The Logic of Warlord Politics

Abstract

This article offers an initial framework for a future theory of warlord politics using a simple game theoretic approach. We address the topic abstractly rather than empirically to develop a set of testable propositions across the wide range of warlord regimes. We discuss the reciprocal military and economic foundations of warlord domination, the structural logic of warlord politics, the stability of warlord regimes, and the circumstances that can be expected to lead to the formation and dissolution of warlord coalitions. We offer a conceptual introduction to these topics and lay the groundwork for a more systematic treatment of these and related themes in subsequent work.

Notes

1 Some exceptions include P Jackson, ‘Warlords as alternative forms of governance’, Small Wars and Insurgencies, 14 (2), 2003, pp 131–150; A Giustozzi, ‘The debate on warlordism: the importance of military legitimacy’, Crisis States Discussion Paper, 13, Crisis States Development Research Centre, London School of Economics, October 2005; A Vici, ‘Conceptualizing warlords as sovereign non-state actors’, London School of Economics, 2006; and Vici, Armed Groups and the Balance of Power, London: Routledge, 2009. Some of the best work on this subject area to date has not been focused specifically on warlords per se but on the more general topic of predatory regimes and state formation. See, for example, the important work by M Levi, Of Rule and Revenue, Berkeley, CA: University of California Press, 1988; and Levi, ‘A theory of predatory rule’, Politics and Society, 10 (4), 1981. See also R Wintrobe, The Political Economy of Dictatorship, Cambridge: Cambridge University Press, 1998.

2 Co-operative behaviour, it has been argued, is wealth-producing and therefore generates a higher net social product than conflict behaviour, which is wealth-reducing. Similarly, co-operation is initiated to generate wealth, conflict is initiated to redistribute wealth (or protect it from being redistributed by others). For a discussion, see CK Rowley (ed), The Selected Works of Gordon Tullock, Vol 8, The Social Dilemma: Of Autocracy, Revolution, Coup d'Etat, and War, Indianapolis, IN: Liberty Fund, 2005, pp 1–8.

3 M Shubik, ‘Game theory: the language of strategy’, in Shubik (ed), Mathematics of Conflict, Amsterdam: North-Holland, 1983, p 3.

4 RH Jackson, Quasi-states: Sovereignty, International Relations and the Third World, Cambridge: Cambridge University Press, 1990, pp 50–51.

5 Sam Bockarie was the notorious military commander of the Revolutionary United Front of Sierra Leone. He was killed in 2003 while ‘resisting arrest’. At the time of his death he was under indictment by a special court for alleged crimes against humanity. When asked what motivated him, he replied that ‘I never wanted myself to be overlooked by my fellow men’. In another interview he proclaimed that ‘I really admire myself’ and that ‘even now’, after many years in the bush, ‘I can do any hair style’. He adopted the nickname ‘Mosquito’ because of his self-proclaimed ability to strike his enemies without warning. See A Toweh, ‘Indicted West African warlord killed in Liberia’, Global Policy Forum, 6 May 2003; and S Coll, ‘The other war’, Washington Post Magazine, at Washington Post.com, accessed 8 November 2008.

6 H Chi, The Chinese Warlord System: 1916 to 1928, Washington, DC: American University Center for Research in Social Systems, February 1969, p 24.

7 General Mohammed Farrah Aideed rose through the military ranks of Mohammed Said Barre's regime before overthrowing him in 1991. Aideed became one of the many de facto faction leaders during the ensuing civil war, fighting competing strong men and impeding international peacekeeping forces throughout the early 1990s. In 1995 Aideed declared himself the president of Somalia. Others, needless to say, disagreed and fighting continued. Aideed was finally taken out of our misery by gun fire on 1 August 1996.

