(English captions & Hindi subtitles available)
About the Seminar:
How do politicians manipulate the allocation of public resources? Prof. Jensenius argues that politicians' choices are influenced by the nature of the party that brings them to power. Politicians who belong to mass parties that are closely tied to social groups face strong pressures to allocate resources to their electoral strongholds. Politicians from catch-all parties are not as constrained and tend to allocate more resources to competitive areas. Prof. Jensenius provides evidence for this using an original polling-station level dataset that manually links voting patterns in the 2009 Indian general elections to the characteristics of 224,000 census villages and to project allocations 2009-14 under India's discretionary constituency development scheme (MPLADS). The motivations of the politicians for their distributional choices are explored through a natural experiment created by the delimitation (redistricting) of electoral boundaries in 2008. The findings contribute to her understanding of how public resources are distributed by politicians.
About the Speaker:
Francesca R. Jensenius is a Professor of Political Science at the University of Oslo and Research Professor at the Norwegian Institute of International Affairs. She specializes in comparative politics, comparative political economy, and research methods with a regional focus on South Asia. She is the author of Social Justice Through Inclusion: The Consequences of Electoral Quotas in India (OUP 2017).
Hello and welcome to the fourth event in CASI's Spring Seminar Series. My name is Nafis Hasan. I'm a postdoctoral research scholar at CASI and along with my colleagues, portrait this series. The seminar series; the Spring, as many of you would know, comprises talks of a diverse range of scholars from across the world on variety of critical and contemporary topics. Our speakers come from a wide variety of disciplines, including political science, anthropology, economics, architecture, and design. We typically have these talks on Thursdays at noon Eastern Time and more information about them is on the CASI's website.
So before I introduce today's speaker, just to put in a plug for next Thursday, that's Feb 17th, we have Professor Christophe Jaffrelot from the King's India Institute, who will talk about his recent work. Please register for that event on the CASI's website. So without further ado, I'm delighted to welcome Professor Francesca Jensenius. She's a professor of political science at the University of Oslo and research professor at the Norwegian Institute of International Affairs.
She specializes in comparative politics, comparative political economy and research methods with a regional focus on South Asia. She's the author of Social Justice through Inclusion: The Consequences of Electoral Quotas in India which was published by Oxford University Press in 2017. Our talk today is entitled Privileging One's Own Voting Patterns and Politicized Spending in India, in which she asked the question, how do politicians manipulate the allocation of public resources.
She provides evidence for her arguments from an original polling station level data set and her findings contribute to our understanding of how public resources are distributed by politicians. Today's talk is in partnership with the South Asia Center & Penn Comparative Politics Workshop. And we thank them all for their support. So before I turn it to our speaker, just to let you know, if you have any questions at the end, please use the chat box to send them directly to me in a free session, and I will call on you to pose your questions to our presenter.
Please keep your questions brief and to the point so we can get to as many as possible. And apologies in advance if I can't get to everyone. Also, please use the chat box only for questions. Finally, please be mindful about muting your mics throughout the duration of the event. And also, please remember that you cannot record this presentation without prior permission from the presenter. Once again, thank you for your interest and for being here today. With that, I'm going to turn it to Professor Jensenius to take us away.
Francesca R. Jensenius:
Great. Thanks so much, Nafis. And thanks so much for inviting me and for having me. I'm going to share my screen. Let's see whether you can see this. You see it, great. As I just told Tariq, I'm very sad not to be there in person with many of you but I do see many names here that I know and I'm happy to meet you online and delighted about the opportunity to share some my research with you today. So the project that I'm presenting on today is co-authored with Pradeep Chhibber at UC Berkeley. This has been a project we've been working on for a while.
Inspiration really comes from an incident during fieldwork about 10 years ago, where I was traveling in UP with an Indian politician who, as part of this gathering in a village talking to villagers made it really clear that if you vote for me, I'll make sure you get a hand pump from my MPLADS funds. MPLADS funds are a discretionary development funds that politicians can use on small projects in their constituencies. And the way he put it, he made it really contingent.
In fact, he asked if someone had a spade and he asked them to go and dig a hole in the ground where they wanted the hand pump to be. And this was two years before the election. So it was definitely a little preemptive little early to dig a hole in the ground. But he made it really clear that if you vote for me, I'll make sure I'll notice and I'll give something back to you. And I thought that was fascinating, not because I was so surprised that a politician would try to use any kind of funding for electoral gain, but to the extent it was contingent.
I wondered whether he actually would go back and dig that hole and ensure that they had that hand pump later on. And how and why politicians manipulate funds, that is of course, something that is a pretty enormous literature across both economics and political science. But trying to summarize it briefly, I think there are three main arguments about it that we're engaging with at least. One is the idea that politicians will tend to manipulate funding by giving goods or money or resources they have access to where support is high.
And of course, you have the basic argument of the core-voter model of Cox and McCubbins which go to a very classic quid pro quo understanding of clientelism. But then in recent years, there's been quite a bit of work suggesting that you also give to core areas or areas we have a lot of support in order to ensure turnout. You want to make sure people come out to vote.
And also, arguments by Stokes et al, for example and Hicken et al suggests that you might also channel to core areas, because you're channeling through your brokers. And those brokers either you're trying to buy brokers who then give stuff to their networks or you give to brokers who you think will support you without really knowing what they're doing with the money. So they're different arguments related to how it's quite beneficial to give to core areas and there's a lot of empirical support supporting that.
On the other hand, you have completely contradictory models, suggesting that it makes sense for politicians to try to sway voters to get new voters by giving money to swing voter areas. And there's also evidence that you will try to do that more in electoral business cycles that you will try to sway new voters especially before the next election. And then you also have a lot of recent work on India, which is fascinating, suggesting that there's a lot of giving that happens in a non-contingent way.
It's not necessarily quid pro quo. A lot of work challenges whether politicians really can monitor what's going on and their argument's related to maintaining big-man status, building long-term reputations, giving non-contingent constituency service, et cetera. And what I find fascinating about that literature is also that the political scientists there have a tendency to think of that as an alternative form of electoral strategizing, while the social anthropologists think of it more as a longer term, maximizing something else; your network, your reputation.
But of course, there are many types of goods one can give out. We shouldn't necessarily expect the same behavior for all of it. And so what I'll be talking about is first to tell you a bit about the data and context related to MPLADS spending in India and village-level voting patterns, which is the data we have. Then I'll show you the association between voting patterns and MPLADS spending, first, overall, on the village level, and then using a natural experiment to try to probe a bit more what's going on.
And then finally, the main argument we have is really also about how politicians strategize differently, and how we find that politicians from mass parties or parties embedded in social organizations or ethnic networks tend to channel more to core areas than politicians from catch-all parties that are less embedded in existing networks. They are freer, we argue to give to swing areas. So what we're trying to think of is how politicians may behave differently even within the same context and with the same resource.
Now, let me bring us back to India, and to 2009. I think a lot of us have gotten used to seeing this map very orange since the BJP has won many more seats. But in the Indian general elections, which is what we're focusing on here, it was actually much more of a fragmented party system with many parties winning elections. And what I'm showing you here is the borders for PCs, where we have 543 MPs selected.
Now each of these MPs have access to MPLADS funds. At the time we were looking at, that was about ₹50 million a year, so about $800,000 a year. The beautiful thing about the MPLADS as a thing to look at for thinking about discretionary spending is that MPLADS are more than anything, a resource that's really visible, because politicians can hand it out to an area. In fact, most of the time, when they give a little MPLADS project, they put a little bronze plaque saying the name of the MP. It's a really visible good.
They also have almost perfect discretion over it. They can specify exactly where they wanted to end up and that's listed and they list that when they allocate the funds. It has to be spent on some development thing. It has to be something that stays there. They can't just give cash, so they can't spend it on anything. But they have a lot of freedom over anything that's development-related in their area, they can allocate to some area.
We know from previous studies that not all politicians use all their MPLADS. Keefer and Khemani showed that MPLADS spending is less in party stronghold. So if politicians feel safe, they spend less of their money overall. We know from Bohlken that there's more spending in areas where MLAs come from the same party as the MP. But we're interested in how strategic is the targeting within constituencies. So what we did was in 2014, we scraped all the MPLADS project that we could get access to, for the period 2009 to 2014.
We covered 11 states that were the 11 states for which we were able to access all the data sources, we needed to put together our dataset. And the result was about 91,000 projects, which we then manually linked up to census village identifier so that we could geo locate it and link it to other data sources. And it has to be manually linked because literally where it is, is written as a sentence in some language. So it's really hard to automate it. So as those of you who engage in this kind of data collection, know that this is the stuff that takes a bit of time.
And so, in 2009, to '14 period, on average, there were 385 MPLADS project in one PC. On average, they spent about ₹92 million which is about then half of the money they were entitled to during this period. And in our data set, about 13% of the villages got at least one MPLADS project. And project include things like electricity, sanitation, roads, community centers. And to give you a sense of it, here's an example of an MPLADS project.
It was a water cooler for a village. We had a fantastic grad student who went and checked up on a bunch of these and actually found, I think all of the ones we sent him to find, which we were quite amazed by. But what he found quite clearly is sure they're for a village, but often they're in someone's courtyard, or they're often fairly personal. So it's often that they don't seem so much a club good or village good, but more a personal good. So water cooler was an example.
Another example is a small stretch of road. And I'd like this one... This was one of my favorites. One solar light near the house of Jayveer Singh, who I believe was the pradhan in the village. So it was very clear often that this was used to give to activist brokers connections in the villages, often specified in this way to make it really clear that I'm giving back to you for the service you gave to me.
It's also clear in the data that politicians specialize in types of projects, which was also interesting that you see that in this PC, the person gave a lot of road projects, while this other person in Andhra Pradesh gave a lot of electricity-related projects. This one is Andhra. Andhra was very, very often community centers or some contribution to community centers. So, is this allocation of these funds non-contingent?
On the books, it's not supposed to be. On the books, it's supposed to be that politicians can go and have their ear to the ground, understand the development needs of people. And then go off and give a project that's needed. So in order to look at whether something is non-contingent, we then moved to looking at village-level voting patterns. And what we put together is village-level voting data for the 2009 general elections. So that is using the so-called Form 20 data, the polling-booth level data for those 11 states that we were looking at.
For 2009, there were about 262,491 polling stations in these 11 states with a range from 300 to 1400 eligible voters. On average about 800 voters per polling station. And then in order to be able to link that up to our data, we collapsed it to the village town level. So our entire data set is at the village town level. We did that by manually again linking polling-booths level data to census identifiers using the electoral rules, also a cumbersome task. By doing it manually, we have a very high level of accuracy though.
About 38% of the census localities that we looked have exactly one polling station. That means that politician has access to online information about exactly how one village voted. In other cases, in 40% of the data, one village shares a polling station with other villages so that for example, two or three villages that vote in one polling station. So there again, it's pretty clear that a certain area voted in a certain way.
And then about 4% of the data has more than three polling stations in one unit. We excluded those because we think it's really hard to know then whether you're really targeting that area or not. And that's usually big towns. In our final data set, we're collapsing all this to the village level. We have about 225,000 villages or towns across 232 PCs and whether or not they got an MPLADS project. I hope that was clear enough.
