models Presidents Obasanjo in Nigeria and Mbeke in South
Africa were prosecuted. Admittedly, those prosecutions were
probably warranted, but you can still claim to be following their
precedents. If corruption is too sensitive an issue to open, you
can try citizenship. Given the considerable ethnic diversity of
most countries of the bottom billion, and the large migrations
of peoples, it should be easy to trump up some ancestry that
debars them from citizenship. Potentially, you can go the whole
hog, like President Abacha of Nigeria, and debar everyone. Im-
plausible as it might seem, it is still possible to hold a contested
election. Failing all else, someone might assassinate your op-
ponent, as happened in the run-up to the Pakistani elections of
2007, which Benazir Bhutto might otherwise have won.
Cons: Unless you go the whole hog, voters inevitably have some
alternative to your own good self, however awful. They may be
sufficiently foolish to opt for it. You think mournfully of Presi-
dent Gueï of Cote d’Ivoire, whose sad story must wait a little.
So banning key opponents makes sense, but cannot be relied
upon to be sufficient. Worried, you wonder whether there is any
strategy that you have overlooked. And then you heave a long, deep
sigh of relief.
Option 7: Miscount the votes
Pros: At last you have found a strategy that sounds reliable.
With this one you literally cannot lose: incumbent one, oppo-
nent ten million; headline: “Incumbent Wins Narrowly by One
Vote.” It also has advantages in reinforcing the other strategies.
36
WARS, GUNS, AND VOTES
Once people get the sense that you are going to win anyway
and that their true votes will not be counted, they have even less
incentive to forgo the bribes and take the risks of opposition.
You can also keep this one in reserve until you see that you are
losing. In the Kenyan elections of December 2007, as one by one
the parliamentary constituency results were declared, the op-
position looked set to win the presidency. Yet by the time these
constituency votes were added up to the national total by the
electoral commission to determine who should be president, lo
and behold, the incumbent president had narrowly won.
Cons: The international community won’t like it if you push it
too far. Better be a bit careful: after the Kenyan election results
the European Union got upset about discrepancies. In one con-
stituency the vote for the president had unfortunately first been
announced as 50,145 before being entered as 75,261 in the final
tally.
Th i s o n e i s d e f i n i t e ly f o r you. Just remember not to push it too far: not 99 percent; it should not look like a Soviet election.
So much for putting oneself in the position of the president.
What struck me was how much superior, from the point of view of
a self-interested political leader, some of the other options were to
the tough and unreliable option of trying to be a good government.
In the typical election in one of the developed countries, as defined
by membership in the Organization for Economic Cooperation and
Development (OECD), the incumbent government has a chance of
reelection of around 45 percent. In the average election held in a
society of the bottom billion, despite the fact that voters usually have
many more grounds for complaint, it is a much healthier 74 percent.
Political scientists have developed a scale of democratic governance
called Polity IV, starting at –10, which characterizes political hell,
Votes and Violence
37
and going right through to +10, which is political heaven. Among
those countries of the bottom billion in the range –10 to zero, the
president has an even healthier chance of electoral victory: an amaz-
ing 88 percent. Somehow or other, incumbents in these societies re-
ally are very good at winning elections.
I decided that it was time to investigate the winning strategies
more systematically, and for this work I turned to Pedro Vicente,
who already had experience from Cape Verde and São Tomé, two
little islands off the coast of West Africa. I persuaded Pedro that
we should be ambitious: little islands provided neat natural experi-
ments, but we should try working on one of the major new democra-
cies. We chose Nigeria, where elections were due during the course
of 2007. Despite its evident importance as Africa’s largest society,
there is amazingly little quantitative field research on Nigeria. It has
a reputation for being a difficult, and indeed a dangerous environ-
ment, and it is also astonishingly expensive.
All the gossip was that the Nigerian elections would be nasty.
President Obasanjo had set his sights on changing the constitution
so as to have a third term. The vice president, who had aspirations to
the top job, set about blocking this strategy, which needed approval
from the Senate. In a close and bitterly contested Senate vote the vice
president succeeded in blocking the third term. This left President
Obasanjo without an heir apparent of his own choosing: for obvi-
ous reasons he had not wanted there to be any alternative to himself.
