Asad Tariq on Electoral Redistricting and Public Goods Provision in India

SHRUTI RAJAGOPALAN: Welcome to Ideas of India, a podcast where we examine academic ideas that can propel India forward. My name is Shruti Rajagopalan, and this is the 2025 job market series, where I speak with young scholars entering the academic job market about their latest research on India. 

Our fourth scholar in the series is Asad Tariq,  who is a doctoral candidate in Economics at the Indian Institute of Technology, New Delhi. His research focuses on the political economy of development, with a particular interest in religion, politics and public service delivery in India. 

We spoke about his job market paper titled, Constituencies of Change: Electoral Redistricting and Public Goods Provision in India. We talked about the 2008 delimitation exercise, especially at the state level, gerrymandering, the median voter versus swing voters and ethnic groups, public service delivery for minorities, especially Muslims, and much more. 

For a full transcript of this conversation, including helpful links of all the references mentioned, click the link in the show notes or visit mercatus.org/podcasts.

Hi, Asad. Welcome to the show. It’s such a pleasure to have you here.

ASAD TARIQ: Thank you. Thank you so much. It’s nice to be here. Thank you.

RAJAGOPALAN: I’m really excited to dig into your research more generally. But in this particular paper, you’re looking at the delimitation exercise at the state level, in particular, the one that happened in 2008, and you look at how India basically redistricts its state-level constituencies. Because of that redistricting, what happens when Muslim voters become more or less concentrated within a given assembly constituency, both before and after redistricting?

The CliffsNotes or the headline finding is that when the Muslim share rises, that is, the concentration of that group increases, you find that these villages are more likely to receive more government-run or public schools. This is historically an area where the Muslim community has been underserved. But you don’t see a similar effect for other kinds of public goods, like clean water or electricity or roads or the classic public goods that we normally think of. You particularly study electricity and roads.

You think this is happening because of the delimitation, which you use as an exogenous shock to then measure this difference. Did I largely get your research paper right? Can you tell us more about it?

Packing and Cracking

TARIQ: Yes, I think you have caught the idea quite nicely. Starting this work, there were two motivations. One was primary and the other was secondary. The primary motivation was to learn: When minorities are packed into a constituency, how does that impact provision of public services towards them?

Packing and cracking are very classic terms in the literature of redistricting and gerrymandering. Packing refers to putting a bunch of similar types of people, so let’s say minorities, into one constituency, and cracking is the exact opposite. If a constituency had a majority of one community, you “crack” it from the middle and you divide it into two separate constituencies. Most of this literature is based in the US because the US is famous (or infamous, I should say) for gerrymandering and bias into constituencies. 

RAJAGOPALAN: Just to add, the US is infamous, and many other countries or democracies are infamous for this because there, the redistricting is a political or explicitly partisan exercise, where the redistricting itself is hoping to pack or crack as the nature of the exercise. Whereas in India, that has historically not been the case.

TARIQ: Yes, we will come to that. In the US, because of that, in most of the literature that looks at it, we are actually observing the impacts of partisan gerrymandering and partisan policies on long-term impacts on public services or policies. We do not actually have an exogenous experiment where you have packed or cracked minorities or certain ethnicities into a constituency. That was the primary motivation in terms of political and economic theory.

The secondary motivation was more rooted in India. I worked with Sam Asher at Imperial College London. I worked under his supervision for a year. He had this amazing paper with Paul Novosad and others, where they have data on 1.5 million neighborhoods in India. They find that Muslim-dominated neighborhoods are the least likely to have public goods, similarly for Dalit-dominated neighborhoods. That is a great paper, but I wanted to extend that in looking at, “How does politics interact with this underprovision of public goods?” That is in terms of how did I start; that is my primary motivation, and this was the secondary motivation.

From Theory to Ballots

RAJAGOPALAN: To connect the dots with the paper that Sam and Paul and their coauthors have written, the question there was just looking at how different minorities or groups experience public goods provisions. Here, you actually explain in a more dynamic sense what might happen if the groups increase or decrease in size. This really starts mattering because we have a first-past-the-post system.

The concentration of a minority increasing a little bit or decreasing a little bit may have larger effects on who actually ends up winning in that particular constituency or, even more generally, who ends up forming the government because of this kind of change in concentration. Am I thinking about this the correct way?

