VoxEU Column Economic history Politics and economics

Democracy by mistake

Most research on the transition to democracy tries to explain why autocrats choose to democratise. Based on two centuries of data on democratisation, this column argues, however, that autocratic rulers overwhelmingly create democracies by mistake. Taking these mistakes into account during analysis may improve the predictive or explanatory power of existing models.

The introduction of democracy in a country means a loss of power for authoritarian rulers. Why would they agree to it?

Among the arguments are that leaders might embrace democracy as a way to forestall revolution (Acemoglu and Robinson 2006), to motivate citizens to fight a war (Ticchi and Vindigni 2008), to nudge future governments away from patronage (Lizzeri and Persico 2004), to outmanoeuvre political rivals (Collier 1999), or to forge a pact between deadlocked social groups (Rustow 1970).

All these arguments assume that incumbents democratise intentionally. In Dankwart Rustow’s words, liberalisation results from a “deliberate decision on the part of political leaders” (Rustow 1970). For one reason or another, authoritarian elites choose to give up control.

Rulers did not choose democracy

Examining all democratisations between 1800 and 2015, I found evidence for a different view (Treisman 2017). While some rulers – such as King Frederick VII of Denmark in 1848 – deliberately limited their power, most did not. Democracy did not emerge because incumbent elites chose it but because, in seeking to prevent it, they made critical mistakes.

In the history of transitions, there is more chaos and miscalculation than rational planning. Lord Derby memorably described Disraeli’s 1867 franchise enlargement as “a leap in the dark.” Historians have used the same phrase to describe Bismarck’s reforms a few years later, as well as those of Franco’s heirs in Spain. Many a ruler facing pressures for change has been – as Tocqueville described France’s King Louis Philippe in 1848 – “knocked flat before he had understood”.

I examined each of the 201 cases that fit one of three common definitions of democratisation, studying history books and articles, newspapers, magazines, web publications, biographies, memoirs, diaries, and published interviews – 1,064 sources in all. For each episode, I recorded whether it could plausibly fit any of the 'deliberate choice' arguments above. I also assessed whether democratisation had occurred because of one or more mistakes by the incumbent rulers.

The results were striking. Depending on the definition of democracy, each of the deliberate choice arguments could apply to between 4% (nudging governments to reduce patronage) and 16-19% (democratisation by pact) of cases. At least one may have contributed to between 28% and 33% of cases. However, in between 64% and 67% of cases, rulers appear not to have intended to share power – they democratised by mistake.

By 'mistake', I mean a choice for which the expected payoff is lower than that of some other feasible option (with expectations taken using objective probabilities). In other words, it is a course of action or inaction that would lie off the equilibrium path in a perfect information game. I distinguish two types of mistake:

  • Mistakes of information may be rational given the actor’s beliefs and yet suboptimal because some of those beliefs are wrong.
  • Miscalculations occur when, despite accurate information, the actor selects a path with a lower expected payoff than some other feasible option.

Not all actions with undesired outcomes are mistakes. One may lose a gamble that was, nevertheless, optimal ex ante. Or, faced with only unattractive options, one may choose the 'lesser of two evils'. Conversely, an action may be a mistake even if all alternatives were bad. All that is required is that some feasible course had a higher expected payoff. So, to be clear, I did not assume leaders always prioritised staying in power. If they stepped down to prevent bloodshed, that in itself would not have been a mistake. However, if, through wiser action earlier, they could have avoided having to step down to prevent bloodshed, their prior failure would fit my definition.

Among dictators’ errors, several patterns predominate. First, many succumb to hubris. In between 13% and 17% of cases, incumbents underestimated the strength of opposition, failed to compromise or repress until too late, and were overthrown, leaving it to their successors to liberalise. France’s Louis Philippe, for instance, refused even minor concessions, and turned a series of reform banquets into a revolution. A collection of left-wingers declared the republic after he fled to England.

Other rulers overestimated their popularity, called an election or referendum, failed to manipulate sufficiently, and lost, splitting the elite and empowering opponents (this occurred in between 24% and 29% of cases). In 1988, learning that he had lost a plebiscite on his regime’s future, General Pinochet of Chile roared: “It’s a big lie, a big lie… Here there are only traitors and liars!” (Constable and Valenzuela 1991).

Other victims of electoral overconfidence include Daniel Ortega in Nicaragua in 1990 and the Communists in Poland in 1989: “‘Never in our darkest nightmares did anyone predict such a shameful rout,’” one high party official spluttered, gulping whiskeys and shaking his head in astonishment” (Meyer 2009).

Another mistake – think Leopoldo Galtieri in Argentina – was to start a military conflict only to lose both the battle and political power (6% to 9% of cases). A fourth (30% to 34%) was to slide – like Mikhail Gorbachev after 1985 – down the slippery slope, introducing partial reforms to strengthen the regime, inadvertently undermining it.

In each of these scenarios, the ruler did not mean to share or surrender power, but his misstep destroyed the status quo. In another category of mistakes (between 7% and 10% of cases), it was not the top official but other members of the ruling group who blundered. They chose a leader – like Juan Carlos or Adolfo Suárez in post-Franco Spain – whom they trusted to preserve the system, but who destroyed it. Finally, in 12% to 15% of cases, excessively violent repression inflames opposition or splits the elite, weakening the dictatorship.

