Paul Collier
Professor of Economics
and Director
Center for the Study of African Economies (CSAE)
Oxford University, United Kingdom

Introduction and overview

Most wars are now civil wars. Although international wars attract enormous global attention, they have become infrequent and brief. Civil wars usually attract less attention, but they have become increasingly common and typically go on for much longer. The average civil war lasts about 10 times longer than the average international war. Civil wars are an important issue for development and global security. War retards development, but conversely, development retards war. Where development succeeds countries become safer, making subsequent development easier. Where development fails countries are at high risk
of becoming caught in a conflict trap, in which war destroys the economy and thus increases the risk of further war. The effects from civil war extend far beyond the country borders. Economic growth in neighbouring countries is reduced and the entire region is destabilized due to the direct and indirect impacts of civil war. In addition, there are global effects. Civil war creates territories outside the control of the recognized government. It is for example estimated that 95 percent of illegal drugs are produced in countries with civil wars. Civil war has also contributed to the rise of terrorist movements, such as Al Qaeda. Large scale terrorist organizations can probably only operate from territories beyond the control of any recognized government. Al Qaeda chose Afghanistan as a main location although the majority of recruits were not Afghans because they could operate effectively in Taliban held territory.

In our input to the round table discussion on 'Development Policy and the Armed Forces' we want to focus on three issues. First, we discuss opportunities of global conflict prevention, second, we analyze instruments aimed at shorting existing conflict and third, we make some recommendations on how to design policies for post-conflict societies. Post-conflict periods are characterized by particularly high risks of renewed civil war and we argue that the best opportunity for development policy is in the prevention of recurring war in post-conflict societies.

Our analysis of possible policy instruments in the reduction of the incidence of civil war is based on a global cost benefit analysis. We thus calculate the cost of the average civil war and contrast the estimated cost of each policy instrument with the benefits resulting from a reduction or shortening of wars.

The benefits of reducing the incidence of civil war

We now consider the global benefits resulting from a reduction in the incidence of civil war. Benefits from reducing conflict and war accrue at three distinct levels:
national, regional and global.

Taking the national level first, one clear cost of civil war is a reduction in economic growth. Using a conservative estimate, one year of conflict reduces a country's growth rate by 2.2%. Since, on average, each civil conflict lasts for seven years, the economy will be 15% smaller at the end of the war than if the war had not taken place. During the post-war recovery, even though the economy on average grows at an annual rate of more than 1% above the norm, it will take roughly ten years to return to its pre-war growth rates (that is, 17 years after the conflict started). 21 years after the start of the original war, the GDP has returned to the level it would have achieved if no war had occurred. The total economic cost, expressed as a present value at the start of the war (using a 5% discount rate), is 105% of the GDP at that point. This seems to be a fairly conservative estimate, as a more detailed analysis shows the actual figure to lie in the range 41-305% (90% confidence limits).

The welfare of a country's population is further reduced because of increased military spending during and after the war. It is estimated that military spending increases by about 1.8% during the war, and only falls back by 0.5% once the conflict has ended. Assuming that this higher level of spending lasts for only ten years after the conflict, the additional cost (expressed again as present value when the conflict started) is 18% of GDP.

In addition, conflict has a severe effect on human health. One way of summarizing this effect is to express the cost in terms of Disability Affected Life Years (DALYs): a measure of the total number of people affected and the period for which their disability lasts. An average war causes an estimated 0.5 million DALYs each year. Assuming they decline smoothly to zero in the 21st year and discounting them at 5% as for the direct economic costs gives a figure of 5 million DALYs as the net present value of health costs when hostilities start. If each DALY is valued at $1,000 (roughly the per capita income in many at-risk countries), the economic cost of harm to human health in a typical war is around $5 billion.

At the regional level, analysis shows that the growth rate of neighbouring countries not directly involved in the conflict is reduced by 0.9% during the war. If they subsequently recover at the same rate as the conflict country, the additional cost (as a present value at the start of the conflict) is 43% of initial GDP. On average, each country has 2.7 neighbours, so the direct effect of a typical civil war on neighbouring countries is 115% of the initial GDP of one country: greater than the direct effect in the conflict country itself.

