Table 1.1: Indicators of Economic and Social Development in Developing-World Democracies
Population Area (sq. km) Literacy Rate (%) GDP (US dollars) HDI Index Botswana 1,443,000 582,000 68.0 2,658 0.741 Colombia 34,520,000 1,142,000 90.6 1,873 0.840 Costa Rica 3,071,000 51,100 94.5 2,729 0.884 India 952,590,000 3,167,000 43.6 311 0.436 Jamaica 2,411,000 11,430 84.1 1,593 0.702 Mauritius 1,070,000 2,040 81.7 2,515 0.825
Papua New Guinea 4,107,000 462,840 70.5 1,058 0.504
Sri Lanka 17,865,000 65,610 89.6 656 0.698
Trinidad & Tobago 1,237,000 5,130 97.6 4,217 0.872
Venezuela 20,712,000 912,050 90.6 2,897 0.859
Source: International IDEA 1997. Figures are based on those present at most recent elections as of May 1997.
In many ways PNG also operates as a salutary correction to some of the more complacent findings of comparative political scientists in the electoral studies field. To take but one example: under Duverger’s well-known ‘sociological law’ concerning the relationship between electoral and party systems, PNG’s first-past-the-post electoral system should have produced a relatively stable two-party system.20 In fact, the PNG party system, such as it exists at all, is characterised by a weak and highly-fragmented multipartism and other factors which, according to Ghai, are usually associated with proportional representation: “multiple parties, large candidature, coalitions, and weak governments” (1988, 73). As noted earlier, one explanation for this divergence between theory and practice lies in the almost entirely Western focus of traditional electoral studies, many of which are based exclusively on examination of elections in industrialised countries only.21 Another reason for comparative politics specialists to give more attention to PNG is because it provides another relatively unusual case: a country that has changed from one electoral system to a different one. While incremental adjustments to electoral laws are relatively common, examples of wholesale electoral system change remain unusual, given the in-built inertia of electoral laws 22
20 First formulated in Duverger 1954.
21 See, for example, Bogdanor and Butler 1983; Farrell 1997. 22 See Rose 1983,42-43.
This has traditionally hampered the comparative study of electoral systems: most discussions require a particular system in one country to be compared with a similar system in a different country. Even for two very similar polities, the vast number of different variables in a country-to-country comparison makes it difficult to assess the effect of a single constant such as an electoral system in isolation from the rest of a political system. This means that it is often difficult to make substantial assessments about the effects of one system or another: we are not comparing like with like. This thesis, by contrast, compares two different electoral systems — one of which (AV) is unusual by world standards — in one polity, PNG, over a period of over 30 years.
Papua New Guinea’s political institutions have not received a great deal of academic attention. There has been an invaluable series of election studies published after every PNG election to date23, but with one or two exceptions these have concentrated overwhelmingly on a seat-by-seat description and analysis of election campaigns, with minimal attention paid to formal election rules or the way these rules influence styles of political competition. This reflects the preoccupations of the respective authors and editors: PNG elections are often interpreted as clashes between traditional culture and the modem world of representative politics, the “counterpoint between traditional and modem bases for status-competition” in the words of one author (Strathem 1976, 283). Most election studies to date have implicitly viewed elections as an arena in which the forces of tradition and modernity can be played out, and in which the social customs and obligations of voters and candidates play a dominant role. The formal rules of the electoral system, by contrast, have tended to be seen as being of marginal importance. In part, this reflects the traditional dominance among the social science disciplines of anthropology in studies of PNG, which has long represented something of a Mecca for anthropologists due to the diversity and isolation from external influences of many of its traditional societies. A considerable amount of research and scholarship on politics in PNG, including electoral studies, has been conducted not by political scientists but by anthropologists, who have understandably given more attention to social forces than political institutions.24 The comparative lack of attention to the impacts of modem political institutions has hampered a comprehensive understanding of the political
23 These are: Bettison, Hughes and van der Veur 1965; Epstein, Parker and Reay 1971; Stone 1976; Hegarty 1983; King 1989, Oliver 1989; Saffu 1996.
24 Amongst the scholars cited in this thesis, Marie Reay, Andrew Strathem and Bill Standish are all anthropologists by training.
process in PNG, and led to a tendency to ascribe the causes for political activity overwhelmingly to traditional ‘cultural’ factors, with insufficient attention given to the rational calculations made by PNG’s political actors as they interact with PNG’s political institutions.
This thesis will attempt to show that many of the features of PNG’s electoral politics usually seen as having social, cultural or other explanations are often influenced by, and sometimes reactions to, the incentives presented by the electoral system. In examining and giving prominence to an ‘institutionalist’ line of inquiry in this thesis, the centrality of political culture to questions of political behaviour and democratic performance may sometimes appear to have been forgotten. This is emphatically not the case. The reasons advanced in Chapter Two for PNG’s somewhat surprising democratic success are essentially cultural ones: the combination of traditional social structures of political leadership with the constitutionalism and willingness to abide by the ‘rules of the game’ inculcated amongst the political elite by the Australian colonial administration. A nation’s political culture is also, however, extraordinarily difficult to manipulate and almost impossible to change from above. By contrast, political institutions such as constitutions and electoral systems are considerably less influential but, unlike a nation’s historical or socio-economic situation, can be changed with relative ease. The aim of this thesis is essentially to explore the question of which political institutions will best work to promote peaceful and effective democratic government in a particular social setting. This type of institutionalist inquiry is necessarily, as Juan Linz has noted, “a modest quest, but a worthy one” (1996, 161).
