Kartikeya Batra on Long-Run Effects of Land Redistribution 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 2023 job market series, where I speak with young scholars entering the academic job market about their latest research on India. 

I spoke with Kartikeya Batra, who is a Ph.D. candidate in economics at the University of Maryland and has a master’s in international affairs from The Fletcher School.

His research lies at the intersection of political economy, development economics, and economic history. We discussed his job market paper titled “Long-Run Effects of Land Redistribution: Evidence from India,” and talked about the erstwhile Zamindari or Taluqdari system and its impact on modern day Uttar Pradesh, land titling versus land redistribution in India, the importance of property rights, relationship between land owning castes and socioeconomic outcomes, 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, Kartikeya. Welcome to the show. It’s a pleasure to have you here.

KARTIKEYA BATRA: Hi Shruti. It’s a pleasure to be on the show, and thanks for conducting this wonderful series.

RAJAGOPALAN: I’m excited to get into your job market paper.

In your job market paper, what you’re looking at are the long-run effects of reorganization of proprietary rights and property rights in land in the early 19th century in what is modern-day Uttar Pradesh, but that time, it was a motley crew of provinces that were strummed together.

BATRA: United Provinces.

RAJAGOPALAN: Eventually became the United Provinces. What you find is that the areas that moved away from the traditional Mughal and East India Company, Zamindari system we are familiar with now, they actually shifted to a Mahalwari system, which is typically at the village level. A lot more landlords as opposed to giant feudal estates is the way I would think about that.

The shift, small bit of it, happens over the late 17th century and 18th century, but really between 1820 and 1860 is the period that you are studying. Now, we are 200 years later. What you find is that, today, the areas that shifted towards the Mahalwari system compared to the erstwhile Mughal Zamindari system have higher wealth, higher non-farm employment, more years of schooling, essentially better economic outcomes and social outcomes in modern-day areas of Uttar Pradesh that reformed themselves almost 200 years ago. So, can you tell us how someone goes about studying this question and what exactly you’re doing in this paper?

Land Redistribution in Uttar Pradesh: Positive Effects

BATRA: Before that, Shruti, that’s actually a very wonderful summary. Taluqdari system was a remnant of the Mughal system, and those Taluqdars were recognized as landowners in one part of Uttar Pradesh. In the rest of Uttar Pradesh, the East India Company took land, which was earlier under the control of the Taluqdars and distributed land titles to local village communities. Now, understand, in the Mughal period, the whole state is under Taluqdars, who are revenue collectors who owe their allegiance to the Nawab.

Although they’re just revenue farmers, but they are exercising a lot of de facto power in their area, and they are stamping out proprietary rights of these village communities who’ve been living there for God knows since when. When the British come in, they annex Uttar Pradesh in a staggered manner. In large parts of Uttar Pradesh, 77%. If anybody here listening to this podcast is aware of the geography of Uttar Pradesh [UP], what I’m talking about is Western UP, Southern UP, which is Bundelkhand and Purvanchal, which is Eastern UP. Those parts are first taken over by the British during the early 1800s.

In the next 20, 30 years, they introduced the Mahalwari system where, as you rightly said, two, three things are done. One, local village communities are identified, who they think were the initial owners of land. They are given proprietary titles, and they’re also made liable for revenue payments. That’s happened in the treated areas. Which is why this is redistribution in the sense that you took land from powerful revenue farmers, paid them off, and you redistributed land—on average, of course—amongst smaller village communities, and you gave them titles.

In the other area, which is the central part of UP, which is Awadh, annexation is held back until the 1850s for political reasons. 1850s, when finally it’s annexed, their initial plan is to actually implement the Mahalwari System even there, which is why this variation is sort of exogenous. But what happens is the mutiny of 1857, which is a national event. Now, they want to ensure that they retain the loyalty of the Taluqdars, who were very politically powerful in their area.

In central UP, to be clear, in the area around Lucknow—let’s say, broadly—what they do is they say, "You Taluqdars, even you were a little not so loyal towards us during the revolt, so we’re going to confer your lands. But wait, if you submit to us, and you pledge your allegiance to us, we’ll make you not just revenue farmers, we’ll give you proprietary titles." Now, you have a situation where land titling has happened in both areas within UP, but in one area, it’s happened amongst a number of people, smaller village communities, but in the other area, the previous de facto system is made de jure in a way, or the old system is continued to function with new proprietary titles with the Taluqdars retaining their political powers now as landowners. That is the basic structure.

