The Fractures Within: The Subclassification, the Yard Stick Problem and the Constitutional Ambiguities of State of Punjab v Davinder Singh

Introduction

The Madiga community of Andhra Pradesh has been sitting at the lowest rung of a society already at the bottom of the caste hierarchy for decades. Denied not just the dignity promised by the Constitution of India to all of its people but even the reservation meant specifically for them as a partial remedy. The underlying problem in their case was never that provisions of reservations did not exist, instead the issue was that a relatively more advanced Scheduled Caste (‘SC’) Community, the Malas, were absorbing a disproportionate share of the SC quota seats, leaving the Madigas with close to nothing. This pattern repeats across India, for instance the Jatavas over Valmikis in Uttar Pradesh, the Paswans over the Musahars in Bihar, where one SC community secures the overwhelming majority in public employment, while the other, even after being larger in number and more acutely disadvantaged, receives a share proportionate to neither their population nor their needs.

The decision in State of Punjab v Davinder Singh works as the Supreme Court’s answer to this issue, by opining that sub-classification of scheduled castes by state governments may be valid for the purpose of directing larger share of benefits toward the more disadvantaged sub-groups and in doing so it overruled the decision in EV Chinnaiah v State of Andhra Pradesh, that had declared SCs a homogenous class and barred any internal differentiation.

Yet while Davinder Singh was constitutionally necessary, it remains doctrinally incomplete. The court was correct in overruling Chinnaiah and reconstructing the relationship between Article  341 and Article 15 and 16, but it failed in being  practically and doctrinally  applicable  by leaving the operative yardstick for identifying inter-se backwardness within the SCs undefined. This eventually created a standard that is constitutionally mandated yet operationally hollow, thereby risking a landmark holding to turn into a mere permission slip, whose conditions are designed to be impossible to fulfill.      

This difficulty is rooted in the nature of caste itself. In the Annihilation of Caste, Ambedkar defined the caste system as one of ‘graded inequality’ marked by an ‘ascending scale of reverence and descending scale of contempt’. Caste hierarchy functions not with a single line separating the oppressor from the oppressed but in a layered structure in which even the oppressed rank themselves against each other. The Balmiki, the Musahars, the Madiga, never got to share the same social positions as the other communities on the SC list, instead they sat at the lowest point of the lowest category. A reservation policy made without acknowledging this gradation isn't Ambedkarite at all. As the court put it in  NM Thomas, it is ‘to treat unequals equally is to deny equality’.

A Door Left Open, Then Shut

The constitutional framework for reservation was built case by case as and when inadequacies were identified. One case framework was Justice Mathew’s holding in State of Kerala v NM Thomas, that further sub-classification within a backward class was not constitutionally barred, provided it was done based on a reasonable basis. This was a door left open that was misread, closed and then forced back open over a course of five decades.

The framework built by the EV Chinnaiah  judgment, held that the Presidential List under Article 341 created a homogenous class and that no state legislature could further differentiate within it. For this it used the Ambedkarite reasoning, that the list was designed to ‘eliminate any kind of political interference’ as grounds for restricting state power and established that a ‘uniform yardstick’ had to be applied in case of SC reservation. However, a critical limitation of this framework was its assumption of a uniformity that was simply unsupported by empirical reality.

The SC population exhibits significant disparities in labour force participation and work distribution. The 2011 Census records substantial inter-community variation within the Scheduled Castes. The 2011 Census highlights substantial inter-community variation within SCs. An instance being findings from the Justice P Ramachandra Raju Commission report that reveal severe employment imbalances within the SC umbrella, where the Malas secured 69% of government positions, compared to Madigas at 22.7% and Rellis at 1.5%. It highlights the underlying socio-economic disparities that Chinnaiah’s one-size-fits-all approach overlooked.  This case was ultimately overruled by the Davinder Singh judgment.

Deconstructing Davinder Singh: Identification v Allocation 

In Davinder Singh, Chief Justice Chandrachud opined  that the preceding Chinnaiah judgment had made a category error by conflating two constitutionally distinct operations- the identification of Scheduled Castes under Article 341, which is an exclusively parliamentary function and the distribution of reservation benefits among those already identified which falls under Article 15(4) and 16(4). He further asserted that sub-classification ‘does not tinker with or modify the presidential list. The State law of preferential treatment to a limited extent does not amend the list. It adopts the list as it is’. Suggesting the standard, he said that adequacy of representation has to be assessed qualitatively, whereas a community concentrated in lower-grade positions and senior administrative roles lack ‘effective representation’, even when the aggregate employment numbers seem sufficient.

A state that sub-classifies SC communities for differential treatment does not vary the list, does not remove any caste, does not add one or alter who is legally recognised as Scheduled Caste. It simply allows certain communities already on the list to be given specific benefits, different from the others, which is precisely the kind of distributive justice that Articles 15(4) and 16(4) authorise.      

The Yard Stick Problem

The fundamental failure of the judgment in Davinder Singh lies in the fact that beyond the ‘effective representation’ language the judgment doesn't go any further to specify what exactly is effective representation and what quantum of disparity amounts to inadequacy. It doesn't categorise grades for constitutional relevance and representation. It doesn't indicate the base line for measuring effective representation, and further even fails to set a temporal frame for assessment or review.

This isn't a minor gap to be overlooked; it weakens the entire operation of the doctrine. Article 14 necessitates that any classification must rest on an ascertainable principle, leaving this principle undefined risks executive discretion that is all the more, unreviewable for all practical purposes.