How does this look? First of all, village-level vote share for the MPs in our data set in 2009. Average voter for an MP was 42.5% but you see a lot of spread. Some villages got a lot of votes, other villages lesser. Similarly, the average margin of victory for an MP in a village was about 4.1%. Margin of victory here is whether they won so if it's more than zero, it's whether they won the election in that village, with a certain percentage. If it's below zero, it's the difference between the MP and whoever got the highest vote share in the village. This is slightly tweaked margin of victory.
So what should we expect? Here's three outcome variables we looked at; whether or not a village gets any project at all, in the entire period 2009 to '14, whether they get... How many funds they get allocated, and whether they had a completed project during that period. And I guess, if they give more to areas that support them in high numbers, we should see more spending on the right side of each of these quadrants. If they give more to areas where they can sway more voters, we should expect more of a curvilinear relationship. And so what we see is this.
We see a curvilinear relationship and definitely more spending in areas that will support them in high numbers as well. So, there's definitely no association between... There is definitely association between voting patterns and spending, suggesting that politicians either look at the Form 20 data. Or that activists in those areas, look at it and inform politicians of how the village was doing. So we can see that areas that supported MPs in low numbers got very little money and then there's more.
The pattern is to some extent driven by the fact that the central areas of the graph, the blobs being larger suggest more observations, they're more places, they're also bigger places in general. More fragmented areas tend to be larger. If you control for population, some of the curvilinearity goes away. You see a more linear relationship. There's still some curvilinearity left, so I'm not going to walk you through this regression table. But there is a regression table, where we show that when you control for population and look at within PC patterns by putting PC fixed effects, you see still some curvilinearity but a stronger linear relationship between voting patterns and spending.
So is this then evidence of short-term electoral strategizing? Or is it evidence for example of politicians simply giving to their brokers, and the brokers tend to be more present in those areas? Well, the 2008 delimitation offers a natural experiment that we try to leverage. So, in the 2008 delimitation, here you see a map. This is Kannod Constituency in UP. I believe the black boundary is the old constituency, the gray boundary is the new one. When the delimitation came into effect in 2008, some areas were moved into the constituency, some areas were moved out, suggesting that they lost their electoral importance.
The neat thing is that that happened in the middle of the term of the politician in power at the time. So when you think of UP particularly, the first meeting with associate members were held in March 2005. Draft delimitation was announced in July 2006. And the final delimitation approved in December 2006. So sometime between 2005 and 2006, politicians were sure about which villages would end up in their constituency next time around and which one would end up somewhere else.
So villages moved out of the PC became electorally unimportant from 2006. So then it's nice to see whether they keep spending there, because they still had brokers who had supported them in those areas. And so what we do here is a border design. First of all, we define what is a stayer and what is a mover. Well, we restrict to PCs with a 75% overlap after delimitation and the same reservation status to call it the same constituency. And whoever doesn't end up within that, we consider moving. That reduces it to 24 PCs.
And then we use matching to link up villages that are right around the border, because it's not random if you're a stayer or mover since urban areas have a larger tendency of staying in a constituency. So we match up villages that were stayers and movers. And so we end up having 9000 matched neighboring villages. What we see is that... You can see, first of all, in 2004, there was a little bit of money spent, but not much. They haven't really gotten started.
Then 2005 to 2006, the movers got less money and 2007 to 2009, movers got no money. In fact, I'm thinking the little money that's there may be coding error, because the overlaid maps are not perfect. What we see is a very, very clear drop in money spent suggesting that politicians were really looking to elections for how they spent their MPLADS money. If the area was moved out, it was not electorally strategic. They don't spend money there. So we believe that provides some further evidence that the pattern we're seeing is indeed evidence of politicians strategically using their MPLADS funds.
So, do all politicians behave in the same way? If you look at this graph, I guess one feels that there's two different patterns going on here; one more curvilinear, and one more linear. And our intuition here comes from interviews with politicians, that politicians we talked to really talked about two different types of parties. And first of all, people were talking about how you have to give to activists. You give to activists and brokers, not to voters, because that's what they expect. They expect something back.
Then we heard things like the BJP being a part of cadres and INC a party of voters. You have to give to the brokers that work for the party. One has to be seen as responsive to the RSS. What we heard again and again, is there's a difference in how much the activists in parties can control the parties. And so what we tried to code up here is two types of parties that we would expect to behave in different ways. And we call the first embedded parties or mass parties, parties that are embedded in social organizations or ethnic networks.
The idea there is that there's a clear expectation from people that you should privilege your own. And the activists are able to hold you to account. So there's combination of an expectation and an ability to hold you accountable there. While there's a lot of parties that don't have such a clear expectation, and much less of an ability to hold people to account, there's always a desire of people getting stuff back. But we believe that a lot of parties are more casual, they rely more on primarily personal party networks. And they have a greater freedom, although not perfect freedom to focus on swing voters if they so desire.
As we tried to code this up, and we'd love to hear your thoughts on it, here we build on great work by Tariq and others on how parties can be coded. Now, in Tariq's work, they focus on ethnic parties, and particularly on how voters link up to party. Well, we are interested in something related but slightly different. It's about how the party brokers and activists relate to politicians and how they're able to hold them accountable. And we're quite clear on our intuition that BJP and SP and also BSP and RJD are really, really tied to clear networks that hold them accountable and expect them to give back to their own.
