simple way of checking the reliability of the results.
We then asked how the amount of time left until the next elec-
tion affected whether policy and governance improve or worsen.
We found a clear and unambiguous relationship. For the first few
years after an election the chances of policy improvement got better,
Votes and Violence
43
year by year. And then, as the next election approached, the chances
of reform started to get worse again, year by year. Two years before
the election the chances of reform slipped, and the year before the
election reform was highly unlikely. What the results were telling us
was that the chances of reform were at their peak when the society
was as far away from an election—in either direction—as possible.
Why might this be? Perhaps, in the first year or so after an election,
the government was too new to be able to implement reform, and
as the election approached it was too preoccupied by the need to
win the election to bother with reform. After all, the payoff to most
reforms takes several years, and any payoff that does not arrive until
after the election has little political benefit.
This was not really encouraging: it suggested that elections
were to an extent a distraction rather than a stimulus. I recalled my
friend Ngozi Nkonjo-Iweala telling me when she became Nigeria’s
finance minister that although the government was at the start of
a four-year term, she had been given only three years for reform.
“The last year will be politics,” the president had explained to her,
and, as I have just described, so it had proved. However, it might
nevertheless be that all the election effect showed was a variant of
the political business cycle. The political business cycle was the game
that rich-country politicians used to play with their own electorates,
pumping money into the economy just before an election: whoever
won would then have to clean up the mess in the next couple of
years. Damaging as the political business cycle was, it did not mean
that democracy was worse than autocracy. It just showed that it
wasn’t perfect. So the election results in themselves did not say any-
thing about whether, if your society needed reform, democracy was
better or worse than autocracy.
To investigate this deeper question, Lisa and I then introduced
measures of the polity. How democratic was it? Was government
power limited by checks and balances? In particular, were elections
well conducted? Fortunately, all these characteristics are now classi-
44
WARS, GUNS, AND VOTES
fied and coded by political scientists. For example, we used a standard
measure known as Polity IV, which you have already come across.
This arrays the degree of democracy on a scale from 0 to 10, alongside
a scale for autocracy, which is ranged from –10 to 0. So the people’s
paradise of North Korea scores –10, the squeaky clean democracies of
Norway and Switzerland score +10, and a messy electoral competi-
tion typical of the bottom billion would score at best around +2 or +3.
Prior to the wave of democratization the societies of the bottom bil-
lion had on average been around –6: in other words, they were mostly
autocracies. Currently, the average score is around zero. When we
added these characteristics they mattered: the electoral cycle overlaid,
and potentially confused, these deeper effects. Elections can poten-
tially spur a government to adopt reforms, but they can also drag it
further down the road to bad governance. Which effect predominates
depends partly upon structural features of the society, and partly upon
the design of the polity. Elections tend to work better in societies that
have larger populations and fewer ethnic divisions. They also tend
to work better in polities with checks and balances on the power of
government, and in particular where the elections are properly con-
ducted. On the evidence, elections without properly enforced rules of
conduct in small, ethnically divided societies typically retard reform
rather than accelerate it.
So the implication is that to date the process of democratiza-
tion in the bottom billion has remained within the range over which
better is worse: the increased democracy has quite probably re-
tarded the reform of economic policies and governance. It has gone
far enough to lose whatever might be the advantages of autocracy,
while not yet having gone far enough to gain the benefits of democ-
racy, and the typical society of the bottom billion remains well short
of the point at which democratization would lead to improvement.
It has proved much easier to introduce elections than checks and
balances. Presidents quite enjoy being anointed by the holy oil of an
electoral victory, whereas they find the prospect of effective checks
Votes and Violence
45
and balances truly alarming. But above all, they have woken up to
implications of the lack of checks and balances for their ability to
survive elections.
Taken together, the results on elections and democratization are
consistent: if democracy means little more than elections, it is dam-
aging to the reform process. I do not like these results. It would be a
much happier story if at every step along the way to fully fledged de-
mocracy the consequences got better and better. But unfortunately
this does not seem to be the world as it is.
