India is often discussed as if its reservation system were unique. It is not. Across the second half of the twentieth century, many countries built group-based preferences to correct historical disadvantage, whether that disadvantage came from slavery, colonial rule, apartheid, sectarian discrimination, or caste. Each country designed its policy differently, and each produced a different mix of success and failure. Studying those experiments is useful precisely because they were run in different legal systems, different economies, and different social structures, yet several of them ran into the same problems India now debates.
This article surveys six cases: the United States, Malaysia, Brazil, South Africa, and Northern Ireland, with a closer look at how the evidence has been evaluated in each. The purpose is not to argue that affirmative action always works, nor that it always fails. The honest record is mixed. The purpose is to identify which design choices tended to help the intended beneficiaries and which choices tended to send benefits somewhere else. Those patterns carry direct lessons for India’s next round of reform.
Why comparison is worth doing
A single country cannot easily separate the effect of a policy from everything else happening around it. When many countries try variations of the same idea, the common threads become visible. If elite capture appears in Malaysia, in South Africa, and in India, then elite capture is probably a feature of how such policies are structured rather than an accident of any one country. If a class-and-race hybrid evaluates well in Brazil, that is worth knowing before India settles its own debate over economic versus caste criteria.
Comparison also guards against two opposite errors. The first is the belief that preferences are a magic solution that only India has failed to implement well. The second is the belief that preferences never help anyone. Both are wrong, and the international record shows why.
The United States: from Bakke to the end of race-conscious admissions
The United States pursued affirmative action mainly through university admissions and employment, and its story is now one of a long rise followed by a legal reversal. The framework that governed college admissions for decades began with Regents of the University of California v. Bakke in 1978, which barred fixed racial quotas but permitted race to be used as one factor among many. That approach was reaffirmed in Grutter v. Bollinger in 2003 and again, narrowly, in Fisher v. University of Texas.
In June 2023, the Supreme Court ended it. In Students for Fair Admissions v. Harvard and the companion University of North Carolina case, decided on 29 June 2023 by a 6 to 3 majority, the Court held that race-conscious admissions violate the Equal Protection Clause of the Fourteenth Amendment, overturning roughly forty-five years of precedent. The Court did leave one door open: an applicant may still discuss how race affected his or her individual life, and a university may consider that account, but the group-based use of race in admissions is no longer permitted. The details and limits of the ruling are summarized by the NAACP Legal Defense Fund.
The American debate also produced a much-cited academic controversy worth understanding, because opponents of reservation in India sometimes borrow it. The “mismatch” hypothesis, associated with Richard Sander and Stuart Taylor, argues that affirmative action can harm the students it is meant to help by placing them in institutions where they cannot keep up, so they learn less and graduate at lower rates than they would have at a less selective school. The claim is prominent, but it is heavily contested. Other scholars have been unable to reliably reproduce Sander’s results, and several studies find the opposite pattern, with higher graduation rates for minority students at more selective institutions, which is the reverse of what mismatch predicts. A readable overview of the dispute appears at SCOTUSblog. The responsible summary is that mismatch is a live academic argument, not a settled finding, and it should not be cited as proven.
The American lesson for India is less about the courtroom outcome, which turns on a specific constitution, and more about durability. A policy that rests on contested legal ground and is never redesigned can be removed all at once by a change in the court. Policies survive when they are grounded in clear public purpose and adapted over time.
There is a second American thread relevant to India. As race-conscious admissions came under pressure, a large body of work examined whether a class-based system, using family income, wealth, and neighbourhood disadvantage, could achieve much of the same diversity. The findings are mixed: class-based systems capture real economic hardship and are legally sturdier, but on their own they do not always reproduce the same group representation, because disadvantage and identity overlap without being identical. That result maps onto India’s own argument over whether economic criteria can substitute for caste, and it counsels against assuming that one axis fully captures the other. The honest reading is that class and group measure different things, and a system that cares about both usually needs to look at both.
Malaysia: large gains, and a warning about permanence
Malaysia offers the clearest cautionary tale. Its New Economic Policy, launched in 1971 after the racial violence of 1969, created wide preferences for the Bumiputera majority, who were the political majority but the economic minority. The preferences covered higher education, public employment, business ownership, and corporate equity, and they were unusually comprehensive.
The gains were real and large. According to research summarized in the Springer and Wiley literature on the NEP, Bumiputera poverty fell from about 65 percent in 1970 to 3.6 percent in 2007, Bumiputera corporate equity ownership rose from 2.4 percent in 1969 to 19.4 percent in 2006, and a sizable Bumiputera middle class came into being where almost none had existed. On its original poverty-reduction goal, the policy delivered.
The problems are equally documented. Over five decades, the benefits increasingly flowed to a politically connected elite rather than to the poorest Bumiputera. Intra-ethnic inequality rose, meaning the gap within the favoured group widened even as the gap between groups narrowed. The country experienced a persistent brain drain as skilled members of non-preferred groups emigrated. Perhaps most important for policy design, the preferences were never sunset and never seriously restructured, which entrenched a sense of permanent entitlement at the top and permanent grievance elsewhere. Malaysia shows that a preference program can succeed at reducing poverty and still fail if it never ends and never checks who is actually collecting the benefit.
