T407016 // Autumn 2026 // Université de Genève
Comparative
Politics
Session 04. Democratization and democratic erosion.
Last week, this week
- Levitsky and Way on regimes that hold real elections and still tilt the field
- Munck and Verkuilen on how democracy indices are built and where they fall short
- Acemoglu and Robinson on why elites hand over power, and when they take it back
- Lührmann and Lindberg on how democracies decline today
- One group exercise
Last week we sorted regimes into types. This week we ask how countries move from one type to another.
One question before the readings
Why would a ruling elite ever hand power to the people it rules?
What has to happen first?
Why not buy off the protesters with money instead?
Once power is handed over, what stops the elite from taking it back?
Four paths
Consolidated democracy
Reform acts from 1832 widened the vote in steps, with no significant reversal.
Unconsolidated democracy
Democracy became real with the 1912 electoral reforms. Coups returned the country to dictatorship more than once.
Nondemocracy, little force
The regime lasts with minor concessions and without significant repression.
Nondemocracy by repression
Apartheid survived on repression and violence until the regime was forced to democratize in 1994.
"Why are some countries democratic whereas others are ruled by dictatorships or other nondemocratic regimes?"
Three building blocks
Interests
People rank regimes by what they deliver. In 1839 the Chartist J. R. Stephens called universal suffrage "a knife and fork question."
Conflict
Every policy creates winners and losers. The elite may be the rich, a racial or ethnic group, or the military. The citizens are more numerous.
Institutions
Political institutions allocate power today and tomorrow. Changing them is a way to commit to future policy.
The models treat regimes as a dichotomy, with free and fair elections and free entry into politics at the core. Compare that with the four attributes Munck and Verkuilen asked for last week.
Two kinds of power
Power that political institutions allocate. In the authors' example, Labour decides policy in Britain because voters put it in office.
What a group can do by force. Excluded citizens are the majority, and they can riot, strike or threaten revolution.
A real threat needs the masses to solve their collective-action problem and build momentum, and it needs a military that fails to stop them. Such moments pass.
Actual power combines the two. Institutions last, while the citizens' de facto power comes and goes.
Why not just promise?
Citizens know the elite would honor the promise only while it fears them. A promise nobody believes will not stop a revolution.
Put future decisions in the hands of the people who want the policy. Democratization moves de jure power to the majority and makes the commitment credible.
"The principle of my reform is, to prevent the necessity of revolution . . . reforming to preserve and not to overthrow."
Earl Grey presenting his electoral reform to Parliament in 1831.
Concede, democratize or repress
Redistribute today and promise more tomorrow. Cheap for the elite, and useless once the promise loses credibility.
Hand over de jure power. The new institutions persist because parties and unions invest in them, and the majority gains some say over the military.
Keep power with no concessions. Lives and assets are destroyed, sanctions may follow as they did for South Africa in the 1980s, and failed repression can bring revolution.
Democracy arises "when concessions are not credible and repression is not attractive because it is too costly."
Coups run the same logic backward
| Democratization | Coup | |
|---|---|---|
| Who holds power for now | Citizens, in a moment of unrest | The elite, allied with the military in a crisis |
| Whose promise is doubted | The elite's promise to redistribute | The citizens' promise to moderate |
| What locks it in | Democracy | Nondemocracy |
| Example | Argentina 1912, after unrest led by the Radical Party | Chile 1973, during an oil shock and a deep depression |
A coup turns the elite's passing de facto power into lasting de jure power. The more democracy redistributes, the more a coup is worth to the elite.
Democracy in a picture
- Region A. Inequality is low, so the poor have no reason to revolt. Singapore stays a nondemocracy with little repression.
- Region B. Revolution is a threat and repression costs too much, so the elite concedes. Britain and Argentina.
- Region C. Democracy would cost the elite so much that repression is worth it. South Africa until 1994.
- Inequality fell, while industry and globalization pushed the boundary down to the pink line. South Africa crossed into Region B between 1970 and 1994.
Transitions are likeliest at middle levels of inequality. Figure 2.2 adds that higher inequality makes coups pay, which separates Britain from Argentina.
What shifts the odds
An organized majority makes the revolutionary threat real and coups harder to pull off.
