Gaurav Chiplunkar on the Frictions in India’s Labor Market

SHRUTI RAJAGOPALAN: Welcome to Ideas of India, where we examine the academic ideas that can propel India forward. My name is Shruti Rajagopalan, and I am a senior research fellow at the Mercatus Center at George Mason University.

Today my guest is Gaurav Chiplunkar, who is an Assistant Professor in the Global Economies and Markets group at the Darden School of Business, University of Virginia.

We talked about labor market frictions in India, understanding the reasons for low female labor force participation in India, the importance of information flows and networks, the role of middlemen, 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, Gaurav. Welcome to the show. It is such a pleasure to have you here and do this in person.

GAURAV CHIPLUNKAR: Thank you so much for having me, Shruti.

RAJAGOPALAN: Yes.We do this all the time. We just don’t record it on the podcast.

CHIPLUNKAR: [laughs] Yes.

RAJAGOPALAN: So I feel like it’s high time we record it on the podcast.

CHIPLUNKAR: Yes. I’m excited.

RAJAGOPALAN: And bunch of your coauthors have been here, like Ritam most recently, Ashwini Deshpande. I’m sure there are more.

CHIPLUNKAR: Yes.

Examining Female Labor Force Participation in India

RAJAGOPALAN: So, I’m excited. I want to start with your papers. You are my go-to person for all things labor market in India, because you’re somehow studying it from all different angles: demand side, supply side, field experiments, RCTs, general equilibrium models. Somehow you’ve managed to cover the gamut.

CHIPLUNKAR: [chuckles] Yes. 

RAJAGOPALAN: But I want to start with female labor force participation.

CHIPLUNKAR: OK.

RAJAGOPALAN: The headline news is, in India, female labor force participation is both low and declining, and also that there’s a fair amount of variation between the states. It’s just difficult to get a sense of what the problem is or what the solution is, because there could be so many different factors, demand-side factors and supply-side factors. I want to start with the one particular paper of yours, your latest in Econometrica—congratulations—

CHIPLUNKAR: Thank you.

RAJAGOPALAN: —with Penny Goldberg, where you build a quantitative general equilibrium model, where you look at India’s entire labor market, and you look at both men and women, and whether they choose to work, whether they choose to run firms, whether they are formalizing, or if they’re hiring workers, and so on.

What you model in particular is gender-specific barriers at each step. You also treat every state as its own closed economy, which is a lovely way to model this because India has so much regional variation. The big takeaway from the paper is that women-owned firms end up hiring more women. 

But before we get to the results, can you walk us through how you actually model something as complicated as the Indian labor market, in particular female-led firms and female employment? And then, what is the exact mechanism through which this multiplier effect is happening?

CHIPLUNKAR: First of all, thank you so much for having me. I’ve been a great fan of the show and your podcast, and it’s so exciting that I’m finally here and we’re doing this. So, thank you for having me.

To your question, I think, let me preface the thought behind that paper with two things. One thing which I learned very early on is that you can never generalize anything in India because what works in one part just doesn’t work in the other part. Making some general claim about, “This is the problem,” or, “That is the problem,” in some cases is true, but in many cases is not because one can always find exceptions. And so, that’s coming to your insight about why we look at different states and not just the aggregate, is because things might be very different.

Two things. One is that the aggregate matters because, of course, there’s a rich randomized control trial literature trying to look at different barriers to female labor force participation, whether it’s work, whether it’s the type of work, whether it’s where you work, how you work, get renumerated for it, so on and so forth.

The aggregate becomes important from a policy perspective because, once you start implementing this at scale, then you have changes in different channels through just market responses, which could be women entering the workforce, starting these firms, or not starting them if they’re facing barriers. And then, might crowd out or crowd in, depending on what the data tells you, either same gender or men or different male entrepreneurs, so on and so forth. Wages, prices. There are lots of other factors that, once you start scaling up, become important to understand.

The model is obviously a simplification of how this world works, so we don’t want to overclaim in terms of “this is exactly how the world works,” but I think it provides interesting insights in terms of understanding this problem.

RAJAGOPALAN: And you do build the model with real-world data— 

CHIPLUNKAR: Yes, absolutely.

RAJAGOPALAN: —on the Indian labor market and Indian firms, just to be clear on that.

CHIPLUNKAR: One hundred percent.

RAJAGOPALAN: It’s not an exact one-on-one map of the real world, but it’s a pretty close approximation.

CHIPLUNKAR: Yes.Absolutely. It’s inspired by this almost, I would say, now 25-year-old paper by Hsieh and Klenow in 2009 or 2008 paper.

RAJAGOPALAN: Oh, it’s been 25 years, yes?

CHIPLUNKAR: It’s been almost—has it? I don’t know. Fifteen years, maybe. With the idea being that, look, if you observe certain types of distributions in the data, can you—those might be because of two reasons. One is that the fundamentals themselves might be different. For example, women might face different prices or different markets, or skill levels might differ, so on and so forth. Or it could be these noneconomic barriers, which could be culture, it could be norms, it could be other things that are very hard to move from a policy standpoint.

Some of these frameworks I find also helpful because they help very transparently, to the extent that you, of course, believe the framework, help you decompose and understand where things are coming from, as opposed to an abstract way of just thinking about “this is how I want to make the world work.” And so, somehow fitting the model as opposed to letting the data tell you which of these forces are actually more dominant in terms of what we are seeing in the data. That’s the motivation behind that.

I think the thing that also caught our attention, which has been very under-emphasized in the literature, and it’s not an India-specific story, though the paper is about India, is that women-owned firms hire more women.

RAJAGOPALAN: Yes.

CHIPLUNKAR: For some reason, we didn’t find a lot of literature trying to look at that across the spectrum. These are not only small, informal firms. These are also large. Once they get larger, these are in the formal sector, informal sector. Then we use the World Bank data to then look across countries, and the pattern is robust across countries, including advanced economies, and so on and so forth. That we were really struck by.

Now, again, you might think about, is this—as you said, do they end up hiring other women because of a lack of option or is that a preference? Because of other things that might make, say, the workplace more—whether it’s safety, whether it is about amenities, all these other things.

RAJAGOPALAN: Yes, and it could also, within entrepreneurship, be the demand versus the supply side, right?  Either men have more opportunities, so they become more competitive, so the wages are higher, so women are cheaper to hire. 

CHIPLUNKAR: Exactly.

RAJAGOPALAN: That’s one possibility. The other is the within-firm transactions cost which is men are harder to manage for women entrepreneurs depending on the social norms, and so on. Or men may simply not want to work for a female boss, and so on and so forth.

CHIPLUNKAR: Exactly.

RAJAGOPALAN: Even there, it’s nice to study this as the broader, general-equilibrium pattern because you can see that the pattern is sticky.

CHIPLUNKAR: Exactly.

RAJAGOPALAN: Now, whether the mechanism is culture, or whether the mechanism is skills, or the mechanism is wages and competition, the broader literature has to pass—

CHIPLUNKAR: I think that’s one of the limitations. I view all of these as tools in a toolbox.

RAJAGOPALAN: Yes.

CHIPLUNKAR: And so, some things are great for some things and not for other things. I think one of the things that one should be cautious about in these models is—at least the ones that this paper talks about—is that it’s not a policy prescription.  So, it’s not going to be able to tell you, should you have Policy A versus Policy B? I don’t think this model or this framework is geared towards that. What it is geared towards is trying to understand how important are these things as opposed to other things that might crowd in or crowd out participation of women along different dimensions. 

It’s very rich in terms of a framework, in terms of looking at this, not only at the state level, but also state sector level. We actually have agriculture, manufacturing services. We have barriers to entry as entrepreneurs. We have barriers to formalization of firms.

RAJAGOPALAN: Yes.

CHIPLUNKAR: Is it really the fact that you have a lot of informal employment where women work, but the real barrier is the fact that they can’t formalize their firms and grow their firms? Or is it really, for example, on the demand side, hiring women versus men? We explore a lot of these channels in this paper, which is why I think it’s pretty insightful in terms of some of the patterns. 

Broad Patterns in Female Employment

RAJAGOPALAN: So, what are the broad patterns that you can share with us without discussing every single result in the paper? What are the broad patterns, especially when it comes to different states and different kinds of firms? I think that might be the most interesting part.

CHIPLUNKAR: Yes.I think two patterns. One is that the women hiring women, I think, is interesting, particularly from an Indian standpoint, because after actually this paper, I have gotten a lot more interested in looking at labor demand-side policies for female labor force participation as opposed to supply-side policies. Because norms and culture, as we have seen, are extremely hard to move. Obviously, there’s little policy can do about it.

RAJAGOPALAN: I’m also not a huge fan of that literature. There’s variation within the literature. We’ve had some very good people, some on this podcast, talk about it. But also, just the idea that women, especially Indian women, in certain states and certain castes are this exotic thing, completely bound and trapped by culture and norms. It’s too much of a deviation for me from Homo economicus. Right?

CHIPLUNKAR: I totally agree.

RAJAGOPALAN: Everywhere else, markets and prices seem to work. Women are rational too. 

CHIPLUNKAR: Exactly.

RAJAGOPALAN: We’ve just got to figure out what that cost barrier is.

CHIPLUNKAR: Exactly.

RAJAGOPALAN: I don’t like the supply-side literature sometimes because I think it makes Indian women a little too exotic in a way that I’m very uncomfortable.

CHIPLUNKAR: Totally. Of course, much of it is true as well.

RAJAGOPALAN: Yes.

CHIPLUNKAR: There’s some water to that story, but I think in many ways, it’s not the entirety of the problem that people are just chained to their houses and they can’t move and so on and so forth. I think it’s a combination of the two, but I think the latter is extremely under-emphasized in terms of how the narrative around female labor force participation in India is built, at least to my knowledge. Of course, there might be other opinions on this.

Again, in this paper, that’s one of the things that I think would be nice to emphasize, which is the labor demand side of things and the reason that female entrepreneurs hire more women. Therefore, this connects back to this policy of, in the data, there are very, very few female entrepreneurs to begin with.

Then the obvious question is, if you have policies that promote female entrepreneurship, apart from just a social-equity standpoint, this can actually be efficient for the economy in terms of generating growth and act as a multiplier effect on female labor force participation. Because by creating one entrepreneur, if that entrepreneur hires five women, then your labor force participation goes up not by one, but by six women, actually. There’s these labor force participation multipliers that can come in.

Then, of course, that can lead to other things in the economy once you start aggregating things up. That was one result that was very neat and robust across states and across contexts that we find. In fact, the other paper we started working on was the India Policy Forum, which is organized every year. They invited us to then look at this with more recent data in the Indian context. Again, we find very similar things in terms of the barriers to entrepreneurship versus work.

RAJAGOPALAN: Yes.

CHIPLUNKAR: The second thing which was interesting is trying to break—when we think of entrepreneurship, I think it’s important to understand whether this is self-employment, whether this is really owner-operated enterprises, or whether this is really firms in the true entrepreneurial sense. The big margin is the fact that, well, there is a barrier to just women not working. Though I think there’s some recent literature trying to look at this from a measurement standpoint. 

RAJAGOPALAN: Yes.

CHIPLUNKAR: Right? There’s obviously work that needs to be done around this. But of course, even if we were to measure all of these things correctly, I don’t—

RAJAGOPALAN: It’s still too low.

CHIPLUNKAR: It is still low. Whether it’s declining or whether it’s on the up in the past few years or not, all of that is a local trend rather than a global optimum. But what we do is then try and look at, is the margin—so, there’s a clear barrier to labor force participation, but conditional on work. We actually find that women are overrepresented in these self-employed, owner-operated enterprises relative to men. It’s not the case that entrepreneurship is hard.