8 Khun Sa began his military career with the Chinese Kuomintang forces in the 1940s. Following the communist victory in 1949, he crossed into Burma with the remnants of his forces to enter the opium trade. Throughout the 1960s he accumulated forces and shifted armies to suit his economic interests. In the 1970s, Khun Sa set up a base in northern Burma and established the Shan United Army. He maintained a vast drug-trafficking operation in the Golden Triangle until the mid-1990s, when he fell out with other Shan leaders. In 1996 Khun Sa surrendered to the Burmese government but was never extradited to the USA. He lived under the protection (and close eye) of the Burmese authorities in Rangoon, where he died at the age of 73 on 26 October 2007.

9 BR Rubin, The Fragmentation of Afghanistan: State Formation and Collapse in the International System, New Haven, CT: Yale University Press, 1995.

10 Dostum was subsequently displaced by the Taliban regime and withdrew with the remnants of his forces into the mountains of northern Afghanistan, where he became a minority member of the Northern Alliance. In October 2001 US Special Forces established contact with him and recruited him to join their efforts to attack the Taliban stronghold in Mazar-i-Sharif. This attack, which took place in late November, resulted in a set of cascading Taliban defeats that resulted in the collapse of the regime in early December. US Special Forces sealed the deal with Dostum by giving him a large cardboard container filled with $1 million in cash, a box of condoms (super-sized), and a bottle of Johnny Walker. Looking into the box, Dostum turned to his contacts and declared: ‘We are friends’. Author interviews, Afghanistan, Winter 2002. Like many of his warlord counterparts, Dostum was given a high-ranking position in the Karzai government (Deputy Defense Minister) in an effort to make him part of the solution rather than part of the problem. He has returned to Mazar-i-Sharif, where he is suffering from syphilis, ergo the condoms.

11 F Bridgland, Jonas Savimbi: A Key to Africa, New York: Paragon House, 1986, p 66.

12 K Stuvoy, War Economy and the Social Order of Insurgencies: An Analysis of the Internal Structure of unita's War Economy, 3, Hamburg: Universität Hamburg, Institut für Politische Wissenschaft (Institute for Political Science), 2002, p 72.

13 Philippe Le Billon identifies several types of war economies which are not mutually exclusive: state, guerrilla, predatory, foreign alliance, humanitarian and commerce-driven. P Le Billon, The Political Economy of War: What Relief Agencies Need to Know, London: Humanitarian Practice Network, July 2000.

14 See, for example, ibid. For a more general discussion, see Levi, Of Rule and Revenue, passim.

15 P Johnston, ‘Timber booms, state busts: the political economy of Liberian timber’, Review of African Political Economy, 31 (101), 2004, p 446.

16 W Reno, paper presented at unu/wider–Queen Elizabeth House, Oxford, meeting on ‘The Political Economy of Humanitarian Emergencies’, October 1996, cited in M Duffield, ‘Globalization, transborder trade, and war economies’, in M Berdal & DM Malone (eds), Greed and Grievance: Economic Agendas in Civil Wars, Boulder, CO: Lynne Rienner, 2000, p 84.

17 Coalition for International Justice, Following Taylor's Money: A Path of War and Destruction, Washington, DC, May 2005, pp 17–18.

18 Chi, The Chinese Warlord System, pp 75–76.

19 Ibid.

20 Osman Ali Atto was already a prominent businessman in the oil and construction industries when civil war broke out in Somalia in the late 1980s. He was one of General Mohammed Aideed's most important financiers until 1995, when he led a factional split from Aideed. Fighting continued with Aideed's forces and other militias throughout the 1990s. In 2001 Ali Atto participated in peace talks with the transitional government, leading to his official support of the government in 2002. In 2004 Osman Ali Atto was appointed minister of housing and public works in Somalia's Transitional Federal Government (tfg), but he resigned in 2006. He remains involved in the khat trade and continues to be a political player.

21 Economist Intelligence Unit (eiu), Somalia Politics: Narcotic Trade Helps to Prolong Unrest, London: eiu, 15 July 2003.