Akali Dal and JD(U), we always go a bit back and forth. In the coding, all our results are robust, keeping them in or out of the category embedded party or mass party. We also have another measure which we call party embeddedness, using the NES 2009, where for each party state, so for each state, we look at each party and the share of a party's voters who are members of the party and also the cost of religious organization. This is normalized by state because of how membership patterns differs dramatically across states.
And those two measures; more intuitive one at the national level and the party states level one, they're correlated at point three, which is pretty strong correlation. It seems to us that they're picking up something on the same thing. So here's the difference in the behavior of these two parties types. What we see is exactly what we should expect that the more embedded parties, the mass parties have a much more linear trend in the data where they give more to core areas, while the casual parties are much more likely to spend in the more competitive areas.
So to wrap up, there are a bunch of recent studies on India that finds some forms of spending to be non-contingent. We don't find that at all in the case of MPLADS. MPLADS allocation seem clearly driven by electoral concerns. It's strongly correlated with voting patterns. And we find evidence of both retrospective and prospective strategizing. I guess our main takeaway is really that politicians have similar interests, and they all want to re-election, but they have different strategies.
And to me, the theoretical interesting thing is also to think further what other things differ from politicians that would make them strategize differently. I think there are a number of things that make politicians strategize differently. Our field work pointed us in the direction of the difference in party type and how embedded parties are. And the fact that parties more embedded in social organizations or ethnic networks would put greater emphasis on core areas. But I believe there might be other variables that could also explain differences.
What we think is that these findings and also thinking further about how politicians behaviors and strategies might differ, offers explanations for some of the contradictory findings in this distributed politics literature. And so I would love to hear your thoughts, and think with you about how we can bring this further.
Thank you so much. Thank you so much, Francesca. That is a great description of the way in which we can understand these differences. I'm inviting people to ask their questions, and you can type it in the chat box, and then I can direct you to asking a question. So, while we're waiting for people to formulate their questions, Tariq would like to start off. Tariq, would you like to ask something?
Yeah. Sure, I can get us kicked off. Francesca, first of all, thanks so much for sharing this rich project. And just yeah, as always with your work, the amount of work that's gone into it at different kinds is just staggering. And I don't even want to think of how many sleepless nights you guys must have had doing all the manual linking of data for both the projects and the polling-booth data. Just a couple of quick clarifying questions and then the most substantive one.
One on these was, how did you calculate margin of victory for a village that shared a polling station? I just was curious about that. And were there areas where there were multiple polling stations for a single unit? Was there ever the inverse in which case I'm guessing you just averaged but I'm just curious for those two types? I guess a couple of just descriptive points that were interesting and I just wondered what you guys thought having worked on this.
First of all, the overall pattern of underspending in MPLADS. I think you said 50% spending. We see this again, and again, with all kinds of budget allocations in India, this patterns of underspending. And just I think I've come to a point where I'm trying to think, the distributive politics literature that I've also written in and you've written in, there's so much about what is spent. But how do we make sense of the fact that so much is not spent? We don't really, I think, have great theories for that. And do we have to go elsewhere or is there something within the electoral logic of all of this that accounts for understanding?
The second question I had was just on whether you think what you guys are finding is a test of contingency, particularly in these patterns here. I guess some of it may depend on whether you're identifying strong areas that are core. If there's a lot of correlation between areas that are large margins of victory in 2009 and that did in 2004 in general, whether what you're seeing is targeting of core areas by particular parties? That's a different distinction that I'm sure you'll get other questions on. But whether what you're really finding evidence of is that there's definitely electoral strategizing, and electoral targeting of core regions.
But I think that's a little bit... Or even of swing regions by other parties. But in either case, whether that's evidence of contingency. So if I'm a politician, I might strategically say I'm going to target a core area or a swing area. But a lot of that depends on how stable these categories are. It may not necessarily be evidence of contingency as much as evidence of targeting. I don't know if that makes sense. To me, those are slightly different things.
Francesca R. Jensenius:
Okay, let me do them in order. First of all, on the margin of victory of when they're more villages, we're doing a very simple calculation of if there are multiple villages voting in one polling station, we divide the polling station data into the two or three villages, weighted by population size and then calculated based on that. But they'll have the same margin of victory, because that's the proportion.
And then if there are cases of villages sharing a polling station of one village having one polling station, or one village having multiple polling stations. Large cities have a bunch of polling stations. That's why the 4% of the data has more than three polling stations in one census unit we kicked out, because that starts becoming a large place where really it becomes noise, whether they're targeting there or not. Did that makes it clear there?
Francesca R. Jensenius:
So if there are multiple polling stations in the village, if there were two or three polling stations, we just added up the vote shares. So it's just a simple calculation there. On the underspending, I think that's a great question, how do we understand the underspending. I mean, from talking to politicians about it, it's really just a matter... Some just say, "This is such little money. I have access to much bigger resources. It's not worth my time."
In order to allocate an MPLADS project, you don't have to do a ton but you do have to fill in a form. And you need to get the right form and fill in the name and put in the thing and hand it to someone and get it to the bureaucracy. It's not a huge cost, but there is some cost involved and you see some of them have 2000 projects. It's actually quite... Even doing a couple a day is work, right? So I guess they could spend bigger sums. They often spend a lot like ₹30,000 or like small sums. They could have spent more. I don't actually know why they don't spend more. I'm not sure it would have been approved or not. I'm not sure.
Keefer and Khemani had an argument about the understanding that when they feel safe, they just can't bother to go there. From my experience in the field, I would say that they're just very constrained for time. They have hundreds of people wanting to talk to them every day and there's a lot of villages out there and there's a lot of stuff, so. But theorizing that better, I... We haven't done that. Someone should do that.
