Portal de Periódicos da CAPES

SPENDING TO SAVE? THE COST‐EFFECTIVENESS OF CONFLICT PREVENTION

2007; Taylor & Francis; Volume: 18; Issue: 1 Linguagem: Inglês

10.1080/10242690600821693

1476-8267

Malcolm Chalmers,

Politics, Economics, and Education Policy

Abstract While the general argument that it is easier and more cost‐effective to prevent conflicts before the outbreak of violence has considerable attraction, a rigorous approach to estimating the cost and benefits of this policy is still lacking. The objective of this study is to contribute to the development of such an approach. The project involves six case studies, three retrospective (the Western Balkans, Afghanistan, and Rwanda) and three prospective (Afghanistan, Uzbekistan and southern Sudan). Its main conclusion is that targeted programmes of conflict prevention are (or would have been) significantly cheaper than cure. Keywords: Conflictconflict preventioncosts of conflictfragile statescivil war ACKNOWLEDGEMENTS The author would like to acknowledge the assistance of Greg Austin, Emery Brusset, Jeremy McKenna, David Newton, Hooman Peimani, Stefaan Smis, Saskia Van Hoyweghen, and Susan Willett in assisting with the project on which this article is based. The author is also grateful to Ron Smith, Andrew Mack, Duncan Overfield and Malcolm Smart for their comments, as well as to participants at seminars at the University of Warwick, University of Oxford, Cabinet Office, MoD, FCO, and DFID. We acknowledge the funding from the UK Department for International Development that made this research possible. Notes 1 For example, the UK Government has argued that 'preventing conflict is better and more cost‐effective than resolving it' (Foreign and Commonwealth Office, Citation2003, p. 3). 2 Collier et al. (Citation2003) and Collier and Hoeffler (Citation2004) have used global econometric tools to examine the relative cost‐effectiveness of different conflict prevention instruments. Brown and Rosecrance (Citation1999) use a methodology based on retrospective case studies, but it is different in important respects from that employed in this study. 3 These are available on the web at www.brad.ac.uk/acad/cics/publications/spending/. 4 Detailed assumptions are available on http://www.brad.ac.uk/acad/cics/publications/spending/ 5 This is in accordance with recommended UK Government practice for the appraisal of public investments (HM Treasury, Citation2003, Annex 6). 6 In principle, it would be possible to apply a different discount rate for directly‐affected territories and neighbours, compared with the international community. (For example, because one might expect higher real discount rates in countries with higher rates of consumption growth and/or lower life expectancy.) This is unnecessarily complicating for a study of this nature, and a uniform discount rate for all actors is used. 7 For the purposes of data comparability, we use the widely‐accepted definition of 'major armed conflict', which refers to all armed conflicts in which at least one of the actors is a government, in which there are at least 1000 battle‐related deaths in total, and in which there at least 25 battle‐related deaths per annum. Armed Conflict Database (2003). This means that large‐scale inter‐ethnic fighting in which the state is not involved (e.g. in Nigeria in the 1990s) is not coded as conflict. In many cases, however, the state may be thought to have played some indirect role in such fighting, raising questions about the correct coding. 8 Data in Collier et al. (Citation2003) do not include inter‐state wars, of which there were 28 during 1960–99, primarily in the period before 1990. Nor does it include six wars against colonial rulers that took place in the period before 1980. (Armed Conflict Database, 2003). 9 Five out of seven LMC's with GNI per capita of $2000–2935 (the upper band of LMC status) experienced a major conflict in the last 15 years. 10 No reduction in the probability of the 2003 Iraq war is assumed, although it could be argued that the events of September 11, 2001 made the US decision to attack Iraq significantly more likely. 11 By comparison, 23% of the costs to Rwanda itself relate to the civil war. 12 The first CP Package would, however, have been cost‐effective if the costs to Rwanda itself are taken into account. 13 The literature on costs of terrorism includes Gupta et al. (Citation2002), Joint Economic Committee (Citation2002), Lenain et al. (Citation2002). On the costs of oil supply disruptions, see US Energy Information Administration (Citation2002) and International Energy Agency (Citation2004). On the international costs of narcotics trafficking, see UN International Drug Control Programme (Citation1997). 