22 DC McNeil, Jr, ‘Lawless land’, New York Times Upfront, 25 March 2002.

23 C Webersik, ‘Fighting for the plenty: the banana trade in southern Somalia’, Oxford Development Studies, 33 (1), 2005, p 87.

24 Chi, The Chinese Warlord System, p 76.

25 Hussein Aideed succeeded his father, the late General Mohammed Farah, after his death in 1996. The younger Aideed began his military career as a US marine in 1987. He served as a translator in Somalia during Operation Restore Hope and became a naturalised citizen of the USA, where he remained until the age of 30. Following his father's death, Hussein became leader of the Somali National Alliance (sna). Throughout the 1990s, Hussein Aideed was involved with various peace processes, and equally involved with interspersed outbreaks of factional fighting when these processes did not go his way. He has spent the past several years serving in cabinet posts within the tfg.

26 S Grosse-Kettler, External Actors in Stateless Somalia, Bonn, Germany: Bonn International Center for Conversion, 2004, p 22.

27 United Nations, ‘Report of the Panel of Experts on Somalia pursuant to Security Council Resolution 1474’, New York, 2003, p 40.

28 S Ellis, The Mask of Anarchy, London: Hurst and Company, 1999, p 164.

29 Jackson, ‘Warlords as alternative forms of governance’, p 140.

30 Testimony of William H Twaddell, Acting Assistant Secretary of State for African Affairs, before the House International Relations Committee, 26 June 1996.

31 H French, ‘Don't send in the Foreign Legion!’, New York Times, 30 October 1994, cited in W Reno, ‘War, markets, and the reconfiguration of West Africa's weak states’, Comparative Politics, 29 (4), 1997, p 500.

32 Padsha Khan Zadran is an example of a local strongman who was able to achieve his position without a prior military background. In contrast to the warlord's typical path to glory, he first gained a resource advantage over his competitors and subsequently translated this into a local force advantage, rather than the reverse. He began his career as a long-haul truck driver, which he was able to parlay into ownership of a fleet of trucks, eventually gaining a position of economic status within his traditional clan area in and around the strategic town of Khost. PK's economic assets permitted him to begin to recruit and equip a militia force which gave him the coercive advantage he needed to control the Khost bowl at its approaches. While his wings have been clipped, he remains in play as a challenge to the Karzai government and a well known local troublemaker. He was given the soubriquet, ‘PK, to distinguish him from two of his regional rivals, Kamel Khan, KK, and Zia Khan, ZK. Collectively, PK, KK, and ZK controlled much of southeastern Afghanistan in the aftermath of the Taliban collapse. KK and ZK remain influential.

33 A Giustozzi, ‘Respectable warlords? The transition from war of all against all to peaceful competition in Afghanistan’, paper presented at the ‘Crisis States Programme’ seminar, Development Studies Institute, London School of Economics, 29 January 2003, p 9.

34 A similar argument is made by Tullock, in Rowley (ed), The Social Dilemma, pp 19–21.

35 The warlord game can also be contrasted with that of a ‘benign dictatorship’, shown in the matrix below. Note, here, that one but not both of the players has an incentive to defect from a co-operative strategy and become Mr Big. For the second player, ‘yield’ is a dominant strategy. Not only is he better off yielding in lieu of a fight, he would always prefer to serve than command, even if the power player is willing to step aside in the face of a challenge. It does not matter to the outcome of this game if the benign dictator is a strong player (who would always prefer to fight than yield) or a weak player (who would always prefer to yield than fight). The fact that the other player in the game always prefers subordination to domination ensures the same outcome in either case. The equilibrium in this game, as we can see, is then YC.