Depressions, debt crises and wars hand temporary power to both sides. Argentina democratized after the Falklands War, and Allende fell during the 1973 oil shock.
Land is easy to tax and its owners lose less from turmoil. Landed elites repress and back coups. Industrial elites concede.
A large, affluent middle class keeps redistribution modest. Compare Costa Rica and Colombia with Guatemala, El Salvador and Nicaragua.
Mobile capital is hard to tax, so democracy frightens elites less. Trade helps or hurts depending on what a country has in abundance.
Conflict can run along race or ethnicity, as in South Africa or Mauritius. Many cross-cutting cleavages keep redistribution low and help consolidation.
Each factor changes what democracy would cost the elite, or what repression and coups would cost it.
Guarantees for the old elite
- Pinochet's constitution under-represented urban areas and kept a military veto. On this account it eased Chile's transition and produced a less redistributive, more stable democracy.
- South Africa's interim constitution gave any party with twenty or more seats cabinet posts in proportion to its seats. The ANC accepted limits on its own power to get the transition.
- Colombia's two traditional parties used the design of proportional representation to keep the left out. Democracy consolidated, amid complaints that it ignores the majority.
"Ironically, the greater security for the armed forces during the initial years of the transition probably reduced the threat to civilian authority in Chile, Turkey, and Korea."
- The limit
- Give the elite too much power and democracy does little for the majority. Revolution or repression returns.
A transition deal has to protect the old elite enough to prevent a coup and leave the majority enough to prevent a revolt.
Why "autocratization"?
Three problems with "democratic backsliding"
An autocracy cannot suffer democratic backsliding into a deeper dictatorship. Stretching the term that far is what Sartori warned against.
The word implies a return to an earlier state. Regimes can also move in a new direction, toward a different form of authoritarianism.
Sliding sounds involuntary and unconscious. Rulers who dismantle democracy act on purpose.
Autocratization is a "substantial de-facto decline of core institutional requirements for electoral democracy."
One umbrella, three kinds
A democracy loses democratic traits and stays a democracy, so far. Hungary under Orbán.
A democracy becomes an autocracy. Fujimori suspended Peru's constitution and parliament in 1992.
An autocracy gets worse. Al-Bashir's 1989 coup turned Sudan's electoral autocracy into a closed dictatorship.
Coding an episode
- A potential episode starts when the Electoral Democracy Index falls by 0.01 or more in one year.
- It continues while the index keeps falling, through up to four years with no further fall.
- It ends after four years with no fall, or as soon as the index rises by 0.02 or more in a year.
- It counts as an episode if the total fall, measured from the year before it started, reaches 0.10.
- Index
- V-Dem's EDI, from 0 to 1, built on Dahl's polyarchy
- Coverage
- 182 countries from 1900 to 2017, or 18,031 country-years
- Example
- Hungary 2006 to 2017, a fall of 0.11
The authors call the 0.10 cut-off "naturally arbitrary." What would Munck and Verkuilen ask about it?
A century of episodes
Autocratization episodes, in 109 countries.
Began in countries that were already autocracies, 142 episodes in all.
Began in democracies.
Of those 75 ended with the country an autocracy.
69 countries had no episode. 33 of them are autocracies with little left to lose, and 36 are democracies such as Sweden, Switzerland, Bhutan and Namibia.
Three waves of autocratization
32 episodes, in democracies and autocracies alike.
62 episodes, almost all in electoral autocracies.
75 episodes by 2017, mostly in democracies.
In 2017 more countries were autocratizing than democratizing, for the first time since 1940. The share of democracies still stood near its peak, at 53 percent.
Behind a legal façade
How episodes that began in democracies unfolded
An executive who took office legally suddenly abolishes elections or parliament. Fujimori in 1992, and Hitler with the Enabling Act in 1933.
An executive who took office legally wears down democratic norms gradually and leaves the key institutions standing. Hungary and Poland.
Slower and shallower
| Episodes in democracies | Before 1994 | 1994 to 2017 |
|---|---|---|
| Most common route | Coups, invasions, self-coups | Democratic erosion |
| Median fastest one-year fall | 31% | 8% |
| Median EDI at the end | 0.13 | 0.45 |
| Median total fall | 0.50 | 0.19 |
Germany's 1930-35 episode included a one-year fall of 26 percent of the scale. Turkey from 2008 and Russia from 1993 never lost more than 7 and 5 percent in a year.
