From formal to substantive equality


back to issue

A popular cartoon shows a child, still in his mother’s womb, thinking, ‘I hope Daddy is SC, ST and OBC’.1 One wonders if the child would still want to be born into a lower caste family if he knew that he could get killed for his choice of ringtone,2 or for plucking flowers,3 or that marriage to a girl he loved might result in either his death, or his fiancée’s death (or both), or his father losing his job.4 Examining the mainstream popular discourse on affirmative action and discrimination, it is easy to be lulled into an illusory world created by the inversion of cause and effect that is so powerfully propagated.

What is proposed as a remedy for caste-based inequalities and discrimination, viz., affirmative action (AA), is perceived in the popular imagination to cause inequalities and injustice in a world where, presumably, none existed prior to the introduction of AA. Anti-reservation outfits have names such as ‘Youth for Equality’; ironical, when juxtaposed against the actual evidence on inequality, which clearly demonstrates persistent, deep-rooted, multifaceted inequalities such that upper castes (those hurt by reservations) are on the right side of the inter-caste inequality divide; they benefit from inequality between caste groups.5


Thus, anti-reservation protests are about the ‘injustice’ of any action that might seek to alter the status quo, or one that would actually pave the way towards greater inter-caste equality, at least between the broad caste groups; by implication, the status quo is always just. The reality is that given the inequality of opportunity, AA or quotas attempt to level the playing field for different caste groups with widely disparate initial conditions, in much the same way that runners racing on a circular track have different starting positions, with the person running on the outermost track being given an ‘advantage’ of a starting line ahead of the person who is assigned the innermost track.

Given that passions run high in any discussion on AA, or more widely on the role of caste in contemporary India, I have advocated (and practiced) evidence-based research and formulated arguments that are backed by data. This essay summarizes the existing economic evidence about the nature of caste inequalities, discrimination and empirically rigorous assessments of AA in order to disentangle the chicken-and-egg question of what came first: caste inequality or reservations? The following sections discuss if the existing AA policy does ‘too much’ or ‘too little’ in the context of the underlying caste inequalities. This also contains a brief discussion of the lacunae in the existing AA policy, and whether the politics of reservations might allow us to refine and hone the policy to better serve its objectives.

The essay also touches upon a new, or perhaps not-so-new, conundrum in the AA literature: how does one view differentiation within beneficiary communities as a result of access to preferred positions, and thus increasing intragroup inequalities, while reducing between-group inequality? The bulk of the essay focuses on caste, as that is the primary arena where AA is applicable; however, given that India has gender-based political reservations, this essay briefly examines the evidence on the working of women’s reservations.


The current reservation policy goes back a hundred years, and is essentially a continuation of the quota principle that was instituted during the colonial period.6 In the early twentieth century, untouchability was recognized as a major social disability resulting in severe economic handicaps, and while quotas were opposed as vociferously then as they are today, in retrospect, quotas might appear, to some observers, as justified for a period when untouchability was rampant and caste played a more obvious role in social life than it does today. It is argued that in today’s India, caste is either dead or dying, and institutions such as reservations keep caste consciousness alive (artificially), and solidify categories that might otherwise be fluid. Therefore, it is important to assess the contemporary nature of material realities to see if caste inequality is basically a hangover from the past, and/or is confined to rural areas alone, and disappearing rapidly from urban India, which, arguably, is the face of the future.


It should be noted that macro data on caste include tribes and other communities, and are available by broad aggregates that are essentially administrative categories: Scheduled Castes, SC; Scheduled Tribes, ST; Other Backward Classes, OBCs; and every-one else, Others. The ‘Others’ category is a residual and heterogeneous one, and includes castes that might be close to the other three categories in social and economic indicators. ‘Others’ includes Hindu upper castes, and is used as a loose proxy for them, even though it goes beyond to include other communities. These broad omnibus categories are often viewed with suspicion as essentializing and missing out or eliminating the nuances of jati and sub-jati interactions.

