Ask someone what a classical liberal believes, and you’ll get a familiar list. Classical liberals stand for private property. They stand for free markets and limited government. They believe in the rule of law, free trade, free speech, and religious toleration. Give or take an item or two, you’ll find something like this list in any textbook summary, or on the “about” page of most classical liberal think tanks. And it’s a good list, as far as it goes. Each of those commitments has a long history in the tradition, and each has been defended by some of its best minds.
The problem with the list is what’s not on it: democracy. It’s not that classical liberals oppose democracy, exactly. Most of us don’t. And almost none of us would trade it for any of the alternatives actually on offer. But democracy isn’t what classical liberals are known for. It’s not part of the brand. Indeed, if you spend much time in classical liberal circles, you’re much more likely to hear about democracy’s defects than about its accomplishments. We might tolerate democracy, but we rarely sing its praises.
My goal in this essay is to argue that democracy belongs on the list, and that it belongs there for the same reason the other items do: because it is a way of keeping power in check. Democracy’s distinctive function is to keep power contestable, so that the people who hold it can be removed without violence. Recognizing this function is especially important today given the specific ways in which illiberal, anti-democratic regimes have operated. For the most part, those regimes aren’t abolishing elections. The elections still run, but they are tilted so heavily against challengers that removing the people in charge becomes very difficult. Political scientists have names for this. Nancy Bermeo calls the process “executive aggrandizement.” Steven Levitsky and Lucan Way call the resulting type of regime, in which elections are real but unfair, “competitive authoritarianism.” The problem they identify is real and growing. By the latest count from the International Institute for Democracy and Electoral Assistance, thirty-two countries, nearly a fifth of those it tracks, saw the credibility of their elections decline between 2020 and 2025.
In regimes of this sort, the democratic form survives even while its function is quietly hollowed out. A tradition that keeps democracy off its list is poorly placed to notice. Unless we understand that function, and see how it goes hand in hand with the things that are on the list, we may not recognize the loss while it’s happening.
***
John Stuart Mill once wrote an entire book arguing that representative democracy was the best form of government, and he was hardly alone in his time. How, for classical liberals, did democracy fall off the list of core values? Part of the answer, I think, is a conflation. Classical liberals have a well-developed critique of what we might call the over-politicization of society. Simply put, the state has taken on too many tasks that markets and civil society ought to perform instead. That critique points to a claim about the proper scope of government. Somewhere along the way, though, that claim about the proper scope of government turned into a claim about democratic control itself. Skepticism about how much the state should do came to be felt as skepticism about letting the public decide what it does.
Jacob Levy has diagnosed the confusion clearly. The distrust of democracy among those sympathetic to markets, he observes, “typically juxtaposes democratic decision-making to market decision-making, as if ‘democracy’ were just a synonym for ‘politics’ or ‘the state,’ as if the contrast to ‘democratic’ were ‘economic’ rather than ‘autocratic.’”
But that’s a hasty move. One can be a radical about reducing the scope of the state and still insist that its agents must be answerable to the people they govern. Or, as Levy puts it, “We should sharply distinguish the question of how much the state ought to do from the question of how much democratic control of the state there ought to be, and thereby stop allowing market-liberal critiques of too much politics to be used as any kind of intellectual cover for the erosion of democratic government.” Once these two questions come apart, the position that most classical liberals already hold becomes both understandable and coherent.
***
There’s a deeper point here too. Classical liberals tend to treat depoliticized markets as a kind of natural starting point or baseline that bad policy disturbs. But this gets the sequencing backwards. For most of history, doing business on any significant scale meant having friends in power. To get things done, you needed a charter from the crown, a guild’s license, or the protection of the right patron. The ability of businesses in a free society to survive without political favors is the product of a particular kind of political order. To see what kind, and where democracy fits into it, it helps to start with what democracy is for.
The classical liberal tradition itself offers an answer. In 1927, Ludwig von Mises defined democracy as “that form of political constitution which makes possible the adaptation of the government to the wishes of the governed without violent struggles.” Democracy, on this view, isn’t about political equality or the general will. It’s about violence and about how power changes hands. Or, as Jason Kuznicki puts the underlying question, if one ruler must sooner or later give way to another, how should it be done? “By a bloody and opportunistic palace coup? By the rising of an angry mob? Or by counting a stack of papers on a duly appointed day?”