Brazil: a hybrid that has evaluated well
Brazil designed its main program differently, and the difference matters. The Quota Law, Law 12.711 of 2012, reserves 50 percent of seats in federal universities and federal institutes for a combined pool: students from public schools, students from low-income families, and Black, mixed-race, and Indigenous students, along with students with disabilities. This is a hybrid of class and group criteria rather than a pure race quota, and it is anchored to the public-school system, which is itself a strong proxy for disadvantage.
The reach has been broad. More than one million students were covered by the quota system between 2012 and 2021, and since 2014 a large share of undergraduates in federal institutions have come from public schools and low-income families. The evaluation record is encouraging. Research by Andrew Francis and Maria Tannuri-Pianto, discussed in outlets such as VoxDev, found that quotas raised the enrolment of low-income Black students without reducing their academic effort, and in some cases prompted applicants to aim for more selective courses and to prepare harder. Later cohort studies extended the analysis into professional outcomes. Brazil’s model is not free of dispute, and the law was reviewed and adjusted in 2023, but it stands as the best-evaluated example of a preference system that pairs group criteria with a clear class anchor.
The lesson for India is direct. A hybrid that combines social group with an economic or schooling test can widen access while keeping the benefit pointed at genuine disadvantage, and it can be studied honestly because the eligibility criteria are measurable.
Brazil also allows a look further down the pipeline than admissions. Because the quota cohorts entering federal universities from 2012 can now be tracked into professional life, researchers have begun to study outcomes such as medical-residency placement for quota entrants, linking large national datasets across several years. Early work of this kind matters because the sharpest criticism of any preference policy is that access at entry does not translate into competence or careers at exit. Where the data can follow students that far, the Brazilian evidence so far does not show the collapse in outcomes that critics predict, though the cohorts are young and the research is still maturing. The wider point for India is procedural: a policy that is built to be measured end to end can answer its critics with evidence, while a policy that is never tracked can only answer them with assertion.
South Africa: representation up, reach still uneven
South Africa built Black Economic Empowerment and employment equity rules after the end of apartheid in 1994, addressing one of the most extreme systems of legal discrimination in modern history. The policies expanded Black representation in management, professions, and business ownership, and helped create a Black middle class that apartheid had deliberately prevented.
South Africa combined two instruments. Employment equity law required larger employers to report on and improve the representation of previously excluded groups across job levels, an approach closer to the monitoring model than to fixed quotas. Black Economic Empowerment went further, scoring firms on ownership, management, procurement, and skills development, and tying those scores to eligibility for state business. The design intent was to move beyond token hiring toward ownership and control of the economy.
The criticism echoes Malaysia. A significant share of the most visible gains accrued to a politically connected minority, a pattern captured in the local term for those who prospered through state contracts. Scorecards proved easier to satisfy through a handful of high-profile ownership deals than through broad-based skills and employment change, so measured compliance sometimes ran ahead of real economic inclusion. Reaching the poorest majority, especially in employment and basic services, has proved far slower than reaching the emerging elite. South Africa demonstrates that formal representation at the top can improve quickly while the conditions of the poor change little, which is exactly the gap a well-targeted policy is supposed to close, and it shows that what a policy chooses to measure is what a policy will get.
Northern Ireland: goals and monitoring rather than rigid quotas
Northern Ireland took a quieter and, by many accounts, more successful path. The Fair Employment Act of 1989 tackled the long-standing employment imbalance between the Catholic and Protestant communities. Rather than imposing rigid numerical quotas, it required employers to monitor the composition of their workforces, to register, and to work toward fair-participation goals, with an enforcement body overseeing progress.
Over time the employment gap between the two communities narrowed, and the approach is frequently cited as effective and comparatively low in polarisation. The reason is instructive. Monitoring and goals change behaviour and create accountability without the zero-sum framing that fixed quotas often produce, and they can be relaxed as the underlying imbalance shrinks. For India, the Northern Ireland case is a reminder that transparency and measured goals are policy tools in their own right, not merely softer versions of quotas.
The cases side by side
| Country | Policy | Main gain | Main problem |
|---|---|---|---|
| United States | Race-conscious admissions (1978 to 2023) | Diversified selective universities for decades | Rested on contested legal ground, ended by the courts in 2023 |
| Malaysia | New Economic Policy preferences (1971) | Bumiputera poverty fell from 65% to 3.6% | Elite capture, rising intra-group inequality, no sunset |
| Brazil | Quota Law 2012 (class and race hybrid) | Over 1 million students, well-evaluated gains | Ongoing debate, needed 2023 review and adjustment |
| South Africa | Black Economic Empowerment (post-1994) | Larger Black middle class and management share | Gains concentrated in a connected elite |
| Northern Ireland | Fair Employment Act (1989) | Narrowed Catholic-Protestant job gap | Slower where segregation was deepest |
What the pattern says
Read together, the six cases point to a small number of design lessons that recur regardless of country.