There is also an effect on military spending in adjoining countries: a neighbourhood arms race often ensues. In the average case considered so far, a 1% rise in military expenditure in the country at war would increase the average spending of bordering countries by 0.23%. In a typical conflict, that means military spending will increase by 0.4% of GDP during the war, and by 0.3% during the post-conflict period: a total net present value of 4.3% of the country's initial GDP. On average, there are 2.7 neighbouring countries; thus the total extra cost of the regional arms race is about 12% of one country's GDP.

Other costs which are too difficult to quantify are incurred both in the country at war and in the region as a whole, including forced migration and increased disease. With the proviso that the figures so far are therefore underestimated to some degree, the total benefit of averting a single "typical" civil war can be calculated. The various national and regional costs covered so far amount to 250% of initial GDP. The average GDP of conflict-affected low-income countries (excluding India and China) just prior to war is $19.7 billion. Therefore, the cost of a single war is around $49 billion. To this we must add $5 billion of health costs, giving a total cost of $54
billion for a single low-income country.

This is already a significant figure, but in addition there is the "conflict trap": countries that have just experienced a civil war are more likely to have further conflict. Looking at the 21 countries in which wars started and ended in the period 1965-99, the risk of conflict over the five years before the war averaged 22.3%, but this rose to 38.6% post-war. Over the 15 year period needed for the risk to reach the pre-war level again, the additional discounted cost is $10.2 billion. Thus the total national and regional cost of a single war is more than $64 billion.

There are additional, global impacts of civil wars, massive in scale but difficult to assign a cost to. Three world scourges over the last 30 years have had civil conflicts as contributory factors: hard drug production, AIDS and international terrorism.

Based on these estimates (and bearing in mind the unquantified but significant global effects of conflict), we can look at the benefits of reducing the number of countries at war at any one time. At present, an average of two civil wars start each year. The net cost of these would be $128 billion. So, an initiative which reduced the chance of a new war by 10%, for example, would generate benefits of around $13 billion each year.

Another class of opportunities are those which shorten wars rather than prevent them. The benefit of shortening the typical war by one year - from seven to six years - turns out to be surprisingly large: nearly $11 billion. For the average of two new conflicts per year, this would generate benefits of about $22 billion annually. There would also be a one-off gain from shortening the present 17 significant civil conflicts by one year. A discounted value of this figure (comparable to the year-on-year benefit) is around $9 billion, so the total annual benefit of shortening civil wars by one year is over $30 billion.

Opportunities for conflict prevention

Analysis shows that a country's political and social characteristics are much less important than economic factors in determining the risk of war. The three most significant characteristics are level of income, its rate of income growth and the degree of dependence upon primary commodity exports.

Of the two wars which on average start annually, 0.7 occur in the poor, relatively small countries with no recent history of war (China and India are excluded). Here we look at the benefits of preventing some of these. Post-conflict situations represent a separate opportunity.

Raising the economic growth rate reduces the risk of conflict both directly and by increasing incomes. As a point of reference, a 1% increase in annual growth rate for a period of ten years was modelled. For the first five years, the model shows a reduction in the risk of war from 13.8% to 12.7%. During the second five-year period, the additional income level effect reduces the risk further, to 12.2%. Assuming that the growth rate returns to normal after the ten year period, the longer term effect of higher income leaves the risk of conflict at 12.7%, as a permanent reduction over the initial level.

In financial terms, the benefit in the first five years is an 8% reduction in the chance of a civil war starting. Since each war costs $64 billion, and on average 0.7 start each year, the annual benefit is $3.6 billion. In the second five-year period, this gain rises to $5.4 billion annually (a 12% reduction in the risk of conflict starting). After ten years, the annual benefit falls back to $3.6 billion. The present value (discounted at 5%) is around $79 billion. To put this large figure in perspective, remember that this is purely a benefit of conflict reduction, over and above the direct benefits of poverty reduction.