Worse, the vice president had entrenched himself as the likely win-
ner, using the vice presidency to benefit from his own powers of in-
cumbency. If there was one person President Obasanjo did not want
to succeed him, it was the vice president. So, with less than twelve
months before the election, he was going to have an uphill struggle
to take someone from zero to victory over the vice president. As the
election campaign approached he told his party it was a “do or die” af-
fair. Everyone understood what “do or die” meant: it meant no-holds-
barred. In turn, this meant “refer to the above list of options.”
38
WARS, GUNS, AND VOTES
On one of my visits to Nigeria I had met Otive Igbuzor, an out-
spoken political activist who impressed me. Although I thought that
some of his views on the economy were wrong, his concerns about
the lack of political accountability were cogent and passionate. He
was also sufficiently open that he did not dismiss me simply because
I was a foreigner. We decided to join forces. I brought a research
group able to conduct a scientific field experiment; he brought a
vocal local NGO that he headed, Action Aid, with a field network
of committed people. Together we designed a field experiment to
measure three of the illegitimate winning options: bribery, intimi-
dation, and vote miscounting. We were also able to join forces with
the team from Michigan State University that runs the Pan-African
Afrobarometer survey of political attitudes. The heart of our experi-
ment was to see whether voter intimidation could be countered. On
a randomized basis across Nigeria, Action Aid organized powerful
local campaigns against intimidation.
Manifestly, a research project aimed at trying to counter po-
litical violence during
a Nigerian election campaign that was an-
ticipated to be particularly nasty was pushing the limits. Quite apart
from the physical dangers for all the participants, Pedro had to di-
vert from the safe strategy of using his time to write up his existing
research for publication into this highly risky undertaking. There
might easily be nothing to show for months of work, and he would
need publications to get another job once the funding for his re-
search post expired. Even I had to find a modicum of courage: reas-
suring research foundations that they were not pouring their money
down a particularly expensive drain. In the event, the elections were
indeed marred by irregularities. The monitors sent by the European
Union described it as “not credible,” and Human Rights Watch de-
scribed it as a “farce.” As I write, five of the governors elected have
been stripped of office by the Nigerian courts. For Nigerians the
election was evidently flawed, but these very flaws made it well-
suited to our research.
Votes and Violence
39
We found clear statistical evidence of all three strategies. The
Action Aid campaign against voter intimidation had a remark-
ably large effect. In those randomly chosen locations in which the
campaign was conducted, more people found the courage to vote.
We interviewed people both before and after the election: where
the campaign was conducted many more people who had initially
decided not to vote changed their minds. What is more, despite this
overall increase in turnout, the vote for those politicians perceived as
espousing violence fell. People who had initially intended to vote for
these candidates changed their minds and stayed at home.
That one campaign by one NGO could have such a big effect
against such an apparently intractable problem is surely remarkable.
But that was not the only surprise. We found that bribery and vote
miscounting went hand in hand: they were complementary strate-
gies. We measured them by asking people how serious they per-
ceived bribery and ballot fraud to have been in their constituency.
We found that ballot rigging favored the local incumbent party.
Evidently, local incumbency is what matters for controlling the vote
count. But the surprise was that voter intimidation was high when
bribery and miscounting were low. It turned out that, at least in the
Nigerian election, violence was predominantly a strategy of the po-
litically weak, perhaps somewhat analogous to terrorism.
S o i n N i g e r i a p o l i t i c i a n s h a d clearly resorted to socially dysfunctional strategies of vote winning. Now think of the implications. With these options available, electoral competition is simply
not going to deliver accountability. Nor, if politicians win by these
unscrupulous means, is democracy going to confer much in the way
of legitimacy. Losing opponents are not going to say, “Fair enough,
you now have a mandate”; they are going to say, “You cheated” and
resort to violence. In other words, democratic elections cannot pos-
sibly, in themselves, be a solution to the problem of violence, or to
40
WARS, GUNS, AND VOTES
the larger problem of decent government. In themselves they are a
recipe for driving political leadership into the gutter. It is not even a
matter of maybe. Electoral competition creates a Darwinian strug-
gle for political survival in which the winner is the one who adopts
the most cost-effective means of attracting votes. In the absence of
restraints the most cost-effective means are simply not going to be
good governance: that option is surely way down the list.