Now, how I got motivated by this question was honestly, Shruti, I was just thinking, why is UP the way it is? I just went on this journey two years back to figure out what can be done to figure out why, because I do a lot of election analysis. This was also the time in the lead up the UP elections. When I was studying the politics and economy of UP, I realized it’s so different in different parts. It’s like studying five different European countries packed into one. Of course, there are several complex factors that go into the long-run evolution of socioeconomic outcomes.

One of the factors I zeroed in was the land tenure system or the land ownership system. That was the motivation. Then I just told you the setup of the policy experiment. Within UP, the two areas were separated by a boundary, which I, in my paper, have tried to convince the readers was arbitrary, which then allows us for a spatial RD methodology. Then I aggregate village-level data from three different village-level data sources, and as you said, we get access to data from population census. We get years of schooling from the SECC, and we get non-farm employment from the economic census. This is all from 2011 to 2013.

Then I run a spatial RD, UP has around more than 100,000 villages. My sample is limited to 10,000 villages around that border. As you rightly said, I find long-run positive effects of that land redistribution in the long run. Interestingly, the literature on land redistribution—although theoretically, is pretty robust—empirically well-identified papers are not that many. One of the problems is this only that there’s endogeneity policy implementation. Which state does it? Usually, it’s done at a state level, or on a national level. You can either do cross-state comparison or cross-country comparison.

Then, that way, this policy experiment, which by the way, has also been used in a previous paper by Priyanka Pandey, in a different context, 2010, is different because it’s intrastate, so it takes care of a lot of factors that probably interstate comparisons don’t take into account or can’t take into account. That’s the basic premise of the paper.

The Link Between the Past and Present: Land Reform, Caste, and Change in Social Norms

RAJAGOPALAN: What is it that you think is driving this result, that the modern-day outcomes are now linked to something that happened in the past. What is the mechanism through which this is transmitted?

BATRA: Exactly. Of course, as I said, land redistribution is like one shock, which then sets several balls rolling. It’s almost impossible to tease out all potential mechanisms, but the mechanism that I went through was caste. Again, this is all because of my interest in electoral analysis. If you’re analyzing elections in any Indian state, you have to figure out what the caste equations are. One thing that one has to understand is how inextricably linked land ownership and the caste system have been over the centuries. Essentially, they mutually reinforce each other.

You are upper caste, so you’ll own land. You own land; therefore, you are in a power position. As I said, there could have been several channels of persistence, but the one that I chose to focus was caste. One other result, which I think is new in this paper, which has not been seen in the two broad strands of literature I’m contributing to, one is land redistribution—maybe it’s a little bit there, but not so much in the Indian context. And second is the literature on historical colonial institutions in India. You have papers of Abhijit Banerjee, Lakshmi Iyer. Lakshmi Iyer herself has done some work on princely state. There is Priyanka Pandey’s work.

Then in Indian context, land redistribution work mostly has been done by Pranab Bardhan and Dilip Mookherjee. There is paper by Besley and co-authors including Rohini Pande.

They talk a little bit about it, but here what I’m also figuring out is, okay, we had this land redistribution: done. We have longer persistent effects. But we have to understand that even back then land was only redistributed amongst the upper-caste households. In both the treated and controlled areas, land was held only by upper-caste households. When I say only, I mean roughly 70% to 80%, which means that redistribution amongst the scheduled caste (SC) was almost next to nil.

I went through colonial gazettes to actually identify [data]. I find no data where it says…Some OBCs [other backward class] did get some land when redistribution took place. But there is virtually no record of any land being redistributed among SCs. You use caste as a proxy for any household which was there back in the day which was a non-beneficiary. You find that in the long run, even the SCs who may not have received land back then have benefited. Now, what might explain that? They did not receive land, so they can’t be direct effects of wealth ownership. There has to be something else that’s happening. Again, land redistribution is such a complex subject, there could be several things happening.

