Fiji’s 2006 coup was, in some ways, curious by international standards. It was not driven by poverty or economic backwardness, although that is often a ‘common denominator’ amongst coups. Samuel Huntington’s claim that successful coups are confined to countries with income per capita at less than US$500 scarcely fits the experience of Fiji, where average income is more than four times that. Fiji would also sit awkwardly alongside Samuel Finer’s crude continuum between ‘less developed’ states, which, he states, are particularly vulnerable to military coups, and ‘more mature’ states, which are not. Fiji is the most economically developed of its Pacific neighbours. Yet none of the others – bar perhaps the Solomon Islands – have witnessed coups. The 2006 Fiji coup was not an expression of economic backwardness: It was a coup of the radicals amongst the westernized elite, who sought to superimpose a national consensus upon a divided social order. It was made possible only by the relatively large size and robustness of the military as compared with the fragmentary character of the civilian political order and the strategic weakness of the deposed government. It was a coup of utopians seeking to transcend, rather than mould, social forces that they deemed responsible for long-run ethnic disquiet and poor governance.
In a more familiar fashion, the architects of the coup said that they were inspired by a strong sentiment of anti-corruption, on the pattern of Thailand or, a decade earlier, Pakistan. Those elites and intellectuals in Fiji who identified with the 2006 coup had long professed sympathies for Singapore’s Lee Kwan Yew’s authoritarian modernism, together with his disdain for democracy, which he saw as promoting graft and inefficiency, and as serving as a restraint on development. Yet Bainimarama’s coup showed little sign of turning Fiji into an Asian ‘tiger economy’, and the botched deliberations of the post-coup Fiji Independent Commission Against Corruption discredited the military’s ‘clean up campaign’. The anti-corruption objectives of the coup of the coup gradually faded from the political limelight. What became more important was the attack by the military and the National Council for Building a Better Fiji (NCBBF), on what has been called the ‘fallacy of electoralism’. Their argument was this: Despite regular elections, substantial sections of the population had been excluded from political power; efforts to promote moderation and inter-ethnic conciliation through democratic means had failed; and only a strong-arm regime, capable of withstanding electoral pressures, could promote modernization, development and inter-ethnic harmony.
Fiji’s 2006 coup needs to be seen in the context of a long-running debate about institutional engineering in the wake of agreement over Fiji’s 1997 constitution. Initial hopes that the alternative vote system, agreed as part of that constitution, would promote moderate parties and candidates came to little. The centrist parties obtained a significant share of the vote in 1999 and 2001, but were greatly disadvantaged under the new electoral laws. Fiji’s voting system, which was supposed to have a ‘centripetal’ influence, in fact helped to ‘hollow out’ the centre, leaving Fiji, most starkly after the May 2006 polls, with two large ethnically-based political parties.
Another element of the 1997 constitution had been power-sharing mechanisms. All parties with over 10 per cent of seats in parliament were entitled to proportional representation in cabinet. Those rules did not work effectively after the 2001 election, but, after May 2006, Qarase formed a multiparty cabinet that included nine members of the Fiji Labour Party. Labour leader Mahendra Chaudhry stayed out, a decision that always appeared likely to serve as a death-blow for the new arrangements. When the 2007 budget was to be put to the vote, Chaudhry demanded that his colleagues in cabinet follow the party whip and vote against it, while Qarase insisted that his ministers accept Westminster doctrines of ‘collective responsibility’ and back the budget. Four ministers followed Chaudhry while Krishna Datt and Poseci Bune were expelled from the Labour Party. In the weeks before the coup, then, power-sharing was faltering as a solution to the country’s problems.
Within Fiji, explanations of the 2006 coup often centre on personal antagonisms – evidenced by four years of regular outbursts by the military commander against Qarase and his government. Many people emphasised the commander’s psychological difficulties in handling the trauma of the November 2000 mutiny, which had steeled his resolve to hunt down, court martial and imprison the perpetrators. Although such explanations may tell us something about personal motivations, they explain nothing about why or how the commander was able to obtain the backing of the Military Council or avoid a mutiny of the rank-and-file.
