ments that resort to it are therefore governments that are desperate:
ones that have almost given up on the future because their struggle
to survive in the present is all-consuming. One likely reason that
President Mugabe has given up on the future is that he is himself
eighty-four years old: the oldest head of government in the world.
Ironically, Zimbabwean citizens are on average among the youngest
in the world, so you might otherwise have expected this society to
be extremely conscious of the future. Unfortunately, even ordinary
citizens have good reason to discount the future: life expectancy in
Zimbabwe is rock-bottom.
Zimbabwe is truly rare in being a peacetime society in full-
blooded hyperinflation. But we wondered whether during civil
war governments might often get desperate. Normal tax revenues
decline as the economy contracts in the face of violence. These de-
clining revenues collide with exploding expenditures as the military
demands a larger budget: during a civil war military spending typi-
cally nearly doubles. So, we reasoned, governments were likely to
resort to the printing press. This supposition turned out to be cor-
Inside the Cauldron
91
rect, but our concern was not what happened to the economy dur-
ing civil war but its implications for the post-conflict recovery. The
legacy of high inflation is that people expect inflation to continue
and learn to economize on holding cash. If the government wants to
get back to where it started—low inflation and people sufficiently
confident to hold the currency—it will need a prolonged period of
fiscal restraint. In effect, during the civil war the government’s in-
flation strategy has been equivalent to borrowing: it has created the
liability of expected inflation. The post-conflict government now
faces the need to wipe out this liability. But it faces its own desperate
fiscal challenges.
Tax revenues will take time to rebuild. Typically the economy
will have become less formal as firms try to escape taxation. Raising
taxes too rapidly retards the process of rebuilding formal activity. In
Sierra Leone the head of the chamber of commerce told me how his
membership had dwindled as firms had gone informal. Yet there
are huge demands for government spending: rebuilding infrastruc-
ture, alleviating crises of collapsed social provision, creating jobs for
unemployed youth. We found that an unnoticed benefit of post-
conflict aid is that it helps to square this circle. The government
no longer needs to use the inflation option so aggressively, and this
enables confidence in the currency to return. Like other aspects of
economic recovery, this is a slow process: this use of aid is an invest-
ment in confidence and it takes more than a decade before citizens
are back to their pre-conflict willingness to hold the currency. But
without post-conflict aid it takes far longer.
Does it matter? Well, it turns out that inflation is particularly
damaging in post-conflict economies for a simple reason. During a
civil war residents shift their assets abroad: it is called capital flight.
Such flight is a major phenomenon across the societies of the bot-
tom billion. With Tara MacIndoe, a Zimbabwean graduate student,
Anke and I have estimated the proportion of Africa’s private wealth
that is held outside the region. By 2004 it had reached the astound-
92
WARS, GUNS, AND VOTES
ing figure of 36 percent: more than a third of Africa’s own wealth is
outside the region. Unsurprisingly, during and after civil war capi-
tal flight is even more severe than this average. For the post-conflict
period the legacy of accumulated capital flight is potentially a life-
line: if only the money can be attracted back. But usually instead of
being a lifeline it is a continuing hemorrhage: faced with a high risk
of conflict reversion, people continue to shift their money out. This
is a collective action problem: in aggregate, capital flight retards eco-
nomic recovery and so makes conflict reversion more likely. Every-
one has an interest that everyone else should keep his capital in the
country, but his individual interest is to shift it out.
So how does this relate to inflation? Victor discovered that
capital flight in post-conflict situations was particularly sensitive to
inflation, much more than in normal peacetime conditions. We are
not sure why: perhaps high inflation is seen as a sign of future insta-
bility, indicating that the government is itself discounting the future
heavily. But the implication is that aid during the post-conflict pe-
riod is particularly effective. By enabling the government to reduce
inflation, aid is geared up by reduced capital flight, and indeed pos-
sibly by capital repatriation.
It is not only capital that is lost during conflict, it is also skills.
With Marguerite I have tried to analyze the effect of the conflict in
Sierra Leone on the skill base of private firms. There are few data
available for post-conflict economies, and so this work was right on
the edge of what is possible. We were given a new survey of firms and
workers that had been conducted by the United Nations Develop-
ment Program (UNDP). Although it included a lot of information
on worker training, by itself it was largely uninformative. We needed
to match it with other information. A team of researchers at Berke-
ley had surveyed households across Sierra Leone, recording the fam-
ily history of deaths through the violence of the civil war. The team
kindly shared their data with us, and this allowed us to build a picture
of how the incidence of violence differed, locality by locality.