Two major policy options might deliver these benefits: aid and better utilization of income from natural resources.

Aid

The effect of aid will depend to a large extent on the political and institutional situation. Additional aid as a means to boost growth is also subject to the law of diminishing returns. On average, for a single country from the group of 32 poor, relatively small and generally peaceful states, an extra 2% in aid would raise their annual growth rates by only 0.2%. The benefits of aid in terms of conflict reduction would therefore be one-fifth of that from the model of 1% additional GDP growth: $16 billion.

At purchasing power parity rates, the combined GDPs of the countries in this group amount to $1,200 billion, so the overall cost of the additional aid amounts to $24 billion annually. The net present value of this over a decade is $195 billion. This indicates that unselective aid programmes are not a cost effective way of reducing conflict: the benefits are less than 10% of the costs. This is not to say that the aid may not be justified; after all, its main purpose is poverty reduction. However, conflict reduction should not be expected to be a major outcome.

Improved governance of income from natural resources

Not only is income from natural resources poorly converted to growth in many conflict-prone countries, but revenue from primary commodities is actually a risk factor in civil war. There are a number of reasons for this, including the fact that valuable resources can encourage regional secessions and provide finance for rebel movements.

There are two ways in which the international community can act collectively to improve this situation: 1. Increase the transparency of revenue streams. A particular campaign, backed by a range of NGOs, the G8 governments and international institutions is known as the Extractive Industries Transparency Initiative (EITA). 2. Encourage greater medium-term revenue smoothing to avoid destabilizing booms and busts.

Such international action would provide a template for government reform, and would give clear guidelines against which poorly-performing governments could be measured. International agreements are difficult to achieve, and only a limited number can be negotiated at one time. There is therefore an opportunity cost of pursuing this particular strategy in preference to some other initiative which could also bring equal or greater benefits.

It should be feasible to halve the adverse effects of poorly managed natural resource incomes (complete elimination may be an overly ambitious target). This would create a knock-on effect: reducing conflicts via increased growth rates. To halve the adverse effects, growth rates would be raised by 0.067%, assumed to be permanent once reforms are in place. The benefit in present value terms would be $12.1 billion.

There is also a much larger benefit realizable. Greater transparency in the revenue flows can reduce regional grievances and the incentives for secessionist groups to take control of the income. If such action reduced the risk from natural resource dependence by 10%, the overall risk of war for a typical country would fall from 13.8% to 12.7%. This would be worth $3.9 billion annually and, assuming it is permanent, gives a present value benefit of $77 billion.

The total benefit from conflict reduction by this route is then $89 billion.

Opportunities for shortening conflicts

Civil wars last typically for seven years. Unfortunately, there is no systematic evidence of the effectiveness of either financial or military intervention by the international community. However, rebel movements tend to be self-financing, so controlling international markets in commodities and armaments could reduce their ability to be self-sustaining.

When the world was alerted to the problem of "conflict diamonds" as a funding source for rebel movements in Africa, a government certification process was put in place. This appears to have been at least partly responsible for the collapse of both the RUF in Sierra Leone and UNITA in Angola, and the concept is now being extended to trade in timber. Certification does not totally exclude rebel groups from the market, but it introduces a two-tier market where uncertified goods are sold at a discount.

In addition to diamonds and timber (and extending this to oil), it is even conceivable to consider creating a two-tier market in hard drugs, 95% of which originate from civil war environments. This would be analogous to the situation in the UK in the 1960s, where registered addicts could obtain drugs legitimately via the country's health service.

In 2002, conflict diamonds were trading at an approximately 10% discount. This figure can be taken as a minimum expectation, since diamonds are difficult to police and proper monitoring was not in place at the time. Using a 10% discount as an estimate for commodities in general, and using a figure of 12% for the conflict shortening effect (derived from estimates of the actual effect of commodity price changes), this would reduce the length of the war by a little under a year. Given the number of countries which significantly depend upon commodity exports, we arrive at a figure of $5.9 billion for the present value of this intervention.

