Caste in our social imagination


back to issue

THE next fifty years will produce a bifurcation in the nature of caste in urban and rural India. In urban India, it will become less salient but more fixed. In rural India, it will become more salient and more fluid. Urban Indians, in other words, will have the choice not to be defined by caste at all – but if they choose caste as a means of self-definition, they will not have much choice over which caste category defines them. Rural Indians, by contrast, will have no choice but to be defined by caste. But within the prison of caste, they will have a great deal of choice over which caste category defines them.

To see why, we must first investigate the roots of the salience of caste in Indian politics. According to a ‘primordialist’ view, outdated in specialist circles but alive and well outside, the salience of caste in India is evidence of the persistence of tradition. According to this view, caste has been part of India’s essential tradition since Rigvedic times and this is not going to change. According to a second view, the salience of caste in India is the product, not of tradition, but of modernity.1 As the Indian polity began to modernize in the twentieth century, according to this view, fragments of the past were reinvented to suit new circumstances. Caste is one of those fragments.

According to a third view, the salience of caste is the product of the policies of the colonial state – and especially the census.2 The colonial census, according to this view, and preferential policies based on it, made caste salient by giving individuals continued incentives to identify with it. Caste remains salient in post-colonial India because the post-colonial state has continued, even intensified, the practices of colonial India.

But there is a fourth view to which I subscribe.3 According to this view, the salience of caste, among other ethnic identities, in India is a product of the underlying system of patronage that has until recently driven India’s democracy in both urban and rural areas. India has for many years been a patronage-democracy – a system in which the government monopolizes access to basic goods and services valued by a majority of the population, and in which government officials have individualized discretion over how these basic goods and services are distributed. In a patronage-democracy, voters decide between politicians, not by assessing their policy positions, but by assessing whether a candidate will favour them in the distribution of patronage.

Politicians too court votes, not by promising policies – the policy differences between most political parties in India are minimal – but by promising to distribute patronage. But these promises, and assessments, take place ‘undercover’. In a modern system of government, no politician can openly promise to distribute patronage. Imagine the furore if a candidate said openly at an election rally that she would provide licenses only to Brahmin voters but not to Kurmis. Consequently, signals about patronage are best sent surreptitiously. Voters too make surreptitious assessments. And the unique visibility of caste – and other ethnic identities – makes them especially advantageous in a surreptitious world.

Consider the roots of the perception in Uttar Pradesh that the Mulayam Singh Yadav government favours Yadavs. This is a widely held perception. As one respondent in Uttar Pradesh put it: ‘The minute Mulayam Singh Yadav (from the Yadav caste) becomes chief minister, the Yadavs will put on their best clothes and show up at the door of the district magistrate, demanding that he do their work... and just to get rid of them he will do it.’4


How does the Yadav government put out that signal – and how do voters come to believe it? It is not by reading about the policies favoured by Mulayam Singh, but by assessing the principles on which he distributes patronage. And the unique visibility of caste and other ethnic identities lends these acts an ethnic interpretation, whether or not caste was relevant to them in the first place.

Take the example of a report that 720 out of 900 teachers appointed by Mulayam Singh Yadav reportedly belonged to the Yadav caste.5 Newspaper reports like this are routine in Uttar Pradesh. How might a reporter have made such an assessment? He or she could easily ascertain caste identity from the list of Mulayam Singh’s appointments. However, it would be difficult to ascertain, on the basis of such superficial information, what the other bases of these appointments might have been. No wonder, then, that calculus of voters in India is routinely described as ‘voting for the surname’.6


The importance of patronage in determining the salience of caste in India does not mean that other variables are not relevant in some way. The modernization of India’s democracy, for instance, has indeed led to the reinvention of caste as the basis of modern political coalitions. But that explanation cannot tell us why caste identities – rather than other aspects of tradition – are suitable for coalition building in modern day India. Why have other traditional forms of community (the village community for instance) not acquired the same resonance? It is in explaining this underlying resonance that patronage is key.

The policies of the colonial and post-colonial state are also important. They are necessary, not in determining the salience of caste, but in determining the categories through which caste finds expression in politics. To see how, consider the roots of the varna system, which associates four principal categories – Brahmins, Kshatriyas, Vaishyas and Shudras – in a hierarchical structure that we commonly know as the caste system. This interpretation of the caste system, as a body of work in anthropology, sociology and history has shown us, is a product of state power.7 Historically, the basic unit of the caste system in India should be seen, not as the varna, but the jati. Jatis often had a very loose correlation with the categories of the varna system. They were more often organized into local hierarchies, which differed from region to region. The colonial state, helped along by colonial ethnographers, imposed a single interpretation on these local hierarchies, organizing them into a single meta-hierarchy of Brahmin, Kshatriya, Vaishya and Shudra. This system of interpretation has persisted for almost two hundred years. But it is now in the process of being transformed. The present-day Indian state is now in the process of recreating this hierarchy as the Forward Castes, Intermediate, Backward and Scheduled Castes, with finer distinctions in between.