More plausibly, the antagonism between Bainimarama and Qarase can be explained in terms of the dynamics of institutional survival and consolidation. Qarase’s SDL, forged in the midst of the 2000 crisis, saw its objective as being to advance a ‘realistic’ alternative to George Speight, who claimed to have intervened to save indigenous Fijians. The Blueprint for Affirmative Action for Indigenous Fijians and Rotumans became central to the Soqosoqo Duavata ni Lewenivanua (SDL) platform when it went to the election in 2001; and almost all SDL policies from then on aimed at placating indigenous Fijians, and at healing the splits that had emerged amongst Fijian parties in 1999 and made possible Mahendra Chaudhry’s election victory. Qarase returned 40 per cent of state land to indigenous ownership in 2002 – the so-called schedule A and schedule B lands – increasing the indigenous share of all land in Fiji from 83 per cent to 87 per cent. He attempted something similar by promoting the Qoliqoli Bill, which would have returned traditional fishing grounds to indigenous ownership; and the Indigenous Claims Tribunal Bill, which would have established a court where Fijians could seek the return of yet more land, including freehold. He introduced the Promotion of Reconciliation, Tolerance and Unity (RTU) Bill in 2005 as a measure on which his government would stand or fall, unleashing a storm of criticism that the bill was actually aimed at giving coup perpetrators amnesty. His strategy throughout was to build a political base by appealing to Fiji’s majority population, the Fijians, and to pose as their protector.
Conversely, the Republic of Fiji Military Forces (RFMF) faced a major split in 2000, when mutineers attempted to kill Bainimarama and reverse the defeat of Speight’s pro-indigenous coup. Bainimarama survived as military leader but only just, and he now knew who his enemies were. To consolidate control in the country, Bainimarama clamped down on the coup-supporting areas of northern Tailevu and Cakaudrove with a harshness that was not forgotten or forgiven. Within the armed forces, traditionally seen as defending the indigenous Fijians, were officers who remained highly sympathetic to the Speight coup and its ethno-nationalist objectives. The task was to purge them from the force. One way was to demand a pledge of personal allegiance and force the resignation of those who would not give it, as happened in 2004. Another way emerged over time. As Qarase pursued his pro-indigenous political program, Bainimarama opposed it issue by issue, creating a series of controversies that he used as loyalty tests for his officers. Would they back Qarase or would they back him? Each controversy compelled officers to take a stand, and as they did so Bainimarama weeded out potential mutineers. In so doing, Bainimimarama destroyed much of the professionalism in the senior officer corps, which coup-theorists like Samuel Huntington take as the major factor keeping the soldier out of politics.
The clash between the two institutions – government and military – became inevitable after the 2001 election. Qarase could become prime minister only with the support of the Conservative Alliance–Matanitu Vanua (CAMV), the very group of nationalist Fijians who threatened Bainimarama’s control of the RFMF. Bainimarama came to see every pro-indigenous move made by the government through this prism: This was a government, after all, that wanted to topple him as commander and halve the size of the institution he headed.
To begin with, Bainimarama was no friend of Mahendra Chaudhry. He declined to restore Chaudhry’s government after Speight overthrew it. But the two came to share a common enemy in Qarase, especially after the emergence of the RTU Bill. The bill had the advantage for Qarase of rallying indigenous Fijians behind him in numbers sufficient to win the 2006 election – but the disadvantage of uniting everyone else against him, from the FLP and the military to moderate political parties such as the National Federation Party and the National Alliance Party, civil society organisations, the legal profession, Indian religious bodies and the Catholic, Anglican and Seventh-Day Adventist churches. Qarase could win but not in a way that brought general acceptance that he had won legitimately. When Bainimarama threw his counter-punch a few months later, he did so in the knowledge that he too had numerous supporters on his side. Qarase had played polarizing politics, and he paid the price.
There were three broad potential outcomes of Fiji’s third coup.
A first possibility was a long-term consolidation of military authority, and the advent of a strong-arm regime. Might Fiji develop into a ‘praetorian’ or ‘garrison’ state, with the military coming to dominate the political system? After the 2006 coup, soldiers rapidly assumed central positions in the government bureaucracy, following a pattern witnessed after the 1987 coup. Bainimarama became prime minister, thereby annihilating any perception that the ‘interim government’ was anything other than a military puppet. Two former military commanders were given positions in cabinet, and another leading officer became minister of works and transport. Senior and junior officers were deployed across the civil service, an appointment process justified on the grounds of alleged greater efficiency than civilian personnel. Fiji’s Military Council exerted a continuing powerful influence over cabinet, as became clear when it successfully urged the sacking of interim Finance Minister Mahendra Chaudhry in mid-2008. On several occasions, Bainimarama voiced his opposition to any civilian taking the role of prime minister, and said Fiji’s quarrelsome politicians were responsible for ethnic disquiet, endemic corruption and economic weakness.
