Merve Demirel on the Costs of Migrating for Higher Education 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 2026 job market series, where I speak with young scholars entering the academic job market about their latest research on India. 

This is the 2026 job market series, where I speak with young scholars entering the academic job market about their latest research on India. I spoke with Merve Demirel, who is a doctoral candidate in economics at the Institute for International Economic Studies, Stockholm University. We discussed her job market paper titled, “Migration Costs and Higher Education: Experimental Evidence from India.”

We talked about the education opportunities available to senior secondary students in Maharashtra, why these students prefer to study close to home despite recognizing the higher returns available from colleges in Mumbai and Pune, measuring the nonfinancial costs of migration, and why those costs are especially high for women, 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, Merve. Welcome to the show. It’s such a pleasure to have you here.

MERVE DEMIREL: Thank you. It’s such a privilege to be here. I’m very excited.

Students at the Crossroads: Understanding Migration Friction

RAJAGOPALAN: I’m excited too. And your paper, I really enjoyed reading it, because this is not an area I’ve thought about much. In this paper, you ask why students choose to remain close to home for college, even when they believe that studying elsewhere will produce better labor-market returns. What you look at specifically in this context is whether access to a peer or a mentor—in this instance, a migrant mentor—will help them overcome the nonfinancial costs and frictions of moving.

This was fascinating to me because I’m an immigrant myself, and I have moved very far from home [chuckles] for educational reasons. So, it was fascinating for me to actually try and understand what kinds of frictions exist. And some of these frictions are really at the district level, right? We’re not even talking about moving 7,000 miles or something, the kind of move that you and I made maybe.

In this particular study, you look at just over 1,600 final-year secondary school students across 52 schools in various districts in Maharashtra. And you’re looking at them at that precise moment in their lives when they’re making this decision about what to do next after secondary school, right? This is about are they going to college, where are they going to college, what field they’re going to look at. And also, what their expectations are in terms of their life and how it’s going to play out. So, what are their expectations when it comes to earnings, when it comes to employment and things like that? 

And you find what I would think of as an extraordinarily large preference to remain at home. I really was not expecting this, but looking at the more aggregate data in India, it adds up pretty well. The average student would need an increase equal to 129 percent of expected postgraduate earnings to accept studying away from home. That’s a very large premium, or the other way of saying it is, there seems to be a lot of stickiness to remain at home. Or there seems to be some friction between the two to be able to bridge that particular gap. 

And this penalty also increases with distance. So, distance does play a factor. It’s not just about stickiness to being home. It increases from about 39 percent for the students’ own district capital to 93 percent for Pune and 196 percent for Mumbai and 303 percent for Nagpur. These are enormous numbers. I really did not expect this going in. 

And then, after you study this, you run a randomized experiment in which half the students are offered mentoring by someone from the same gender who’s a university student who has migrated to Pune or Mumbai. This is a long-distance, over-the-telephone mentoring. Then, you look at what assignment to mentorship does. 

The especially interesting result in this area is that applications to nearby district capitals rise by about 4.9 percentage points, but moves which are further outside, let’s say to Pune or Mumbai generally, don’t see this big uptick, thanks to mentorship. 

There is something very interesting going on in this particular marketplace. This is the marketplace for education and eventually for post-education employment. It is also the marketplace for migration and migration-related choices. So, it’s a fascinating study that you’ve done, and I can’t wait to dig into it. 

But largely, have I summarized your paper well? Because you’re looking at this at a lot of depth, plus, there’s a randomized experiment. So, am I getting this somewhat right? Feel free to correct me.

DEMIREL: I think you’re explaining the paper beautifully and very intuitively. I think you’ve gotten it. Yes, great. [chuckles]

RAJAGOPALAN: So,I guess, Merve, my first question would be then what was the motivation for you to study this or look at this, and in this particular way? Because I’ve not seen too many studies like this one.

DEMIREL: So, in this project, what I really, really wanted to study was this longstanding puzzle that we see in a lot of different contexts, namely, the puzzle where there are large opportunity gaps across space, but it is accompanied by low migration rates. People seem to be not moving to where opportunities are, to a large extent, and it has really puzzled economists for a really long time.

So, I studied this question, and I studied in a specific migration setting, and that is the question of where to study for college. And I think this is a really interesting and high-stakes setting to study because, essentially, this is a one-time decision that a young adult is going to be making very early on in their lives with very limited information, and they’re weighing up a lot of really consequential costs and benefits together.