The conception of democracy at work here is a minimalist one, according to which democracy is understood as a system in which people gain and lose power through competitive elections. Some democratic theorists argue that democracy involves much more than this. Others stop right here. Adam Przeworski’s influential definition holds that democracy is “a system in which parties lose elections.” What else it requires depends on which theorist you ask, but typical candidates include good-faith deliberation, political equality, and even limits on economic inequality. My point isn’t that these claims about democracy are mistaken. My point is that even if we set them aside, democracy still performs a function that classical liberals have reason to care about. It makes rulers answerable to a public that can remove them without violence.
Understood in this way, democracy is a vital complement to the other items on the classical liberal list. Courts, federalism, a free press, and the freedom to leave all constrain the people who hold power. But elections are what give ordinary citizens a regular, routine way of removing them.
Still, they can’t do this by themselves. Voting can remove a ruler only if voters are free to organize, able to find out what the government has been doing, and able to have their votes counted honestly. Even Joseph Schumpeter, the founder of the minimalist view of democracy, recognized that free competition for votes would normally require “a considerable amount of freedom of discussion for all.” This is why democratic backsliding rarely targets elections in isolation. It goes after the press, the opposition, the courts, and the whole machinery of electoral administration. Would-be tyrants fear those institutions because they make the threat of removal real. And, as we’ll see, democracy helps keep them secure in turn.
***
Why should it matter so much that power can be lost? One answer can be found in the work of the economic historians Douglass North, John Joseph Wallis, and Barry Weingast. In their sweeping and ambitious book, Violence and Social Orders, they argue that for most of human history, communities have solved the problem of social violence by means of what they call a “limited access order.” A coalition of elites agrees not to fight, and that agreement is held together by privilege. Only members of the coalition may hold certain kinds of property, enter certain trades, or form certain kinds of organizations. “The creation of rents through limiting access,” they write, “provides the glue that holds the coalition together.”
This is precisely the world of chartered monopolies and guild privileges that Adam Smith attacked in The Wealth of Nations. But it wasn’t just the economic logic of Smith’s critique that brought that world to an end. In the countries that made the transition to “open access,” economic and political openness arrived together, “within relatively narrow windows of historical time.” General incorporation laws replaced corporate charters that had required “an explicit grant by the legislature [or] the crown.” The reform occurred during the same era that competitive political parties formed and the franchise began to widen. This was no accident. On North, Wallis, and Weingast’s account, the two kinds of openness depend on each other. Closed economies and closed political systems propped each other up, and economic openness could be sustained only alongside political openness.
There’s more to political openness than the right to vote. Openness requires the freedom to form parties, to organize opposition, and to compete for power on impersonal terms. But all of this overlaps heavily with the contestability that democratic elections are meant to secure. A political order in which power can be challenged and lost is one in which privilege is much harder to lock in, in politics and in markets alike.
* * *
Political scientists have identified a mechanism that helps explain why. In The Logic of Political Survival, Bruce Bueno de Mesquita and his coauthors argue that rulers who depend on a large coalition have to buy its support with goods that benefit many people. Rulers who depend on a small coalition, in contrast, can get by on private favors.
Coalition size isn’t quite the same thing as electorate size. But they are related, and studies of what happens when the franchise widens demonstrate the same logic at work. After American states gave women the vote, Grant Miller finds, legislators quickly changed their behavior. Local public health spending rose sharply, and child mortality fell by eight to fifteen percent. Similarly, when Brazil’s switch to electronic voting effectively enfranchised large numbers of less-educated citizens, Thomas Fujiwara finds that public spending shifted toward health care in a way that produced real health benefits for the newborns of less-educated mothers.
This doesn’t mean that a wider franchise always produces better policy, or worse. The historical record is more complicated than that. The point is simply that who a government has to answer to shapes what it does, and a government that must answer to a broad electorate has less room to govern for the benefit of a favored few. Electoral breadth is thus a constraint on precisely the kind of rent-seeking and predation that classical liberals abhor. Women’s suffrage widens the coalition needed for victory and makes a closed economic or political order more difficult to achieve.