One, for instance, is just the local economy changes: wages go up, credit constraints go down because somebody in your neighborhood owns land, they set up an industry or something. The second is, broadly speaking, that land redistribution somehow changes institutional factors in your area. Does that impact the politics of the area? Does that impact the sociology of the area? That’s where then I focus. My hypothesis was—and I did a field survey—imagine that you were used to living in an area where you had a bunch of feudal lords, very few. By the way, between control and treatment, the number of upper-caste landowners by the 1940s, the ratio was one to three: 17 landlords per 1,000 people in the control areas versus 49 in treated areas. Now, this shock to the land system happened because of the reform. Imagine you living in the treated areas, you were the control area for a very long time. You were used to the 17 landlords per 1,000 people: That there is one landlord who is huge, feudal Taluqdar, and that person is godly in nature, and I have to submit to them. Suddenly, this land system is shocked. Now you do not have one, but on average three landowners around you.

Of course, their landowning size also goes down. Of course, their political power also goes down. Does that in any way impact your compliance with social norms? This is coming from a scheduled caste perspective that scheduled caste houses are told, or they are conditioned that, "You can’t study beyond the point, you have to do this work, your child cannot do this job," et cetera. Once the higher echelon of that system starts shaking or their power gets reduced, now you have more outside options, there is more competition amongst them, et cetera, et cetera. Does that in any way affect your compliance with these restrictive norms, which may have been holding you back socio-economically?

In my field survey, I compiled a bunch of questions from different surveys, World Value Survey, et cetera. Tailored them to the Indian reality of the caste system, and I asked them questions that, "How much do you agree with the following: that if there is a government job, it should be reserved for the upper caste?" for example or something like that. Then combine indexes. I also asked them, "What do you think your grandson or son should become when they grow up?" Also asked them, "Do you agree with the current functioning of the caste system?"

Shruti, we found that in the treated areas, the support for the existing caste system was lower and aspirational mobility or aspirations and the desire for mobility was higher, compliance with these norms was lower.

My limited point here is that I think there could have been a bunch of complex mechanisms at play. One of the mechanisms at play, it seems, was that immediately after you redistributed land, there was a change in norms because the powers that be in the treated areas, their number went up, their influence per individual went down or per household went down. The folks who you were used to following them probably felt something, and that led to a change in social norms amongst them. I think I can pre-empt your next question, but I’ll let you ask it because a lot of people ask me, "Oh, but you know, how can you say norms changed first, and then socioeconomic indicators changed?"

Land Reform or Return of Property Rights?

RAJAGOPALAN: Yes, my question is not so much about norms actually. I want to go back to history for a minute. My question is more about the way you characterize what happened in UP in the 1820s. The way I think about property rights and proprietary rights in India especially as they were set up by the East India Company—it of course starts in the 1760s to 1790s in the Diwani of Bengal. India has never had the Anglo-Saxon system of property rights, right?

They were trying to import it, but to import it they would have had to do serious cadastral surveys, have proper titling the way they did in the south. Now I’m thinking what [Thomas] Munro did in the south for instance, almost 50-60 years later. But they don’t have the state capacity to do it in East India Company, of the Diwani of Bengal. What they do is they just transfer the entire system to the existing Mughal Zamindari system, and the Zamindars who were just very powerful but still just revenue collectors actually become property owners.

The right to buy and sell the Zamindari becomes de facto the right to buy and sell the land associated with it. Overnight, all the original property owners get converted into tenants where now the Zamindar is the landlord, and the East India Company is a super landlord mixed in with the sovereign of some sort. Now, there’s a little bit of this permanent settlement in the Jaunpur side, but leaving that aside, when we come to Uttar Pradesh, my major question on your historical research design is, it seems to me that this is not really act of redistribution. It seems to me that places that actually had a sensible village group council system, they decide to actually title that and recognize that instead of imposing some crazy thing like what permanent settlement did almost 50, 60 years ago.

In a sense, they’re learning from their mistakes, and they are reassigning or titling in a uniform way the property customs that already existed there.

BATRA: They did a survey in UP as well.