There are risks of state governments commissioning surveys with a friendly methodology, restricting the pool of comparator groups, or selectively drawing from historical data just in order to favour a particular community. There is nothing in the judgment’s opinion in Davinder Singh that stops this. The case of Haryana illustrates this danger. Where the Backward Classes Commission selected villages for survey without disclosing the basis of selection. It was revealed by the National Commission for Backward Classes that the survey considered Jat representation in government services but omitted substantial representation in the armed forces. Further, in the case of Murari Lal v State of Haryana, this selection was alleged to have been influenced by extraneous conditions. This highlights the need for a uniform and objectively defined yardstick.

A Data Problem Being Misquandered as a Legal One

The standards and framework established by the judgment are effective if applied, however for any practical application, the judgment fails to acknowledge the fact that the infrastructure needed to apply this standard does not exist yet.      

India currently lacks granular up-to-date data for tracking caste-wise data on public employment. This gap is compounded by the difficulty of even producing reliable sub-caste data. The 2011 Socio-Economic Caste Census recorded nearly 46 lakhs raw, caste, sub-caste, synonym, surname, clan and gotra entries that required further extensive reclassification. Due to the survey being open-column and largely relying upon self-reporting, the data was riddled with spelling variations and duplicate entries, rendering it unusable. Hence, it was never officially published. While the upcoming census is expected to include comprehensive caste enumeration for all communities, the methodology is still unsettled. Ultimately the core issue is not merely the lack of statistics but the absence of a stable data architecture capable of producing reliable sub-caste metrics that the court’s standard presumes.      

Further this creates a paradox, the more marginalised a community is, the lesser institutional capacity it holds to generate the statistical proof the court demands. Eventually the communities that most need the benefits of this sub-classification are the least likely to be able to fulfill the evidentiary threshold attached to avail it.

Constructing a Workable Framework

Whatever be the current state of infrastructure, none is a reason to walk back the constitutional power the majority has correctly recognised. It is a pertinent reason to build a framework to materialise the suggested framework. An independent commission inquiry, participatory evidence gathering, and published findings open to public challenge before any notification takes place should be at place. A workable framework would work in four parts.

First, a measurable constitutional metric for ‘inadequacy of effective representation’ should be given a standardised quantitative definition, specifying what proportion of a SC community’s   population share in Group A and Group B services is considered sufficient, thereby moving beyond the vague judicial language currently at place. This threshold is backed by Thorat and Attewell’s empirical work on labour market discrimination that supports that stance that exclusion from senior roles perpetuates caste subordination in ways that lower grade employment simply does not work. Their findings support assessing caste representation not merely by the number of persons employed but also their representation across occupational grades.    

Secondly, states should be given the burden of proof for establishing this disparity using public data. This must necessitate the integration of disaggregated, granular data, since all the currently available data fail to capture the fluctuations in the demographic realities essential for policy calibration.

Thirdly, any sub-classification in place must undergo mandatory periodic review. To ensure that once parity is reached for a community, the sub-classification lapses unless the state can show continuing inadequacy. Ensuring that preferential allocation responds to continuing disparities without treating backwardness as a permanent condition.

Fourth, on judicial review, courts should look into the reasoning relied upon by the commission, and solid real-world evidence to back up its facts. Crucially, this review must extend to assessing whether the methodology employed for sub-classification adheres to established quantitative thresholds, ensuring that the state's reliance on empirical data is not merely a pretext for arbitrary exclusion or political preference.

Essentially, Article 14 of the Constitution requires intelligibility and rationality not uniformity. As long as States adopt a workable framework that is publicly stated, consistently applied, and rationally connected to the goals of effective representation, it is very well acceptable within the threshold of Article 14. What goes against the constitutional norms is a threshold that is not stated at all, because without a declared standard there is nothing to test neither intelligibility nor rational nexus applied.   

In order to institutionalise this framework, the existing National Commission for Scheduled Caste should be appropriately empowered to conduct and oversee independent surveys, assess inter-se backwardness, and issue recommendations on the basis of a common methodology. To further bridge the gap, the Parliament should mandate standardised, disaggregated, SC employment reporting across state services, preserving each state’s role in collecting its own administrative data. This mechanism would operate within the existing Article 341 list and would not confer any power on the state to alter or add to the list. Instead, its function would be to provide an independent evidentiary basis for determining whether differential allocation among listed SC communities remains justified.  Without this the powers recognised by Davinder Singh can never be realised.

Acknowledgement is not a Remedy

At its heart, the Davinder Singh judgment is historic as it brings the law into reality. Its significance lies in what it finally admits: that the promise of reservation cannot be honoured by treating the most marginalised as legally identical to those only marginally less marginalised than the general population.

For years the judicial system had treated the Scheduled Caste community as a single, uniform group. On paper, the entire community was given equal access to affirmative action. However, reality told a different story, some communities were far more disadvantaged than the others. As a result of the blanket quota, those communities who were slightly better off, ended up absorbing most of the opportunities, creating hidden inequalities.

The Davinder Singh judgment matters a great deal for the Madigas, the Musahar, and the Valmiki, for all of whom the reservation provisions had been a right in name and a vacancy in practice. A right that existed on paper, but barely resulted in a job, a seat in a classroom or a path upward. This judgment gives the government’s power to route support directly to those standing at the very back of the line.

However, this is not the total cure, and acknowledging a wrong does not automatically fix it.  Whether this judgment brings changes it seems to bring, will depend on the lawmakers. Whether the Parliament and state governments build the real-world machinery part of which is collecting data fairly, while ensuring that the process isn’t hijacked by political biases, will determine the possibility of reform. At its best, constitutional law shouldn't just be a dull diary from the past, but a bold practical tool to build a future we actually hope we owe.

---

This blog has been authored by Pragati Singh, a student at Gujarat National Law University, Gandhinagar. It is a part of RSRR's Rolling Blog Series.