homogenous villages were others that had been created as settle-
ment schemes at various times, in which there were varying degrees
of ethnic mix. She was able to show that controlling for other char-
acteristics, the more ethnically mixed villages revealed a lower level
of trust through the strategies people chose to play. Another result
is that people are more willing to pay taxes to benefit people with
whom they have an affinity than if they know that much of the ex-
penditure will benefit people who are very different. When these
results were reported in Europe, they produced a frisson of concern
that immigration and the resulting change to multiethnic societies
would erode the welfare states that characterize the continent.
There is also evidence that the public good of scrutiny of gov-
ernment breaks down. I have already recounted the anecdote about
the Mercedes bought for the speaker of the Nigerian House, but
there is more systematic evidence. A particularly convincing study
compares the functioning of school boards in various parts of ru-
ral Kenya. The school boards, composed of parents, can raise funds
and manage the school: they thus have an important role to play in
determining school quality. A clever study by Edward Miguel and
Mary Kay Gugerty found that where the board was diverse, man-
agement was worse: specifically, the members of the board were not
prepared to criticize people from their own ethnic group who were
failing to make contributions.
Fortunately, there is a silver lining to ethnic diversity. The adverse
effect of diversity on public provision is offset by an advantage that
it confers in private economic activity. Why does diversity enhance
private sector productivity? There is now pretty good experimental
evidence that reveals what is going on, somewhat along the lines of
Abigail’s work in Zimbabwe. Basically, diversity raises the productiv-
ity of a team because it increases the range of skills, knowledge, and
perspectives, and these help problem solving. Although diverse teams
60
WARS, GUNS, AND VOTES
do not get along as well, they are better at achieving results. There is
also some evidence that suggests that this scales up so as to affect the
overall performance of an economy. I am not particularly proud of
my own effort in this direction: I pushed to the limits of data avail-
ability, and so the results are probably not that solid. But, for what it is worth, I constructed estimates, country by country, of the public and
private capital stock and then investigated whether the productivity
of these two types of capital was affected by the degree of ethnic diver-
sity of the society. Each of these steps is precarious, but what came out
was that ethnic diversity reduced the productivity of public capital,
while increasing the productivity of private capital. While this may
be spurious, it is at least consistent both with the micro-level evidence
and with other macro-level results.
One implication is that diverse societies should play to their ad-
vantage and place as many activities as possible in the private sec-
tor. This is clearly consistent with the contrast between America
and Europe: the as yet more homogenous societies of Europe have
a larger public sector. The societies of the bottom billion, with their
high diversity, were particularly ill suited to the socialism that un-
til recently has been overwhelmingly their predominant ideology.
Their adoption of socialism was understandable: most of the first
generation of political leaders had been educated in France and
Britain in the 1950s. Not only had socialism been at its apogee, but
to their great credit Europe’s socialists were the first politicians to
support decolonization struggles. And beyond European socialism,
imitating the Soviet model carried the sweetener of a ready access
to armaments to address their security problems. One aspect of the
so-called Structural Adjustment Programs of the 1980s was that
African governments were encouraged or coerced into shifting ac-
tivities from the public sector to the private sector. Though heavily
criticized both for being coercive and for being ideologically driven,
the direction of the shift was appropriate given the diverse composi-
tion of Africa’s societies.
Ethnic Politics
61
Since diversity has beneficial effects as well as adverse ones, it
sounds as though with appropriate choices it might be a case of six of
one and a half dozen of the other. The net effect might be negligible.
However, you have already seen that an effect can be very different
in high- and low-income societies. Recall that democracy reduces
political violence in high-income societies but increases it in low-
income societies. Could the effect of ethnic diversity be similar?
Unfortunately for the societies of the bottom billion, it is. The
beneficial effects of diversity only set in at higher levels of income:
diversity is good news for America, and while Europe’s rising di-
versity may well weaken its welfare state, it will be compensated by
a more vibrant private economy. But it is bad news for Kenya and
the other societies of the bottom billion. At low levels of income,
diversity is a substantial net economic disadvantage, and it shows
up in slower growth: a highly diverse low-income society on aver-
age grows a full 2 percentage points less rapidly that a completely
homogenous one. Why might diversity help high-income societies
but hinder low-income societies? Perhaps it is because the key ad-
vantages of diversity come from skills and knowledge. In an econ-
omy with high levels of skills and knowledge, the larger the pool of
diverse skills and knowledge, the better. But in economies where
skills and knowledge are more rudimentary, there is less scope for
diversity and less use for it.