There are several reasons why a ‘praetorian state’ is an unlikely outcome. Fiji’s coup, like that in Thailand, was justified as a ‘guardian coup’, to use Samuel Huntington’s phrase, aimed at improving public order, cleaning up corruption, promoting economic development and ridding the state of ethnic cleavages. It was not what Huntington once described as a ‘breakthrough’ coup, in which military officers seize power purporting to act as an instrument of some mobilized social class or group. Fiji’s Indians, although mostly supportive, did not mobilize behind the 2006 coup. Nor was this what Huntington called a ‘veto’ coup, where the army conservatively intervenes to forestall some threat of social revolution. Guardian coups purport to be temporary expedients. They may entail a violent clampdown and suppression of dissent, but the army soon returns to barracks – as it did in Fiji in early 2007, with the dismantling of roadblocks and checkpoints. Normally, ‘guardian’ coups entail a lesser risk of violent loss of life than ‘breakthrough’ or, particularly, ‘veto’ coups, but, nevertheless, enhance the likelihood of protracted instability and further coups.
Fiji’s coup was an unlikely catalyst for long-term military control over the state for another reason: It did not initially entail abrogating the constitution and, from the start, its perpetrators justified their usurpation of power only as a temporary expedient. The coalition that came together in support of the coup – Fijian moderates, the FLP, civil society activists, Catholic social justice advocates and others – were unlikely to have rallied easily behind an overthrow of the 1997 constitution. Judges, such as acting Chief Justice Anthony Gates, would have had to discard long-cherished ideals forged in the battle against those in the judiciary who, they felt, had consented to the attempted abrogation of the constitution on 29 May 2000. Preserving the constitution at all costs, and whatever the consequences, became an element in judicial considerations. Gates justified his 9 October 2008 verdict in the Qarase v Bainimarama case by claiming that to have ruled the interim government to have no legal standing would have encouraged military abrogation of the constitution. True, this coalition was just a convenient embellishment for the military commanders, and could have been dispensed with easily enough. But Bainimarama’s government claimed legitimacy through appeals to anti-corruption, multiracialism and, oddly, the rule of law – all of which would have sat awkwardly alongside construction of a garrison state. The Qarase v Bainimarama decision, by upholding the legality of post-December 2006 interim rule by presidential decree, in fact laid open a purportedly ‘legal’ course for the commander to assume extraordinary powers, if he were able – officially as well as unofficially – to capture the office of the president. But, as of late 2008, little suggested the likelihood of a Burma-style military regime, enduring for decades.
A second possibility was a reasonably speedy restoration of some kind of constitutional government, following the pattern after the 2000 coup or, perhaps more plausibly – if in a more protracted and difficult way – after the 1987 coup. The conclusion to the 2000 coup was exceptional and seemed unlikely to be repeated. Ultimately, the RFMF itself defeated George Speight’s insurrection and the law courts restored constitutionality. In the aftermath of the 2006 coup, the RFMF showed fewer open signs of divisions than in 2000, although these likely festered beneath the surface. Leading anti-coup officers were purged prior to the December 2006 takeover, and their replacements owed their positions to Commodore Bainimarama. The courts always seemed unlikely to follow the post-2000 coup trajectory. The 2001 Chandrika Prasad judgement, which overturned the post-George Speight interim regime, was an oddity in international jurisprudence. Worldwide, few coups have been reversed by court judgements. In still fewer cases have governments accepted verdicts from the courts that declared them unconstitutional. Besides, Fiji’s courts were thoroughly transfigured by the 2006 coup in ways that did not happen after 2000. All six of the judges on Fiji’s Court of Appeal resigned, and other judges who accepted appointments or renewal of contracts knew that, in doing so, they were swearing allegiance to the new order. Such appointees are scarcely likely to pass down a ruling that Bainimarama’s government is illegal.