RAJAGOPALAN: Yes.

CHIPLUNKAR: Right? It’s not the case that I just can’t start, say, a sewing shop in my house.

RAJAGOPALAN: And that is the classic one. Most female entrepreneurs are actually doing sewing work.

CHIPLUNKAR: Exactly. It’s not that. So, there is a barrier of participation, but conditional on participation. It’s not really entrepreneurship that is the problem. It is really growing your firms to be larger which seems to be the barrier. 

RAJAGOPALAN: Yes.

CHIPLUNKAR: Again, as I said, these are not policy prescriptions, but it points you towards where we should be thinking about from a policy standpoint. That’s the second result that we find interesting.

Now, the third result, which is coming back to this “women hiring more women,” I find it to be fascinating. Because I think there are two sides to this in terms of a preference as opposed to a constraint story. I would love to, in the future, dig into this a little bit more in terms of trying to understand, what are these constraints or are these actually just preferences? 

There’s a homophily literature on this, like men hiring more men, men promoting more men. There’s the entire literature of homophily on this as opposed to constraints to the ones that you were talking about, like do men not want women bosses? Do women not want to hire junior men? It can go in many ways.

RAJAGOPALAN: Or it could be something as simple as, if most of these women who are entrepreneurs are one- or two-people shops in a simple room in the home or adjoining the home, they may not want a strange man to come to the house. It’s just much easier to have a stranger who’s a woman come into the house.

CHIPLUNKAR: Exactly.

RAJAGOPALAN: So, the constraints could sometimes be that basic and we just don’t know what they are yet.

CHIPLUNKAR: Absolutely, and I think this is ripe ground for randomized controlled trials to take place to be able to actually figure out, at a very micro level, trying to understand some of these barriers and learn from what are the barriers for hiring or growing their firms. That’s where the Econometrica paper is situated and the follow-up IPF paper on this, which is using more recent data.

Barriers to Entrepreneurship for Indian Women

RAJAGOPALAN: One other question on the Econometrica paper itself: What are the gender-specific barriers to entrepreneurship which are different for men versus women? One, you said, obviously, is scale. India is a country of microenterprises, so that’s, I think, true across the board, but a microenterprise is under 10 people. I imagine women cluster more around the zero to five or zero to four level. There’s obviously a scale barrier.

What are the other kinds of barriers? Is it regulatory at all? Is it just the same barriers you have towards workforce participation, which is you can’t leave the house, or there isn’t good-enough public transport? Or is it access to credit?

CHIPLUNKAR: Yes, the IPF paper actually digs into that a little bit more because what we do there is we go to this other dataset called the Global Entrepreneurship Monitor. There, they actually ask people questions about aspirations, about constraints, about whether it’s society, media, prestige, all of the other things. 

The World Bank Enterprise Surveys also—and they now have this nice informal firms model as well—you can look at the owner of the firm, or in many cases, whether a majority of the owners in the firm are women or not. Then they have these modules on different types of inputs, like electricity, water, licensing. Then you can look at, is that a barrier or not, and so on and so forth.

There, we actually find that many of these also come through just accessing services, accessing credits, where women report facing larger barriers to accessing these services. And again, we need to dig into more about whether these are informational constraints, whether these are other constraints in terms of just going to a bureaucratic office and sitting there and getting things done. 

That’s where I feel like there is a nice match to be made between these aggregate patterns and models with really micro-founded, well-done work using either RCTs or even descriptive data would be very helpful, to be honest, to just understand what are the functional barriers that people face on the ground. It’s a combination of many. I think that’s the hesitation in terms of hanging your hat on one. Because it’s going to be a combination of all of these things. But I think just to be able to understand which context, where, how, I think just doing some deep research on some of these would be really unlocking this last-mile barrier in terms of where the constraints are.

The IPF paper does talk a lot about this. It’s not really skills. It’s not really confidence. It’s not really all of these other things you might be worried about. Are women under-confident? They’re not. Are women under-skilled? They’re not. Of course, this could be self-reported, but if anything, self-reported data should bias it even more.

