---

Summary

Key Takeaways from the Research

• Direct democracy–the people making laws directly through ballot propositions–has been part of American democracy from the beginning.
• Its use has gradually expanded across the country over time, and other nations routinely hold referendums to decide important public policies; it is a time-tested, accepted democratic tool.
• The evidence suggests that direct democracy allows citizens to bring public policies into better alignment with popular preferences, especially in situations where special interests are influential in the legislature.
• The preponderance of evidence does not support the fear that voters lack the competence to participate in making important public decisions, that direct democracy empowers special interests, or that minority rights will be undermined by direct democracy.

Important Questions We Need Better Answers To

• Does direct democracy, by empowering the majority, pose a potential risk to minority rights? While the existing evidence has yet to identify compelling examples of this danger, the evidentiary base is thin, allowing some doubt about the magnitude of this risk.
• What would be the collateral effects of holding national referendums in the United States, given what a dramatic change it would represent for American democracy? We can make intelligent conjectures about how it would work based on experience in the states and other nations, but the United States is sui generis, so these are only informed conjectures.
• To what extent would more direct democracy relieve populist pressures? While there are plausible stories for why it might have this effect (and perhaps also reduce polarization), the evidence on these specific causal relationships is yet to be securely established.

---

Introduction

In states where ballot measures are common, they can rival the legislature as a lawmaking process.

Democracy typically involves electing representatives to make laws, but that is only part of it. In most democratic nations, voters also make laws themselves through ballot measures. This is called direct democracy, the people legislating directly without intermediation by public officials.

Almost every democratic nation holds referendums to vote on issues of importance, as do all American states and most American cities. In the American states, the issues have ranged from women’s suffrage, old age pensions, workers’ compensation, and tax reform in the early 20th century; to taxes and spending, abortion, same-sex marriage, minimum wage, renewable energy, and criminal justice reform in recent years. Direct democracy has been especially important for reforming democracy itself, putting issues such as term limits, open primaries, citizen redistricting, and campaign contribution limits on the agenda.

In states where ballot measures are common, they can rival the legislature as a lawmaking process. In California, for example, the amount of money spent on ballot measure campaigns from 2000 to 2020 was $4.2 billion, almost three times the $1.5 billion spent on state legislator campaigns over the same time period.¹ If money is any indicator of policy importance, direct democracy may even dominate the legislature in the state.

History and Basic Facts

Direct democracy is an ancient form of democracy. More than two millennia ago, the ancient Greek city-states and the Roman Republic had vibrant citizen assemblies that were responsible for important public decisions, such as passing laws, declaring war, and approving peace treaties.

Although citizen assemblies are still used in small towns in New England and Switzerland, it isn’t feasible for nations, states, and most cities to convene their citizens in one place. Instead they make decisions by voting on ballot measures (or “ballot propositions.”)

In ballot measure elections, citizens vote “yes” or “no” on a proposed law. Although all ballot measures are about making laws directly, there are important differences in how they reach the ballot.

Initiatives

An initiative is a law or constitutional amendment proposed by citizens that is qualified for the ballot by collecting signatures from other citizens, called “petitioning”. The required number of signatures is set by law, for example, 5% of the votes cast in the previous general election.

Veto referendums

A veto referendum is also qualified for the ballot by citizen petition, but instead of proposing a new law, it proposes to repeal an existing law passed by the legislature.

Legislative measures

Most ballot propositions are placed on the ballot by the legislature, not by citizen petition. Sometimes they are required by law, for example, almost all states require a constitutional amendment proposed by the legislature to be approved by the voters. Sometimes they are advisory in nature, such as the UK’s Brexit referendum which asked voters if they wanted to leave or remain in the European Union.

“Referendum” is an umbrella term that is used to describe any type of ballot measure, whether initiative, veto referendum (somewhat confusingly), legislative, or other. Referendum is synonymous with plebiscite.²

Voting on important public issues is an American tradition that goes back to the founding of the Republic, when the state of Rhode Island held a referendum on whether to join the Union. (Voters said no in that referendum). The states began holding referendums to approve amendments to their constitutions in the early 1800s, and throughout the 19th century gradually required voter approval on other matters of interest, such as location of the state capital and issuance of state debt. The most dramatic expansion of direct democracy occurred during the Progressive Era of the early 20th century when many states adopted the initiative and veto referendum. The initiative and veto referendum were one plank of a broad reform platform targeted at rooting out corruption and special interest influence in government.³

