The study of electoral politics and political behavior across the developing world has grown into a substantial body of scholarship over the past several decades. A hallmark of this literature has been to revise models first developed to explain the politics of advanced industrial countries. While it would be unwise to speak of a unified “consensus” that has materialized out of this rich and diverse scholarship, a set of conventional wisdoms has emerged that structure our understanding of electoral dynamics in the developing world. In particular, three pieces of received wisdom stand out.
First, electoral politics in the developing world is seen to be dominated by various forms of “distributive politics” (Stokes et al. Reference Stokes, Dunning, Nazareno and Brusco2013), which stand in marked contrast with programmatic politics linked to tax-and-transfer policies. Prevailing notions of distributive politics are often premised on the idea of quid pro quo contingent exchange, whereby voters select leaders based on targeted benefits and leaders, in turn, deliver these benefits to voters. The commitment problems governing such discretionary exchanges require parties to deploy local brokers who monitor electoral compliance. The distribution of promised goods is typically understood to take place during elections, leading to a focus on campaign season “vote-buying” as the dominant form of distributive politics (Stokes Reference Stokes2005; Nichter Reference Nichter2008).
A second, related element of the received wisdom is that ethnicity is one of the most—if not the most—crucial determinants of vote choice for large sections of the developing world, particularly in South Asia and Sub-Saharan Africa. Many scholars acknowledge that voters might have expressive preferences for co-ethnic candidates and parties (Horowitz Reference Horowitz1985; Carlson Reference Carlson2015). Perhaps even more influential, however, is the belief that ethnicity can provide a useful heuristic for parties and voters to credibly support one another, thereby solving the commitment problems at the heart of discretionary distributive politics (Fearon Reference Fearon1999; Chandra Reference Chandra2004; Posner Reference Posner2005).
Third, political parties in much of the developing world are characterized as relatively weak in organizational terms. They lack both strong norms that guide how power and responsibility flow internally as well as the organizational wherewithal to exert a meaningful presence in the daily lives of citizens between elections. Consequently, citizens are thought to attach to candidates rather than parties, which only reinforces the weak incentives for programmatic politics to develop. Parties mobilize to contest and win elections, but fade into the distance once those elections conclude.
As the world’s most populous—and the developing world’s most enduring—democracy, India has provided both inspiration for, and validation of, these three stylized facts. First, India has widely been seen to characterize non-programmatic distributive politics. Indeed, the country has been dubbed as the epitome of a “patronage democracy.”Footnote 1 Second, India has long been invoked as an exemplar of an ethnicized democracy, with caste and religion structuring the vote (Kothari Reference Kothari1970; Chandra Reference Chandra2004). Finally, Indian political parties are largely described as institutionally weak, organizationally shallow, and overly personalistic (Kohli Reference Kohli1990; Keefer and Khemani Reference Keefer and Khemani2004; Manor Reference Manor, Adeney and Saez2005).
The objective of this article is to consider research on India that challenges the contention that India neatly fits the prevailing wisdom on each of the three issues highlighted earlier—distributive politics, ethnic voting, and political parties. In so doing, it forces us to scrutinize afresh the very underpinnings of that conventional wisdom for research on developing countries more broadly. Thus, this article is not about how comparative politics can inform our understanding of India, but what India can teach us about how to reframe some of our core assumptions as comparativists. We also argue that the research highlighted here suggests that scholars of comparative politics must be more prudent in how they invoke India to support their claims. To be clear, our aim is not to suggest that existing characterizations of India are wholly erroneous, but rather to bring nuance to what we perceive to be the prevailing wisdom. Finally, we discuss some possible new research frontiers opened up by relaxing some of the standard assumptions that presently shape our thinking.
Distributive Politics
Over the past two decades, a burgeoning literature on distributive politics has modelled politician-voter linkages and their implications for the allocation of government-provided goods and services and politician-enabled election-time handouts (Golden and Min Reference Golden and Min2013). Scholars have analyzed politicians’ decisions to allocate public resources across space (Dixit and Londregan Reference Dixit and Londregan1996; Porto and Sanguinetti Reference Porto and Sanguinetti2001; Arulampalam et al. Reference Arulampalam, Dasgupta, Dhillon and Dutta2009; Cammett and Issar Reference Cammett and Issar2010; Nathan Reference Nathan2019), their strategies over the distribution of private benefits between elections (Auyero Reference Auyero2000; Calvo and Murillo Reference Calvo and Murillo2013; Diaz-Cayeros, Estévez, and Magaloni Reference Diaz-Cayeros, Estévez and Magaloni2016), as well as their decisions over the allocation of cash and other gifts during elections (Wantchekon Reference Wantchekon2003; Brusco, Nazareno, and Stokes Reference Brusco, Nazareno and Stokes2004; Schaffer Reference Schaffer2007).
A major strand of the literature on distributive politics focuses on clientelism—iterative, quid pro quo relationships between politicians and voters in which access to goods and services is conditional on political support.Footnote 2 Although research on clientelism is diverse (see Hicken Reference Hicken2011 for a review), much of this literature is based on four core assumptions about politician-voter relations: contingency, electoral timing, voter monitoring, and hierarchy in politician-voter interactions. In this section, we unpack these assumptions and reevaluate them in light of recent evidence from India.
Core Assumptions
First, clientelism revolves around contingent transactions. The implications of models that emphasize such contingency (Stokes Reference Stokes2005; Rueda Reference Rueda2016; Nichter Reference Nichter2018; Gans-Morse, Mazzuca, and Nichter Reference Gans-Morse, Mazzuca and Nichter2014) are worth spelling out. If distributive politics is primarily about quid pro quo transactions, moral and programmatic elements should rarely guide allocation decisions and voters should infrequently choose candidates on the basis of programmatic promises.
Second, the transactions in question typically involve the distribution of targeted goods during elections—and research focused on this asks whether and how such efforts sway voters and boost turnout. Across diverse settings such as Argentina (Auyero Reference Auyero2000; Szwarcberg Reference Szwarcberg2015), Brazil (Gay Reference Gay1994), Hungary and Romania (Mares and Young Reference Mares and Young2019), India (Piliavsky Reference Piliavsky2014), and Kenya (Kramon Reference Kramon2016), it appears that communities are flooded with material inducements in the runup to elections. An implicit assumption is that politicians expect to directly benefit from these efforts.
Third, contingent transactions require local political actors who can facilitate distribution and observe voters’ preferences with reasonable accuracy. Most analyses emphasize the existence of party networks enlisting brokers to carry out these quid pro quo transactions (Stokes Reference Stokes2005; Stokes et al. Reference Stokes, Dunning, Nazareno and Brusco2013; Camp Reference Camp2015). In many studies, these party networks are modeled as relatively fixed, comprising long-term partisan actors. As a result, brokers—as in the quintessential Peronist example from Argentina—emerge as committed to a specific party network. Relatedly, the literature frequently relies on these brokers’ ability to distinguish between core supporters and swing voters (Stokes Reference Stokes2005; Nichter Reference Nichter2008; Stokes et al. Reference Stokes, Dunning, Nazareno and Brusco2013). This assumption implies a partial failure of the secret ballot.
Fourth, models of clientelism also often emphasize hierarchy in politician-voter interactions—the ability of leaders to successfully wield the upper hand with voters, resulting in a power asymmetry in patron-client relations. The contingency at the heart of clientelism implies a form of “perverse accountability” that robs individuals of choice over their vote (Stokes Reference Stokes2005). The implications of this narrative are bleak for democratic politics.
Prevailing Wisdom
India has typically been assumed to exhibit these same characteristics. In fact, the country is frequently described as an archetypical “patronage democracy” in which the distribution of state services is discretionary, whether driven by electoral, material, or cultural expectations (Bailey Reference Bailey1970; Chandra Reference Chandra2004; Chatterjee Reference Chatterjee2004; Piliavsky Reference Piliavsky2014), and where citizens often turn to intermediaries to help them navigate otherwise dismissive, capricious state institutions (Reddy and Haragopal Reference Reddy and Haragopal1985; Oldenburg Reference Oldenburg1987; Manor Reference Manor2000; Corbridge et al. Reference Corbridge, Williams, Srivastava and Veron2005; Harriss Reference Harriss2005). Thus, descriptions of non-programmatic politics, rent-seeking politicians, and porous state institutions have long been applied to India (Krueger Reference Krueger1974; Rudolph and Rudolph Reference Rudolph and Rudolph1987; Bardhan Reference Bardhan1998). Herring (Reference Herring and Woo-Cummings1999), for instance, described how governance in India exhibits forms of “embedded particularism” wherein the actions of state officials are twisted to local political interests. Gupta (Reference Gupta1995) described the “blurred lines” between state and society in India, highlighting the negotiated exchanges that unfold in local governance.
In such contexts, citizens can appeal to politicians to intervene in bureaucratic procedures and have the rules bent to advance their material interests (Berenschot Reference Berenschot2010). Intermediaries proliferate in this environment to mediate access to state services and collect rents and patronage for such activities (Reddy and Haragopal Reference Reddy and Haragopal1985; Manor Reference Manor2000; Krishna Reference Krishna2002; Mitra Reference Mitra2002). India is thus depicted as an “intermediated democracy” (Berenschot Reference Berenschot2010), where individuals similar to the “brokers” at the center of the comparative literature on distributive politics abound (Bardhan et al. Reference Bardhan, Mitra, Mookherjee and Sarkar2009).
New Approaches
Such strong evidence of local intermediation might lead observers to assume that India is a setting where we should observe many of the dynamics previously highlighted as central to clientelism. Yet many studies of Indian politics upend several of the key assumptions embedded in the comparative literature.
First, the assumption of quid pro quo politics—particularly the ability of politicians or brokers to monitor voting behavior—is challenged by a robust secret ballot, meaningful voter autonomy, and substantial electoral volatility (Heath Reference Heath2005; Uppal Reference Uppal2009). Indeed, one recent study in rural Rajasthan found that sarpanch (elected village officials) are not particularly competent in “guessing” the voting behavior of local villagers (Schneider Reference Schneider2019). Another study in an urban metropolitan setting (Mumbai) finds that brokers tasked with allocating election-time money for candidates do not even attempt to monitor votes (Chauchard Reference Chauchard2018). Both suggest that the exchange of support for goods between voters, brokers, and patrons in India is probabilistic, at best. An evaluation of state legislator behavior by Bussell (Reference Bussell2019) shows that politicians offer constituency service similar to what is observed in Global North democracies, without attention to contingency or partisan leanings.
Second, recent work suggests that the universe of intermediaries engaged in distributive politics is less partisan than what previous work suggests. Whereas party activists are often described as having strong and stable ties to parties in the Latin American context (Magaloni Reference Magaloni2008; Stokes et al. Reference Stokes, Dunning, Nazareno and Brusco2013), recent work suggests the party-broker relationships in India are quite fluid.Footnote 3 These actors are not just geographically proximate to voters, but they are also embedded and active outside of elections, when their primary purpose is to help citizens access the state (Krishna Reference Krishna2002; Kruks-Wisner Reference Kruks-Wisner2018). In addition to traditional party workers (Harriss Reference Harriss2005; Jha, Rao, and Woolcock Reference Jha, Rao and Woolcock2007; Berenschot Reference Berenschot2010; Auerbach Reference Auerbach2020), middlemen can include entrepreneurial, non-partisan individuals who will work for the highest bidder (Manor Reference Manor2000; Krishna Reference Krishna2002). These naya neta (new leaders) are thus often not expressly tied to any one political party (Krishna Reference Krishna2002).
The assumption that politicians inundate voters with handouts because they think it will swing elections also finds limited support. Studies of electoral handouts suggest that their impact is marginal and that politicians are well aware of these limitations (Björkman Reference Björkman2014; Chauchard Reference Chauchard2018). For politicians, these efforts may instead be about revealing targeting preferences to voters (Schneider and Sircar Reference Schneider and Sircar2020), generating reputations for efficacy (Auerbach and Thachil Reference Auerbach and Thachil2018), signaling electoral viability or credibility regarding the promise of future transfers (Björkman Reference Björkman2014; Muñoz Reference Muñoz2014) or simply seeming “glamorous” (Jensenius Reference Jensenius2017).
Finally, research on Indian politics forces us to question the perception of voters as passive recipients of targeted goods (Stokes Reference Stokes2005; Nichter Reference Nichter2008). Recent research on Indian politics, however, finds substantial voter agency, often activated via bottom-up forms of organization and associational activity (Dasgupta Reference Dasgupta2017; Auerbach Reference Auerbach2020). Voters often hedge their bets by diversifying their claim-making strategies (Kruks-Wisner Reference Kruks-Wisner2018) and can circumvent non-responsive politicians and officials in India’s multi-tiered, federal democracy (Bussell Reference Bussell2019). As Bussell’s work demonstrates, citizens who are “blocked” from accessing public services because they lack connections to local patronage networks often petition higher-level politicians who have an incentive to respond to their grievances in order to expand their personal following. Poor voters are also pivotal in the construction of patron-client hierarchies (Auerbach and Thachil Reference Auerbach and Thachil2018), challenging studies that see such networks as structures imposed from above (e.g., Calvo and Murillo Reference Calvo and Murillo2013).
As a result of these findings, studies of Indian politics have pivoted away from examining episodic forms of “vote-buying” toward more quotidian forms of distributive politics (see Bussell Reference Bussell2012; Thachil Reference Thachil2014; Bohlken Reference Bohlken2017; Dasgupta Reference Dasgupta2017; Bussell Reference Bussell2019; Auerbach Reference Auerbach2020; Schneider Reference Schneider2020). The services being analyzed—roads, sewers, streetlights, and water taps—are politicized in their allocation and frequently involve networks of intermediaries. Yet they are high-spillover goods, undermining the ability of politicians and brokers to exclude non-supporters. This suggests that politicians may be targeting groups or localities rather than individuals. In this respect, recent work more closely approximates the questions examined and findings unearthed during an earlier wave of research in Latin America (Ray Reference Ray1969; Gay Reference Gay1994). Moreover, these services are often provided not through election-time spending, but rather through everyday allocations that are harder to selectively withdraw than campaign handouts (Bussell Reference Bussell2019).
Implications
If the conventional wisdom in the scholarship on clientelism and patronage imperfectly applies to India, there are good reasons to question how well it applies to other contexts too, as argued in recent work on Latin America and Africa (see, e.g., Lawson and Greene Reference Lawson and Greene2014; Kramon Reference Kramon2016). This suggests that theories of distributive politics in developing countries need to be updated, expanded, or nuanced. Specifically, we identify two potential avenues for future research.
First, research could explore variation in the roles and characteristics of brokers in mediating distribution. The depiction of the broker as a partisan activist embedded in a strongly organized machine is just one of many forms in which intermediaries emerge and operate. The Indian case suggests that brokers need not be partisan activists in strongly organized machines, but rather may hedge their bets and refrain from investing in particular parties. New research should consider what types of actors have discretion over distribution and how this varies across contexts.
Second, another exciting avenue of research could explore patterns of allocation in contexts where partisanship and ethnicity have insufficient explanatory power. How does variation in citizen-level mobilization and claim-making strategies impact the chances of securing public services? And does variation along these lines lead to more just allocations of benefits, or perpetuate inequalities? Answering these questions will provide important insights into the ways in which citizens access the state and, crucially, the relative importance of agency in distributive politics.
Ethnic Voting
While the study of clientelism has focused on how electoral incentives shape politicians’ choices about the distribution of state resources, another extensive strand of literature has focused on the role of ethnic identity in shaping vote choice. While definitions of ethnicity can themselves be contentious (see Brubaker Reference Brubaker2004; Chandra Reference Chandra2006), we follow broad convention in viewing ethnic groups as based on ascriptive categories such as race, tribe, religion, or more subjectively, as “self-identification around a characteristic that is difficult or impossible to change, such as language, race, or location” (Birnir Reference Birnir2006, 66). Contributions from India to the study of ethnic politics have come from scholars of social movements (Baruah Reference Baruah1999; Jaffrelot Reference Jaffrelot2003), identity formation (Pandey Reference Pandey1993; Chatterjee Reference Chatterjee2004), and violent conflict (Varshney Reference Varshney2002; Wilkinson Reference Wilkinson2004; Brass Reference Brass2011). Given the focus of this essay, we focus here more narrowly on scholarship examining ethnic voting.
Core Assumptions
Theories of ethnic voting can be roughly partitioned into two camps: “expressive” and “instrumental” theories.Footnote 4 Theories of expressive voting draw on social-psychological models of intergroup behavior, especially social identity theory, which predict in-group favoritism to manifest under even the most minimal of conditions (Tajfel and Turner Reference Tajfel, Turner, Austin and Worchel1986; Shayo Reference Shayo2009). Expressive theories anticipate that individuals in ethnically fragmented societies will seek affirmation of group self-worth by voting along ethnic lines (see Horowitz Reference Horowitz1985). In the most extreme manifestations of this dynamic, elections will simply reflect the demographic strength of different groups.
In instrumental theories, ethnicity’s political salience is explained as a consequence of its utility in maximizing anticipated access to material benefits. One influential set of arguments views ethnicity as an informational shortcut, or heuristic, in circumstances where voters have little information about a party’s programmatic or distributive agenda. Sticky, visible ethnic markers are perceived as especially useful in solving commitment problems plaguing the discretionary exchange of goods for votes. Thus, voters support parties and candidates associated with their ethnic group, not because of a psychological attachment to their in-group, but because they see co-ethnics as their best chance of claiming state resources (Chandra Reference Chandra2004). Other models focus more on ethnicity’s relative utility in crafting minimum winning coalitions (Posner Reference Posner2005; Huber Reference Huber2017). A third set of studies highlight the instrumental behavior of political elites, who sometimes deploy divisive campaign strategies intended to polarize ethnic divisions, in the hope of inducing ethnic voting (Wilkinson Reference Wilkinson2004; Ferree Reference Ferree2006; Nellis, Weaver, and Rosenzweig Reference Nellis, Weaver and Rosenzweig2016).Footnote 5 Instrumental models of ethnicity have quickly risen to prominence within the study of electoral behavior across the global South.Footnote 6
Prevailing Wisdom
Whether interpreted as expressive or instrumental, ethnic voting is considered an important force molding electoral behavior in low-income, multi-ethnic democracies. India has been central to motivating and confirming this conventional wisdom. Pioneering scholarship on India’s postcolonial politics outlined how ethnic groups, especially caste groups, rapidly adapted to the imperatives of democratic politics (Kothari Reference Kothari1964; Weiner Reference Weiner and Kohli2001; Shah Reference Shah2004). Voters of the same localized sub-castes (jatis) were mobilized into electoral blocs or “vote banks” through various mechanisms.Footnote 7 Caste-based associations swiftly transformed from traditional social organizations to vehicles for interest group politics—whether as a means to organize voters or to place demands on the state for greater welfare, educational, and economic development (Rudolph and Rudolph Reference Rudolph and Rudolph1960). Localized caste-based voting blocs were integrated into multi-ethnic coalitions headed by powerful elites who delivered the votes of their bloc to the dominant party Indian National Congress party (Srinivas Reference Srinivas1955; Bailey Reference Bailey1970; Kothari and Maru Reference Kothari, Maru and Kothari1970; Jaffrelot Reference Jaffrelot2000).Footnote 8
The nature of ethnic voting shifted in the competitive multiparty system that replaced Congress dominance. No longer content to be subsumed within upper caste-led factions, lower castes sought to aggregate local jatis of similar status into broader social blocs that could fuel their own political parties (Yadav Reference Yadav1999; Pai Reference Pai2002; Michelutti Reference Michelutti2008; Ahuja Reference Ahuja2019). This “silent revolution” (Jaffrelot Reference Jaffrelot2003) dramatically reshaped the social composition of the country’s legislatures. In many Indian states, the faltering Congress was replaced by a fragmented set of regionalized political players seen to draw support from specific linguistic or caste communities (Ruparelia Reference Ruparelia2015; Ziegfeld Reference Ziegfeld2016).
The political transformation of caste-based interests in the early years of Indian democracy can help us understand how the political salience of ethnic identities are activated and continually reshaped in a dynamic social and political context. In particular, the Indian experience informs our comparative understanding of how a status-based system, similar to other ancien régime cases, adapts and finds new relevance. As Yadav (Reference Yadav1999, 2398) notes: “politics has affected caste as much as caste affects politics.” Some scholarship has theorized the microfoundations of ethnic voting in India. Chandra (Reference Chandra2004) links ethnic voting to India’s “patronage democracy,” whereby stable and visible ethnic identities effectively structure clientelistic exchange. Consequently, voters choose parties with the highest “head counts” of leaders from their own ethnic group.Footnote 9 Other accounts have emphasized the psychological benefits of descriptive representation, which especially fuel ethnic voting among low-status groups (Pai Reference Pai2002). Beyond caste, scholars have argued that India’s Muslims vote cohesively for particular types of parties and candidates, based on their overall demographic strength (Rudolph and Rudolph Reference Rudolph and Rudolph1987, 195).
