One answer is that an extended state bureaucracy can become relatively independent of elected officials and resist their agendas. In its extreme form, it is sometimes called “the deep state.” This even more pejorative term found its source in Turkey half a century ago. It gained currency in America about a decade ago when the MAGA crowd (later picked up by Trump himself) used it to criticize career civil servants pursuing their own agenda with their legal protections and their powerful agencies (The Economist 2017). But the US version of the concept, without the label, dates back at least to Richard Nixon’s “administrative presidency” effort to take over the administrative state (Nathan 1976; Howell & Moe 2025).

Beyond issues of constitutional interpretation in the United States (which will not be discussed here, but see Wallison 2026), the administrative state has long been a source of concern, at least in the free world. This article examines the administrative state in light of public choice theory and its offshoot, constitutional political economy (Brennan & Buchanan 1985), which focuses on the design of institutions in a free society.

The Invasive Administrative State

At the end of the second volume of his Democracy in America (1840), Alexis de Tocqueville tried to discern “what type of despotism democratic nations have to fear.” He concluded it will be a “sort of servitude, regulated, mild, and peaceful,” an “administrative despotism.” “Combined with the external forms of liberty,” it could be “established in the very shadow of the sovereignty of the people.” Tocqueville saw administrative despotism as covering “the surface of society with a network of small, complicated, minute, and uniform rules.”

Start with the Federal Register, the daily repository of all proposed and final federal regulations (called “rules”) and other documents that have regulatory implications, such as notices, standards, and presidential documents (including executive orders and proclamations). A frequently used measure—imperfect but objective—of the volume of annual federal regulation is given by the total number of standardized pages in the Federal Register. In 1976 (the first year we have standardized page counts), the total was 50,505 pages. By 2024, the last year of Joe Biden’s presidency, the annual output had more than doubled to 106,109. (Except when otherwise mentioned, all statistics on regulation come from Crews 2026.) In between was a nearly unbroken string of ever higher page counts.

In 2025, the first year of Trump’s second term, the Federal Register totaled just 60,917 pages (of which 7,648 pages were due to the last breath of the Biden administration before Trump’s inauguration in late January). This was a remarkable decrease in rulemaking activity. Except for Trump’s page count of 61,314 in 2017 (another atypical decrease), this was the lowest addition to regulations (measured this way) since 1992.

Have Americans thus found in Trump a deregulator-in-chief? Not necessarily, if only because the Federal Register only describes the flow of (or change in the level of) regulations. Even if the flow is small, the stock of regulations continues to grow. On the other hand, a deregulatory action also requires new regulation, which means that the Federal Register also includes deregulatory rules, so perhaps Federal Register pages aren’t the best metric.

A measure of the stock of regulations is given by the size of the Code of Federal Regulations (CFR), which consolidates the final rules of departments and agencies, and some presidential regulatory texts. During Trump’s first term, the number of pages in the CFR increased from 185,131 pages (in Barack Obama’s last full year as president) to 186,645 pages (the last full year of Trump I), a growth of 0.7 percent. That is remarkably little growth compared to other recent presidencies, but it is still growth. In 2023, the last year the CFR is available, it totaled 190,627 pages, filling 245 bound volumes.

The Mercatus Center’s RegData database provides another measure of federal regulation by counting the number of “restrictions”—that is, occurrences of the terms shall, must, may not, required, and prohibited in the CFR (McLaughlin et al. 2023). This measure is tightly correlated with the number of CFR pages up to 2023, and it also provides data for 2024 and 2025. From 1970 to 2025, the number of restrictions in the CFR tripled, from 441,276 to 1,367,625. Figure 1 shows the RegData figures from Obama’s last year to the first year of Trump II. According to this measure, Trump I roughly plateaued the administrative state after initially boosting it 2.6 percent from Obama’s last year. In the first year of Trump II, this measure increased by 1.2 percent from Biden’s final year.

From all these figures, it is not clear at all that Trump reduced the weight of the administrative state’s regulations.

Dark matter and other critiques / Beyond the formal rules are what regulatory policy analyst C. Wayne Crews calls “regulatory dark matter.” This consists of various “dos and don’ts” that result from memoranda, guidance documents, bulletins, and circulars, only some of which appear in the Federal Register and generally don’t make it to the CFR. Moreover, all the numbers above don’t count the innumerable enforcement actions made by the state, both administrative and political, on antitrust, trade, finance, energy, health, and so forth.

