The Bureaucratization of the Federal Government in 15 charts

If you've been following my work, you know that I've long been concerned about the steady bureaucratization in the U.S. economy. My focus has recently shifted to understanding how this trend is sapping the vitality of public institutions and, more importantly, what can be done to reverse it.

One of the most telling signs of bureaucratization is the proliferation of managers and administrative functions such as HR, Finance, Contracting, Legal, and Program Management. As these roles multiply, they create an ever-expanding web of layers, silos, rules, and processes. This post presents evidence of this dynamic in the federal government based on detailed occupational data, illustrated with fifteen handy charts. (For more on methodology, see below)

Overall trends

 
ONE. Four out of every ten federal employee is a manager or an administrator. This includes 250,000 managers directing the workforce (93,000 of whom are administrative supervisors). Non-managerial staff in HR, Finance, Legal, Contracting, and Program Management functions make up 25% of federal employment. The largest administrative category, “Other Administration and Program” is a catch-all that includes administrative officers, program specialists, program managers, and coordinators.

TWO: Since 1998, management and administrative roles have driven all employment gains in the U.S. federal workforce. While this mirrors a broader pattern in the U.S. economy, the trend in the federal government is more pronounced. For those who think the ranks of federal employees are multiplying, they are—but not, for the most part, in “line” roles that directly deliver services to the American people.

THREE. There’s little evidence of managerial and administrative economies. In fact, the opposite is likely. Discretionary spending (annual funding that Congress must actively approve, such as defense and education) per manager and administrator fell 15% in the last ten years compared to the prior decade. The “dollar per bureaucrat” ratio is marginally better for non-discretionary programs like Social Security, but this is probably due to mandated payment increases rather than administrative efficiency (see the COVID-related spike in 2020-21).

FOUR. The managerial and admin productivity decline is clear at agencies like the National Institutes of Health, which rely on discretionary budgets. Here, the growth in core administrative, financial and contracting occupations has substantially outpaced funding over the last twenty years. The same pattern emerges in research awards per manager and administrator.

FIVE. The Department of Defense, which receives the largest share of discretionary spending, tells a similar story. Each civilian manager and administrator today handles significantly fewer procurement and R&D dollars than 15 years ago. Effectiveness doesn’t seem to have increased either. The ranks of DoD finance, legal and contracting staff have grown by more than 50% over the last few decades, and yet the Pentagon just failed its seventh audit in a row. The bureaucratic footprint has also grown relative to military staff: today, there are half as many soldiers per manager and administrators as in 1998.

Bonus chart: The trend becomes clearer when we track the evolution of staffing and spending over time. As Defense spending declined after the Iraq and Afghanistan deployments, you might expect administrative staffing to shrink as well. Instead, the opposite occurred. Program management, contracting, and other administrative roles grew steadily even as the budgets they managed shrank.

Administrative occupation trends

 
SIX. Administrative growth is pervasive. HR staffers, budget analysts, program managers, coordinators, contracting specialists, lawyers, and related roles have increased their share employment across agencies.^[1]

SEVEN. All administrative functions have grown substantially over the last 25 years. Finance expanded by 50%, notable given technology should have improved accounting efficiency. Program management positions had the steepest rise, from 46,000 to 100,000 staff. Legal and contracting positions grew nearly as much, increasing from 68,000 to 128,000 people—an 88% jump.

EIGHT. This shift has elevated administrative expertise over technical capability. Administrative positions increased by 220,000, triple the growth in engineering, mathematics, science, and IT jobs. Removing IT—which includes systems administrators—widens this gap. The government added more finance and HR specialists than engineers, statisticians, and scientists in physical and biological sciences combined.

NINE. This trend applies across all agencies that rely on STEM talent: the National Institutes of Health (NIH), the Food and Drug Administration (FDA), the National Science Foundation (NSF), the Centers for Disease Control (CDC), DoD's central units, as well as NASA.

Management occupation trends

 
TEN. The manager^[2] share of federal employment increased three percentage points, from 12 to 15%. Administrative managers grew their ranks more than twice as fast as their line counterparts. There’s little doubt that senior administrators have become more influential over the last few decades.

ELEVEN. Managerial growth means that the span of control has shrunk by almost a quarter. Up until the early 2000s, there were 7.3 employees for every manager; that ratio now sits at 5.7.

TWELVE. Management positions have increased across federal agencies. Several organizations show a 4-5 percentage point increase in supervisor and manager employment share over the past 25 years.