The importance of this long lineage of work is indisputable. These studies also highlight how identity interacts with various causal factors—political agency, institutional design, patronage, ideological conflicts, group size, and cross-cutting social cleavages—to explain political outcomes. Yet while these studies are individually nuanced in their explanations of ethnic voting, their collective weight often gives the impression that India is primarily characterized by ethnicized political behavior. As Herring (Reference Herring, Kohli and Singh2013, 137) critically notes, many observers continue to describe Indian politics as “the moving about of blocks on a chess board—this caste supports X, this caste, Y, and so the election went.” Ethnic voting can thus easily be construed as the past and present of Indian democracy. Importantly, studies of other parts of the world often invoke the example of India as a paradigmatic example of a polity dominated by ethnic voting (Ferree Reference Ferree2006; Gao Reference Gao2016; Ejdemyr, Kramon, and Robinson Reference Ejdemyr, Kramon and Robinson2017).
New Approaches
Recent studies of Indian politics have challenged this stereotype. Beyond identifying variations in the electoral preferences of members of the same group as past studies have done, current scholarship has more centrally focused on the explanations for why such heterogeneity in within-group preferences emerges across states and time periods. Thachil (Reference Thachil2014) analyzes the roots of variation in support from low-status ethnic groups in India for the Hindu nationalist party BJP, which has historically been identified as representing wealthy upper-caste Hindus. Suryanarayan (Reference Suryanarayan2019) explains how the shared social rank of socially privileged upper caste voters can explain why poor voters from high-status groups join their wealthy counterparts in supporting the conservative BJP. Ahuja (Reference Ahuja2019) explores variation in the cohesion of voting among low caste Dalits and shows how it is more prevalent in states with weaker Dalit social movements; social mobilization of the marginalized, he argues, weakens ethnic voting because it makes other political parties co-opt their agenda. Heath, Verniers, and Kumar’s (Reference Heath, Verniers and Kumar2015) work on electoral preferences among Muslims, long considered India’s most cohesive ethnic vote bank, shows that Muslims only support co-ethnic candidates who have a realistic chance of winning. Devasher (Reference Devasher2014) arrives at similar conclusions through an analysis of Muslim communities in Uttar Pradesh.
There is also increasing evidence of performance-based—or economic—voting in India (Verma Reference Verma2012; Gupta and Panagariya Reference Gupta and Panagariya2014; Vaishnav and Swanson Reference Vaishnav and Swanson2015). Cole, Healy, and Werker et al. (Reference Cole, Healy and Werker2012) show that voters are less likely to punish the ruling coalition for weather emergencies when the government responds effectively to them.
These findings in no way demonstrate that ethnicity is irrelevant in Indian politics. They do suggest, however, the need to view ethnicity as one of many factors influencing, rather than the central foundation of, political behavior in India. In fact, recent studies that explicitly compare the relative salience of ethnicity and non-ethnic indicators of efficacy, find the latter to be as—if not more—significant than the former. Chauchard (Reference Chauchard2016) draws on experimental data from Uttar Pradesh, widely regarded as a bastion of ethnic voting in India, to show that while ethnicity shapes voters’ evaluations of hypothetical candidates, so does information about performance in office, knowledge about their criminal records, and overall party evaluation. At a more localized level, Auerbach and Thachil (Reference Auerbach and Thachil2018) provide experimental evidence that poor urban slum residents—often portrayed as prototypical ethnic voters—weigh markers of efficacy more strongly than shared ethnicity when selecting informal settlement leaders. Perhaps most starkly, Vaishnav (Reference Vaishnav2017) finds that many voters in the northern state Bihar cannot even identify the caste of the politician for whom they voted just days earlier.
Implications
The nuanced findings of early studies of Indian elections, further buttressed by more recent scholarship, cut against easy stereotypes of the country’s voters as solely or even primarily motivated by ethnicity. Instead, this legacy suggests several openings for exciting research in India and beyond. The first is more studies of how parties and candidates develop cross-ethnic reputations for competence within developing democracies. For example, Power and Ready (Reference Power and Ready2018) study how village leaders in Tamil Nadu construct reputations for generosity that help garner support from across caste lines. Such work can contribute to emerging comparative efforts on understanding dynamics of constituency service and credit claiming (see, e.g., Harding Reference Harding2015), that have received less attention than theories of ethnic patronage.
Second, there is considerable potential to develop more nuanced frameworks acknowledging the interplay between class and ethnicity in multi-ethnic democracies. Conceptualizing vote choices as dichotomous—either ethnic/clientelistic or programmatic/class-based—is unnecessarily limiting. Models of ethnic voting rightly pushed back against spatial models that presumed the universality of a Western-style left-right programmatic axis. However, their critiques may have been too quick to abandon class entirely, neglecting the fact that class politics need not be conceptualized solely in terms of traditional tax-and-transfer policies.
Class can inform sectoral voter preferences and political mobilization strategies in economic policy, including in the realms of targeted subsidies, agricultural prices, and other policy agendas. Again, class and ethnic politics might intersect within such strategic efforts. Anderson, Francois, and Kotwal (Reference Anderson, Francois and Kotwal2015) find that landlords from a locally dominant caste are able to wield their economic influence to ensure support for their preferred candidate from poor villagers of numerous, less privileged castes. Huber and Suryanarayan (Reference Huber and Suryanarayan2016) use group-wide caste and sub-caste voting patterns across Indian states to show that ethnic voting is greater in places with greater inter-group economic differences. In other words, there is a strong class component to ethnic voting in ways that echo the “ethnopopulism” witnessed in other parts of the world (Madrid Reference Madrid2012). Contemporary studies of earlier periods of Indian politics provide similarly intersectional insights. For example, Lee (Reference Lee2019) finds that levels of education within caste groups informs their degree of mobilization during British colonial rule, as measured by petitions for name changes submitted to the colonial government.
Finally, where models of ethnic clientelism repeatedly emphasized the lack of ideological and programmatic politics in India, future efforts can help uncover where and when ideology matters, including in service of constructing multi-ethnic coalitions. Chhibber and Verma (Reference Chhibber and Verma2018) argue that citizens’ views of who the state serves, and how it functions, have created ideological cleavages in Indian politics that cut across caste lines. The image of members of various ethnic groups joining in programmatic opposition to a rival coalition that includes their own co-ethnics is an exciting step in moving past depictions of Indian voters as mechanically assembling into caste-based vote banks.
Political Parties
Our third topic concerns the key building blocks of democratic elections: political parties. Much like parties across the developing world, Indian parties have been characterized as organizationally weak. Indeed, many Indian parties are weak according to traditional metrics.Footnote 10 Yet, emerging research amply demonstrates that they excel at two core functions: campaigning and connecting citizens to the state. These seemingly competing depictions can be reconciled by conceiving of parties as networks rather than as vertically integrated organizations. In other words, there is more than one way for a party to be “strong.” This framework incorporates what we know about Indian parties into broader discussions of party strength and offers lessons for studies of political parties elsewhere.
Core Assumptions
In comparative politics, party organization conventionally consists of two elements: institutions and infrastructure.Footnote 11 Institutions structure how power and responsibility are distributed within a party: the authority vested in particular positions and how these positions relate to one another. Party infrastructure, meanwhile, refers to a party’s “brick and mortar” presence: its working offices, paid staff, stable elite membership, and financial assets. According to these criteria, a party is said to be strong when: a) clearly delineated and consistently enforced rules allocate power and responsibility within the party and formalize the party’s relationship to civil-society based affiliates (e.g., churches and labor unions); and b) the party boasts a widespread physical presence, has sufficient personnel, a stable cadre of candidates and leaders, and ample coffers. We dub this understanding of party strength the parties-as-organizations approach. It stipulates that parties are strong insofar as they resemble a vertically integrated firm, construed as a hierarchically-structured organization capable of producing all the inputs needed to achieve the organization’s goals (Williamson Reference Williamson1971).
In contrast to this idealized characterization, scholars routinely portray parties active in the developing world as amorphous entities lacking serious organizational backbone (Lupu and Riedl Reference Lupu and Riedl2013). The focal concept in this literature is party system institutionalization (Mainwaring and Scully Reference Mainwaring and Scully1995). Party systems are considered highly institutionalized when parties have strong roots in society, voters maintain strong attachments to parties, and political elites treat parties as legitimate political actors. According to this view, party systems in poorer countries often appear weakly institutionalized—marked by high degrees of electoral volatility and personalism and low degrees of ideological linkages between parties and voters (Mainwaring and Torcal Reference Mainwaring, Torcal, Katz and Crotty2006). Reviewing the state of the party system in the Philippines, Hicken (Reference Hicken2009, 156) writes that parties “can be set up, merged with others, split, resurrected, regurgitated, reconstituted, renamed, repackaged, recycled or flushed down the toilet anytime.” Similar descriptions emerge from Latin America (Van Cott Reference Van Cott2007; Calvo and Murillo Reference Calvo and Murillo2013; Novaes Reference Novaes2018), sub-Saharan Africa (Riedl Reference Riedl2014), and post-communist Europe (Bielasiak Reference Bielasiak2002; Tavits Reference Tavits2005).
Exceptions to this pattern certainly exist. The relatively institutionalized party systems in Ghana (Riedl Reference Riedl2014) or Taiwan (Hicken and Kuhonta Reference Hicken and Kuhonta2011) are prominent examples. Moreover, even if entire party systems are not well institutionalized, individual parties may be (Randall and Svåsand Reference Randall and Svåsand2002). Brazil’s PT, or Worker’s Party, (Samuels Reference Samuels1999), as well as dominant parties in semi-authoritarian countries like Singapore, Malaysia, Indonesia, and Tanzania fall into this category (Smith Reference Smith2005; Slater Reference Slater2010). Nevertheless, these counterexamples do not undermine broader claims that political parties tend to be organizationally weak across the Global South.