RAJAGOPALAN: Even more, yes.

CHIPLUNKAR: Right, and—

RAJAGOPALAN: If that’s the main issue.

CHIPLUNKAR: If that’s the main issue. The fact that we don’t find this—and of course, this part is not causal, but it’s still informative and descriptive. But it’s access to services. It’s access to credit. It’s access to taxes, being able to pay them. You’ll have to bribe, you’ll have to deal with these inspectors, as you know a lot, on the labor regulation standpoint. There’s a lot of this that happens, and you see that in some of these data.

RAJAGOPALAN: Yes, but I find the bribe thing is quite interesting because if a traffic cop stops me in Delhi, which is where I’m from, he’s more likely to have a conversation with the driver, even though I’m the one who’s actually going to pay the bribe.

CHIPLUNKAR: [chuckles] Yes.

RAJAGOPALAN: But they don’t want to have this conversation with what they call “ladies log.” [women]

CHIPLUNKAR: Yes.

RAJAGOPALAN: You know?

CHIPLUNKAR: Yes.

RAJAGOPALAN: They just don’t want to have—they won’t make the demand. I can’t even elbow my way in and have the conversation.

CHIPLUNKAR: Absolutely.

RAJAGOPALAN: Right? So, it’s really on both sides. There’s a friction, but it could also be like, this is where we have so much rich descriptive work on the supply-side constraints but we don’t have it on the demand-side constraints. It could be something as simple as maybe the local bank officer doesn’t see the potential in work that is sewing and embroidery. He just doesn’t understand the market. He understands a cycle or a mechanic shop a little bit better. Maybe the constraint might be we need better-educated bank officers, or we need more women loan officers. It could be something that basic, and we just don’t have a good sense of it. 

Which are the states that are doing well relative to the ones that are not?

CHIPLUNKAR: Actually, Northeast is doing excellent in terms of female labor force participation. South is doing much better. It’s the usual suspects that you might think about where it’s a concern.

RAJAGOPALAN: And the places that do well on labor force participation also do well on entrepreneurship, and you’ve just explained to us why because of the multiplier effect, right?

CHIPLUNKAR: One hundred percent. Coming back to your earlier point, which I want to double-click on a little bit, is this idea about if you have the men won’t talk to you, or loan officers won’t give you loan. I think there are randomized controlled trials that show these biases exist. I feel where the conversation needs to move is not so much towards documenting there is a bias, but what is the source of this bias. Which is obviously hard to measure. But I don’t think we have a good handle on that. In a very simple way, is it just taste-based or statistical? If it is taste-based, then certain things that we can think about, statistical, certain things that we can think about. But also, loan officers giving or not giving loans to women versus men for what—the problem with working with labor demand, as I have routinely faced, is that just getting firms is hard.

RAJAGOPALAN: Yes.

CHIPLUNKAR: No entrepreneur, especially for these big firms, wants to risk . . .. It’s very blatant once you talk to them.

RAJAGOPALAN: Yes.

CHIPLUNKAR: They’re like—

RAJAGOPALAN: They tell you.

CHIPLUNKAR: “Oh, gender is a big issue. If you come out and tell us that we’re not doing things right, we’re going to get into trouble.” Really partnering with these firms is extremely hard, especially for RCTs, but I feel even them releasing their own data, because many firms actually do have a lot of great data. I’ve worked with a bunch of job portals. They have rich data on applications, on postings, on this and that. They just don’t want to share it. Many of them just don’t want to share it, which is a shame.

RAJAGOPALAN: Yes, but that’s also because we’ve pushed this completely on the firms and not enough on the structural, broader issues, right? Because if the answer is, “It’s public safety,” the immediate media and policy lens will be on the firm, of, “Why aren’t you providing safe transportation them?” Right?

CHIPLUNKAR: Exactly.