TARIQ: Yes, exactly. Also, I think in Sam’s paper, it was more about neighborhoods in terms of that the neighborhoods are enumeration blocks defined in the census, and people are living in there. Then you look at, “In enumeration block one, what are the number of schools this block gets?” In my paper, I have villages instead of enumeration blocks, so I’m looking at rural India. They remain the same. At least I assume that people are not moving in and out of the villages significantly.

What is happening is that the boundaries are being redrawn. It becomes more dynamic, as you said. I look at underprovision of public goods, and I interact that with politics. How does that impact it in the long run, because I have a 22-year panel?

Median Voter Logic: A Mechanism in Play

RAJAGOPALAN: What is the exact mechanism, because you are looking at this over a period of time, and also more dynamically, and bringing in the politics and not just standard fiscal distribution of resources or public goods provisioning? What, according to you, is the exact mechanism which links the delimitation exercise to minority concentrations and finally to the provisioning of public goods?

TARIQ: When the delimitation or the redistricting happened in 2008, what it actually did was it rearranged certain villages. Let’s say village A belongs to constituency 1. Now it belonged to constituency 2. Because this pattern changed, but the village’s population, that remained the same. Within village, it is the same, but the constituency it was assigned to, that changed. I’m using that as an arguably exogenous variation. I’ll discuss more on that.

In terms of what is the mechanism, what I’m saying is that—let’s say there’s a politician who wants to win the election. Political theory suggests that they usually cater to the median voter. If more Muslims enter the constituency, that median shifts towards Muslim preferences. Therefore, I argue that this is a plausible mechanism because what is happening is that if more Muslims are entering, and the politician is catering to the median voter, the median is shifting towards whatever the Muslims, as a community, prefer. That is one of the potential mechanisms.

Empirically, I also later on show that the constituencies where the Muslim population increased are obviously much more likely to elect Muslim candidates as their member of legislative assembly.

Delimitation as an Exogenous Shock?

RAJAGOPALAN: Which is not unusual in India at all. Now, to start at the beginning. For you, you talk about how a largely exogenous shock is the delimitation commission. For me, that’s the most interesting or unusual part of the paper, especially given current times, though you look at this a while ago. You argue that because the commission is chaired by a retired Supreme Court judge, the design and the implementation is largely insulated from partisan control. It’s not that this is not a political exercise. Everything is a political exercise, but it’s not a partisan exercise in India the way it is in many other countries, where there’s obvious gerrymandering.

Now, can you tell us more about how this exercise happens, especially given that in modern day, we’ve heard a lot of news reports on how the delimitation commission is no longer apolitical or nonpartisan. Especially in border states, in Jammu and Kashmir, in Assam, we’ve had lots of issues just this year. One, can you walk us through the process of delimitation at the state level, and then what happened in 2008?

TARIQ: I completely agree with what you said last. There have been a lot of news reports about the much more recent delimitation in the border states. This is not a commentary on what is happening right now. I only particularly look at the 2008 delimitation because that is my entire identification or the exogenous treatment. The last delimitation before 2008 happened in 1971, and after that, it was supposed to happen in 2001. The Parliament passed a Delimitation Act in 2002.

However, that Delimitation Act—what it did was it made a committee, which was chaired by a retired Supreme Court judge and other members of the Election Commission and state election commissions. Then they started collecting data and tried to bifurcate constituencies. One of the key things that happened is that the Parliament froze the number of constituencies. Fundamentally, across all democracies—whether it be the US, the UK, Canada, India—the idea of redistricting is the idea of one person, one vote. You equalize the number of people across all constituencies.

Because of a plethora of political reasons in India, primarily, it had to deal with that some states said that, because we have a lower population, it is impacting our representation in the legislature, so you cannot penalize us for having lower population. Because of those reasons, in each state, the number of constituencies was frozen. In India, redistricting did not imply equalization of population across the country, across all constituencies. It only equalized population within each state.

RAJAGOPALAN: This didn’t affect parliamentary constituencies, only legislative assembly constituencies, basically?

TARIQ: No, it affected both. It affected both. Even the parliamentary constituencies for each state remained frozen.

RAJAGOPALAN: No, I’m saying within each state, you will equalize within the state for legislative constituencies. Between states, you won’t equalize for parliamentary constituencies.

What I mean is, within each state, they are still going for equal-sized constituencies, largely, because each state is its own thing and is not affected by the 2001 freeze, which then got kicked down to 2026 and so on. For the legislative assembly seats, we still have one person, one vote within each state.