While the use of these categories in state-run affirmative action policies may well make caste more salient than it might otherwise be, it is not necessary to making caste salient. Caste will continue to be salient in Indian politics as long as patronage persists, notwithstanding these policies.


A key feature of the signals sent about caste – and ethnicity more generally – in patronage transactions is that there is no single correct category to which the beneficiaries of patronage can be said to belong. Instead there are many interpretations. Take the example of someone with the last name Yadav. If voters know that someone called Yadav received patronage, there is still room for many interpretations about the category to which he belongs: the name Yadav could signal that ‘Hindus’ are beneficiaries, or that ‘North Indians’ are or that ‘Backward Castes’ are or that ‘Yadavs’ are, among other options.

The stakes attached to each interpretation are high. Each interpretation produces categories of different size, with different insiders and outsiders. Thus, each interpretation has different consequences for who ends up on the winning side in an election and who loses.

Where the stakes are high, we should expect politicians to constantly manipulate the interpretation of ethnic signals to produce winning categories for themselves. Those who produce an interpretation that puts them on the winning side should seek to stabilize their interpretations. Those who end up on the losing side should try to generate interpretations that transform the outcome. Consequently, when caste is salient in politics, we should expect to see a high degree of fluidity in the actual caste categories that people use to define themselves. Conversely, where the stakes attached to the political manipulation of caste identities are not that high, we should expect categories to remain fixed. If there are no incentives attached to change in these categories, who would want to change them?


Indeed, caste politics in India has historically been marked by a very high degree of fluidity in exactly those states in which it is most salient. In Uttar Pradesh in the ’60s and ’70s, Charan Singh attempted to forge the political aggregate AJGAR out of the Ahir, Jat and Gujjar categories. Later, this old AJGAR category was supplemented by the emergence of others – most notably, the meta-category of ‘Other Backward Caste’ in 1989. The OBC aggregate, however, was just as quickly broken down by subsequent political competition. In Uttar Pradesh, parties that stood to lose from the consolidation of lower castes behind the OBC category have attempted to carve out from it the new category of the ‘Most Backward Castes (MBC)’, and the ‘Forwards among Backwards’.

In Bihar, the Rashtriya Janata Dal (RJD) initially aggregated the support of the Muslims and Yadavs into what is now commonly described as the ‘MY’ coalition. Subsequently, that MY coalition was augmented by further distinctions between the ‘extremely backward castes’ and everyone else. In each of these examples, it would have been difficult to predict in advance how politicians might combine and disaggregate ethnic building blocks into a winning coalition. And once created, it would be equally difficult to predict how long any aggregate might last.


Since the economic reforms of the 1990s, two parallel processes are transforming the Indian state: shrinking opportunities for patronage in the urban economy and expanding opportunities for patronage in the rural economy. The urban economy is dominated by the industrial sector, where patronage opportunities have declined post-1991. The government no longer sets quotas, and licenses are fewer and more transparently allocated. The state does remain involved in land transactions, in regulatory activity and in a large public sector; but it is smaller now in the urban, industrial economy than ever since independence, and it will probably continue to shrink.

The rural economy is a different story. Here, the reforms have been accompanied by an expansion of patronage opportunities associated with the state, especially during the term of the current government. The UPA, in its first term, introduced a major expansion of the state in rural areas, both through new schemes such as the National Rural Employment Guarantee (NREG) and the Rashtriya Swasthya Bima Yojana and through continued implementation of schemes introduced by previous governments, such as the NDA’s Prime Ministers Gram Sadak Yojna. However, despite the best efforts of their designers, schemes such as NREG are still riddled with loopholes in implementation. This has expanded opportunities for patronage in the rural economy; and since both major formations promise an expansion of rural state-led schemes through expanded subsidies on food, credit schemes, state-provided rural infrastructure, and a revamped public distribution system, among others, rural patronage opportunities are only likely to further expand.

This bifurcation in the importance of patronage to India’s democracy has begun to produce two parallel forms of politics in urban and rural areas: an issue-based form of politics in our cities and towns and an identity-based form of politics in our villages.


Why? Patronage is a game in which individual political candidates have a lot of power to target voters in policy implementation. Wherever the scope for patronage is large, voters need to pay attention to the attributes of individual politicians such as caste and religion and tribe. But when space for patronage shrinks, so does the importance of individual politicians. Voters are then freed to attend to the content of laws and policies in addition to their implementation. Individual attributes based on caste and religion and tribe are not entirely sidelined, but what matters more is the candidate’s position on issues, and their ability to work with others to create and implement policies based on these issues.

