Going in When it Counts: Military Intervention and the Outcome of Civil Conflicts∗

Conventional wisdom suggests that biased military interventions in civil conflicts should increase the probability that the supported side will win. However, while this is the case for rebel groups, the same is not true for governments. The explanation for this surprising finding becomes clear once one considers the decision of a third party intervener. Since interveners want to impact the outcomes of civil conflict, governmentand rebel-biased interventions will be more likely when the government is facing a stronger rebel group. Given that governmentbiased third parties intervene in the “toughest” cases, empirically they appear to be less effective than rebel-biased interveners. In recent years, there has been great scholarly interest in the effects of military interventions in civil conflicts, particularly the impact that interventions have on the duration of civil conflicts. The conventional wisdom drawn from the literature appears to be that military interventions increase the duration of civil conflicts. However, a closer look at previous studies indicates that there is no consensus on the relationship between military intervention and civil war duration. In fact, only one published study finds a positive relationship between biased military intervention and civil war duration (Balch-Lindsay and Enterline 2000). On the other hand, Collier, Hoeffler, and Söderbom (2004) find that rebelbiased military interventions decrease the duration of civil wars, while government-biased interventions have no significant effect. Finally, Regan (2002) and Regan and Aydin (2006) find that biased military interventions have no statistically significant effect on the duration of intrastate conflicts. This lack of empirical consensus concerning military intervention and civil war duration is likely due to the fact that military interventions do not primarily affect the duration of civil wars. Instead, the primary effect of an intervention is on the outcome of a civil war. When a third party intervenes militarily to support one side in a civil conflict, it provides additional military support for the targeted side. This additional support shifts the balance of power more in favor of the targeted side. On average, then, biased military interventions should increase the probability that the targeted side will win a conflict. Third parties, however, do not randomly intervene in civil conflicts. Since potential interveners care about the outcome of a civil conflict, and intervention is costly, they should be more likely to intervene militarily when it would have the greatest marginal effect on producing a more preferred outcome. The potential effect of a military intervention depends, in part, on the characteristics of the domestic groups engaged in a civil war. Previous studies indicate that additional military resources have the greatest impact on battlefield success when neither side has a significant advantage in capabilities over the other (Dupuy 1987; Hirshliefer 1989, 2000). Civil wars differ from international wars in that they are often asymmetrical (Zartman 1995). Governments generally have structural advantages unavailable to rebel groups. Thus, rebel groups are rarely significantly stronger than their government opponents. Given this, the effectiveness of providing additional resources in a civil war is greatest in cases with strong rebel groups. If potential interveners care primarily about influencing civil war outcomes, military intervention should be most likely when governments face strong rebel groups. Consider an intervener biased in favor of the government. While governments can often defeat weak rebel groups with or without outside help, strong rebel groups provide a legitimate threat to the survival of the government. Thus, intervention on behalf of the government can have a greater effect on shifting the outcome of the civil conflict toward government victory when there is a stronger rebel group. On the other hand, consider the case of a rebelbiased intervener. In general, weak rebel groups are unlikely to be able to achieve military victory over government forces on the battlefield whether or not they receive outside military support. Instead, outside military support will be more efficiently used when there is a strong rebel group that is more effective at fighting the government. Since intervention is more likely when rebel groups are strong, government-biased interventions may appear to be less effective than rebel-biased interventions. However, this does not necessarily imply that government-biased intervention has no effect. Instead, interveners are strategically choosing to intervene in the “toughest” cases: those in which the rebels have a high level of military capabilities and are likely to be effective at waging war. On the other hand, rebel-biased interveners choose to intervene in cases where rebel groups have the highest likelihood of winning, which reinforces the positive effect of intervention on the probability of rebel victory. The theoretical argument outlined above incorporates strategic decision making. Given this, game theory provides a useful method to ensure both the logical consistency of the argument and identify other empirical implications of the theoretical argument. In the analysis that follows, I develop a game theoretic model of third party intervention in a civil