teric, it can be given a very clear meaning. The ideal scale is reached
when a further increase in peacekeeping expenditure would gener-
ate additional benefits that just equal their cost: beyond that, ex-
pansion would be wasteful. In principle, the analysis that yields the
ratio of costs to benefits also yields the ideal scale of peacekeeping
operations. Obviously, peacekeeping has to be tailored to particular
98
WARS, GUNS, AND VOTES
circumstances, and so the question is not answerable statistically on
any realistic array of data.
Given the primitive nature of my own model it is definitely
a bridge too far, but let me take it nevertheless to illustrate the ap-
proach. The point at which additional benefits match additional
costs is reached when the peacekeeping force is somewhere between
$100 million and $200 million. At $200 million the risk of conflict
reversion is reduced by around 25 percentage points and so is worth
around $5 billion. Its cost over the decade is $2 billion: effective
peacekeeping is not cheap. While peacekeeping on this scale has a
much less impressive ratio of benefits to costs, it still leads to a large
overall gain of around $3 billion. These estimates are conservative
because they are based on a cost of war that omits many important
elements, and so the likely overall gain from a peacekeeping opera-
tion is, I suspect, even larger. The core role of politicians is to mobi-
lize the collective action that supplies public goods with benefits far
in excess of their costs. Peacekeeping is such a public good.
Is this quantification a complete fantasy? I think that in the
Democratic Republic of the Congo there is a reasonable case that
the presence of peacekeepers has averted a catastrophe. If so, it has
surely been good value for money. Quantification forces you to flesh
out such judgments. Nobody is going to be so foolish as to base pol-
icy on numbers. But it is surely useful, given that such huge amounts
of money are being spent and that lives are on the line, to try to
move beyond gut instinct. Indeed, peacekeeping is quite unpopular
with aid agencies: they see large amounts of money being channeled
through their ministries of defense and would like it diverted into
their own budgets. Such decisions should not be taken on the basis
of turf wars: in the end, the question of whether peacekeeping gives
value for money has to be faced.
While the initial maintenance of peace cannot credibly be done
without troops on the ground, the British experience to date in Si-
erra Leone suggests that it may be possible to phase out the bulk
Inside the Cauldron
99
of international troops after, say, five years, replacing them with
a guarantee made credible by a rapid-reaction force. The French
security guarantee for Francophone Africa prior to the late 1990s
reduced the risk of a civil war breaking out in the typical country
of Francophone Africa from around 10 percent in any five-year pe-
riod to around 3 percent. The risk reduction achieved by the French
security guarantee can perhaps give some guidance as to whether
over-the-horizon guarantees are cost-effective.
When I started to think about how to do a cost-benefit analysis
of the French guarantee, I expected to need three components. One
would be a figure for the reduction in the risk of conflict: I have just
given you that, from 10 percent to 3 percent. The second would be
the cost of achieving this reduction in risk. I asked the French Trea-
sury for an estimate of approximately how much its rapid-reaction
force had cost, and they gave me a ballpark figure. As a ballpark it
must be treated with heavy caveats, but they thought that it might
have been around $1 billion per year. This is equivalent to a super-
force of peacekeepers in a single country, but this very scale pre-
sumably added to its credibility. Indeed, the guarantee force must
evidently be at least as large as that needed in the largest envisaged
operation. The third component in a cost-benefit estimate would
normally be the cost of an averted conflict. For peacekeeping I used
the figure of $20 billion, but for over-the-horizon guarantees, I hit
on a way of avoiding the need for any figure at all. This was to value
the over-the-horizon guarantee not relative to the absence of any
peacekeeping but relative to the continued presence of peacekeepers
in the country. This way I could pose the question as “How many
peacekeepers could be brought home if a guarantee was put in place
while leaving the risk of conflict unaltered?” An advantage was that
posed in this way, I did not need to value any change in the risk of
conflict: by design there would be no change.
The gain from the release of peacekeepers on the ground clearly
depends upon the size of forces and on the number of situations
100
WARS, GUNS, AND VOTES
covered by a single rapid-reaction force. For example, I estimated
that if the initial forces were costing $500 million, then they could be
scaled back to around $100 million. This estimate is purely illustra-
tive. There are simply not enough instances of peacekeepers being
partially withdrawn and replaced by an over-the-horizon guarantee
for us to be able to make a defensible estimate. But the illustration is
a guide to the judgments that will need to underpin actual decisions.
