Political Parties and Social Structure in the Developing World

Vicky Randall. Handbook of Party Politics. Editor: Richard S Katz & William Crotty. Sage Publications, 2006.


The relationship between political parties and their putative social base raises complex questions, even in the context of developed democracies and abundant information. Outside this context, while losing none of its importance, the relationship becomes even more difficult to pin down. This chapter attempts to capture themes and implications of the existing literature while suggesting some of the analytical difficulties entailed and stressing that much relevant research is at an early stage.

First a couple of words are necessary about the parameters of this discussion. The chapter is concerned with the non-Western or developing world, which is conventionally understood to include Africa, Asia, the Caribbean, Latin America and the Middle East. The coherence of such a huge category, even in the heyday of Third Worldism, was always in doubt. These days, as the shared colonial legacy recedes and economic differentiation between and within its regions increases, the diversity of the developing world—and of its political systems and societies—becomes ever more apparent, making generalization increasingly problematic.

In addition, it is concerned with ‘social structure’, a term which has been understood in many different ways, ranging from a Marxist-inspired focus on economically based social classes to the social institutions of functional sociology. Here it will be interpreted quite broadly to include social classes, categories, groups, and even, where appropriate, institutions and associations.

The chapter proceeds as follows. The next section comments on political parties in the developing world, both what we know about them and how they have been studied. The different possible ways in which one could think about the relationship between parties and their social base are then suggested and provide the framework for the subsequent discussion, looking at parties and social cleavages, parties and electoral behaviour, appeals by party leaders and links between parties and particular social organizations, descriptive and substantive representation and parties themselves helping to create cleavages.

Political Parties in the Developing World

Whilst doubtless queries could be raised about how political parties are to be defined and recognized in a developing world context, these are not the most pressing problems facing the present discussion, which are more to do with the availability of data and scholarly analysis. On the one hand there has long been a relative shortage of such information, and it has been very uneven in terms of geographical coverage. Whilst in some countries, notably India, research into political parties has been solid and continuous, elsewhere, following a number of important pioneering studies, with the onset of authoritarian systems of government in the 1970s there was a sharp decline in scholarly interest. (Re)democratization and the particular emphasis on the contribution political parties can make to democratic consolidation has only relatively recently rekindled interest in this subject. Now there is a great deal of research, with an ever widening range of countries included. Much of this focuses primarily on analysing elections and their results, but other research has taken up themes in the wider democratization literature about parties’ contribution to democracy, institutionalization of parties and party systems, and so on, though there is still little direct discussion of the relationship between parties and their social bases. So a more recent problem is simply keeping up to speed with this proliferating literature, and the present discussion makes no claim to comprehensive coverage.

Second, and at the risk of stating the blindingly obvious, the great number of political parties and party systems in the developing world are hugely diverse. Some, like Congress in India, have existed a very long time, others only appeared with the most recent wave of democratization; some, like Congress and the Bharatiya Janata Party (BJP) in India, the Kuomintang (KMT) in Taiwan, the Institutionalized Party of the Revolution (PRI) in Mexico and the African National Congress (ANC) in South Africa, are agreed to be highly ‘institutionalized’ and/or organized, whilst others scarcely exist beyond their name, and so forth. There have been few systematic attempts to categorize them, although Gunther and Diamond (2003) have made a valuable start in this direction.

Having said that, and always bearing in mind the important exceptions, to which we return below, certain features are regularly observed, no less in recent years than in the earlier phase of scholarly interest in such parties, features also apparent in some Western parties but to a lesser extent. Thus parties in the developing world are regularly observed to be weakly institutionalized, a confused concept (Randall and Svåsand, 2002) but often understood to include ‘rootedness’ in society. Parties have frequently been constructed simply as ephemeral vehicles for their leaders’ personal ambition, and where links with social sectors do exist these are often of a clientelistic nature. Parties generally lack a distinct membership or clearly defined membership criteria, which might give some practical and/or symbolic linkage with society or a section of it. Again typically they do not seek to differentiate themselves by developing coherent programmes or policy debates through which distinct social interests could be identified and promoted. All these features seem likely to reduce the possibility of strong, meaningful links with society.