I totally agree with you that what we're picking up is definite electoral strategizing. I'm not sure it's contingency. Yes, I use the contingency in response to the recent literature like John's book on non-contingency. In the sense, it's saying... Well, fair enough, I don't think we're showing evidence of contingency. I think we're showing evidence of clear electoral strategizing, and a clear eye to the next election. And I think starting out, I wasn't sure. I thought it was possible.
Like, given for example the work on Maholi and Bombay, and how people just go out to create a good atmosphere and build a reputation, I wasn't sure whether people would always strategize the election, but it seems in this money they do. They don't necessarily do it in other money. I mean, I think there's plenty of work suggesting they don't do it in everything. But in this particularly high visibility type of spending, it seems that they are looking to the next election.
Thanks. You have a bunch of questions so thanks a lot for that. Yeah.
Francesca R. Jensenius:
Yeah, [inaudible 00:33:07].
Thank you. There are a couple of questions that have come in, so I'm going to ask people to unmute their mics and ask. So, can Anurag go ahead? Anurag, can you unmute your mic and ask your question? All right. I think, Anorak, you're not able to unmute your mic, it seems. Alan, can I ask you to just unmute everyone so that they can ask their questions as I call on them. All right. Let's try. Let's go to Tania. Tania are you able to unmute and ask your question?
Okay. Please go ahead.
Hi. I basically I'm very curious to understand how incumbency plays a role here because if I understood correctly, this analysis is taking place for a specific time period. So A, could we be seeing incumbency playing a role in these results? And B, if not, what is your sense of how this would look if we separated by incumbents versus non-incumbents?
Francesca R. Jensenius:
That's a great question. It's been a while since I did that analysis but I have done it. From what I remember, I'm pretty sure that new politicians spent less in the beginning. It's almost like they hadn't understood how to do it or learned it but then they pick up and become very similar. So I didn't see any systematic huge patterns on incumbency as far as I remember.
I definitely ran it and I haven't talked about it here, which is just I didn't find it that interesting. But I do believe that new entrants were less and less, and they were, sort of, didn't know the game entirely. And we also know that people who are very safe in their seat tend to spend somewhat less, that's the key from commodity finding.
Okay. Thank you.
Francesca R. Jensenius:
Can I ask Aditya Dasgupta to ask his question.
Yep, sure. Hi, Francesca. Really cool paper and data work. I was wondering, do politicians have access to information about their village level support and voting shares? And if they do, how do they get that kind of fine grained electoral information? And does that potentially vary across political parties with different levels of organizational capacity?
Francesca R. Jensenius:
Great question. First of all, thanks for being here. Good to online see you. So the Form 20 data is online and it's very easily accessible and has been for a long time. So when we got this data, we just downloaded it and scripted it. There's still a lot of work putting it together as a whole dataset, but getting a single constituency is actually really simple. In 2009, 2010, when I talked with parties, they had access to Form 20 data and talked about it. I learned about it from congress people.
There's probably a difference in how actively they use it. BJP, I believe we're much ahead of the game with other parties in actually utilizing data. But it's definitely accessible and my guess is that if not the politicians themselves sitting looking at stuff, someone else would be doing so. And so my guess is actually that party workers who try to mobilize for a village would go and check up on it, or ask about it, or figure it out in some way. So I believe these data are quite easily accessible and have been for a long time.
Yeah, thank you for that. And I've seen in my own work with the district and the subdistrict level bureaucracy that a lot of this data is consolidated at a particular place, and then that becomes very easy to access them. So thanks for that. Rashi, would you like to ask your question?
Yeah. Thank you so much for the talk. It was really insightful and fascinating. I have two questions. So one, is that I was just curious about the theory of the mass party versus catch-all party in the spending differences. I think you just alluded to this briefly, but I just wanted to understand like is a theory that brokers are holding? Like, what is the embedded network theory, is that brokers are holding the party accountable, is it citizens are holding the party accountable? And so I was just curious about that.
And then the second question was should we be understanding MPLADS funds differently from how other types of funds are spent because they're so discretionary. And so is there something about the characteristics of the MPLADS funds that are driving this kind of spending? And should we expect to see this in other types of funding or no? Yeah. Thank you.
Francesca R. Jensenius:
That's great. Thank you. So are MPLADS is different? I think MPLADS are different in the sense that they're extremely high visibility. And in the political economy literature, one definitely talks about that is the difference between high visibility and low visibility goods. This is very high visibility, if you can actively allocate something, you're the one in charge of it, there's no one else to blame for process and you put a plaque with a new name on it. It's like, that's a high visibility good.
And so to some extent, I don't think we should expect this level of strategizing to occur in all of the forms of spending. But for the MPLADS one, I think it is an easy case for electoral strategizing. On the other hand, I mean, on paper, it's supposed to be a development fund meant to fill gaps in the state and they're definitely not using it for that. And we also see that more funds are spent in places with higher literacy and more development, so it's definitely not filling the intended on paper purpose.
And as for the theorizing the parties, I mean, we're basing this on interviews, and it's not extremely precise. But what people kept telling us is that you have to please the party workers, and some people feel more beholden to the party workers than others. And it was really about not voters holding parties accountable. But brokers being able to hold the person accountable. And particularly the ones embedded in an organization, they talked about how...
Well, you had the RSS people, for example. You have RSS members at every level of the organization. So if an MP doesn't comply and behave in the way they want to they can go and complain to someone above them, or their peers, or they can mess with their chances of becoming a candidate in the next election. So you can think of it as an electoral strategizing including also the step of becoming a candidate, that if you mess with the party workers, they might not support you for becoming a candidate in the next round.