All of this is rather different from the arguments for democracy typically made by modern liberals. The most familiar of those arguments are based on ideals of political equality or collective self-government. The classical liberal case, in contrast, focuses on accountability and the containment of power. Yet there is nothing mutually exclusive in these arguments. Both arrive at the same institution, albeit for different reasons. This redundancy is a strength in the case for democracy. Even if you doubt that your vote does much for you, and even if the act of voting doesn’t leave you feeling all that dignified, the accountability argument still goes through.
* * *
None of this requires pretending that democracy’s critics have been wrong about everything. Voter ignorance is real and well documented. Rent-seeking is real. And, of course, there are some questions that shouldn’t be put to a vote at all, a fact recognized by anybody who supports the idea of constitutionally protected rights. Proposals to improve on existing democratic institutions, including citizens’ assemblies chosen by lot and expanded opportunities for what Ilya Somin calls “foot voting,” deserve a serious hearing.
The problem isn’t the criticisms themselves. It’s that, over time, opposition to democracy has become for many in our tradition less a set of arguments than a kind of identity. You can see it in the titles of what libertarians and classical liberals write about democracy. Somin’s thoughtful essay on this site is called “Liberalism’s Uneasy Relationship with Democracy.” Jason Brennan, apparently less conflicted, called his book simply Against Democracy, even though it concedes that “existing democratic countries tend to do a better job protecting citizens’ civil and economic liberty than nondemocratic ones,” and its practical conclusion is a call to experiment. Farther to the fringes lies Hans-Hermann Hoppe’s Democracy: The God That Failed, which sets out the advantages of monarchy, and which Curtis Yarvin credits with opening “a kind of door to the pre-revolutionary world” for him.
These are very different works, and it would be unfair to lump them together. Somin’s essay is a careful account of democracy’s genuine problems. Hoppe’s book argues that monarchy was better than democracy, even though both are worse than having no state at all. The book has become a founding text for the neoreactionary movement. But that’s exactly the trouble with a brand. What people latch onto is the label, “anti-democracy,” not the reasoning or the qualifications behind it.
The label also obscures Levy’s distinction from earlier in this essay, between who wields political power and how far that power extends. That distinction is one that Hayek often insisted upon, writing for instance that “it is not democracy but unlimited democracy which I regard as no better than any other unlimited government.” Hayek’s objection was to treating a majority’s will as the last word on anything it cared to decide, not to the practice of making rulers answerable to the people they govern. Indeed, on that second point Hayek was willing to go further than most of his readers remember: “the consistent application of liberal principles,” he wrote, “leads to democracy.”
But Hayek’s own example is also a warning about what happens when misgivings about democratic choice harden into a posture. In a 1981 interview with Chile’s El Mercurio, Hayek allowed that “a dictatorship may be a necessary system during a transitional period,” and added: “I personally prefer a liberal dictator to a democratic government lacking liberalism.”
In a sense, this is just the same contrast between limited and unlimited democracy that Hayek had been drawing for decades. The obvious problem is that we cannot hope for a benevolent dictator; nothing about a dictatorship ensures that absolute power will be used, as Hayek put it, to “avoid and limit any absolute power in the future.” The expectation that a dictator would limit his own power rests on little more than a hope about the character of the man in charge, precisely the sort of hope Hayek spent his career warning against.
The fact that a thinker as characteristically smart and careful as Hayek could make such an obvious and momentous mistake is deeply troubling. Hayek’s writings over the course of his career demonstrate that he understood the dynamics of political power perfectly well. The lesson is less about Hayek than about the posture. Once suspicion of democracy hardens, it can carry even the best liberal minds somewhere their own arguments would never have taken them.
Michael Munger wrote here recently that liberalism “is sustainable only so long as it is broadly supported by the citizens it protects.” I’d add that it is sustainable only so long as its own advocates are willing to defend it under its own name. If democracy is a liberal achievement, the people who built it ought to be willing to say so out loud, above all while it’s under attack.