THIRTEEN. The higher up you are, the better your performance ratings. This chart breaks out non-managers into three categories used by the Office of Personnel Management: those with managerial titles but no significant oversight responsibilities (CSRA), a small group (about 40,000 people) in 'leader' or 'team leader' roles who monitor clerical workers or guide professional staff through projects, and all other employees (1.4 million). Over time, higher ratings mean more promotions--and a lopsided pyramid.

FOURTEEN. The “thickening” of the managerial hierarchy is greatest at the very top. Executive Schedule positions—key leadership roles that require Presidential appointment and often Senate confirmation—have multiplied in the last quarter century. The Department of Defense is a good example of this growth: its leadership positions expanded from 34 to 61, mainly through new Assistant Secretaries and Deputy Undersecretaries, which added 16 and 7 slots respectively. Each new senior box in the org chart generates cascading layers of subordinate boxes—Principal Deputy Assistant Secretary, Director, Deputy Director, etc—which lead to an exponential growth of management roles throughout.

FIFTEEN. New York University professor Paul Light found that there are nearly five times more organizational layers among senior roles in the federal government since the 1960s. His 2020 count found 1,070 deputy assistant secretaries, 236 assistants to the assistant secretary, 204 deputy deputy assistant secretaries, and 153 deputy assistant assistant secretaries.

Why is this happening?

 
Fully doing justice to this question is a topic for another day. But let's consider five explanations for why the ranks of administrators and managers are expanding across federal government, beyond the obvious shifts I've already noted like increases in spending, or the creation of a new large department like Homeland Security (not explicitly covered in this analysis since it doesn't change the results).

Hypothesis 1: More complex work

Some might contend that the proliferation of administrators and managers is a natural outgrowth of an increasingly knowledge-driven economy. As routine clerical and technical tasks are automated, the argument goes, a larger share of the remaining workforce is engaged in skilled, judgment-based work that demands greater coordination and "coaching" from experienced managers.

This view fails to explain why administrative employment has outpaced growth in high-skilled professional roles in engineering and science occupations (see chart 7). It's also far from self-evident that a knowledge economy requires more support staff, managers, and layers. Indeed, thinkers like Peter Drucker have made the opposite case, as have we. Large organizations as diverse as Haier (appliances), Nucor (steel making), Vinci (infrastructure), WL Gore (materials), and Buurtzorg (health care) show that it's possible to be highly effective and efficient with super-lean administrative functions and minimal hierarchy.

More layers and process is the prevailing response for handling complexity at scale for a number of reasons: it's what most other large institutions do, so it's viewed as inevitable; it's the typical "best practice" prescription from consulting firms and professional schools of business and public administration; and it conforms with expectations of regulators who mistake the trappings of bureaucracy–a chief compliance officer, compulsory training, comprehensive reporting–for true accountability and control. But it's definitely not the only response available, and almost surely not the best one.

In fact, there’s little evidence that more federal administrators and managers translate into improved organizational effectiveness. A 2015 paper by researchers at Duke, Princeton, and Vanderbilt found that increased numbers of administrators were actually associated with lower performance, measured by agency scores on the Performance Assessment Rating Tool, or PART. PART was developed by the Office of Management and Budget during the George W. Bush administration and has since been discontinued, so this isn't conclusive evidence. But it's consistent with plenty of other evidence which I hope to get into in future posts.

Hypothesis 2: Title and grade inflation

Job changes within the federal workforce could explain some of the growth in administrative positions, especially when white collar support roles (which the Office of Personnel and Management defines as "Technical") are reclassified into administrative ones, as seems to have happened with roles like Tax Specialist. Alternatively, people could be switching from technical occupations to adjacent administrative roles (e.g., from Legal Assistant to Paralegal Specialist).

The same paper I cited earlier about the link between administrative growth and performance also looked at this dynamic, and found some evidence that these shifts are happening, especially among younger and relatively more educated technical workers. But the effect is modest, especially given that we're focused on a narrow and stable set of "core" admin occupations. For instance, only 17,000 employees made the move from the technical "General Clerk and Assistant" role to "General Administration and Program" over 23 years (1988-2011).