The results on the dysfunctional consequences of partial de-
mocracy for reform are also consistent with the evidence on how
elections are actually won in the societies of the bottom billion. The
six nefarious options for winning an election not only dominate the
option of trying to be a good government, collectively they consti-
tute an alternative. So why don’t more governments hedge their bets
and do both: win by foul means, but improve their chances that ex-
tra bit by also trying to be a good government? I think it is because
the other options depend upon bad governance. If you want to use them you have to sacrifice the strategy of being a good government even
if you recognize that otherwise it might be worth doing.
One reason for the conflict between decent governance and the
other options is money. When President Obasanjo realized that he
would not be able to stand for a third term, he knew that he was in
for a very tough contest. How do you win a Nigerian election for an
unknown candidate in only a few months, facing an entrenched op-
ponent? The answer is you probably need a lot of money. Yet over the
previous three years President Obasanjo had started to put in place
the rudiments of accountable government finances. He had entrusted
the ministry of finance to Ngozi Nkonjo-Iweala, and public procure-
ment to Oby Ezekwesili. These two tough, able, Christian women
had shut down the sources for the sort of slush money that a political
campaign was likely to need. Within a month of the Senate decision
to deny President Obasanjo a third term, he had shifted bo
th of them
46
WARS, GUNS, AND VOTES
well away from control of government money. The one high-profile
fighter against corruption who was not shifted was Nuhu Ribadu,
head of the Economic and Financial Crimes Commission. Bravely,
in late 2007, he launched a prosecution against James Ibori, the key
financial backer of President Obasanjo’s chosen successor. Ribadu
lasted only a further three months before being ousted.
The more effective strategies are also incompatible with the
rule of law. When President Mugabe discovered that he had lost
the referendum on removing the term limit, and so realized that he
would lose the next election, he set about the process of dismantling
the rule of law, starting by forcing the chief justice into early retire-
ment and appointing a placeman. As the rule of law was gradually
lifted, new options opened up for snatching revenues at the expense
of the economy and President Mugabe duly took them: property
rights were ignored, and finally he resorted to hyperinflation. In
other words, the government needs to remove checks and balances
in order to use the other electoral options, and with checks and bal-
ances removed, other policies are very likely to deteriorate.
This is, unfortunately, consistent with some new work by Mas-
ayuki Kudamatsu that carefully tries to investigate whether the
introduction of elections in Africa has led to a reduction in infant
mortality. Reducing infant mortality is surely about the most basic
concern of ordinary citizens, and across the bottom billion, infant
mortality has been avoidably high. An election should surely em-
power citizens to force governments to reduce the risk that their
young children will die. He concludes that only following those rare
elections in which the incumbent president was defeated did infant
mortality fall. In the more normal situation of incumbent power,
elections achieved nothing.
So both the evidence on how elections are actually won, and the
actual policy performance of democratic governments in the bot-
tom billion, point to the same conclusion: in the conditions of the
bottom billion, electoral competition is not producing accountable
Votes and Violence
47
government. I started by noting that there had been a considerable
improvement in economic policies and governance across the bot-
tom billion coincident with the spread of electoral competition. So,
if elections have not caused the improvement, what has?
I think there are two likely explanations. The simpler, and
therefore probably the better, is that societies have learned from past
mistakes. Experience is a hard school, but all the societies of the bot-
tom billion have been through it. The high-income world has obvi-
ously learned from its mistakes: the inflation of the 1970s is a thing
of the past because electorates in the high-income societies will no
longer put up with it and governments have learned how to tame it.
Quite probably the same process has been going on in Africa. With
the exception of Zimbabwe, inflation rates are now far lower than
they used to be. Whether or not electorates in the bottom billion
have much influence on their governments, elites may well have
woken up to the fact that inflation and some other dysfunctional
economic policies are not worthwhile.
The other possible explanation is that donor conditionality has
imposed discipline on governments, forcing them into reform even if
they did not want to do it. I do not entirely discount this explanation.
It is very difficult to sort out the motivations behind actions. To an
extent donor conditionality might have forced reform. But the statis-
tical evidence if anything suggests that it has delayed reform rather
than accelerated it. Governments do not like being made to do some-
thing against their will and they are remarkably ingenious at finding
ways of not doing it. Donors are also amazingly bad at enforcing their
agreements with governments. So my own judgment is that donor
conditionality on economic policies is not the explanation for policy
improvement. I would put my money on learning from failure.
