lence is enormous. Even were it to lead to healthy political change,
we would need to ask whether the eventual benefits were worth
these massive costs. But the final tragedy of civil war is that it does
not usually lead to any such political legacy. If we take as our mea-
sure of the polity the Polity IV index, civil war leads not to improve-
ment but to deterioration. Instead, as we have seen, the most likely
legacy of a civil war is a further civil war.
I f t h e f e as i b i l i t y h y p ot h e s i s i s right it has a powerful implication: violent conflict cannot be prevented by addressing the
problems that are likely to motivate it; it can only be prevented by
making it more difficult. Whether rebellion is easy or difficult basi-
cally comes down to whether rebels have access to guns and money,
and whether the state is effective in opposing them. Most of the guns
and money that finance rebellion come from outside the societies
that are plagued by civil war. The effectiveness of a state increases
with its level of development. This gives the international commu-
nity some scope to reduce the incidence of war. It can squeeze rebel
organizations by curtailing the guns and money that presently reach
them so readily, and it can try to break the impasse that has frus-
trated development.
Should the international community try to discourage rebel-
lion? When that iconic poster of Che Guevara first came out, I was a
student. For my generation, support for armed struggle in develop-
140
WARS, GUNS, AND VOTES
ing countries was a natural extension of our support for liberation
movements. But liberation from colonialism and rebellion are not
the same thing: one unites the society against an external oppressor;
the other divides it against itself. Painful as it is to revise cherished
old beliefs, armed struggle is usually development in reverse.
C h a p t e r 6
C O U P S : T H E U N G U I D E D
M I S S I L E
The coup d’état as a technology of political violence
will play a central role in this book. The ghoulish glam-
our widely associated with political violence has focused
almost exclusively upon rebellion—armed struggle, as it is called by
its aficionados. Rebellions should be turned into history as rapidly
as possible because the consequences of civil war are so dire. But
coups are a different matter. The challenge posed by coups is not to
eliminate them but to harness them. Coups have the ready potential
to deliver what armed struggle was supposed to achieve but seldom
did. But I am not about to give you a eulogy for coups: to date, they
have usually been awful. It is time to look.
Suppose that you are the president of a country that is one of
the bottom billion. Although life for ordinary citizens is tough, for-
tunately a grateful nation has made your own life remarkably com-
fortable. In the developed countries presidents have to wait until
they step down and write a best-selling memoir before they come
into money. Even in the more successful developing countries politi-
cal power often does not lead to wealth: former president Muhatir
of Malaysia—one of the most successfully transformed countries
on the planet—is not a rich man after many years in power. When
142
WARS, GUNS, AND VOTES
President Quett Masire of Botswana, Africa’s most successful econ-
omy, stepped down, he feared that he might be declared bankrupt.
But in the societies of the bottom billion, political leaders have a long
tradition of accumulating wealth while in office. Far from offering
the prospects of book tours, losing power looks decidedly scary. So
evident is the fear of losing office that a public-spirited African busi-
nessman, Mo Ibrahim, has now introduced a $5 million prize for
African presidents who voluntarily step down. Quite possibly, over
time this will indeed change behavior.
Recall that most presidents have learned how to live with the
menace of elections. The prospect of facing the electorate once every
few years is not what brings presidents out in a cold sweat in the
early hours of the morning. What brings on the fear is, ironically,
the system that is supposed to keep the country secure: presidents
fear a coup d’état from their own army. Since independence, suc-
cessful African coups have been running at a rate of around two a
year. Unlike an election, it could happen at any time of any day or
night. If a coup succeeds, sometimes the president can get out in
time, but not necessarily. The successful coup leaders who toppled
President Doe in Liberia not only tortured him to death but made a
video of it. So presidents are right to be scared. This is the analysis
in which some of them will be most interested. It is based on work
with Benedikt Goderis and Anke Hoeffler.
While the prospect of a coup obviously matters to presidents,
it is not clear that it should matter to the rest of us. If the only re-
gimes that are threatened are themselves dictatorships, a coup is
not something to feel outraged about: perhaps it is the only way of
disciplining dictators. Of course, it would be a different matter if
democracies were also threatened by coups. Even if coups replace
dictators, as an economist my instinct is to say, “What does it cost?”