Be that as it may, there are two significant features about these omnibus categories that should be noted: one, they enable comparative assessments of broad material indicators, and two, any comparison based on these aggregate categories underestimates the actual extent of caste disparities between the upper castes, in particular Brahmins, and the most stigmatized within the Dalit jatis. Thus, gaps revealed by this broad four-way comparison are underestimates of the true extent of caste disparities. Working with successive rounds of the National Sample Survey (NSS) or Census data one had always suspected this to be the case; the availability of data from the India Human Development Survey (IHDS), which provides data for Hindu upper castes, and specifically about Brahmins confirms this suspicion.7

Disparities between caste groups in material indicators are common knowledge. The more interesting question is that of changes over time. Are the gaps increasing or decreasing over time? That would provide empirical verification to the belief that caste is either dead or dying. It turns out that not only are caste disparities, as evidenced from this four-way comparison, significant in contemporary India; over time, the trend has not necessarily been of convergence across caste groups for all indicators. We find that in terms of basic literacy and primary education, caste groups have converged towards one another. However, across all other educational categories, the gap between ‘Others’ on the one hand, and SC-ST-OBCs on the other hand, has not closed in the post-independence period.


In order to gauge inter-generational transmission of educational status, a key facet of educational advantage or disadvantage, we compare the effect of fathers’ education level on their sons’,8 and find that while for all groups the effect of fathers’ education in determining sons’ education levels has gone down (which demonstrates the possibility for inter-generational mobility), the effect is the largest for the ‘Others’, where the fathers are more educated than the fathers in the other three social groups.9 Thus, sons of upper caste fathers getting an education level higher than their fathers drive the increase in inter-generational educational mobility observed in the last few decades in India.


In occupational indicators too, the trend is similar, except that younger cohorts of OBCs are converging with their counterparts from the ‘Others’ group in terms of access to white-collar jobs (i.e. Others continue to have a higher proportion in white-collar jobs compared to OBCs, but the gaps are becoming smaller). There is no convergence between SC-STs and ‘Others’. Similarly, in land ownership, monthly per capita expenditure, wages and other key material indicators, the picture is predominantly of divergence (gaps becoming larger), rather than convergence (gaps becoming smaller).

All of this is as true in urban areas as it is in rural. Recall that these broad comparisons underestimate the true extent of inter-caste disparity. We started this section with how untouchability was a major social issue at the start of the last century. Where do we stand vis-à-vis untouchability, a practice that has been illegal for the last 65 years, at the start of the 21st century? Thorat and Joshi, based on self-reported answers from the IHDS data, show that in 2011-12, 20 per cent of all urban households, 52 per cent of Brahmin households and 24 per cent of households with at least one adult with graduate or more education practiced some form of untouchability.10


Several commentators are concerned about caste disparities in material indicators, and in this context AA is often criticized for not reducing the gaps between caste groups in say poverty levels, or at lower educational levels. I believe this criticism is misplaced. This is because AA is not, and cannot be, a cure-all policy to end all manner of caste disparities. The gaps are too many, and too multidimensional, to be cured by one single policy instrument. First, we see disparities in the presence of AA, but we cannot assess what these might have been in the absence of AA. It is entirely possible that the gaps might have been larger. Second, and more importantly, AA is a policy meant to desegregate the elite. By providing access to higher education and to public sector jobs to groups that are marginalized, stigmatized and discriminated against and, therefore, are likely to be under-represented in these positions, AA makes the composition of the elite more representative of the underlying social composition, thus giving groups a potentially larger say in decision making and governance.

Another argument against AA is that by providing quotas in both education and jobs we are doubling the ‘advantage’ for beneficiary groups. Why not have quotas for education so that the playing field is levelled, and then let everyone compete on an equal footing in the job market? Indeed, if equal access to education could level the playing field, this would be the way to proceed. However, the reason AA is needed in employment is the presence of discrimination in job markets that results in equally qualified job aspirants being treated differently in the job market, such that upper caste candidates are much more likely to get superior and better paying jobs compared to their Dalit peers. A set of four studies attempting to uncover pathways of caste discrimination in urban, formal sector labour markets (Thorat and Attewell,11 Madheswaran and Attewell,12 Deshpande and Newman13 and Jodhka and Newman14) found that getting access to education, or graduating from a good university does not eliminate labour market discrimination. Employers are conscious of the complex and overlapping categories of caste, class, religion and family background. They pay lip service to ‘merit’: in reality, they are completely blind to the unequal playing field that produces ‘merit’ in the first place and both consciously and unconsciously use social stereotypes to assess the merit of job market candidates.


The most common objection to AA is that it conflicts with the consideration of merit, and thus introduces mediocrity. Additionally, it is criticized on the grounds that it places poorly qualified candidates from beneficiary groups into positions to which they are ill suited, and are therefore likely to drop out, and thus does more harm than good (the ‘mismatch hypothesis’). An analogous argument states that AA negatively impacts the productivity or efficiency of enterprises, as it replaces more efficient workers with less qualified and therefore less efficient ones. Another view believes that by targeting higher education and government jobs, it is too little, too late.