F u n da m e n ta l ly, t h e r e s u lt s s o fa r suggest that ethnic diversity makes social cooperation more difficult, and that at low
income levels this effect is sufficiently strong to be a substantial im-
pediment to prosperity. It is tempting to conclude from this that
diverse societies cannot afford to rely upon cooperation to achieve
the collective effort that is necessary for success in any economy.
The alternative to achieving collective effort through cooperation
is to achieve it through coercion. Someone is needed to direct the
62
WARS, GUNS, AND VOTES
coercion: step forward the benign dictator. The recent evidence of
China illustrates a more widespread phenomenon, that a society can
make rapid economic progress if collective effort is guided by a suf-
ficiently sensible and relatively benign autocratic leadership. Is this
the answer for ethnically diverse low-income societies?
The case for autocracy appears to be strengthened once we get
back to the fundamental issue of security. You have already seen that
in the bottom billion, democracy increases political violence in all its
main forms. While dem
ocracy makes such societies more danger-
ous, repression seems to work. We are back with the ugly fact that
Saddam Hussein kept the peace in Iraq more effectively than Jalal
Talabani. So better public goods—dictators make the trains run on
time—and better security: the case for dictatorship looks disturb-
ingly strong.
While I do not want to discount the benefits from a sensible
and benign autocrat, I think that for ethnically diverse societies, this
solution to the problem of collective action is very dangerous. Ethnic
diversity generates bad autocracies as well as bad democracies. In
an ethnically diverse society dictators usually play the ethnic card,
building their power base on their own ethnic group. As a result,
their patronage base is almost inevitably narrow, not extending be-
yond their ethnic group. The narrower the power base, the stronger
is the incentive to retain power by raping the national economy and
transferring the proceeds to their own ethnic group rather than by
building the national economy and benefiting everyone. So, on this
analysis, ethnically diverse societies would be supremely ill suited to
dictatorship.
Again, it is best to look at the evidence, but this particular ques-
tion is far from straightforward. I started with a rough-and-ready
attempt. What I found was that, judged by economic performance,
ethnically diverse societies needed democracy more than those that
were homogenous. If this result was correct, then far from needing
a dictator, ethnically diverse societies were peculiarly ill suited to
Ethnic Politics
63
them. While the result was sufficiently new to get published in a
respectable academic journal, it was manifestly only a first step at an
important question and quite possibly a misleading one. Recently
Eliana La Ferrara and her distinguished coauthor Alberto Alesina,
head of the economics department at Harvard, have revisited the
issue and published a considerably more thorough analysis. I read
their study with the mixed emotions of delight that the topic had
engaged such a heavyweight team, admiration that they had pur-
sued possibilities that I had missed, and, of course, trepidation that
my own work might be revealed as a house of cards. In the academic
world you are never more than one demolishing article away from
humiliation.
One important result that I had missed was that diversity was
less damaging at higher levels of income. But potentially this spelled
the death knell for my own result on democracy. Since democracy
is more common at higher income, my result, which did not control
for the level of income, may simply have been due to this correla-
tion. They built up their analysis step by step, first replicating my
result, then revealing their own, and finally combining the two pos-
sibilities. Happily both for ethnically diverse democracies and for
my own peace of mind, they found that both effects survived: ethnic
diversity is less problematic at higher levels of income, and is differ-
entially well suited to democracy. Even their analysis is only prelim-
inary. As they acknowledge, there are various ways in which these
results could be spurious. Nevertheless, the results caution against
the leap from the problems of ethnic democratic politics to the infer-
ence that what is needed is a dictator.
Following their work I have tried to take the analysis a little
further. Both my previous study and that of Alesina and La Ferrara
took economic growth as the measure of performance. For some
purposes this is not a bad measure. If there are indeed two offsetting
effects of diversity, the key issue is whether the net effect is positive
or negative. As a composite measure of performance, growth is as
64
WARS, GUNS, AND VOTES
good as any. However, if, as seems likely, for the societies of the bot-
tom billion the net effect is indeed negative, then we need to drill
down. The adverse effects on public goods that are doing the dam-
age must run through political or social choices. I therefore decided
to switch from looking at the effects on growth to a more direct
measure of these choices.