Without the vast administrative apparatus of the state, regulations could not be as invasive. Elected officials could not, every year, write the roughly 2,200 final regulations issued by the Trump administration in 2025, not to mention the more than 3,000 issued in Biden’s final year and the more than 4,000 per year before 2005 (Crews 2026). (A “final rule” is the revised version of a proposed rule submitted to public comments.) In contrast to that output, elected federal officials are barely able to vote annual budgets and incapable of controlling the growing budget deficit.

Public choice analysis of government bureaucracies has shown the many inefficiencies and dangers of government bureaucracies. To summarize: Like ordinary individuals, the civil servant is primarily pursuing his own interest, although he may take an interest in other people’s welfare as he interprets it (Moe 1997). The original model by William Niskanen (and many of its successor models) argues that the bureaucrat (at least at a certain level) tries to expand the mission and size of his bureau because this is likely to improve his career opportunities, pay grade, perks, recognition, and prestige. Elected officials struggle to control the bureaucracy because it controls the information and influences the political agenda. Moreover, government bureaucrats form a non-negligible part of the electorate and are biased toward state power.

The Politicians’ State

Although a good case exists against the administrative state, we should not forget that it was created by politicians via statute. A majority of the voters may have agreed. Moreover, only powerful central agencies could administer the programs wanted by the progressives of the early 20th century. The administrative state can also be thought of as a commitment device that allows politicians to make credible promises because bureaucrats will be enlisted to fulfill them over time. Another factor is that the administrative state seems preferable to the political patronage and party machines of the 19th century (Moe 1997; Firey 2024). In short, bureaucrats are not the only problem; politicians are, too—and a big one at that.

Bureaucrats or politicians? / The most important question is, what would substitute for the administrative state? Because regulations at least aim to be non-arbitrary rules, the North Korean or Russian equivalents of the Federal Register and CFR don’t exist; they are replaced by the naked commands of the rulers. The United States and most Western countries are, of course, much closer to the ideal of a free society, but could it be partly because the administrative state is restraining the politicians’ state?

Commenting on the seeming (but not true) deregulation in the first year of Trump II, Crews writes:

The Trump administration has reduced rule output in unprecedented fashion while expanding industrial policy and federal steering more generally. These swamp things—also present during Trump 1.0—do not always show up in rule tallies, but they entrench government control just as effectively, and sometimes more so. (Crews 2025b)

This suggests that unrestricted politicians are more dangerous than the rule-bound administrative state.

Strongmen / The attack on the administrative state, or “deep state,” by such Western strongmen as Viktor Orbán (the recently defeated Hungarian prime minister who espoused “illiberal democracy”) or Trump (an ideological friend who openly supported Orbán) is meant to replace it with their decrees and commands. A strongman is an autocrat or a would-be autocrat. Howell & Moe (2025) characterize Trump as “the nation’s first strongman president,” although Nixon was a more timid and constrained precursor. (Franklin D. Roosevelt could also be characterized as a strongman, but the administrative state was ideologically on his side.)

The Trump I experience should have shown classical liberals that the administrative state is less dangerous than the whims of an autocrat or autocrat-in-training. The slower growth of formal regulations in America since around 2010, and especially under Trump, can be explained as the substitution of direct commands and threats for a heavy bureaucracy. In 2025, Trump issued 225 executive orders, the most by far in any year since at least 2016 and exceeding those of John F. Kennedy and George H.W. Bush during their entire four-year terms (Crews 2026). In short, naked commands and threats are a substitute for rule-based regulations and the rule of law.

Imagine a country where the president could, at his pleasure, change tariffs on imported goods and violate any previous trade treaty; regularly issue threats or commands to private domestic companies, even without legal authority; regularly pretext emergencies to grab more power; take government participations in private companies; threaten, sue, or otherwise intimidate private universities; start a war on his own decision, with fuzzy objectives and threats of war crimes; use his pardon power as a political wand, including for immunity promises to his henchmen (and henchwomen); let the Justice Department display on its headquarters a large banner of his portrait; order this Justice Department to prosecute his perceived political enemies; run a federal police force whose masked agents apparently need no judicial warrant nor are bound by habeas corpus, and apparently will receive immunity if they commit crimes and even murders; intimidate journalists and personally sue news outlets; surround himself with yes-men (and yes-women); purge or otherwise undermine bureaus that resist his demands; and act to overrule the electorate if they try to remove him from power.

