When Egyptian troops overthrew President Mohamed Morsi on July 3, 2013, a number of observers proposed that the coup might be just what Egyptian democracy needed. After all, many Egyptians supported the military’s actions, the Islamist elected government had shown little respect for minority rights, and the military-led interim government announced a clear timetable for a return to democratic rule just a few days after seizing power. Despite initial optimism, less than a year later, the military’s own Gen. Abdel Fatah al-Sissi won 97 percent of the vote, in a race that none of the major opposition parties contested.

Though the Egyptian coup did not usher in democracy, “good coups” — or those against dictatorships that lead to democratization — appear to have dramatically increased in number since the end of the Cold War, at least partially because of the incentives created by international pressures for democratization. Examples include coups in Mali in 1991, Guinea Bissau in 2003, and Niger in 1999 and 2010.

This trend has generated arguments that coups — traditionally seen as a sign of democratic breakdown — may actually be a tool to usher in democracy. By creating a shock to the political system, the argument goes, coups can generate opportunities for political liberalization that would otherwise be absent. As Paul Collier wrote in 2009 for the New Humanist, “coups and the threat of coups can be a significant weapon in fostering democracy.”

Can coups really foster democracy? In a recent study, we weigh in on this question. We look at the political systems that follow coups against autocrats, as well as the ensuing levels of repression.

We emphasize that though the most basic goal of coups is to bring about changes in leadership, coup plotters often seek more-significant political change. Successful coups against autocrats can therefore lead to three distinct outcomes: no regime change (such as when the Nigerian military replaced Gen. Yakubu Gowon with Brigadier Murtala Muhammad in 1975, without changing the identity of the group in power or the rules for governing), ouster of the incumbent dictatorship and establishment of a new one (such as when Gen. Idi Amin toppled Milton Obote’s dictatorship in Uganda in 1971), and ouster of the dictatorship followed by democratization (such as the two “good coups” in Niger in 1999 and 2010).

We find that since the end of the Cold War, regime change of some sort increasingly follows successful coups (68 percent pre-1990 compared with 90 percent afterward, with the rest simply reshuffling the leadership). Though more of these changes now end in democratization, the most common outcome is still the replacement of one dictatorship by a different group of autocrats. As Figure 1 shows, about half of all coups — 56 percent during the Cold War and 50 percent since the end of it — install new autocratic regimes. On the contrary, only 12 percent of coups during the Cold War installed democracies; that increased to 40 percent post-Cold War. Finally, 32 percent of Cold War coups and 10 percent of post-Cold War coups merely reshuffled the regime’s leadership (no regime change). In short, more often than not, coups in dictatorships simply install new dictatorships.

Figure 1: Outcomes of successful coups, pre-vs-post Cold War

A bevy of statistical tests that take into account a host of potentially confounding factors unearth a similar message: Coups increase the chance of a new dictatorship but do not exert a noticeable effect on the chance of democratization.

The same is true when we incorporate failed coups in our analysis. Though some have argued that coup attempts — whether successful or not — can create opportunities for democratization, our study indicates that this, too, is an unlikely outcome (see Figure 2).

Figure 2: Outcomes of coup attempts, pre-vs-post Cold War

“Good coups” may grab our attention, but the data indicate that they are not the norm. For example, though Nigerien coups in 1999 and 2010 imposed democracy, coups in 1974 and 1996 led to the establishment of new dictatorships.

The bad news does not end there. Using annual data on repression, we find that coups that launch new dictatorships lead to higher levels of repression in the year that follows than existed in the year leading to the coup. Moreover, in daily event data for the 49 coup attempts that have occurred since 1989, we find that there is only one case of a coup followed by a drop in state-caused civilian deaths during the subsequent 12 months.

Figure 3 summarizes our analysis of the 49 coup attempts. The dark lines in the boxes display the median change in state-sanctioned deaths in the 12 months after the coup, versus the 12 months before the coup. The width of the boxes reflects the spread in the distributions. Though we cannot be statistically confident that repression increases after coups — even for reshuffling coups — two things should be noted. First, there is only one case of a (failed) coup followed by a drop in deadly repression. Second, post-coup increases in state violence are common.

Figure 3: Coups and state-sanctioned civilian deaths

The experience of Guinea illustrates the typical pattern. After the death of longtime dictator Lansana Conté, Cpt. Dadis Camara staged a coup on Dec. 23, 2008. Citizens initially welcomed the coup as a chance for greater freedom, but the new government began a campaign of repression soon after. State violence peaked in September 2009, when security forces killed scores of citizens participating in anti-government protests. Rather than opening the door for democracy, the 2008 coup instead brought a new dictatorship to power and plenty of bloodshed.

Though democracies are occasionally established in the wake of coups, our research indicates that more often coups initiate new dictatorships and more human rights violations.

Joseph Wright is an associate professor in the political science department at Pennsylvania State University and a visiting international affiliate in the political studies department at the University of Cape Town.

Barbara Geddes is a professor in political science at UCLA.

Erica Frantz is an assistant professor in political science at Michigan State University.

George Derpanopoulos is a graduate student in political science at UCLA.