Conclusion
Because of the relative obscurity of PNG for many political scientists, and its unusual status as a competitive democracy in the developing world, part of this thesis will necessarily deal with the history of competitive democracy in PNG. Chapter Two argues that PNG is in many ways a deviant case for students of comparative politics, and one which challenges a number of well-established theories concerning the relationship between democratic sustainability, ethnic fragmentation and political institutions. Chapters Three and Four of this thesis present a description and analysis of elections in PNG, from the first elections in 1964 until the most recent elections in
1997. By accumulating and calculating data on voting trends such as party vote shares, turnover of MPs, margins of victory, turnout rates, party hopping, seats won and other electoral outcomes, this thesis provides the data and analysis to enable the PNG case to be incorporated more fully into the wider field of comparative politics.25 Chapter Four concentrates on the core hypothesis of this thesis: the extent to which co-operative political behaviour and accommodative outcomes were encouraged by PNG’s centripetal political institutions in the pre-independence period, and the extent to which this behaviour changed as PNG political actors responded to the new ‘rules of the game’ after independence.
While Chapters Two through Four are exclusively devoted to PNG, Chapters Five to Eight are comparative. Despite the evidence from PNG in support of centripetal theories presented in Chapter Four, it would be unwise to put too much weight on the PNG example alone, especially considering the unusually fragmented nature of PNG society and the colonial environment in which the relevant elections were held. For this reason, it is important to examine other ethnically-divided societies in which similar electoral arrangements have been adopted to attempt to encourage inter-ethnic accommodation. There are two such cases: Fiji, which in 1997 introduced an AV electoral system for all future elections for this very purpose; and Sri Lanka, which has used a CV system to elect its President since 1978. These two cases are examined in Chapter Five and are found to provide mostly speculative but still important evidence to support the findings of the PNG case.
Having presented the key data from those divided societies in which centripetal institutions have been used, the remainder of the thesis evaluates the implications of the material presented. Chapter Six assesses the specific implications of the PNG, Fiji and Sri Lanka cases for the arguments for and against centripetal theories of democracy in divided societies as advanced by Horowitz, Lijphart, Sisk and others. It also looks at the experience of AV elections in Australia, the longest-running example of the system. Although Australia is an ethnically-diverse society, it is not an ethnically-divided
society, in the sense of ethnicity representing a fundamental political cleavage around which political interests are formed and mobilised, and so cannot be used to directly
25 For example, the second edition of Arend Lijphart’sDemocracies (forthcoming), which utilises data from this thesis.
evaluate arguments for or against centripetalism. Australia does offer, however, the case which best illustrates the way preference-swapping can promote tangible policy changes on the part of parties and governments. Chapter Six concludes by examining the strengths and weaknesses of the theoretical case for AV as an instrument for managing ethnic conflict, finding it to be most applicable to situations where there is a degree of fluidity to ethnic identities and relationships.
Although most of the evidence presented in the earlier chapters concerns the alternative vote, this is not the only preferential electoral system that has been recommended by centripetal enthusiasts. Chapter Seven therefore examines the comparative experience of preferential voting systems around the world. There are three forms of preferential electoral system used for national elections: the alternative vote; the single transferable vote; and the contingent vote. These systems can themselves be further broken down according to whether the system makes the numbering of all preferences compulsory or leaves the decision as to whether to mark preferences beyond the first in the hands of the voter; and, in the case of AV and CV, whether it is used in single-member districts or multi-member districts. Examination of the experience of these variations suggests that not all preferential systems are equally effective at promoting accommodation in divided societies, and that apparently minor technical differences between systems can have major effects in terms of outcomes.
The concluding Chapter of this thesis attempts to put the collected evidence for centripetalism in context. On the basis of the evidence and analysis presented in the preceding chapters, some general facilitating conditions for centripetal approaches to democracy in divided societies are suggested. The central conclusion is that the application of the centripetal model is highly dependent upon socio-structural conditions — particularly the size, number and geographic dispersion of the contending ethnic groups. To be effective, centripetal electoral approaches require ethnically- heterogeneous electoral districts, and are thus likely to work effectively only in situations of extreme ethnic fragmentation (such as Papua New Guinea) or in situations where competing groups are widely dispersed and inter-mixed (such as in Fiji). While uncommon in some regions, there are numerous examples of this type of ethnic group distribution in the Asia-Pacific region, due in part to the influences of colonialism and the presence of large Chinese and Indian diasporas. In addition, all the identified
examples of centripetalism occur in Asia-Pacific states. It is thus likely that centripetal institutions may offer the most appropriate choices for engineering accommodative politics in many of the ethnically-divided states of the Asia-Pacific region in particular, and more generally in situations of high ethnic group inter-mixture or fragmentation in other regions.