#271 Some Slopes Are Slippery For Real
A Debate on Caste-based Census, A Trade BATman, and An Algorithm for Fiscal Federalism
India Policy Watch #1: A Matter Of Numbers
Insights on current policy issues in India
— RSJ
As the Rashtriya Swayamsevak Sangh (RSS), the ideological mothership of the BJP concluded its 3-day-long Samanvay Baithak in Palakkad, it came up with its view on the political hot-button issue of the moment, a caste-based census.
It couldn’t have been easy for it. After all, the core objective on which it was founded was to organise and consolidate the Hindu society that was otherwise riven with fault lines, none bigger than caste. The political success in the past decade of the BJP has been in achieving this consolidation through a combination of targeted welfarism, nationalism and the othering of the non-Hindus. This hasn’t been easy. It has required a relentless focus on launching big-ticket welfare schemes (Jandhan, Ujaala), creating multiple narratives of a new dawn (Amrit Kaal) and future prosperity (Viksit Bharat by 2047) and enabling tentpole milestones like the building of the Ram temple and abrogation of Article 370 that are dear to the so-called core Hindu community. As I have written in the past, this consolidating force has to contend with the centrifugal tendency that is natural to any large majority group like the Hindus in India. So, any concession by the RSS-BJP to an idea that could shake this hard-earned limited consolidation of the Hindu society would be an anathema to it.
A caste-based census is that idea. So, it was surprising to hear this from the publicity in-charge of the RSS at the press conference marking the end of the Baithak at Palakkad:
“The RSS thinks that definitely, for all welfare activities, particularly those targeting such communities or castes which are lagging behind – for whom special attention is needed – for that, if sometimes the government needs the numbers, it is a well-established practice. Earlier also, it (the government) has taken (such data), and so it can do it again. But it should be done only for the welfare of those communities and castes. It should not be used as a political tool for elections. So we put this forth with a line of caution for everyone.
As a Hindu society, caste and caste relations are sensitive issues. It is an important issue for our national unity and integrity. So, it should be dealt with very sensitively and not on the basis of elections or electoral practices and politics."
The demand for a caste-based census has been led by the Congress since the surprising results of the general elections in 2024. What the opposition took away from the results was that peak Hindu consolidation had been reached, and with rising economic inequity, unemployment and a sense of fatigue on nationalistic issues, the time was ripe to unwind the majority-based coalition that the BJP had built over a decade. The government was dragging its feet on the census itself, and the Congress found in caste-based census an easy stick to beat the government with and to paint it as a party run by and for the privileged upper castes and elites who don’t want their cosy lives to be upended by such a census. Whether the Congress has a genuine belief in the benefits of such a census or has fully appreciated the long-term political impact of such a move on its own political fortunes is somewhat irrelevant now. As the opposition that has been out in the political wilderness for long, they see this issue as an expedient tool to accelerate the de-consolidation to hurt BJP politically. They aren’t too exercised by all the possible consequences of such a move. That’s for the party in power to worry about. So, they have thrown their weight behind the caste-based census and put the ball in the court of the RSS-BJP to respond.
And that statement at the press conference is the RSS response. As you can see, it is going to be a tough balancing but by giving its broad assent to such an exercise, it has yielded to the agenda set by the opposition. It didn’t have a choice because the electoral risks of being against it were high. So, how will they play it from here? I can foresee a planned set of op-eds and public statements by ideologues and loose canons aligned to RSS-BJP to oppose the caste census so that the core base, largely upper caste, isn’t alienated. This two-track strategy will continue for some time to allow the core to vent for a while. But in some ways, the die is now cast. There will be a caste-based census. It is only a matter of time.
There are actually two different issues on hand here. One, should there be a caste based census? Two, what do you do once you have that data with you?
On the first question, as I said, arguing against including caste details beyond SC/ST in the census is futile. More data availability should lead to better analysis of a policy problem, giving us better insights on making a policy more effective in its design and implementation. Data, by itself, is apolitical. Its analysis and the conclusions one draws can and will be abused politically. But that possibility shouldn’t hold us back from collecting more accurate and comprehensive data. After all, it isn’t as if in the absence of such data, there’s no analysis being done. In fact, more flawed and politically slanted conclusions get drawn without a challenge when there’s no tested and official data available. Any further resistance to collecting caste data will be painted as a conspiracy of the minority upper caste elite to deny the legitimate claims of the historically oppressed majority. It is a battle no political party will find worth fighting.