RAJAGOPALAN: I’m sure they did. To me, the way the British acted in the early part of the rule, which is the East India Company rule is, places where they had the capacity to actually enforce the rules and understand the rules and then collect revenue directly are quite different from the places where they didn’t have the capacity to do so. Now, this is where my question of endogeneity, finally I’m getting to it, which is, is there something unique about Oudh, as they call it, or Awadh in Indian parlance, is there something fundamentally different about Awadh where it is just so difficult to develop state capacity in that area for various historical reasons, for the mutiny and other things, that that’s the reason it ends up going to the Taluqdars, and in other places where you say it was actually redistributed from Taluqdars to the farmer, I don’t think that’s exactly a redistribution. In fact, it is a non-distribution because in other places, what East India Company effectively did was it took away the rights of the cultivator and instead made it a proprietary right for Taluqdars and Zamindars and so on. I guess that’s my question, would you characterize this as a redistribution, and if not, then do you suddenly face an endogeneity problem with the research design?

BATRA: I think I’ve mentioned this in passing in my earlier response, that if not for the mutiny, Awadh would have also faced a similar scenario. They were ready to do something very similar in Awadh. If you had Munro in south, the champion of this system in UP was Holt Mackenzie. He came up with this. There was this whole speech or letter, whatever you may call it and he said, "Oh, you need to do this," in 1822. They were ready to do this in Awadh, but the mutiny happened, which is why you can say that there was nothing special about Awadh, except for the fact that the timing probably went awry for them.

If Awadh had been annexed say 10, 15 years before 1856, probably the same system would have been in place.

Now to the point of use of whether this is redistribution or not. How would we generalize it in today’s framework? You are the government, and you find that a bunch of people have taken over land or control of land, which used to belong to some other community, and that’s happened in a pretty big area. As a dictatorial power almost, you step in, you take control of those land first away from these guys who are controlling the land. Then you say, "Let’s figure out who are the people who used to live here earlier."

We do a survey; we check exactly what happened. Of course, it wasn’t perfect, limited capacity, but it was there. Then whoever we think used to own this plot of land or had property rights before these Mughal invaders came in, we are going to assign them rights. Therefore, we will create smaller patta, as you would call them in India, patta being a land title. To my mind, this is different from a classic land redistribution policy where you have a land ceiling, and the excess land is redistributed amongst landless people. No, this is different. Of course, it’s very different.

In fact, it’s closer to what happens in, say, Colombia. There’s public land. The government just says, "I’m going to redistribute this land." Except that in this case, the government does expropriate land and says—this is probably in this transitionary phase, "It’s my land, but by the way, it used to belong to someone. Let’s identify who those people are." Then those people, when they take control over their pieces of land, they practice their own agricultural methods. Some communities are the ones who own and cultivate, some other ones who own but don’t cultivate, so they hire other laborers.

I would still say that titling happened in both places, but the distribution of land amongst people whose rights had been taken takes place only in the treated areas, not in the controlled area.

RAJAGOPALAN: Fair enough. I just don’t think it as traditional land reform. I just think it as recognition of property rights as opposed to de-recognition, which they do in other places.

BATRA: Of course. Titling happened in both areas. You can make the argument that Taluqdar’s titling was more like as a de-recognition or de-distribution, whereas that in the case of the treated areas is more like you’re giving it back to those. I think you may define it whichever way. There are two, three ways of defining it, but at the end, the first order effect.

RAJAGOPALAN: Yes. Is that there is a lot less concentration—

BATRA: Exactly.

RAJAGOPALAN: —of land in the hands of the few?

BATRA: Yes.

RAJAGOPALAN: I totally understand that. The reason I was getting into the land reform question is because that clearly sets your project apart from what happens in the post-colonial papers, which are studying the same question. Because that land reform and land redistribution is of a completely different category—

BATRA: Nature. Exactly,

RAJAGOPALAN: —than what you are fundamentally talking about. It feels similar because in both cases we’re talking about reducing concentration of land in the hands of the few and so on, but we don’t want conflation on these things. That’s the reason for nitpicking.

BATRA: That’s absolutely perfect. I agree with that. In fact, I’m working on this section contribution to literature, and these are the questions staring me in the face. One thing is that this policy is also different in the following way, that post-independence, a lot of policies happen under a democratic setup. There was only so much the government could do in terms of expropriation. Even if you read the work by Bardhan and Mookherjee, for instance, they make the point that in Bengal where land ceiling was implemented, redistribution was probably not very effective because poor quality land was redistributed.

