Fukuyama's Defense of the Deep State
Why bureaucratic autonomy matters to administrative capacity
Francis Fukuyama of Stanford University has been one of the most perceptive scholars of how state power is constructed, used and abused, most notably in his books, The Origins of Political Order, and Political Order and Political Decay. Earlier this year, he gave the Donald Stone lecture to the American Society for Public Administration, which was reprinted in the Asia Pacific Journal of Public Administration. With permission from the editor of the journal, Professor James Perry, and Fukuyama, I am including excerpts from that lecture, “In Defense of the Deep State” below.
Here are some key takeaways for me:
the idea of a “deep state” does not reflect how the American administrative state operates
delegation to administrative actors is necessary for a complex, modern state to function
state actors have many tools to manage the bureaucracy to keep it accountable
sometimes under-delegation or “vetocracy” is a barrier to getting things done
the goal should be to find a balance between delegating too much power to state actors (as is the case in China) versus destroying state capacity (as Trump era-populists seem poised to do)
The phrase “deep state” originated in countries like Turkey and Egypt, where a complex of military and security agencies manipulated the political system and operated in a completely non-transparent way to affect politics. This phrase was then appropriated by American conservatives like Steve Bannon and used to characterize the American bureaucracy; total destruction of the “deep state” has become a central agenda item for many populists on the right.
The United States does not have a “deep state” in the Middle Eastern sense of the term. It has a large and complex civil service at federal, state, and local levels that is responsible for providing the bulk of the services that citizens expect from their government, what is known as the “administrative state”. This administrative state is in fact highly transparent when compared to those of other liberal democracies, not to speak of Middle Eastern dictatorships, and it can be readily controlled by elected political leaders if the latter choose to exercise their powers. That bureaucracy necessarily operates independently of its political masters, however, since elected leaders cannot possibly specify the thousands of daily decisions that are needed to keep the government running. They delegate authority to bureaucrats, which is what we call bureaucratic autonomy.
Protecting a sphere of bureaucratic autonomy is extremely important if the government is to function properly. I want to defend the continuing need for an administrative state that is allowed to exercise a wide degree of judgement in its implementation of the people’s will. We do not want elected politicians to make decisions, for example, on setting interest rates or deciding which banks to bail out, to determine schedules for Air Force aircraft maintenance, or to certify particular drugs as safe and effective. When politicians start to make these kinds of interventions, the results are almost always harmful because they do not have the expertise or knowledge to do so effectively, and will be tempted to use these powers for their own narrow advantage. Nor do we want those politicians to constrain bureaucratic decision-making excessively by issuing thousands of ex ante rules; bureaucratic red tape is one of the great scourges of government.
On the other hand, we do not want bureaucrats to make big decisions regarding policy. In a liberal democracy, citizens elect leaders who are responsible for setting the overall direction of policy: how much money to allocated to guns versus butter; what kinds of social services the state should offer; or when and where to use military force. This issue is a sore point for many conservatives, since bureaucrats as a group tend to lean to the left in their political preferences. Finding the right degree of bureaucratic autonomy is one of the biggest challenges in any modern liberal democracy.
In democratic theory, the people are sovereign and have the authority to make decisions regarding their own self-government. They are the principals in a principal-agent hierarchy in which they choose representatives via elections, who then issue mandates to the agencies and bureaucrats responsible for carrying out the people’s wishes. Democratically chosen political principals should control bureaucratic agents.
The problem with this theory is that this simple normative hierarchy has never been possible to implement in practice. As Herbert Simon and his co-authors pointed out many decades ago, authority in bureaucracies often flows in the opposite direction. Bureaucratic agents often have the detailed knowledge and expertise that political principals lack, and therefore end up instructing the latter on necessary policies and their requirements for implementation.
There are other important reasons why substantial delegations of authority are necessary for the functioning of any bureaucracy. As Friedrich Hayek pointed out many years ago, the vast majority of information in any modern society is local in nature. Hayek in this article was concerned with price-setting in a market economy, but his observation applies to bureaucratic decision-making as well. It is the low-level agents and not their higher level bosses who have best access to local knowledge, and are therefore best able to understand and interact with their environment. They are able to react more quickly to changes in that environment, and, when the proper mechanisms exist, can be held more directly accountable for the decisions they make.
