them: in addition to being British I am English, more particularly
Northern English, and getting right to the bone, I am a Yorkshire-
man; I have taught my son Daniel to sing our anthem, “On Ilkley
Moor bar t’at.” The British general election of 2001 pitched a York-
shireman against a Scot for prime minister. Yet like most people
from Yorkshire I supported the Scot. A society can function perfectly
well if its citizens hold multiple identities, but problems arise when
those subnational identities arouse loyalties that override loyalty to
the nation as a whole. As the Luo vote suggests, in the societies of the
bottom billion, ethnic identity usually trumps national identity.
The societies of the bottom billion are for the most part far more
ethnically diverse than those of the high-income countries. Often
this diversity verges on being a taboo subject: it is just too upsetting.
I think that it poses genuinely tough, but not insuperable, problems.
They will not be overcome unless they are faced.
Ethnic diversity compounds the problems that the societies of
52
WARS, GUNS, AND VOTES
the bottom billion would in any case face in making electoral com-
petition work. Yet more fundamentally, diversity impedes the basic
role of the state, the provision of public goods. It is tempting to con-
clude that what an ethnically diverse society of the bottom billion
needs is a strongman. Wrong: bad as democracy is in ethnically di-
verse societies, dictators are even worse. But there is a vital role for
political leadership: leaders must build the nation before they can
build the state.
Wh at was t h e o r i g i n o f these strong ethnic loyalties? In the absence of states, ethnicity was the obvious basis for collective action, and in a rural society bumping along at a subsistence level of
income, one form of collective action was supremely important: in-
surance. Life at subsistence is risky: if you fall sick when you should
be plowing, planting, or harvesting, your income will collapse. If
vermin eat your food stores, you face starvation. You need catas-
trophe insurance. The problem with insurance is what economists
coyly term moral hazard: if I’m insured, what the heck! If you could
insure yourself against a decline in income, why get up in the morn-
ing? And so such insurance does not exist unless the moral hazard
problem can be solved. The solution to moral hazard is not indig-
nantly to protest that the insurer should not doubt your good faith,
it is to make your behavior observable. Only if the insurer can see
that you are trying your best does the insurance become feasible. For
a private insurance company such observation would be prohibi-
tively expensive, but for a community it is feasible. Nosiness, gossip,
friendly intimacy, all the ingredients that are natural to a commu-
nity also happen to be just what is needed for insurance.
Observability is necessary but not sufficient. The right to rely
upon other people in the community when faced by a personal ca-
tastrophe depends upon a reciprocal obligation to provide such as-
sistance to others: but who is in and who is out? If anyone can join
Ethnic Politics
53
or leave the insurance group at any time, then it will be in perpetual
deficit: people will declare themselves to be members of the com-
munity when they fall on hard times and declare themselves fancy-
free when things are going well. This is known in economics as the
problem of adverse selection: unless insurance companies take care,
instead of getting a random selection of clients from the popula-
tion, they get people who know that they are bad risks. That is why
insurance companies use some device for restoring a random selec-
tion, such as offering much better terms for all the employees of a
firm than they offer to individuals who turn up at the door. This
is where ethnicity comes in: you do not choose your ethnic group.
If you are not a member of the ethnic group, you cannot choose to
become a member when times are hard. If you are a member, you
cannot choose to exit the group when things go well. That is the
economic basis for strong ethnic loyalties: it enables income insur-
ance to work in the high-risk, low-income conditions under which
it is supremely valuable. Over time, loyalty to the group becomes
reinforced by all the normal power of morality: it is morally good to
meet your obligations.
Insurance sustained by loyalty helps everyone within the group
and is not at the expense of other groups. However, even in the tra-
ditional economy loyalty to the group is sometimes at the expense of
other groups, most obviously in respect of violence against enemy
groups. But ethnic loyalties have far more scope for being at the
expense of other groups when they are transferred to the context of
the modern economy. The public purse becomes the common pool
resource that the collective action of one group can capture at the ex-
pense of other groups. It is at this stage that moral obligations to the
ethnic group collide with moral obligations to society as a whole.
My friend John Githongo, anticorruption commissioner in the
Kenyan government, blew the whistle on the corruption at the heart
of the government, becoming internationally famous and an exile
in the process. Even I could see that what John did took courage.