When you have a colonial project like this, it’s so absolute in nature that probably some of those frictions that are a result of a democratic institution, which I’m all for, are not there. In that sense, you’re absolutely right, the policy is different.

RAJAGOPALAN: The uniformity that they were able to impose is quite different. I completely understand.

BATRA: The scope for probably fudging also, a manipulation of the policy by the landowners probably also was not as high, of course.

The Differences Between the Mahalwari and Taluqdari Areas

RAJAGOPALAN: A second question now, and here I want to dig into the difference between the Mahalwari areas and the Taluqdari areas.

The difference between the controlled and the treated area. In the Mahalwari areas, one of the things that’s going on is there is a collective action problem, so to speak, has already been solved by the community. The community has already got the resources where they’re able to figure out who is cultivating what, who owns what. They’re able to sort out the low-level disputes that can occur when it comes to a farming community or a village community, and they’re able to figure out the revenue collection system in a way that—that’s the reason it is decentralized.

Once again, places where that could be done automatically since the very beginning where the Mughal Taluqdars may not have had as much power to begin with, I imagine, which is why this system was working reasonably well. Does that pose a endogeneity problem again for what you’re trying to study? Because it’s quite clear that in Awadh, something was going on where it was just much harder for both the Awadhi cultivator and the company and crown rule to overthrow this union or cartel of Taluqdars. They were just really hard to get rid of.

Once again, is there something different in these two bottom-up approaches, even in minor regional differences? Because Uttar Pradesh as a state is only our modern-day imagination. I guess what I’m asking is, is western UP more like Punjab than it is like eastern UP? We’ve just characterized it and bundled it all in one place together. For instance, is western UP fundamentally different because it was closer to Delhi? We know now that places that were farther removed from Delhi, the Taluqdars and Zamindars had much a much higher degree of power because that’s how the Mughals delegated power. Places that were closer to Delhi, they didn’t have to delegate as much. They could have a lot more oversight. Do you think some of this stuff is going on here?

BATRA: The first empirical answer is that it’s a spatial RD within a very narrow bandwidth. Around the border, that’s the idea behind an RD that,10 kilometers here and 10 kilometers there, it shouldn’t be that different. Second, I’m controlling for distance to four important historical cities in UP in my main specification— Kanpur, Lucknow, Agra, and Allahabad— Three, because there have been capitals, and one, because it was a important commercial center. They’re mostly for precision. Even if I remove those controls, things don’t change a lot. Third thing is that—thank you for giving me an avenue to show off a little bit—I compile data from Ain-i-Akbari which is probably the only pre-colonial record that colonial gazettes cite.

I was able to link colonial tehsils to modern tehsils, and then colonial tehsils to the pre-colonial Mughal tehsils. The latter was a little easier because the gazette had the mapping. I controlled for four or five different factors which were present in Ain-i-Akbari. One was tax demand, second was presence of infantry, cavalry, and there was something called religious donations. When I controlled for those factors, assume that they are pretreatment and there is some imbalance, and they’re impacting the long run outcome, then that should change the positions. That doesn’t happen. Even after controlling for those pre-colonial factors the results remain robust.

Of course, the precolonial factor data is available at the sub-district level, so it’s not a perfect measure, but it’s the best we got. Also, Shruti, what I was able to document using some data from a paper by Dave Donaldson was that actually, initially after the Mahalwari system was implemented, they were doing not as well as one might think. Their initial performance from the period of 1880s through 1930, for which I have data, just compared to our just basic means comparison of the bordering districts around this border, actually Mahalwari areas and treated areas were doing badly, if at all.

So, something happened later. One of the reasons was that once you gave land titles to these communities, a lot of people started dividing land. They said, "Give me my piece, I’m leaving. Bye." This ties into modern literature, but again, Bardhan, Mookherjee, this other paper I think 2010, where they show how land redistribution leads to further distribution, and it leads to migration, et cetera. I would still say, Shruti, to whatever I have understood, of course, we are all learning even after the paper is published. From what I have understand is that, a) the boundary takes care of a lot of the factors that you just mentioned, in general, I don’t find any reason why Awadh was very different from—at least within that bandwidth, but then generally also.