A more plausible parallel is the 1987 coup, a fully-fledged military coup like that of 2006. Five years passed before Fiji returned to elective democracy in 1992. In the early 1990s, the prospects for a return to a more equitable and consensual settlement appeared bleak. The 1990 constitution, the brainchild of that coup, reserved the positions of prime minister and president for indigenous Fijians and gave the indigenous Fijians 37 seats in the 70-member assembly, leaving only 27 seats for the Fiji Indians. Ethnic Fijian paramountcy had been institutionalized and the military appeared to have assumed the mantle of its protector. Few could, at that time, have envisaged that coup leader Sitiveni Rabuka, who became prime minister in 1992, would, by the mid-1990s, embrace a compact with Fiji Indian political leaders. Few back then expected that yet another new constitution, that of 1997, would restore some balance between the Fijians and the Indians, and remove the core ethno-nationalist provisions of the 1990 constitution. Many emphasized the role of external pressure in encouraging Rabuka’s Soqosoqo ni Vakavulewa ni Taukei (SVT) party to accept a constitutional compromise. Ruling without the consent of half of the population also proved an unattractive proposition by the mid-1990s, particularly for those more cosmopolitan ministers who yearned for a more stable and internationally respectable political order. Even with Fijian paramountcy built into the constitutional architecture, Rabuka’s SVT had found itself having to broker deals first with Chaudhry’s Fiji Labour Party and then with Jai Ram Reddy’s National Federation Party.
Similarly, after the 2006 coup, Bainimarama discovered that efforts to legitimize his new order required an awkward accommodation with the victims of his own coup. In deliberations over the ‘People’s Charter’, the absence of endorsement from key ethnic Fijian-backed institutions featured like a gaping sore. Negotiations soon ensued with Methodist leaders and even with Qarase himself, although, predictably, with little initial consequence since Bainimarama wanted abject acquiescence and not compromise. Efforts to change the constitution, in the absence of parliament, were unlikely to be accepted at home or abroad unless these had the endorsement of indigenous Fijian parliamentary leaders. Of course, had there been mass popular support for the coup and had Bainimarama become the demagogue he aspired to be, restoration of constitutional authority would have likely been rapid. The difficulty was that such backing was absent. The reason why Bainimarama reneged on the electoral timetable that was agreed with Pacific Islands Forum leaders was that an election seemed highly unlikely to deliver a government sympathetic to the ideals of the 2006 coup.
Back in the mid-1990s, Rabuka had envisaged that his constitutional deal would get popular endorsement from both ethnic Fijians and Fiji Indians. If he had known that hostile parties, on both sides, would successfully collaborate to defeat his new coalition, would he have embraced that settlement? Might Bainimarama see a post-People’s Charter election, under new voting laws, as a sure route to the triumph of his intended social revolution? Might hubris again turn to ashes, as the party he deposed in 2006 emerged triumphant at the polls? That was also the experience in Thailand, much to the dismay of the Thai army. Electoral reforms, Bainimarama hoped, would avoid such an outcome in Fiji. Yet, in truth, Qarase’s SDL had nothing to fear from the proposed reformed voting laws. Provided they proved able to retain the 80 per cent of the Fijian vote they obtained in 2006, the SDL was likely to win the first post-coup election.
The post-2006 coup interim government always seemed unlikely to be long-term. Although it breached the commitment to an election by March 2009, there seemed little likelihood – as of late 2008 – that it would endure beyond 2010 or 2011. Its ministers were alleged to be unambitious technocrats who would not stand at the next elections. Bainimarama seemed unlikely to be able to repeat Rabuka’s achievement of recycling himself as a civilian prime minister. In any case, he publicly professed a lack of political ambition, even if he evidently warmed to the exercise of power. Divisions between the interim ministers were carefully concealed, but the departures of Bernadette Rounds Ganilau, Poseci Bune and, most importantly, Mahendra Chaudhry and his Fiji Labour Party colleagues testified to the shifting political character of the coup. Would these regular ministerial reshuffles, combined with a malleable political agenda, eventually lead Fiji back to constitutional rule? Would the impossibility of achieving a stable model for post-coup governance in any other way eventually lead the military commander to seek a more robust and consensus-based settlement with those he had overthrown? If so, following the experience of Turkey, Brazil and Chile, it seemed likely that military leaders would seek for themselves a permanent and substantial influence over the governments that replaced them. This surely was what was meant by the passage in the draft People’s Charter that sought to redefine ‘the role of the Republic of Fiji Military Forces to bring it closer to the people’.
