Affirmative Action in India with Hierarchical Reservations

arXiv:2602.09189 · econ.TH · Submitted 2026-02-09 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Economics. The day in economics: theory, mechanisms and the evidence beneath them.

Ruth: I'm Ruth.

Warren: And I'm Warren, here with Iris.

Ruth: We're looking at the paper: "Affirmative Action in India with Hierarchical Reservations".

Warren: Here's the gist of the paper. India implements the world’s most complex affirmative action program through vertical and horizontal reservations Although applicants can belong to at most one vertical category, they can qualify for multiple horizontal reservation categories simultaneously.

Ruth: We examine resource allocation problems in India, where horizontal reservations follow a hierarchical structure within a one-to-all horizontal matching framework.

Warren: The Nature of Indian Affirmative Action The Indian affirmative action program has been implemented through a highly regulated reserve system that consists of vertical and horizontal reservations.

Ruth: Reservations for Scheduled Castes (SC), Scheduled Tribes (ST), Other Backward Classes (OBC) and Economically Weaker Sections (EWS) are called vertical reservations.

Title and authors: Ruth: The paper sets up the main tension right away by focusing on how India manages its most complex affirmative action program through vertical and horizontal reservations. It establishes that applicants can fit into multiple vertical categories, but only one vertical category, while they can qualify for several horizontal ones at the same time.

Warren: That complexity is what makes the resource allocation tricky. The paper then zeroes in on how those horizontal reservations follow a specific hierarchical structure within a matching framework that goes from one candidate to all of them.

Iris: It’s important because it moves beyond just listing the rules and starts modeling the actual choices applicants make under these constraints. They introduce this concept called the hierarchical choice rule, which they claim selects the most meritorious set of applicants among those who qualify for horizontal reservations.

The paper's summary: Ruth: The summary explains that by using this hierarchical choice rule, the authors can show that it actually picks the most meritorious set of applicants when you consider all the vertical categories together. They then build an aggregate choice rule out of those individual rules across all vertical categories.

Warren: That aggregate choice rule is what they analyze next because they prove it satisfies some key conditions—the substitute condition and size monotonicity—and even that aggregate rule satisfies a fairness condition, which is pretty significant for their model.

Iris: They connect this to mechanism design by showing that when you pair this aggregate choice rule with the generalized deferred acceptance mechanism, you get a unique result: it’s stable and strategy-proof while eliminating justified envy.

The paper's improvements: Ruth: The authors suggest some ways to strengthen their initial model. One key improvement is extending the model to handle OBC de-reservations, which means they allow unfilled positions in public universities to be filled later by looking at all the remaining applicants.

Warren: That extension requires them to use an aggregate choice rule with transfers that allocates those unfilled spots as open-category positions at the end, and they show that this modified choice rule is fair for those specific cases.

Iris: What's interesting here is how they broaden the scope by incorporating both horizontal and vertical reservations along with these de-reservations, which means the domain of what individuals actually prefer gets much richer.

Conclusion: Ruth: So, to wrap up, the paper on "Affirmative Action in India with Hierarchical Reservations" shows how a specific hierarchical choice rule and its aggregate version can select the most meritorious applicants in this complex system.

Warren: The mechanism design part is also solid because they show that coupling it with that aggregate rule results in a stable and strategy-proof mechanism that avoids justified envy.

Iris: It really expands the practical application by including those de-reservations, which shows how the framework holds up when you factor in real-world adjustments to the rules.

Ruth: It gives us a robust way to model these allocation problems where vertical and horizontal constraints overlap. We'll be keeping an eye on this paper as we explore other ways complex preference domains are modeled.

BOGAZIıCI ˘ UNIVERSITY · Iowa State University

econ.TH

Submitted: 2026-02-09

Updated: 2026-11-14

Comments: arXiv admin note: text overlap with arXiv:2310.02660

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 92/100

The gist: India implements the world’s most complex affirmative action program through vertical and horizontal reservations Although applicants can belong to at most one vertical category…

