How Democracy Works — and Why It's Under Threat

Half a century ago, democracy was the exception. By the V-Dem Institute's count, democracies were a small minority of the world's states in the mid-1970s. Then came what the political scientist Samuel Huntington called the "third wave" of democratisation, starting with Portugal's Carnation Revolution in 1974 and continuing through the end of military rule across Latin America, the fall of communist regimes in Central and Eastern Europe, and transitions in East Asia and Africa. In 1989, Francis Fukuyama's essay "The End of History?" captured the mood: liberal democracy seemed to have no serious ideological rival.

That optimism has faded. Most major organisations that measure democracy report a sustained decline, although the Economist Intelligence Unit's 2025 index, published in April 2026, found scores had stabilised. V-Dem's Democracy Report 2026 finds that, for the average person in the world, the level of democracy is back to where it was in 1978. Freedom House reports a 20th consecutive year in which more countries lost freedom than gained it. And the way democracies decline has changed. Military coups still happen, but the more common pattern is gradual erosion by governments that came to power through elections.

This article explains what democracy requires, how researchers measure it, what the four leading indices currently report, where those indices disagree, and what political science says about why democracies erode or survive. It describes research findings, not judgements about any particular party or politician.

Key takeaways

• Democracy is more than elections. Most definitions also require free expression, independent media, an independent judiciary, the rule of law and limits on executive power.
• Four indices dominate measurement: V-Dem (expert-coded), Freedom House (analyst ratings), the EIU Democracy Index (expert assessments plus surveys) and International IDEA's Global State of Democracy indices.
• Three of four still report decline; the EIU reports stabilisation. V-Dem counted 92 autocracies and 87 democracies at the end of 2025; Freedom House found 54 countries declined and 35 improved in 2025; IDEA found 2025 was the 11th straight year in which more countries declined than improved; the EIU's global average was 5.17 in 2024 (its lowest since the index began) and 5.19 in 2025, which the EIU describes as a stabilisation.
• The dominant pattern is gradual erosion ("autocratisation" or "democratic backsliding"), in which elected governments weaken the institutions that check them, rather than sudden coups.
• The indices disagree on details, especially about individual countries close to the line between democracy and autocracy, because they use different definitions and methods.

What democracy actually requires

"Democracy" is often used loosely to mean holding elections. Political scientists usually treat competitive elections as necessary but not sufficient. The tradition associated with the political scientist Robert Dahl, widely used in measurement, describes a set of institutions that must work together:

Free and fair elections. Voters can choose between real alternatives without coercion, votes are counted honestly, and losers accept the result. In practice this depends on independent electoral bodies, equal media access for opposition parties, protection from voter intimidation, and courts that enforce electoral law.

Freedom of expression and alternative sources of information. Citizens need accurate information to make choices. A free press is the main check on government between elections. Our ranking of press freedom by country shows how unevenly this is protected.

Freedom of association. People must be able to form parties, unions, and civic groups. Without it, opposition cannot organise.

Rule of law and judicial independence. Laws apply to everyone, including the government. Courts can rule against those in power without retaliation.

Checks and balances. Executive power is limited by legislatures, courts and constitutional rules. No single office controls everything.

Civilian control of the military. Armed forces obey elected civilian authorities and stay out of politics.

Protection of minority rights. Majorities govern, but within limits that protect individual rights and minorities from majoritarian abuse.

Researchers often distinguish two levels. An electoral democracy has meaningful, competitive elections with universal suffrage and the freedoms needed to make them work. A liberal democracy adds strong constraints on executive power, judicial independence and the rule of law. The gap between the two is where much of the recent erosion has occurred.

How democracy is measured

Measuring democracy means turning complex political realities into numbers. Four projects dominate, and their methods differ in important ways.