Inside the Cauldron
93
Our idea was to see whether the violence had destroyed jobs
and skills by matching this picture of the incidence of violence
against the survey of firms and workers. There was, however, a fur-
ther problem: we needed to know what the economy had been like
locality by locality before the violence. I sent Marguerite off to the
libraries to search through the archives: one of the useful features
of Oxford is that it has good collections of musty documents, and
I guessed that there must at some time in the past have been a sur-
vey of firms in Sierra Leone. Sure enough, after a lot of searching
Marguerite discovered that there had indeed been such a survey, but
Oxford had only a partial set of the report, done some thirty-seven
years ago. Our library ran a national search, and the missing vol-
umes were eventually found. I suppose that in a few years we will
be able to download all these old documents from the Web and such
searches will become a thing of the past.
At last we thought we were in business: how had the violence
affected firms, jobs, and skills? And then we started to worry: sup-
pose that the violence had been targeted on the poorest places; su-
perficially it would look as though the violence had impoverished
them, but it would be sp
urious. We were in danger of getting cau-
sality the wrong way around. There is a way around such problems:
you need to find something that increases the risk of violence but
that does not directly affect the economy. Fortunately, we hit on one:
the RUF rebel force had been based in Liberia: one lawless state had
served as a safe haven for spewing violence into its neighbor. And so
within each district of Sierra Leone differences between localities in
the distance from Liberia turned out to be a pretty good predictor
of the scale of the violence; equally important, other than through
violence it did not much affect the economy. We were at last truly in
business. It took around three months to reach that point and about
three days to move from the data to the results. Until that point you
cannot tell whether all that effort has been wasted. It was, of course,
Marguerite who was facing the risk: had I marched her up a blind
94
WARS, GUNS, AND VOTES
alley, or was she going to find results that could be defended in a
doctorate?
What we found looks interesting. Seven years after the end of
the conflict there was no trace of any effect of the violence on either
the number of firms or the number of jobs. By looking at the size of
firms and the year when they were established, we found that where
violence had been intense, firms had shrunk. Although to an extent
these firms had bounced back once the fighting was over, it had left
a significant legacy: it had sharply reduced worker productivity. Re-
sponding to the problem of low productivity, firms in the previously
violent districts were more likely to be undertaking basic training of
their workers. Evidently, violence had deskilled the workforce. The
overall picture was of a flexible private economy that had been rav-
aged: firms reestablished and workers could find jobs at some pitiful
wage, but the skills that would have justified higher wages had been
destroyed. More than forty years ago Nobel laureate Ken Arrow
had the key insight into the process of skill accumulation in a soci-
ety. He called it “learning by doing”: productivity rises in an activity
with practice. Conceptually, there is a counterpart to learning by do-
ing: “forgetting by not doing.” In the prosperous economic context
of Arrow’s work, this logical possibility was not even sufficiently
important to warrant a footnote. But for the economics of civil war
it matters. Civil war is development in reverse, and Arrow’s model
is running backward.
Are there any forgotten skills that are particularly important
for the post-conflict recovery? I think that there are. In fact, they
are too mundane for the aid agencies to notice them. The agencies
spend a fortune on the rather ambitious agenda of inculcating at-
titudes of reconciliation, which recent studies find to be ineffective.
But they neglect the obvious. During civil war the sector that col-
lapses most severely is construction: the society is hell-bent on de-
struction and so nobody is investing in buildings and infrastructure.
The construction sector uses a lot of unskilled labor, which has the
Inside the Cauldron
95
potential for job creation among the unemployed youth that con-
stitute a post-conflict powder keg. But even the construction sector
needs skills: you cannot build a wall with only unskilled labor. And
so the fact that the basic skills needed for construction have atro-
phied during the war becomes an impediment to rapid economic
recovery. Showing Marguerite the half-collapsed roof of a building,
her guide explained, “My grandfather would have known how to
repair that, but I don’t.”
As donors and the government try to rebuild, their spending
pushes up construction prices and so gets dissipated: the skills short-
age is a bottleneck in reconstruction. For example, in Liberia the
cost of constructing a school has roughly doubled. The construction
bottlenecks need to be broken. Some donors do this by hiring Chi-
nese contractors: the Chinese face no bottlenecks because they rou-
tinely bring in absolutely everything, including the entire workforce.