Deshpande assesses all these claims in detail, including problematizing the singular notion of ‘merit’ as a single number to be captured by marks in an entrance examination.15 The notion of AA conflicting with merit assumes that labour markets in the absence of AA are meritocratic, which they are not. Empirically rigorous evaluations find these claims to be invalid. Bertrand, Hanna and Mullainathan show how the mismatch hypothesis does not hold in education16; Deshpande and Weisskopf demonstrate that productivity in enterprises is not negatively affected by AA17; Cassan shows how AA in higher education led to an increase in educational attainment at the primary level.18 Pande finds that political reservations increase transfers towards beneficiary groups.19 This is important as these are groups, which are likely to be marginalized even in allocation of funds.


Similarly, studies that assess reservations for women in panchayats and urban local bodies find that political reservations lead to an altered focus in policy making, e.g. panchayats that are led by women tend to invest more in drinking water, compared to roads, as it is an acute problem for women.20

As far as the claim of ‘too late’ goes, interventions during the early education years, and AA at the higher education level are not mutually exclusive. A holistic approach to redress inequalities could conceive of interventions at different levels that are compatible with each other. Additionally, we saw earlier that AA in higher education and public sector jobs fulfils a separate purpose – that of desegregating elites.


Could AA be a one-shot answer to caste inequalities and discrimination? The simple answer is ‘no’, and therein lies the rub. The continued presence of disparities is often used as a stick to beat AA with, but precisely because it is not designed to solve all material disparities, it is unfair to expect it to be a magic wand. I believe the biggest weakness of the current policy is that the announcement of the policy is the beginning and end of AA. There is no assessment (by the government) if it is being implemented adequately across all states and at the Centre, and there are no penalties for non-compliance. Thus, institutions can carry on with unfilled quota vacancies.

If the problem is a lack of unqualified candidates to fill seats, there is no thought given to how a pool of qualified candidates might be created. Finally, giving individuals access to jobs and educational institutions is only the beginning. Given the accumulated disadvantage, especially for first generation beneficiaries, provisions for special remedial classes, on-the-job training and other features are needed that make quotas effective.

Finally, ‘outside the box’ measures must be considered that go beyond the scope of the current AA programme: free, compulsory and good quality primary education, provision of basic health facilities, adequate nutrition, vigorous expansion of non-farm employment, land reforms wherever feasible, subsidies/support for Dalit business/self employment and establishment of a strong anti-discriminatory framework. Given that scavengers are the most stigmatized of all Dalit communities, the elimination of the degrading practice of manual scavenging is a must. However, it is important to recognize that this can only happen when we move towards a modern sewage system, including toilets. This will not only provide the basis for de-stigmatizing a large community, but will also provide significant health benefits. All these measures will benefit a much larger section of Dalits than the current AA programme.

The political discourse on AA, while it is strongly focused on proportions of seats (i.e. the extent of quotas) and claims and counterclaims from individual groups, does not pay adequate attention to the interconnections between these various elements that would make quotas more effective.


Could social mobility due to AA increase inequality? Given the nature of quotas, it is natural that the better placed among beneficiary groups are able to avail of quotas. An illiterate poor person cannot be expected to enter college, regardless of whether quotas exist. Assessments of AA have found that several beneficiaries, even when they were better-off compared to other Dalits in their communities, would not have been able to access higher education at all in the absence of quotas.21 Thus, AA has enabled them to pursue occupations very different from their parents, and break the historical link between their caste status and deeply degrading and stigmatizing work. This is precisely the social mobility that AA is expected to provide.

Given that a large proportion of Dalits continue to be mired in conditions where studying up to the point where they could possibly become eligible for quotas is a pipe dream, this could lead to differentiation within beneficiary communities. This is also often used to berate AA by invoking the ‘but what about the poor, illiterate, rural Dalit’ argument. As we noted above, neither AA nor any other single policy measure, can possibly be that magic wand to solve all caste related problems.


