It’s an astonishing statistic that took the internet by storm: 94% of jobs created in 2021 went to racial minorities. It’s also completely untrue.
The conclusion from Bloomberg — which credits five reporters and four editors for its report — is that on the heels of the Black Lives Matter protests, corporations followed through on their promise to diversify their ranks. The outlet reported that companies in the S&P 100 added 323,094 new employees in 2021, and that only 20,524 of those new jobs were filled by whites.
If true, such a statistic could launch a legion of victorious racial discrimination lawsuits — blacks, Asians, and Hispanics make up roughly 40% of the population, and there is no way they could obtain 94% of new jobs without some level of discrimination.
But the takeaway from Bloomberg is an example of innumeracy by journalists at an outlet once relied upon by businessmen for savvy financial analysis — or of willfully setting aside common sense in favor of what the publication acknowledged is an ideologically-driven “project to hold companies accountable to [diversity] pledges.”
A Daily Wire analysis of the same numbers examined by Bloomberg found that, in reality, the demographics of hiring figures for 2021 were barely different from previous years. The percentage of new jobs that went to whites was likely about 46%, eight points below the 54% white makeup of companies’ existing workforces. That’s to be expected given demographic changes in the United States since the time that the currently-retiring baby boomer generation first entered the workforce.
Though Bloomberg spun the tale as a victory for Black Lives Matter, blacks benefited the least of any racial group from the slight decline in whites, according to the analysis. The percentage of black hires was up from the status quo by 1%, while Asians were up by 2%, and Hispanics were up by 4%. That’s also explained by demographics — decades ago, when baby boomers entered the workforce, the U.S. was mostly white and black; in the decades since, the numbers of Hispanics and Asians in the United States have increased.
Here’s how Bloomberg got the story wrong, and how numbers actually work.
Bloomberg based its analysis on a form companies file to the Equal Employment Opportunity Commission listing the racial breakdown of their U.S. employees. The forms are filed annually, but they don’t break out stats for employees hired that year; they just provide the total headcount of all employees by race.
Bloomberg, reaching for a way to isolate recent numbers, focused on the fact that companies increased their cumulative headcount by some 320,000 in 2021. Then they made a flawed leap of logic: They took the increase of minorities across the entire workforce, and divided it by the number of new positions — not the number of actual hires, which overwhelmingly come from replacing people who leave the company. In short, they got the denominator wrong.
“Bloomberg obtained 2020 and 2021 data for 88 S&P 100 companies and calculated overall US job growth at those firms…. Overall, these companies increased their headcount by 323,094 employees in 2021,” the outlet wrote. “Bloomberg then analyzed the racial makeup of those additional workers, finding that 94% of them were people of color.”
But it’s not possible from the data to say that those additional “people of color” took the 320,000 newly created positions. Most of them were almost certainly hired as part of a much larger group: replacements for existing jobs that were vacated by retirees or people changing jobs.
The 88 companies in Bloomberg’s study together employed 8.9 million workers in 2020. Based on a typical annual turnover rate of 18%, an estimated 1.6 million people would have left and been replaced in 2021. So the increase in minorities shouldn’t be calculated as 300,000 out of 320,000 — it’s 300,000 out of roughly 2 million (1.6 million plus 320,000).
Some companies may have had no employment growth at all, just turnover from retirements. Bloomberg’s methodology could capture minor demographic changes across companies like this, then apply all those changes to a small number of newly-created positions from entirely different companies.
Take this hypothetical: There are two companies, each staffed by 100 white people. One of them replaces 20 retiring white employees with 20 black employees. The other is very profitable, so it creates 10 new positions, all given to whites. Bloomberg’s methodology would conclude that the companies, as a group, hired 20 black people into 10 new jobs, and that minorities received 200% of jobs. This is obviously impossible, and in fact, all of the new jobs went to whites.
Chart from Bloomberg News based on its flawed study
In 2020, the S&P companies were just under 54% white. If every retiree was replaced with someone of the same race, and whites also got the same share of the new jobs, you’d expect to see 173,000 whites added. Bloomberg observes that there were only 20,000 whites added.
What this tells us is there was a shortage of approximately 153,000 whites — but that’s not across the 320,000, it’s across 2 million jobs filled. Instead of hiring just over 1 million whites (54%), they hired about 884,000 (45%).
On the “people of color” side, Bloomberg said that Hispanics grew by 131,000 in the 320,000 jobs. But they actually simply got their share of newly-created new jobs (54,000), plus out of the 864,000 whites retiring, 77,000 were replaced by Hispanics.
In all, we can compute that of the 2 million jobs filled in 2021, about 400,000 went to Hispanics — 20%, which is only slightly higher than the 17% in the legacy workforce.
Following the same logic, the number of 2021 hires who were Asian was 12%, slightly higher than the 10% of the legacy workforce, while blacks went up 1%, from 17% to 18%.
Though blacks benefited the least from the decline in whites, that’s expected because blacks, who make up 12% of the population, are actually overrepresented in the workforce studied by Bloomberg.
Robert VerBruggen, an economic policy fellow at the Manhattan Institute, agreed that Bloomberg’s analysis failed to take into account the workers leaving and their demographics.
“The country’s demographics are changing — older generations are much whiter than younger ones,” VerBruggen told The Daily Wire. “If whites are disproportionately retiring and non-whites are disproportionately getting jobs, that will skew the numbers, giving the appearance of a serious commitment to equity — or mass violations of civil-rights law, depending on one’s perspective.”
A telltale sign that Bloomberg’s “percentage of the net increase” methodology is flawed, VerBruggen explained, is that, if the departures of whites had been just a little higher, the net change in whites would have been negative instead of the actual small growth of 20,000. Bloomberg’s methodology would then assert that whites took a negative percentage of the new 320,000 jobs, a mathematic impossibility.
The study also received pushback from pollster Patrick Ruffini, who said Bloomberg’s analysis was “statistically illiterate” and pointed out the outlet’s failure to take into account employee turnover.
“This stuff could be the product of generational turnover in the workforce,” Ruffini said. “If retirees are 80% white and new workers are 55% in line with demographics of young workers, you get numbers like this.”
Normally in journalism, there is a safeguard that comes in the form of calling the companies mentioned for comment. But Bloomberg said “none of the half-dozen companies with the most notable diversity gains wanted to discuss them in detail.”
The companies could have pointed out that Bloomberg’s methodology was flawed, and prevented them from publishing an error. But the error overstated their diversity gains, and it was clear from the reporters’ bias that big diversity gains would be portrayed favorably. (The story was headlined “Corporate America Promised to Hire a Lot More People of Color. It Actually Did.”)
Asked about the methodology for its study, a Bloomberg spokeswoman said “We stand by our reporting and analysis.”
The Daily Wire’s calculations based on the data used by Bloomberg can be viewed below:
