Reservation is the most argued-about social policy in India, and also one of the least read about honestly. Supporters treat any criticism as an attack on social justice. Critics treat any defense as a denial of merit. Both sides tend to argue from position rather than from data. This article does something narrower and more useful. It asks what the measurable record actually shows: where reservation moved real numbers, where it did not, and which questions remain open after seventy-five years.
This is the first part of a five-part series. It sets the factual baseline. The later parts examine the specific reform debates that follow from this record: the creamy layer, caste versus economic criteria, how other countries approached group-based disadvantage, and the foundations that quotas alone cannot build. The aim across all five is not to reach a verdict for or against reservation, but to describe the policy accurately enough that a reader can form a considered view.
What reservation is, and how it took its present shape
Reservation is a constitutional system of quotas in education, public employment, and legislatures for groups the Constitution recognises as historically disadvantaged: Scheduled Castes (SC), Scheduled Tribes (ST), and Other Backward Classes (OBC). The framework began with the Constitution in 1950, which permitted the state to make special provision for the advancement of socially and educationally backward classes.
For the first four decades, reservation applied mainly to SCs and STs. The major expansion came in 1990, when the government accepted the Mandal Commission recommendation to reserve 27 percent of central government posts for OBCs. That decision triggered the most consequential legal challenge in the policy’s history.
In 1992, a nine-judge bench of the Supreme Court decided Indra Sawhney v. Union of India. The judgment shaped almost everything that followed. According to the Supreme Court’s record and legal summaries, the Court did three things that still govern reservation today. It upheld the 27 percent OBC quota. It fixed a ceiling of 50 percent on total reservation, except in extraordinary circumstances. And it introduced the idea of the creamy layer, directing that the socially advanced and wealthier members of the OBCs be excluded so that benefits reach the genuinely backward. The 1992 judgment also held, at that time, that reservation should apply to initial appointments and not to promotions, a position later modified by constitutional amendment.
Two features of this legal frame matter for reading the evidence. First, the 50 percent cap means reservation was never designed to guarantee proportional representation; it was designed as a bounded corrective. Second, the creamy layer principle was built in from 1992 for OBCs, which tells us that the concern about benefits being captured by the better-off is not a recent objection from opponents. It was a condition the Court itself attached.
The frame has kept moving since. After Indra Sawhney restricted reservation in promotions, Parliament amended the Constitution to restore it for SCs and STs, through the 77th amendment and later amendments that also protected consequential seniority. In 2019, the 103rd amendment opened a new front by adding a 10 percent quota for the Economically Weaker Sections of groups not already covered by caste-based reservation, the first time economic status alone became the basis for a quota. In 2024, a seven-judge bench allowed states to sub-classify Scheduled Castes, recognising that the category is not uniform. Each of these shifts is a subject in its own right, and the later parts of this series return to them. For now, the point is that reservation is not a fixed settlement. It is a policy that courts and legislatures have been adjusting for the whole of its existence.
The scale of the system
The size of the system is often underestimated. Reservation covers three broad categories: Scheduled Castes, at roughly 16 percent of the population; Scheduled Tribes, at roughly 8 percent; and Other Backward Classes, a far larger and more diverse category that the Mandal Commission estimated at 52 percent using colonial-era data, and that Bihar’s 2023 caste survey placed at 63 percent within that state. Together with the newer EWS quota, the majority of Indians now belong to a group eligible for some form of reservation.
That scale has two consequences for reading the evidence. First, reservation is not a narrow carve-out for a small minority; it is a structuring feature of Indian public life, touching admissions, hiring, and elections across almost every institution. Second, the categories are internally diverse to a degree the headline percentages hide. The OBC category alone contains roughly 2,600 communities with very different histories and circumstances. Treating any of these categories as a single block, whether to praise the policy or to attack it, misreads how benefits actually move within them. This internal diversity is the hinge of the reform debate, and Part 2 examines it directly.
What the evidence says worked
The strongest and most consistent finding in the research literature is that reservation increased the presence of SC and ST citizens in places they had been almost entirely absent from. The economist Ashwini Deshpande and others who have studied the policy over decades reach a broadly similar big-picture conclusion, summarised in outlets such as The India Forum: reservation has benefited SCs and STs as a whole, even where implementation has been uneven.
Three specific gains are well documented.
- Representation. Reservation put SC and ST citizens into universities, government offices, and legislatures in numbers that would not have occurred otherwise. This is the outcome the policy was most directly designed to produce, and it is the one it delivered most clearly.
- A new middle class. Access to public education and secure public employment enabled the growth of a Dalit middle class that did not exist at independence. Salaried government work, in particular, offered a route into the formal economy that private hiring often closed.
- Occupational entry. The policy expanded access to occupations that had been effectively inaccessible to entire communities, breaking, at least partially, the historical link between birth and permitted work.