The second question is more interesting - what do you do with this data once you have it? The obvious answer among those supporting such a census in India is to relook at reservations. The simplistic argument for this goes something like this. The Other Backward Castes (OBC) quota in most states is now capped at 27 per cent based on the SC judgment that the total reservation should not exceed 50 per cent. What if the caste-based census shows the OBC percentage in India is, say, 55 per cent? Should we not relook at the 27 per cent reservation for the OBCs in light of this data? After all, as the slogan goes - ‘jitni aabadi, utni bhagedaari, utna haq’. Should we not then use the caste data to push for reservations in non-government sectors? Because it will be clear from the data collected that the skew in favour of the elite upper castes is even higher outside of the government. Should we introspect and build consensus on the best use cases of such datasets that go beyond reservations, or should we let it be used for the instrumental purpose of bringing social parity and ushering in a genuine affirmative action revolution in society?
My suggestion would be to follow the first principles of policymaking when you have access to better and more data. That is to use that data and look back at what policies of the past have worked or not worked. Before formulating new policies, it is important to know what has been the track record of what we have already implemented. Why have some worked while others didn’t? Who have these policies benefited? What objective functions should we set for ourselves based on this retrospective analysis so that we have the right outcomes? Reservations might only be one among such objective functions. And possibly the one that needs a better understanding of its past effectiveness than other objectives.
A quick set of existing hypotheses that could be empirically tested around reservations itself can be listed here to make the benefits of such an exercise clear. So, we could have better answers to these questions:
How do we determine the duration of affirmative actions to help a community to improve its representation in the public sphere that compensates for the wrongs suffered in the past? How quickly or slowly has the impact of preferential programmes been seen in the financial or social metrics of these communities?
The old question of who has benefitted from these programmes in the past. Is it the so-called ‘creamy layer’ who have cornered most of these benefits across generations? How should one think of such layers as caste based census becomes available to policymakers? It is almost impossible to pull back a perk or an advantage given to a specific group.
What specific interventions seem to have worked more than others? For instance, have reservations in educational institutions worked better than job reservations or vice versa?
How should one think of what percentage should be reserved for a particular subgroup or community? Is it always the moral case that your share in society and economy should be the same as your population share? If we go with that, it is possible a lot of positions will go empty because there might not be that percentage available or interested in a particular education stream or a job from that community. This is already seen in India where thousands of reserved seats go unclaimed today.
There are many more such questions that could have better answers once we get the caste-based census data. But I am not too hopeful that this is how the data will be first used. Like with the implementation of the Mandal Commission report and subsequent expansion of reservations, the primary motivation of this data would be a demand for greater affirmative action in the same vein as has been done in the past without attempting to learn any lesson to make such policies more effective. The spectacle of further demands for caste groups to be recognised as OBC, fraudulent claims of belonging to such groups and increasing polarisation on sub-caste lines will play out. We will learn the wrong lessons.
P.S.: I thought it would be useful to put the critique of the conventional way of thinking about affirmative action by Thomas Sowell here. Sowell is perhaps the greatest living economist and public intellectual. He’s black, has seen discrimination first hand and has studied the impact of affirmative action deeply. Here’s an extract from his seminal study on affirmative actions around the world from the website of the Hoover Institution:
Alternatives to Affirmative Action
Concern for the less fortunate is entirely different from imagining that we can do what we cannot do. Nor is the humbling admission of our inherent limitations as human beings a reason for failing to do the considerable number of things which can still be done within those limitations. In America, at least, history has demonstrated dramatically that it can be done because it has already been done.
Americans need only look back to the beginning of the twentieth century to see what enormous social and economic progress has been made by some of the poorest and apparently least promising segments of the population. At the beginning of the twentieth century, only about half of the black population of the United States could read and write. Jews lived packed into slums on the lower east side of New York, with more overcrowding than in any slums in America today. As late as the First World War, the results of mass mental testing of American soldiers led a leading authority on mental tests to conclude that it was a myth that Jews were highly intelligent. The situation of Chinese Americans looked so hopeless that a popular expression of the time described someone facing impossible odds as having “not a Chinaman’s chance.”