And so there's something about the ability of the party workers to... The expectation they hold of the politicians' behavior and the ability to hold politicians accountable. That's the mechanism we're trying to get out here. And that's why... So we go back and forth between calling it mass and catch-all parties because that's labels in the party literature and they're similar, but we also sometimes call them embedded, non-embedded. We go back and forth the labels a bit.
Thank you. Shikhar Singh had a question. Shikhar would you like to ask?
[Inaudible 00:40:49] there's two questions. One, does a friendly or a hostile state government or administration constrain the sort of discretionary spending? Presumably, the MP has some interface with the local administration, DM, et cetera, and therefore through that the state government can exercise some constraint? And second, is there any evidence that mass parties in general may be winning with, let's say, larger margins and therefore can lavish more resources on their, let's say, for constituents or workers? As compared to, let's say, catch-all parties that are barely scraping through and therefore are more constrained in moving out from swing areas? Thank you.
Francesca R. Jensenius:
Those are great questions. Thank you. First of all, on the interface with the bureaucracy... The MPLADS rules are quite clear that the bureaucracy is supposed to sanction any valid project within 45 days and then implement within a year. It's not perfectly done, but it's not terrible either. We didn't see any association between... We looked at sanctioning time as one outcome, we didn't see any clear patterns there so we ended up not including it, because it just seems to be pretty similar for everyone.
In terms of the size of the parties. Our mass parties here include BJP with the catch-all include into congress. So both sides of these have ones of large and several smaller parties. There's no major difference in the competitiveness or the margin of victory at the PC level. So there's no association that we have found in the PC level characteristics and their behavior. And so all of the models we run with these parties, where we look at these patterns are within PC patterns. Although, what I'm showing on the screen right now is just the raw data.
Thank you. Tanushree Goyal had a question. Tanushree would you like to ask you a question?
Hi, Francesca. It's really nice to see you.
Francesca R. Jensenius:
Thanks for this really interesting talk. I just have a question about when people in the broader surrounding see party members attract all these resources, I wonder what it does for political participation in those communities. And if you see any spillovers for just the broader community where this is going on. So there's variation, as you said, in parties that are shoveling these resources on their party members. So I was curious to hear about those spillovers. I was curious are women party members... And I don't know if you actually have so much data on that. I was just curious, is this ability to attract these resources equal for men and women, and if women are also able to benefit from this patronage as much as men are able to do? Thanks.
Francesca R. Jensenius:
First of all, good to see it, honestly. And I don't have any data on individual level data on who gets this stuff. All the names that we see in the data, like the example I gave, have all been male names. And also all the people we interviewed were men. I've actually always had a hard time getting access to female politicians, there aren't that many of them and I find them harder to access, although I've tried. So this is very male story although there might be gender dynamics here that we don't have, and we're not picking up.
In terms of attracting the resources. I mean, I assume that when you get an MPLADS project, put in your house and you can say like, "Look, the state is doing something for me." So I'm sure they leverage it in campaigning in the future. At some point in life when I have the energy I will look at the 2014 data which we have collected, and see whether I can see any result from MPLADS in the next election. Blame COVID for not having had time to do that.
Thank you. Kumar had a question. Sorry, Kumar. I think I missed you. You had put a question earlier on. Would you like to ask it?
Yeah. Thank you, Nafis. Hi, thank you for insightful presentation. It was really fascinating. I have few small questions. One, is would you consider Shiv Sena as embedded party or not? Second, I'm sure you might have small sample size on that, but I still wanted to know what are your thoughts on how swing politicians can change their funding? Because I was thinking, maybe if politicians are changing party in between their tenure, that can change how they are spending money and where is going. Yeah. Thank you.
Francesca R. Jensenius:
Thank you. I don't think I fully understood that last point. So what parties were moving between what?
So like, if there are swing politicians who are changing parties in between their tenure, how does that change where they are putting their MPLADS funds?
Francesca R. Jensenius:
Ah. So they're switching candidates? That's a great question. I have spent a lot of time looking at switching candidates, and I haven't thought about doing it in this project. So I will look at that but I don't know. I'm not sure what my expectation would be. I guess, my expectation would be that if you're truly embedded in a party and you care about a party, you don't switch as easily. So if you switch that suggests you're less embedded. I mean, I don't pretend that all of this division into two types of parties is perfect.
In fact, one of the most core voter behaving politicians at all in our data set is [inaudible 00:46:36]. It's fascinating. He spends absolutely all his money on places that support him in really high numbers. He does not spend a penny on places that doesn't support him in high numbers. He has a very clear electoral strategizing in his MPLADS spending. And he's congress, which we categorize as not so embedded.
I think our intuition here is that politicians will have different incentives to strategize in different way. We shouldn't expect them all to behave in the same way, and this is one broad way of categorizing it. But I think if one starts thinking about this further and testing it further, I think one can find other stuff too that correlates with how politicians choose to strategize. But I actually would love to look at the switchers how they behave. On the Shiv Sena I would say that that's definitely an embedded party and mass party. But we don't have that have Maharashtra on our data set.
There's a question from Vijay Prakash. Vijay Prakash, would you like to ask a question?
Thank you, ma'am of this insight. Because I'm from UP so this kind of pattern at ground we not witness but your data it very clearly visible. But one thing I would like to say, for example, in UP the MP fund we not witness in villages at this extent how they is spending. But we really witness when chief minister regions determine a lot of funding kind of thing. For example, when a [inaudible 00:48:24] belonging to Western UP, then more spend in Western UP. This time Yogi did not become chief minister, so more spending come to Purvanchal region Eastern UP kind of thing. So that thing.
