Typology of autocracies (Geddes, Wright and Frantz)

Dictatorships differ as much from each other as from democracies. Barbara Geddes (1999) proposed to classify them by the group that controls access to office and influences policy: in single-party (dominant-party) regimes a party does; in military regimes a group of officers does; in personalist regimes power rests with one individual; and Geddes, Joseph Wright and Erica Frantz (2014) later added monarchies and hybrids such as party-personalist or military-personalist regimes.

The categories matter because each type faces different internal politics. Military officers value the unity of the armed forces above holding office, so when factions split they often negotiate a return to the barracks. Party cadres want to stay in office and co-opt rivals, so party regimes tend to survive economic shocks. Personalist rulers close ranks around a narrow circle and often leave power violently. Geddes's 1999 survey ranked single-party regimes as the most resilient, military regimes as the least, and personalist ones in between.

A key move in the 2014 dataset is to separate leaders from regimes. A regime is the set of basic rules identifying who chooses leaders and policies; it can outlive many leaders, and one autocracy can be replaced by another. The dataset covers 280 autocratic regimes in existence from 1946 to 2010 and records how each ended, how violent the transition was and what followed.

• Single-party: the Soviet Union, Mexico under the PRI (1929–2000), China and Vietnam today.
• Military: Argentina 1976–83, Brazil 1964–85, Myanmar's juntas.
• Personalist: Mobutu's Zaire, Saddam Hussein's Iraq, Turkmenistan under Niyazov.
• Monarchy: Saudi Arabia, the other Gulf monarchies, Eswatini.

Classifying real regimes requires judgement, since most combine features, and personalism is better treated as a matter of degree; the authors themselves later built a continuous measure of personalization. Alternative schemes classify differently: Cheibub, Gandhi and Vreeland code dictatorships as monarchic, military or civilian according to the ruler's background, and Kailitz distinguishes communist and other ideocracies by their claim to legitimacy. Results on durability depend on the scheme used: Kailitz and Stockemer find ruling monarchies and communist ideocracies to be the most durable autocratic types.

The typology reshaped the study of authoritarianism by showing that roughly half of all regime changes are from one autocracy to another rather than to democracy, and that regime type predicts how dictatorships end. It is widely used, alongside V-Dem's Regimes of the World, to study repression, conflict, growth and the survival of leaders.