- Political representation. Reserved seats in Parliament and state assemblies, and the reservation of local-government seats for SC, ST, and women through the 73rd and 74th amendments in 1993, placed hundreds of thousands of citizens from these groups into elected office. Research on local-government quotas finds measurable effects on the mix of public goods that reserved councils choose to provide, which suggests representation can change decisions and not only faces.
These are not small achievements. In a society where caste had for centuries assigned occupation and status by birth, moving significant numbers of people into new roles is a real structural change, not a cosmetic one.
The strongest positive case: Tamil Nadu
If the question is whether high reservation is compatible with strong institutions and good social outcomes, Tamil Nadu is the most important case in the country. The state’s total reservation stands at 69 percent, well above the national 50 percent ceiling, protected by a special law placed in the Ninth Schedule. The working split is roughly Backward Classes 30 percent, Most Backward Classes and Denotified Communities 20 percent, Scheduled Castes 18 percent, and Scheduled Tribes 1 percent.
Tamil Nadu also has, by many measures, the best social outcomes among India’s large states: among the lowest maternal mortality, high female literacy, strong nutrition and primary education indicators, and a socially representative healthcare workforce. This record is examined in detail in the companion piece on how Tamil Nadu built India’s best social outcomes.
The Tamil Nadu case carries an important lesson that both sides of the debate tend to miss. Reservation there did not work in isolation. It was paired with sustained public investment in schools, primary health centres, nutrition programmes, and a broad-access administrative culture built over decades. The reservation broadened who entered these institutions; the investment built the institutions worth entering. Neither alone explains the result. This distinction, between reservation as access and investment as capacity, runs through the rest of this series.
It is also honest to record that Tamil Nadu’s success is not uniform. Caste-based exclusion and atrocities have not disappeared in the state, which shows that even high reservation and strong services do not by themselves dissolve deep social hierarchy.
What the evidence says did not work, or remains unresolved
A fair scorecard has to record the failures and the open questions with the same seriousness as the gains.
Implementation gaps, especially at the top. Reservation on paper is not reservation in practice. Research on central universities finds that reserved positions are frequently left unfilled, with the problem most severe at the faculty level, where large backlogs of reserved posts persist year after year. A quota that is not filled delivers representation only on paper. The higher and more powerful the position, the weaker the record of actual filling tends to be.
Concentration of benefits. A recurring finding is that benefits flow disproportionately to households that are already relatively better off within an eligible group, rather than to the poorest members. This is the concern the creamy layer principle was meant to address for OBCs, and the evidence that it persists is the subject of Part 2 of this series, on the creamy layer problem. The point here is only that the concern is empirical, not rhetorical.
Reach to the poorest. Representation and poverty are different targets. A policy can raise the number of SC officers and graduates substantially while still largely missing the poorest households in the same category, who often lack the schooling, information, or family stability needed to reach the point where a quota becomes usable. Whether reservation reaches the bottom of the distribution, and not only the middle, is a live and unsettled question.
Persistence of caste harm. Perhaps the most sobering finding is that quotas have not ended caste discrimination or violence. Studies of the labour market show that caste identity continues to shape outcomes even for the educated. Reservation changed who holds certain positions; it did not, on its own, change the social attitudes that produce discrimination.
Why the evidence is thinner than it should be
Any honest scorecard has to admit a limitation: India measures the effects of reservation with data that is decades out of date. The last full caste count in a national census was in 1931. Since then, the country has run its entire OBC reservation system on population estimates rather than an actual count, and it has assessed the results without a reliable denominator for who belongs to which group and in what numbers.
This gap weakens both sides of the argument. Supporters cannot show precisely how far benefits have reached, because the base population is estimated. Critics cannot show precisely how concentrated benefits have become, for the same reason. The Rohini Commission’s finding on OBC benefit concentration, examined in Part 2, had to work around exactly this absence of current data. The case for finally counting caste, and the objections to it, are set out in the companion piece on India’s caste census. The relevant point here is that much of the reservation debate is conducted with weaker evidence than the stakes deserve, and that better measurement would narrow the range of honest disagreement.
Reading the scorecard
The table below summarises the record against the policy’s stated goals. It is a simplification, and every row carries qualifications discussed above, but it captures the shape of the evidence.
| Stated goal | What the evidence shows |
|---|---|
| Raise representation of SC/ST/OBC in institutions | Achieved, clearly, though unevenly by level |
| Create upward mobility and a new middle class | Achieved for a meaningful share, mainly through public employment |
| Reach the poorest within each group | Weak and contested; benefits skew to the better-off |
| Fill reserved positions in practice | Partial; large backlogs persist, worst at faculty and senior levels |
| End caste discrimination | Not achieved; discrimination and atrocities continue |
| Build strong institutions | Not a goal reservation can meet alone; depends on investment |
The pattern is consistent. Reservation performs well on the goal it was directly designed for, which is access and representation. It performs poorly, or not at all, on goals it was never structurally equipped to meet on its own, which are reaching the very poorest and building institutional quality.