Index	Publisher	First published	Coverage	Method	Scale
V-Dem	V-Dem Institute, University of Gothenburg	Dataset 2014; annual Democracy Report since 2017	179 countries classified in latest report	Over 4,200 country experts code hundreds of indicators; statistical model aggregates them	0–1 indices; Regimes of the World categories
Freedom in the World	Freedom House (US-based NGO)	1973	195 countries and 13 territories	In-house and external analysts score 25 indicators	0–100; Free / Partly Free / Not Free
Democracy Index	Economist Intelligence Unit (EIU)	2006	167 countries and territories	60 indicators in 5 categories, from expert assessment and public-opinion surveys	0–10; four regime types
Global State of Democracy	International IDEA (intergovernmental)	2017	173-174 countries, data from 1975	Aggregates existing datasets into four categories	0–1; performance bands

Sources: V-Dem Democracy Report 2026; Freedom House, Freedom in the World 2026; International IDEA, Global State of Democracy 2025. EIU details from the publisher's report series. Founding dates and indicator counts for Freedom House and the EIU are from their published methodologies.

V-Dem (Varieties of Democracy) is based in Sweden. It asks multiple country experts, more than 4,200 in total, to answer detailed questions about their country, then uses a statistical measurement model to combine their answers and estimate uncertainty. Its headline Liberal Democracy Index combines electoral institutions with checks on the executive, judicial independence and civil liberties. Its "Regimes of the World" scheme sorts countries into four categories: liberal democracy, electoral democracy, electoral autocracy and closed autocracy.

Freedom House rates political rights (up to 40 points) and civil liberties (up to 60 points) in a 100-point score, using a team of analysts and advisers. It classifies countries as Free, Partly Free or Not Free. Because it measures rights and freedoms rather than democratic institutions as such, it is sometimes described as measuring "freedom" more than "democracy."

The EIU Democracy Index scores five categories: electoral process and pluralism, functioning of government, political participation, political culture, and civil liberties. Countries are classified as full democracies, flawed democracies, hybrid regimes or authoritarian regimes. It is unusual in including survey data on public attitudes.

International IDEA, an intergovernmental organisation based in Stockholm, groups its indicators into four categories: Representation (elections and effective parliaments), Rights (including freedom of expression, press and assembly), Rule of Law (including judicial independence) and Participation (citizen involvement during and between elections).

What the latest reports find

Index (edition, year covered)	Headline finding
V-Dem Democracy Report 2026 (end of 2025)	92 autocracies and 87 democracies; 74% of the world's population (6 billion) live in autocracies; only 7% live in liberal democracies; average citizen's level of democracy back to 1978
Freedom in the World 2026 (2025)	20th consecutive year of global decline; 54 countries declined and 35 improved; 88 countries rated Free and 59 Not Free
Global State of Democracy 2026 (2025)	11th consecutive year with more countries declining than improving; 98 countries (57%) declined in at least one factor compared with five years earlier, against 55 (32%) that improved. (The 2025 edition, covering 2024, found a ninth consecutive year and 94 countries, 54%, declining.)
EIU Democracy Index 2025 (2025)	Global average score 5.19, up from 5.17 in 2024; the EIU describes 2025 as "a stabilisation of democracy scores". The 2024 score had been the lowest since the index began, down from a 2015 peak of 5.55

Sources: V-Dem Democracy Report 2026 (executive summary); Freedom House Freedom in the World 2026 summary page; International IDEA, Global State of Democracy 2026 launch materials (15 September 2026) and the September 2025 press release for the 2025 edition; EIU Democracy Index 2025 press release (7 April 2026) and EIU Democracy Index 2024 as reported at release in February 2025.

V-Dem's report adds several details. Liberal democracies, by its count, fell from a peak of 45 in 2009 to 31 in 2025, making them the least common regime type in the world, while electoral democracies rose from 46 to 56. That shift, V-Dem argues, shows that the liberal features of established democracies, such as checks on the executive and the rule of law, are wearing away even where elections continue. It counts 44 countries as "autocratising" and only 18 as democratising, and reports that freedom of expression is the most frequently attacked aspect of democracy, worsening in 44 countries.