One reason why parties are often weakly institutionalized is that long periods of authoritarian rule have meant that party development was repeatedly interrupted. Furthermore, there has often been a close association between political parties and the state. Development and democratization could be expected to increase the possibility of party autonomy from the state and, through improved communications, of party leaders strengthening links with their ‘grassroots’. Paradoxically, however, it is argued that in countries like Chile (Munck and Bosworth, 1998) or Brazil, as in the West, party leaders place increasing reliance on television to reach the electorate which limits the development of the party as an organization and thus the possibility of group representation within and through it.

Conceptualizing the Relationship between Parties and Social Structure

Panebianco (1988: 3-4) has famously warned against the ‘sociological prejudice … that the activities of parties are the product of the “demands” of social groups, and that, more generally, parties themselves are nothing more than manifestations of social divisions in the political arena’. Nonetheless the literature on political parties is surprisingly vague about the sense(s) in which and the mechanisms through which a party can be said to be linked to or based on a social sector. This is, however, true of the literature as a whole, not just that concerned with parties in the developing world. To discuss the relationship between parties and social structure(s) we need first to distinguish different possible forms or aspects of that relationship, around which the analysis can be organized.

Much analysis of the relationship between parties and social structure in the Western world takes as its point of departure the Lipset and Rokkan (1967) model discussed in Chapters 29 and 30 of this Handbook. How relevant or helpful is that model for understanding party-society relations in the developing world? More simply, we can ask about patterns of electoral support for parties on the one hand, and the ways in which party leaders seek to project particular party identities on the other. We can also consider links between parties and particular social institutions and organizations. But this is still not quite the same question as whom the party actually does stand for. This requires us to explore the thorny issue of representation, especially in its descriptive and more substantive forms.

Parties and Cleavages

The Lipset-Rokkan model was not devised to explain the emergence of parties in the developing world, but reflected a particular western European history. We can nonetheless ask how far either its categorization of cleavages or its thesis both about the sequence and the manner in which parties were originally formed is of relevance in a developing world context. The model identified a series of potential cleavage lines—between church and state, centre and periphery, urban and rural sectors and labour and capital—which do indeed have relevance for the developing countries, but only up to a point. In the first place, the church-state opposition needs to be extended to include competing confessional parties and the centre-periphery opposition needs to include conflict based on ethnic/regional rivalry as well as opposition to the centre. Second, even then these categories may not adequately convey distinctive qualities of social structures, for instance the depth of attachment to ethnic or religious identity, or the scale of social inequality.

They also omit a cleavage or basis of identification that has been very important in the formation of political parties in the developing world, that of the (oppressed) nation versus the (neo-)imperial power. So many of the biggest, most powerful parties have grown out of the movement for national liberation—Congress, the PRI, a succession of parties in sub-Saharan Africa such as KANU (Kenya), the TANU-Chama Cha Mapinduzi (Tanzania), the ANC (South Africa), FRELIMO (Mozambique), and SWAPO (Namibia). It is possible to argue, as Sinnott (1984) has claimed for the Republic of Ireland, that such parties fall into the center-periphery cleavage category because at the time of their formation they represented the oppressed periphery of a metropolitan centre -but really, given its prominence, this cleavage deserves a category of its own (for a fuller discussion see Randall, 2001).

The Lipset-Rokkan model implies that, with economic development and national integration, increasingly the class-based cleavages will come to the fore. In terms both of predominant cleavages and of the direction of change, Latin America is the developing region that comes closest to this pattern. There, with the possible exception of the PRI in Mexico, parties have not arisen out of a national versus (neo-) colonial cleavage, and to varying degrees by the mid-20th century it was possible to place parties on a left-right axis, reflecting the prominence of social class divisions. Chile in particular, according to Scully (1995), is almost a classic case of party system evolution on Lipset-Rokkan lines. Originating in the 19th century from conflict between clerical and secular forces, by the early 20th century the system incorporated parties reflecting urban class divisions and later a new centre party based on the peasantry.