The DoD's conversion of military into civilian jobs between 2004 and 2010 is another factor to consider. The Congressional Budget Office (CBO) notes that this added 32,000 civilian roles (the General Accounting Office reports that in this period some contractors also became permanent staff, though it's not clear whether these are in addition to the 32,000 new roles mentioned by the CBO). We don't have a clear breakdown of what jobs former military or civilian contractors moved into, but even with the generous assumption that two-thirds of the "civilianized" positions were administrative, that accounts for only 21,000 jobs. That's about a third of the admin growth that time period, and less than a fifth of the total expansion of the administrative ranks between 2004 and 2023. This influx might explain some, but certainly not the majority, of the dynamic (and it certainly doesn't account for any of the administrative growth outside of DoD).

For managers and supervisors, grade creep is likely inflating the count. A report by the Merit System Protection Board makes the point that a supervisor’s grade is pegged to the grade of their subordinate: in other words, the more high-level people you oversee, the higher your status (and the bigger your salary). This creates a powerful incentive for agencies to shift certain employees into managerial roles. I'm sure that in many cases this is a well-intentioned attempt at rewarding and retaining capable employees, given the lack of viable alternatives. But this is hardly the best way to recognize contribution, and it ends up creating unnecessary roles and layers.

Hypothesis 3: More contract work

NYU’s Paul Light and others have documented the federal government's greater reliance on private sector contractors, state and local governments, and other third parties to carry out its functions. This can create additional administrative work for the permanent workforce—setting requirements, drafting contracts, ensuring compliance, and reviewing performance.

There are several reasons behind the federal government's embrace of contracting: a sluggish hiring process, skills gaps in critical occupations, political pressure to limit the size of the federal workforce, and a belief that offloading work to third parties improves efficiency and flexibility. Unfortunately, this "asset light" strategy doesn't always pay off, since it can erode institutional knowledge and practical expertise. There are several examples of this in the public sector, which are wonderfully documented by Jen Pahlka (founder of Code for America and U.S. Deputy Chief Technology Officer during the Obama Administration) in her excellent book, Recoding America. This pervasive trend also affects the corporate sector (just look at Boeing).

The contracting explanation has some merit, but it doesn't account for the growth in administration and management over the last decade, as contract work has stabilized or even declined in real terms. Consider again the DoD data in the “bonus” chart: procurement spend is currently in at 2007 levels, yet the number of program managers has increased by 60%.

Hypothesis 4: Increased legalism

There's a growing consensus that America's legal framework for government action relies excessively on prescriptive rulebooks and formal procedures, and subjects agencies to layers upon layers of challenges and review. It's no wonder we often get inflexible policy making and slow and kludgy implementation (the National Environmental Policy Act, or NEPA, is a good example). Philip K. Howard has long made this argument (for instance, here and here), along with other voices across the political spectrum like University of Michigan's Nicholas Bagley and Niskanen's Brink Lindsey.

It's easy to see how a legalistic approach to policy making and execution can lead to more administrative staff. Agencies need more lawyers to anticipate or respond to risks and more contracts officers to ensure agreements are properly formulated (consider that there are over 2,000 pages in the Federal Acquisitions Regulations manual).

On the other hand, many of the legal changes in question happened in the 1960s and 1970s, well before the time frame of our analysis (another bastion of legal proceduralism, the Office of Information and Regulatory Affairs or OIRA, was established in 1980). The raft of regulations also seems to have slowed down since those days. For instance, the number of final rules published in the Federal Register (the government's official publication for rules, executive orders, and notices) in 2023 was 2,803, 40% lower than in 1998 and 65% lower than 1980. A Congressional Research Service report finds that the number of "major" rules that need to be reviewed by Congress (such as those having an impact in excess of $100 million) oscillates in the 60-80 range for most years between 1997 and 2018. It’s worth keeping in mind that legal and contracting personnel has kept expanding even as contact spend (and presumably the number of contacts and related legal matters to engage on) has plateaued and even declined in the case of the Department of Defense.

This isn't to say that overbearing legalism isn't a problem or to discount the pernicious effect of a steady stream of rules accumulating over time. But my best guess is that the impact is more indirect–and reciprocal. As Howard observed, the prevailing statutory and regulatory paradigm was inspired by bureaucratic principles like standardization (uniform and precise rules) and formalization (clearly circumscribed and fixed roles). It's not surprising that a legal framework informed by bureaucracy provides fertile soil for bureaucratic management practices. And this brings us to the final hypothesis...