Imagine also that this president is obsessed with increasing the money supply, notwithstanding the well-known danger of inflation in strongman regimes. How would you label such a country?

The experience of the Orbán regime’s substitution of a strongman’s will and hidden deals for administrative regulation sheds light on the detrimental effect of public corruption on market competition. It has been reported that a new plutocracy developed among Orbán’s cronies and their families (Dunai 2026).

The permitted and the forbidden / Formal regulations are called “rules” because that’s what they’re supposed to be, as opposed to ad hoc decrees from some central ruler. The multiplication of these rules is suspect because their enabling statutes are already supposed to be rules in the tradition of the rule of law (Hayek 1960) and should arguably be sufficient. If not, it may be that the legislature has delegated too much to the administrative state.

Take the “waters of the United States” (WOTUS) rule allowing federal regulation of certain water bodies, defined in the Clean Water Act of 1972 as “navigable waters.” (One might think that, at least in most cases and accounting for ordinary nuisances, private property incentives would be sufficient to “regulate” water, but this apparently does not occur to legislators.) Through regulations, the administrative state extended “navigable waters” to encompass often non-navigable tributaries and adjacent wetlands.

On this specific case, Trump came down on the right side of the issue (Adler 2019) with a revised regulatory definition in 2020, probably because farmers—who are especially burdened by WOTUS regulations—form an outsized component of his base. A 2023 Supreme Court decision also restricted the meaning of “navigable waters.”

The multiplication of regulations, promoted, written, and often enforced by administrative agencies, plus the dark regulatory matter that adds more interpretation have already undermined the liberal principle that everything not explicitly forbidden by law is allowed. With their whimsical commands, strongmen further strengthen the illiberal acceptance that everything not permitted by the government is prohibited. In a sense, strongmen are the administrative state gone mad.

Some Benefits of the Administrative State

The foregoing suggests another way to look at the administrative state. To the extent that it is, and remains, constrained by rules, it may block strongmen’s ambitions. George Mason economist Tyler Cowen has argued that tyranny could not be established in the United States because the federal bureaucracy is just too heavy:

The American government is so large and unwieldy [that] it is simply too hard for the fascists, or for that matter other radical groups, to seize control of. … Big government is useful precisely for (among other reasons) helping to keep government relatively small. … It is easier to take over a smaller and simpler state than it is to commandeer … a diverse, decentralized, and complex bureaucracy. (Cowen 2018; emphasis in original)

In his view, this check on power explains why Italian fascism was more moderate and less destructive than Nazi fascism.

Note how bringing the administrative state to heel is important for strongmen. Just like Orbán in Hungary, the populist PiS government in Poland used the politicization of the civil service as an effective weapon. Cayton and Rosenfeld (2025) observe, “Both Hungary and Poland’s governments have purged managers and professionals of the state sectors, replacing them with loyalists.” Trump has been trying to do the same.

Paradox / The paradox of the administrative state is that it can also prepare the way for a strongman. Observe how Trump’s takeover attempt on the American state is helped by many existing regulations—products of the administrative state—that lend a sort of varnish to the strongman’s progress. Examples include the Federal Communications Commission’s jurisdiction over free speech in broadcasting, the sprawling federal antidiscrimination regulations, the flexibility of antitrust legislation, the sedimentation of government surveillance, and so forth. Given the sheer number of laws and regulations, a strongman can always find one to help him.

The Federal Reserve System is among the few bureaucracies that have openly and successfully resisted Trump and provides a counterexample. The Fed is an economically detrimental institution (Lemieux 2018), but it is not as bad as would be a system where politicians and a fortiori a strongman were freely toying with the money supply, inflation, and all economic research within the government. This supplies grist for Cowen’s mill.

What political scientist Hans Morgenthau called the “dual state,” a forerunner of the deep state, raises another danger of a powerful administrative state and especially powerful central agencies (Goldsmith 2018). The label described the national security state that developed in the United States after the creation of the FBI, CIA, and NSA. Through its capacity to collect and leak information on individuals, the dual state was able to control many politicians. It was partly brought to heel by the 1976 Church Committee Report, but the Trump administration is resurrecting it.