Figure 1: States Providing Either Initiative or Veto Referendum

In the United States today, 26 states permit initiatives or veto referendums (Figure 1), and all 50 states use legislative measures from time to time. At the local level, 82% of cities allow initiatives for local laws, including 97% in the west, 82% in the south and northeast, and 59% in the central part of country.⁴

---

The main rationale for direct democracy is to give the people a fallback option if their elected officials fail to represent their interests.⁵

There are several reasons why elected officials might not represent voters. They may become captured or unduly influenced by special interests, such as corporations, unions, and individuals that provide them with campaign money. During the Progressive Era, presidents of both parties, Republican Theodore Roosevelt and Democrat Woodrow Wilson, called for adoption of the initiative and veto referendum to help the people retake their government from the power of industrial corporations and urban political machines. In Roosevelt’s words, “The trouble has been . . . that at many times and in many localities, there have held public office in the states and in the nation men who have, in fact, served not the whole people, but some special class or special interest.”

Another reason that representatives might not accurately represent the electorate is gerrymandering. Gerrymandering is the practice of drawing district lines to give one party a greater share of seats in the legislature than the share of votes they receive from the voters. This overrepresentation of one party can lead to extreme partisan policies that don’t reflect the views of the general public. Initiatives and referendums allow voters to correct such distortions and bring policy back to the center.

Direct democracy improves representation by allowing voters to override their elected legislators, but also by providing a threat to override them. In states where initiatives and referendums are available, legislators may adjust their policy choices specifically to avoid attracting a hostile ballot measure. This implies that direct democracy has an effect on policy even when it is not directly used. The existence of this threat effect is supported by anecdotes and some statistical evidence.

Direct democracy also helps solve the representation problem that stems from the fact that a representative takes positions on a bundle of issues, not just one. Voters may disagree with their representative’s position on one issue but agree on all of the others. They could remove their representative from office at the next election to correct the noncongruent issue, but there may be no alternative candidate who is congruent on all issues. Direct democracy allows voters to override their representative on the one issue where they disagree without having to remove the representative from office, essentially unbundling the one issue.

Some direct democracy advocates also argue that voting on issues creates a more informed and engaged electorate, and builds trust in government. However, this research relies on cross-sectional methods, so does not conclusively rule out that a common factor (such as political culture) causes both adoption of direct democracy and the information and engagement outcomes of interest.⁶ There is also some evidence, again mostly based on cross-sectional correlations and therefore with the same limitations, that citizens trust government more and are more satisfied with their lives in states and cities where direct democracy is often used.⁷ There is more convincing evidence that having propositions on the ballot increases turnout in low-information elections (such as midterm elections), but not necessarily during presidential elections. This evidence comes from panel data, essentially using within-state variation in the number of propositions to predict turnout.⁸ There is also evidence that citizens view decisions made by direct democracy as more legitimate.⁹

Concerns about Direct Democracy

Decades of political science research have established beyond any doubt that ordinary voters have limited knowledge about the most basic facts of government and politics, and more to the point, they know little about the details of the ballot measures on which they vote. Can they be trusted to make important public decisions?

The answer is yes, because, somewhat counterintuitively, voters do not need to know the details of the laws on which they vote in order to make good decisions. They can vote their interests by relying instead on information “cues” or “shortcuts” in the form of endorsements.¹⁰ Consider, for example, a green voter facing a ballot measure titled “Forest Preservation Initiative.” The voter needs to know if the measure actually preserves forests, or if it is a stealth law that allows the timber industry to cut more trees. The voter could read the detailed legalese of the text and try to parse out the effects, but this would be challenging for anyone who is not a lawyer or a regulatory expert on the industry. The Sierra Club, though, has lawyers that can read the law, and therefore knowing the endorsement of the Sierra Club allows green voters to accurately register their interests at the polls, even though they don’t end up understanding the details of the proposal. Similarly, voters in logging communities can vote their interests based on the recommendation of the timber industry. If enough information shortcuts are available, voters can accurately vote their interests, and research and casual observation shows that many voters behave in just this way.¹¹

Some critics argue that voters are irrational, and if given the power to set policy, would simultaneously cut taxes and increase government spending. This is not what happens: there is an enormous literature on fiscal policy and direct democracy spanning many countries, levels of government, and time periods showing that differences in taxes and spending are modest between states and cities with and without direct democracy (initiative states actually tend to spend less in the postwar period), and debt levels are not detectably different. Most of these studies rely on panel data methods, attempting to control for economic, demographic, and cultural factors that might otherwise drive differences. Some recent research uses regression discontinuity methods exploiting the fact that in some nations direct democracy becomes available when the population crosses a statutory threshold. On social issues, there is a modest tendency for initiative states to choose more conservative policies, but there is a lot of variance around this central tendency. Most of these findings come from panel data regressions as well.¹²

An important group of policies that are frequently the subject of direct democracy are government reforms. Initiative states have been more likely to adopt term limits, nonpartisan redistricting, and open primaries than noninitiative states. Whether these are good or bad reforms is a subjective call, of course, but it is hard to view them as irrational. Early in the 20th century, initiatives were used to advance women’s suffrage, equal-population redistricting (before the one-person-one-vote rule was established), and direct election of U.S. senators.

