How Democracies Die

by Steven Levitsky

The tragic paradox of the electoral route to authoritarianism is that democracy’s assassins use the very institutions of democracy—gradually, subtly, and even legally—to kill

Historically, our system of checks and balances has worked pretty well—but not, or not entirely, because of the constitutional system designed by the founders. Democracies work best—and survive longer—where constitutions are reinforced by unwritten democratic norms. Two basic norms have preserved America’s checks and balances in ways we have come to take for granted: mutual toleration, or the understanding that competing parties accept one another as legitimate rivals, and forbearance, or the idea that politicians should exercise restraint in deploying their institutional prerogatives. These two norms undergirded American democracy for most of the twentieth century. Leaders of the two major parties accepted one another as legitimate and resisted the temptation to use their temporary control of institutions to maximum partisan advantage.

Today, however, the guardrails of American democracy are weakening. The erosion of our democratic norms began in the 1980s and 1990s and accelerated in the 2000s. By the time Barack Obama became president, many Republicans, in particular, questioned the legitimacy of their Democratic rivals and had abandoned forbearance for a strategy of winning by any means necessary. Donald Trump may have accelerated this process, but he didn’t cause it. The challenges facing American democracy run deeper. The weakening of our democratic norms is rooted in extreme partisan polarization—one that extends beyond policy differences into an existential conflict over race and culture. America’s efforts to achieve racial equality as our society grows increasingly diverse have fueled an insidious reaction and intensifying polarization. And if one thing is clear from studying breakdowns throughout history, it’s that extreme polarization can kill democracies.

Building on Linz’s work, we have developed a set of four behavioral warning signs that can help us know an authoritarian when we see one. We should worry when a politician 1) rejects, in words or action, the democratic rules of the game, 2) denies the legitimacy of opponents, 3) tolerates or encourages violence, or 4) indicates a willingness to curtail the civil liberties of opponents, including the media. Table 1 shows how to assess politicians in terms of these four factors.

A politician who meets even one of these criteria is cause for concern. What kinds of candidates tend to test positive on a litmus test for authoritarianism? Very often, populist outsiders do. Populists are antiestablishment politicians—figures who, claiming to represent the voice of “the people,” wage war on what they depict as a corrupt and conspiratorial elite. Populists tend to deny the legitimacy of established parties, attacking them as undemocratic and even unpatriotic. They tell voters that the existing system is not really a democracy but instead has been hijacked, corrupted, or rigged by the elite. And they promise to bury that elite and return power to “the people.”

Nobody likes smoke-filled rooms today—and for good reason. They were not very democratic. Candidates were chosen by a small group of power brokers who were not accountable to the party rank and file, much less to average citizens. And smoke-filled rooms did not always produce good presidents—Harding’s term, after all, was marked by scandal. But backroom candidate selection had a virtue that is often forgotten today: It served a gatekeeping function, keeping demonstrably unfit figures off the ballot and out of office. To be sure, the reason for this was not the high-mindedness of party leaders. Rather, party “bosses,” as their opponents called them, were most interested in picking safe candidates who could win. It was, above all, their risk aversion that led them to avoid extremists.

Our founders were deeply concerned with gatekeeping. In designing the Constitution and electoral system, they grappled with a dilemma that, in many respects, remains with us today. On the one hand, they sought not a monarch but an elected president—one who conformed to their idea of a republican popular government, reflecting the will of the people. On the other, the founders did not fully trust the people’s ability to judge candidates’ fitness for office. Alexander Hamilton worried that a popularly elected presidency could be too easily captured by those who would play on fear and ignorance to win elections and then rule as tyrants. “History will teach us,” Hamilton wrote in the Federalist Papers, that “of those men who have overturned the liberties of republics, the great number have begun their career by paying an obsequious court to the people; commencing demagogues, and ending tyrants.”