A third, and seldom considered, alternative was a continuing cycle of coups, reinforced by the mounting political uncertainties and deepening economic decline that each unleashed. The 2006 coup advertised its merits as a ‘coup to end all coups’, but in reality it introduced a new dimension to the problem of seizures of power in Fiji. In so doing, it further weakened the Fiji state. Previous coups in Fiji had followed a reasonably predictable pattern, one envisaged by the colonial authorities in the dying days of the British administration. Whatever the intent of their leaders, they gained one-sided popular support from those who professed a desire to resist native peoples becoming outnumbered in their own land. Fiji’s Indians might control economic power, so the argument went, but they would be prohibited from acquiring political power. The 2006 coup, unlike those of 1987 and 2000, had no robust and durable mass social base, and, despite vigorous protestations to the contrary, it lacked a coherent social program or philosophy. In line with the bitter ethnic mechanics of Fiji’s previous coups, it initially captured considerable Indian support. But, as the economy collapsed, as food and fuel prices soared, and as the interim government reconfigured itself to adjust to the departure of the Fiji Labour Party, the prospects for longer-term retention of Fiji Indian backing waned. Without that support, the coup would become more narrowly dependent on the force of arms for its survival. In the process, the prospects for some kind of counter-coup increased, as did the dangers of a violent retaliation by those whose authority might be threatened.
‘Guardian’ coups – for example, in Ghana, Nigeria or Uganda – have often led to a recurring cycle of military takeovers, interspersed with phases of weak civilian control. Once the ice is broken, Finer wrote in The Man on Horseback, coups tend to weaken the structures of civilian authority and render more likely a pattern of recurrent coups. Coups spawn counter-coups. Those who acquire power by force tend, internationally, to be more vulnerable to a similar fate than those who gain office by legal means. And if one usurpation breeds another, it opens a Pandora’s box: Each logically becomes more cut-throat and poses more of a ‘life or death’ conundrum for its perpetrators. Fiji’s future coup-makers, should they fail, are unlikely to face merely a prison sentence on Nukulau Island, as did George Speight after 2000. Amnesties, like that obtained by 1987 coup-leader Sitiveni Rabuka, seem less likely to present themselves as viable alternatives for those who fail to firmly establish their control. Prior to the 2006 coup, the military commander held a complacent and naïve view as regards the ease of inverting the political order and mounting his coup. As the realities of the inevitable post-coup legitimacy crisis became clearer, the dangers of failure also became more apparent and, with this, the political order became correspondingly more brittle.
In other words, the three alternatives considered above are not mutually exclusive, but any movement in one direction may reinforce the possibility of another. The avenue of reverting to constitutional rule and consensus may threaten newly acquired political powers, leading to the type of situation witnessed in September 1987, when the military intervened a second time to derail the Deuba talks and establish a strong-arm regime. The potential for counter-coups itself generates fuel for coup-instigators who feel safer under continuing authoritarian rule. Those incentives might logically lead historians of the coup d’état to imagine a global future of increasing coups, clampdowns and the spread of garrison states, as anticipated by Finer in The Man on Horseback. But that prediction proved false, judging by the diminishing global frequency of successful coups and the global spread of a ‘third wave’ of democratization. The expectation of a coup-ridden global future proved wrong because it underestimated the central dilemma encountered by military regimes; difficulties associated with constructing consensus and establishing a legitimate and durable political order.
Those worldwide trends of the ‘third wave’ provide no cause for complacency in Fiji, which, after all, has departed from international patterns in numerous respects. The 2006 coup, more than its predecessors, introduced new and dangerous potential catalysts for future instability. Before that coup, it was possible to imagine that the coup phenomenon was dying out. First, despite some glaring exceptions, the core 2000 coup leaders – George Speight and his associates – had been sent to prison, serving as a seemingly strong deterrent to would-be future coup-makers. Second, as the Fijian majority grew, the likelihood of coups to establish Fijian paramountcy dwindled. Third, with the advent of a power-sharing government, Fiji seemed briefly to have found the seedling of a solution to long-run and bitter antagonisms. After the 2006 coup and the associated destruction of power-sharing, such optimism was no longer possible. What was certain, however, was that Fiji’s political future would depend on the place found for its military forces: On whether or not they could be reduced in size and whether or not they could regain the professionalism and subordination to civilian control that appeared to be growing over the 1990s. If civilian control, judicial integrity and constitutional authority were to be durably re-established in Fiji, the process of doing so was sure to be measured in decades, not months or even years.