Elsewhere the model is less helpful and has been less frequently used in explaining either the main underlying social cleavages or the direction of change. However, it may still shed light on the process through which parties were originally formed. Lipset and Rokkan observed that, in Western Europe, party systems formed in the critical period on the eve of mass suffrage reflected or embodied the most salient political cleavages at that time, which then acquired a degree of institutional fixity, making it difficult for parties representing new bases of cleavage to form and break in. This has arguably also been true in many developing countries. Moreover, and anticipating what will be said later about parties creating their own cleavages, some parties or party systems have managed to survive in this way, even when forced underground during periods of authoritarian rule—the party tradition going back to Danquah and Busia and which has resurfaced in the recently victorious NPP in Ghana (Nugent, 2001) is an example.

This leads, finally, and as indeed recognized by Lipset and Rokkan (1967: 3), to the possibility that parties can themselves ‘produce their own alignments’ independent of pre-existing geographical, social or cultural differences. The political institution—the party—could to some degree act back on social structure creating, for instance, a family of new quasi-social institutions that play a part in socializing their participants and institutionalizing the party. An example would be the succession of parties in the Peronist tradition, of which the latest manifestation is the Justicialist Party (JP) in Argentina.

Parties and Electoral Behaviour

The Lipset-Rokkan model, whilst positing a historical link between the origin of party systems and politicized cleavages, does not greatly reflect on the nature of the link between party leaders and their perceived social base. In practice those following in this tradition of analysis have largely relied for evidence of relationship on patterns of voting behaviour, including estimates of their volatility or continuity, supplemented perhaps by independent measures of party identification. This is true more generally of attempts to answer the question of which social group(s) a party stands for. Patterns of electoral support are inferred either through aggregate ecological analysis, exit polls or voting intention surveys. This kind of analysis, at its most sophisticated in India perhaps, has been applied increasingly to elections throughout the developing world. The discussion that follows can only highlight emerging themes and makes no claim to be exhaustive with regard to this proliferating literature.

As already implied, social structures and thus the range of potential social constituencies for parties vary greatly between and even within regions of the developing world. Social class differences are paramount in Latin America. This is partly a reflection of relatively advanced levels of industrialization and urbanization. It is also a consequence of its colonial history. This bequeathed a hegemonic Roman Catholic church strongly identified with the state. It also virtually eliminated the native population in many parts of Latin America, and elsewhere generally resulted in a system where ethnic groupings were ‘ranked’ in Horowitz’s term, or aligned with rather than cross-cutting lines of social stratification (Horowitz, 1985). Ethnicity, then, has been largely absent as an independent basis of party support, although Donna Van Cott has recently produced interesting work on the difference that having large Indian communities can make to party formation, with particular reference to the case of the Movimiento Nacionalista Revolucionario (MNR) in Bolivia (Van Cott, 2000). Despite the relative salience of class divisions sociologically speaking, however, these do not necessarily determine patterns of party support. A decade ago Dix (1989) argued that the alignment of class divisions with voting patterns apparent in many Latin American countries in the 1960s had weakened everywhere but in Chile and Argentina. The trend, in the era of redemocra-tization was for what he chose to call ‘catch-all’ parties, deliberately seeking to maximize their cross-sectional appeal. Myers (1998) provides an example of this in Venezuela where in the 1950s his own ecological analysis suggested that the two main parties, Acción Democrática and COPEI, had relatively differentiated bases of social support but by 1978 public opinion research found that demographic variables had little bearing on voting intentions. The reasons for this regional trend are doubtless complex and locally varied but include, as in the West, the restricted programmatic options for historically left parties.