But resorting to the Chinese throws out the main short-term benefit
from the recovery of the construction sector, which is to generate
jobs for young men. This looks likely to be critical in bringing down
the risks of conflict. So the solution to the skill bottleneck is training.
Post-conflict situations need squads of bricklayers, plumbers, weld-
ers, and so forth, who set about training young men. Unfortunately,
it is too mundane for the development agencies to organize it. We
need Bricklayers Without Borders.
Wh e r e w e h av e g ot t o is that post-conflict societies are fragile, and that there does not seem to be a simple political solution.
The strategy that makes a difference, really bringing down risks,
is peacekeeping, evolving into an over-the-horizon guarantee, for
which the exit strategy is economic recovery speeded by aid.
Each of these components seems to make a significant differ-
ence, but this does not mean that they are necessarily worth doing:
they might simply be too costly to be worth the candle. How can we
96
WARS, GUNS, AND VOTES
judge whether an intervention is worthwhile? The answer provided
by my profession is cost-benefit analysis, which is just a fancy term
for weighing up the pros and cons. So let’s weigh them up. Since any
military intervention is now controversial, I am going to focus on
peacekeeping. Is it cost-effective? Two yardsticks are useful: what is
the ratio of costs to benefits, and what is the net gain?
The reduction in risk achieved by peacekeeping forces depends
upon their scale of deployment. Our estimates should be seen as
rough approximations: the precise results depend upon technical
choices, and while we think our particular choices are defensible,
they could surely be improved upon. We estimate that an annual
expenditure of $100 million on peacekeepers reduces the cumulative
ten-year risk of reversion to conflict very substantially from around
38 percent to 17 percent. If peacekeeping forces are scaled up, the
risk falls further. At $200 million per year the risk is down to around
13 percent, and at $500 million it is down to 9 percent. The next step
is to convert this reduction in risk into a benefit. For that we need an
estimate of the costs of conflict. I am going to use the figure of $20
billion. Although this seems enormous, for the typical civil war in a
country of the bottom billion it is at the lower bound of the range of
realistic costs: it is likely to be considerably too conservative. If a civil war inflicts costs on the society of around $20 billion, then of course
avoiding one confers benefits of $20 billion. By extension, a strat-
egy that replaced an inevitable war with one that occurred only if a
flipped coin came up heads would therefore be worth $10 billion.
More generally, each percentage point reduction in the risk of
a civil war is worth around $200 million. Recall that peacekeeping
at the level of $100 million per year sustained over the post-conflict
decade reduces the risk of civil war by 21 percentage points. So the
value of the benefit is around $4.2 billion. Since the peacekeepers are
there for a decade, their total cost is $1 billion. We are at last ready
for the punch-line number: the ratio of benefits to costs is better
than four to one. Peacekeeping looks to be very good value. Given
Inside the Cauldron
97
the difficulties of estimation, this is likely to be a long way out. Tech-
nically, it is possible to estimate statistical confidence intervals, and
I have duly done so. But a much surer way of building confidence
is the challenge of rival researchers. As other estimates build up we
will learn the range of credible answers. I make no claim for the
numbers that I have just presented, other than that they can be very
wrong before peacekeeping starts to look a waste of money. I was
invited to present these results, and the case for peacekeeping, to
the panel of judges for the 2008 Copenhagen Consensus. The panel
assesses ten rival research teams making the case for international
public money to be spent on something. The process is terrifying:
the panel consists of Nobel laureate economists, and defending my
work before it reminded me of my doctoral viva more than thirty
years earlier. The verdict of the panel was that peacekeeping was
selected as one of the approved expenditures. In their words: “The
expert panel found that peacekeeping forces in post-conflict societ-
ies could provide fair value for cost.”
While the ratio of benefits to costs is a useful guide to action,
it is not in fact the end of the story. As will be evident from these
figures, peacekeeping forces appear to be subject to diminishing re-
turns: as you keep on expanding the force, you get less bang for the
buck. Scale is not everything, of course; quality matters as well. The
large United Nations force originally assigned to Sierra Leone was
useless because its soldiers lacked the mandate and the willingness to
fight. But, given the level of quality, scale matters. Since the bang for
the buck diminishes, there is at least potentially some ideal scale of
intervention. Although the notion of an ideal scale might seem eso-
Paul Collier - Wars, Guns, and Votes Page 12