TARIQ: Yes. Within each state, the one person, one vote stands for the legislative assembly. I look at the legislative assembly constituencies. After this was done, the Election Commission started redrawing these boundaries. Once they redrew it, they made it public through the Indian government gazette and the state gazettes. Then they asked people or political parties to challenge it. At times, people did challenge it, and then they redrew it. This process took some time. Eventually, the final boundaries were made public in 2008.

RAJAGOPALAN: One of the criticisms—and I don’t know how much this impacts your paper because you’re very specifically looking at Muslim concentrations, but if I remember correctly, in 2008, a major criticism of the delimitation exercise was that there was a lot of imbalance between urban and rural voters. Because since the previous census, within each state, the urban and rural share had dramatically changed.

One major criticism was, most migrants were either from the highest decile of the socioeconomic order or from the bottom two deciles of the socioeconomic order. At which point, there is a question of concentration, not so much maybe based on religion or caste, but there was a question of inequality and how we are redrawing the lines between the rural and the urban areas. How do you think about that question interacting with this question of Muslim concentration?

TARIQ: Ideally, I would love to include the urban areas in my data as well. The problem was that the data for urban neighborhoods did not exist. That is why in my paper, I primarily look at rural India, which is the villages. For example, you said that a lot of people have migrated. I cannot say much about that in terms of people migrating to urban areas, because I am only focused on what is happening in the rural areas, primarily because of data limitation.

RAJAGOPALAN: Sure. What I mean is that you may have changes in concentrations because Dalits, because of poor opportunities in the village, may have left in larger numbers, therefore changing the concentration of Muslims in a particular village or something like that. Are you worried about those kinds of effects?

TARIQ: For the village-level population, I observe it in the year 2013. It is roughly in the middle. My pre-period goes from 2000 to 2008, and the post is from 2008 to 2020. I do not have dynamic population changes in the village level because the census does not release the Muslim population [data] at the village level. What I had access to through Sam was this scrape data from the SECC [Socio Economic and Caste Census], which did have population estimates for the village level. I have data in 2013.

What you are saying is true that a lot of Dalits might have migrated to urban areas, but that would be true if I were looking in the year 1976 and then comparing it to 2008. I have data from 2013. I do not think there would be very significant migration in terms of Dalits or Muslims from 2008 to 2013. Moreover, I also do this very basic exercise of looking at the correlation between rural Muslim population and urban Muslim population in a constituency. These two are very highly correlated with each other.

RAJAGOPALAN: Fairly stable from what I understand?

TARIQ: Yes. At least in the 20-year period data . . .

RAJAGOPALAN: In the 20-year period that you’re looking at.

TARIQ: One of the reasons why I specifically look at Muslims is, as compared to other marginalized groups in India—let’s say the Scheduled Castes or the Scheduled Tribes—is primarily because there have been political reservation norms for the Scheduled Castes and Scheduled Tribes. There is a plethora of literature on how this reservation has impacted their development outcomes. I did not want my results to be confounded by political reservation as well as minority concentration.

For example, I could have done the same exercise for, let’s say, Scheduled Castes. I could have looked at their constituency-level population share and their village-level population share. What I was afraid of is that they have reservation at multiple levels. There is reservation at the Panchayati Raj level. There is reservation at the constituency level. There is also reservation at the parliamentary constituency level. Especially at the Panchayati Raj, that keeps on changing. If I do not keep track of that, what exactly am I observing?

Am I observing purely the change in Scheduled Caste concentration, or am I observing the impact of reservation? Or let’s say, even if it went out of reservation, there might be lasting impacts for that. To isolate the effect of simply the packing of minorities or cracking of minorities, I look at Muslims because historically there has been no political reservation for Muslims.

RAJAGOPALAN: You also have the problem on the other side, which is the public goods provisioning side. Just like you have reservations at the parliamentary level, state level, and the local level, you also have public goods which are financed by the union government under a centrally sponsored scheme, but they’re a state subject, but they’re actually implemented at the Panchayati level. You have this crazy confusion going on. Then it becomes very difficult to match which elected representative is actually responsible for which public good provisioning in India, sometimes.

TARIQ: Yes, that is quite tricky. Honestly, I think, at all levels, people do get a say. There is a certain level of discretion at all levels, and that discretion also includes bureaucrats. In my paper, I use MLAs [Members of Legislative Assembly], who are political representatives at the state level, simply because there is existing literature on it. Sonia Bhalotra and Lakshmi Iyer, they have a paper which looks at if a Muslim MLA is elected, how does that impact the education outcomes and health outcomes for Muslim children? There was some evidence which pointed towards the fact that, look, MLAs are important.

