The Economics of Peace and Security Journal, ISSN 1749-852X P. Collier, War and military expenditure in developing countries p.

10
© www.epsjournal.org.uk – Vol. 1, No. 1 (2006)

average country a doubling of military expenditure reduced the growth rate for a
War and military expenditure in developing period, eventually leading to a reduction in the level of income of 20 percent. For
countries and their consequences for developing countries, the adverse effects of a given level of military expenditure on
income are probably even more costly than for the global average. In developed
development countries such expenditure may in part be concealed routes for providing subsidies
to high-tech firms, hence the term “military-industrial complex.” In the poorer
Paul Collier developing countries military equipment is imported, rather than produced
domestically and so does not offer any side-benefits to technical progress.

D
eveloping countries have enough problems without either the waste of I now turn to the costs of war.
resources constituted by military expenditure, or the social and economic For developing countries by far the The cost of civil war is considerable.
destruction brought about by warfare. I briefly review the evidence on the most common form of war is civil During the war the growth rate is
adverse consequences of military expenditure and warfare for development. I then war. Whereas international warfare typically reduced by around 2 percent.
turn to the question of why, since military expenditure is so costly for low-income is often quite brief, civil wars last a
countries, it is nevertheless so high. I show that some of the strongest empirical long time – typically around seven
influences on military expenditure years. A recent analysis finds that such wars are getting longer – they now appear to
reflect either neighborhood arms continue for around three times as long as the civil wars prior to 1980.2 The cost of
Development, not deterrence, is the races, or the patronage demands of civil war is considerable. During the war the growth rate is typically reduced by
most effective strategy for building politically powerful military around 2 percent. The losses can sometimes continue post-war: for example, people
safe societies. establishments. Both of these may continue to move their wealth out of the country due to perceived high risks of
p r o b l e ms a r e p o t e n t i a l l y further conflict. Such perceptions would often not be misplaced. One model estimates
addressable. One of the other major that there is a 50 percent risk of conflict renewal during the first five post-war years.3
influences on military expenditure in developing countries is internal rebellion. Hence, as I will discuss further below, countries can get stuck in a conflict trap.
Where civil wars are ongoing military expenditure is greatly elevated. Further, there Finally, there is new evidence that the cost of a civil war spills over to the whole
is evidence that governments set their defense expenditure at levels designed to deter region in the form of reduced growth rates.4 One route for this might be increased
such rebellions. I then discuss why the incidence of rebellion is so high and show that perceptions of risk on the part of investors; another might be regional reductions in
the risk of rebellion is strongly linked to economic causes – a lack of development demand following from the fall in income in the country that is directly affected.
is a major risk factor. I further discuss whether military expenditure achieves its In summary, even where military expenditure is not associated with conflict, it is
intended effect of deterring rebellion and find that it does not. Indeed, since poor a drag on development. Active military conflict can lock a country into a sustained
economic performance is a major risk factor, high military expenditure, by phase of economic contraction.
contributing to such poor performance, may inadvertently contribute to the risks that
it is attempting to reduce. I conclude by suggesting that development, not deterrence, Why do the governments of developing countries have such high levels of
is the most effective strategy for building safe societies. military expenditure?

How do military expenditure and war affect development? Developing countries have astonishing levels of poverty, yet their governments
choose to devote a significant proportion of their resources to military spending
Both military expenditure and war retard development. This is not surprising, but which, as discussed above, actually retards growth and so accentuates that poverty.
there is now reasonable quantitative evidence on the scale of the effects. Military I now discuss why governments choose to use their resources in this way.
expenditure diverts government resources that could be put to better use – public With Anke Hoeffler, I have analyzed the global pattern of military expenditure,
services, infrastructure, or lower taxes. A joint analysis by the research departments trying to understand why some countries spend a far higher proportion of GDP on the
of the World Bank and the International Monetary Fund (IMF) estimated the cost to military than do others.5 The global average for military spending is around 3.5
growth and the level of income of military expenditure.1 The study found that for the percent of GDP, but the ranges from virtually zero, to an astonishing 45 percent. We
The Economics of Peace and Security Journal, ISSN 1749-852X P. Collier, War and military expenditure in developing countries p. 11
© www.epsjournal.org.uk – Vol. 1, No. 1 (2006)

