarily translate to more people employed. For instance, a doctor working at a hospital, a private practice, and doing research could add three payrolls while being one working person. This was particularly relevant last month when 35,000 healthcare professionals returning from a strike were counted as new payrolls, not new employment opportunities for previously jobless individuals. Second, the methodology is outdated. The BLS’s model for collecting payroll data, designed for a time when most people worked in factories or offices, struggles to accurately account for remote work or jobs that span multiple locations. This leads to increased double-counting, undercounting, and misunderstandings, contributing to embarrassingly large revisions. Last year, the BLS quietly revised away 800,000 previously reported jobs. The "births and deaths model," used to estimate new and closed businesses, is also skewed by the rise of gig work, where a new "business" could be an individual starting to drive for a ride-sharing service, counted as a job-creating enterprise despite often representing unstable employment. Third, the payroll survey fails to capture a significant and growing segment of the American workforce: private contractors. This includes gig workers, consultants, freelance engineers, and the entire creator economy. These individuals are counted by the household survey but are invisible to the payroll survey. They are also often the first to lose work during economic downturns, as companies cut contractors before employees with formal contracts. These job losses do not appear in press releases or the establishment survey and may not immediately register in unemployment claims, as many contractors are ineligible. This explains how the labor force can shrink by nearly 400,000 people in a month while payrolls simultaneously increase. Despite its flaws, the payroll survey is still used for two main reasons: speed and "reliable unreliability." It provides monthly data quickly, crucial for timely economic decisions. Additionally, businesses are less likely to fabricate payroll numbers than individuals might be to misrepresent their employment status in a survey due to embarrassment or confusion. The household survey also has timing issues; a person losing a job after the 12th of the month will still be counted as employed for another four weeks. The accuracy of these economic statistics is vital because they inform critical decisions, including Federal Reserve interest rate policies, government stimulus, and business hiring plans. Inaccurate data can lead to misguided policies and undermine public faith in the economy. Economists at the BLS are aware of these issues, and other developed economies have already adjusted their methodologies. However, the U.S. has not due to structural reasons. Changing the methodology would create a discontinuity in historical data, which markets dislike. It would also require the BLS to admit its previous numbers were misrepresentative, something federal agencies are reluctant to do, especially with potential funding cuts from Congress. Furthermore, administrations have a quiet incentive to maintain a "politically convenient noise generator"—a monthly number that can swing significantly based on various factors, allowing for selective interpretation in press releases. Consequently, the most quoted economic statistic in the United States continues to tell a story that fewer and fewer working people recognize.