Because as I said, if not for the mutiny, which was a national event, Awadh would have faced the same fate as the rest of the state. Plus, this is also the period when the East India Company is waning, and the crown is rising in India. The ideology is also different. The East India Company was being seen, from my reading of the literature, as too much of a social player, that you are interfering in traditions and practices in India. "Why are you doing all this?" The crown wants more status quo approach.

RAJAGOPALAN: They want stability.

BATRA: Exactly because they are a more political animal. Whereas the East India Company is more of a venture that has turned into a political creature. Again, there is no perfect answer, but in the paper, I have tried my best to convey that these factors should be more—then the last thing I’ve also done is, I used this world atlas from 1914, and I’ve shown that at least within the bandwidth, if you start from 2nd century BC until the 1815, you will find that both the treatment controlled area are probably falling in the same kingdom.

Except for one period in time starting 1398 when the southern boundary of this boundary that I am exploiting, collides with—it is the same as the southern boundary of a kingdom called the Jaunpur Kingdom. As a robustness check, I removed that boundary, and I re-estimate my results, so they don’t change. That boundary is also important because it’s the boundary of the Ganga’s flow so removing that serves two purposes. It’s robustness to the exclusion of the river, as well as the southern boundary of the Jaunpur Kingdom.

What About Jaunpur?

RAJAGOPALAN: Just as an aside, this is not crucial to your paper. Can we just talk for a minute what was happening in Jaunpur because that happened well before 1820? And that’s more of the permanent settlement of the Bengal Diwani system than it is of what’s happening in 1820s in Uttar Pradesh or in the south and so on. What’s going on in Jaunpur?

BATRA: There’s a treaty of 1775 which I think you’re referring to in Banaras where the British—this is probably their first steps into UP. They come in, and they take over. Again, I’m not 100% clear about this, but what happens eventually is, it is touted as a Zamindari, but de facto it is also Mahalwari. It’s the treaty of Banaras. Again, just so that I’m not conflating results, in my robustness check and remove that part of the boundary, so that just in case that’s driving the results, which is not, thankfully. That part of history had been settled back in 1775. That’s a very small portion, though. Majority of annexation happens between 1801 and 1803. Parts are annexed.

My treated control area, the way it’s done—again, that’s another reason why endogeneity my not be that big of a concern. So the British are trying to squeeze the Nawab of Awadh, and they are increasing his tax payment. There comes a point when he’s unable to pay them, and the British pretty well know this. Then they say, "You can’t pay tax? Why don’t you ease some land to us?" He goes, “You know what, take some land along the periphery of my kingdom." The treaty of Awadh is signed in 1801. In my paper, I show a snapshot of it. In that treaty what happens is, areas outside the border, which I’m exploiting, they somehow gives parts of it to the British. When you look at the treaty of Awadh, it’s not very clear how they’re doing it. It’s very random. Some complete districts are going. Some small blocks are going.

Anyway, the way that it’s finally determined is that the Nawab owes the company a few crores or something, some money. He says, "You take land which is worth that much." This is, again, just telling you that at the end of it, this was all not very systematic in nature. It’s just that the Nawab wanted to retain his central piece of the kingdom, which is very natural, but he said, "You just take a belt of the land, which is on the periphery of my kingdom, whose value amounts to the revenue that I owe you."

Focusing on a Region Rather than the Entire Nation

RAJAGOPALAN: Coming back to the underlying mechanism that you’re exploring, I just wanted to place it more broadly in the literature. Your mechanism is fundamentally one of concentration of assets, specifically in land, and concentration of assets in particular castes within a social structure. Now, like Alexander Lee’s paper, for instance, talks about the difference between systems. And therefore, the long run outcomes as a function of state capacity because what he’s looking at is Zamindari versus Ryotwari or British provinces versus the princely states.

What you find is that, in systems that are not Zamindari or Taluqdari where the British is more directly involved in revenue collection. Now, if you have a revenue collection system that you need, then you’re going to have officers placed there, which means you’re going to have some kind of district collector office or sub-district office, which is also going to do a whole bunch of other things, and that’s how state capacity gets developed. That’s the general story there. Why is that not potentially the story in the area that you’re studying?