Not even most optimists would have predicted at that time how much all these groups would rise over the next half century—before there were preferences or quotas. Even for blacks, at the center of current controversies about affirmative action, the decline in their poverty and their rise in the professions were both more dramatic before the federal government created affirmative action in the 1970s. With all these American ethnic groups—and others—what happened was not a transfer of benefits from the rest of the population, but a rising contribution from these minorities to the growing prosperity of American society as a whole, from which both they and the larger society benefited, as the less educated became more educated, as farm laborers and domestic servants acquired the skills and experience to take on more challenging work. This was not a zero-sum process, while redistribution is at best a zero-sum process, if it somehow manages to avoid disincentive effects and intergroup turmoil.
Why is this social process, with a proven track record, so little appreciated, or even noticed—and sometimes dismissed as a policy of “doing nothing”? Perhaps that is because, whatever its economic and social benefits, it offers few rewards to politicians, activists, and intellectuals or to those who wish to seem morally superior by denouncing society. The heroes of these groups’ rise are anonymous individuals, not public figures. Here is some history worth repeating—but only if the goal is the advancement of the less fortunate, rather than the aggrandizement of those who would be their guardians or spokesmen or elected officials.
A Rejoinder
— Pranay Kotasthane
I’m glad we have finally found an issue of divergence after more than 250 editions of near concurrence. In this rejoinder, I counter its three underlying assumptions and argue that a caste-based census is unwarranted.
But first, the points of agreement. We indeed seem to be inching towards a caste-based census. If discrediting the opposition doesn’t work, the only way to puncture its political salience is to bandwagon with the opposing view. That’s where the politics of a caste-based census seems to be heading. So, it might well be politically futile to argue against it.
However, when I put on the policy analyst’s hat, I can’t help but think about the undesirable consequences of this move. The model I have in mind is that a caste-based census is a leverage point in a complex system (the Indian society), meaning that a small change in this policy can have non-linear, disproportionately large effects on the entire system. In this case, most of those effects are negative. At the very least, we must reflect deeply on the assumptions underlying the push for a caste-based census.
It is from this perspective that I want to challenge three assumptions.
#1 All Data is Good For Society
It is well-known that Indian governments have become more cagey about releasing data over the last decade. The no longer decennial census is a case in point. Given this dismal situation, the idea that a caste-based census will fill important data gaps seems enticing for well-meaning researchers, analysts, and observers. The underlying belief is that any data made public by the government can only improve things, and hence, regardless of the political consequences, a caste-based census makes ample sense.
That’s a wrong assumption. There’s a vast difference between government data on public service delivery and government data on society's social makeup. While the former is necessary for keeping governments accountable, the latter can dismantle, create, and reaffirm identities, the effects of which are felt over decades. Deborah Stone’s Policy Paradox succinctly explains the political effects of counting through these eight points.
Thus, all counting is political. But who counts is equally vital. A political party counting castes is qualitatively different from a State counting castes. Due to its authority, a State counting any identity lends it a sense of semi-permanency. Thus, once the State starts counting castes, we can pretty much discard the hope of annihilating castes even a century later. We must think deeper: how can an exercise that hardens a tragic aspect of Indian society eliminate it?
There’s also a second-order effect—the consolidation around caste will lead to another round of consolidation around religion. Mandal and Kamandal are not substitutes; they feed off each other.
#2 What’s Done With the Caste-based Data Can Be Left for Later
Another assumption is that the use of caste-based census data is independent of its generation; let’s have the data first, and we shall analyse the consequences later. This line of argument seems naïve because the sides supporting this exercise have made it clear that their eventual goal is to extend quotas to all spheres—public and private. Caste compositions are being sought in photos of finance ministry officials on the floor of the Lok Sabha, and speeches are being made hinting at caste-based quotas in “media and big companies”. Soon, caste-based compositions will be sought in our sports teams and armed forces.
Some might say that I’m being alarmist here. Political parties, after all, keep making promises they hardly implement once they are in the hot seat. Even so, this leverage point will set in motion a chain of events that will further drown out the larger Indian identity.
#3 A Caste-based Census Should be Welcomed by all, Whether Pro-reservation or Anti-reservation
Yogendra Yadav made this argument in his article “Ten half-truths of caste census.” However, I don’t quite understand how this data could be used to alter the reservation policy. If anything, the underrepresented sub-castes will seek quotas within quotas, while the overrepresented will try to find ways to fudge their caste and income certificates to position themselves favourably. New sub-castes will be conjured up, and demand for newer quotas will rise.