Prevailing Wisdom
Judged by standard metrics employed in comparative politics, Indian political parties are undoubtedly weak, particularly compared to parties in other longstanding democracies.Footnote 12 Most Indian parties are highly centralized and run autocratically by a single leader or family and their close associates. Outside ruling cliques, titular officeholders within the party tend to wield minimal authority—a point brought out in Chandra’s (Reference Chandra2004) analysis of the Bahujan Samaj Party (BSP), Wyatt’s (Reference Wyatt2009) description of party “entrepreneurs” in Tamil Nadu, and Hansen’s (Reference Hansen2001) ethnography of the Shiv Sena in Maharashtra. Transparent rules for candidate selection and intra-party promotions are either altogether absent or widely flouted (Roy Reference Roy1965; Farooqui and Sridharan Reference Farooqui and Sridharan2014). In terms of infrastructure, too, Indian parties appear hollow. Local party offices (at the district- or block-levels) either do not exist or remain shuttered outside of election time. Parties have few permanent staff members, and party switching among politicians is rampant (Kashyap Reference Kashyap1970; Kamath Reference Kamath1985). Descriptions of numerous major parties emphasize the extent to which they are loosely organized collections of local notables or regional factions (Erdman Reference Erdman1967; Fickett Reference Fickett1976).
There are anomalous cases. For decades, India’s main communist parties possessed the trappings of classically strong parties but they have proven electorally viable in just two major Indian states.Footnote 13 The BJP’s ties to a Hindu revivalist organization, the Rashtriya Swayamsevak Sangh (RSS), have provided the party with some measure of organizational presence (Andersen and Damle Reference Andersen and Damle1987; Graham Reference Graham1990), although its organizational strength is often overstated (Manor Reference Manor, Adeney and Saez2005). Notably, India’s former dominant party, the Indian National Congress, boasted an extensive nationwide organization and well-developed internal institutions in the immediate decades after independence (Weiner Reference Weiner1967), though this atrophied from the 1960s onward (Kohli Reference Kohli1990).
Amid this landscape of organizational torpor, two stylized features of India’s partisan politics appear surprising. The first is that, come election time, Indian parties mobilize vast amounts of human and financial capital and launch vigorous campaigns requiring armies of volunteers on the ground (Banerjee Reference Banerjee2014; Björkman Reference Björkman2014; Jha Reference Jha2017; Palshikar, Kumar, and Lodha Reference Palshikar, Kumar and Lodha2017). To illustrate, the ruling BJP reportedly hired one panna pramukh (“page leader,” referring to a page of the voter list) for every thirty to sixty voters during the last two national election campaigns. Indeed, according to the 2014 Indian National Election Study, 61% of respondents reported that a member of a political campaign had come to their house to solicit their vote.Footnote 14
The second surprising feature is the scale at which parties effectively intermediate between citizens and the state. State capacity is low in India. Accessing state services and critical social safety nets is hard for most citizens. Parties supply the critical link in this unpredictable environment. Voters habitually turn to elected officials—and the parties to which they belong—for assistance in navigating the central, state, or local bureaucracy (Chopra Reference Chopra1996; Kruks-Wisner Reference Kruks-Wisner2018), and distributive politics often occurs through partisan channels (Dunning and Nilekani Reference Dunning and Nilekani2013).
In sum, Indian parties fall short when matched up against traditional notions of party strength, but nonetheless shine in performing at least two advanced tasks: waging vast election campaigns and advocating for citizens in front of the state in between elections. How can India’s parties appear both weak and strong at the same time?
New Approaches
In lieu of the dominant parties-as-organizations paradigm, this discussion suggests that India’s parties are better understood using a parties-as-networks approach. Rather than relying on formal, internal structures to achieve party goals, Indian parties typically contract out to informal social networks. According to our conceptualization, a strong party-as-network relies on a broad set of interconnected members endowed with extensive physical, financial, and human assets. A weak party-as-network, meanwhile, draws on a comparatively small network of activists, working in relative isolation with few formal assets. The informal, social-network basis of many Indian parties frequently enables efficacy in campaigning and citizen-responsiveness, even as it coexists with—and may in fact promote—a weak organizational structure.
To highlight the distinction between strong parties-as-organizations and strong parties-as-networks, consider how these different types of strong parties campaign and connect citizens to the state. When parties rely on traditional organizations, they campaign using their permanent physical infrastructure. In the case of party machines, a cadre of loyal, partisan-committed brokers develops close relationships with citizens and seeks to extract state benefits on their behalf (Auyero Reference Auyero2000; Calvo and Murillo Reference Calvo and Murillo2004).
Parties whose strength derives from informal social networks perform these functions differently. During election time, such parties campaign by mobilizing large, extant networks—for instance, groups of voters united by geography or kinship networks and often aligned with local (rather than partisan) brokers or intermediaries—that provide physical, financial, and human capital. Members’ homes become de facto campaign offices; personal wealth finances election expenditures (Sircar Reference Sircar, Kapur and Vaishnav2018); and friends, family members, and other associates engage in assorted campaign activities. These “movable” assets are the lifeblood of election campaigns, but only because network members choose to deploy them; they are not resources attached exclusively to the party. A member can transfer her wealth or followers to another party if she so desires.
A strong party-as-network can also effectively champion citizens’ material needs between elections. Citizens approach a party member for assistance, but select this “helper” more on the basis of social proximity than formal party attachment. Party members then transmit community demands upward via their parties (Auerbach Reference Auerbach2020) and draw on the full menu of social ties available to them to address those demands (Bussell Reference Bussell2019).