- Elite capture is the common failure mode. In Malaysia, in South Africa, and in India’s own creamy-layer debate, the benefit drifts toward those within the eligible group who are already better off. A policy that does not actively guard against this will tend to reward the top of the disadvantaged group.
- Pairing preferences with investment matters. Preferences open a door, but whether people can walk through it depends on the schooling, health, and family resources they bring. The countries that treated preferences as one part of a wider development effort saw broader results.
- Time-bound review and good data matter. The policies that aged worst were the ones that were never sunset and never measured. The policies that could be defended were the ones that were evaluated openly.
- Hybrids evaluate well. Brazil’s combination of class, schooling, and group criteria produced measurable gains and could be studied honestly, which strengthened rather than weakened public trust.
- Rigid permanence invites backlash. Fixed, unreviewed quotas tend to harden into grievance on all sides. Goals, monitoring, and periodic review lower the temperature.
Two things India is already testing
India is not standing still, and two recent developments line up closely with the international lessons above. In 2019 the country added a ten percent quota for economically weaker sections outside the existing categories, a purely economic criterion whose constitutional validity the Supreme Court upheld in 2022. Whatever one thinks of its design, it moved the system toward the class axis that Brazil and the American class-based debate both point to, and it reopened the question of how economic and caste disadvantage should be weighed together.
In August 2024 a seven-judge bench allowed states to sub-classify Scheduled Castes, recognising that the category is not uniform and that some communities within it remain far more excluded than others. Several judges also urged states to identify a creamy layer even within the Scheduled Castes. Both moves are attempts to fix the elite-capture problem that recurs across Malaysia, South Africa, and India, by pointing the benefit at the least advantaged within the eligible group rather than the most advanced. These are exactly the targeting questions the comparative record says determine whether a preference policy reaches the people it names.
None of this settles India’s debate, which has its own constitutional history and its own data. That history and evidence are examined in the first article in this series on what the Indian data actually shows. What the international record adds is perspective: India’s difficulties are not signs of a uniquely broken system, they are the predictable stresses of a policy family that every country has found hard to target well.
Where this leaves the reform question
The comparative evidence does not support the claim that group-based preferences never help. Malaysia cut poverty sharply, Brazil widened access without lowering standards, South Africa built a middle class that apartheid had blocked, and Northern Ireland narrowed a sectarian divide. The evidence also does not support the claim that preferences automatically reach the people who need them most. In several countries they did not, and the reasons were structural.
The useful conclusion is about design, not about whether any group deserves support. Preferences work best when they are targeted at real disadvantage, paired with investment in the foundations that let people succeed, measured with honest data, and reviewed on a schedule rather than treated as permanent. That is also the argument for shifting attention upstream, to the schools, clinics, and institutions that decide who ever reaches the starting line, which is the subject of the final article in this series on fixing the foundations.
Frequently asked questions
Did the United States ban all consideration of race in admissions? Not entirely. The 2023 ruling ended the group-based use of race in admissions, but the Court said an applicant may still describe how race affected his or her life, and a university may weigh that individual account. Broad racial preferences as a system are what the decision prohibited.
Is the mismatch theory proof that affirmative action hurts its beneficiaries? No. Mismatch is a contested academic hypothesis. Independent scholars have struggled to reproduce its results, and several studies find the opposite, with better graduation outcomes at more selective institutions. It should be described as disputed, not proven.
Which country’s model is closest to a needs-based approach? Brazil’s, because its quota is anchored to public schooling and family income alongside group identity. That hybrid targets disadvantage directly and has been evaluated more favourably than pure quota systems.
What is the single most common problem across these countries? Elite capture, meaning the tendency for benefits to concentrate among the better-off members of the eligible group rather than reaching the poorest. It appears in Malaysia, South Africa, and India alike.
Does international experience say India should end reservation? No. It suggests India should redesign for better targeting, pair preferences with investment in foundations, and review the system on evidence, rather than either abolishing it or freezing it in place.
Why include Northern Ireland, which did not use quotas? Because its monitoring-and-goals model is widely regarded as effective and less polarising than rigid quotas, which makes it a useful alternative design for India to study rather than a straight quota comparison.
Do these countries suggest a time limit on preferences? The record favours periodic review over a fixed expiry date. The policies that aged worst, such as Malaysia’s, were never reassessed, while the ones that adjusted on evidence kept public trust. The practical lesson is to build in scheduled review and honest measurement rather than to promise an arbitrary end date that politics will not honour.
What should India copy, and what should it avoid? Worth copying: Brazil’s class-and-group anchor and its commitment to measurement, and Northern Ireland’s transparent monitoring. Worth avoiding: Malaysia’s permanence without review and South Africa’s reliance on scorecards that a small elite can satisfy without broad change.