FORCE AND BIAS: TOWARDS A PREDICTIVE MODEL OF EFFECTIVE THIRD‐PARTY INTERVENTION

Abstract

Throughout the 1990s multilateral interventions often deviated significantly from traditional peacekeeping in terms of mandate complexity, level of force, and the absence of consent and impartiality. This paper develops a formal model of biased intervention and specifies propositions regarding its effects on combatant behaviour. We find that the response to the intervener depends on the how the combatants divide their labour resources between production and fighting, the amount of resources the intervener transfers between the combatants, and the degree to which the intervener’s military efforts affect the effectiveness of the combatant’s military forces. NATO’s intervention in Kosovo is then used as a case study to illustrate the strengths and weaknesses of the model. Finally, we identify the theoretical and policy‐relevant implications of the model and outline directions for future research.

Keywords:

• Peacekeeping
• Conflict intervention
• Biased intervention
• Third‐party interveners

Notes

¹ Today, the collective management of complex intrastate conflicts tends to unfold selectively, when those who have the capability to respond also have the motivation to do so (Rotberg and Albaugh 1999). Rothchild and Lake (1998) see an emerging norm of collective intervention in a wide range of situations, including genocide, interference with the delivery of relief, violation of ceasefire agreements, collapse of civil order, and irregular interruption of democratic governance.

² Kleiboer (1996) argues that modeling produces an incentive for more theory‐driven empirical research, which, by her assessment, is more informative for the researcher and the policy maker.

³ Saideman (1997) and Davis et al. (1997) examine the role of partisan states in ethnic conflict, emphasizing the pressing sense of obligation to support a minority brethren’s claims. The consequent coalitions and linkages are often steeped with powerful affective elements, including regional identity and religious affinities (Heraclides 1991; Suhrke and Noble 1977).

⁴ For a complete assessment of the role of bias in mediation see Kydd (2000). See also Weeton and Pruitt (1987) and Kressel and Pruitt (1989). Carnevale and Arad (1996) provide useful definitions that distinguish between content bias and source bias in mediation.

⁵ Though Roberts (1997) sees ‘loss of consent’ problematic for more peaceful missions, as in the UNTAC mission to Cambodia from 1991–1993.

⁶ Betts uses the examples of UN intervention in Bosnia, US intervention in Somalia, and US and UN intervention in Haiti to underscore his argument. He argues that impartial intervention can work in more limited instances – such as the ceasefire mediation between Iran and Iraq and the political receivership of the UN Transitional Authority in Cambodia ‐ but that these instances prove only that impartiality works best where intervention is needed least, where wars have ‘burned out’ and the fighting factions need only the good offices of mediators to end the fighting formally.

⁷ For Zartman and Touval (1996) power is the basis for this process. Power translates into leverage in the form of persuasion, extraction (producing a favourable outcome for each party); termination (the ability to withdraw from a negotiation); deprivation (the ability to affect a hurting stalemate by withholding resources from one side or shift them to another); and gratification (the ability to add resources to the outcome). They emphasize that interveners make as much of a calculation based on interest in deciding to mediate as is the case for adversarial parties when deciding to engage in war.

⁸ Combatants could also engage in pre‐intervention strategies to avoid, delay, accelerate or encourage an intervention, or to influence its size, composition, goals or direction of bias. Unless such strategies involved commitments of an irreversible nature, however, their direct effect on the interactions considered here is limited once the nature of the intervention has been determined.

⁹ In this regard Hirshleifer (1988: 228) cites Clausewitz’s dictum that ‘War is merely the continuation of policy by other means’.

¹⁰ The model also shows that interventions can also affect fighting effort by altering the marginal product of labour in productive activities. Increasing the marginal productivity of labour in production would generally make a shift away from fighting more desirable. Further analysis of this element may prove interesting for generating economic policies to complement military intervention.

¹¹ While FRY forces claimed to have ‘destroyed the core’ of the KLA at this time, more independent reports noted that the actions appeared to have ‘swelled their ranks beyond belief’ (Sebak, 1998). Judah (1999a) noted that ‘following violent events in the Drenica region [in late winter/early spring 1998] the KLA suddenly found itself in command of an uprising.’ Others concurred: ‘no one disputes the fact that since the first Serbian operation against them in March, the KLA’s influence has grown’ (Lungescu, 1998). British Foreign Secretary Robin Cook was quoted as saying that Milosevic’s tactics had backfired, leaving the KLA in control of more territory, with greater strength and more money (Lungescu, 1998). KLA military and political strength, supported by the Albanian diaspora, often expanded in response to atrocities attributed to Serb security forces.

¹² Certainly, reports of the order in which Serb forces attacked and cleared areas and cities seem to indicate a decisive plan aimed at clearing out the province of Kosovo: ‘Western officials now believe that the aim of attacking in the south and west [in mid‐March] was to clear a path to the border for a mass expulsion. ‘It was very intentionally done to drive large masses of Kosovars,’ said a NATO official’ (Smith and Drozdiak, 1999). It is noted by Hedges (1999: 25) that the Serbian ethnic cleansing and aggression in Kosovo was ‘to a large degree tactical, designed to deny the rebels succour from civilians and therefore aimed primarily at the inhabitants of KLA strongholds.

¹³ Support for the KLA continued to grow amongst the Albanian diaspora over the period of negotiations, and open fund‐raising in the United States appeared to many to have the tacit approval of Washington (Leyne, 1999). The new arms and money reaching the KLA did ‘not change the military balance [with amassing Serb forces clearly outpowering the ethnic Albanians], but they did give the KLA leadership the self‐confidence to play tough in the peace talks’ (Leyne, 1999). Just prior to the bombing the KLA was estimated to number from 7000 to 8000 people (ICG, 1999c).

¹⁴ Formally this function is also assumed to be smooth and continuously twice‐differentiable to permit subsequent mathematical manipulations. In this paper, the variance of the income is not examined directly; variance, risk attitudes and intervention are examined in Rowlands and Carment (1998).

¹⁵ The diagram resembles Figure 3 in Hirshleifer (1995) and Figure 1 in Becker (1983).