Hypothesis 5: Bureaucratic capture

Expansion in management and administration is the inevitable outcome in organizations that are run with a bureaucratic operating system. Here are the defining elements of that model:

As bureaucratic organizations become larger (and older), power almost invariably shifts to the “suits”—the taskmasters, controllers, accountants, and analysts who are charged with maximizing control, alignment, and efficiency Here's how:

• When the organization expands, layers get added and the ratio of managers to front-line staff goes up. Over time, the proportion of employees who have a direct customer-facing role goes down. As Mark Zuckerberg recently wrote, you end up with “a management structure that's just managers managing managers, managing managers, managing managers, managing the people who are doing the work.”
• In most organizations, your power and compensation are the product of headcount and budget. The tendency is therefore to find all sorts of reasons to expand both. No one ever downsizes their empire voluntarily, either. This explanation is consistent with managerial grade inflation (see Hypothesis #2).
• Internal boundaries multiply and become more rigid over time, creating the need for roles and processes for obtaining approvals and coordinating action. For a great real-world illustration, see this quote from a Defense Business Board report on program managers:

Many DoD program managers, senior officials, and others who were interviewed believe that PMs spend the majority of their time going from meeting to meeting and answering the same questions amongst the various offices. PMs spend too much time managing the politics and the “process” within DoD rather than managing their specific program.

• Powerful staff groups blanket the organization with policies and rules, which typically don’t have a sunset clause. As a result, the clog of red tape grows ever bigger. Moreover, internal service providers can’t be fired by their so-called customers, which means their incentives for improving the efficiency and quality of their offerings are muted. (Note: It's undoubtedly true that external factors, in this case guidance from Congress and the White House, often drive the creation of new positions and oversight requirements. But in my experience, the majority of process and rules are generated internally).

• Every new challenge begets a new CxO or head office unit (remember those new 23 new Assistant Secretaries and Deputy Undersecretaries the DoD added over the last few decades?) These soon become permanent fixtures. Similarly, authority moves to the center during a crisis– and stays there.

• Finally, an administrative aristocracy emerges. What often distinguishes senior people from frontline colleagues is not creativity, foresight or technical expertise, but mastery of administrative arcana—how to develop a plan, set performance goals, build a budget, interpret financial results, coordinate a project, conduct a performance review and give employee feedback. An organization that selects for these skills will, over time, become dominated by administrators.

The last item is subtly related to the hollowing out of expertise in government (Hypothesis #3). Administrators will have a penchant to solve problems via administrative solutions. For instance, when faced with a challenge that requires deep engineering or software development capabilities, decision-makers who lack technical expertise might prefer outsourcing the work to outside of contractors. It's just easier to model the project in a spreadsheet and send it out for bids than to wrangle a team of "geeks."

An increasingly influential administrative class, likely composed of people who got ahead by not rocking the boat, will also tend to favor narrow interpretations of existing rules and regulations (Hypothesis #4), thereby creating more inflexibility and conservatism. If the organization selects for and promotes people who excel at ensuring compliance and being compliant, it will fill up with risk-averse, micromanaging administrators. The "irregular" people who are most likely to drive positive change through initiative and off-script ingenuity are marginalized or end up leaving. Pahlka illustrates this point perfectly in Recoding America, when she writes that existing pathways to expedite procurement decisions are largely ignored because of "a culture in which if even one person prefers safe over sorry, we’re back to the way it’s always been done."

As these dynamics play out over time, the end result is not only managerial and administrative overhead but also reduced vitality – daring, resilience, creativity, and engagement. Bureaucratic institutions have a tendency to ossify, as RAND's Michael Mazarr aptly puts it.

More to come on this front. In the meantime, I'd love to hear your thoughts:

• Which of these data is most surprising? Most concerning?
• How do these trends jibe with your own experiences in government or other large organizations?
• Which explanations are most convincing? Are there others you’d add?
• What am I missing?

I'm eager to get your take in the comments section. Alternatively, feel free to reach me at micheleatmanagementlab.org.

A note on methodology

This analysis was made possible by the support of the Office of Personnel Management (OPM), which helped assemble a detailed dataset on the federal workforce over time (a great example of responsive government!). A few notes on scope: the OPM numbers do not include military, foreign service, intelligence personnel, the Postal Service, and the presidentially-appointed government executives (though the final few charts will address this last group). With one exception, the Veterans Administration is also not part of the analysis, as its supervisory role data shows inconsistencies over time. The manager count doesn't include 45,000 federal employees with managerial titles who don't directly supervise staff.^[3] My approach to categorizing administrative occupations differs from OPM's—they include operational roles like park rangers, FBI agents, IT specialists, and compliance officers,which are excluded here.^[4] These choices make both the manager and administrator counts deliberately conservative.