On the one hand, if a student is able to migrate, they’re going to be able to access many more colleges, many more opportunities, potentially better colleges, and colleges that fit their personality or fit their ambitions better. And they might be able to set up their adult lives in cities or locations where the jobs are paying higher or the quality of life is higher, and so on and so forth.

So, you have this big, long-term, large investment that the students are able to make, but you also have these immediate costs that they have to go through with it to be able to access those opportunities. And those costs can be really daunting. It can be financial costs of moving from point A to B, or it can be the nonfinancial or nonpecuniary costs, as I call them in my paper, that can be really things like moving away from your family home, or leaving your existing friends behind, and starting to take charge of everything in your daily life by yourself. These can be really daunting or overwhelming for some students. It can be also very exciting for some others who really want to move away from home, but it really depends on the student. 

So, I was really interested in studying this. I think it’s a finding that consistently comes up in the literature, that demand for higher education falls with distance. So, the further away educational opportunities are—especially for higher education, because that’s typically when we start thinking about migrating away from home—the less likely are students to take up that offer.

We see this across a number of different contexts. It’s certainly true in India. I have also some other work with a set of coauthors that really shows that distance is a first-order cost in higher education decisions in Indian students. So, my paper is really trying to understand what is that cost and how should we think about it. Should we think about it as barriers? Or should we think about it as just misinformation? Or should we think about it as just students’ preferences that we should respect, and how do we approach it? So, that’s how I started studying this topic.

The Decision Pressures Facing Prospective College Students

RAJAGOPALAN: This is fascinating. Thank you for the context. But now, we are looking at not the general college student, but a very particular kind of college student, right? So, you’re looking at a couple of districts in Maharashtra. And you’re looking at the college decision for that particular student in grade 12 or senior secondary. For those of us who are unfamiliar with that setting, what does that college decision really look like for that representative higher secondary student there?

DEMIREL: Right. So, for a student that’s in my sample, the decision to study, what to study, and where to study after class 12 is—it’s a really complicated decision. They are really looking at a number of different options. They’re weighing up whether they should go to college, should they do vocational training, what kind of exams they should sit for. They’re really going through a very stressful period in their times. Class 12 is really stressful. It’s end-of-high-school exams, but also a lot of these different centralized exams that they’re sitting through.

And they’re doing it—at least my students are doing it—in a low-information, low-support system setting. The students in my sample, they’re more likely to come from under-resourced schools, at least compared to the average school in their district. They report very low availability of resources that would help them think through college decisions. Most of them don’t have any support system at school, so they’re really navigating a complex choice by themselves.

But on the other hand, they have a large opportunity gradient available to them if they were to start looking at colleges outside of their immediate locality. Of course, that changes for every single student. I like to illustrate the example of a student who is wanting to study engineering and is looking at just in their immediate environment versus their districts and in the whole state. And adjusting for competition—so, how many other engineering students are there that can take up that spot—the available spots for them can go from 70 to 10,000, or something. So, it’s really a crazy number. Just the number of seats that they can access by looking further is really large.

And of course, those better-quality educational institutions are more likely to be at places that are large cities, so places like Mumbai and Pune. Those are the places where opportunities are constructed. So, they’re in that setting. They have potentially large returns from migrating for college, but, overwhelmingly so, they’re not very interested in doing it.

RAJAGOPALAN: Yes, and these are not students in the outskirts of Mumbai or Pune, right? You’re looking at students in Sangli and Ahilyanagar, where it’s not the back of the beyond. They do have some college opportunities and educational opportunities and some potential post-university work opportunities. But at the same time, these are not huge urban agglomerations where you will really get access to this phenomenal peer group or access to the innovation economy or access to the largest labor market in your state or anything like that.

DEMIREL: That’s exactly right. So, actually, thank you for reminding me. It really depends on what you want to study. If you’re instead looking at, for example, vocational training, then these students are already living in places where they can access that type of education. So, it’s really not a choice between not studying at all or not having any access to any type of education, but it’s just accessing better opportunities, or the best possible opportunity that you can get to.

And yes, some of the students in my sample, they’re more in towns, they’re in district capitals. Some of them are in more rural places. There’s a bit of heterogeneity, what’s available to each and every one of them locally. But, yes, this is roughly what they’re weighing up.

RAJAGOPALAN: Yes, and you alluded a little bit in your earlier answer on what makes this decision and this issue of distance so salient, but I wanted to get into that a little bit more. So, what is the central challenge or friction? Do these students just not have information? Do they not perceive the economic returns to studying elsewhere? Or they do have the information, and they can understand the expected returns are going to increase, but as you pointed out, there’s just a strong preference to remain at home for whatever reason? And so, what exactly is going on below the surface of this particular choice?

DEMIREL: Right, so, for the students in my sample, they do actually understand the gradient, at least to some extent, because when I collect data from them about their expectations of the expected salaries if they were to study at different destinations. So, studying at home versus studying at district capital or Pune or Mumbai, they do expect to earn somewhere between 16–30 percent higher just comparing what they expect [at] home versus these destinations.

And this is an expectation, or this is an increase, that for the majority still stands when I account for the higher living cost or the cost of studying at these different destinations, or, to some extent, the risk of them not being able to get a job after. So, to me, it’s definitely the case that the students expect some economic returns to materialize after this investment takes place, but it simply doesn’t seem to be enough to make them want to pursue that.

That could still be because of a number of different things, right? They could still have other constraints that I’m not capturing super well. Or it could be that they simply don’t think they would get into these colleges, as they are more selective and competitive places. 

But none of these—as far as I can tell with the data that I collect—none of these seem to be the final answer. I really find a gap between students’ intentions and the economic returns, and that makes me conclude that there must be something else going on some other costs that are making them behave this way, or that are making them decide to study at home. And that’s the central object of interest that I pursue in my paper.

Measuring Willingness to Migrate

RAJAGOPALAN: Exactly. Yes. Let’s get into that. One of the most interesting things about the paper is actually measuring what you’re talking about. How does one measure a student’s unwillingness to migrate, right? And what does it mean to say when you write in the paper ‘that the average migration penalty is 129 percent of expected earnings’? How do you even think about putting that together, measuring that, calculating that?

DEMIREL: Right, so, what I want to estimate—so, I can measure somehow the objective returns or the subjective expected monetary returns from the student or from household data, and so on. What’s really difficult is to understand what is in this move for the students. What is the cost of separating from home, separating from friends, and so on and so forth? That’s not typically something we can measure from data that’s available. For that, I have to speak to students and get an estimation of what it’s worth to them.

So, again, just to restate, my goal is trying to understand what is the disutility that’s attached to moving away from home to another destination, holding all of the financial aspects of this move constant. That means I need to find a way of introducing experimental variation into what destinations cost and what they mean to students. And that’s why I’m going to introduce this discrete choice experiment, which is step one in my paper.

I’m going to essentially ask students to—I’m going to present the students with a set of colleges. It’s going to be a repeated choice experiment, and for each college, I’m going to experimentally vary the destination and the expected earnings that the students should think about after graduating from that college. And that’s going to give me a dataset where I can calculate the disutility of moving from one destination to the other. That’s going to allow me to price it in terms of the utility that an additional percentage of earnings brings to the student.

So, it’s really a way to put a price on the nonpecuniary utility differences that the student is expecting to get. And to really make it concrete what exactly this choice experiment is, you can think of a student who’s filling my survey in a tablet, and they’re presented with two colleges. These colleges are hypothetical, and each of them has five attributes: It’s the field of college, the location of college, the average salaries after graduation, and what it would cost in terms of living expenses and tuition to attend these colleges.

So, I ask the student, “Look, here are two colleges. Imagine that you’ve gotten the marks to gain admission in both of them. Which one do you prefer?” And the student can choose college one, college two, or they can choose, “Actually, I don’t like either of these. I’m not going to study at any college at all.” And to make it realistic and feasible for the student, I’m going to use information that I have elicited from the student to fill in the characteristics of that college.

The colleges are always going to be fixed in the field of what the student wants to study. So, let’s say it’s engineering. They’re always going to see engineering colleges, and they’re always going to see colleges that are feasible for them.

The cost of living and tuition is going to be fixed and within the budget constraint that the student has previously told me earlier in the survey. So, it’s really an exercise where I’m trying to shut down different feasibility channels in terms of budget constraints, in terms of, “Can I get into it? What’s the risk associated with it?” And all of these things. I’m just trying to capture what is it about living away from home, in nonfinancial terms, but priced in the utility of 1 percent earnings. So, yes, that’s it, what I do. 

Then, the 129 percent estimate comes from basically dividing the disutility of a destination that’s not home by the utility of 1 percent increase in earnings. That is what essentially tells me, “You know, these students on average would be indifferent from studying at home versus studying somewhere else, but only if that somewhere else gives them 130 percent more in expected earnings after graduation.” So, it really doesn’t have anything to do with what they believe or what they think they’re going to get or something. It’s just a way to price the disutility.