The Electoral College, made up of locally prominent men in each state, would thus be responsible for choosing the president. Under this arrangement, Hamilton reasoned, “the office of president will seldom fall to the lot of any man who is not in an eminent degree endowed with the requisite qualifications.” Men with “talents for low intrigue, and the little arts of popularity” would be filtered out. The Electoral College thus became our original gatekeeper.

These dual imperatives—choosing a popular candidate and keeping out demagogues—may, at times, conflict with each other. What if the people choose a demagogue? This is the recurring tension at the heart of the presidential nomination process, from the founders’ era through today. An overreliance on gatekeeping is, in itself, undemocratic—it can create a world of party bosses who ignore the rank and file and fail to represent the people. But an overreliance on the “will of the people” can also be dangerous, for it can lead to the election of a demagogue who threatens democracy itself. There is no escape from this tension. There are always trade-offs.

On the other hand, the convention system was an effective gatekeeper, in that it systematically filtered out dangerous candidates. Party insiders provided what political scientists called “peer review.” Mayors, senators, and congressional representatives knew the candidates personally. They had worked with them, under diverse conditions, over the years and were thus well-positioned to evaluate their character, judgment, and ability to operate under stress. Smoke-filled back rooms therefore served as a screening mechanism, helping to keep out the kind of demagogues and extremists who derailed democracy elsewhere in the world. American party gatekeeping was so effective that outsiders simply couldn’t win. As a result, most didn’t even try.

Some political scientists worried about the new system. Binding primaries were certainly more democratic. But might they be too democratic? By placing presidential nominations in the hands of voters, binding primaries weakened parties’ gatekeeping function, potentially eliminating the peer review process and opening the door to outsiders. Just before the McGovern–Fraser Commission began its work, two prominent political scientists warned that primaries could “lead to the appearance of extremist candidates and demagogues” who, unrestrained by party allegiances, “have little to lose by stirring up mass hatreds or making absurd promises.”

Judges who cannot be bought off may be targeted for impeachment. When Perón assumed the presidency in 1946, four of Argentina’s five-member supreme court were conservative opponents, one of whom had called him a fascist. Concerned about the court’s history of striking down pro-labor legislation, Perón’s allies in congress impeached three of the justices on the grounds of malfeasance (a fourth resigned before he could be impeached). Perón then appointed four loyalists, and the court never opposed him again.

So they changed the rules—and did away with democracy. “Give us a [constitutional] convention, and I will fix it so that…the Negro shall never be heard from,” former Georgia senator Robert Toombs declared as Reconstruction was coming to an end. Between 1885 and 1908, all eleven post-Confederate states reformed their constitutions and electoral laws to disenfranchise African Americans. To comply with the letter of the law as stipulated in the Fifteenth Amendment, no mention of race could be made in efforts to restrict voting rights, so states introduced purportedly “neutral” poll taxes, property requirements, literacy tests, and complex written ballots. “The overarching aim of all of these restrictions,” historian Alex Keyssar observed, “was to keep poor and illiterate blacks…from the polls.” And because African Americans were overwhelmingly Republican, their disenfranchisement could be expected to restore the Democrats’ electoral dominance. The goal, as a state senator from North Carolina put it, was to write a “good square, honest law that will always give a good Democratic majority.”

Both sides in America’s early partisan battles—John Adams’s Federalists and Thomas Jefferson’s Republicans—regarded each other as a threat to the republic. The Federalists saw themselves as the embodiment of the Constitution; in their view, one could not oppose the Federalists without opposing the entire American project. So when Jefferson and Madison organized what would become the Republican Party, the Federalists regarded them as traitors, even suspecting them of harboring loyalties to Revolutionary France—with which the United States was nearly at war. The

Some polarization is healthy—even necessary—for democracy. And indeed, the historical experience of democracies in Western Europe shows us that norms can be sustained even where parties are separated by considerable ideological differences. But when societies grow so deeply divided that parties become wedded to incompatible worldviews, and especially when their members are so socially segregated that they rarely interact, stable partisan rivalries eventually give way to perceptions of mutual threat. As mutual toleration disappears, politicians grow tempted to abandon forbearance and try to win at all costs. This may encourage the rise of antisystem groups that reject democracy’s rules altogether. When that happens, democracy is in trouble.