These considerations are apparent in military organizations. One of the reasons the US military has become one of the world’s premiere fighting forces has to do with its adoption of the doctrine of “mission orders” or “commander’s intent” in the course of its post-Vietnam reforms. This doctrine, based on the German military’s Auftragstaktik, maintains that military commanders should delegate authority to the lowest possible command level consistent with the overall mission. Commanders set broad operational objectives, but it is lower-ranking officers and NCOs in direct contact with the enemy who are responsible for figuring out how to implement those instructions.
All political systems therefore need to delegate authority, and in many cases, the more delegation, the better the bureaucracy will run. Delegation, however, comes with risks that the agent will make mistakes, take needless risks, be corrupt, or use that authority for purposes other than those intended by the political principal. One of the central issues in administrative law is which authorities to delegate, how much to delegate, and what mechanisms to use to control agencies. In general, the better trained and more professional the agent, the more authority can safely be delegated, though there is not necessarily optimal degree of delegation that works in all conditions. Furthermore, all bureaucracies necessarily run on a degree of trust.
The degree to which different societies are willing to trust bureaucrats varies widely. It is safe to say that those in East Asia exhibit the highest degree of trust. One of the deepest Chinese cultural traditions has to do with the training and respect given to bureaucrats; meritocratic recruitment into the bureaucracy in China goes back a couple of millennia, and the practice has spread to other societies in that country’s cultural sphere of influence like Japan and South Korea.
At the other end of the scale are numerous low-income countries with corrupt or low-capacity governments, which are often seen as predatory or serving the interests of hidden elites. Among developed democracies, most European nations have relatively high degrees of trust in government, though levels are higher in the north and western parts of the region. The United States is something of an outlier among developed democracies, given its deep tradition of anti-statism. Many Americans across the ideological spectrum are suspicious of government despite their dependence on the state for many essential services. For this reason, the American state developed later than its European counterparts, has been less extensive in scope, and has had to survive numerous efforts to roll back its authority.
The concept of a modern state was articulated most notably by Max Weber at the turn of the 20th century. In contrast to patrimonial states that were the outgrowth of the ruler’s household and were staffed by the latter’s friends and family, a modern state is impersonal, treating subjects as equal citizens and serving public interest rather than the private interest of the ruler. Weber understood the ideal type of bureaucrat as a non-partisan, expert administrator, a public servant who was given autonomy and oriented towards public interest.
The early American state was not modern in Weber’s sense, particularly after the 1828 election that brought Andrew Jackson to power. Jackson is often described as the father of American populism, having benefited from the expansion of the franchise to all white males that occurred in most US states during the 1820s. It turned out that individualized benefits like a bottle of bourbon or a job in the post office were the easiest ways to mobilize masses of new voters; Jackson argued that ordinary Americans were qualified to run the government, and that he should get to appoint them since he won the election. Thus began the period of American history known as the “patronage” or “spoils” system, under which virtually every federal employee owed his job to a politician. This system was modernized only in the 1880s with the passage of the Pendleton Act, which established a US Civil Service Commission that sought to introduce merit-based hiring and promotion. While this effort to professionalize the US bureaucracy was largely consolidated by the time of the First World War and expanded with the New Deal’s proliferation of federal agencies, political appointees continue to be much more common in the US than in the democracies of Europe or Asia.
Indeed, the administrative state has been the target of conservatives for decades now, who allege that the country is being ruled by secretive bureaucracy that has escaped the control of democratically elected leaders. Many present-day American populists have defined their political objective as the total destruction of the “deep state”. This agenda surfaced briefly at the end of the Trump administration in 2020, when the White House issued Executive Order 13,957 creating a “Schedule F” category of federal employees who could be fired at will, and ordered administrative agencies to move their employees into this designation. While this order was quickly rescinded by the incoming Biden administration, it remains on the agenda of many Republican lawmakers. Some conservative intellectuals maintain that the Pendleton Act was unconstitutional and seek to roll back hiring and promotion practices to the kinds of clientelistic practices of the 19th century.
In light of these attacks, it is critical to defend the principle of bureaucratic autonomy, and to define its proper bounds. There is in fact some justice to conservative complaints about an “out-of-control” federal bureaucracy. But in other cases that same bureaucracy is overly constrained, particularly by the masses of ex ante rules and procedures that it has to follow. Those procedures were put in place out of distrust of bureaucratic authority, but they have contributed to the de-legitimising of that same authority because they slow down or block entirely the state’s ability to achieve desirable outcomes.