In stark contrast with Latin America, in Africa, by common agreement, ‘ethnicity’, whatever exactly is understood by this term, has much greater social salience than social class (except perhaps in South Africa) or, hitherto, than religion. And it is widely assumed to underpin both party formation and party strategy: Ottaway (1999: 311), for instance, refers to the ‘overt or covert ethnic character of the majority of the emerging political parties’. This is true despite legislation or constitutional provisions, in a number of countries such as Tanzania, Cameroon and Ghana, designed to prevent the formation of parties or electioneering on this or other particularistic bases. The role of ethnicity in shaping patterns of electoral support is confirmed, for instance, in Nugent’s (2001) analysis of the simultaneous presidential and parliamentary elections in Ghana in 2000. In analysing patterns of electoral support, it is admittedly sometimes difficult to differentiate between ethnicity and regionalism, given the tendency for ethnic groups to be geographically concentrated, an issue explored in Kaspin’s (1995) excellent analysis of the 1994 presidential and parliamentary elections in Malawi. There is also some debate as to how far tropical Africa’s parties actually are ethnic, in the sense of based on a single ethnic group, as opposed to multi-ethnic (Scarritt and Mozaffar, 2002). Certainly in most national contexts, and depending on the nature of ethnic cleavages, both traditional ruling parties and emerging opposition parties with any realistic chance of winning power or even becoming significant players, need to draw support from more than one such community.

In much of Asia, and to a lesser extent the Middle East, social structure is more complex, with significant fracturing on ethnic, religious and class/caste lines. Asia of course covers a very diverse range of countries. In South Korea the picture is simplified: given that it is largely homogeneous in ethnic terms and that there have been severe ideological and constitutional constraints on trade union involvement in party politics, the primary determinant of electoral support for parties, since democratization got under way in 1987, has been region (Kim, 1998). India, however, epitomizes Asian complexity, which is compounded by its federal system of government. Given that competitive party politics has prevailed since independence in 1947, systematic study of voting behaviour goes back a long way (for a valuable discussion of the methods, strengths and findings of ecological analysis in the 1960s and 1970s, see Brass, 1985). One distinctive feature of Indian social structure is the caste system, whose character has of course profoundly changed, partly as a consequence of its encounter with democratic forms of politics, but which is still recognized to be of major relevance to voting behaviour. The Congress Party, originally based on the movement for national independence, remained electorally dominant at the national level into the 1990s. It drew support from all sectors of society, although especially from both the upper castes and least privileged sectors (the Untouchables or ‘Dalits’, tribals or ‘adivasis’, and Muslims). However, Chhibber and Petrocik (1990) found that when analysed at individual state level, its electoral support was much more homogeneous, and based on particular, if locally varying, caste and other groupings, a view more recently echoed by Heath and Yadav (2002). The 1990s saw the rise of a number of state-based parties that draw disproportionately on the electoral support of particular caste blocs, such as the Bahujan Samaj Party (BSP) based in Uttar Pradesh which draws a large share of the Dalit vote. At the same time at the national level Congress has been successfully challenged—until the 2004 general election—by the BJP, widely perceived as a Hindu chauvinist party. Electoral support for the BJP obviously comes primarily from Hindus; however, they constitute around 83% of India’s population, and interest has focused rather on the disproportionate rates of support from upper castes, especially in the north-western ‘Hindi heartland’. The success of the BJP has involved widening its social bases of support to include lower castes and regions further south as well as allying with parties with contrasting electoral bases (Heath, 1999).

Given the chequered history of competitive party politics in the Middle East, the scope for analysing patterns of electoral support for parties has inevitably been limited. Of particular potential interest, however, are the increasingly numerous Islamist parties. These obviously differ in many respects one from another and several may coexist in a single country. Some are conservative and closely linked with the prevailing regime. But others are more oppositional in character, have often developed out of a wider movement, and have relatively strong social ‘roots’. Again obviously their electoral supporters have in common a Muslim religious identity which they may perceive to be threatened, even in such overwhelmingly Muslim countries, by secularizing policies of government. Nonetheless, whilst hard psephological analysis is hard to come by, it appears that support comes disproportionately from particular social sectors and for different reasons. The role of the professional middle class and students in supporting more radical Islamic movements is often noted. In his study of support for Turkey’s Welfare (Refah) Party, Gülalp (2001) also underlines the importance of the small-scale, provincial business sector which suffered under state-centred Import Substitution Industrialization (ISI) type development policies in the 1960s and 1970s but has expanded in the era of economic globalization. But purely in terms of numbers most significant has been support amongst the rapidly growing urban poor due in no small measure to the extensive welfare work undertaken by such parties and associated organizations. It was in the poorest neighbourhoods, where the Turkish Welfare Party ‘spoke the language of socio-economic justice and equality’ (Gülalp, 2001: 442) that its vote grew most rapidly in the 1990s. Likewise, in Algeria, the welfare network of the Islamist movement provided the Front Islamique du Salut (FIS) ‘with a significant and loyal pool of potential voters’ in 1992 (Willis, 2002: 10).