Not all errors are equal

Of course, not all errors prompt democratisation – and those that do so contribute directly only to the fall of an incumbent dictator or authoritarian regime. Democracy then emerges only if structural and other conditions are conducive to it. The parameters of the study – examining only episodes of democratisation – did not permit general claims about what caused such transitions (although I suggested causal interpretations of particular cases). The data did reveal, however, that democratisations followed popular mobilisation in between 85% and 87% of episodes, economic crises in 65% to 67%, and foreign pressures to democratise in 62%.

Deliberate choice better explains democratisation in the first wave that took place before 1927 – but still only one argument (reform to outmanoeuvre an elite rival) would fit in more than a quarter of cases. (Mistakes, by contrast, played a role in up to three-quarters.) The results are similar for democratisations that were not subsequently reversed, or those for which sources were relatively plentiful and detailed.

Why did so many autocrats miscalculate?

Familiar cognitive biases probably contribute – from over-optimism and overconfidence to the illusion of control (Langer and Roth 1975) and the ostrich effect, a tendency to avoid exposure to bad news (Karlsson, Loewenstein and Seppi 2009). While such biases affect all humans, dictators have some more distinctive problems. Deliberately or not, many isolate themselves in echo chambers, excluding discordant views. Among Nicaragua’s Sandinistas, “bad-news bearers were stigmatised as Contra allies or weaklings. The lower-level Party cadres and militants… filtered out the negative from their reports” (Guillermoprieto 1990). Zambia’s Kenneth Kaunda “surrounded himself with conmen” (Dowden 1991). Venezuela’s Pérez Jiménez “depended for advice on sycophants and third-rate generals” (Burggraaff 1972).

Shielded from criticism, some autocrats – Pinochet, Ferdinand Marcos, Than Shwe, Idi Amin, Jean-Bédel Bokassa, and Francisco Macías Nguema, to name a few – turned to fortune tellers. Even those with more scientific approaches have had trouble forecasting political events. Non-responses and preference falsification reduce the validity of polls. If media restrictions were loosened and opposition permitted, attitudes could change dramatically. The apparently solid support when a ruler calls a semi-free election can metamorphose into an opposition landslide.

From this perspective, it is not surprising dictators made mistakes. In fact, what’s odd is that they didn’t make more. Intuitive judgements tend to be more accurate when the setting is rule-based and offers opportunities to practice (Kahneman, 2011). Few environments are less regular and conducive to practice than regime transitions, which are rare events that involve multiple actors in complex configurations. In other contexts of institutional change – electoral system choice, the negotiation of international human rights treaties – outcomes have often surprised the participants. The year before his disastrous plebiscite, Pinochet signed the UN Convention Against Torture. He clearly did not expect a Spanish judge to use it 11 years later against him.

Although mistakes do not lend themselves to formal modelling and unifying theory, they are inescapable in the real world. Incorporating them into analysis, as behavioural economics has done in other areas, could improve predictive power. The is particularly true if particular types of mistakes have distinctive triggers. At the same time, such analysis, combining rational and irrational elements, would yield a richer and more accurate interpretation of history than stylised simplifications with imperfect purchase on the facts.


Acemoglu, D and J Robinson (2006), Economic Origins of Dictatorship and Democracy, New York: Cambridge University Press.

Burggraaff, W J (1972), The Venezuelan Armed Forces in Politics, 1935-1959, Columbia, MO: University of Missouri Press.

Collier, R B (1999), Paths Toward Democracy: The Working Class and Elites in Western Europe and South America, New York: Cambridge University Press.

Constable, P and A Valenzuela (1991), A Nation of Enemies: Chile Under Pinochet, New York: Norton.

Dowden, R (1991) "Profile: The Man Who Was Zambia; Through 27 Years, Kenneth Kaunda Spoke for Black Africa. Last Week's Vote May Mean his Africa Has Had its Day." The Independent, 3 November.

Guillermoprieto, A (1990), "Letter from Managua", The New Yorker, 26 March.

Kahneman, D (2011), Thinking, Fast and Slow, New York: Macmillan.

Karlsson, N, G Loewenstein and D Seppi (2009), "The Ostrich Effect: Selective Attention to Information", Journal of Risk and Uncertainty 38(2): 95-115.

Langer, E J and J Roth (1975), "Heads I Win, Tails it's Chance: Illusion of Control as a Function of Sequence of Outcomes in a Purely Chance Task", Journal of Personality and Social Psychology 32: 951–955.

Lizzeri, A and N Persico (2004), "Why Did the Elites Extend the Suffrage? Democracy and the Scope of Government, with an Application to Britain's 'Age of Reform'", Quarterly Journal of Economics 119(2): 707-765.

Llavador, H and R J Oxoby (2005), "Partisan Competition, Growth, and the Franchise", Quarterly Journal of Economics 120(3): 1155-1189.

Meyer, M (2009), The Year That Changed the World: The Untold Story Behind the Fall of the Berlin Wall, New York: Scribner.

Rustow, D A (1970) "Transitions to Democracy: Toward a Dynamic Model." Comparative Politics 2(3): 337-363.

Ticchi, D and A Vindigni (2008), "War and Endogenous Democracy", IZA working paper No. 3397.

Treisman, D (2017), "Democracy by Mistake", NBER working paper No. 23944.

8,609 Reads