A caste-based census will reduce the logic of “social justice” to one and only one item: expansion of OBC reservation. If the goal were to reduce deprivation and provide more economic opportunities to the marginalised, politics would centre on improving our public education, health, and employment systems. And as Pratap Bhanu Mehta points out, none of these goals require a caste-based census. In his words:
“The most important things that are required for social justice do not require caste data. Making quality education available to all, the creation of public goods in which all can participate, the design of welfare or other cash support schemes, the best mix of subsidies and income enhancing measures, and most importantly, an expanding economy that creates mobility do not require the framework of caste. The mistake of the social justice agenda was that it forgot Ambedkar’s lesson that to effectively attack caste you have to (for the most part) strongly but indirectly attack the range of material deprivations that make its logic so insidious.”
These are my challenges to the core assumptions that make the caste-based census seem a benign counting exercise. It’s anything but.
PS: Observe how the goal of caste-based census has shifted from correcting grave historical injustices to merely privileging caste as a method for sharing resources in society. More on this topic here.
PPS: We recorded a Puliyabaazi on a caste-based census a few months back. There, too, the three of us co-hosts had differing viewpoints.
India Policy Watch #2: An Algorithm for Fiscal Federalism
Insights on current policy issues in India
— Sarthak Pradhan and Pranay Kotasthane
(This is a draft version of an article that was first published in ThePrint on 4th September)
Many other states expressed concerns when the 2024 budget provided special grants to Andhra Pradesh and Bihar. The Karnataka Chief Minister alleged that the budget overlooked Karnataka's interests; DMK announced state-wide agitations; and the Telangana CM regretted biased treatment. These reactions to the budget highlight long-standing apprehensions among states on fiscal allocations. Yet, there doesn’t seem to be a pathway to address such concerns amicably and sustainably.
Framing the Problem
All federalism debates focus on problems of horizontal devolution, i.e., the formula for sharing resources between states. A powerful variant is the claim that the South is disproportionately subsidising the North. But the real problem lies in vertical devolution, i.e., how the tax resources are split between the Union government and all states as a whole.
India tends to have a large vertical fiscal gap, which has been increasing. The reason is that while the Constitution assigns the most buoyant taxation powers to the union, it allocates more spending responsibilities to the states. The union can raise resources through broad-based taxes such as those on personal and corporation income, but the states have large expenditure obligations, particularly in important sectors such as education, health, police, law and order, forests, drinking water, etc. In some sectors, the share of state expenditure is disproportionately high. For instance, states incur 77% of the total government spending on health (2023-24) and 75% of the same on education (2020-21).
The Fifteenth Finance Commission pointed out that although the union collected 62.7% of the combined revenues of both the union and state governments, states alone were responsible for 62.4% of the total expenditure. Moreover, even as the union’s share in the revenue collections remained more or less the same over time, the states’ share in the expenditure has increased by around 10% in the last three decades.
To address this vertical fiscal gap, the broad-based taxes collected by the union are shared with the states according to the recommendations of the Union Finance Commissions (FC). However, the increase in the states' share from 32% to 42%, as proposed by the 14th FC, has not resulted in a proportionate rise in devolution to the states. This paradox is due to the increasing share of cesses and surcharges in the union tax pool, which are not shared with the states. Due to the lack of effective devolution, the states’ expenditure commitments suffer, which in turn affects the quality of public services, resulting in inter-state squabbles.
So, what’s the way out?
The focus on inter-state sparring on horizontal devolution distracts us from addressing the underlying problem of vertical imbalance. We propose an algorithm that goes to the roots of the problem, as shown in the decision tree below. The solutions are arranged in decreasing order of the fiscal benefits the states receive.
The ideal solution would be to increase the vertical devolution share to the states to 50%. This recommendation is not as radical as it sounds. It aligns with what the current Prime Minister said when he was the chief minister of Gujarat. If this option had been implemented in FY 2023-24, the states would have got an additional fiscal space of ₹2.24 lakh crores, increasing the combined revenue pool of all states by 5.42 per cent. Implementing this change all at once might be difficult for the union, so a less radical option would be a gradual increase of 1% per year to make the transition easier.