But have you accessed the recent data after 2014 when people going to vote not seeing their MP case but to the larger, one leader kind of [inaudible 00:48:51] and big leader case. They are not giving importance to their MP personality and their spending its style, but they are voting for larger [inaudible 00:49:03] parts then how this spending pattern change. Thank you.
Francesca R. Jensenius:
Wow. That's a fascinating question, whether the spending patterns would have changed when you get more focused on the leaders? I don't know what my theoretical expectation is on that. I think I'd have to think about it a bit. But I think it's a really, really interesting thing to potentially explore. So I don't actually have a good answer from you. I totally agree with the observation that when you see leaders from some region, you see more spending in general going to certain regions. And I think we see it in many types of data there. At various levels, you see money being politicized. And so what you're saying completely confirms existing studies.
What I think is fascinating about the MPLADS funding in particular is that it's really so specific that you can really... That it's one person deciding and it's targeted to one area. So it's a really, really clean test of how they're choosing to do it. But also I must say, I do believe that some of it is happens very fast without much thinking, I don't think all of it is super strategized. And we also have politicians who say like, "Oh, I just give to whoever asked because I don't have time to think about it." So although we see patterns, that doesn't mean that every single decision is super strategic either.
Thank you. So there are a couple of questions from the the Trivedi Centre for Political Data, and so I invite them to ask the questions.
Hi, Francesca. Good to see you and great to see that data into a paper, fantastic work. So you're finding is that these are indeed strategic in the way they allocate their MPLADS funds, which is an intuitive finding. But and I don't know, maybe you do that in the paper, which I have not seen but it might be useful to qualify what kind of strategy we're talking about, because you could think of a different way they could strategize. They can strategize as a form of reward. And that seems to be the case here. They seems to be using MPLADS not so much as a way to gain new voters, but as a way to reward their voters. Particularly, when there's a pre-existing bond between those voters and the MPs rather than a strategy of conquering new voters. And once you have data for 2014, and then after 2014, it's going to be really fascinating to see how that eventually evolve as individual candidate strategies have changed quite a bit in elections.
Otherwise, a couple of very quick questions. I was curious whether you considered looking at possible correlation between types of MPLADS expenses and the reelection prospects of rerunning MPs? I mean, does it actually help to... And does it... Is there any correlation or any connection here? I have a similar question as Tania asked before about new MP versus the [inaudible 00:52:58] which you addressed. I was curious also to see whether there were any difference noted between spending patterns across constituency types reserved versus non-reserved. And finally, should we assume... I mean, I understand the way your dataset is structured and you look mostly at village level data. But in the MPLAD data that you have, is there any sense of the shared resources allocated to towns versus villages?
Francesca R. Jensenius:
Well, first of all, good to see you. Great questions. To answer the last one first. Generally more money is given to more easily accessible and more populous areas. So it's sort of the opposite of the intention of filling gaps by the state in villages. But that's not very surprising, I think. Reserved versus not reserved, I haven't checked, although I have worked on reserved seats before in my life. Actually, it's not something I thought about. New versus old, I talked about. I think we show... I don't think one can fully distinguish between the rewarding and they're looking to the next election. I do think politicians think of it as a game where they build reputations and also look to the next election. I mean, for the UP part where we're looking at the delimitation, that is looking to the next election. Well, the rest of it is sort of rewarding. There are no clear electoral business cycles in the data. So there's no big difference in how they spend early and later in their periods, which I actually found surprising. So no clear patterns there on whether they care more about the retrospective or prospective. My sense is that they do both. I think that was it. And oh, yeah. And on the effects in 2014, I have the 2014 data, I just haven't had time to run the analysis. So when I get this paper out and get some time in life, I will give the answer.
Thank you. So we have another question from Sofie. Sofie Heintz, would you like to ask a question?
Yes. Hi. Thank you so much for the really interesting talk. I have a question that's not really related to the core argument but something I thought was really interesting where you talked about the specializing of politicians in providing certain types of development projects which is, to me, a bit counterintuitive. Because it seems that then you don't really listen to what people really want, but just offer what they can get. But I was wondering, so whether you explore it in any way, whether there are also some forms of patronage links to like construction companies. So like influential people that are connected to influential companies and sort of these companies that expect MPLAD money coming in just because they hold a lot of influence in the region.
Francesca R. Jensenius:
Thank you. That's a really interesting question.
Francesca, just a second. So there's another question by Varun which is on the similar line. So I'm just going to ask-
Francesca R. Jensenius:
Yeah. Varun, would ask your question. Hi, Varun, are you there?
Like he's saying he has a bad connection, so do you want to just read his question now?
Yeah, I can read it. So Varun's question is, your paper looks at the allocation of funds towards those areas types of projects et cetera. Does your data set also have a list of contractors who are awarded the funds to implement the projects? If yes, I would assume that there are patterns and who benefits monetarily from the project rather than looking at the development outcomes? Yeah, so thank you.
Francesca R. Jensenius:
I think that's a great point. We don't have a list of the companies that get them, that's allocated later on by the bureaucracy. And I don't have information about it. Most of these projects are really small though, so this is not a big kickback type project. Most of them are really like less than a lakh, building a little wall or doing a tiny thing. So I don't think the MPLADS fund is the biggest kickback opportunity you have around. But it would still be fascinating to see the implementing because we don't have that.
Thank you. So another question from Rachel Brule.
Hi, Francesca. This is such exciting work. I guess, I had a quick question. I really am enjoying just hearing as you're developing the theory. And I wondered for the mass parties or the embedded parties is there a way in which you could use variance in party age? So some of the parties can look at are quite new, others have been around for quite some time to see how these kinds of organizational accountability processes, either grow stronger, or weaker over time, or get more or less effective.