BATRA: As I said, to be honest, there could be dozens of mechanisms, and it is very complex dated in 200 years. What somebody in my shoes can do probably is to focus on one or two, and then rule out a couple of others. What I do on state capacity, and I read this Lee paper back in 2021, what I do in this paper is, I test for state capacity using two different metrics and I compile this from the gazettes. One was more straightforward than the other. One was just looking at tax demand per acre at the sub-district level, so I compiled that. I find that it’s balanced between the two.

A, the tax collection is not different. One would assume that spending would also be the same. How would we do that? Colonial gazettes are a treasure trove of a lot of information, and colonial gazettes from the 1900s, 1901, 1905 at the district level, they contain data on the list of schools. With locality name, and locality could be a village, or a town, or a city. What I did over the summer very painstakingly was manually map around 2,500 localities to modern day villages. I think I have gotten 85% match rate for now. I find balance on that as well.

The probability of receiving a school, by the early 1900s, mind you, there is a 30, 40 period when the treated areas are under the British, and Awadh is under the Nawab. 40 years after the annexation of Awadh, it seems that if there was any difference, there was catch up. Catch up was completed because, statistically, there is no difference between the probability of receiving the school within the bandwidth. Again, to be very clear, using the archival records that I have been able to collect. State capacity is definitely a potential factor, but it seems that in my case—

RAJAGOPALAN: That’s not the driving factor?

BATRA: —it’s not a very important factor. Again, one of the things that differentiates this paper from, say, Alexander Lee’s paper or broadly Abhijit Banerjee, Lakshmi Iyer’s paper is also that those papers take a very national view or macro view, which also constraints their ability to implement very robust empirical strategy. I think they admit to it. I’m not saying it myself. The tradeoff here is, this paper’s empirical strategy is better, but of course, the generalizability is lower because you do [crosstalk]

RAJAGOPALAN: Yes, it’s extremely regional.

BATRA: Exactly. It’s very localized. That’s a tradeoff. Probably what Lee finds may still be valid, but at a more nationally aggregated level.

RAJAGOPALAN: It’s not the driving force in this region.

BATRA: Exactly. [crosstalk]

RAJAGOPALAN: Especially in the border areas.

BATRA: It seems so, yes.

RAJAGOPALAN: Can you walk me through the size of these effects? What are we really talking about here? What is the big difference between the birth accident of being born on this side of the Awadh?

BATRA: There are three variables that you initially mentioned. One is asset ownership, which is basically an aggregate index that I create. Then one is average years of schooling. Then there is non-farm employment for hundred weeks. There are three very different indicators indicating different things. Household wealth is more asset, human capital formation is years of schooling. You can think of non-farm employment as an indicator of structural transformation for an area and moving away from agriculture to industry. In schooling and asset ownership, the average treatment effect is 6% to 7% of the control. 200 years later, there’s still a 6% to 7% effect. Despite all the redistribution policies and affirmative action, et cetera. On non-farm employment, it’s 50% of the control.

BATRA: That’s huge, right? Just to give you perspective and I’ll have to refer—

RAJAGOPALAN: It’s basically they hasten the structural transformation—

BATRA: It seems so.

RAJAGOPALAN: —by setting something in place 200 years ago.

BATRA: Exactly. Again, there could be so many things happening, so many things happening. Then when we come down to scheduled castes, the magnitudes actually go up further in terms of percentages. These effects are sizable, even for a normal, I would say, treatment. To think of it as something that happened 200 years back, I think it’s fairly robust in the sense that it’s something you can’t ignore.

RAJAGOPALAN: No. Absolutely not. If today we had a policy intervention which gave you these effects, you’d be like, "Yes, do it."

The Ultimatum Splitting Game

RAJAGOPALAN: Anything it takes to get these things done. These effects are pretty large. It’s very difficult to think of modern-day policy implications based on this historical work. Is there any modern-day policy implication that you can think of other than the fact that the state should really have very sensible land titling, sensible understanding of rights, and have a systematization in some sense of what’s happening in different sectors of the economy?

BATRA: I’ll take a different route to this answer, because there’s some results I haven’t discussed with you. I think a policy implication, if any, might just come from there because I think we are past the point of land redistribution policies, programs, at least in India at this point, honestly speaking. There is an important lesson to be learned, which is the point. In the survey, we also caught some other things. I told so far the story is a treated area. Scheduled caste households are more aspirational, they’re more mobile, intergenerational. What does that do to their relationships with the upper caste households?