 Londregan, J.B. and Poole, K.T. 1990. ‘Poverty, the Coup Trap and the Seizure of Executive Power’, World Politics, 42(2);151–63.
 Huntington, S. 1995. ‘Reforming Civil-Military Relations’, Journal of Democracy, 6(4):15. ‘Countries with per capita GNPs of $1,000 or more do not have successful coups; countries with per-capita GNPs of $3,000 or more do not have coup attempts. The area between $1,000 and $3,000 per-capita GNP is where unsuccessful coups occur, while successful coups in Nigeria, Sudan, and Haiti were in countries with per-capita GNPs under $500’; the Asian Development Bank gives Fiji’s per capita income as $US2,172. http://www.adb.org/FijiIslands/country-info.asp, while the UN Human Development Report 2007/2008, gives a figure, based on purchasing power parity, of $US6,049.
 Finer, S. 1962. The Man on Horseback; The Role of the Military in Politics, Transaction Publishers, New Brunswick and London, (2002 edition).
 The partial exception is the coup in the Solomon Islands in June 2000, but this was followed by a speedy restoration of constitutional rule, and the courts – when they sat – did not rule the post-coup government illegal.
 Diamond, L. 1996. ‘Is the Third Wave over?’, Journal of Democracy, 7(3):22.
 Biman P. and Tisdell, C. 2006. Institutions, Economic Performance and Economic Development: a Case Study of the Fiji Islands, Nova Science, New York, p. 81.
 Huntington. S. 1957. The Soldier and the State, Harvard University Press, Cambridge, pp.82–85. For a critique, see Feaver, ‘Civil-Military Relations’ Annual Review of Political Science, 2, 1999, pp.235–6.
 Those strategic difficulties call to mind Pippa Norris’s distinction between ‘bonding’ parties, such as Qarase’s SDL, which appeal to a narrow ethnic base, and ‘bridging’ parties like Chaudhry’s FLP which ‘bring together heterogeneous publics into loose, shifting coalitions’ (See, Norris, P. 2004. Electoral Engineering: Voting Rules and Political Behaviour, Cambridge University Press, Cambridge, p.10). As Fiji indicates, the difference between these party types may be more strategic than real. The core support base for Chaudhry’s FLP was solidly Indo-Fijian, but its political strategy was to achieve power by building a ‘bridging’ coalition. Qarase’s SDL also sought to build broader coalitions, but by the time it did so – ahead of the 2006 polls – this was too late. The SDL already had a reputation, entrenched in public perception, as a ‘Fijian’ party. The only way that it might have shifted those perceptions, at that stage, was by substantial concessions to any coalition parties or other allies, however small in number, that were publicly identified with other communities.
 Perlmutter, A. 1969. ‘The Praetorian State and the Praetorian Army: Towards a Taxonomy of Civil-Military Relations in Developing Countries’, Comparative Politics, 1(3):383; Stanley, J. (ed.) 1997. Essays on the Garrison State, Transaction Publishers, New Brunswick, NJ.
 ‘Democracy Defiled: The Coup Makes a Comeback’, The Economist, 27 December 2006.
 Huntington, S. 1957. The Soldier and the State: The Theory and Politics of Civil-Military Relations.
 National Council for Building a Better Fiji, Draft Peoples Charter for Change, Peace and Progress, 5 August 2008, p.15.
 Statement by H.E. Commodore Josaia Voreqe Bainimarama, Prime Minister of the Republic of Fiji Islands, 62nd Session of the UN General Assembly, New York, 28 September 2007.
 Finer, S. The Man on Horseback.
 Bienen, H. and Van De Walle, N. ‘Time and Power in Africa’, American Political Science Review, 83, 1989, p.26.
 ‘…military intervention, of which we are now seeing a good deal, will not only continue but become very much more common’ (Finer, S. The Man on Horseback [2002 edition], p.238)
 Feaver, P. 1999. ‘Civil-military Relations’, Annual Reviews of Political Science, 2, pp.217–18; Huntington, S. 1991. The Third Wave; Democratization in the Late Twentieth Century, University of Oklahoma Press.