The American republic was not born with strong democratic norms. In fact, its early years were a textbook case of politics without guardrails. As we have seen, norms of mutual toleration were at best embryonic in the 1780s and 1790s: Far from accepting one another as legitimate rivals, Federalists and Republicans initially suspected each other of treason. This climate of partisan hostility and distrust encouraged what we would today call constitutional hardball. In 1798, the Federalists passed the Sedition Act, which, though purportedly criminalizing

false statements against the government, was so vague that it virtually criminalized criticism of the government. The act was used to target Republican Party newspapers and activists. In

Mutual toleration, in turn, encouraged forbearance. By the late nineteenth century, informal conventions or work-arounds had already begun to permeate all branches of government, enabling our system of checks and balances to function reasonably well. The importance of these norms was not lost on outside observers. In his two-volume masterpiece, The American Commonwealth (1888), British scholar James Bryce wrote that it was not the U.S. Constitution itself that made the American political system work but rather what he called “usages”: our unwritten rules.

The American system of checks and balances, therefore, requires that public officials use their institutional prerogatives judiciously. U.S. presidents, congressional leaders, and Supreme Court justices enjoy a range of powers that, if deployed without restraint, could undermine the system. Consider six of these powers. Three are available to the president: executive orders, the presidential pardon, and court packing. Another three lie with the Congress: the filibuster, the Senate’s power of advice and consent, and impeachment. Whether these prerogatives are formally stipulated in the Constitution or merely permitted under the Constitution, their weaponization could easily result in deadlock, dysfunction, and even democratic breakdown. For most of the twentieth century, however, American politicians used them all with remarkable forbearance.

The immense powers of the executive branch create a temptation for presidents to rule unilaterally—at the margins of Congress and the judiciary. Presidents who find their agenda stalled can circumvent the legislature by issuing executive orders, proclamations, directives, executive agreements, or presidential memoranda, which can assume the weight of law without the endorsement of Congress. The Constitution does not prohibit such action. Likewise, presidents can circumvent the judiciary, either by refusing to abide by court rulings, as Lincoln did when the Supreme Court rejected his suspension of the writ of habeas corpus, or by using the prerogative of the presidential pardon. Alexander Hamilton argued in Federalist 74 that because the power of pardon was so far-reaching, it would “naturally inspire scrupulousness and caution.” But in the hands of a president without scruples or caution, the pardon can be used to thoroughly shield the government from judicial checks. The president can even pardon himself. Such action, though constitutional, would undermine the independence of the judiciary.

Given the vast potential for unilateral action, nearly all of which is either prescribed or permitted by the Constitution, the importance of executive forbearance is hard to overstate. George Washington was an important precedent-setting figure in this regard. Washington knew his presidency would help establish the future scope of executive authority; as he put it, “I walk on untrodden ground. There is scarcely any part of my conduct which may not hereafter be drawn into precedent.” As the occupant of an office many feared would become a new form of monarchy, Washington worked hard to establish norms and practices that would complement—and strengthen—constitutional rules. He energetically defended his designated areas of authority but was careful not to encroach on areas within the domain of Congress. He limited his use of the veto to bills he regarded as constitutionally dubious, issuing only two vetoes in eight years and writing that he “signed many bills with which my Judgement is at variance,” out of “motives of respect to the legislature.” Washington was also reluctant to issue decrees that could be seen as encroaching on congressional jurisdiction. In eight years, he issued only eight executive orders.