One of the things that I went to the field with was I thought that guys in the control areas would be fed up of the upper caste household. They’ll be like, "You spoiled our lives," blah, blah, blah. "We don’t like you anymore." I played an ultimatum game with them. What we did was we gave them unfair offers. We randomized the caste of the proposal, so a list was read out to them—

RAJAGOPALAN: What’s the offer for?

BATRA: It’s like saying that we gave them ₹40, and they send you this offer. We’re splitting that money.

RAJAGOPALAN: It’s the ultimatum splitting game.

BATRA: Exactly. Ultimatum game. I can share the list with you. There was a Rajput, Brahmins, there was a Yadav, so there were OBCs SCs, upper caste, all three, and for placebos, we add some Sikh names. To the contrary, what we found was that actually in the control area, the level of cooperation or acceptance of unfair offers is higher amongst the lower caste households. In a way there is more social capital or trust or cooperation, whatever you may want to call it in the control area.

RAJAGOPALAN: That’s one way of looking at it. The other way of looking at it is that the aspirations have been stifled generationally or intergenerationally.

BATRA: I’m coming to that. Exactly. That’s exactly the point. The other thing we found was there is less cooperation in treated areas between the upper caste and scheduled caste on things like sharing water from the same well. What explains that, Shruti, there is a model by Acemoglu and Robinson and what you just said is perfect: that when you are in a position of power, such that your sphere of influence is so big, you adopt different strategies to block changes to things like social norms. It seems like in the control areas because you did not redistribute too much of land, the powerful Bahubali as you call them in UP, most of them come from that area.

If you heard of Raja Bhaiya, Brij Bhushan, Sharan Singh these are all Bahubali or powerful muscular leaders. Not in a good way, of course. They all come from that area. Anecdotally, that makes sense. Now, if you put it empirically and then theoretically the way this works is because they have so much to begin with, they have all the incentives to be strategic about their behavior towards the lower caste households. On the other end, the treated areas, you anyway shrunk the land size, you’ve shrunk their power they don’t have much to save. They become more spiteful towards the lower caste households because the socioeconomic gap reduces.

The policy implication that I took was that at the end, are we trading growth for social harmony in a society like Uttar Pradesh, which is so socially split or fragmented along caste line that are the welfare effects positive, negative, or zero, we don’t know. Any policy that seeks to take something away from the elite and redistribute it, what role do the elites play in that policy formulation is probably something that this paper throws up as a question at the end. Because you have now generated two different sets of incentives for the upper caste households to behave with the lower caste households.

Where the lower caste households need to be doing much better in the treated area socioeconomically, but it’s not as cordial. In the control area, there is cordiality, but it’s a very perverse cordiality. Because that cordiality, as you rightly said, is allowing the upper caste to stifle the lower caste. This ties into another paper by Acemoglu, Reed, and Robinson on Sierra Leone. Where they found that paramount chiefs who were the power centers in Sierra Leone, they use higher social cooperation and capital to make sure that they retain informal control over people. In India, Siwan Anderson and coauthors find in Maharashtra that the Maratha landlords in democratic societies, the old power centers find new ways of retaining informal control.

In the Maharashtra, the Maratha landlords have resorted to something called Clientelistic vote trading. They want to retain that sense of power. What they do in order to achieve that, they will not indulge in violence. They will not arm twist you. They’ll just strike a deal with you, you as voter. In Sierra Leone, they will indulge in higher social cooperation. We have something similar has happening in the control areas here. I’ll just end by saying that at the end, I also see that if you look at political candidate selection, upper caste households are more likely to get nominated in the control than the treated areas probably because of their influence in these areas.

RAJAGOPALAN: This is really fascinating, and I understand that these are not like exactly policy implications, but they’re more like social implications or just a better understanding of what is going on in these places other than just the measurement of how much land is owned or how many people own a radio or a TV and so on. This is really fascinating, and I really enjoyed reading the paper. Thank you so much for doing this. This was a lot of fun.

BATRA: Thank You, Shruti. I really appreciate it, and it’s been fun to speak one’s mind out at times. The right answers come when the right questions are asked. Thank you for asking the wonderful questions.

RAJAGOPALAN: No, thrilled to have you here.

Image credit: Environment & Society