RAJAGOPALAN: So, it’s just the way we think about the problem is a little bit twisted. Because we’ve given up on the state solving a lot of structural battles. [chuckles]

CHIPLUNKAR: No, exactly. In fact, there are papers that have nondisclosure agreements that need to—let alone the firm, not even mention India. They have things like “there’s a large developing economy” or something like that. Obviously, one knows what you’re talking about, but that’s the extent to which legal teams require you to go to hide identities in many cases.

RAJAGOPALAN: Yes, and I don’t blame them.

CHIPLUNKAR: Exactly. But that limits our understanding of unpacking—I’ve gotten excited really a lot about the labor demand side of things in the last many years because I think that’s the margin we can actually tangibly move and will have long-term spillover effects once we start—a great paper was the Gurgaon paper by Rob Jensen back in the day that shows when call centers opened up neighboring villages, women started skilling themselves because there were these jobs that were accessible, that were “female-friendly” in these MNCs.

RAJAGOPALAN: They provided transportation.

CHIPLUNKAR: At MNCs. Exactly. Exactly. And so, playing around with labor demand or having policy that is directed towards that can actually have these, again, spillover effects on labor supply. Maybe not in the short run, but definitely in the medium run. I think that’s a policy push that we need to go—does not mean we study a lot on regulation, does not mean overburdening the onus on—there needs to be a needle that needs to be threaded very carefully when it comes to addressing these issues from the labor-demand standpoint.

Well-Intended Labor Policy and Unintended Consequences

RAJAGOPALAN: The other reason I think labor demand is so critical is, I think we are a little too obsessed with labor supply, not just in the case of women, but overall.

CHIPLUNKAR: Yes.

RAJAGOPALAN: This brings me to your paper with Ritam and Vidhya. If you obsess so much over labor supply constraints, then you’re like, “Oh, they don’t get paid enough,” or, “They don’t have period leave or maternity leave,” or this, that, and the other. Without looking at the labor demand side, you actually implement policy. What you find is that everything becomes upside down.

For those who haven’t heard the episode with Ritam, we briefly discussed the paper. It’s a fantastic paper. You look at something so specific, which is contract workers versus regular workers, and once you start tagging on all these benefits to contract workers, what firms do is not just hire fewer contract workers; they prefer to be informal rather than formal. Actually, the supply-side-motivated regulation can sometimes be so burdensome that we actually flip it. You really tilt every firm, and you move them back into informality, which is actually worse for both when it comes to labor for men or women.

CHIPLUNKAR: Yes.

RAJAGOPALAN: This, I think, is one thing I’m learning from your work on why we should pay more attention to labor demand. But there are other things you find like that?

CHIPLUNKAR: Especially with this paper, I think another margin which I found fascinating is, I think, apart from the fact that I’ve written it, so I should like it.

RAJAGOPALAN: No, it’s a great paper. This is the second time I’m discussing it in two months, so I must really like the paper. [laughter]

CHIPLUNKAR: Apart from that, I think one of the learnings for me was this is exactly the test case for why aggregate GE things matter, because if you were to purely do a policy evaluation—so, the policy just to tell your listeners is, Andhra Pradesh, back in the day, basically said, “Oh, firms are evading formal, regular workers because once with regular workers, if you’re in the formal sector, you have to give all of these benefits, yada, yada, yada.” We know all of that. How do firms get around this is hire a bunch of people on contracts so they don’t have to comply with all of these.

RAJAGOPALAN: And this is still all formal employment.

CHIPLUNKAR: Exactly.

RAJAGOPALAN: This is not informal contract employment which is a whole other beast.

CHIPLUNKAR: Absolutely. Absolutely. And so, Andhra Pradesh one day decides that “we are going to ban firms from hiring contract workers”—they are formal-sector firms—ban them from hiring contract workers, push them to hire more regular workers. 

Again, a very well-intentioned policy because it solves a real problem of workers getting their due benefits within firms. From a “reduced form perspective,” one would actually evaluate this and find a lot of success, because that’s what we find. Formal-sector firms actually do move towards more regular workers, and they do shed their contract workers.