Alternative ways to Identify a Party’s Social Base

In addition to looking at patterns of electoral support to identify a party’s social base, one could focus on what the party leadership does to attract support in terms of the manipulation of symbols, public statements, targeted campaigning, programmes, manifestos (where these are available) and so forth. To which group(s) do they appear to be appealing? A useful resource would be something like the Manifesto Research Group (Budge, 1994) to collect and analyse such material. In a developing world context, little systematic work of this kind has yet been undertaken and further difficulties arise from the fact that in many cases parties either do not produce programmes or they barely differ from one another. Thus according to Wanjohi (2003: 251), ‘nearly all party manifestos in Kenya look alike, often using the same phraseology, and even identical paragraphs’.

Alternatively, a clue to the social character of a party may lie in the nature of the social institutions or organizations into which it is historically and/or actually linked. For instance, parties like the PT in Brazil and successive Peronist parties in Argentina, or the MMD in Zambia and the ANC in South Africa, have had close links with trade unions. India’s BJP is closely linked to Rashtriya Sevak Sangh (RSS), a highly disciplined militant Hindu organization, and its ‘family’ of voluntary associations.

Parties and the Representation of Social Groups

But none of these indicators of a party’s social base necessarily tells us who the party really stands for. This raises the wider issue of representation, which is of course as relevant to Western political parties as to those of the developing world. Which social groups do parties represent, and in what sense? From the perspective of normative political theory, almost nothing seems to have been written about the role parties ideally should play in representation of social groups (a partial exception is Birch, 1972), perhaps because this is actually a very difficult question to answer (Randall, forthcoming).

Pitkin (1967) and others, writing about representation more generally, have distinguished different forms. Most relevant for us is the distinction between descriptive and substantive representation. In the former case, in a party context, specific attributes of the social group in question are reflected, say, in the composition of party membership, its key decision-making bodies or its candidates for the legislature. In the latter, the perspectives and interests of that group are voiced and carried forward in internal debate and/or party policy. A problem of which post-structural arguments have made us more aware here is of course how such interests come to be identified and articulated—if they are not ‘objectively’ to be read off, what part does the party itself play in constructing them?

Liberal theorists of representation have questioned the value of descriptive representation, arguing that there is no guarantee delegates selected on this basis will promote the interests of the group whose characteristics they share. However, more recently, feminist theorists in particular have suggested that measures to increase descriptive representation—or ‘the politics of presence’—may be valuable symbolically, to correct past injustices, and membership of a group may be necessary to understand and articulate its distinctive experience and concerns (Phillips, 1995). Whether or not this is so, descriptive representation is often a very real concern, explicitly or implicitly, for parties in developing countries. It is an intrinsic feature of ethnically based political parties in sub-Saharan Africa that their leaders, parliamentary candidates and suchlike should themselves be members of the ethnic group they claim to stand for. In India, all parties, including communist parties, have had to take descriptive representation, in terms particularly of caste, into account when selecting candidates for local, state and national elections.