Anecdotally, as well, I felt that the MLAs are politically large enough to have a lot of discretion and to exercise a lot of power, but they are also not too big like a Member of Parliament, who commands 2, 3, 4 million people in their constituency, and they usually, at times they reside in Delhi. Half of the time they are not even in their constituency. Just thinking about these things, I thought that MLAs might be a reasonable middle ground in terms of—if I go very minute, let’s say, Panchayati Raj.

That was one of the reasons, apart from the fact that the existing literature does point towards the fact that they have been important, specifically for Muslims in terms of provision of education and healthcare.

Does Identity of the elected leader matter?

RAJAGOPALAN: One question I have, I think it’s important to discuss at this point, is: Do you find that with an increase in concentration of Muslims in a particular legislative seat constituency, does that impact public good provisioning conditional upon electing a Muslim leader, or is it irrespective of electing a Muslim leader?

TARIQ: I find it to be irrespective of electing a Muslim leader because, as I said . . . In a very recent iteration of the paper, I also wrote a theoretical model to further explain the results more mathematically. At least what I feel is happening is that the median is shifting. If more Muslims enter the constituency, the median preferences of the overall constituency shift towards Muslims. Even if the candidate is not Muslim, they want to win the election. If you want to win the election, you need to cater to whoever is the median voter. In that context, you will be much more inclined towards serving the Muslims.

Enter: Swing Voters 

RAJAGOPALAN: Is it that the median is shifting, or is it that in a first-past-the-post system, a small group—even if a small group swings—could change the electoral outcome? Which of the two is really going on? Because my intuition would have been the latter and not so much the median voter. This is weird in India because in India, we very much vote along ethnic lines or religious lines or caste lines, which may not always be the case in other parts of the world. It’s a multiparty system very much based on ethnic, linguistic, caste, and religious lines.

TARIQ: There are primarily two points. One is the median voter, and another is the swing voter. What the median voter says is that a politician caters to whatever is the median preference because that has the highest payoff in terms of the votes that that politician can get. The swing voter says that each politician has their core vote secured. No matter what they do, they will get that core vote bank anyway. There is a small percentage in each constituency which might swing one way or the other. That is what the politician targets to achieve or to please.

Empirically, because data limits are there, I cannot observe community-level voting preferences. I cannot observe if Muslims are much more “swingy” than other groups. Although there is some political science literature by CSDS Lokniti, where they do look at Muslim voting patterns. There might be some evidence, at least qualitatively and not at a large scale, but still, there might be some evidence that Muslims have shifted—I don’t want to use the term loyalty, but let’s say they have shifted preferences. 

For example, if you look at the Uttar Pradesh legislative assembly elections, Mayawati had the Dalit vote bank. The Samajwadi Party had the OBC Yadav vote bank. Both of them won one election each before the BJP came in power. What was happening was that Muslims were shifting.

RAJAGOPALAN: Muslims supported the Dalits once, and the Muslims supported the Bahujans the other time. The individual Muslim was not necessarily the swing, but as a group, when you vote along ethnic lines, the group may actually be the swing group, which changes the electoral outcome.

TARIQ: It is also up to the perception of the politician. As long as the politician perceives that this group might swing, then that works in my model. Mathematically, I show that if a group is much more swingy in nature, then withan increase in their population or concentration at the constituency level, they might receive more public goods, especially the kind of public goods that they prefer. If you want to look at it from the median voter perspective, that is also justified because as more Muslims enter the constituency, the median vote shifts.

Schools, Roads, and Wires: Evidence on Public Goods

RAJAGOPALAN: Either way, you’re saying that because this is fundamentally about voter preferences and politically self-interested politicians who will cater to those preferences, here the dominating preference will be greater access to education because that’s the underserved public good or the underserved need. They are much more likely, knowing that there’s a higher concentration of Muslims, and they would either shift the median or they would vote as a group. Did I get the mechanism right?

TARIQ: Yes. In my paper, I look at three types of public goods primarily because these are the three public goods for which I could find village-level data. One is schools from the UDISE, which is a dataset maintained by the Ministry of Education, and it’s the universe of all the schools in India. The other is the rural roads from Pradhan Mantri Gramin Sadak Yojana, which is a rural road program sponsored by the central government. The third one is electrification of villages, which is also through a central government-sponsored scheme.