It is obvious that a civil war is immensely costly, both in the narrow
terms of income forgone and in the deeper terms of mortality and
the breakdown of social cohesion. But that is because civil wars are
Coups
143
prolonged, destructive, and usually indecisive. A coup is a surgical
strike: perhaps it is a cheap and effective way of ousting bad re-
gimes. Rather than speculate, I decided to investigate. Since at this
point presidents will be tempted to throw the book down in disgust,
let me reassure them that soon enough I will get to something of
more evident importance: what makes a coup less likely. Impatient
presidents can skip to the next section.
To get a sense of the cost of a coup, a useful starting point is
its impact on the growth of the economy. We found a clear and
straightforward effect of a coup: in the year of the shock it reduced
income by around 3.5 percent, but after a couple of years the econ-
omy reverted to normal. So, taking the aftereffects into account, this
particular cost was around 7 percent of a year’s income. We realized
that this cost might be merely the tip of an iceberg. Economists have
found that political instability is bad for the economy, and coups
seemed likely to be an important form of instability. The main costs
of coups might not be the brief aftereffects of successful coups, but
the consequences of the continuous fear of them. In countries at
high risk of a coup d’état, investors may keep away. To investigate
whether the risk of a coup is damaging, you first need to estimate
the risk. In the process you discover what makes a country prone to
a coup, which I will describe shortly.
We introduced this risk into
our analysis of growth, trying to see whether it reduced economic
activity over and above the actual coups themselves. We could not
find an effect, and while that does not mean it isn’t there, it probably
means that any effect is fairly modest.
A price of 7 percent of a year’s income is not a cheap way of re-
placing a government, but if the government is truly terrible and it
is replaced by a better one, then it is probably a bargain. Would the
Iraqis have paid 7 percent of a year’s income to oust Saddam Hus-
sein, thereby avoiding a war? Would the Zimbabweans have paid it
to oust President Mugabe, thereby avoiding an economic meltdown
and mass emigration? This is an important difference between a
144
WARS, GUNS, AND VOTES
coup and a rebellion. A rebellion with its consequent civil war in-
flicts such high costs on the society that in my view there should be
a strong presumption that rebellion is undesirable. Armed struggle
may be romantic, but it is usually a menace. It is sometimes argued
that if governments and rebels are equally bad, as I think is often
the case, then the international community should stay neutral. I
strongly disagree with this view. Unless the rebels are unquestion-
ably a whole lot better than the government, then the cost inflicted
on the society for the one-in-five chance that the rebellion will lead
to the government being overthrown is far too high, and so the re-
bellion should be discouraged. Neutrality is inappropriate when the
issue is war or peace. But coups are a different matter: they have to
be judged predominantly by whether they improve governance.
It is easy to come up with reasons that a coup might improve
matters, and sometimes they clearly do so. The mere threat of a
coup may act as a restraint upon the government. For example, one
of the few instances in which an election in one of the societies of the
bottom billion resulted in the defeat of the incumbent and a change
of president was in Senegal in 2000. The incumbent was persuaded
to accept defeat because the army told him that if he did not, they
would mount a coup. In turn, the Senegalese military had been em-
boldened by the successful coup d’état in Cote d’Ivoire a few months
earlier. So, in that instance, one country’s coup was another coun-
try’s safeguard of the democratic process.
Not only might the threat of a coup discipline a government,
but in extremis, a coup might be the only way of replacing a dysfunctional leader. Colonel Ely Ould Mohamed Vall led a surgical coup
in Mauritania in 2005, promising clean elections in which he would
not be a candidate, and duly kept his promise. The elections, prop-
erly conducted, ushered in what currently looks to be an excellent
government. But unfortunately, even good coups that replace ter-
rible rulers can end up further degrading the polity. Emperor Hailie
Selassie of Ethiopia built a regime in which power was entirely con-
Coups
145
centrated in his own person. By 1974 he was a senile octogenarian
ruling with disastrous incompetence over the poorest country in the
world. Visiting him, the emperor’s retired adviser John Spencer was
so shocked that he predicted a coup within six weeks. In fact it hap-
pened the next day. So far, so good for everyone except the emperor:
the coup replaced a senile emperor with a respected general, Aman
Andom. But this was not the end of the story. Easy as it would have
been to improve on the performance of a senile emperor, the coup
ended up producing an even greater catastrophe: Colonel Mengistu
Haile Mariam. The general who led the first coup was replaced in a
further coup, and the new leader marched the country into a disas-
trous war, in the process creating one of the world’s most repressive
and economically ruinous regimes.