find that five factors are driving these large differences: with the minister of finance, that their own country should spend approximately at
the same level. Whatever the interpretation, the consequence of this regional spillover
1. Active international warfare effect is that military expenditure is, in effect, a regional public bad. Each time one
2. Peacetime military budget inertia country raises its military expenditure there will be a ripple effect across the region.
3. Neighborhood effects (arms races) Further, as neighbors respond to the initial increase, the country that increased its
4. Internal rebellion or civil war military expenditure may itself respond with further increases – the classic process
5. Beneficiaries and vested interests of an arms race. We estimate that the typical multiplier from an initial increase in
spending in one country to the new neighborhood equilibrium may involve both the
The most obvious is that high country and its neighbors having increased the level of spending by around three
If a country is at international war it military expenditure is sometimes a times the initial increase.
spends around an additional 2.5 response to active warfare. We find While the threat of international
percent of GDP on the military. If it that, controlling for other factors, if war is clearly one concern that Governments’ precautionary spending
has a civil war, it spends around an a country is at international war it might motivate military spending, to reduce the threat of rebellion is
additional 1.8 percent. Countries that spends around an additional 2.5 for most developing country considerable. A country with a 30
have a history of international war percent of GDP on the military, governments internal rebellion is a percent risk of civil war during the
spend 1.3 percent more of GDP than while if it has a civil war, it spends far more likely threat than coming five years would raise its
those that do not. around an additional 1.8 percent. international war. Currently, civil spending by around 1.2 percent of
Hence, one indirect explanation for wars are around ten times as GDP.
military expenditure is whatever common as international wars.
causes war, something I discuss in the next section. Thus, military expenditure may
There are also large differences in military spending among countries that are at often be motivated by the desire to defend the government from the threat of
peace. We find that one important influence on spending is if there is a past history rebellion. As I discuss more fully in the next section, Hoeffler and I have developed
of international war. Countries that have such a history spend around 1.3 percent of a model of the risk of civil war. We use this model to construct a predicted risk for
GDP more than countries that have not. Possibly this reflects an assessment of the each country, and for each time period. We then investigate whether military
higher risk of future conflict. However, it may also reflect inertia or political interests expenditure is related to this risk – do governments make a realistic assessment of the
– once a country has built a large military, as happens during war, there are internal risk of civil war and set their military expenditure accordingly? We find that the
forces maintaining the level of government expenditure. Such persistence would not predicted risk of civil war is significant in explaining military expenditure –
be surprising; it is indeed common in other areas of public expenditure. governments indeed anticipate the threat of rebellion and raise military expenditure
To the extent that a past war raises military expenditure because of a perceived in an attempt to reduce the risk. On our analysis, this precautionary spending is
higher risk of further war, it reflects fear of neighbors, or aggressive intentions considerable: a government of a country with say a 30 percent risk of civil war during
towards them. We might therefore expect that the level of military expenditure the coming five years would raise its spending by around 1.2 percent of GDP relative
chosen by a government would, to an extent, be influenced by the level chosen by its to an otherwise identical country without such a risk.
neighbors. This is indeed what we find. That is, the average level of spending of The above motivations for military expenditure have either been to fight a war or
neighboring countries significantly influences the level chosen by a government. This to deter it. However, these are not the only motivations for military spending. As with
can be interpreted in various ways, the most obvious of which is that of a other forms of public expenditure, military expenditure has beneficiaries.6 In
neighborhood arms race. For most countries the most serious external threat comes developed countries these beneficiaries are largely industrial companies that produce
from their neighbors and so the appropriate level of deterrence is set by the behavior military hardware. Developing countries largely import such hardware and so the
of neighbors. A different interpretation of the same phenomenon is that military domestic beneficiaries are predominantly military employees. We might therefore
expenditure is set by regional norms of behavior, in a form of emulation. If the expect that where military employees have a lot of influence over government
neighbors are spending a particular share of national income on defense, then the decisions, the government will be persuaded to choose a higher level of military
chiefs of the military, or the minister of defense, have a relatively easy case to argue expenditure. This is a natural tendency – if professors were in charge of a government
The Economics of Peace and Security Journal, ISSN 1749-852X P. Collier, War and military expenditure in developing countries p. 12
© www.epsjournal.org.uk – Vol. 1, No. 1 (2006)