***
If we grant that the criticisms of democracy are real, and yet still insist that democracy belongs on the list of core classical liberal commitments, the question becomes one of how we might reform democracy to make it better. The argument I have presented so far suggests a test: Does a proposed reform preserve broad contestation, and with it our ability to discover and correct political mistakes?
The most serious challenge comes from epistocrats like Brennan, who argue that citizens have a right to a competent electorate, and that political power should therefore be tilted toward those who can demonstrate political knowledge. Brennan considers a range of designs, from restricting the franchise to giving extra votes to the well informed. And he is admirably candid about how speculative the case is. Borrowing a thought experiment from Munger, he notes that democracy is an ugly pig. But it’s hard to know if his own epistocracy is any less ugly since, as he concedes, “I don’t have a second pig to look at.”
Pointing out that contested elections are important to prevent the concentration of power is not enough to defeat Brennan’s case. An epistocracy can keep contested elections while still restricting the political power of ordinary voters. The real disagreement is about how widely the power to hold rulers accountable should be shared.
Any attempt to implement a right to competent government through voter qualifications requires someone to decide what counts as competence and how it should be measured. The problem is that this creates a significant new prize worth capturing. If political power is, in Levy’s words, “the key instrumental good in the pursuit of rents,” then control over who gets to exercise it is more valuable still. The history of literacy tests in the American South illustrates the danger.
On their face, these were neutral tests of reading and constitutional understanding. In practice, of course, local registrars used their discretion to keep Black citizens from voting. Elizabeth Cascio and Ebonya Washington found that when the Voting Rights Act suspended literacy tests, increases in both turnout and state transfers were greater in counties with larger Black population shares. The history shows how a formally neutral qualification can become an instrument of exclusion in the hands of people with something to protect.
Brennan has responded to a version of this worry raised by David Estlund, arguing that literacy tests “were administered in bad faith.” And of course he’s right to note the difference in motivation. Showing that literacy tests were simply racism in epistocratic disguise doesn’t discredit epistocracy itself. Brennan also concedes that a genuine competence test could become an object of partisan manipulation. His answer is comparative: democratic institutions get abused too, so the question is which system performs better once the abuses of each are counted.
But as Samuel Bagg has argued, the real problem with epistocratic tests is structural, not motivational, and the structure is unusually dangerous. A system that allows incumbents to set epistemic qualifications would let them “entrench their rule by choosing their own voters.” Ordinary gerrymandering is limited by geography and a fixed electorate. This new form “could continuously compound incumbent advantages through ever-more-targeted qualifications,” altering the electorate itself as it goes. Brennan’s own analogy is telling here. He tries to show that epistocratic proposals are no more objectionable than medical licensing. But the most compelling objection against that sort of licensing, one embraced by classical liberals and progressives alike, is that it often functions mainly to protect incumbents from competition.
Other reforms do better by the test of contestability. A citizens’ assembly chosen by lot gives a randomly selected group the time to study a question closely, which the rest of us will never do, and then the assembly reports back. That maintains broad accountability while addressing voter ignorance in a meaningful way. Decentralization and easier exit work differently, by lowering the stakes of political decision making in general, and by lowering the cost of losing in particular. Neither of these reforms takes the vote away from anybody. And given how little any of us knows about how these arrangements would perform, there is a lot to be said for the ones that let us notice our mistakes and undo them.
* * *
Even the most doctrinaire classical liberals already know why private property, free speech, the rule of law, and limited government belong on the list of their core political commitments. Each of these institutions helps to keep political power from becoming too dangerous. My argument is that democracy belongs there for the same reason, as a positive good, and not as an exception to the rule. It gives ordinary citizens a regular way to remove rulers, and it spreads the power of accountability broadly enough that government cannot answer only to a favored few.
Democracy will not guarantee wise decisions any more than competitive markets guarantee wise investments. As Janet Bufton has argued, criticizing democracy for what it produces is much like criticizing markets for what they produce. What both do, even in failure, is to ensure that the people who get it wrong can be replaced. A liberal tradition concerned with concentrated power has every reason to claim as its own the institution that makes political power contestable. Democracy belongs on the list.