The strong party-as-network is particularly adept at furthering citizens’ interests in at least three ways. First, social networks transmit information to parties about citizens’ needs at low cost because citizens can more readily approach an acquaintance from their social network rather than an unknown party functionary. Second, dense networks composed of many members allow party activists to exploit the wide range of contacts often necessary to extract goods from an ill-functioning bureaucracy. Third, because citizens can exert social pressure on party leaders with whom they are in close social proximity, such party brokers are highly accountable.Footnote 15
Implications
The parties-as-networks conceptualization has several implications for future research in democracies where political parties have not followed the trajectories of Western Europe’s mass parties. For one, the presence of strong parties-as-networks calls into question the conventional wisdom that party systems in much of the developing world are inchoate and unstable (Mainwaring and Torcal Reference Mainwaring, Torcal, Katz and Crotty2006). The perpetual state of political flux observed in India may mask a surprising degree of structure and order. Often, electoral volatility exists alongside considerable stability in the groups of voters that vote with or against one another. Indeed, relative to countries in the developing world, India’s party system exhibits considerable institutionalization.Footnote 16 One potential strategy for measuring the strength of parties-as-networks in India and elsewhere may be to examine the extent to which certain types of voters—whether defined in ethnic, geographic, or other terms—routinely align with or against other groups, even if the specific parties for whom they are voting might change over time.
Additionally, future research should explore how parties build and maintain strong networks. What resources do they use to attract social networks, and how do they retain the loyalty of such networks? Some strategies—whether selective material incentives, ethnicity, ideology, or leadership charisma—may be more effective in attracting high-quality, committed workers, and instilling loyalty to the party brand. The broader party system might also inform how parties engage with their constituent networks, as parties operating in a highly competitive context may behave differently than in places where the party is electorally dominant.
Conclusion
Emerging research on electoral behavior in India challenges some of the received wisdom in the comparative politics literature and the way in which the Indian case is refracted through it. While theories of patronage politics, identity-based voting, and organizationally weak parties capture important aspects of politics in developing countries, the Indian experience helps illuminate that there is considerably more variation on each of these than we often acknowledge.
With regards to distributive politics, new research suggests that brokers and politicians in India are highly constrained in their ability to monitor voters thanks to a large, heterogeneous electorate and the relative sanctity of the secret ballot. These constraints raise doubts about whether politicians and voters can genuinely engage in a contractual quid pro quo. Many of the brokers encountered in the Indian context, moreover, are not the canonical partisan intermediaries observed in contexts of Latin American party machines. The relatively weak and volatile partisan ties of these actors generate more fluidity in vertical political linkages than is typically described in analyses of distributive politics. Further, recent studies have documented Indian villages and urban neighborhoods as intensely competitive brokerage environments, wherein multiple intermediaries continually vie with each other for a local following—both within and across party lines. Significant voter agency, coupled with competitive local brokerage environments, leaves conventional models of rigid clientelism with diminished analytical purchase in the Indian context.
Recent studies of voting behavior in India, in turn, question the dominance of accepted narratives regarding ethnic voting. Ethnicity does not always neatly overlap with political preferences; in fact, some research suggests that ethnic groups in India are remarkably heterogeneous in the expression of their political preferences. Furthermore, the extent to which ethnicity emerges as salient in voting behavior is conditioned by other types of group characteristics such as economic or social standing. In some cases, ethnicity appears to take a backseat to other electoral considerations, such as the state of the economy—a sign that the retrospective economic voting model popular in advanced industrial democracies could be at play. Another mainstay of politics in well-established democracies—constituency service—is also highlighted as something politicians prioritize and voters reward during elections.
Finally, recent studies concur with past assessments regarding the weak organizational foundations of Indian parties. However, they also question whether formal characteristics—central to the study of Western party systems—are the best metrics through which to assess the robustness of political organizations in developing countries. Research from India suggests that the tendency to measure party strength using metrics of legislative discipline or physical presence gives short shrift to alternative conceptions, such as viewing parties as rooted in social networks. Indian parties often rely on personal networks to achieve their core goals, which means that they outsource many core party functions to individuals who are not full-time party workers. This suggests that a more profitable way to study parties in India, and perhaps other developing democracies, is to examine the underlying strength of their associated social networks.
Our hope is that the findings and hypotheses discussed in the preceding pages will enrich the broader study of political behavior in the developing world. Not only does this research question the repeated invocation of India in comparative politics for arguments the country only imperfectly exemplifies, but it also suggests exciting new directions in the study of comparative politics.
Acknowledgements
This article is the product of a workshop on “Rethinking Electoral Politics in India,” hosted by the Carnegie Endowment for International Peace, Washington, DC, in September 2016 with financial support from the Hurford Foundation. We thank Devesh Kapur and Irfan Nooruddin for constructive feedback at this workshop. The authors are grateful to Rachel Osnos for helping organize the workshop and Rebecca Brown, Matthew Lillehaugen, Megan Maxwell, and Jamie Hintson for editorial and research assistance. We thank Pradeep Chhibber, Devesh Kapur, Ashutosh Varshney, the editor, and three anonymous reviewers for incisive comments on earlier drafts. All errors are our own.