Reducing post-conflict risks

During the first decade following a war, there are very high risks of repeat conflict: around half of all civil wars arise in this way. However, typically there are only about 12 countries in this post-conflict category at any given time, making it relatively easy to direct resources to them. Two policy approaches to reduce post-conflict risks are the use of aid, and military spending.

Post-conflict aid

Countries typically have higher growth rates in the post-conflict decade, but these are neither assured nor evenly spread. Growth is, in fact, driven entirely by the amount of aid, and is concentrated in the middle years of the decade. Although the need is great immediately post-conflict, there is limited capacity to use aid effectively; by the middle years, resources can be managed better and the needs addressed properly. The opportunity identified is then to provide increased aid at the time when it is most useful: the analysis is for an aid increase of 2% of GDP for the middle five years of the decade.

There are typically 12 countries in their first post-conflict decade at any one time, with a combined GDP of about $163 billion. The cost of the additional aid averages $1.6 billion per year over the decade, with a net present value of $13 billion. The gain in growth rate in the years when extra aid is received would be 1.1%, more than five times the increase seen in normal non-conflict situations. The benefits of this targeted aid in terms of avoiding conflict can be calculated as $31.5 billion across the 12 countries. This intervention is clearly cost-effective for the additional security benefits alone, in addition to its main purpose of poverty reduction.

Military expenditure

In a country's vulnerable years immediately post-war, government military spending is maintained at a high level and actually appears to be counter-productive in terms of peace-keeping. There is a case for international military intervention, on condition that the government makes deep cuts in its own military expenditure. Merely by reducing this spending to pre-conflict levels, GDP would increase by 2% in the decade.

Assuming that an international peace-keeping force completely avoids the outbreak of another war during the ten years it is in place, a risk of 38.6% in the first five years and 31.9% in the second is averted. With the average civil war costing $64 billion, the present value of this intervention is $29.9 billion. In addition, a gain of $3.2 billion can be attributed to the reduced risk of war because of reduced military spending and increased GDP, making a total of $33.1 billion.

Costs will vary with the individual country's situation, but as a typical example the current British force in Sierra Leone has cost, on average, $49 million each year. If we assume that such forces will be in place for the full ten years, the net present value comes to only $397 million. The benefit/cost ratio is therefore enormous: an expenditure of less than half a billion dollars securing a benefit of $30 billion. Extending this approach to all 12 typical post-conflict situations would deliver a benefit of $397 billion for a cost of less than $5 billion.

Conclusion and comparison

All of the above estimates in are gross approximations. Nevertheless, they enable us to distinguish between those policies offering very high rates of return and those which are uneconomic. Clearly, the option of international military intervention under Chapter VII of the United Nations Charter offers enormous benefits, but is also the most difficult politically. At the other end of the scale, aid has very limited effectiveness in conflict prevention unless it is much better targeted. Reforms of the commodity trading system fall between the extremes, offering significant benefits at reasonable cost.

References:

P. Collier and A. Hoeffler. 2004. Conflicts. in: Lomborg,
B. (ed) Global Crises, Global Solutions. Cambridge
University Press, forthcoming.
P. Collier and A. Hoeffler. 2004. Greed and Grievance in
Civil Wars. Oxford Economic Papers, forthcoming.
P. Collier and A. Hoeffler. 2004. Aid, Policy and Growth in
Post-Conflict Countries. The European Economic Review 48:
1125-1145.
P. Collier and A. Hoeffler. 2004. Unintended Consequences:
Does Aid Promote Arms Races? Mimeo.
P. Collier and A. Hoeffler. 2004. Military Expenditure in
Post-Conflict Societies. Mimeo.
P. Collier and A. Hoeffler. 2004. Murder by Numbers: On the
Socio-Economic Determinants of Homicide and Civil War.
Centre for the Study of African Economies WPS 2004/10.

* Co-author: Anke Hoeffler, Research Officer, Centre for the Study of African Economies (CSAE), Oxford University, Oxford, United Kingdom.

Top of page

Contents	Speeches/Issues Notes	EF Homepage
Preface	Programme	EF Texts and Reports
Summary of Discussions	List of Participants