For example, in my example a soldier based over-the-horizon in a
rapid-reserve unit is much less effective than one based in the con-
flict situation, and this seems plausible. However, the rapid-reaction
force might nevertheless be more cost-effective if the same force can
cover more than one situation. A rapid-reaction force is analogous
to a fire brigade, whereas a force in situ is analogous to a sprinkler
system. In my example, a rapid-reaction force would be cost-effec-
tive if it could provide cover for three post-conflict situations. And
this is before counting the benefit that for most of the time soldiers
would not need to be away from home.
What should we conclude from all this? Post-conflict situations
are fragile and there is usually no simple political fix. Peacekeeping,
phasing into an over-the-horizon guarantee, looks to be the key instru-
ment for post-conflict peace. The disquiet that it now evokes both in
the countries that send troops and in the regions that receive them is
understandable but misplaced. Even peacekeeping is not a quick fix: it
needs to be sustained for around a decade. Peacekeeping is the hand-
maiden of economic recovery, not its rival, and so budget wars between
the aid agencies and defense ministries of the high-income countries
are misplaced. Building post-conflict peace is expensive, and both will
need large budgets. The case for a large budget for peacekeeping is
that it looks to be very good value. We should learn to support it. Post-
conflict situations also need big aid. The aid-assisted economic recov-
ery is the true exit strategy for the
peacekeepers.
Part II
F A C I N G R E A L I T Y :
N A S T Y, B R U T I S H ,
A N D L O N G
C h a p t e r 4
G U N S : F U E L I N G T H E F I R E
As we are told: guns don’t kill people, people kill
people. The genocide in Rwanda showed that mass kill-
ing does not need guns: the Hutu government managed
to generate a slaughter of more than half a million people, largely
with machetes. But guns become necessary if your opponents have
them. The Hutu government could slaughter Tutsi civilians with
machetes because they were unarmed, but rebels face government
armies and so they need guns. No rebel guns, no rebellion, and so no
nasty, brutish, and long civil war. And because the government next
door has guns, our government needs guns: a government without
guns cannot defend its citizens against a neighbor with guns. That is
the message that many of our politicians thrive on: national security
is the ultimate national public good and military spending is the way
to achieve it.
Like many issues, whether the ready availability of guns makes
a society more dangerous or less dangerous is an empirical matter.
Despite the overheated political positions, there are three perfectly
sensible possibilities. Cheap and plentiful guns may increase the risk
of violence. Alternatively, they may make violence so dangerous
that they deter it. Finally, guns may be plentiful where there is a
lot of violence, but this may be because in societies that are violent
104
WARS, GUNS, AND VOTES
for whatever underlying reason, people make sure they have guns
around: the guns are a consequence, not a cause. Ideologues seem to
think that such issues can be settled a priori: turning to the evidence
challenges superficially plausible belief systems with the possibility
that they are make-believe.
P ro b a b ly t h e m o s t i m p o rta n t q u e s t i o n to be asked about guns is whether they work as a deterrent. However, in order
to answer this question, we first need to know why governments
buy so many of them. Somehow we have to be able to sort out the
chicken-and-egg causality problem: the risk of violence affects mili-
tary spending, and military spending affects the risk of violence. If
military spending keeps a society safe it is worth every cent. But be-
fore slipping into such easy thought patterns I decided to investigate
what has actually been driving it. By this I do not mean answers
such as “the military-industrial complex” about which the Marxists
of my youth used to fantasize.
During the Cold War there used to be an academic industry
that studied the arms race between NATO and the Warsaw Pact.
But with the end of the Cold War, that collapsed, and nothing
has replaced it. There were scarcely any recent studies on military
spending in developing countries, so Anke and I decided to do it
ourselves. Because we were new to this question it took a long time
to get it right: we finally published the results in 2007. No sooner
had we done so than President Arias of Costa Rica, the Nobel Peace
laureate, asked us to develop their implications as a support for his
initiative for coordinated reductions in military spending. Costa
Rica has led the world in virtually eliminating military spending,
and we were delighted to be able to add our evidence to support his
efforts.
