The Federal Workforce Is Smaller Than You Think, and That Is the Problem

Here is a fact that ruins half the speeches you have heard on both sides of the aisle: the federal civilian workforce today is only modestly larger than it was when Eisenhower left office. Conservatives who say "bloated bureaucracy" and liberals who say "dedicated public servants" are both arguing about a number that has hardly moved.

Meanwhile the thing the workforce administers has grown past recognition. That mismatch is the story. If you want to shrink the government, you cannot do it by counting desks. You have to count programs.

The claim, defined

Let me define terms, because this subject is full of numbers that mean different things.

Federal civilian employment counts people on the federal payroll who are not in the armed forces. It can include or exclude the Postal Service, which is large, self-funded in the main and a different kind of enterprise. The Office of Personnel Management publishes historical tables of federal civilian employment, and the usual way to compare eras is to look at the executive branch outside the Postal Service.

Federal outlays are what the government spends, measured in the budget tables that the Office of Management and Budget publishes as the Historical Tables, which run back to 1940 and include Table 1.1, "Summary of Receipts, Outlays, and Surpluses or Deficits."

The proxy state is a phrase I take from the political scientist John DiIulio, who wrote the 2014 book Bring Back the Bureaucrats and has argued that the federal government carries out much of its work through a "proxy" workforce: contractors, grantees, and state and local governments paid with federal dollars. The count of people who actually draw a federal paycheck is the visible tip. Most of the work is done by people who do not.

My claim has three parts. One: the number of federal civilian employees has grown by only about a fifth in sixty years. Two: federal spending has grown enormously over the same period. Three: the gap between them is filled by contractors, grantees and the states, and that is why cutting positions does not cut government.

The arithmetic

Two figures from OMB's historical tables, which I would ask you to check for yourself because I can only give you what I am confident of and you should not take my word for it.

Federal outlays in fiscal year 1960 were about $92.2 billion. In fiscal year 2019, the last full year before the pandemic distorted the tables, they were about $4.45 trillion.

Divide. $4,450 billion divided by $92.2 billion is about 48.3. So nominal federal spending grew roughly forty-eight-fold.

Now be fair about inflation, because a point-by-point argument does not let the other side say you ignored it. Consumer prices rose by a factor of about 8.6 over that span, from an index of 29.3 in January 1960 to 251.7 in January 2019. Divide forty-eight by eight and a half and you get something near six. Even after adjusting for inflation, spending grew by a factor of five or six. Population grew too, from about 181 million to about 331 million, a factor of 1.8 or so. Divide six by 1.8 and you still get a real increase of more than threefold per person. The government spends more than three times as much per American, in constant dollars, as it did in the Eisenhower era, and I have used generous round numbers in the other side's favor.

Now headcount. The Bureau of Labor Statistics' all-employee series for the federal government, which includes the Postal Service, shows about 2.33 million in January 1960 and about 2.81 million in January 2019. Over those same years, outlays grew roughly forty-eight-fold while that count grew by about a fifth. OPM publishes its own historical tables that leave out the Postal Service, so check the table for the exact year and definition you want. For the arithmetic below I use a round two million, which is on the low side, so treat the per-employee figures as illustrations; the true per-employee multiple is a little smaller.

Put the two together. If the workforce is about two million and nominal spending in 2019 was $4.45 trillion, the spending per federal employee is $4,450 billion divided by two million, which is $2.2 million. In 1960 the same division gave $92.2 billion divided by two million, or about $46,000. Nominal, yes, so not a fair comparison without the inflation adjustment I made above. After it, each federal employee's desk sits over several times as many real dollars as in 1960: using the BLS counts above, roughly four to five times. The workforce did not grow with the budget. The budget grew around the workforce.

Where the money goes, and who spends it

Most of that spending is not salary. Big federal programs are transfer programs: checks to individuals, as in Social Security and the retirement and disability programs; payments to providers, as in Medicare; and grants to states, as in Medicaid. A relatively small team at the Social Security Administration administers benefits to tens of millions of people. A federal agency may send a grant to a state, which hires its own employees to run the program under federal rules. The federal government writes the rules, writes the check and takes the political credit. The state or the contractor does the work and takes the blame.

DiIulio's point is that this arrangement is the hidden structure of the modern federal government. The president and Congress can tell you the number of federal civilian employees with some precision. They cannot tell you with equal precision how many people work on federal money for a private firm or a state agency, and the figure is a multiple of the civil service. Paul Light, another scholar of the public service, has written about a "shadow" government of contractors and grantees alongside the official one. I will not give you his figures, because estimates vary with how they count, and I would rather you look them up than trust my memory.

The strongest objection

The best case against my argument goes like this.

The objection says: if the federal workforce is flat while the country's population nearly doubled and the budget rose many times over, then the federal workforce has become more efficient, which is what conservatives are supposed to want. A flat headcount is a sign of discipline. Cutting it further would reduce the government's capacity to deliver the benefits Americans have been promised, to inspect food, to land planes and to process claims. The right place to look for savings is the programs themselves, the objector says, and Congress, not the civil servants, decides which programs exist. The employees are not the problem. They are the people who carry out the laws.

I agree with almost every sentence of that. It is a correct statement of where the decision lies, and it makes my case for me. The workers did not create the programs. Congress did. The problem is not that the people in the buildings are too many. The problem is that Congress has created and funded so many programs that the government cannot run them without an enormous shadow workforce and that Congress does not review whether the programs work.