Decomposing the Migration Penalty

RAJAGOPALAN: Yes.The way you’re looking at this, I love the setup so much, because it’s really looking at the relevant tradeoff, right? This is not some hypothetical counterfactual that may or may not exist in their life, that they’re just playing a game with. This is a very realistic choice between two options that may almost present themselves in a few months in their life.

But in this particular instance, how do you manage to calculate some of the costs involved in this disutility? So, for instance, are they just worried about the fact that, hey, maybe the costs of food are uncertain, or are the costs of food large? Is the quality of housing uncertain, or is the quality of housing poor? 

What is a good way to think about that? Because that’s really hard to capture. You just know what they are choosing. But how do you decompose that, or what is below this? What is making up this very large migration penalty?

DEMIREL: Right. That’s an excellent question because that is exactly the nature of this object that I’m estimating, right? Conceptually, it’s a very large object. It contains everything that I might think about moving away from home, and those are going to be—by definition, it’s going to be multidimensional. It’s going to be subjective, and it’s going to be a combination of what I think my life is going to look like across a number of different outcomes and how much I care about those different outcomes in different places. So, it’s really a combination. It’s a composite object that captures many different things. 

But my whole design conjectures that it captures the nonpecuniary costs associated with movement, so I do a validation exercise. I go out to students and ask them about their beliefs about very specific things that they expect to happen if they were living and studying at these different destinations.

So, that can be something like me asking you, “Shruti, imagine if you were studying in Pune, how safe do you think you would feel in your daily life just moving around in the city, from 1 to 10?” And I ask you the same question for a number of different destinations and for a number of different outcomes that I think capture some of the nonpecuniary outcomes. So, that can be emotional outcomes, social outcomes, or just practical aspects of moving.

I do this, for example: “How much do you think your parents will approve? What will those who stay behind in your hometown think of you? How lonely do you think you’ll be? How many friends will you have? How much will you be able to keep in touch with your friends and family back home?” I also ask about just overall thinking about all the practical things associated with a move. “How difficult do you think it’s going to be to move to a different destination?” And that one’s a bit of a special item because it’s right away asking you about a relational question. So, it’s defined for home as not difficult at all, because home is not a move. 

So, I measure these subjective expectations, as I said, for a number of different things. Then, when I first just look at what these beliefs, or subjective expectations, look like for a number of different destinations, I find that, overall, students expect all of these to decrease the further away they are from home.

So, a student simply thinks that life away from home, studying at these different destinations, will look worse on average compared to studying at home. Then again, that tells me these things correlate with the migration penalty. And if it is the case that it captures some aspects of the migration penalty, then controlling for them in the choice model that I’m estimating should attenuate some of the disutility that students attach to migrating away from home.

I also do that validation exercise, and I find that once I control for those beliefs, the disutility coefficients that are attached to each of these destinations shrink by approximately half or 60 percent. It’s a little bit dependent on the specification. And that simply tells me, these beliefs that I’m measuring, they seem to be relevant for this object. It’s a sanity check. It’s a statistical decomposition that I then do. I try to understand to what extent each of these beliefs accounts for the attenuation of the coefficient. And then, I find that they approximately capture 15, 20, 30 percent of this attenuation. 

So, that’s a long explanation to just say I measure specific beliefs, and then I validate them against this object that I estimate from the discrete choice experiment. Both of them tell me, “You seem to be capturing some relevant objects for these students.” And they do explain an OK share of the migration penalty, or the disutility, that I’m measuring.

The Campus Trail Randomized Trial

RAJAGOPALAN: Yes.This is super helpful because, again, as I mentioned, when I first started reading the paper, 61 percent home preference, these numbers are just so large. So, it’s very useful to be able to unpack it the way you’ve done—decompose what’s going on at each of these levels. I also want to talk to you about the gender difference. 

But before we get into that, I want to get to the meat of your paper, which is this Campus Trail intervention that you designed and that you implemented, right? So, this is the randomized trial where you have this migrant peer that is attached to the student. Can you tell us exactly what this Campus Trail intervention was and what happens during this peer or mentor interaction, and maybe, if you can take us inside, how both sides perceive this conversation? Whether it’s the student, or what kinds of peers are more likely to be attached to them and so on? That would be super interesting.