President Harry Truman complied with the Supreme Court’s blocking of his 1952 executive order to nationalize the steel industry in the face of a strike that he viewed as a national emergency. Eisenhower enforced the Supreme Court’s Brown v. Board of Education decision despite his own displeasure with it. Even Nixon acceded to congressional demands that he turn over his secret tapes after the Supreme Court ruled in Congress’s favor. So although the office of the American presidency strengthened during the twentieth century, American presidents demonstrated considerable restraint in their exercise of that power. Even in the absence of constitutional barriers, unilateral executive action remained largely a wartime exception, rather than the rule. A similar story can be told about presidential court packing. Court packing may take one of two forms: impeaching unfriendly Supreme Court justices and replacing them with partisan allies, or altering the size of the Court and filling the new seats with loyalists. Both of these maneuvers are, strictly speaking, legal; the Constitution permits impeachment and does not specify the size of the Supreme Court. Presidents may purge and pack the Court without violating the letter of the law. They have not done so, however, for well over a century.

By the 1920s, British journalist H. W. Horwill concluded that there existed an informal norm “strong enough to prohibit the most powerful President and Congress, whatever the provocation, from taking a course which would make the Supreme Court the plaything of party politics.” President Franklin Roosevelt, of course, violated this particular norm with his 1937 court-packing effort. As constitutional scholars Lee Epstein and Jeffrey Segal wrote, Roosevelt’s norm-violating proposal was “extraordinary in its hubris.” Equally extraordinary, however, was the resistance it generated. At the time, Roosevelt was extremely popular—he had just been reelected in a historic landslide, and his Democratic allies enjoyed solid majorities in both houses of Congress. Few American presidents have ever enjoyed such political strength. Yet court packing triggered across-the-board opposition. Media criticism was fierce—the San Francisco Chronicle described the plan as an “open declaration of war on the Supreme Court.”

Norms of forbearance also operate in Congress. Take the U.S. Senate. As a body whose original purpose was to protect minorities from the power of majorities (which, the founders believed, would be represented by the House), the Senate was designed, from its inception, to allow deliberation. It developed a range of tools—many of them unwritten—that enabled legislative minorities, and even individual senators, to slow down or block projects put forth by the majority. Prior to 1917, the Senate lacked any rules limiting discussion, which meant that any senator could prevent a vote on (or “filibuster”) any legislation indefinitely by simply prolonging debate.

Or take another example: In 1901, a routine White House press release was issued on behalf of new president Theodore Roosevelt headlined, “Booker T. Washington of Tuskegee, Alabama, dined with the President last evening.” While prominent black political leaders had visited the White House before, a dinner with a leading African American political figure was, as one historian has described it, a violation of “the prevailing social etiquette of white domination.” The response was immediate and vicious. One newspaper described it as “the most damnable outrage which has ever been perpetrated by any citizen of the United States.” Senator William Jennings Bryan commented, “It is hoped that both of them [Roosevelt and Washington] will upon reflection, realize the wisdom of abandoning their purpose to wipe out race lines.” In the face of the uproar, the White House’s press operation first denied the event happened, later said it had “merely” been a lunch, and then defended it by saying that at least no women had been present.

When American democracy has worked, it has relied upon two norms that we often take for granted—mutual tolerance and institutional forbearance. Treating rivals as legitimate contenders for power and underutilizing one’s institutional prerogatives in the spirit of fair play are not written into the American Constitution. Yet without them, our constitutional checks and balances will not operate as we expect them to. When French thinker Baron de Montesquieu pioneered the notion of separation of powers in his 1748 work The Spirit of the Laws, he worried little about what we today call norms. Montesquieu believed the hard architecture of political institutions might be enough to constrain overreaching power—that constitutional design was not unlike an engineering problem, a challenge of crafting institutions so that ambition could be used to counteract ambition, even when political leaders were flawed. Many of our founders believed this, as well.

History quickly revealed that the founders were mistaken. Without innovations such as political parties and their accompanying norms, the Constitution they so carefully constructed in Philadelphia would not have survived. Institutions were more than just formal rules; they encompassed the shared understandings of appropriate behavior that overlay them. The genius of the first generation of America’s political leaders was not that they created foolproof institutions, but that, in addition to designing very good institutions, they—gradually and with difficulty—established a set of shared beliefs and practices that helped make those institutions work.