An example from the Nigerian gubernatorial elections stared me
in the face. This had been the campaign for reelection by the incumbent
minister for the federal capital territory of Abuja, Nasir el-Rufai. Con-
trary to most of his colleagues, he had governed well. His ability was
recognized by ordinary Nigerians: in 2006 he had won the prestigious
Silverbird Man-of-the-Year Award. Indeed, by any standards he was
competent. He had managed to get into Harvard Business School, no
mean feat for a young Nigerian, and had duly come to the top in his
year. Also exceptionally, he had decided not to exploit the potential mul-
tiple advantages of incumbency and conducted an honest campaign. He
lost: in fact he didn’t even manage to win the nomination of his own
party in the primary that preceded the gubernatorial election. Given the
potency of the dishonest options, the honest and decent have so much
stacked against them that that is all too often their fate.
S o fa r I h av e c o m e at this from the perspective of how to game an election. The punch line I have been working toward is that in the
typical society of the bottom billion, electoral competition, far from
disciplining a government into good policies, drives it into worse
ones. But even if incumbent politicians resort to mischief when they
come to an election, in the meantime they might also decide to do
their best. In other words, being a good government and all the
other options may not be alternatives they may be complementary: a
scared politician may try them all. It is time to look, and for this we
need to observe not the electoral strategies but the policy choices.
Votes and Violence
41
Undoubtedly, during the period of electoral competition which
began in the early 1990s, economic policies in the countries of the
bottom billion have tended to improve. Is this causal: has democracy
driven governments into better economic policies despite the mis-
chief over how they win elections? It seems a plausible hypothesis.
I had already worked on the preconditions for the reform of poli-
cies and governance with a young French economist, Lisa Chauvet.
The issue of how democracy and elections affected the chances of
reform was a natural extension of this earlier work, and so she was
the obvious person to work with. The only problem was that she
was pregnant. We raced against the arrival of little Diego to get the
results I now report.
Our universe of observations was all the countries that at some
stage or other had been impoverished and had had seriously dys-
functional policies and governance. From this universe, the task is to
try to explain why some countries at some particular time managed
to reform out of the mess, and in particular, to investigate whether
democracy in general, and elections in particular, seemed to help or
hinder the process. The phrase “policies and governance” is easy to
write, and within reason people can agree on what they mean by it.
But it is a difficult concept to measure with any precision. Further,
we needed a measure of policies and governance that was available
on a consistent basis for as many countries as possible, for as long a
period as possible. There are only two possibilities, one put together
by the World Bank, called the Country Policy and Institutional As-
sessment, and the other put together by a commercial company, the
<
br /> International Country Risk Guide. They are both based on judg-
ments of a professional staff, a little like the process by which Stan-
dard and Poor’s assigns credit ratings to country debt. We chose the
World Bank rating, mainly because it started seven years earlier
than the commercial rating agency and so covers a longer period.
We had already found that there were some clear preconditions
that made reform easier. The larger a country’s population was, the
42
WARS, GUNS, AND VOTES
faster it reformed. I think that this is because a large population sup-
ports a market for specialized publications that can discus economic
policy. India has a newspaper, The Economic Times, with a circula-
tion of 1.2 million. It can afford to send its correspondents around
the world. If Zambia had the equivalent with the same density of
circulation, sales would be under ten thousand, and so Zambia has
no Economic Times. It also helped to have donor technical support,
the much-despised form of aid whereby skilled foreigners are sent
to help governments. But now we looked at elections and democ-
racy.
One problem with elections is that they are often not held ac-
cording to a set calendar but occur due to circumstances that might
themselves affect the chances of reform. Let me give you a simple
example of how an unsuspecting researcher could get into trouble.
Suppose that what is really going on is that periodically the peo-
ple pressing for change in their society manage to break through
politically. They believe in economic reform and they also believe
in democracy. So they hold an election and they also reform the
economy. If the researcher is not careful, this is going to look as if
elections cause reform. So how can the researcher be careful? The
answer is to find something that is a reasonable predictor of when
the next election will be held but that does not itself influence cur-
rent prospects of reform. The best we could think of for this was to
predict the timing of the election on the basis of the time lapse be-
tween the two preceding elections. The idea is that in many societies
there is a fairly fixed frequency of elections. Indeed in some it is set
in concrete, as in America. Repeating the analysis just using these
countries where the government cannot choose the election date is a
Paul Collier - Wars, Guns, and Votes Page 5