DEMIREL: Right. So, the idea behind this intervention is that we’re talking about very subjective assessments of living at different destinations and being a student in different places. This is not a space in which we can think of, in my opinion, objective information interventions and things like that. But most of the time, when in real life, if we’re moving somewhere, if we’re considering moving somewhere, we try to think of a person that we know there maybe we can have a conversation with. We try to ring up them and say, “Hey, what is your experience like living in this destination?” 

This is where it really came from. I designed this program, Campus Trail, as a mentorship program that helps students make more informed decisions. And of course, there are many, many different mentorship programs that are aiming to do exactly the same thing. What is really different, or maybe a bit more special about this program, is that it’s also targeting these nonpecuniary aspects of choosing a college. And that also includes moving away from home and managing everything that comes with moving away from home.

RAJAGOPALAN: Yes.

DEMIREL: What I do is, I recruit students who are either current or recent college students in Mumbai and Pune. Most of them have moved away from home to come and study at these colleges. I train them to be migrant mentors. I say, “train,” with almost quote marks because I’m not the one doing any training. They’re the ones who know about the experience of moving away from home and navigating this really complicated period of choosing a college.

But we essentially go through the point of this program, the safeguarding. Yes, some safeguarding rules, some general tips on mentoring. We established, basically, the ground rules of what this mentorship is going to look like. 

These students come with a lot of knowledge and a lot of insights into what it is to be moving away from home, and they can support the students who are currently going through that choice themselves. They can provide a lot of information on how to apply to colleges, how to find information about colleges. They can do that irrespective of the destination. But they can also offer specific tips and tricks and information about budgeting, or finding housing, or navigating the city in a safe way, or arranging travel, and anything you can think of.

RAJAGOPALAN: This really makes sense, right? Because those of us who’ve been to university, you see this as a university student experience. So many times, when you migrate from a particular location, you see student housing where everyone’s from the same country or city.

And that’s just a matter of networks, or everyone’s from the same course. They’re all studying botany or something, and that’s how they found each other. And then, that becomes the apartment that is rented to botany students. [chuckles] Each time, when someone graduates, the next one comes in. So, you actually do see this even on the other side. And now this makes perfect sense on exactly how this pipeline moves, both in terms of easing the frictions, but also for those who actually end up taking that leap and making that choice.

DEMIREL: Right, exactly. We also often see this as something that schools or universities do after a student accepts their offer. I think that is also because people do recognize that these things are important. What I’m trying to understand is that having access to this beforehand, can that influence the outcomes or what the students are aiming to do with their lives to begin with? So, that is roughly the idea behind the intervention. 

What ends up happening specifically is that these mentors are available for phone conversations, and each of them commits a certain number of hours for each mentee. I randomly offer to half of my sample phone conversations with these mentors. And I make sure that the mentors and mentees are always sharing a gender and academic stream. That’s partly to make sure that the information exchange is relevant for the student, given that everything is going to be very personalized. We didn’t get to talk about the gender differences now yet, but spoiler alert, it’s a very gendered decision whether or not— 

RAJAGOPALAN: It’s very large.

DEMIREL: [chuckles] Migrating away from home, it looks different if you’re a man or a woman. Those are the things that I have in mind when I make sure that they share a gender and academic stream. 

The conversations are very personalized. It’s not scripted. It’s up to the mentee to follow up with the mentor and ask them about different things that they’re curious about. It’s also really coming from a place of—these students can use the mentors as a source of information. The mentors are not there to advertise, in a way, their own choices or their own universities or anything like that, which I think is the beauty of mentorship interventions usually. It’s really tailored to what students think is important for them.

The Study’s Most Surprising Result

RAJAGOPALAN: This is the most surprising result for me in your study, other than just the size of some of these results, which is that mentors in Pune and Mumbai don’t end up inducing the students who they are attached to to move to Pune or Mumbai. Instead, the response is a little bit more concentrated in the nearby district capital. So, what is going on with the Pune and Mumbai mentors or these cities or the opportunities in those cities?

DEMIREL: Right, so, I agree. I was also surprised. At the end of the intervention, I look at applications, I look at whether the students are moving to different cities, and I look at how the students are thinking about different cities. Throughout all of these outcomes, it’s really shining through that students—a subset of students at least—are getting more excited about moving to district capital or applying there.

But really, nothing seems to be changing for them on how they see the faraway cities: whether they think it’s a good opportunity, whether they think they’ll be happy there, and so on. None of these things seem to move for the further away cities.