On first appearances this may not look any different from a warlord outcome; one player after all dominates the other. The structure of the game in each case, however, is different, resulting in a similar form of governance, but with very different social consequences. If we look again at the structure of the warlord game we see that force (or the threat of force) is the arbiter of success. The dominant player, in every case, is the one who is ultimately able to prevail over his competitor on the battlefield. This is clearly revealed in Case I, it explains why the outcome of the game is indeterminate in Case II without a first move advantage, and it explains why neither side can prevail against the other in Case III. This is not the basis of authority in the game of benign dictatorship. While the dominant player in this game may or may not prefer to fight than yield, his authority is not ultimately a function of his battlefield advantage, but the conferred legitimacy of his rule. He governs by the consensus of the ruled rather than at the point of a gun. Although the subordinate player, in this variation of benign dictatorship, prefers strategy set YY to YC, he also prefers YC to CY. It is this feature of benign dictatorship that distinguishes it from the structure of warlord politics. Clearly, the warlord game and the game of benign dictatorship offer the subordinate players a choice; they can submit or they can choose to fight. The nature of the choice, however, is subtly different in each case. In the first case, the yielding player chooses a strategy of deference because of the costs of not doing so; in the second case he submits himself to the power player by preference.

36 The reader should note that as long as the subordinate remains a ‘weak’ player, the value of S will always be greater than F. D is also clearly greater than F. Thus, as long as the would-be challenger is weak, which is the only condition under which his decision to challenge turns on his estimate of the warlord's type, the fraction to the right of the inequality is positive. As long as D is greater than S, furthermore, which is true by definition in the warlord game, the value of the fraction will always be less than 1. This expression, therefore, poses a realistic constraint on α, which must then be between 0 and 1. In the unlikely event that the subordinate player has become a ‘strong’ player, he has a dominate strategy, which is to fight. While his payoff, in these circumstances, depends on the warlord's type, his decision to challenge the warlord does not; it is always his best course of action. In this case F is greater than S (but still not greater than D) and the fraction is negative. Note that, even if there is no chance that the warlord is a weak player and will back down to a challenge, ie α = 0, it is still greater than a negative value and the subordinate player is better off acting. This shows that acting is a dominant strategy. What if S is greater than D? In this case, as noted in an earlier explanatory note, the players are no longer in the warlord game, but in a benign dictatorship. The subordinate player has no desire to make a bid for power because he would rather serve than rule. In this case the value of the fraction to the right of the inequality is always greater than 1. Since α cannot be greater than 1, there is no feasible value of α that will satisfy the criterion to act. Even if it is known with certainty that the warlord is weak and will back down to a challenge, ie α = 1, the subordinate player will not act. Subordination is, thus, a dominant strategy.

37 See, for example, T Caplow, Two Against One: Coalitions in a Triad, Englewood Cliffs, NJ: Prentice-Hall, 1968; Caplow, ‘A theory of coalitions in a triad’, American Sociological Review, 21 (4), 1956, pp 489–493; and Caplow, ‘Further development of a theory of coalitions in a triad’, American Journal of Sociology, 64 (5), 1959, pp 488–493. Subsequent work, building on Caplow's findings, included J Chertkoff, ‘A revision of Caplow's coalition theory’, Journal of Experimental Social Psychology, 3, 1967, pp 172–177; WA Gamson, ‘A theory of coalition formation’, American Sociological Review, 26 (3), 1961, pp 373–382; and Gamson, ‘An experimental test of a theory of coalition formation’, American Sociological Review, 26 (4), 1961, pp 565–573. While our own discussion on warlord coalitions is developed in the spirit of this early work on coalitions in a triad, the specific set of rules that guides the formation of warlord coalitions is distinct from those advanced by these authors and those that followed them.

38 Thus ‘major’ and ‘minor’ are contextual rather than objective designations. A warlord is a major or minor player in relation to his competition. This is similar to the earlier distinction we made between ‘strong’ and ‘weak’ players. A player is ‘strong’, we said, if he would rather fight than yield and ‘weak’ if he would rather yield than fight. In every case, however, it is a question of one's willingness to fight or yield in relation to a particular opponent. A ‘strong’ player in one set of circumstances might be a ‘weak player in another. The same is true for ‘major’ and ‘minor’ warlords.