One thing to note is that electrification does not mean that the electricity actually reaches the village. It is just the first-order condition, the bare minimum condition that the electricity poles and the wires are there. That is it.

RAJAGOPALAN: The pipes have reached the village, but it’s not like if you flip the switch, everyone will have 24-hour electricity.

TARIQ: Yes, or even 5 hours. I do not observe the hours of electricity.

RAJAGOPALAN: Yes, depending on what happens.

TARIQ: Yes, it’s zero and one. That’s it. It’s just binary. Among these three, for the two that I do not observe effects, I have two answers why I don’t. The first one is what you have already stated. In the baseline, I do not observe a significant bias against—not bias, but let’s say higher Muslim villages were not under-provided with rural roads or electrification, but they were under-provided with or were less likely to have a government school in the baseline period. Because of that, I also use the REDS data, which is a survey of 250 villages.

They do ask the village head to rank. They give them a set of nine public services and they ask, assign a number of importance to it. They need to rank it from one to four, one being the lowest importance and four being the highest importance. The more Muslims in a village, the higher the importance for schools and education. However, I do not find this significant for rural roads or electrification. Using these two data points, I say that, okay, it might be the case that because Muslims were under-provided with schools, you crave what you don’t have. You want schools because you did not have schools. 

The second thing to note is that the two schemes, the rural road schemes and the electrification schemes, were sponsored by the central government. Because they were sponsored by the central government, relatively speaking, MLAs might actually have less discretion. I’m not saying they will have zero discretion, but they might have less discretion. That might also be coming into play there.

Crunching the Numbers

RAJAGOPALAN: Now, when you look at this, how do you measure something like this? 

TARIQ: Sure. I have the Muslim population at the village level. I assume that remains fixed—not the population, but the population percentage. It remains fixed because I cannot observe it dynamically. I only observe it at a point in time. For the constituency population, I use maps. I have digital maps through SHRUG, which is a very good resource for data users, if anyone is listening and is interested in Indian data—again by Sam Asher, Paul Novosad, and their team. I use digital maps from SHRUG to overlay the village maps with the constituency maps, both before and after delimitation. Through this, I link the village to constituencies, again, before and after delimitation.

I have the village-level population for Muslims. I calculate the constituency-level population for Muslims by aggregating, again, before and after. For each village, I subtract the after-Muslim population from the before-Muslim population. That becomes my continuous treatment in the sense that if it is zero, then no change. If it is positive, then it increases. Negative, it decreases. Village population proportions remain the same. I interact it with a binary variable, which takes value one after the year 2008, which is a classic different-issue strategy.

RAJAGOPALAN: What do you find?

TARIQ: I find that for higher Muslim villages where there has been an increase in Muslim population at the constituency level, they are much more likely to get a government primary and a government above-primary school. I do not find any impact from private schooling, which makes sense because it would be weird if private schools increased. I do not find any impact on electrification of villages and rural roads.

If a 100% Muslim village shifts to a constituency where the Muslim population share increases by one percentage point, it raises the probability of getting a primary school by 0.24 percentage points and then an above-primary school by 0.19 percentage points. Relative to the mean, this is equal to a 1% and a 1.4% increase, respectively.

RAJAGOPALAN: Now, you mentioned a second ago that this shouldn’t affect private school provisioning. I found that quite curious. If Muslims have largely been underserved by government schools, wouldn’t you expect that private schools fill that gap? And when the government school gets built, there might actually be a decrease in private schools, or does that not matter because the populations have also increased?

TARIQ: In the baseline period, in the correlation that I observed in terms of schooling, the higher the Muslim population in the village, it is negative all across the board. Less government schools, less private schools. The only positive correlation is with madrasas.

RAJAGOPALAN: They have their own schooling being provided, which is religious schooling, but both the government and the private sector are underserving Muslim majority communities when it comes to schooling.

TARIQ: Yes. Also, I think madrasas are, one could argue, although they are recognized by the government, most of them are run through charity. It is some form of collective private-public good.

Drawing the Lines: Gerrymandering Then and Now

RAJAGOPALAN: It’s also difficult to measure the differences in the kind of education they provide, and so on and so forth. Whereas with the government school, this is what a government elementary school will provide. This is what a government secondary school will provide. Now, I guess my next question is, to go back to our head of the conversation where we talked about how you had assumed that the delimitation exercise in 2008 was largely exogenous in this particular study because it was done by a Supreme Court judge, and it was nonpartisan.