I make sense of it in two different ways. The first one is to go back to the discrete choice experiment. To the extent that the migration penalty object that I’m capturing, if we take that very seriously, as you also mentioned, for a lot of the students moving to the nearby district capital, which on average is about a two-, three-hour move for most of my sample, that is the first available move that costs them less in some sense.

The nonpecuniary penalty, or the migration penalty, associated with such a move is much smaller than moving outside of the district to some of these bigger cities. Compared with such a large penalty, my intervention, which essentially, it is a very light-touch intervention, it is for majority of the students—it just becomes a short phone call with a mentor who’s farther away. It doesn’t do much to change the real costs or the real barriers that are associated with them.

Now, I have to say, I was surprised to see not even, “Oh, it sounds great to study at Pune, but I don’t think I’ll do it because my grades aren’t enough,” or so. It’s really that students, I think, are coming out of these conversations thinking, “You know what? I think it’s great to stay at home or at least close by to home,” and they’re not even getting excited about these outside options.

Which then leads me to think, yes, these students have talked to a real student who has done that move, and they just reinforce their beliefs about how difficult or uncomfortable or unsafe or any of these other undesirable things that such a move will consist of. So, that, I think, is an interesting finding that I have, [chuckles] that it seems to be that most of this disutility that I’m capturing is quite difficult to move, and it is potentially consisting of real nonpecuniary costs or barriers that the students are seeing accurately.

RAJAGOPALAN: The way I think about the Pune–Mumbai result—and maybe we can center it around something a little bit more familiar to you and me, which is doing an economics PhD. I feel like at different points in time, if you ask someone who’s doing their qualifying exams, they’ll be like, “Oh, good God, don’t do a PhD, it’s just a lot of work.” Or if they’re in the middle of the job market, it’s like, “Oh, this is so hard, the job market’s terrible, please don’t do a PhD. Get a private-sector job, you’ll be better.” [laughter] Then, a few years after that passes, it’s like, “Oh, you must do an economics PhD; it was the greatest time of my life.” Basically, they’ve forgotten.

DEMIREL: Right.

RAJAGOPALAN: Right? In fact, most people who are mentored to go on to do an economics PhD, it’s usually by their undergraduate econ professors, who have enough distance from that time. So, one way I think about what may be getting picked up with the Pune and Mumbai mentors is, they are still in the thick of figuring out how difficult that choice is to navigate. Right?

Maybe if it’s someone many years older, or someone who just moved there and is still super excited and in the honeymoon phase, it might be a different conversation. But the other way I think about that same conversation is, someone who manages to make that leap and move to Mumbai or Pune suddenly seems like a very different type of person. 

So, sometimes it may also be when you’re talking to these mentors, you’re like, “Wow, I could never be that. I could never do that,” and that becomes the barrier. Almost like there are lots of people who look at PhD students, they’re like, “Wow, that person is so smart, I could never be that.” Whereas all that person has done is taken two, three years of classes [chuckles] and seminars. So, what do you think is going on within this subset? I’m not asking for exact conversations, but what is your hunch on what’s going on?

DEMIREL: Right, so, I think what you say about the relatability and this distance that students have to mentors, this shows up in my take-up measure, actually. Out of all the students who are offered to have a conversation with these peers, almost all of them do have one conversation, but roughly only half go on to have a conversation that lasts for more than 15 minutes.

So, I think this is where that relatability, or, “How is this information relevant to me? You’re just putting me on the phone with some really highly educated, really big-shot student from Pune.” I think some of it is potentially driving—some of the students are just not taking up because of that. They think it’s a daunting thing to be on the phone with this person.

But among the students that do take up the intervention and have that conversation, I actually see overwhelmingly positive reflections of what they thought of the mentor. They thought that the mentor gave them good advice, that it was relatable, and that they trust the mentor, and so on. The other thing is that it’s not that they just talk about the academic opportunities and so on. They do talk about safety or making friends or cost of living. All these other dimensions, they do come up in the conversations. What could be happening possibly is that the students, depending on what it is that they have their mind set on, they take up different topics or they ask about different destinations. And, because the mentors are specifically instructed to help the students achieve the goals that they want, it could be that the information that they’re giving to students who are wanting to go to district capital is simply about how to find colleges there or how to apply for scholarships there and to make the cost of living more manageable or things like that. Whereas the topic of how to move to Pune is less likely to come up.

So, that’s one conjecture that I have. I’m still working on trying to understand it more deeply. But, yes, I think, it is possible that some of the students could have benefited from this intervention if it was a bit more structured, but it’s also true that some of the students benefited precisely because it was not structured and it was just tailored to their wants and needs.