The argument, in numbered steps

1. The budget is a measure of commitments, not of staff. The forty-eight-fold nominal increase measures the promises Congress made and the transfer programs it created. Those are statutes. They run on autopilot, and the payroll is a small share of the cost.
2. Cutting headcount without cutting programs does not save money; it shifts the cost. If an agency loses a hundred federal employees and still must deliver the same program, it will hire contractors, who cost the taxpayer more per hour in some cases and who are not subject to the same oversight. Washington has a long record of this: when a hiring freeze or a headcount cap is imposed, the work flows to contractors and the budget does not fall.
3. A "proxy" government is less accountable than a direct one. A federal employee can be called to testify, is covered by the Inspector General Act, is subject to the Freedom of Information Act and the civil service rules. A contractor or a grantee is reached only through the contract or the grant agreement. The more of the work is done by proxy, the harder it is for a voter or a member of Congress to find who is responsible.
4. The Founding was explicit that a legislature should answer for what it enacts. Madison wrote in Federalist 62 that the laws must not be so voluminous that they cannot be read, and in Federalist 51 that the government must be obliged to control itself. A Congress that creates programs and hands the work to grantees and contractors has found a way to take credit and avoid the blame.
5. Therefore reform is about programs, sunsets and reviews. Every program should have a defined purpose, a measurable result, and an end date unless Congress votes to renew it. That is how you cut government: you stop doing things, and you stop paying people, in agencies and outside them, to do those things.

What a serious cut would look like

If someone tells you they will cut government by cutting positions, ask three questions.

First: which programs will end? A statement that "we will reduce the workforce by ten percent" is a statement about people, not about programs. If the programs continue, so will the work, and so will the cost.

Second: what happens to the work? If it moves to contractors, you have changed the label on the cost. If it moves to the states, you have changed who sends the bill. Both can be good ideas, but they are not savings.

Third: who decides? If the answer is the agency, the answer is wrong. The Constitution gives the power of the purse to Congress. Article I, Section 8 gives Congress the power to lay and collect taxes "to pay the Debts and provide for the common Defence and general Welfare of the United States." Article I, Section 9 says no money shall be drawn from the Treasury but by appropriations made by law. A real cut is a vote.

I want to be careful on the other side of this ledger, too. There are real arguments for fewer federal employees in certain places. Duplicated functions are real. A thicker layer of management than the work requires is real. A hiring system that takes months to fill a post is a real failure and, as it happens, one that the civil service rules help produce. A conservative reformer should want a federal workforce that is lean, well paid for good work, and accountable. Those goals do not conflict with the claim in this post. They are the claim.

What the debt clock says about it

If you want a reason to take this seriously, look at the national debt clock. The Treasury's figures put the total public debt at $40.56 trillion on November 5, 2026, up from $38.05 trillion on November 4, 2025. The difference is about $2.51 trillion in a single year: $40.56 trillion minus $38.05 trillion equals $2.51 trillion. No headcount reduction at the margin, even a deep one, comes close to that scale. If every one of two million federal civilian employees were paid $100,000 a year in salary and the entire workforce were eliminated, the saving would be two million times $100,000, or $200 billion. That is about eight percent of one year's debt growth. I chose a round figure that is higher than a realistic average salary, and I did it in order to give the headcount argument every advantage. The point survives. The debt is a story about programs and promises, not about people at desks.

For a wider look at why government grows when no one is counting, see Conservatives Are Not Against Government, We Are Against Rule by Bureau. For a reform that forces the question of each agency's existence, see Every Agency Should Have to Justify Its Existence Again, the Texas Sunset Way. And for what happens when the work is handed to private hands under a grant, Block Grants or Categorical Grants, and Why the Strings Matter Most explains why the strings matter.

Measured against the Founding

The Framers feared an executive with a standing army of dependents. They also feared a legislature that could hide its choices. Madison's Federalist 10 warned of factions, and a proxy government is a government of factions: contractors, grantees, associations and agencies that depend on a program and have every reason to see it continue. When a program has a constituency that is paid by the program, the program gains a lobby, and the lobby does not appear in the federal employment tables.

That is how a republic ends up with a government much larger than its visible payroll. Not by conspiracy. By incentives. Every program creates a set of people with a living at stake, and every one of them will argue, sincerely, that the program is essential.

The answer is not contempt for those people. It is a rule that every program must justify itself on a schedule, in public, to the elected representatives of the people who pay for it.

The case in brief

1. The federal workforce grew only about a fifth between 1960 and 2019 while nominal federal outlays grew about forty-eight-fold, from roughly $92.2 billion in 1960 to roughly $4.45 trillion in 2019.
2. Washington governs through a proxy workforce of contractors, grantees and state governments, as John DiIulio argued in 2014, so headcount understates the size of the government.
3. Cutting positions without cutting programs shifts costs rather than saving them, and it moves work away from the people who are easiest to hold accountable.
4. The Constitution assigns the power of the purse to Congress, so the real lever is the program: sunsets, reviews and votes.
5. The debt grew by about $2.51 trillion in one year, a scale that no payroll reduction can reach.

What you can do: Choose one federal program you pay for, look it up on the numbers page, and ask your representative through My Representatives when Congress last voted on whether it should continue.