Alarm or calm?
- More than half of all countries were democracies in 2017
- Third-wave episodes end at a median EDI of 0.45, against 0.13 before
- Open coups and stolen elections are costly at home and abroad, so would-be autocrats "play a game of deception"
- The wave mostly hits democracies
- Of 33 erosion episodes, 18 ended in breakdown and 10 were still running in 2017. Only 5 stopped short.
- Gradual decline gives citizens fewer clear moments to mobilize
"As it was premature to announce the 'end of history' in 1992, it is premature to proclaim the 'end of democracy' now."
Two readings, one puzzle
| Acemoglu and Robinson 2006 | Lührmann and Lindberg 2019 | |
|---|---|---|
| Question | Why do elites concede democracy, and when do they overturn it? | How does today's autocratization differ from earlier waves? |
| Actors | Social groups, elites against citizens | Incumbents, most of them elected |
| Evidence | A formal theory, illustrated with cases from Britain to South Africa | V-Dem data on 217 episodes since 1900 |
| How regimes change | In jumps. Democracy under threat of revolution, coups in a crisis | Mostly by gradual erosion behind a legal façade |
| Last week | Two categories, democracy and nondemocracy | A continuous index, plus a 0.10 threshold to call an episode |
In Acemoglu and Robinson, a coup turns passing power into lasting power. Is an elected leader who packs the courts doing the same thing?
A democracy-support fund sends your group four country files
Our rules open a review once an autocratization episode is confirmed. Which of these countries qualify?
| Country | File | 2017 | 2018 | 2019 | 2020 | 2021 | 2022 | 2023 | 2024 |
|---|---|---|---|---|---|---|---|---|---|
| A | A president elected in 2017 replaced four constitutional judges in 2018, put the media regulator under a minister in 2019 and won a 2021 referendum lifting term limits. The opposition holds a third of the seats. | 0.74 | 0.71 | 0.68 | 0.68 | 0.64 | 0.63 | 0.63 | 0.62 |
| B | An elected government passed land reform in 2018 amid a wave of strikes. In 2019 the army closed parliament. A few families tied to the officer corps own most export plantations. | 0.67 | 0.67 | 0.21 | 0.20 | 0.23 | 0.24 | 0.24 | 0.25 |
| C | The ruling party has won every election since independence. It jailed the main opposition leader in 2018, passed a foreign-agents law for NGOs in 2019 and closed the last independent daily in 2022. | 0.38 | 0.34 | 0.32 | 0.32 | 0.31 | 0.27 | 0.27 | 0.27 |
| D | A coalition in office since 2018 has cut the public broadcaster's budget twice and in 2020 arrested two investigative journalists under a new security law. Courts still rule against it, and elections are due in 2025. | 0.80 | 0.78 | 0.77 | 0.75 | 0.75 | 0.73 | 0.72 | 0.71 |
Electoral Democracy Index by year. Start on a fall of 0.01 or more. Continue through up to four flat years. End early on a rise of 0.02 or more. Confirm once the total fall reaches 0.10. Treat 0.50 and above as a democracy.
- Code each country with Lührmann and Lindberg's rules. When does the decline start, how far has the index fallen, and is the episode confirmed? Name the kind (recession, breakdown or autocratic consolidation) and the tactic.
- Explain Country B's coup with Acemoglu and Robinson. Which details in its file raised the elite's gain from a coup or lowered its cost?
The files and scores are invented and describe no real country.
Before Thursday 15 October
Week 05. Executive design. Presidents and parliaments.
Linz, Juan J. 1990. "The Perils of Presidentialism." Journal of Democracy 1(1), 51-69.
Cheibub, José Antonio, and Fernando Limongi. 2002. "Democratic Institutions and Regime Survival: Parliamentary and Presidential Democracies Reconsidered." Annual Review of Political Science 5, 151-179.
- No class
- Quiz 2 opens the class. It covers Sessions 03 and 04.
- Week 6 class on electoral systems
- Complement on concepts, measurement and validityM2170
Progress saved
Questions?
Office hours Thursdays 14:30 to 16:00 in M4328
Rate each group member's participation in today's exercise. Required before you leave the room.