IDEA's 2025 report (covering 2024) emphasises elections. In the 2024 "super-cycle" of national elections, its Credible Elections measure fell to its lowest level in 30 years, and its Representation category reached its lowest level since 2001. It also found one bright spot: its Absence of Corruption measure recorded a 15th consecutive year of advances. For context on corruption trends see our ranking of the most and least corrupt countries.

Where the indices disagree

Three of the four indices agree on the overall direction (the EIU's latest edition reports stabilisation), but they often disagree on individual countries and on how severe the decline is. There are several reasons.

Different definitions. V-Dem's Liberal Democracy Index weights checks on the executive and judicial independence heavily. Freedom House measures rights and liberties. The EIU includes political culture and participation, which reflect public attitudes. A country with free elections but a weak judiciary can look quite different on each.

Different thresholds. Each index draws the line between democracy and non-democracy in its own place. V-Dem itself flags countries in the "grey zone" near the line, noting that some electoral democracies could plausibly be classed as autocracies and vice versa. Small shifts in scoring can move a populous country across the line and change global population shares dramatically.

Counting countries versus people. Population-weighted figures, such as V-Dem's 74% living in autocracies, are dominated by a handful of large countries. V-Dem notes that four of the world's five most populous countries are classed as autocracies in its data. If one of them were reclassified, the headline percentage would change sharply even with no change elsewhere.

Expert judgement and bias. All four rely on human judgement at some stage. Critics, including some governments rated poorly, argue that the indices reflect Western or liberal assumptions, or that expert coders share particular political views. The index producers respond that they use multiple coders, published methods and statistical checks. V-Dem in particular publishes uncertainty estimates and revises past scores when new data arrive, which means figures in older reports can change.

The "perception gap" debate. Some political scientists, notably Andrew Little and Anne Meng in a 2023 paper, have argued that objective indicators such as incumbent election losses show little global decline, and have suggested that coder bias in expert-coded measures may help explain the gap. Others, including V-Dem researchers, have challenged that argument and found no evidence of such a bias. The debate is unresolved, but it is a reminder that democracy scores are measurements with error, not direct observations.

Readers who want country-by-country scores can compare them in our ranking of the most and least democratic countries.

Why democracies form and persist

Political science has identified several factors associated with democratic survival, though none guarantees it.

Economic development. Richer countries are much more likely to be stable democracies. In an influential 1997 study in World Politics, Adam Przeworski and Fernando Limongi analysed 135 countries between 1950 and 1990. They found that rising income did not reliably cause countries to become democracies, but that once a country was democratic, higher income made breakdown much less likely. Explanations include a larger middle class with economic independence, more education, and lower stakes in losing power.

Elite commitment. Democracy works when those with power, including parties, the military and business, accept losing elections as preferable to breaking the rules. That commitment weakens when losing appears to carry very high costs, such as prosecution or expropriation.

Civil society. Independent associations such as unions, professional bodies, religious groups and civic organisations give citizens ways to organise and resist government overreach.

Inequality and trust. High inequality is associated with lower trust in institutions and more polarised politics, though the causal links are debated. Data on wealth inequality and its causes show how unevenly gains have been shared within many democracies.

International environment. Neighbouring democracies, trade ties and membership conditions in organisations such as the European Union have helped anchor democratic transitions, especially in Central and Eastern Europe.

Participation. Turnout is only one sign of democratic health, but sharp declines can indicate disengagement. Our table of voter turnout by country shows wide variation even among established democracies.

How democracies erode in the 21st century

In How Democracies Die (2018), the Harvard political scientists Steven Levitsky and Daniel Ziblatt argued that most recent democratic breakdowns have come not from coups but from elected governments. V-Dem's researchers describe the same process as "autocratisation." The pattern described in the research literature typically includes several steps, not always in this order:

1. An election victory, often reflecting real public grievances.
2. Changes to the courts, such as altering appointment rules, retirement ages or the size of courts in ways that give the governing side control.
3. Pressure on the media, through regulation, licensing, state advertising, ownership changes or legal action against critical outlets.
4. Changes to electoral rules that favour incumbents, such as redrawing districts, changing campaign finance rules or reshaping electoral commissions.
5. Use of state power against opponents, through investigations, tax audits or prosecutions.
6. Control of independent institutions, such as prosecutors, central banks or audit bodies.
7. Constitutional changes that lock in the new arrangements.