If the ideal solution proves to be politically unfeasible, the next-best option would be to include cesses and surcharges in the divisible pool to be shared with the states. The total amount of cesses and surcharges for the fiscal year 2023-24 was around ₹5.01 lakh crores. By including this amount in the divisible pool, the fiscal space for the states would have increased by ₹2.06 lakh crores, increasing the combined revenue pool of all states by 4.97 per cent.
If this option doesn’t succeed, the next-best option would be to reduce the number of Centrally Sponsored Schemes (CSS)—centrally designed schemes that are jointly funded by the union and states but wholly implemented by states. While CSS does augment states' resources, it has many problems. For instance, in 2023-24, the actual transfer through CSS was ₹1.12 lakh crore less than the allocated transfer as many states could not provide matching funds from their side. The lower absorption capacity stemming from weak governance and low GSDP of poorer states prevent them from deriving substantial benefits from the CSS. Beyond these inherent limitations, another fundamental challenge of CSS is the union's interference in the states' constitutional domain.
So this option would reduce CSS only to early education, healthcare, poverty alleviation, and perhaps rural employment guarantee. This restraint will decrease the expenditure burden on the union, create more fiscal space for the union, and curb the need for cesses and surcharges. If the CSS is limited to only education, healthcare (both preventive and curative) and social assistance, states can get an additional ₹1.69 lakh crore, i.e. around 4.09% of the combined revenue pool of all states. They can deploy these funds according to their own needs.
If none of the above options are feasible, the union government must commit to including non-tax revenues in the divisible pool. The union's non-tax revenues include interest receipts, dividends and profits from public sector units, and service charges. The union’s non-tax revenue collection for FY 2023-24 was ₹3.01 lakh crores. Sharing these non-tax revenues with the states would have led to an increased devolution of ₹1.24 lakh crores to the states, a 2.99% increase in the combined revenue of the states.
While the above options seem mutually exclusive, combining one or more approaches is possible. But first and foremost, there should be due recognition of the problem. The problem is not North vs South but that all Indian states have limited revenue streams for their spending responsibilities.
PolicyWTF: Shooting Oneself in the Foot
This section looks at egregious public policies. Policies that make you go: WTF, Did that really happen?
— Pranay Kotasthane
The Union Commerce and Industry Minister has proposed a border adjustment tax (BAT) to ‘protect’ India’s steel industry from cheap imports. This was in response to concerns by a steel industry delegation that cheaper imports from trade partner countries like South Korea, Japan, Vietnam, and Indonesia were hurting their profitability.
The facts are as follows: Steel prices haven’t increased since the pandemic. The price decline is attributed to low domestic demand and increased imports. Now, cheaper steel should be great news to Indian manufacturers of machine tools and engineering goods. They can now sell their wares cheaper, thereby improving their export competitiveness. But it is obviously a problem for domestic steel makers. As you would expect, they are clamouring for a surge in tariffs. Nothing new here.
What’s interesting is the chain of logic highlighted in a Business Standard report on this issue:
“The idea behind BAT is that Indian steel producers pay taxes like electricity duty, iron ore duty, and coal cess, while imported steel is often cheaper because other countries may not have similar taxes, giving their steel a price advantage. BAT would level the playing field.”
So, India's self-inflicted tax structure is the reason for declining domestic competitiveness. But why address the root cause? Instead of tackling these cost disabilities that make producing in India difficult, the government has taken its favourite short-cut: increasing import costs.
The unintended consequences are easy to anticipate. With this protectionist policy in place, our steel producers will have no incentive to reduce their costs, while Indian products using costly steel will become uncompetitive. The import substitution vibes are unmistakable.
PS: Do we need a Ministry of Steel in the Information Age? Shouldn’t it be a component of the Industry Ministry instead?
HomeWork
Reading and listening recommendations on public policy matters
[Podcast] Puliyabaazi completed 250 episodes this week. We began six-plus years ago on a whim. I’m energised that over this time period, we have captured some counterintuitive ideas and foundational concepts in an approachable language and format. The 250th episode features a conversation on doing business in India with Manish Sabharwal.
[Book] A new biography of M Visvesvaraya is out; a perfect read for the upcoming Engineer’s Day.
[Paper] Development through Economic Growth is a thought-provoking report by Stefan Dercon.
[Article] Dan Drezner rues “How Everything Became National Security” in his must-read Foreign Affairs article.
Amazing arguments on caste based census . Enjoyed this a lot .
Easily one of my favourite editions in recent times..good on you Pranay and team!