Delegation in US administrative law
Conservative legal scholars sometimes cite a supposed “non-delegation” doctrine that limits the ability of Congress to delegate authority to administrative agencies. As a constitutional matter, this is highly questionable; no such clause exists in the Constitution. Delegation has been practiced since the first days of the Republic; the first US Congress delegated to Alexander Hamilton’s Treasury Department the job of cleaning up the nation’s Revolutionary War debt. It is a fantasy to think that the American people through their elected representatives can manage a modern government without substantial delegation. They perhaps have in mind the image of a New England town meeting from a couple of centuries ago, where citizens could debate public affairs and play direct roles in local government. This is the way that local government should continue to work today, but it doesn’t work for a federal government that needs to provide services to a country that spans a continent and is home to more than 330 million people.
In US administrative law, there are five basic mechanisms by which political principals can control bureaucratic agents. These include:
Ex ante procedures. The most important of these is the 1946 Administrative Procedure Act (APA), which requires notice-and-comment for any new rule promulgated by an agency. In addition, there are organic laws defining the procedures that agencies much follow in carrying out Congress’ wishes.
Ex post review. There are numerous mechanisms to enforce bureaucratic accountability, the most important framework for which was established in the 1983 Chevron Deference decision by the Supreme Court. The latter imposes a two-stage rule for when the courts can intervene in reviewing bureaucratic decision-making, and limits that intervention on the grounds that the courts in general do not have the expertise to over-rule agency judgements. Chevron Deference has itself become a target of many conservatives, who have chipped away at the ruling through avenues like the “major questions” doctrine. There are numerous other ex post methods for reviewing bureaucratic decisions, such as the APA’s Administrative Judges who can review agency decisions; in addition, Congress frequently holds hearings in which bureaucrats must account for their actions.
Appointment power. In any government, the ability to control personnel is one of the most important means of controlling the bureaucracy. No set of ex ante rules can foresee all possible future states of the world, so human judgement is required to carry out the people’s will in light of changed conditions. The US Constitution specifies that the President may appoint cabinet officers with the “advice and consent” of the Senate, and can set rules for the appointment of “inferior officials”.
Removal power. One might think that removal power is comparable to appointment power, but it has never been the case in US law. The Founding Fathers held a prolonged debate on removals, leading to the “Decision of 1789” wherein the President was given wide latitude to dismiss senior federal officials. James Madison himself argued that, in contrast to appointment power, the ability to remove recalcitrant officials was fundamental to executive authority. Removal power was weakened for lower level officials, however, in the post-World War II period by the imposition of procedural rules regarding the dismissal of federal workers. These procedures have led to complaints by government managers that it is extremely difficult to hold poorly performing workers accountable.
Ad hoc interventions. There are other legal ways in which political principals can intervene to limit bureaucratic autonomy. The US federal Constitution does not provide for emergency powers, but some state constitutions do, and Congress has enacted a federal emergency powers act. The president can use his pardon power to override executive branch decisions of various sorts, as when former President Trump pardoned Navy SEAL Eddie Gallagher’s conviction by a Navy court.
In recent decades, there have been clear cases of bureaucratic overreach, where agencies have arguably gone beyond the statutory authority granted them by Congress…However, it would seem hard to make the case that Americans are living under a tyranny led by out-of-control bureaucrats. America’s Constitutional system of checks-and-balances provides remedies for executive branch overreach, which were ultimately applied by the courts in three of the four cases cited. Progressives may denounce the substantive outcomes of these decisions. The West Virginia v. EPA decision in particular has set a poor precedent for dealing with the urgent problem of carbon emissions. But procedurally, it was clear that the system could check bureaucratic power.
A question of balance
It should be clear that any well-functioning liberal democracy should balance bureaucratic autonomy against the need for democratic accountability; neither extreme of unchecked state authority or thoroughgoing politicisation is appropriate. Finding that balance is, however, can be very challenging.