Each step can usually be presented as a reasonable reform, which is what makes the process hard to resist. Researchers stress that any single change of this kind can be legitimate; it is the cumulative pattern that indices try to capture. V-Dem reports that freedom of expression is currently the most frequently attacked component of democracy among the countries it tracks.

How the indices classify specific cases. V-Dem's data have classified Hungary as an electoral autocracy since 2018, and Freedom House has rated it Partly Free since its 2019 edition (65/100 in 2026), making it the first EU member state to lose Free status in that index. Hungary's Fidesz government lost the April 2026 parliamentary election to the Tisza party, so these ratings describe the period through 2025 and may change in later editions. Turkey, an EU accession candidate since 1999, has been rated Not Free by Freedom House since its 2018 edition (32/100 in 2026), following a failed coup attempt in 2016 and a 2017 referendum that replaced the parliamentary system with an executive presidency. Poland is often cited as a case of partial reversal: its scores fell after 2015 and began to recover after a change of government in 2023. These classifications are disputed by the governments concerned.

The United States has also become part of the debate. V-Dem's 2026 report states that the US has lost its status as a liberal democracy in its data, and Freedom House listed the US among the Free countries with the largest declines in 2025. Both remain contested assessments, and the US remains classed as a democracy in all four indices; the claims describe changes in specific indicators, not a regime change.

Why now? Explanations researchers debate

There is no single accepted explanation for the current wave of decline. The main candidates are:

Economic shocks. The 2008 global financial crisis and subsequent slow wage growth in many countries are associated with falling trust in mainstream parties and institutions.

Cultural and demographic change. Rapid social change, migration and shifting values have produced political backlash in some countries, which researchers such as Pippa Norris and Ronald Inglehart have studied.

Information environment. Social media lowered the cost of reaching voters directly, bypassing traditional gatekeepers such as editors and party organisations. Researchers debate how much recommendation systems, like the one described in our explainer on how the TikTok algorithm works, contribute to polarisation. Classic psychology findings on how misinformation alters memory help explain why false claims can be persistent once spread.

Polarisation. When supporters of rival parties see each other as threats rather than opponents, they become more willing to tolerate rule-breaking by their own side. This is one of the mechanisms Levitsky and Ziblatt emphasise.

International factors. The influence of large autocracies, the decline of democracy promotion as a foreign-policy priority in some countries, and changing alliances all affect the environment in which democracies operate.

Historically, changes in communication technology have repeatedly reshaped politics. The printing press both spread new ideas and fuelled religious and political conflict; the debate about digital media echoes that history.

Can democracy recover?

The history of democracy includes many recoveries. Spain moved from dictatorship to democracy after 1975. South Korea and Taiwan transitioned from authoritarian rule in the late 1980s and 1990s and are now rated as democracies by all four indices. V-Dem's data identify "U-turns," in which countries that began to autocratise reversed course, and its 2026 report counts 18 countries currently democratising.

Researchers point to several features of successful recoveries: broad opposition coalitions that set aside differences, independent courts or electoral bodies that held firm, strong civil society, and high public participation in elections. The evidence also suggests recovery becomes harder the longer erosion continues, because the rules governing elections and the media become less fair.

Common misconceptions

"If a country holds elections, it is a democracy." Most indices classify countries with regular but unfair elections as "electoral autocracies" or "hybrid regimes."

"Democratic decline means democracy is collapsing everywhere." By V-Dem's count, democracies were far scarcer in the 1970s than the 87 it counted at the end of 2025. Decline is real but uneven.