Does direct democracy allow wealthy and organized special interests to dominate policy?

Rich business groups and individuals dominate ballot measure campaigns. Of the $4.2 billion spent on ballot measure campaigns in California during 2000-2020, $2.0 billion was spent by corporations, $647 million was spent by unions, and $641 million was spent by rich individuals.¹³ While wealthy businesses, groups, and individuals likely have disproportionate influence in direct democracy elections, those same groups and individuals also likely have disproportionate influence under representative democracy, so the question is not if the playing field is tilted in their favor but whether it is more or less tilted in their favor under direct than representative democracy.

The evidence is fairly clear that businesses and unions fare worse under direct than representative democracy. A recent study examining the content of every business-related state-level initiative in the United States going back more than a century found that anti-business proposals were much more common than pro-business proposals.¹⁴ Direct democracy counteracts rich special interests primarily because it shifts lawmaking away from elected officials where special interests have a comparative advantage. With elected officials, interest groups can lobby, make campaign contributions, give jobs to family members, and so on, but none of these levers can be used on a mass electorate under direct democracy. The fact that business groups are usually opposed to direct democracy is also revealing.

Does direct democracy threaten the rights of numerical minorities?

The purpose of direct democracy is to empower the majority. This creates a risk that the majority will use its power to oppress minority groups, whether defined on the basis of race, sexual orientation, religion, etc. or on the basis of property (the have-nots voting to confiscate the wealth of the haves.) This is a concern for representative democracy as well because a tyrannical majority can act through its representatives.

There is no bulletproof protection against this risk, but there are safeguards to mitigate it. One of the safeguards is judicial review, which applies equally for direct as for representative democracy. Another safeguard is separation of legislative from executive power, which also applies to direct democracy, since voters can pass laws but do not administer them.

We only have limited and mixed evidence on whether direct democracy leads to more antiminority laws, and most of it is limited to racial, language, and sexual orientation groups.

Empirically, we only have limited and mixed evidence on whether direct democracy leads to more antiminority laws, and most of it is limited to racial, language, and sexual orientation groups. Somewhat reassuringly, one study found that racial/ethnic minority voters were just as likely to be on the winning side of ballot measure elections as white voters, and opinion surveys show that substantial majorities of racial and ethnic minorities support the initiative process.¹⁵

---

Pro-Democracy Reforms and  Trojan Horse Reforms

Almost alone amongst its peers, the United States does not allow direct democracy at the national level, and has never held a national vote on an issue.

The country that pioneered democracy is now using somewhat antiquated forms of democracy. European nations have held over 50 national votes on matters related to European integration, and often try to resolve controversial issues with national votes. Examples include Brexit in the UK, and votes on abortion and divorce in Ireland and Italy. Switzerland has voted on hundreds of national initiatives since adopting the process in 1858. National referendums are also used in Latin America, such as Colombia’s vote on whether to approve a peace agreement with a militia group, and in Asia, where Taiwan adopted a national initiative law in 2003, and recently voted on same-sex marriage, food imports from Japan, nuclear power plants, and lowering the voting age.¹⁶

One reform would be to hold national votes on issues of importance in the United States. Opinion surveys consistently indicate that supermajorities of Americans overwhelmingly would favor doing so.¹⁷ The easiest path would be for Congress to call occasional advisory votes on matters of importance, for example, asking the people how they would like to address the immigration crisis at the border. This would not require a constitutional amendment. Holding national referendums would help counteract the growing sense among ordinary Americans that they have lost control of their government.

Another reform would be to make the initiative and veto referendum available in all states. Opinion surveys show that majorities in every state would like to have this right in their state. Adopting the initiative and veto referendum in a state requires a state constitutional amendment, which for the most part requires the legislature to take the first step and propose one. The challenge is that most legislators appear to dislike direct democracy these days.