Now we have a lot of reports, increasingly, that the delimitation exercise has not just become political and partisan, but that, in particular, one of the targets is the Muslim community. We find the most evidence for this in the post-2019 Jammu and Kashmir and Ladakh areas, and also in Assam. These are, of course, very complicated places which I want to underline. This is not one of the interior or one of the larger states. They are border states. Assam has had this huge citizenship registry exercise. Jammu and Kashmir and Ladakh, of course, have lost their statehood and become union territories.There’s been a revocation of Article 370. 

There’ve been a lot of bigger political changes, not just delimitation. Delimitation is happening under the shadow of those bigger political changes. We do find that the latest delimitation exercise in these two states has actually disadvantaged Muslims in that they have reduced their concentration in more constituencies than one would expect otherwise, or at least that’s what the news reports say. How do you square that with what happened in 2008? Is there a good explanation for what’s happening? Because to me, it feels like now your study is even more relevant, given what just happened.

TARIQ: In 2008, I’m not ruling out that there has been no gerrymandering or not a single constituency in India has been affected by partisan bias. What I’m saying is that first, by and large, across India, there is not enough significant evidence quantitatively to say that there was partisan bias. That’s not just me. I’m standing on the existing literature in this area. I can cite some of the papers. Lakshmi Iyer and [Maya] Reddy, they have a paper where they show that there has been no partisan bias in the sense that redistricting, it did not favor incumbents, neither it affected the seat vote curve responsiveness.

A seat vote curve is a very classic measure of measuring gerrymandering. Let’s say in an ideal proportional world, if a party gets 35% of the votes in a state, then 35% of the seats in the legislative assembly should belong to them. In a real world, especially in the first-past-the-post system and in India, we know that this is not often the case, and also in the US. In the US, seat-vote curve in the literature has been used to measure gerrymandering. They look at the seat-vote curve. What they find is that there is not a significant difference before and after. The key thing is that: before and after. If before it was already—let’s say in 1976 it was already messed up; it is not more or less messed up post-2008. The way that DID [difference-in-differences] works, that should not impact me because I’m only observing before and after. If that remains the same, it is not messing up my results. 

Second, there have been some claims, even in the Sachar Committee report, that Muslim-dominated constituencies were disproportionately reserved for Scheduled Caste. Francesca Jensenius has a paper where, quantitatively, she shows that this is not the case. That is also ruled out. Then there are a couple more papers where they show that even the influential politicians could not gain significantly favorable boundaries.

There is also a paper by Resuf Ahmed that shows that delimitation also successfully equalized voter population. It achieved the goal that it set out to achieve. Now, what I’m trying to say is that in the most popular ways that gerrymandering could be done, or the literature says in the US, that is not the case in India. At least that’s what the literature says. I further do this exercise where I also look at the compactness score of the polygons, which essentially means how compact a shape is. I do not find a significant jump before and after in terms of the weirdly shaped constituencies.

RAJAGOPALAN: If they are contriving to put a certain village in this side or the other side, then you should get very odd shapes, which is not what we see in India. 

TARIQ: The thing is, if I say, “Okay, here are 10 ways I tested gerrymandering does not exist,” someone would come up and say, “Look, there is an 11th way which you can test.” What I’m saying is that the existing literature and all the popular ways that I can think of, I looked at it, and I could not find any evidence. I’m actually working on a proposal where I’m trying to dig deeper into the facets of gerrymandering in India and what more we could do to explore it.

As of now, the existing literature points towards the fact that by and large, across the country, there is not significant gerrymandering or partisan bias happening in terms of political boundary. Now, coming to what is happening right now—all this literature is for 2008. Right now, what is happening does need to be analyzed quantitatively, but the reports are quite concerning to say the least.

RAJAGOPALAN: We’ve been reading the same stuff, right? No one’s done a polygon analysis or anything yet, and the elections haven’t happened yet to test for the outcome either. It does seem like suddenly there’s a lot more talk about delimitation exercise in these two border states, especially.

TARIQ: One could also argue about how independent the Election Commission is as compared to 2008. The concerns are there, and the concerns should be there. The last thing I would want is someone to take my paper and cite it as “No evidence of gerrymandering in 2025.” Yes, that is definitely not the case. At least that is not what my paper says.

RAJAGOPALAN: No, no, I totally understand. The reason I want to dig into it this specifically is because now I want to get into the policy implications of your paper.

