Who Are These Federal Workers Trump Wants to Fire?
The civilian workforce has barely grown in recent decades, is scattered across the country and accounts for a small fraction of US spending.
The US government is an insurance company with an army.
Photographer: Jay Mallin/Bloomberg
Federal workers are uncomfortably in the spotlight, with President-elect Donald Trump threatening to strip civil-service protections from tens of thousands of them and Vivek Ramaswamy, co-chairman of Trump’s planned “Department of Government Efficiency,” having suggested that cutting a million federal jobs within a year is a reasonable target.
As with anything coming out of Trumpworld, it’s hard to know what to take literally, what to take seriously and what to dismiss as totally implausible. What’s clear from an examination of where things stand now is that, while federal spending has grown sharply and regulation is arguably in need of correction, there is no such long-run trend in federal employment, and firing lots of federal workers would not in itself result in much savings or deregulation.
My inspiration here is Syracuse University political scientist Emily Thorson’s recent book, The Invented State: Policy Misperceptions in the American Public, in which she shows that such misperceptions linger in part because journalists — especially opinion journalists — tend to play down basic facts about government because they’re not news. So how about some basic facts about federal workers?
The most basic one in this context is that there are 4.3 million federal government employees — 3 million civilians and 1.3 million active-duty members of the military. It’s non-Postal Service civilian jobs that the prospective Trump administration seems to have in its crosshairs, and there were 2,395,900 of those as of October, according to the US Bureau of Labor Statistics. This number has grown over the past 25 years, including an increase of 76,300 during Trump’s first term, but the growth followed sharp declines in the mid-1990s. Current non-Postal Service federal employment is slightly higher than in the late 1980s and early 1990s. Include the Postal Service, and it’s slightly lower.
Federal Employment Through the Decades
It’s lower now than in the late 1980s and early 1990s
Source: US Bureau of Labor Statistics
Note: Spikes every 10 years represent temporary Census hiring.
Meanwhile, the US population and workforce have continued to grow. Civilian federal workers (Postal Service excluded) now make up just 1.5% of nonfarm payroll employment, down from more than 3% in the 1950s and 1960s and more than 2% through the early 1990s.
The Shrinking Federal Jobs Share
Non-Postal Service civilian federal employment as a percentage of nonfarm payroll employment
Source: US Bureau of Labor Statistics
The share did stop falling two decades ago, but it remains quite small. There’s been some talk about the significant role government employment has played in recent job gains in the US, but that’s almost all state and local government jobs, which together make up almost 13% of nonfarm payroll employment. Federal government employment just doesn’t amount to much.
State and Local Governments Have Been Driving Job Growth
Government share of total growth in nonfarm payroll employment
Source: US Bureau of Labor Statistics
How to square the shrunken federal employment footprint with federal spending that, at more than 23% of gross domestic product over the recently ended fiscal year, is higher than it ever was from 1947 through 2008? One answer has to do with jobs created through federal contracts and grants, which by the occasional estimates of Paul C. Light, a New York University professor emeritus of public service, make up about as big a share of the overall US workforce now as they did in the early 1980s. But the main explanation is that the US government has become, as the saying goes, “an insurance company with an army.” Most federal spending is transfer payments either directly to individuals (Social Security, food stamps, etc.) or to their health-care providers (Medicare, Medicaid, etc.). The employee payroll is by comparison insignificant.
Paychecks Are a Small Part of Federal Spending
Compensation costs and other federal spending, fiscal year 2022
Sources: Congressional Budget Office; Department of Defense; White House Office of Management and Budget
Note: Military personnel compensation does not include spending on the Veterans Administration, which was $273.8 billion in FY 2022.
While these statistics show stagnant federal job growth and limited savings potential, they don’t really answer whether federal employment or pay are too high or too low. I’m not sure any statistics could — in reaction to a column of mine years ago about whether federal employees are over- or underpaid relative to their private-sector peers, the libertarian economist Arnold Kling wrote a blog post with the headline “Justin Fox Inadvertently Makes the Case Against Empiricism.”
Even if empiricism doesn’t offer all the answers, though, it can be pretty interesting. For example, wouldn’t you like to know where all those government workers work?
The White House Office of Management and Budget publishes a handy breakdown of full-time-equivalent civilian executive-branch employment every year as part of the president’s budget proposal, so that covers most of them. But I was curious about the others. The budget also contains scattered FTE numbers for the judiciary branch and legislative branch agencies other than Congress itself, which the Congressional Research Service has covered. The Defense Manpower Data Center has numbers on active-duty military that I could have broken down more than I have below, but I figured they’re not really the focus of this analysis. For the Postal Service, I’ve used BLS numbers. The semi-private Federal Reserve System publishes its own employment numbers, and for the Central Intelligence Agency and National Security Agency — for which such information is classified — I’ve taken estimates found on Wikipedia and rounded to the nearest 10,000. Also, yes, I realize that this and subsequent tables don’t look great on mobile phones, but figured even readers on the go might appreciate all the detail.