"The indices are all saying the same thing." They agree on direction but differ on many individual countries and on the size of the decline, for the reasons set out above.

"Erosion only happens in new democracies." V-Dem reports backsliding in some long-established democracies, as well as breakdowns in newer ones and deepening autocracy in already autocratic states.

Frequently asked questions

Q: What are the main ways of measuring democracy? A: Four indices dominate: V-Dem, based on more than 4,200 country experts; Freedom House's Freedom in the World, scoring political rights and civil liberties out of 100; the EIU Democracy Index, scoring five categories out of 10; and International IDEA's Global State of Democracy indices, covering 173 countries.

Q: Is democracy declining worldwide? A: Most indices say yes, though not all agree on the latest trend. V-Dem counted 92 autocracies and 87 democracies at the end of 2025, Freedom House recorded a 20th consecutive year of net decline in 2025, and IDEA reported an 11th year in which more countries declined than improved. The EIU's global average hit a record low of 5.17 in 2024 but edged up to 5.19 in 2025, which the EIU calls a stabilisation.

Q: Why do democracy rankings disagree about some countries? A: They use different definitions, indicators and thresholds. V-Dem emphasises checks on executive power, Freedom House measures rights and liberties, and the EIU includes political culture and participation. Countries near the boundary between democracy and autocracy can be classified differently.

Q: What is democratic backsliding? A: It is the gradual weakening of democratic institutions by those in power, usually through legal or quasi-legal means such as changing court appointments, media rules or electoral laws, rather than through a sudden coup. V-Dem calls the process autocratisation.

Q: What is the difference between an electoral and a liberal democracy? A: An electoral democracy has free, fair and competitive elections with the freedoms needed to make them meaningful. A liberal democracy also has strong limits on executive power, independent courts and robust rule of law. V-Dem counted 31 liberal and 56 electoral democracies in 2025.

Q: Does economic development protect democracy? A: It helps. Przeworski and Limongi's 1997 study of 135 countries found that wealthier democracies were much less likely to break down, though rising income did not reliably cause transitions to democracy in the first place.

Q: Can a democracy that has eroded be restored? A: Yes. Spain, South Korea and Taiwan all moved from authoritarian rule to democracy, and V-Dem identifies countries that reversed an autocratising trend. Researchers find recovery is easier when it begins early, before electoral and media rules have been reshaped.

The bottom line

Democracy is not simply a matter of holding elections. It depends on a web of institutions and norms, including free media, independent courts and protected civil liberties, that allow citizens to choose and replace their governments. Despite different methods and frequent disagreements over individual countries, three of the four leading measurement projects report that this web has been weakening globally for well over a decade, mostly through gradual erosion rather than sudden coups, while the EIU's latest edition reports that its scores have stabilised. They also show that the world remains far more democratic than it was 50 years ago, and that recoveries happen. Understanding how the indices work, and what they can and cannot show, is the best starting point for judging claims about democracy's health.

Sources

• V-Dem Institute, Democracy Reports, including Democracy Report 2026: Unraveling the Democratic Era?
• Freedom House, Freedom in the World 2026: The Growing Shadow of Autocracy.
• International IDEA, Global State of Democracy 2026: Democracy in an Era of Conflict (September 2026); Global Democracy Report: Majority of Countries Worsen as Press Freedom Hits 50-Year Low (September 2025), and The Global State of Democracy Initiative.
• Economist Intelligence Unit, EIU Democracy Index 2025 press release (7 April 2026), "Democracy stabilises after eight years of decline".
• Przeworski A, Limongi F. "Modernization: Theories and Facts." World Politics 49(2), 1997, 155–183. Cambridge Core.

Related Reading

• How Evolution Works — and the Evidence for It
• What Is Quantum Mechanics — and Why Does It Matter?
• How Nuclear Weapons Work — and Why They Haven't Been Used Since 1945
• How Space Exploration Works — and What's Next
• What Is the Stock Market — and How Does It Actually Work?