The caste based mosaic of Indian politics


back to issue

WHILE caste politics all over India during the 1990s has been dominated by the Other Backward Classes (OBC) phenomenon, these aggregates – the OBC and India – have gradually lost their relevance, with individual jatis and states becoming the real units of analysis. Many of the regional parties, which now represent half of the Indian voters, are associated with a single caste and/or religious community – and this is also true of some state units of national parties. Such a development is largely due to the fact that the states themselves are often identified with a few jatis. This is the logical outcome of the redrawing of the map of India according to the linguistic criterion, a process which started in the 1950s. One could even argue that this redrawing of the frontiers of the state in large part stemmed from the demands of dominant castes.

In 1947, India had inherited from the British Raj provinces whose borders had questionable cultural coherence. Most of them had been decided arbitrarily and gathered together several linguistic groups. Though the demand for linguistic states was spelt out primarily in cultural terms – linguistic communities asked for homogenous entities for the sake of communication and education – it was mostly articulated by the dominant castes which had spread throughout the area/domain of one vernacular language. As a result of the redrawing of the Indian map, the Marathas, for instance, became the dominant caste of Maharashtra.1 However, in most states, there were usually two or more dominant castes. In Andhra Pradesh, Kammas and Reddys were locked in some rivalry. Similar, in Karnataka, was the case of the Vokkaligas and the Lingayats, who opposed each other.

After the redrawing of state boundaries, these castes often tended to align themselves with one party. For instance, the Kammas opted for the Communist Party of India while the Reddys favoured the Congress, so much so that while the CPI became the brand name, the sub-text was ‘Kamma’. After some time, a few of these dominant castes decided to form their own party. Kamma leaders, to continue with this example, founded the Telugu Desam Party in the early 1980s, the very reference to the state language reflecting the claim of the party to represent Andhra Pradesh at large. The party chief, N.T. Rama Rao, a popular film actor, won the elections for the first time in 1983. But the two party system constituted by the Reddy-supported Congress and the Kamma-supported TDP was disturbed by the emergence of a newcomer in 2008, the Praja Rajyam Party, in turn supported by a caste benefiting from rapid upward social mobility, the Kapus.


Once again a popular film actor, Chiranjeevi, was responsible for launching a party that was conceived as a caste-based electoral machine. ‘Primarily, it emerged out of the aspiration of the elite among the Kapu community to see their man at the helm of the state government. In recent times they seem to have nurtured a grievance that under the rule of the Congress and the TDP, only two communities, namely, the Reddys and the Kammas, exercised supreme political power although the Kapus enjoyed analogous social status and were numerically as large as these communities. They were resentful of the asymmetry between their rise in social and economic prominence and their role as junior partners in power.’2 

A few months after its creation, the PRP was able to win 16% of the valid votes polled in Andhra Pradesh in 2009, mainly due to the support of the Kapus. 59% of them voted for the PRP, 59% of the Kammas voted for the TDP, and 59% of the Reddys voted for the Congress,3 a graphic illustration of the role of caste groups behind the electoral performance of parties at the state level, despite claims that their ideology and politics are universalistic or multicultural.

None of the castes associated with political parties, however, vote en bloc for one party; 59% being far from 100%. There are many reasons why so many chose to vote for a party not identified with their caste. First, in some constituencies, two parties or more might have nominated a candidate of the same caste, a common practice in constituencies where intra-caste factionalism is pronounced. Second, citizens may choose a party which is not associated with their caste simply because they like it – since governance related issues (among others, charges of corruption, programmes and past record) do influence individual choice.


Third, factions play a significant role within a caste group, especially at the local level. In Uttar Pradesh, G.K. Lieten noticed that among the Jats – one of the state’s dominant castes – ‘the vote is decided collectively, according to the general tendency in the Jat "biradari" (local sub-caste).’4 Lieten even shows that in 1991, the Jats delinked their vote to the state assembly from their vote to elect Members of Parliament at the Centre in order to keep the two main factions of the caste happy. Thus, they decided to vote for the Janata Dal at the general elections, and for the Congress in the state elections. Fourth, as James Manor points out, ‘Shrewd politicians (...) constantly engage – often simultaneously – in efforts to divide social groups (including jatis and jati-clusters) that oppose them and to unite those that offer (or may offer) them support.’5 One of the techniques used by these politicians to divide the vote of castes, which they know are not supportive, consists of promoting (and financing) a ‘dummy’ candidate of the same caste as their main rival(s).6 

But the general pattern of voting behaviour in India suggests that as recently as 2009, ‘caste or community continues to be the primary building block of political affiliation at the micro level.’7 Indeed, there are many states in which the majority of a caste group aligns itself to one party. This is more true in states where a two-party system is well entrenched. In Rajasthan, 74% of the Brahmins, 59% of the Jats and 55% of the Rajputs voted for the BJP while 66% of the Dalits voted for the Congress.8 In Punjab, 58% of the upper caste Hindus voted for the Congress and 36% of them for the BJP and its Sikh ally, the SAD (Shiromani Akali Dal). For the Jat Sikhs, it is the reverse: 37% voted for the former and 54% for the latter.9 In Bihar, the Yadavs continued to vote for the RJD of Laloo Prasad Yadav, whereas the upper castes stuck to the BJP and its state ally, the JD(U): in six elections out of eight, between 63% and 77% of them have supported this coalition.10 


One may wonder why and how a caste somewhat aligns itself with a party. The ‘how’ can be elucidated more easily. At the local level, especially in a rural context, opinion leaders who are better educated than the average peasant help to shape collective forms of voting behaviour, as James Manor pointed out in his study of the 1980 elections.11 But members of one caste may decide to vote together for one party/candidate without any hesitation because they share not only the same identity but also the same interests. Paul Brass assumes that ‘members of the same caste often carefully calculate how their votes may best serve their collective advantage within any given constituency, and they tend to coalesce with other castes in order to achieve that goal.’12 


How can we define this ‘collective advantage’? To answer this question implies a shift from the ‘how’ to the ‘why’. We’ll try to respond by focusing on the lower castes which form the bulk of the voters and which tend to have a higher turnout than the average.13 We need first to understand what elections mean for these plebeians. In her ethnographic work in West Bengal, Mukulika Banerjee convincingly argues that elections offer them ‘an opportunity for the expression of citizenship, and for an understanding of the duties and rights involved in living in a democracy.’14 Banerjee emphasizes that ‘the egalitarian mechanics of the poll afforded (them) particular pleasure.’15 But, besides the symbolic gratification of standing in the same queue as dominant caste members, rights are at stake. One of her interviewees said: ‘The fact that I can cast my vote is a right. But we have also got our rights through voting.’16 What rights?


In the villages where she did field-work to understand the resilience of the communist domination, Banerjee discovered not only that the land reform implemented by the CPI(M) since its first electoral victory in 1977 had an economic effect – no more famine – but also that ‘the social effect was even more dramatic, genuinely transforming the feudal character of labouring relations, with the lower castes becoming increasingly self-assured. In my research villages, the Syeds (upper caste Muslims) themselves were driven by the new labour costs to sweating in the fields, a heavy symbolic blow given their traditional abhorrence of the plough and manual labour. The sharecroppers could now look them in the eye when talking. Previously, even riding a bicycle in front of a Syed was considered taboo.’17 

In such villages, low caste Hindus and Muslims voted together for the CPI(M) and its allies in order to obtain the symbolic as well as material advantages that land reforms were bound to bring them. The same development is now taking place among the Dalits, not only at the state level – as evident from the case of UP – but at the national level. Our final case study focusing on the BSP will help us respond even better to the ‘how and why’ question.

The building of a pan-Indian Dalit party had been the goal of B.R. Ambedkar for twenty years. He successively initiated the Independent Labour Party (1935), the Scheduled Castes Federation (1942) and the Republican Party of India (1956). But none of them could make an impact, largely because individual Dalit jatis were not prepared to join hands and together support the same party. In fact, the Mahars – Ambedkar’s caste fellows – were the only ones who tended to support his party.


The Bahujan Samaj Party (the party of the masses), which was founded in 1984 by a militant Ambedkarite, Kanshi Ram, has gradually overcome this handicap. In the course of the last two decades it has emerged as a full-fledged Dalit party, largely because it was in a position to cash in on the development of a Dalit counter-culture (including a poignant literature) and programmes of positive discrimination which have given birth to a Dalit middle class whose dedicated members were keen to organize themselves and their caste fellows.

In its stronghold of Uttar Pradesh, where the BSP could rely on the Backward and Minority Communities Employees Federation (BAMCEF), a movement of mostly Dalit public sector employees that Kanshi Ram had organized from the 1970s onwards, the BSP became the third largest party in 1989. It then benefited from being a coalition partner of the Samajwadi Party at the helm of the government of Uttar Pradesh in 1993-95. In this capacity, it advocated policies intended to promote the interests of the Dalits at large, including the Ambedkar Village Scheme.18 Mayawati, the BSP leader, repeated the same strategy when she became chief minister of UP with the support of the BJP in 1996-97 and 2003.19 Eventually, the BSP won a majority of the seats in the state assembly in 2007.


Whenever it has been in office in UP, the BSP has combined both symbolic and substantial measures. On the one hand, it has given the name of Dalit personalities to districts and stadiums and built a large number of Ambedkar statues. On the other hand, it has ‘accomplished the fullest implementation of quotas ever achieved in the state.’20 It has also made decisions which are both symbolic and substantial, like the strict implementation of the Scheduled Caste and Scheduled Tribe (Prevention of Atrocities) Act (1989), which has deeply transformed social relations at the local level.

In her remarkable doctoral thesis on the BSP in East and West UP, Sohini Guha cites one of her Rajput informers as follows: ‘Earlier, these low caste men would stand up when they saw us (i.e. Thakurs) coming from a mile away. Now, forget standing up, they invite us to sit on their charpai (string beds) and drink tea! The cheek of it! Times have changed, these days are not good for us. We have to swallow these insults. Earlier, we would have knocked their teeth out, or worse, for this behaviour.’21 


The BSP has gradually reinforced its standing among Dalits at large, so much so that it is not, any more, a party of one jati, the Chamars (the caste of Mayawati), but a party of the majority of UP Dalits and even a party for Dalits living out of Uttar Pradesh, as evident from the accompanying table.

The Dalit Vote for the BSP in Seven States in 2009 (in %)


% of Dalit Votes







Madhya Pradesh

Jatavs: 27

Other Dalits: 6


Mahars: 15

Buddhist Dalits: 37

Other Dalits: 9


Hindu Dalits: 21

Sikh Dalits: 14

Uttar Pradesh

Jatavs: 85

Pasis: 64

Other Dalits: 61

Source: Rahul Verma, ‘Dalit Voting Patterns’, Economic and Political Weekly XLIV(39), 26 September 2009, p. 97.

Since no individual Dalit jati represents more than 20% of the Scheduled Castes in any state of India, the figures suggest that in Chhattisgarh, Delhi and Haryana, non-Chamars (or Jatavs) Dalits have also voted for the BSP in 2009. The same table makes this point explicit in the case of Madhya Pradesh, Maharashtra, Punjab and Uttar Pradesh where the party is now a full-fledged Dalit party since more than 60% of all Dalit jatis, which used to be locked in rivalry (especially the Chamars or Jatavs and the Pasis or Dusadh) now together largely support the BSP.

With 21% of the Dalits voting for the BSP (against 27% voting for the Congress), the party has become the third largest political party in India, ahead of the Communist Party of India (Marxist) in 2009. This is the first caste-based and caste-oriented party to climb on to the podium of national parties. (To be recognized as a ‘national’ party – and not merely as a state party – a party needs, among other things to win more than 5% of the valid votes; the BSP receiving more than 6% of the valid votes in 2009.)

In February/March of this year, the BSP was shaken by a serious setback. Certainly its defeat is not as dramatic as its loss of seats suggests: while the party tally was reduced from 206 to 80 seats, its share of valid votes remained very high. In fact, the BSP’s performance was only reduced by 4.5 percentage points from 31.5 to 26 % of the valid votes. But the disturbing dimension of this result – if one goes by the CSDS surveys – lies in the declining popularity of the BSP among its Dalit supporters, including the Jatavs (from 86 to 62%) and the Balmikis (from 71 to 42) – interestingly, the Pasis rallied behind the BSP in larger numbers (from 53 to 57).22 In spite of this setback, the BSP remains clearly identified with the Dalits, and more especially the Jatavs.


Our final case study about the BSP allows us to draw three conclusions: First, Dalits tend to vote for their caste for both symbolic and substantive reasons. This example shows that the distinction between a rational choice-oriented approach and an analysis emphasizing the role of symbols to explain voting behaviour is not tenable. Certainly, Dalit voters are attracted to the BSP because of the pride they derive from one of their own being at the helm of the largest state of India but, equally, they would stop supporting Mayawati if she failed to deliver any material benefits. Interestingly, while Kanchan Chandra considers the rise of the BSP in UP as emblematic of the Indian brand of ‘patronage democracy’,23 she assumes that ‘individual voters and elite in patronage-democracies are motivated by a desire for either material or psychic goods or some combination of both.’24 The case of the Dalits’ relation to the BSP shows that the debate between the proponents of identity politics and ‘patronage democracy’ is largely artificial, somewhat expected in a society where caste and class coincide to a great extent and where, therefore, the plebeians are interested in material benefit and symbolic recognition in terms of status.


Second, this case study enables us to state our response to the ‘how and why’ question more systematically in respect of the techniques a party may use to attract caste groups. The BSP has not been identified by the Dalits as ‘their’ party in UP only on the basis of the social profile of its candidates. In 2009, out of 80 BSP candidates only 18% were Dalits. The Dalits voted for the party both because its leader was a Dalit, and because its programme was ‘pro-Dalit’. In fact, most of the ‘caste parties’ are placed in this category because of the caste of their leader, as is evident from the ‘Yadav parties’. But the BSP played an additional caste card in 2009 by nominating a record number of Brahmin candidates (20%) in order to send a signal to this caste group which had joined hands with the Dalits against the Yadav-Rajput alliance in 2007. This shows that an affinity between one party and caste groups can come from the personality (and programme) of its leader more than from the social profile of its candidates.


Third, the case of the BSP in UP suggests that the moment a caste group successfully supports a caste-oriented party, the other castes and parties are forced to adjust. The fate of the Congress has been partly sealed in UP and Bihar – at least in state elections – because of its eagerness to remain a catch-all party. While the BJP did not openly speak against the Mandal report, its studied silence and elitist profile led the upper castes to rally around the party in larger numbers to build a counter force. As a result, the Brahmins left the Congress in the 1990s to support the BJP.

The transformation of castes into interest groups, resulting in the erosion of the caste system, mostly under the influence of a process of ethnicization and positive discrimination programmes, has made a strong impact on the Indian political system. Certainly, while the two largest parties – the Congress and the BJP – are not associated with an individual caste (or even a grouping of castes) at an all India level, they usually are at the state level where regional parties are even more related to specific jatis. In fact, state politics is, in part, taking over national politics because of the decline of mainstream parties vis-à-vis regional parties which are often associated with one caste, especially with one dominant caste. Besides state politics, the pervasiveness of caste considerations also stems from the emergence of a-more-than-one-state-and-one-jati party, the BSP.

While the 2009 general elections have been analyzed as marking a re-nationalization of Indian politics, this is clearly a symptom of an ‘interpretative deficit’ that Palshikar and Yadav regretted in their own account of these polls.25 In fact, the success of the Congress which gave rise to this incorrect interpretation was itself a result of the growing fragmentation of the party system, the largest party being in a position to win with less votes when it faces more opponents in a first past the post electoral system.26 And interestingly, several of the new parties are caste based, like the PRP, which in Andhra Pradesh was intended to give voice to an upwardly mobile caste, the Kapus.


All the members of one caste never vote together because of factionalism, dummy candidates and the (growing?) importance of issue based politics at the expense of identity politics and ‘patronage democracy’. But why do they vote together to such a large extent? The case of the lower castes, and more especially of the Dalits, suggests that they do so for symbolic as well as substantive reasons. Here, to distinguish identity politics from interest related motivations is virtually impossible. Such a conclusion was something we could expect. It derives logically from our analysis of the transformation of castes into ethnic interest groups over more than a hundred years: since the late 19th century, the ethnicization of caste reflects the quest for status related recognition while quota oriented mobilizations are articulated in class terms – voting behaviour was bound to largely flow from these societal patterns.



1. Something Dr. Ambedkar had anticipated and feared.

2. K.C. Suri, P. Narasimha Rao and V. Anji Reddy, ‘Andhra Pradesh: a Vote for Status Quo?’ Economic and Political Weekly XLIV(39), 26 September 2009, pp. 110-111.

3. Ibid., p. 112.

4. G.K. Lieten, ‘Inclusive View of Religion: a Rural Discourse in Uttar Pradesh’, Economic and Political Weekly, 8 June 1996, p. 1412.

5. James Manor, ‘Caste and Politics in Recent Times’, in Rajni Kothari (ed.), Caste in Indian Politics. Orient Blackswan, Hyderabad, 2010 (second edition), p. xxxvii.

6. See Christophe Jaffrelot, ‘Indian Democracy: The Rule of Law on Trial’, Indian Review 1(1), January 2002, pp. 77-121.

7. Yogendra Yadav and Suhas Palshikar, ‘Between Fortuna and Virtu: Explaining the Congress’ Ambiguous Victory in 2009’, Economic and Political Weekly XLIV(39), 26 September 2009, p. 38.

8. Sanjay Lodha, ‘Rajasthan: Performance and Campaigning Paid Dividends’, Economic and Political Weekly XLIV(39), 26 September 2009, p. 189.

9. A. Kumar and J.S. Sekhon, ‘Punjab: Resurgence of the Congress’, Economic and Political Weekly XLIV(39), 26 September 2009, p.185.

10. S. Kumar and R. Rajan, ‘Bihar: Development Matters’, Economic and Political Weekly 39, 26 September 2009 , p. 143.

11. These ‘school teachers, clerks, voluntary association and cooperative society workers, development board members or employees or – apparently less often – small lawyers’ usually ‘had assisted villagers in dealing with supra-local authority in some way, whether in minor matters like filling in form or in more serious cases defending villagers interests...’. J. Manor, ‘The Electoral Process Amid Awakening and Decay: Reflections on the Indian General Elections of 1980’, in P. Lyon and J. Manor (eds.), Transfer and Transformation: Political Institutions in the New Commonwealth. Leicester University Press, Leicester, 1983, p. 98.

12. Cited in J. Manor, op. cit., p. xxxiii.

13. In India, the elite groups vote less often than the plebeians. On this atypical Indian pattern see C. Jaffrelot, ‘Why Should We Vote?’ The Indian Middle Class and the Functioning of the World’s Largest Democracy’, in C. Jaffrelot and Peter van der Veer (eds.), Patterns of Middle Class Consumption in India and China. Sage, New Delhi, 2008, pp. 35-54.

14. M. Banerjee, ‘Sacred Elections’, Economic and Political Weekly, 28 April 2007, p. 1561.

15. Ibid., p. 1560.

16. Ibid., p. 1561.

17. M. Banerjee, ‘Democracy, Sacred and Everyday: An Ethnographic Case From India’, in Julia Paley (ed.), Democracy: Anthropological Approaches. SAR Press, Santa Fe,2008, p. 82.

18. See Sudha Pai, Dalit Assertion and the Unfinished Democratic Revolution: The Bahujan Samaj Party in Uttar Pradesh. Sage, New Delhi, 2002.

19. C. Jaffrelot, ‘The BSP in Uttar Pradesh: Whose Party is it?’, in S.M. Michael (ed.), Dalits in Modern India: Vision and Values. Sage, New Delhi, 2007, pp. 260-286.

20. Sohini Guha, ‘A Symmetric Representation and the BSP in UP’, Seminar 571, March 2007, p. 59.

21. Cited in S. Guha, Ethnic Party, Material Politics and the Ethnic Poor: The Bahujan Samaj Party in North India. PhD dissertation, McGill University, 2008, p. 99.

22. ‘Special Statistics: 2012 State Elections’, Economic and Political Weekly XLVII(147), April 2012, p. 81.

23. That is ‘a democracy in which the state monopolizes access to jobs and services, and in which elected officials have discretion in the implementation of laws allocating the jobs and service at the disposal of the state.’ Kanchan Chandra, Why Ethnic Parties Succeed: Patronage and Ethnic Head Counts in India. Cambridge University Press, Cambridge, 2004, p. 6.

24. Ibid., p. 11.

25. Yogendra Yadav and Suhas Palshikar, op cit., 26 September 2009, p. 33.

26. For more details on this point see C. Jaffrelot and G. Verniers, ‘India’s 2009 Elections: The Resilience of Regionalism and Ethnicity’, SAMAJ 3, 2010.