Amid a pretty good jobs report last week, there were some disturbing signs for Black Americans. The Black unemployment rate reported by the Bureau of Labour Statistics shot up to 6.8 per cent in June from 6 per cent in May — and from an all-time low of 4.8 per cent in April 2023 — while the overall unemployment rate fell to 4.1 per cent from 4.2 per cent. And while unemployment rates for minority groups can jump around a lot from month to month, the less-noisy employment-population ratio for Black Americans has been sliding for more than a year now, and as of June was two percentage points below its 2023 average.

Because the employment-population ratio is to some extent determined by the age of the population, in particular the growing share that is 65 or older, the so-called prime-age ratio for ages 25 through 54 is a better metric — probably the single best measure of the health of the labour market. But the BLS does not publish it in seasonally adjusted form for most demographic groups, and the seasonal swings can be quite large, so I’ve made do here with 12-month moving averages of the unadjusted ratios.

The prime-age lines never converge while those for the entire working-age population — defined in US labour statistics as those 16 and older — did in 2023 and 2024 mainly because Black Americans are younger than Americans overall, with a median age of 35.1 years as of 2024 compared with 39.1 for the total population. Also, older Black Americans did not permanently leave the workforce during the pandemic to the extent that others in their age range did.

The two charts agree, though, in showing a divergence over the past year between a still-healthy employment picture overall and weaker numbers for Black Americans. What could be driving it? President Donald Trump’s efforts to stamp out diversity, equity and inclusion programmes and reduce the federal workforce, in which Black workers are overrepresented, may be playing a role. The decline in Black employment started before he took office, though, which raises suspicion that it could be an early sign of underlying labour market weakness.

When times have been tough for US workers in the past, Black Americans have usually suffered most. They are more likely to get laid off, and less likely to be rehired, than others. During the Great Recession and its aftermath, the prime-age employment-population ratio fell much farther and the unemployment rate rose much higher for Black Americans than for those overall.

By 2019, though, a strong job market had driven the unemployment rate for Black Americans to its lowest level since the data series began in 1972 — a development that Trump with some justification patted himself on the back about. After the brief pandemic recession again hit Black workers harder than others, the rapid recovery brought new employment-population highs and unemployment lows for Black Americans in 2023 — a development that Joe Biden with some justification patted himself on the back about.

In mid-2025, overall unemployment, layoffs and jobless claims are still low, and the prime-age employment rate for Black Americans is quite high by historical standards. But hiring is slow, which may be affecting Black would-be workers disproportionately. A lot of the disparity in employment outcomes by race can be explained by structural factors such as education levels and the industries people work in, but there is also strong evidence that, all else being equal, it remains harder to get hired in the US if you are Black. Again, it is conceivable that Trump Administration policies and public statements are exacerbating these hiring disparities, but because the shift in Black employment rates predated Trump’s inauguration, it also seems like a potential early warning of what is coming for other demographic groups if hiring demand slows more.

If the statistics on employment rates for Black Americans just starting out in the labour market can be believed, alarm bells should be ringing. I have written a lot about the 20-to-24 age group and its employment struggles since the start of the Covid-19 pandemic. But while the 1.7 percentage-point, employment-population-ratio decline for all 20 to 24-year-old Americans since early 2020 is disturbing, the 6.1 percentage-point decline for Black Americans in that age group looks downright recessionary.

The decline is so big, though, that one has to wonder whether it is entirely real. These statistics are compiled by the BLS from the Census Bureau’s monthly Current Population Survey of about 60,000 households, and as one examines smaller slices of that survey population the potential for error grows. The survey’s response rate has also fallen from 82.3 per cent just before the pandemic to 68.1 per cent as of April, and appears to be even lower for young people in general and young Black people in particular¹. A lot of the age-group changes in employment-population rates for Black Americans since the pandemic seem too big to believe, with most overshooting in a positive direction.

One more factor that probably is not having much effect on the numbers yet but could render historical comparisons less meaningful in the future is shifting racial definitions. The Census Bureau began several years ago to ask more open-ended questions about race and ethnicity in its annual American Community Survey and decennial census, and the Biden Administration decreed last year that all government statistics gatherers head in this direction.

The monthly CPS is still conducted the old way, with survey takers instructed to ask whether a person is “White; Black or African-American; American-Indian or Alaska Native; Asian; or Native Hawaiian or Other Pacific Islander” and not mention any other possibility, but the number of respondents volunteering that they are of two or more races has been growing anyway, and the surveyors are beginning this month to experiment with open-ended questions. To get a sense of where that may lead, here’s the percentage of Americans of two or more races as determined by the ACS and decennial census on the one hand and the CPS on the other.

This fading of the lines dividing racial categories is surely a healthy development overall but will make it harder to interpret the labour market statistics for racial groups. Right now, those statistics for Black Americans seem to be pointing to trouble, although how much is not entirely clear.

Notes:

1, I base this surmise on the weights assigned to individual responses in the Current Population Survey microdata released by the Census Bureau. In May, the average Black 20 to 24-year-old in the survey was assumed to represent 5,464 people, while the average White 60 to 64-year-old was assumed to represent 3,072

• Justin Fox is a Bloomberg Opinion columnist covering business, economics and other topics involving charts. A former editorial director of the Harvard Business Review, he is author of The Myth of the Rational Market