I don't know how much variation you believe you have there. But it would be really, really exciting if... I don't know if there's anything you could say about it now or if that's something that that comes out as you go forward. I really enjoyed it. I thought it was easier to follow the embedded versus non-embedded title of the groups, but that might just be because I'm outside a lot of the literature on parties. But thanks.
Francesca R. Jensenius:
Well, great to see you Rachel. I hope all is well. Thanks a lot for the input on party age. I mean, Congress is our primary example of a not so embedded one and they're sort of old, so I'm not sure how much leverage I have there. Part of me is thinking our life is easier if we just say they're ethnic parties because it's very overlapping. But then we have to say BJP is an ethnic party, which is, of course, something people aren't so enthusiastic about. But we're still definitely in a discussion on what to call it. I mean, our intuition about what we're trying to pick up is quite clear, but we're lacking the labels and exact... We don't have great measures for it and we don't have much better measures for it than what we have right now, so.
Francesca, on that, just because you got some other questions on this, do you have an intuition that the more embedded parties should have more less defections, less party switching at the candidate level? And is that something, because that was the one thing that struck me is that we do see at least some of that with the BJP. I'm just trying to think, theoretically, if the model is where the super embedded party with a stable organizational core and that's why we target core bases. Then a lot of party switching would go against that model, because you're being held accountable to this broken network.
That might be something you can test even just for the Congress, and BJP is the biggest drivers of each category like... I don't know. Just like these observable implications for the concepts that you're thinking of developing if the existing concepts aren't that helpful. I don't know that's getting to what Rachel's talking about but...
Francesca R. Jensenius:
No, that's great. Thank you, Tariq. I mean, to some extent, I think of defection more as a measure of party organizational structure. While I think of the embeddedness as a two step thing of whether people really expect you to privilege your own. It's sort of that's your business model, and whether you're able to hold people to account doing that. And so that's not the same as the strong party organization, I think. But actually, it's easy enough to run the correlation with the factions. But I don't think that's picking up the same thing. But I really appreciate you thinking with me here.
Can I follow up on that, because I do think what if... If you're talking more about what broker behavior looks like in this case, could you look at broker defections? I don't know if you have that data, but to see do brokers stick around longer where they have more of a [crosstalk 01:01:27].
Francesca R. Jensenius:
Tariq, do you and Adam have that data for me by any chance.
I mean, we do for two cities, but we find a lot of defection on both sides. We find pretty comparable levels of defection. But it's self-reported. So but even when we track through party membership, it's about equivalence, about 20% for both.
Francesca R. Jensenius:
Right. Anyway, sorry, I took us off the rails. My bad.
Francesca R. Jensenius:
No, no, this is brilliant. I mean, this is so helpful. Gilles, the data guru has input here.
Well, I mean, it seems to me that there is also an expectation of the role played by MPs and regardless of their party affiliation, which intuitively might also create a bit of uniformity in the way they behave with their MPLADS funds. And defections, it seems to me, as a more function of how candidates are recruited in the first place. And the propensity of parties to recruit outside their organization, which the BJP didn't used to do in the past but now does as a strategy of expansion.
And so in a way to present that in a simplistic manner is that every party tend to recruit their candidate more or less in the same way using the same criteria. And candidates across major parties tends to be also sociologically interchangeable, except that they may belong to more of certain casting party X or Y and so forth. And so, in a way, there's a candidate effect or any legal effect on behavior, and then there's a party effect. What you're trying to look at is, is there an impact of the party on individual behavior, but how do you weigh that with individual drivers?
So, Francesca, before you respond to that, I just keeping being mindful of the time. I just want to end with asking Biswajit to ask his question. He's had his hand raised for a long time. And then after his question we could the rest the audience could go overtime. But anyone who wants to stay around and have a chat could do too that as well. So Biswajit, would you like to ask the final question?
Yeah. Am I audible?
Francesca R. Jensenius:
Thanks, Francesca, for your interesting presentation. And thanks to Tariq also to organizing this CASI Lecture Series and nice to see you all. Basically, I've not any kind of question but I have a query that, Francesca, inputs that when the MP is actually spending their money with their core centers of their assembly sections and which assembly part is not very much dominated by their same party. So if the opposition party MLAs are there, so how they are spending at that that kind of assembly areas. And means so after the four and a half years so they have to win their constituency again. So how they respond to this kind of spending money or projecting some kind of welfare schemes in that area. So how you can manage is and dealing with the oppositional ground with their MLAs and MPs. So this is my queries for the Francesca.
Francesca R. Jensenius:
No that's great. Thank you so much. So it's very clear that there's actually a great paper by Anjali Bohlken showing that they spend more MPLADS funding in areas segments where they have a copartisan, and so they're clearly spending more money in those areas. I mean in any locality there is opposition in the very few places where they have full support, so there's always opposition in any place. In fact looking at these village level data one of the most fascinating thing is how fragmented is even at the village level I think.
But also remember that the MPLADS funds is not the biggest money they're giving out, it's a very visible source but it's not at all the funding that there is distributing. And most places, so 13% of villagers get one project or some a few more but it's not it's not huge money that's being spent. But there's definitely patterns going on between the MP level and the MLA level that, so they're mindful of.
Thank you so much. I'm going to end the formal session right now. Thank, Professor Francesca Jensenius, for this wonderful talk and to the audience for all these wonderful questions. I'll put in a link for next week's talks so if anyone is interested please go onto the CASI website and register for that talk. I'd also invite you now to stay around if you'd like to have a chat, if Francesca has the time to have a chat around these wonderful and interesting issues that have been raised here. Thank you.