Take the case of public health during the Covid pandemic. In the early stages of the outbreak, public health authorities, even in red states, were given considerable latitude in setting policies regarding social distancing, shutdowns, masking, and the like. Indeed, the California state constitution granted county health officers final authority in these matters. In the first months of the pandemic, there was a widespread view among liberals that it was illegitimate to even consider the existence of a tradeoff between public health and other social goods like jobs or economic growth. As the pandemic evolved, however, public attitudes began to shift. The costs of strict enforcement of shutdowns became more evident, particularly to parents who were frustrated by state-level decisions to keep public schools closed. Covid itself had evolved into less deadly variants, and large parts of the population had been vaccinated. By the end of 2022, China’s “zero-Covid” policy (an even stricter version of US shutdowns) was widely mocked in the US for failing to take into account the economic and social costs of the policy.
These choices then got caught up in the broader political polarization. There was a backlash against earlier deference to public health authorities and, among some conservatives, conspiracy theories about their underlying intentions. Many red state politicians then moved to the other extreme, passing statutes dictating detailed restrictions on what kind of public health measures could be taken in future outbreaks. These new rules will hobble the ability of public health authorities to flexibly deal with a future crisis.
A proper response to the Covid crisis should have involved the highest political authorities carefully balancing different social goods. Public health authorities should have had significant inputs into this decision, but they should not have been the final arbiters of the tradeoff, since they will typically prioritize maximization of health over other interests. Unfortunately, under the Trump presidency we did not have a political leader at the federal level who was able or willing to make this decision based on the best available evidence; rather we had a politician who looked primarily to his own re-election interests above all else.
The West Virginia v. EPA decision reflected a similar political failure. The case in my view was procedurally correct but led to a bad outcome, i.e., the failure to regulate, at least in principle, carbon emissions. The fundamental problem lay not in the Supreme Court, but rather in Congress’ failure to legislate on this issue and provide the EPA with proper statutory authority in this domain. In light of this failure, it was understandable that the executive branch tried to extend its authority unilaterally, but also proper for SCOTUS to cut back that authority. Justice Kagan may be correct that the current conservative court was substituting its judgement in place of the expertise of the EPA and Congress, but the courts have been used extensively in recent decades to achieve outcomes that could not be achieved legislatively. Since Brown v. Board of Education, this has largely been the work of progressives, but conservatives have now shown that they can make use of this judicial power as well. In the American system, it is much better to have the elected representatives in Congress make fundamental decisions about tradeoffs between social goods, and then hand off implementation to expert agencies. But the legislative branch has been failing to exercise its proper powers, with bureaucracies and courts then seeking to fill the vacuum.
In addition to the problem of over-delegation, the United States also suffers from a problem of under-delegation. As noted above, Americans typically don’t trust their bureaucrats, and have consequently wrapped the bureaucracy in layers of procedural controls to hold them accountable. This began with the APA but over the years has extended to virtually all public decisions. Major infrastructure projects, for example, are subject to extensive environmental reviews and public hearings under the National Environmental Policy Act (NEPA), many of which are then duplicated and extended at a state level. Alone among advanced democracies, the United States frequently employs a “right of private action” to enforce its laws, in which individual citizens can act as “private attorneys general” and initiate litigation against parties accused of violating a law. This has led to a common-law process in which litigation through (mostly) state courts has steadily expanded the reach and scope of government rules independently of any legislative action.
This steady expansion of procedural checks has led to an outcome I labelled “vetocracy”, in which veto power is so widely distributed among stakeholders that it becomes extremely difficult to make and implement decisions that serve collective interests. As Nicolas Bagley has pointed out, excessive proceduralism has hobbled government and prevented progressives from using government effectively to achieve their aims. On the right, Philip Howard has inveighed against excessive bureaucratic red tape that defies “common sense” in government decision-making. This logically implies that bureaucrats ought to be given more discretion in implementing legislative mandates, where they can use “common sense” to get to desirable outcomes.
If we are to permit greater bureaucratic autonomy, we still need to put in place strong accountability measures to make sure that autonomy is not abused. One method might be to replace ex ante procedures and the mountains of red tape they impose on bureaucrats, with stronger ex post review as a method of holding agencies accountable. Ex post review can be just as obstructive as ex ante procedures if they are allowed to proliferate endlessly, so we need to find ways of streamlining them. For any bureaucratic system to work, the society would have to make adequate investments in the training and professionalization of public servants.