54
WARS, GUNS, AND VOTES
But in talking to him I got a surprise: it had taken more than cour-
age. John is a Kikuyu, the tribe that dominates the government. Un-
surprisingly, when he blew the whistle against the government, his
Kikuyu friends had accused him of betrayal. But the surprise was
that John had felt this himself as an inner struggle of conflicting
loyalties. Like many of the finest African reformers, John is a com-
mitted Christian: religious faith gives a moral framework that helps
people to put their ethnic obligations in perspective. Another com-
mitted Christian reformer, Oby Ezekwesili, who bravely slammed
the door shut on Nigeria’s public procurement scams, described the
prevailing morality: “They see themselves as good if they benefit a
few thousand kin at the expense of the nation.” It is that prevailing
morality that gives ethnic loyalties a whole new possibility for being
valuable to the group yet damaging for the society.
Far from the transition to the modern economy weakening
ethnic bonds, there are strong forces intensifying them. Occasion-
ally an event gets under the skin of a society and reveals much more
than its direct importance. Here is the story of a Kenyan funeral.
Like Raila Odinga, Mr. Otieno was a Luo. However, he had left
his native region in his youth and moved to Nairobi, where he had
become a successful businessman and married a Kikuyu woman.
So far we have a standard story of the melting pot. On his death in
1986, his widow, in accordance with Mr. Otieno’s last will and testa-
ment, arranged for his burial in Nairobi. At this point Mr. Otieno’s<
br />
Luo relatives objected: they wanted him buried back home. Indeed,
they wanted him home so badly that they took the matter to court.
Faced with a choice between adhering to the wishes of the dead
man and his wife, or the wishes of the relatives, the court was in no
doubt: he was duly buried back in his Luo village.
What on earth was going on? Think back to the key policing
roles of ethnicity: entry and exit from the obligations of the group.
There is little difficulty in policing entry: “Very sorry but we won’t
help you because you’re not one of us.” But policing exit is rather
Ethnic Politics
55
harder. The people who will want to exit from their obligations are
the successful: and how do you stop them? This is where the place of
burial comes in. The spirits of ancestors loom large in the belief sys-
tems of most ancient societies, and spirits are usually localized. Mr.
Otieno might have managed to exit his obligations during life, but
he might now be getting his comeuppance in death. Consciously, or
subliminally, an enforcement mechanism for ethnic loyalty was at
work; and unlike Mr. Otieno it was alive and well.
S o w h at h a p p e n s i f t h e r e are many ethnic groups, each with powerful loyalties? How does it affect the politics?
Electoral competition is an activity with powerful economies of
scale: if I can get 51 percent of the votes I win. Indeed, in the absence
of restraints on the use of power, I win everything. To reap these
economies of scale, power seekers group together into political par-
ties that develop brands and try to build voter loyalty. In ethnically
homogenous societies with winner-take-all voting systems, this pro-
cess tends to be driven to the extreme in which everyone amalgam-
ates into only two parties. Although the leaders of these parties are
chosen only by their respective supporters, once chosen, both leaders
chase the median voter to get elected. This produces a politics of
moderation that broadly describes how modern democracies func-
tion. One hallmark is that the activists within each political party are
usually dissatisfied with the moderation of their leaders. We see the
process played out most clearly in America, but with minor varia-
tions it is the general pattern.
When I first tried to work out how this process was affected by
ethnic diversity I came away quite heartened. Of course, if voters
had strong ethnic identities then politicians would organize their
parties on ethnic lines: this would simply be the cheapest way of at-
tracting voter loyalty. The election itself would sound very different
from an election in an ethnically homogenous society: leaders would
56
WARS, GUNS, AND VOTES
simply be mobilizing their own ethnic base rather than reaching out
to the median voter. But after the election these ethnic parties would
need to form coalitions. Any ethnic group that got too demanding
would not be able to bid itself into the winning coalition. Ethnic pol-
itics might produce a merry-go-round of changes in governments,
but each group would get close to its fair share of power.
Since publishing those ideas in 2001 I have started to have
doubts. First, ethnic politics seems likely to contaminate the content
of the election campaign. Policy choices get crowded out by iden-
tity. Let’s go back to those winning electoral strategies: remember
the one that involved playing the ethnic card. Playing on ethnic
fears and hatreds is truly the politics of the gutter: unfortunately, it
works. The holy grail of modern economic field research is the ran-
domized experiment, something that medics have been doing for
years, but it is usually more difficult to arrange with economic in-
terventions. When it comes to the content of an election campaign,
you would imagine that the scope for conducting a genuinely ran-
domized experiment is decidedly limited. Not one bit of it: Leonard
Wantchekon, a remarkable economist from Benin now working in
America, did just that. He managed to persuade the politicians of
Benin randomly to adopt different campaign messages in different
localities. This alone tells you most of what you need to know about
the election campaign in Benin, but Wantchekon’s story is yet more
depressing. Not only were politicians willing randomly to adopt
either a campaign message that they would provide good national
governance or a message that they would provide ethnic favorit-
ism, but once the results were subject to statistical analysis, it became
clear that favoritism was more effective at pulling in the votes.