Small wonder then, that many urban candidates are focusing on issue-based politics in their local level campaigns: Ajay Maken in Delhi speaks of a master plan for Delhi, Meera Sanyal in Mumbai raises questions of national security policy and reform in municipal law, and Ananth Kumar campaigns in Bangalore on infrastructure development. Nor is this restricted to the metropolises: in Vishakapatnam, candidates discuss industrial pollution and the displacement of fishermen while in Chevella, near Hyderabad, the campaign is about industry and irrigation. Caste, region, and religion do come up – but when they do, it’s more about what candidates think about issues related to caste and religion, less about what their own caste and religious identities are. This is in marked contrast to election campaigns in urban areas just ten years ago.

This is not to say that caste is not important in urban areas. It remains alive and well for instance in the matrimonial ads or to social relations. But it is now being shifted into the private sphere.


In rural areas, by contrast, caste remains important in the public as well as the private sphere. In local campaigns in rural constituencies, by contrast, the identity of individual candidates matter far more. During the 1996 elections, I asked a Congress campaign manager in Western UP which issues the party was emphasising in his constituency. ‘Our candidate,’ he said, ‘is a Ghosi Yadav, while the other candidate is a Kamaria Yadav.’ ‘And what issues will your candidate emphasise?’ I asked? His reply: ‘Caste is our only issue.’

More than a decade later, election rhetoric is changing in cities and towns but appears to be the same in rural areas. In my travels in Western UP during these elections, I found the same focus on caste and religion: questions about the ‘profile’ of a constituency typically produced a breakdown by caste and religion, and questions about the campaign typically produced an account of the caste and religious identities of individual candidates. Dausa in Rajasthan is another example. In 1996 Rajesh Pilot, India’s former home minister who could have emphasized issues raised by his experience in government campaigned instead based on his identity as a Gujjar. Thirteen years later, the campaign strategies of individual candidates in Dausa remain focused on their Gujjar or Meena identities.

Rural voters have some choice in deciding which ethnic categories define them – a Lodh in Uttar Pradesh, for example, may choose to call himself a Lodh or an OBC or an MBC. Politics has been the primary medium leading to change in caste categorizations in India. As caste becomes less salient to politics in urban areas, it follows that it will also become more fixed. But in rural areas, as caste remains politically salient, we should expect to see a continuation of the high level of fluidity that has historically marked caste politics in India.


But there is virtually no choice for rural voters in whether ethnic categories define them. As one Jatav voter put it to me: ‘We are going to vote for the BSP – not necessarily because of what they have done for us but because we are Jatavs and no one else will trust us if we say we are going to vote for them.’ The continuing power of identity politics in rural constituencies is often mistakenly read as the enduring influence of tradition. It is not. It is a modern phenomenon produced by the vast reach of a patronage-based state in rural India, and will persist as long as the reach of patronage persists.

Over time, if and as the bifurcation of politics in urban and rural areas continues, we should see this bifurcation in the role of caste continue as well. If urban areas were a small part of India’s democracy, as they have been for so long, this bifurcation may not have been terribly important. India lives in its villages, Gandhi said, and politicians have usually paid far more attention to building a rural rather than an urban base. But this is changing: with high rates of urban migration, India has now begun to live in its towns and cities. And as the urbanization of India continues, we may end up with a fundamentally different nature of politics in both areas.


What does this mean for Indian democracy? One possibility is that it might lead to greater conflict. India has periodically seen mass movements such as those led by Mahendra Singh Tikait that pitch rural against urban interests. If identity-based politics now intensifies in the villages and issue-based politics in the towns and cities, urban-rural conflicts may acquire a more sustained nature. But in my view, the changing nature of India’s party system makes such conflict unlikely. The one-party system we had for the first three or four decades after independence – even one led by a party as inclusive as the Congress has historically been – would have been taxed by having to aggregate such diverging interests. But the fragmented multi-party system that we now have in India is more flexible. It easily accommodates new interests through the emergence of new parties.

As rural and urban interests diverge, therefore, we may well end up for the first time with distinct urban and rural parties, with distinct issue-based and identity-based platforms. But the institutionalization of these interests in the party system should insulate the system against conflict. The most disturbing feature of this bifurcation in the nature of our politics then, may not be the spectre of greater conflict, but the restrictions that continue to be imposed on both the economic freedom and the freedom of self-definition of rural citizens by a continuation of the politics of patronage.



1. See for instance Lloyd I. Rudolph and Susanne H. Rudolph, The Modernity of Tradition, University of Chicago Press, Chicago, 1967.

2. See for instance Bernard Cohn, An Anthropologist Among the Historians and other Essays, Oxford University Press, New York, 1987.

3. Kanchan Chandra, Why Ethnic Parties Succeed, Cambridge University Press, New York, 2004.

4. Interview, Allahabad, 14 November 1997.

5. India Today, 15 October 1994, 37, cited in Christophe Jaffrelot, ‘The BJP and the 1996 General Election.’ Paper presented at the National Conference on Political Sociology of India’s Democracy, 14-16 November 1996.

6. The Week, 18 January 1998.

7. See, for instance, Dipankar Gupta, Interrogating Caste, Penguin, New Delhi, 2000.