Key concepts

Vertical Reservations
These are mandatory quotas set aside for specific marginalized groups such as Scheduled Castes (SC), Scheduled Tribes (ST), Other Backward Classes (OBC), and Economically Weaker Sections (EWS). They operate on an 'over-and-above' principle, meaning positions filled by these groups don't reduce their reserved quota.
Horizontal Reservations
These are reservations designated for other marginalized categories, like individuals with disabilities. Unlike vertical reservations, horizontal reservations are not mandatory quotas but rather mechanisms used to address specific needs within the system.
Hierarchical Choice Rule
This rule ranks applicants by prioritizing choices that satisfy horizontal reservation requirements first. It is designed to be merit-undominated among all choice rules that meet these constraints, ensuring the selection process favors the most meritorious candidates while respecting legal mandates.
Aggregate Choice Rule
This rule combines hierarchical choice rules across all vertical categories. It is crucial because it satisfies fairness conditions and ensures that when an applicant is rejected, those selected must have higher merit or specific reservation benefits that the rejected applicant lacks.

Terminology

Summary

India implements the world’s most complex affirmative action program through vertical and horizontal reservations Although applicants can belong to at most one vertical category, they can qualify for multiple horizontal reservation categories simultaneously. p. 0

Where the paper says it

AFFIRMATIVE ACTION IN INDIA WITH HIERARCHICAL RESERVATIONS ORHAN AYGÜN1 AND BERTAN TURHAN2 ABSTRACT: India implements the world’s most complex affirmative action program through vertical and horizontal reservations. Although applicants can belong to at most one vertical category, they can qualify for multiple horizontal reservation categories simultaneously.

We examine resource allocation problems in India, where horizontal reservations follow a hierarchical structure within a one-to-all horizontal matching framework. p. 0
Where the paper says it

Although applicants can belong to at most one vertical category, they can qualify for multiple horizontal reservation categories simultaneously. We examine resource allocation problems in India, where horizontal reservations follow a hierarchical structure within a one-to-all horizontal matching framework.

The Nature of Indian Affirmative Action

The Indian affirmative action program has been implemented through a highly regulated reserve system that consists of vertical and horizontal reservations. Reservations for Scheduled Castes (SC), Scheduled Tribes (ST), Other Backward Classes (OBC) and Economically Weaker Sections (EWS) are called vertical reservations. p. 1

Where the paper says it

1 The Supreme Court of India (SCI) has been regulating the implementation of vertical and horizontal reservations through successive verdicts since 1992. Reservations for Scheduled Castes (SC), Scheduled Tribes (ST), Other Backward Classes (OBC) and Economically Weaker Sections (EWS) are called vertical reservations.

Mandated by the SCI in the Indra Sawhney (1992) verdict, vertical reservations have operated as over-and-above, meaning any open category positions secured by individuals belonging to SC, ST, OBC, or EWS are not to be deducted from their respective vertical reservations. p. 1
Where the paper says it

Mandated by the SCI in the Indra Sawhney (1992) verdict, vertical reservations have operated as over-and-above. That is, any open category positions secured by individuals belonging to SC, ST, OBC, or EWS are not to be deducted from their respective vertical reservations.

Horizontal reservations are designated for other marginalized groups such as individuals with disabilities. p. 1
Where the paper says it

To implement vertical reservations as over-and-above, institutions fill open category positions, considering all applicants, including members of reserved categories, before filling reserved positions. Horizontal reservations are designated for other marginalized groups, such as individuals with disabilities.

The Hierarchical Choice Rule

The paper introduces the hierarchical choice rule, which is defined in Step 1 through Step n of the process. This rule is merit-undominated among the choice rules that satisfy the horizontal reservations. p. 3

Where the paper says it

We formulate the hierarchical choice rule that minimally deviates from fully merit-based outcomes. 11 Theorem 1 shows that the hierarchical choice rule is merit-undominated among the choice rules that satisfy the horizontal reservations.

The aggregate choice rule, which comprises hierarchical choice rules across all vertical categories, satisfies the substitute condition and size monotonicity. p. 3
Where the paper says it

We propose an aggregate choice rule for institutions in which applicants are selected within a vertical category via a hierarchical choice rule. In Proposition 2, we show that this aggregate choice rule satisfies the substitute condition and size monotonicity, two desirable properties in matching theory.