I r e a l i z e d t h at i f t h i s critique of electoral competition was right it had huge implications. The whole modern approach to-
48
WARS, GUNS, AND VOTES
ward failing states had been based on the premise that they would
be rescued by democratic elections. The approach had seemed to be
vindicated by the enthusiastic take-up of elections even in the most
unpromising circumstances. Afghanistan, among the most back-
ward societies on earth, was able to run an election within months
of the expulsion of the Taliban. Iraq, about the most violence-torn
place on earth, was able to conduct an election with quite a high
turnout. The Democratic Republic of the Congo, a society with the
staggering misfortune of Belgian colonialism, followed by Mobutu,
followed by civil war, was still able to hold a competitive election.
The dread shown by the Soviet authorities to any form of competi-
tive election has, I think, confused us into thinking that achieving a
competitive election is in itself the key triumph. The reality is that
rigging elections is not daunting: only the truly paranoid dictators
avoid them.
Why is it so easy to hold elections even in unpromising cir-
cumstances? Surely it is because both political parties and voters
face strong incentives to participate in them. For political par-
ties the incentive is that the election is the route to power. For
the governing party it is a fair bet that the election will consoli-
date power and gain legitimacy in the eyes of the donors. For
opposition parties there is at least a chance of power, and with
the governing party mobilizing its supporters, even if victory is
unlikely it is important to have a countermobilization of sup-
port, otherwise it will drain away. Why do voters bother to vote?
Economists have tied themselves in knots here, missing the obvi-
ous. We are so wedded to the notion that people’s actions must be
in their material self-interest in order to be rational that our ap-
proach is largely confined to what is known as instrumentalist—
or, more colloquially, “What’s in it for me?” A young Northern
Irish economist at Oxford, Colin Jennings, helped me to think on
more realistic lines. Obviously influenced by his Northern Irish
experience, he emphasized the satisfaction that people get from
Votes and Violence
49
using their vote as a way of expressing their identity: voting is
satisfying in the same way that wearing a football scarf is satisfy-
ing. And so voter turnout is likely to be particularly high where
identity politics rules. Paradoxically, the less politics is about
policies, the stuff of instrumental voting theories, the stronger is
people’s incentive to vote. In America voting may be instrumen-
tal. Indeed, perhaps that helps to explain the low turnout; but
in
the divided societies of the bottom billion, voting is likely to be
primarily expressive.
I t i s t i m e t o s u m up where we have got to, and it is not attractive. Democracy, at least in the form it has usually taken to date
in the societies of the bottom billion, does not seem to enhance the
prospects of internal peace. On the contrary, it seems to increase
proneness to political violence. Probably related to this failure to se-
cure social peace, democracy has not yet produced accountable and
therefore legitimate government.
Incumbent politicians have won elections by methods that re-
quire them to misgovern. This is supported by the evidence that
democracy seems to retard reform.
In promoting elections the rich, liberal democracies have basi-
cally missed the point. We want to make the bottom billion look
like us, but we forget how we got to where we now are. We did
not do it in a single leap: dictatorship to liberal democracy. We
have been unrealistic in expecting that these societies could in one
step make a transition that historically has been made in several
distinct steps.
Perhaps, in encouraging elections, we have landed these so-
cieties in an unviable halfway house that has neither the capacity
of autocracies to act decisively nor the accountability of a genuine
democracy. Soon I am going to argue that it is not as hopeless as it
might appear. But we have not yet done with the upsetting material:
50
WARS, GUNS, AND VOTES
the story is going to get worse before it gets better. I will close with
the comment, made the day after the Kenyan election by Michael
Ranneberger, the American ambassador. “It’s a sad day for Kenya,”
he lamented, but then came the acid: “My biggest worry now is vio-
lence, which, let’s be honest, will be along tribal lines.”*
* “Tribal Rivalry Boils Over in Kenyan Election,” New York Times, December 30, 2007.
C h a p t e r 2
E T H N I C P O L I T I C S
In those Kenyan elections the opposition candidate,
Raila Odinga, was a Luo, one of Kenya’s forty-eight ethnic
groups. He secured 98 percent of the Luo vote. This was iden-
tity voting with a vengeance. Does it matter?
Everyone has some subnational identity and usually several of
Paul Collier - Wars, Guns, and Votes Page 6