Not only does identity trump policies, but to the extent that
policies do enter, instead of a race to capture the vote of Ms. Moder-
ate the All-Powerful Median Voter, there is a race to the extremes.
Colin Jennings introduced me to this tendency via the expressive
voting idea. His work analyzes how electoral competition is likely
Ethnic Politics
57
to work out in ethnically divided societies. Voting for the extremist
parties offers the strongest identity fix. It also selects the most ar-
dently sectarian leaders, so that when it comes to the stage of reach-
ing compromise in a grand coalition, the starting point for the ne-
gotiations is as far toward the position of your own ethnic group as
possible.
One graphic instance of this unattractive process is evident in
Northern Ireland, where electoral competition was meant to force
moderation, with parties heading for the center ground in order to
build a coalition. Instead, precisely the opposite happened. There
are four major political parties in Northern Ireland, two Protes-
tant and two Catholic. On each side one of these parties is moderate
and one is extreme. Prior to power sharing, the largest parties on
each side of the Protestant-Catholic divide were the moderate par-
ties: indeed, they were the parties that brokered the power sharing.
But once power sharing was introduced, voters polarized; now the
dominant parties on each side are the extremists. The ruling coali-
tion is a coalition of the extremes headed by grinning bigots who
cannot believe their luck. This seems to be a likely consequence of
identity politics more generally. Indeed, it happened in the Kenyan
elections of December 2007. The forty-eight ethnic groups coalesced
into pro-Kikuyu and anti-Kikuyu coalitions.
I also came to see that electoral competition is not the only as-
pect of democracy that matters. Electoral competition needs to be
complemented by checks and balances. In turn, checks and balances
are public goods: that is, they have to be supplied by cooperation.
Ethnic politics makes such cooperation to build checks and balances
much harder. I came across this graphically in the aftermath of the
Nigerian elections of 2007. The new speaker of the House of Rep-
resentatives, Patricia Etteh, was soon caught misappropriating the
funds meant for her office. She had, for example, acquired twelve
&nb
sp; Mercedes. I do not want to get Scandinavian about the odd dozen
Mercedes: I am quite prepared to believe that any self-respecting
58
WARS, GUNS, AND VOTES
speaker needs them. But many Nigerians seemed not to think so:
they regarded it as outrageous and she was pilloried in the media.
So far there is nothing remarkable in this story: a relatively minor
infringement that met its fate. It is the reaction to the criticism that
is of significance. As soon as she was criticized in the media, the
other politicians from her own ethnic group, the Yoruba, leaped to
her defense. Quite explicitly, their message was “Hands off: she’s
our only representative at the trough.” If a corruption charge can be
deflected by playing the ethnic card, then standards of public con-
duct are bound to be low.
So ethnic electoral politics may not be as benign as I had previ-
ously thought. This would certainly gel with the evidence on ethnic
diversity and public goods, much of which is derived from contexts
in which political choices are the result of vigorous electoral compe-
tition, such as North American cities.
M a n y s t u d i e s h av e f o u n d t h at public services are systematically worse as a result of ethnic diversity among citizens. The
association is causal: it is not just that ethnically diverse societies
happen also to have poor public services. Controlling for other char-
acteristics, greater diversity implies worse public services. Not only
that, but expenditure on channels suited to ethnic patronage, such as
the public payroll, is higher. Why does diversity make public goods
provision harder? For that we need to turn to micro-level evidence
on how collective decisions are taken.
One result clearly established by research is that trust is weaker
across ethnic groups than within them. One rather clever way of
demonstrating this was the work of Abigail Barr, a researcher in
my group, who investigated variations in the level of trust among
the communities of rural Zimbabwe. Trust is difficult to measure,
but she followed a recent line of research and used experimental
games played by volunteers who could win small amounts of money
Ethnic Politics
59
depending upon the strategies they adopted. Zimbabwe was par-
ticularly well suited to the investigation because alongside ethnically
Paul Collier - Wars, Guns, and Votes Page 7