Furthermore, in Theorem 2, this aggregate choice rule satisfies a fairness condition. p. 3
Where the paper says it

In Proposition 2, we show that this aggregate choice rule satisfies the substitute condition and size monotonicity, two desirable properties in matching theory. More importantly, in Theorem 2, we show that the aggregate choice rule satisfies a fairness condition, meaning that when an applicant is not selected, all those who are selected must either have higher merit scores or benefit from vertical or horizontal reservations that the rejected applicant lacks.

Mechanism Design Results

The generalized deferred acceptance mechanism is shown to be the unique stable and strategy-proof mechanism that eliminates justified envy when coupled with the aggregate choice rule. The paper establishes that this mechanism is stable with respect to the aggregate choice rule profile (C h s)s∈S, strategy-proof for individuals, and eliminates justified envy. p. 4

Where the paper says it

AFFIRMATIVE ACTION IN INDIA WITH HIERARCHICAL RESERVATIONS We show in our Theorem 3 that the generalized deferred acceptance mechanism is the unique stable mechanism (with respect to the aggregate choice rules comprising the hierarchical choice rule in each vertical category) that is strategy-proof for individuals and eliminates justified envy. 12 The prevalence of the hierarchical structure of reservations makes our contributions very useful for other applications beyond India.

Extensions and Contributions

The model extends to OBC de-reservations implemented in admissions to public universities using an aggregate choice rule that allows unfilled OBC positions to be filled at the end by considering all remaining applicants. The aggregate choice rule with transfers allocates unfilled OBC positions as open-category positions at the end, and Theorem 4 shows this choice rule is fair. p. 4

Where the paper says it

We introduce an aggregate choice rule for institutions that uses the hierarchical choice rule in each vertical category and that allows unfilled OBC positions to be filled at the end by considering all remaining applicants. We show in Theorem 4 that this choice rule is fair.

This work contributes by incorporating both horizontal and vertical reservations alongside de-reservations, broadening the scope of individuals’ preference domain. p. 4
Where the paper says it

13 This formulation captures real-world phenomena, such as the stigma associated with reserved positions, substantially enhancing the model’s behavioral realism. The preference domain provided to applicants has crucial consequences, and recently there has been interesting work in market design focusing on the impact of the preference domain to which individuals have access, such as Hakimov and Khanna (2025).

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Hierarchical Choice Rule Implementation: The AI system can implement a hierarchical choice rule that is merit-undominated among the choice rules that satisfy horizontal reservations, ensuring it selects the most meritorious set of applicants.

  2. Aggregate Choice Rule Fairness: The AI can utilize an aggregate choice rule, which includes the hierarchical choice rule across all vertical categories, to ensure it satisfies the fairness condition, meaning when an applicant is not selected, all those who are selected must either have higher merit scores or benefit from vertical or horizontal reservations that the rejected applicant lacks.

  3. Mechanism Design for Stability and Strategy-Proofness: The AI can employ the generalized deferred acceptance mechanism coupled with this aggregate choice rule to ensure the resulting mechanism is stable, strategy-proof, and eliminates justified envy.

  4. Preference Modeling: The system can model preferences over a richer domain where individuals express preferences over institution-vertical category pairs, capturing real-world phenomena, such as the stigma associated with reserved positions.

  5. Handling Complex Constraints: The AI can handle complex constraint structures involving vertical and horizontal reservations, including OBC de-reservations, by extending the model to formulate an aggregate choice rule that allows unfilled OBC positions to be filled at the end by considering all remaining applicants.

  6. Soft Reservation Policies: The system can operate effectively under soft reservation policies, as demonstrated by Theorem 5 in the context of aggregate choice rules with transfers, which maximizes merit among all feasible soft reservation schemes.

Abstract

We study the design of affirmative action mechanisms, focusing on India's system of horizontal reservations within vertical reservations. We consider a one-to-all policy, under which an applicant eligible for multiple horizontal categories simultaneously contributes toward each category's minimum guarantee. When horizontal reservations have a hierarchical structure, we introduce a choice rule that selects the unique merit-optimal set of applicants subject to these guaranties, satisfies the substitutes condition and size monotonicity. In a matching-with-contracts framework, we show that the generalized deferred acceptance mechanism is stable, strategy-proof, and eliminates justified envy under aggregate choice rules using the hierarchical choice rule in each vertical category. We establish a fundamental trade-off between merit and affirmative action between the one-to-all and one-to-one counting policies.

Sources

Related papers