If there is this greater benefit to a disenfranchised community through better provisioning of public goods to serve the preferences of that underserved community, now suddenly gerrymandering or contriving to reduce the concentrations of these populations in given constituencies is going to have some impact on public goods provisioning and their overall access to health, education, electrification, clean water, and so on. That’s where I wanted to reach after this discussion about what happened in 2008 and what’s happening in 2025.

TARIQ: Yes. In the future, if the boundaries are drawn in a manner where the minorities are further cracked and they lose whatever little political power they have, then yes, I do think that whatever public good provision they are getting would be even lower. Yes, that would have severe long-term developmental consequences for them.

Also, to backtrack in terms of delimitation, I forgot to cite this book. There is this book called Fixing Electoral Boundaries in India by Sanjeer Alam. It has a ton of chapters coauthored by other people. It’s a very good book if someone is interested in reading more about the redistricting procedure, especially the 2008 redistricting procedure.

Policy Stakes and What’s Next

RAJAGOPALAN: This is super helpful. Now, I’d love to talk about another paper of yours, which is quite interesting because we had your coauthor, Aarushi Kalra, who was here on the same series last year. She discussed this paper very briefly. We were talking about her job market paper, and we talked about this paper where she was looking at the famous, or rather the infamous, Rath Yatra as an identification strategy. I noticed when you were joining us on the podcast that you are now a coauthor on that paper. One, can you walk us through what has changed on that paper since Aarushi appeared last year?

TARIQ: Initially, that paper dealt with the Ratha Yatra. There is existing literature by Blakeslee which shows that the Ratha Yatra was positively correlated with incidents of riots through each district that it went. What that paper does is, it exploits this relationship between the Ratha Yatra and the riots. It shows that the districts through which the Ratha Yatra went, the closer you are, there is a positive correlation with a higher level of education for Muslims. This is a very counterintuitive result.

The way it was being explained earlier in the paper is, Ratha Yatra went through certain districts. It was correlated with riots. As a result, segregation increased. Through this residential segregation, over time, there was private provisioning of public goods. That resulted in a higher level of education for Muslims. That was the initial setup of the paper.

RAJAGOPALAN: Yes. That’s what Aarushi came and talked with us about.

TARIQ: I have been working on this with Aarushi for, I think, since February of this year. I had worked on the UDISE data, the school data extensively for my job market paper. I had a sense of that. I linked that to the PIN codes. We got our hands on a PIN code shift file for all across India—so, maps of PIN codes. Because Aarushi had the map for the Ratha Yatra, so we were able to calculate the distance of the Ratha Yatra from the centroid of each PIN code.

We used that to show that actually, after the Ratha Yatra, the private provisioning of schools—so the private schools actually shot up, and the government schools did not. What was being intuitively explained earlier, there is some substance to that now in terms of data. We also wanted to look at how occupation changed in the long run. We do also find some evidence of occupational shifts in the long run using the NSS occupation codes across multiple rounds of NSS. 

RAJAGOPALAN: The mechanism is still through segregation, right? That post-riots, there is a tendency for both communities, Hindus and Muslims, to segregate themselves. Now, as a consequence of that, there is a tighter targeting of whatever we’re talking about. It could be employment opportunities, it could be education opportunities, it could be other kinds of public goods, but that’s really the mechanism.

TARIQ: Yes. Right now, we are also thinking that it has to do with where you invest capital. Let’s say if in the pre-riot period, you were investing in physical capital—land, shops, cars, whatever—you stop investing in that because that capital is not mobile. What is mobile is human capital.

RAJAGOPALAN: There’s a big literature on this with the Jewish community in Europe, right?

TARIQ: Yes.

RAJAGOPALAN: Which is they’re more likely . . . I think there was a famous quote that said, “Why are more Jews not farmers? Instead, they are violin players and physicists and goldsmiths and so on.” It’s basically because the human capital is just far more mobile when you’re a persecuted community.

TARIQ: I think we do argue on those lines. We are still working on it. I think I’m to blame here because I’ve been very busy with the job market paper.

RAJAGOPALAN: No, this is fascinating. I’m so glad that you both combined forces and managed to look at this in an even more granular way to confirm the original findings and make them more robust. Asad, thank you so much for coming on the show. It was a pleasure to speak with you. Thank you for sharing your research with us.

TARIQ: Thanks a lot. It’s been extremely great to be here. Thank you.