In addition to making the initiative and referendum available where they are now unavailable, there are potential reforms to the processes where they exist. Many process reforms years have been proposed by legislators in recent years, and unfortunately, many of them are Trojan Horses – allegedly intended to improve the processes, but in fact likely to hamstring them. This steady chipping away at the initiative and referendum as “direct democracy backsliding” has been going on for at least for half a century.¹⁸

Potential process reforms address the following issues:

Campaign spending

The amount spent on ballot propositions can be enormous, especially in a big state. The record for a single campaign is $224 million spent in 2020 on California’s Prop. 26, an initiative that overrode a union-backed state law prohibiting drivers for Uber, Lyft, DoorDash, and other app-based companies from working as independent contractors. This was a battle between rich corporations that contributed $205 million in support and rich labor unions that raised $19 million in opposition, with voters/consumers making the decision.

While there have been attempts to restrict campaign spending, the Supreme Court’s “Bellotti” decision in 1978 essentially prohibited any limits on spending. As a result, there are no realistic reform proposals to limit campaign spending. One important reform, however, would be to provide limited public funding if there is no spending on one side of an issue. As discussed earlier, voters are capable of making reasoned decisions if they have access to information shortcuts, but if one side has no money to communicate, voters may not receive the information and fail to register their interests at the polls. The proposal is not to equalize spending, but simply to make sure that each side has a minimum amount available. The UK provided public funding to both sides of the debate in its Brexit referendum; its system offers a road map for administering such a program, especially ensuring that the money gets into the hands of legitimate groups associated with both sides.

Petitioning, requirements, distribution, paid petitioners

To qualify an initiative or veto referendum for the ballot, sponsors must collect signatures. The regulation of petitioning is an actively contested area, with legislators often proposing rules that raise the cost of collecting signatures. Most petitioning is done by paid petitioners who work for signature collection firms. Courts have rejected bans on paid petitioners, but some states ban payment by the signature. Proponents of such bans claim that paying by the signature leads to fraud, but this looks like a Trojan Horse reform because there does not appear to be any reliable evidence that bans mitigate fraud–but they certainly make it harder to qualify measures for the ballot. Some states have attempted to ban outof- state petitioners, but that has also been rejected by courts. Another Trojan Horse reform is requiring petitioners to be certified, undergo training, and jump through other bureaucratic hoops with no apparent benefit to the processes.

A very significant way to limit the use of initiatives and referendums is to require a geographic distribution of signatures: for example, requiring petitioners to collect signatures from 10% of the electorate in the state and also from 10% in every congressional district within the state. Proponents of this Trojan Horse reform claim that it prevents one populous region of the state from dominating the lawmaking process. But since final approval of any proposal must attain a statewide majority, imposing this requirement at the petition stage protects against a problem that is more apparent than real, while at the same time, significantly hamstringing citizen petitioners.

The regulation of petitioning is an actively contested area, with legislators often proposing rules that raise the cost of collecting signatures.

Approval rates

In most cases, ballot measures become law if they receive a majority of votes. In a few cases, supermajorities are required, such as in Florida, which requires 60% approval for constitutional amendments. In 2023, the GOP-controlled legislature in Ohio attempted to amend the state constitution to raise the threshold for approval of amendments to 60%, with the blatant goal of making it more difficult for voters to approve an anticipated pro-choice initiative amendment. Voters saw through the ploy, rejected the supermajority amendment, and approved the abortion amendment. There is a principled argument for requiring more than 50% plus one vote to adopt a constitutional amendment, so states should consider such reforms, but they should not be adopted simply to forestall an anticipated proposal.

---

Conclusion

The goal of democracy is for the people to govern themselves. This goal is never fully realized, but in search of it, democracies employ a rich array of institutions such as regular elections, term limits, and fragmentation of government power across branches to bring policy into alignment with voter preferences and obstruct antidemocratic special interests. Direct democracy is one more tool that many nations, states, and cities have found a useful addition to their democratic toolkits.

Direct democracy is one more tool that many nations, states, and cities have found a useful addition to their democratic toolkits. Direct democracy’s solid record in supporting popular rule provides a solid justification for numerous reforms. Governments that do not have initiatives and referendums can add them, and those that have them can expand their scope and make them more accessible to groups without deep pockets.

Direct democracy’s solid record in supporting popular rule provides a solid justification for numerous reforms.

At the same time, would-be reformers should keep in mind that sitting elected officials are often hostile to direct democracy and actively work to roll it back where possible. Just in the last year, several states have seen legislatures and election officials seek to keep initiatives off the ballot, write deceptive ballot descriptions for those on the ballot, increase signature requirements to qualify initiatives, and make it more difficult for voters to approve measures, among other things. Pro-direct-democracy reformers will have to invest resources not only in expanding the availability of the process but in defending it against rollbacks from established interests.

---

References

---