Where the Federal Workers Work
About 60% of federal jobs are defense-related
Sources: White House Office of Management and Budget; Defense Manpower Data Center; US Bureau of Labor Statistics; Congressional Research Service; Federal Reserve Board; Wikipedia
Note: Numbers are full-time equivalents as of FY 2023 except for active-duty military (Aug. 2024 headcount), USPS (total employees, Oct. 2024), House (total employees, 2023), Senate (total employees, 2022), CIA and NSA (undated estimates).
One persistent and bipartisan policy misperception discussed in Thorson’s book is that the US government spends more on defense than on health care. In fact, it spends more than twice as much on the latter. But if you look at how employment is distributed, defense truly is No. 1 (albeit with health care for veterans accounting for many of the jobs).
There’s a similar disconnect between employment and regulatory reach among the different parts of the federal government, one that would-be bureaucracy slashers ignore at their peril. In a Wall Street Journal op-ed this week, Ramaswamy and his DOGE co-chairman, Elon Musk, focused on their hopes of slashing the country’s regulatory burden and argued that mass job cuts would naturally follow. “The number of federal employees to cut should be at least proportionate to the number of federal regulations that are nullified,” they wrote. “Not only are fewer employees required to enforce fewer regulations, but the agency would produce fewer regulations once its scope of authority is properly limited.”
That doesn’t really follow, though, as one can see by comparing employment at cabinet-level agencies with how many words they’re responsible for in the Code of Federal Regulations — an imperfect but nonetheless revealing measure of regulatory reach. In fact, there’s a very slight negative correlation between department-level employment and regulatory output, with an r-squared of 0.009 (where one means perfect correlation and zero no relationship at all).
No Connection Between Regulatory Output and Employment
Share of total by cabinet-level department, 2022
Sources: QuantGov; White House Office of Management and Budget
Note: Regulations jointly proposed by multiple departments not included
I left out independent agencies for simplicity’s sake, but the independent Environmental Protection Agency makes this point even more strongly, if anything. It’s the second most-prolific regulator after the Treasury Department, responsible for 12% of the words in the CFR in 2022, but accounts for just 0.7% of executive-branch civilian employment. A small group of federal workers can produce a ton of regulations, while most federal workers don’t spend their days writing or even enforcing regulations.
What do they spend their time on? The annual Occupational Employment and Wage Statistics compiled by the BLS offers another imperfect but revealing portrait. Here’s a breakdown of civilian federal employment by the 22 major BLS occupational categories:
What Federal Workers Do and Earn
Civilian executive branch employees by occupational group, May 2023
Source: US Bureau of Labor Statistics, Occupational Employment and Wage Statistics
These occupational categories are defined for the labor force as a whole, which is how we get the somewhat incongruous result of “business and financial operations” topping a ranking of government workers. It’s bean counters, to a large extent. The fact that this category plus management and office and administrative support together make up three of the top four occupational groups and 46% percent of the federal workforce does tell you something important: Federal work is disproportionately office work. Nationwide, just 26% of workers fall into those three groups. This lends some support to the notion that the federal government could make do with fewer workers, especially as large language models and other artificial-intelligence technologies threaten to supplant much office work.
Looking at the most common occupations among federal workers gives a different picture, although a lot depends on the occupational definitions. For example, I’ve chosen to rank “broad” occupations here rather than “detail” ones because so many of the top detail occupations are things like “business operations specialists, all other.” But there were an estimated 191,790 business operations specialists, all other, in the federal executive branch in May 2023, more than any of these broad occupations.
What Federal Workers Do, Another View
Biggest federal executive-branch occupations, March 2023
Source: US Bureau of Labor Statistics
Note: Sorted by "broad" occupational categories.
What this list highlights, really, is the diversity of federal employment — highly paid doctors and lawyers alongside low-paid health aides and cleaners. These people also work and live in diverse locations, with less than a quarter in the District of Columbia, Maryland and Virginia. A recent report from the Congressional Research Service broke down federal employment by congressional district, and while the biggest concentrations are found near Washington (in the District of Columbia proper, 18.5% of those employed work for the federal government), there are also high-employment zones in the South and Mountain West, among other places.
Where the Federal Workers Are, by Congressional District
Civilian federal workers as a percentage of total workers, 2023
Sources: Congressional Research Service; US Census Bureau
The federal workforce is less diverse by age. That is, it’s quite heavy on workers in their 40s and 50s.
The Federal Workforce Skews Late Middle-Aged
Age group shares of total employment, March 2024
Sources: US Office of Personnel Management; US Bureau of Labor Statistics
There are tons more such basic facts about federal workers available, and I must admit that I’m not quite sure where they all lead. The ones detailed here do leave me with the impression, though, that Vivek Ramaswamy will find cutting a million federal jobs in a year much harder than he claims to think it is.