14 This methodology is similar to that used in Collier and Hoeffler (Citation2004). The use of GDP loss also helps to avoid potential problems related to the proper valuation of capital losses. The current economic value of capital assets is the NPV of the income flows that they will generate in future. In practice, however, estimates of asset destruction often use historic cost valuations, which overstate potential income flows in cases where assets would have soon become obsolete, even in the absence of conflict. Most of Eastern Europe and the former Soviet Union suffered a sharp GDP decline in the early 1990s as the peaceful transition to a market economy destroyed the value of assets designed for a different pricing structure. In order to estimate the economic loss incurred by conflict in the Balkans case study, therefore, it was necessary to assume that it would have incurred significant GDP losses even in the absence of conflict as a result of the transition to a market economy, and the related obsolescence of much industrial capital. The costs of conflict were then estimated as the difference between this peacetime GDP trend and the actual war‐affected GDP levels. 15 Estimating the effects of conflict on growth rates in neighbouring countries is especially difficult, and it is possible that the studies may have erred on the side of underestimation in this respect. For example, we have not included the impact of a renewal of conflict in southern Sudan on both the economic development and potential for internal conflict in Uganda. Collier and Hoeffler (Citation2004, p. 133), estimate that the growth loss to neighbours exceeds the loss to the country itself. If our chosen case studies are reasonably representative, therefore, we may have underestimated the costs of conflict to neighbouring states. See also Murdoch and Sandler (Citation2002). 16 One possible way to compensate for this 'middle‐income bias' would be to give a higher weight to absolute GDP benefits in low‐income countries, compared with those in middle‐income countries, based on the assumption that marginal utility is inversely proportional to average income (Cowell and Gardiner, Citation1999). This would mean that a $127 (or 10%) reduction in an average individual's annual income in Bosnia would be valued as equivalent to a $23 (or 10%) reduction in Rwanda. We have not made this adjustment in Table VI because it would require a decision to be made as to whether to adjust Bosnian GDP figures to a Rwanda baseline, or vice versa. In principle, one could apply marginal utility theory to the world as a whole, in which case a loss of $1 in per capita income in Rwanda would be worth 1000 times a similar loss in the US. In humanitarian terms, this may be a reasonable assumption. Since our studies show a high degree of cost‐effectiveness for conflict prevention when costs to directly affected territories and neighbouring states are included at current prices, the benefits of introducing extra weighting based on utility theory had to be balanced against the extra complexity involved. A similar argument applies to the use of purchasing power parity (PPP) figures for GDP losses, which would also have the effect of increasing the total costs of conflict substantially. 17 An example of such a calculation is contained in Arunatilake and Jayasuriya (Citation2001, p. 1494). Also see Ghobarah et al. (Citation2003), Guha‐Sapir and van Panhuis (Citation2002), Stewart et al. (Citation2001). 18 For a useful survey, see Soares (Citation2004). One of the central methodological problems involved is the extent to which deaths and DALYs as a result of conflict or violence should be valued differently from those as a result of other causes, such as communicable disease. See the discussions in Lomborg (Citation2004, pp. 2–6, 108–109, 132–133). One possible way forward would be to differentiate the values of lives and DALYs averted according to the income level of the country concerned (using foregone personal income and/or the amounts that individuals/governments might be imputed as willing to pay to reduce their risks). By valuing the lives of citizens in low‐income countries at levels much lower than in OECD countries, however, such a methodology offends against the ethical solidarity that is at the centre of international concern at conflict deaths. 19 The global number of active armed conflicts was 30 in 2003, compared with 51 in 1991. This is the lowest number of armed conflicts since the early 1970s. Since the number of countries has increased, the probability that a country is conflict‐affected is now the lowest since the early 1950s (Harbom and Wallensteen, Citation2005).