A note on the merit objection
No scorecard is complete without addressing the objection raised most often against reservation: that it lowers merit and therefore weakens the institutions that use it. This deserves a factual answer rather than a reflexive one, in either direction.
The claim rests on an assumption that entry scores predict later performance closely, and that a candidate admitted at a lower entry rank will underperform for good. The evidence on this is mixed and contested. The international version of the argument, known as mismatch theory and associated with the legal scholar Richard Sander, holds that beneficiaries placed in highly selective institutions learn less than they would elsewhere. It has been influential, but a large body of research has been unable to reproduce its central results, and other studies find the opposite, that access to more selective institutions raises graduation and later outcomes for beneficiaries. Part 4 of this series examines this global debate in detail.
Within India, studies that follow reserved-category entrants over time generally find that initial gaps in entry scores narrow as students progress, and that graduates go on to perform their jobs competently. The entry rank is a starting point, not a fixed ceiling. This does not make the merit concern meaningless. It means the concern is an empirical question with a mixed answer, not a settled fact, and that the honest framing is about how to combine access with the support that helps entrants close early gaps, rather than treating a lower entry score as proof of permanent lower capacity. Framing an entire community as less capable is not supported by the evidence and is precisely the generalisation this series avoids.
Why the honest reading matters
Two conclusions follow, and they are uncomfortable for opposite camps.
For those who argue that reservation failed, the data does not support a blanket claim. A policy that moved large numbers of historically excluded citizens into education, employment, and office, and that coexists in Tamil Nadu with the country’s best social outcomes, has not failed on its own terms. Dismissing it wholesale ignores the record.
For those who argue that reservation is sufficient, the data is equally unforgiving. Representation is not the same as reaching the poorest, and it is not the same as institutional quality. A reserved seat in a weak college, or a reserved post that is never filled, delivers far less than the headline percentage suggests. Treating the quota as the end of the work, rather than the beginning, leaves the hardest problems untouched.
The useful debate, therefore, is not whether to keep or scrap reservation. It is how to make it reach the people it was meant for, and how to pair it with the investment that turns access into capacity. Those are reform questions, and they are the subject of the rest of this series: the creamy layer and who actually benefits, caste versus economic criteria after the EWS quota, what other countries learned from their own programmes, and the foundations in schooling, health, and institutional quality that no quota can substitute for. The debate over counting who is disadvantaged in the first place is examined separately in the piece on India’s caste census.
Frequently asked questions
Did reservation achieve its goals? It achieved its central goal of raising the representation of Scheduled Castes, Scheduled Tribes, and Other Backward Classes in education, public employment, and legislatures, and it enabled real upward mobility for a meaningful share of these groups. It did not, on its own, reach the poorest households within these groups or end caste discrimination, and it was never structurally equipped to build institutional quality by itself.
What is the 50 percent cap? In Indra Sawhney (1992), the Supreme Court held that total reservation should not exceed 50 percent except in extraordinary circumstances. This means reservation was designed as a bounded corrective rather than a guarantee of proportional representation. Tamil Nadu exceeds the cap at 69 percent under a special law placed in the Ninth Schedule.
Does Tamil Nadu prove reservation works? Tamil Nadu shows that high reservation is compatible with strong social outcomes, but it does not show that reservation alone produced those outcomes. The state paired reservation with decades of sustained public investment in schools, primary health, and nutrition. The access and the investment together explain the result.
Why are reserved posts left unfilled? Research on central universities finds persistent backlogs of unfilled reserved positions, worst at the faculty and senior levels. Causes include weak enforcement, narrow eligibility pools at the highest qualification levels, and institutional resistance. An unfilled quota delivers representation only on paper.
Is the concern about benefits reaching the better-off a recent political attack? No. The Supreme Court itself built the creamy layer principle into OBC reservation in 1992, precisely to prevent benefits from being captured by the already-advanced within a group. The concern is part of the policy’s own design, and the evidence on how well it works is examined in Part 2 of this series.
Does reservation end caste discrimination? The evidence says it does not, by itself. Quotas changed who occupies certain positions, but studies show caste identity still shapes labour-market and social outcomes, and caste atrocities continue even in states with high reservation. Reservation addresses access to institutions, not the attitudes that produce discrimination.
Does reservation lower merit and weaken institutions? The claim is contested rather than settled. Indian studies that track reserved-category entrants over time generally find that early gaps in entry scores narrow as students progress and that graduates perform their roles competently, while the international mismatch argument that beneficiaries underperform has largely failed to reproduce. The evidence supports pairing access with academic support to close early gaps, and does not support treating a lower entry rank as proof of permanent lower ability, still less generalising that about a whole community.
Should reservation be based on income instead of caste? This is the central reform argument of the past decade, and it became concrete with the 2019 EWS quota, the first quota based on economic status alone. Whether income should replace or supplement caste as the basis for reservation depends on what the policy is meant to correct: caste-based exclusion, economic poverty, or both. Part 3 of this series examines the caste-versus-economic question directly, using the EWS experience as evidence.