they would probably increase would therefore expect that as aid frees up government resources, the government
Military dictatorships have much expenditure upon universities. This would choose to use some of them to augment its military expenditure. Since this is
higher military expenditure than is a testable proposition because evident, donors already exert pressure on aid-recipient governments to contain their
democratic governments – an there is one readily observable military expenditure. We investigated whether these efforts have been successful,
additional 2 percent of GDP. circumstance in which military testing whether aid inadvertently raises military expenditure. Fortunately, there
employees indeed have appears to be no leakage: aid is not significant as an explanation of the level of
considerable influence over military expenditure. An important implication of this result is that donors appear to
government decisions, namely if the government is a military dictatorship. We find be able to exert effective pressure and scrutiny on governments so that their desire
that, controlling for the risks of internal and external conflict, military dictatorships to spend some of the freed-up resources on the military is frustrated. In effect, donors
have much higher military expenditure than democratic governments – an additional are able to force governments to have less military expenditure than they would have
2 percent of GDP. Such spending is essentially a matter of patronage rather than the liked. In turn, this suggests that donors might have the power to reduce such
purchase of efficiently delivered services. Indeed, where the military is in charge of spending, rather than merely to contain it.
the government, military efficiency is likely to decline since there is no independent
source of scrutiny and evaluation of performance. Why is the incidence of civil war so high in developing countries?
Some of these motivations suggest that military expenditure could be
considerably lower without sacrificing interests that are worthwhile. To the extent Civil war is an important
that high spending reflects neighborhood arms races, it is potentially feasible to impediment to development both Conflict prevention is usually seen as a
negotiate mutual reductions in spending. Since most developing countries receive aid directly and through its effects on purely political matter, but empirically
inflows, it might conceivably be possible to strengthen confidence in such military expenditure. It is therefore the major determinants of the risk of
agreements by linking them to the conditions for aid eligibility. This might be done important to determine what actions civil war are often economic.
in the context of voluntarism: that is, a neighborhood might request the international are effective in conflict prevention.
community to assist the enforcement of its agreement. However, aid donors might While this is usually seen as a
also reasonably require that countries in receipt of aid inflows should avoid large purely political matter, empirically the major determinants of the risk of civil war are
levels of military expenditure. Such a use of donor conditionality could arise both to often economic. Hoeffler and I find that the level of income, its rate of growth, and
promote neighborhood arms reduction, and to discipline military governments that its structure, all have substantial effects on risk. Countries with low per capita
would otherwise indulge their own sector in excessive expenditure. income, slow or negative growth, and dependence upon primary commodity exports,
are considerably more at risk than other countries. Thus, the relationship between
Aid and military expenditure conflict and development works in both directions. As well as conflict being
detrimental to development, development reduces the risk of conflict. This
Whether donors play such an overt interdependence creates a trap. Time is needed for development, but in each time
Three findings: role in reducing military period there is a risk of conflict. If a country starts from poverty, slow growth and
1. Aid is not significant as an expenditure, they have a direct primary commodity dependence, it is likely to lapse into conflict before it has had the
explanation of the level of military responsibility to contain it at least to time to develop its economy. In turn, the conflict can sufficiently retard development
expenditure. the extent of preventing aid from so that, even when the country returns to peace, it is likely to fall back into conflict
2. Donors can do exert effective being diverted into military before having had sufficient time to develop.
pressure. purposes. Aid is usually “fungible.” This is part of the rationale for aid in post-conflict countries. Fortunately, aid
3. Donors may have the power to That is, by financing expenditure turns out to be particularly effective in accelerating growth during the first decade
reduce such spending rather than that the government would post-conflict.7 By targeting large aid inflows to post-conflict countries the
merely contain it. otherwise need to make, it releases international community can raise growth and hence reduce the risk of conflict
government resources that it can renewal.8 Cumulatively, the growth raises the level of income and the economy tends
then use for its other priorities. We to diversify, so that with luck the country can get through the phase of high risk
The Economics of Peace and Security Journal, ISSN 1749-852X P. Collier, War and military expenditure in developing countries p. 13
© www.epsjournal.org.uk – Vol. 1, No. 1 (2006)

