The Bureau of Labor Statistics’ (BLS) summary of the Employment Situation for February underlined the cautious behavior of U.S. employers rather than the mid-winter optimism that last month’s jobs report appeared to indicate. BLS reported a decline in non-farm employment of 92,000 (this number will be revised in each of the next two months). The Bureau revised down the large gain in January from 130,000 to 126,000 (this estimate will be revised once more), and the final revision for December pushed that month into negative territory: from +48,000 to -17,000 jobs, which is a revision of -65,000.
The final revision for December laid in place the verdict on 2025 job growth: non-farm employment last year was the slowest non-recession year increase since 2003. Total employment grew by 69,000 last year (not seasonally adjusted) and by 124,000 in 2003. In fairness, much of this slow growth could be due to policy decisions by the administration. The federal workforce shrank by -277,500 jobs (not seasonally adjusted). Also, reductions in immigration, both legal and illegal, reduced labor force growth which most likely reduced employment gains. That said, other economic metrics indicate that the economy slowed. Tariffs may have affected manufacturing employment, which declined by 119,000 in 2025. Likewise, the trade sensitive transportation and warehousing sector declined by 123,800.
Signs of a Recession?
Given the high likelihood that much of last year’s sluggishness was driven by policy choices, some of which were calculated to result in long-term gains for the U.S. economy, it would wrong to assume that the job market signals point to recession. The tame unemployment rate reinforces that view. BLS’s principal metric of unemployment, U3, grew by a weak 0.1 percent in February, which settled it at 4.4 percent (the same reading it had in December).
While the rate is elevated and will further focus the attention of the Federal Reserve on measures to bring it down, the rate in February stood only 0.2 percentage points above its February 2025 rate of 4.2 percent. That same “little changed” label applies to most sex, race, and ethnicity unemployment rates. The exceptions are African American unemployment, which now stands at 7.7 percent, compared to 6.0 a year ago; and, Asian unemployment, which has grown from 3.2 percent in February 2025 to 4.8 percent last month. The labor force participation rate (percent working or looking for work) and the employment population ratio (percent working out of the population 16 and above) remained very steady from last month’s reading, as they have for most of 2025.
That does not mean, however, that public expressions of economic distress are without merit. Average hourly earnings for all employees rose by 3.8 percent in 2025. At the same time, the “typical consumer unit” (think family) faced an inflation rate of 2.8 percent (not seasonally adjusted) last year. In short, earnings were barely keeping up with inflation.
However, about 60 percent of the population probably had an inflation rate higher than 2.8 percent. Research conducted by BLS in 2024 indicates that the bottom three-fifths of the income distribution faced an inflation rate higher than the “typical consumer unit”, and the top two-fifths had an inflation rate lower than the official rate. In other words, analysts should not rely on the announced CPI when drawing conclusions about affordability across the income distribution.[1]
Table 1
This distributional consideration also holds for earnings. The average hourly earnings rate for all employees is just that, an average. Those workers in higher income brackets commonly enjoy increases in earning above the average rate of change, while those in the lower ranges of income often see slower gains.[2]
A Note on the New Population Controls
Finally, the annual population benchmarking by BLS usually is not newsworthy. Unlike the employment benchmarking that almost always makes headlines, the annual resetting of the civilian population estimate does not. This year, however, Census sent BLS some dramatically different population numbers from last year, and they significantly change how analysts understand the size of relevant labor force populations.
What are population controls? The basis for estimating the number of people working or unemployed is a concept called the civilian non-institutionalized population. This is the number of people resident in the U.S. who are age 16 and above and not institutionalized in a prison or similar facility. In other words, this population could be working, looking for work, or not in the labor force for any number of reasons. The survey collects data from the survey sample of 60,000 households (the Current Population Survey) on the number of people looking for work. This small number then is multiplied to be comparable to the size of the civilian population. Thus, a number like 3,800 survey respondents who say they are looking for work becomes something like 7,500,000 once the survey result if plussed up to the national population control number.
The new population controls from the Census Bureau, which BLS is required to adopt, are based on new estimation techniques. These techniques use new information on the population changes in 2020 and new estimates of the changes in immigration during the second half of 2024 and throughout 2025. Higher than typical mortality in 2020 and abnormal movement of the population made estimates of that year’s population very difficult. The Census Bureau faced these challenges in collecting the decennial census that year, and by all measures failed to produce a high-quality enumeration.
Table 2
Table 2 shows the differences these estimation changes made. The new civilian population contains fewer men and more women. The biggest change is in the population of the white race: -5,642,000 fewer than in the previous estimate, which trickles down into all the major labor force metrics. Also significant is the smaller population of African Americans and the larger number of Asians and Hispanics.
Clearly it will take some time to digest what, if any, labor force implications stem from these population changes. In any event, policy researchers should focus on this development both as it affects government programs that depend on BLS population counts and, more importantly, those that depend on what Census says about particular segments of the U.S. population.
[1] Note that Q1 is the lowest income quintile and Q5 is the highest.
[2] EPIC will be publishing research on this topic in the near future.
[3] Note that detail may not sum to totals because of rounding. Estimates for the above race groups (White, Black, or African American, and Asian) do not sum to totals because data are not presented for all races. People whose ethnicity is identified as Hispanic or Latino may be of any race.