Governments are not exactly forthcoming with data on their
military spending. That is scarcely surprising, but it makes the task
Guns
105
of analyzing their spending rather harder. I almost persuaded the
American government to provide its own private estimates of the
spending levels of other countries, but it balked at this. Instead we
relied upon the estimates made each year by the Stockholm Peace
Research Institute (SIPRI). We decided to measure spending as a
proportion of national income: averaged across all the countries of
the world for which SIPRI has data, and the forty-year period from
1960 to 1999, the global average was 3.4 percent. Expressed as a per-
centage it is a small number, but expressed in dollars it is large: by
2006 it had grown to $1.2 trillion. That is around ten times the global
aid budget. The countries of the bottom billion alone are spend-
ing around $9 billion. This compares with their total aid inflows
of only around $34 billion. Our question was why did some coun-
tries spend a higher proportion than others, and why was spending
higher at some times than at others: the highest proportion we found
was 46 percent of income, and the lowest a mere 0.1 percent.
We started with the obvious and gradually got more elaborate.
The most blindingly obvious reason for a government spending a
lot on the military was if it was fighting a war against some other
government. I figured that if we did not find that in the data we
should give up and work on something else. There it was in the
data: controlling for everything else, if a country is engaged in an
international war its military spending increases by 1.5 percentage
points of GDP. However, much as you might be forgiven for think-
ing otherwise, international wars are now so rare that this reason
for military spending accounts for only a tiny proportion of global
spending: most spending occurs during peacetime.
Just because a country is not currently at war with some other
country does not mean that it regards itself as being free of external
threats. We scratched our heads trying to think of a good observable
proxy for the perception of an external threat. One idea we came up
with was once bitten, twice shy. Expressed more professionally, if a
country had once had to fight an external war, maybe it would be
106
WARS, GUNS, AND VOTES
more fearful that it would have to do so again. Perhaps the country
had a particularly dangerous neighbor, maybe it had a particularly
aggressive leadership, or perhaps it saw itself as an international po-
liceman, riding to the rescue of distressed regimes. We decided to
try it, focusing on the history of warfare since the end of the Second
World War. Sure enough, once a country had been engaged in an
international war, its military spending was permanently higher by
around 1.8 percentage points. We tried to see whether this faded
with time. Presumably at some point it does, but we could not find
any such tendency: as far as we could see, wars fought years ago
were still leaving their legacy in the form of higher spending. If this
is right, a disturbing implication is that much of the costs of an in-
ternational war accrue after it is over: the society continues to be
burdened with higher military spending.
A previous war is one reasonable proxy for a perceived external
threat, but we decided to try another one that was even more obvi-
ous and rather neatly complemented it: the Cold War. The Cold
War was evidently a period of perceived threat, but unlike our pre-
/>
vious proxy, it was a threat that did not materialize. Further, it had
a clear ending, namely the collapse of the Soviet Union. The end
of the Cold War therefore constitutes a revealing natural experi-
ment for the coordinated removal of a perceived threat. As you will
see, such an experiment is useful: it simulates the effect of ending
the cold wars in Lilliput that bedevil the societies of the bottom bil-
lion. So what happened once the Cold War ended: did it show up
in global levels of military spending? It most certainly did: with the
end of the Cold War global military spending fell by an astonishing
35 percent. The collapse of the Soviet Union delivered a huge global
peace dividend.
The nature of the threat during the Cold War was, however,
unusual: America and the Soviet Union could threaten each other
despite the fact that they did not have a common border. This was, of
course, due to nuclear missiles. Pretty well all other external threats
Guns
107
come from neighbors: no border, no genuine threat. Even the sub-
sequent proliferation of nuclear missiles has barely changed this
state of affairs: it is because India and Pakistan fear each other as a
result of their common border that they now point nuclear missiles
at each other.
How big a threat is your neighbor? Well, other things equal, it
depends upon how much they are spending on the military. There
is a whole body of fancy economic theory that has modeled warfare.
Economics has a knack for ugly terminology, and these models are
termed contest success functions. The punch line of this work is that
if your enemy spends more, then it is wise for you to spend more.
You might have sensed that that was quite likely, but you will doubt-
less be relieved to know that economists have done the mathematics
to prove it. So, armed with the cast-iron certainty that comes from a
theorem, we decided to investigate whether there really were arms
races in Lilliput. First we had to get countries organized according
to who their neighbors were. We found a data set that purported to
do this, cleaned it up so that China no longer bordered on Uganda,
and got to work. Incidentally, this tells you why our sort of research
requires patience: it is necessary to check, check, and check again
Paul Collier - Wars, Guns, and Votes Page 13