Especially following the 1995 UN Women’s Congress in Beijing, the issue of women’s representation, and specifically of gender quotas, has come to the fore. The trend has gone furthest in Latin America, where between 1991 and 2000 twelve countries passed laws requiring that 20-40% of candidates for national elections should be women. Such provisions in practice have had mixed success in terms of actually increasing the number of women candidates, one important factor being the good-faith compliance of parties themselves. Several major parties also use gender quotas for internal party positions, including El Salvador’s FMLN and Nicaragua’s FSLN (Htun and Jones, 2002). In sub-Saharan Africa a number of parties, including three based on movements of national liberation—the ANC in South Africa, FRELIMO in Mozambique and SWAPO in Namibia—have adopted gender quotas (Yoon, 2001). While adoption of quotas in a number of cases reflects the lobbying of women activists themselves, it has also often coincided with the party’s desire, in a (re)democratizing era, to establish its pluralist, democratic credentials. As to whether gender quotas enhance women’s substantive representation in and through the party, this varies from case to case, and feminist commentators further disagree amongst themselves. For instance, Goetz and Hassim (2001) see gender quotas in the ANC as indirectly helping to bring about three ‘path-breaking’ pieces of legislation for women—the 1996 Termination of Pregnancy Act and, in 1998, the Domestic Violence Act and the Maintenance Act, but Vincent (2002) is less sanguine. Htun and Jones conclude that gender quotas have not necessarily directly affected policy but have helped to transform collective gender awareness by stimulating debate, a view confirmed in Sacchet’s (2003) recent study of gender quotas in Brazil’s PT.

Substantive Representation

The question of how far political parties further the substantive representation of particular social groups could take us into political and philosophical terrain well beyond the implicit brief of this chapter. What is intended here is the more limited but necessary discussion of some of the factors tending to constrain such representation. Many parties in the developing world are primarily vehicles for individual politicians hastily scrambled together for electoral purposes with the vaguest of social bases and minimal attempt to elicit the views of their potential supporters. But even where parties make more efforts to cultivate such a base, the extent of communications with supporters may be constrained by, for instance, geographical distance and the town-rural divide (it is regularly noted that in tropical Africa opposition parties have difficulty extending their organizations beyond the main urban centres) and by social distance. Van Cott (2000) describes how from the late 1950s Bolivia’s MNR appealed to highland Indians, helping them to recover their communal lands. But thereafter, Van Cott suggests the Indians did not seek representation in conventional political terms, wanting rather to be left alone; in return their vote provided the party with its ‘conservative anchor’. Of course it could be said that in this case the Indians got what they wanted—until the economic situation deteriorated.

Clientelism is a pervasive feature of parties in the developing world; accounts so far suggest that neither democratization nor measures to constrain the public sector and public expenditure in keeping with neo-liberal prescriptions have significantly eroded it. A relatively underexplored question is the implication of clientelism for representation: certainly, as Piattoni (2001), writing in a European context, has noted, in clientelism the level of discourse is never lifted up from the particular. Some accounts imply that it is incompatible with and indeed inimical to representation as commonly understood. Alternatively, it might be possible to see clientelism, as some Africanists do, as a distinct form of representation. According to van de Walle and Butler (1999: 26), ‘In Africa today. … parties do not really serve to aggregate interests; rather they serve a representation function in a context of clientelistic politics’ (see also Chabal and Daloz, 1999). That still leaves open the question of whether and in what sense (short or long term) this clientelistic exchange of their votes for favours is in the voters’ interests.

An alternative, though not necessarily incompatible, model would be of some form of corporatist representation of different social groups—women, youth, peasants, unions and so forth—within a party, or through affiliated organizations. A well-known historical example would be Mexico’s PRI. In contrast to clientelism, this model does give formal recognition to groups; however, it often goes with a very centralized system of party decision-making. Fears, for instance, have been expressed that South Africa’s ANC is moving in this direction.


Analysis of the relationship between political parties and social structure in the developing world is flawed by the same conceptual evasions and confusions as in the West. In addition, while research on parties has burgeoned in the era of (re)democratization, there is much less research directly focused on this question. There are also particular fields where more basic information would be valuable, such as the social bases of Islamic parties, and also of the small but growing number of Green parties in the developing world. Cleavage theory has something to offer for the analysis of society-party relationships, especially in Latin America. But much more work needs to be done in investigating and conceptualizing the representation of social groups, including the implications of clientelism.