The issue was evidently on the boundary between economics
and politics, and I was fortunate to be able to team up with Robert
Bates, like Alesina a professor at Harvard, and the undisputed doyen
of political scientists working on Africa. For some years we have
been part of a large team under the auspices of an African-directed
research network. The team had been investigating the fraught
topic of why Africa’s economies had largely stagnated during the
forty years after 1960. Poor choices were by no means the only ex-
planation for stagnation. For example, the fact that so many African
countries were landlocked was a fundamental impediment to pros-
perity. However, choices were clearly contributing factors, and the
team decided to focus on those that were manifestly dysfunctional.
We met up at Stanford one summer and went through the narra-
tives of economic history, country by country. What emerged was a
consensus on a few syndromes. One, for example, was the misman-
agement of booms, gearing them up by borrowing and then squan-
dering the proceeds. We found that where countries stayed clear
of these syndromes they always avoided economic collapse, even if
they did not grow rapidly.
Bates and I decided to use these killer syndromes as our mea-
sure of performance. Did ethnic diversity make a country more
prone to these highly dysfunctional choices? We found that the ru-
inous combination was high diversity together with severe political
repression. This was the cocktail that had produced Africa’s dys-
functional social choices. Indeed, it was only through this lethal in-
teraction that diversity and dictatorship made a society more prone
to the syndromes. It was not dictatorship in itself, or diversity in it-
Ethnic Politics
65
self, but only their combination. This result, which is entirely based
on variation among African societies, is clearly consistent with the
globally based results: ethnically diverse low-income societies are
particularly ill suited to autocracy.
Finally I turn to what is to my mind the most insightful study:
an as yet unpublished paper by Tim Besley and his student Masayuki
Kudamatsu, provocatively entitled “Making Autocracy Work.”
They show that performance in autocracies is far more dispersed
than that in democracies: autocracies can be extremely successful
but also utterly ruinous. Their question is what drives the differ-
ence: why were none of the successful autocracies in Africa? They
build their answer around the notion of a selectariat. A selectariat
is what a dictatorship has instead of an electorate: it is the limited
group of people on whom power rests. These are the people who
could therefore potentially oust the dictator if he performs badly.
Besley and Kudamatsu d
iscovered that the difference between suc-
cessful and ruinous dictatorships is whether the selectariat is willing
to use this power. Where selectariats routinely ditch incompetent
dictators, the autocracy performs well.
This result is important, but it raises a further question: what
determines whether the selectariat is willing to ditch a failing dicta-
tor? They come up with a simple answer. The selectariat will only
dump the dictator if it is confident of retaining power, replacing him
with one of their own. Here, I think, lies the explanation for why
autocracy goes wrong in societies stratified by strong ethnic identi-
ties: in such societies political change is risky. The current selectariat
will be drawn from the ethnic group of the dictator, but if it ousts
him, it may trigger a chain of events in which power passes to a rival
ethnic group and thus to a new selectariat. When I discuss coups you
will find evidence consistent with this: in Africa ethnic polarization
strongly increases the risk of a coup. An ethnic selectariat would be
right to be fearful of disturbing the status quo. Consistent with this
argument, Besley and Kudamatsu find that ethnic diversity reduces
66
WARS, GUNS, AND VOTES
the chances that an autocracy will work. But they also find evidence
that ethnic diversity is far from being the whole story: its effects can
be overridden. A strong ideology such as Marxism makes autocracy
more likely to succeed even in the context of ethnic diversity. If the
selectariat consists of the Communist Party, whoever heads the dic-
tatorship, the party is going to remain in power. The societies of the
bottom billion do not need another dose of Marxism. But they do
need something that gives a sense of common identity.
So, while neither economic theory nor statistical analysis has
yet been able decisively to nail the issue, as far as we can tell it looks
as though the tough autocrat who rules by fear is precisely what the
diverse societies of the bottom billion most need to avoid. Although
they are able to keep the lid on political violence other than their
own, measured on a wider array of criteria, they are a disaster. Di-
versity may make democratic politics deteriorate, but it is likely to
make dictatorship lethal.
S o h ow c a n e t h n i c d i v e r s i t y be overcome? A sense of national identity does not grow out of the soil: it is constructed by
Paul Collier - Wars, Guns, and Votes Page 8