The Effects of Gender on Students’ Migration Decisions

RAJAGOPALAN: Yes, and now I really want to come to the gendered part, because there are two different things going on with male and female students. One is just the treatment effects, and there seems to be a difference between male and female students just on how they look at the intervention and their response to this mentorship network, or mentorship intervention. The second is, of course, the difference in the size of the effects, right? So, maybe now we could get into that.

DEMIREL: Right. So, as I mentioned, as I hinted earlier, it is a very gendered decision whether or not to move away from home. Just when we’re already looking at the willingness-to-pay estimate, so, what do I need to pay a female student to make them equally happy to study away from home, and comparing that with a male student, I find that that willingness-to-pay estimate is approximately three times as large as the male students’.  And my sample is majority female students. Approximately 60, 65 percent of them are female. So, that is also somehow one of the driving factors behind those extremely large willingness-to-pay estimates that we see. 

So, what does that mean, comparing men and women? We need to remember that the willingness-to-pay estimate that I’m calculating is essentially the disutility attached to a destination divided by utility attached to a 1 percent increase in earnings. And if I look at these two objects for men and women, I actually find that the disutility attached to the destinations isn’t very different for men and women. What is really different is how much a 1 percent increase in expected salary is worth for men and women. It’s worth much less for women. 

One of the reasons behind that is because women are just expecting to earn much, much lower salaries compared to men. Even if the additional value of $10 or ₹1,000 rupees is the same for men and women, that represents a much higher increase for women. That is really behind this large difference of willingness to pay. So, yes, I would essentially need to compensate men and women the same amount, but that same amount would represent a much higher increase for women. So, that’s on that one side.

But at the same time, I also find that the women’s subjective expectations about what life away from home is going to look like, well, they expect it to look much worse than men. Actually, the expectations that I’m capturing do a better job of explaining women’s migration penalty than men. I think that is also helpful in trying to make sense of what my results look like at the end of the intervention.

As you hinted already, I find that men are responding in applications away from home and moves away from home. And women really don’t seem to be responding that much on those margins. I tie it back to this. Going back to the baseline discrete choice experiment, I find that women are just much less likely to be at the margins, so the intervention is trying to close a much larger gap. Whatever the intervention is going to do, whether it’s correcting misperceptions or decreasing real costs, it has to do a massive shift. Simply, that’s not what the migrant mentors, I think, are able to achieve.

Family, Marriage Expectations, and Other Factors

RAJAGOPALAN: And it may not be a mentorship problem at all, as you pointed out, right? They’re not yet at that point where mentorship is the thing that changes the decision on the margin. It could be family, parents, so many other things going on. Is the expectation that the girl is going to get married very early, at which point, she’ll be forced to drop out of university, as opposed to be able to continue things if she’s married within the same district. So, there are so many specific things going on, that it seems very intuitive and contextually obvious to me when you point this out. The moment you point it out, that makes sense.

DEMIREL: Right. Again, it’s when I look at the specific beliefs that I even see a worsening that’s attached to destinations away from home. So, I think some of these things are coming through. The girls are really not seeing studying away from home as a beneficial opportunity in any margin that I’m looking at. 

I think a lot of people have in the back of their minds that even if girls are excited about moving away from home, they’re going to hit so many other constraints on the way, which absolutely is true in this setting—maybe not necessarily for my sample, because they’re essentially a selected bunch of students who are responding to this invitation to join the program. I essentially go to their schools and tell them, “Hey, I have this mentorship program. If you’re interested in studying further, you can join it.” So, it is not necessarily going to be the representative class 12 student.

But what I’m finding is that even before we get to all of those parental constraints or other marriage constraints and so on, the students simply start thinking that, “Well, I don’t think this is a great decision for me. I actually start thinking that staying home is going to give me a more comfortable and more preferable environment to study in.” 

So, I think that’s important to note that, even though the parental constraints might be there, at the end of the day, I don’t have any evidence that, that’s where the students are getting stuck on. It seems that the students themselves don’t really see it as a good opportunity. Yes, I think it’s just important to know that.

RAJAGOPALAN: Yes. On the parents thing, so you’re not finding this difference between parents’ preference when it comes to explaining the difference between the two genders, male versus female choices in this instance, but your parent survey does suggest that parents are substantially more reluctant than their children to have their kids migrate, right?

DEMIREL: Yes.