Charybdis and Scylla

Is mankind condemned to a choice between the Charybdis of the administrative state and the Scylla of the strongman’s state? Are human societies condemned to be either devoured by the latter or drowned by the former? Is the only alternative between, on the one hand, modern administrative tyranny and, on the other, arbitrary despotism—the latter having been the millennial condition of mankind?

Not necessarily, because a third alternative exists: classical liberalism (or libertarianism), a product of the 18th century Enlightenment and the source of the 19th century Industrial Revolution. The problems of the administrative state and the strongman’s state would be avoided if the state could be constrained through institutional means such as private property, freedom of contract (the “private fortresses” mentioned by Anthony de Jasay), and the rule of law.

It is true, of course, that a bloated government bureaucracy would also slow down a classical liberal reform. But this danger is less problematic than the takeover of the state by an illiberal strongman. If we adopt a constitutional political economy viewpoint, constitutional change or merely a major policy change under a constitution takes time. The reason is that a practical indication of the required unanimity on underlying rules comes from a broad consensus across political institutions. In the United States, these institutions include the two houses of Congress, the bureaucracy they created, the executive, and the courts (plus, if a constitutional amendment is needed, at least three-fourths of the state governments). We can arguably reach a similar conclusion if we adopt a Hayekian approach.

If this consent requirement looks like an evil, we should reflect on Montesquieu’s aphorism: “Since a despotic government is productive of the most dreadful calamities to human nature, the very evil that restrains it is a good.”

None of this is an argument for the administrative state. It is an argument for clear eyes about what can replace it. The classical liberal goal remains equal individual liberty and, thus, a state constrained by law, not a state constrained by its own bureaucratic inertia. But in the absence of that better world, a rule-bound bureaucracy at least operates within a framework of impersonal rules rather than personal commands. The strongman offers the opposite: liberation from administrative red tape paired with subjection to his will. For those who understand that the rule of law matters more than the identity of the ruler, that is not a deal worth taking.

Readings

• Adler, Jonathan H., 2019, “Redefining ‘Waters of the United States,’ ” Regulation 42(2): 16–23.
• Brennan, Geoffrey, & James Buchanan, 1985, The Reason of Rules: Constitutional Political Economy, Cambridge University Press (Liberty Fund, 2000).
• Cowen, Tyler, 2018, “Could Fascism Come to America?” in Cass R. Sunstein (ed.), Can It Happen Here: Authoritarianism in America, HarperCollins, pp. 37–55.
• Crews, Clyde Wayne Jr., 2025a, “Beyond Rules: The Dreck Equation for Quantifying Unseen Federal Regulatory Burdens,” SSRN no. 5576111.
• Crews, Clyde Wayne Jr., 2025b, “Deregulation’s Year-End Illusion,” Competitive Enterprise Institute, December 22.
• Crews, Clyde Wayne Jr., 2026, Ten Thousand Commandments, 2026: An Annual Snapshot of the Federal Regulatory State, Competitive Enterprise Institute.
• De Jasay, Anthony, 1997, Against Politics: On Government, Anarchy, and Order, Routledge.
• Dunai, Marton, 2026, “Hungary’s Business Elite Pivots Away from Viktor Orbán,” Financial Times, April 28, 2026.
• Firey, Thomas A., 2024, “Schedule F: The Phantom Menace,” Regulation 47(1): 40–45.
• Goldsmith, Jack, 2018, “Paradoxes of the Deep State,” in Cass R. Sunstein (ed.), Can It Happen Here: Authoritarianism in America, HarperCollins, pp. 105–133.
• Hayek, Friedrich A., 1960, The Constitution of Liberty, University of Chicago Press.
• Howell, William G., & Terry M. Moe, 2025, The Trajectory of Power: The Rise of the Strongman Presidency, Princeton University Press.
• Lemieux, Pierre, 2018, “A Bad Solution to Very Real Problems,” EconLog (blog), January 30.
• McLaughlin, Patrick, et al., 2023, “RegData US 5.0 User Guide.”
• Moe, Terry M., 1997, “The Positive Theory of Public Bureaucracy,” in Dennis C. Mueller (ed.), Perspectives on Public Choice: A Handbook, Cambridge University Press, pp. 455–480.
• Nathan, Richard P., 1976, “The ‘Administrative Presidency,’ ” Public Interest 44 (Summer): 40–54.
• Wallison, Peter J., 2026, “A Threat to the Constitutional Order,” Law & Liberty, April 16.