Even worse, coups might not be provoked by bad governance
but by the opportunistic greed of the army. No sooner had dem-
ocratic São Tomé discovered oil than the army attempted a coup.
The nighttime coup that ousted President Sir Dawda Jawara in the
Gambia originated when a group of drunken soldiers decided to go
to the Presidency building to demand higher pay and found it un-
defended. The Thai coup of 2006 deposed a democratically elected
government that duly got reelected once citizens were given the
chance to vote. So, for the moment, we will have to park the ques-
tion of whether coups are surgical strikes against bad governments
that are cheap at the price, or a menace posed by greedy gunsling-
ers: either might be the norm. I want to get back to what a worried
president might want to read.
P r e s i d e n t s , r e s u m e r e a d i n g h e r e : I am going to investigate what determines a coup. As usual, my approach is to gather
as much data as possible on coups and then try to explain their oc-
currence statistically. There is a standard international data source
on successful coups d’états around the world. This was promising,
146
WARS, GUNS, AND VOTES
but then I hit upon a new data set that had been put together by
Patrick McGowan, a political scientist in Arizona. His innovation
was to have recorded not only the successful coups but the failed
attempts, and even those that had never got as far as an attempt but
had been nipped in the bud at the stage of a plot. His data were only
for Africa, but this still added up to a large number of failed plots,
failed attempts, and successful coups. We reasoned that all attempts,
whether successful or not, had at some stage been plotted. This gave
us an amazing 336 coup plots, of which 191 made it through to the
stage of an attempt, and 82 made it all the way to a successful coup.
Our task now was to explain what determined each of these stages:
why plot; what got a plot through to an attempt; and what made an
attempt succeed?
Since the true purpose of this section is not to help worried pres-
idents retain power, but to discuss how to curtail this form of politi-
cal violence, I will start with the key issue. Does democracy makes
coups less likely? Controlling for other influences, unfortunately it
does not: coups are at least as likely to break out in democracies as
in autocracies. I say “at least” because severe repression significantly
reduces the risk of a coup. So precisely when a coup would be most
justified, it is least likely to occur. We checked to see why repression
made governments safer and found that it enhanced the ability of
governments to detect plots. Repressive regimes did not face more
plots than other sorts of regimes, but the plots were more likely to be
aborted before they could reach the stage of an attempt. Behind that
bland statement is the grim reality of repression: torture, fear, agents
provocateurs, and spies. They work, which is presumably why they
are used so enthusiastically by dictators around the world. It is back
to Herodotus and the preemptive weeding of possible opponent
s.
The worried president sets down Wars, Guns, and Votes for a mo-
ment and jots down a memo on the pad beside his bed: increase the
budget for military intelligence.
A second rather disturbing feature of coups is that one leads to
Coups
147
another, just as it did in Ethiopia. The baseline risk for a coup at-
tempt in Africa is 4 percent per year. Following a coup attempt the
risk of further attempts is greatly elevated: a year after an attempted
coup there is a 10 percent risk of a further coup. Evidently, the same
arguments that General Andom used to justify his seizure of gov-
ernment by force could be used by Mengistu to justify seizing the
government from Andom. The very act of usurping power destroys
the defense of legitimacy. Perhaps more potent than the lack of le-
gitimacy is that a coup sets an example. General Andom inadver-
tently demonstrated to younger army officers that they could trans-
form their lives from the squalid setting of a barracks to the luxury
of the presidential palace by one audacious act. Although General
Andom was probably motivated by the best interests of his country,
Mengistu was probably motivated by interests rather less lofty: soon
enough he was being driven around Addis Ababa in a red Cadillac.
What is more, if the coup removes the people currently at the top,
everyone else has a chance of moving up, so it is relatively easy for
new coup leaders to gain support among their colleagues.
Taken together with the previous result, coups are less likely
to throw out truly bad governments than to throw out better ones.
They are also likely to lead to further coups, each of which incurs
costs. Coups are beginning to look less attractive than we might
have hoped.
Do ethnic divisions matter? This is one of the relatively few
respects in which Africa is distinctive. Usually I find that Africa
conforms closely to global patterns of behavior: the outcomes are
distinctive only because the characteristics that globally drive
behavior are distinctive in Africa. But this is not so in respect of
ethnicity and coups. Globally we can find no effect of ethnic po-
larization or diversity on the risk of a coup. But in Africa ethnic
polarization sharply increases risk.
What else determines the risk of a coup? Well, the economy is
Paul Collier - Wars, Guns, and Votes Page 18