Summing up, caste inequality is alive and present in contemporary India, and its presence in the private sector indicates that AA could not be causing it. AA is a necessary element of a redress mechanism, but is not sufficient. We need several other supplementary measures in place. The important thing to note is that the existing AA programme and these supplementary measures need not be considered mutually exclusive. They can strengthen and reinforce each other. What is needed is a strong political will to end caste discrimination and reduce caste disparities. Admittedly, there would be costs to all these measures, but the benefits of integrating large sections of nearly 160 million Dalits, thereby unleashing the suppressed reservoir of talent, is the need of the hour for a whole range of reasons – for equity, for efficiency, for a just and humane society – and even if none of these matter, then simply for higher growth.


* Ashwini Deshpande is the author of Grammar of Caste: Economic Discrimination in Contemporary India, Oxford University Press, Delhi, 2011, and Affirmative Action in India, OUP/OISI, 2013.


1. (accessed 20 May 2015).

2. (accessed 23 May, 2015).

3. beaten.htm (accessed 20 May, 2015).

4. (accessed 20 May, 2015).

5. Since AA in India predominantly takes the form of quotas or reservations, the two terms will be used interchangeably in the Indian context.

6. For a brief history of the reservation policy, see Ashwini Deshpande, Affirmative Action in India. Oxford India Short Introductions, Oxford University Press, New Delhi, 2013.

7. Sonalde Desai and Amaresh Dubey, ‘Caste in 21st Century India: Competing Narratives’, Economic and Political Weekly 46(11), 2011, pp. 40-49.

8. The NSS collects data on co-resident members of a given household (defined as a unit with a common kitchen). The data are organized by households, not families, and do not allow tracing members of a family if they have left the parental household. We can only identify father-son pairs residing in the same household. Since daughters typically marry early and move to the marital home, NSS data does not allow us match daughters with either fathers or mothers, after they have left. Most resident daughters are minors, with many still in school, so the survey cannot determine their ultimate educational level. This constraint is less severe for men, as the joint family system is very widespread, and adult men often live in the same households with their parents, spouses and children.

9. Ashwini Deshpande and Rajesh Ramachandran, ‘How Backward are the Other Backward Classes? Changing Contours of Caste Disadvantage in India’, Centre for Development Economics, Working Paper No. 233, 2014.

10. Amit Thorat and Omkar Joshi, ‘The Continuing Practice of Untouchability in India: Patterns and Mitigating Influences’, mimeo, 2015.

11. Sukhadeo Thorat and Paul Attewell, ‘The Legacy of Social Exclusion: A Correspondence Study of Job Discrimination in India’, Economic and Political Weekly 42, 2007, pp. 4141-4145.

12. 12. S. Madheswaran and Paul Attewell, ‘Caste Discrimination in the Indian Urban Labour Market: Evidence from National Sample Survey’, Economic and Political Weekly 42(41), 13 October-19 October 2007.

13. Ashwini Deshpande and Katherine Newman, ‘Where the Path Leads: The Role of Caste in Post University Employment Expectations’, Economic and Political Weekly 42, 2007, pp. 4133-4140.

14. Surinder Jodhka and Katherine Newman, ‘In the Name of Globalization: Meritocracy, Productivity, and the Hidden Language of Caste’, Economic and Political Weekly 42, 2007, pp. 4125-4132.

15. Ashwini Deshpande, Grammar of Caste: Economic Discrimination in Contemporary India. Oxford University Press, Delhi, 2011.

16. M. Bertrand, R. Hanna and S. Mullai-nathan, ‘Affirmative Action in Education: Evidence from Engineering College Admissions in India’, Journal of Public Economics 94(1-2), 2010, pp. 16-29.

17. Ashwini Deshpande and Thomas E. Weisskopf, ‘Does Affirmative Action Reduce Productivity? A Case Study of the Indian Railways,’ World Development 64, December 2014 (2015), pp. 169-180.

18. Guilhem Cassan, ‘Affirmative Action, Education and Gender: Evidence from India’, mimeo, 2014.

19. Rohini Pande, ‘Can Mandated Political Representation Increase Policy Influence for Disadvantaged Minorities? Theory and Evidence from India’, American Economic Review, 93 (4), 2003, pp. 1132-51.

20. Raghabendra Chattopadhyay and Esther Duflo, ‘Women as Policy Makers: Evidence from a Randomised Policy Experiment in India’, Econometrica 72(5), 2004, pp. 1409-43; T. Besley, R. Pande, L. Rahman and V. Rao, ‘The Politics of Public Good Provision: Evidence from Indian Local Governments’, Journal of the European Economic Association 2(2-3), 2004, pp. 416-426.

21. Op. cit., fn. 7.