relatively quickly. at http://econ.worldbank.org/files/629_wps1577.pdf.

Does military expenditure reduce the risk of civil war? 2. Collier, P. and M. Soderbom. 2001. “On the Duration of Civil War.” World Bank
Policy Research Working Paper 2681. Available at http://econ.worldbank.org/files/
A possible justification for military expenditure is that it acts as a deterrent to war. 2455_wps2681.pdf.
Hoeffler and I test this, investigating whether countries with high military
expenditure have a reduced risk of civil war. Such a test is not straightforward. For 3. Bigombe, B., P. Collier, and N. Sambanis. 2000. “Policies for Building Post-
example, as the risks of civil war mount a government is likely to increase its military Conflict Peace.” Journal of African Economies Vol. 9, No. 3, pp. 323-348.
expenditure. Unless this effect is taken into account, increased military spending will
spuriously appear to increase the risk of war even if in fact the direction of causation 4. Murdoch, J. and T. Sandler. 2001. “Economic Growth, Civil Wars, and Spatial
is the opposite. Once due allowance is made for this problem, we find that military Spillovers.” Journal of Conflict Resolution Vol. 46, No. 1 (February).
expenditure has no effect on the risk that a civil war will be initiated: high spending
does not appear to deter rebellion. This is surprising, but it may indicate that a 5. Collier, P. and A. Hoeffler. 2002. "Regional Military Spillovers." Mimeo. World
conventional military presence, such as soldiers in barracks, is largely ineffective in Bank, Washington, DC.
arresting the incipient stages of a rebellion. Good rural policing, or simply a good
rural administration, may be more effective than an army. This is not to imply that 6. See Neil Cooper’s chapter in this issue for a discussion of how the economies of
military force has no role in conflict prevention, but rather that expansion of forces conflict countries are linked into the world trade system, as well as David Gold’s
beyond those conventional in peacetime may be neither necessary nor even effective. chapter on conflict diamonds as a case study.

Conclusion 7. Collier, P. 2002a. "Aid, Policy and Growth in Post- Conflict Countries." Mimeo.
World Bank, Washington, D.C.
That military expenditure and conflict have adverse consequences for development
is unsurprising but important. The policy challenge is to reduce them. I have 8. For a case study where this advice was not followed, see Tilman Brück’s chapter
suggested that substantial components of military expenditure could be reduced on Mozambique in this issue.
without jeopardizing security interests. Military expenditure does not appear to be an
effective deterrent of rebellion, and, if it is reduced in a coordinated manner across
a region then external security interests would be unaffected. The resources released
by reduced military expenditure could be used to increase growth rates, and this in
turn would gradually but effectively reduce the risk of internal conflict. Development,
not military deterrence, is the best strategy for a safer society.

Notes

At the time of writing Paul Collier was at the World Bank. He is now back at Oxford
University where is he a professor of economics. The findings, interpretations, and
conclusions expressed in this article are entirely those of the author. They do not
necessarily represent the views of the World Bank, its Executive Directors, or the
countries they represent.

1. Knight, M., N. Loayaza, and D. Villanueava. 1996. “Military Spending Cuts and
Economic Growth.” World Bank Policy Research Working Paper 1577. Available