The US bureaucracy suffers from any number of dysfunctions, problems that have been analyzed at length in a number of detailed studies. For example, removal power has indeed been hobbled by layers of procedural protections for public servants that have accumulated over the years, making it difficult to discipline poorly-performing workers. Whatever the proper solution, it is not the wholesale destruction of the professional civil service in the manner envisioned by many conservatives today. Schedule F will not solve the problem of political bias in the bureaucracy but will exacerbate it, and will come at the expense of government competence and effectiveness.
Currently, an election and change in administrations leads to the replacement of between 3000–4000 senior officials throughout the Federal government. This is a far higher number than in any other developed democracy, where political turnover leads to a few dozen replacements at most. It is nonetheless normatively in line with democratic theory. The backers of Schedule F envision increasing the number of political appointees to tens or perhaps hundreds of thousands. In their view, the will of elected political principals is being stymied by the left-leaning bias of ordinary civil servants, and conservatives are compiling lists of politically sympathetic candidates to replace them.
This plan will run into several insuperable problems. Ordinary civil servants are today not overtly selected on the basis of their political leanings but on supposedly objective merit criteria. Any political bias exhibited by those selected reflects biases in the underlying populations: people wanting to be public school teachers or are qualified to be university professors tend to lean left; those wanting to be policemen, firemen, or prison guards tend to lean right. Finding enough people who meet political loyalty tests and have appropriate skills will be hard and will severely compromise the quality of public servants. Moreover, conservatives need to think ahead to what will happen if they lose a subsequent election: their proteges will lose their jobs and be replaced not by neutral non-partisan civil servants, but by liberal appointees meeting explicit liberal political tests. This was the nature of American government under the 19th century patronage system prior to passage of the Pendleton Act.
Distrust of government has existed for as long as governments have existed, and in many cases was fully justified by government behavior. But many contemporary populists have taken this critique to extreme lengths, attacking not just specific abuses of state power, but the very idea of a government built around merit and expertise. They have taken to heart Andrew Jackson’s view that “any ordinary American” could perform the functions of government. This was scarcely true in the 1820s when the federal government did little other than deliver the mail and operate customs houses. But most ordinary Americans today are not capable of managing monetary aggregates, running large-scale predictive weather models, controlling air traffic, teaching at state universities, undertaking employment surveys, or supervising randomized trials for new pharmaceuticals.
Selection of bureaucrats on the basis of merit has been threatened not just from the right, but from the left as well. There are progressive critiques of meritocracy that argue it is a technique used by existing elites to preserve their power and influence; well-educated parents get their children into the right schools and give them a leg up on entering the world of power and influence. Progressives have sought to introduce diversity criteria into federal hiring practices to ensure representation of racial and ethnic minorities, women, and other marginalized populations.
From either perspective, abandoning meritocratic hiring and promotion in the US government will have huge consequences for government effectiveness. A recent cross-national study shows that there is a strong relationship between meritocratic recruitment and promotion and government performance across a wide variety of agencies and countries. The United States could decide to de-prioritize government effectiveness in favor of greater political control, but it is not clear that this is an outcome favored by a majority of citizens who are already unhappy with the way the government works.
If people with the requisite skills to run a modern government have political biases, the solution is to mandate different policies, and correct specific instances of overreach rather than demonizing the idea of expertise in general. Populists today are simply not aware of the mechanisms available to them to control the “deep state”. Their problem lies in the fact that most of these mechanisms run through a Congress that they do not fully control. As a result they will seek to use executive power to undermine the executive branch itself through a mass purge of civil servants deemed not to be sufficiently loyal to their agenda.
The United States is today engaged in a long-term competition with China. The latter constitutes a civilization unto itself that for many centuries has been built around high-quality bureaucracy and respect for expertise. What it lacks are institutions that can check its powerful state, like a rule of law and democratic accountability. The United States sits at the other end of the spectrum, with powerful check-and-balance institutions, an expansive but often ineffective state, and a political culture that is very distrustful of state authority. It is hard to see the United States prevailing in this global competition if it prioritizes the dismantling of its existing bureaucracy and the replacement of experts by political loyalists. What it needs to strive for in the end is balance, a balance between bureaucratic autonomy and political control, and between procedural compliance and effective outcomes.
Can We Still Govern? is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.