RAJAGOPALAN: Is the intervention in some sense almost for the wrong person, in the sense that these students are still relatively young; their parents’ influence, or how they think about it, or the parent preferences are still going to loom very large? If they were older, it might matter a little bit less. If they were younger, the student preference hardly matters at all. It’s all the parents’ preference. But at this interesting age that you’re looking at, maybe what you’re picking up on is that the parents’ opinions really matter a lot and their reluctance must come in some way.

DEMIREL: Right. So, I think that’s exactly right. Of course, I’m looking at 17-, 18-year-olds. What parents think is a big, important part of the decision. And I do find—especially within my sample of male students, I do find some evidence of the treatment effects that I estimate being contained in the group of students who, at baseline, have told me, “I’m going to be the one deciding where to study and whether I study. Not my parents, or not even together. I’m going to be the only one who’s deciding.”

So, I think that is a bit in line with what you’re saying is that, it is possible that other students were also more excited about going away to study at these destinations but they got stuck at what parents are thinking about. For men, I think that’s possible. 

For women, I really don’t find any evidence of that either. For women, it could, of course, be the case that women heard something from their mentors, but then the parents were telling them the exact opposite. Then, they choose to believe what the parents are saying and not necessarily a mentor who they don’t know well and so on.

That is not something that I can speak to. But I really don’t have any evidence of students’ wishes and parents’ wishes for my girls going in opposite directions. It really seems to be that everyone is aligned that it’s just not a good outcome for the girls to move away, which I think is not—I don’t think that means that that’s where we should leave the conversation. I think we should try and think a lot more about what that means for these girls, that they are perceiving really, really large nonpecuniary cost of migration, and it doesn’t seem to improve with contact with mentors who can tell them about what life is like. 

So, what do they consist of? And what can policy do to change those real costs? I think there’s just a lot more to work on there.

RAJAGOPALAN: Oh, yes.

DEMIREL: And I find it exciting to think more about that.

Welfare and Policy Implications

RAJAGOPALAN: Yes. Thank you for explaining the paper so well, but before I let you go, I do want to ask you about what your paper and your findings mean for both welfare and policy in this setting. What is a good way to think about it? Second,  when I finished reading your paper, my first instinct was, is more migration necessarily better? I’m obviously a migrant, and I’m a huge votary of it, and I’ve moved countries many times, cities many times, but it’s not that obvious to me that more migration is necessarily better for most people other than me.

DEMIREL: Right. I’m also an educational migrant, and I moved away from university for my master’s, for my PhD. It does seem to be the case that it’s a certain type of person who either doesn’t necessarily value these nonpecuniary losses very highly. They just seem to value the economic returns very highly and they decide to do it. In that case, migration becomes a very positive change in someone’s life. But for most people, it really seems to be that that’s not necessarily the case.

I think as economists, we care a lot about accessing economic opportunities, and we make sense of the spatial immobility by saying, “Oh, there must be a lot of costs of migration, and these must be nonpecuniary in nature.” I think that’s right, and I think that should really guide our thinking when we design policy interventions, that we should target those things together.

We think a lot about correcting misperceptions or getting people to opportunity. I think what my paper is saying is that, well, if we just do those things and not address any of the real nonpecuniary costs, first of all, we might be doing something that’s not going to be super effective. And I think, second of all, we might be doing something that’s not necessarily welfare-enhancing if it comes at the cost of large nonpecuniary decreases.

Instead, I think we can think about a lot of different interventions that target both of those things separately, so both getting people practical, or financial, support to help them go after the opportunities. while at the same time targeting these nonpecuniary costs. Safety is something that comes up a lot throughout my paper. It’s a very salient worry for girls.

One can think of it’s just a fear of the unknown. It’s just something that they would get used to. But I think what my paper says is that, it’s a real, real cost that the students are perceiving accurately, and they’re pricing that into their information. What we should be thinking about is how do we make girls migrate safely, and not just how do we make girls migrate.

RAJAGOPALAN: Oh, yes. No, absolutely. The women’s safety is—at some point, we’re all pricing this in. You’ve done field work in India. I grew up in India. Sometimes it’s literal dollars and cents, right? The difference between taking a metro or a bus late at night, versus getting into an Uber, which is much more expensive. Finding a better apartment or a shared apartment in a better neighborhood. A lot of the safety is priced in and is much more expensive for women to be safe in these conditions than for men to be safe. Everything you said makes perfect intuitive sense. Again, thank you so much for the paper. I had a great time reading it. I learned a lot. I’m really excited about all of your findings.

DEMIREL: Thank you so much for having me. It was a pleasure to discuss my paper here.