RAJAGOPALAN: Yes.

CHIPLUNKAR: Now, I think where the aggregation and equilibrium effects come in is once you start building up and saying, “Now let’s endogenize a firm’s decision to be formal or not,” and there you actually find the margin that the informal sector is just growing because obviously no crazy entrepreneur is going to want to formalize and get in the ambit of all of these inspectors. And so, what do you do? To some extent, you do find the informal sector basically growing, which this marries nicely these margins along. Here’s a well-intentioned policy that wanted to solve a real labor problem.

RAJAGOPALAN: And you just made contract workers more precarious as opposed to less precarious.

CHIPLUNKAR: Exactly.

RAJAGOPALAN: One thing that comes from that particular paper in Andhra Pradesh is that it’s very, very sensitive to price and wage, right? The products are clearly in a highly elastic market. A little bit higher wage is fundamentally difficult for them to pay, so the firms prefer to be informal. Or the regulation is so onerous that even in not a very highly elastic market—

CHIPLUNKAR: Correct.

RAJAGOPALAN: —it’s not feasible. It could go either way.

CHIPLUNKAR: Correct.

RAJAGOPALAN: But there’s another issue about skill, that the firms are not willing to pay a certain price because the skill level is simply not there, and they have to spend a lot of money skilling. Now, is this a problem across genders? Because there’s a lot of work that’s done on female labor force participation and skilling. But to me, it seems like men and women in India, when it comes to the labor market, there is both less skill and there is mismatch of skill. That is complete misallocation in the labor market. What can you tell us about something like that from this kind of an experiment?

CHIPLUNKAR: Yes, there has been some work that has shown—for example, Ach Adhvaryu, Namrata Kala, Anant Nyshadham, they have this paper with the Good Business Lab that shows that firms really underinvest in skilling their own workers, and the returns to skilling are huge.

RAJAGOPALAN: Yes.

CHIPLUNKAR: Right? Yet, I’ve spent a lot of time talking to entrepreneurs across the spectrum, and they’re just like, “If we skill them, they’re going to leave.” It’s as simple as that. They’re going to leave, especially in sectors where there is a lot of churn, especially in urban areas, for example. 

People are just afraid that the returns to skilling are not going to be borne by the firm themselves. That gets even more precarious once you have contract workers where you don’t have any way of holding workers back for a few months or a considerable amount of time for the firm to actually reap the benefits of the skilling itself.

RAJAGOPALAN: Or women, if they go on maternity leave, things like that.

CHIPLUNKAR: Exactly. Now, coming to the gender aspect of this, this multiplies in many ways because there are other constraints that interact with the skill acquisition itself and then constraints that might impact the way you supply labor to that firm. It’s a complicated question in terms of what should be done about it, but I think that the basic reason is everybody is appreciative of the fact that more skilling—in fact, we ran a small survey, this is back in the day in Uttar Pradesh, and you ask a youth, 21-year-olds who are just entering the labor market. They’re skilled. They’re coming out of these undergraduate colleges, degrees, vocational training. They’re clearly skilled. “Where do you want to find work?” They name all these skilled firms and where they want to work. Now, you go to the flip side and ask these firms, “What’s your biggest problem?” The number one thing they’ll tell you is skilled workers. Right?

RAJAGOPALAN: Exactly.

CHIPLUNKAR: Again, coming back to this, there is, of course, a skill acquisition problem, but I think there is a major problem in terms of matching and more than that, retention. It’s not obvious always that retention or the lack of retention is a bad thing if people are climbing up the job ladder, but that does not seem to be the case. It’s not like people who are working for delivery jobs are suddenly quitting their jobs to be able to go and become some other highly qualified job somewhere six months down the line. No, it’s basically because they’re doing something else which is parallel in terms of the skill requirements, in terms of pay, in terms of—then I think, there’s also a problem about just documenting panel-level data in India. I haven’t seen a good dataset that has a high-frequency collection of data on just how people are moving around in jobs for an extended amount of time.

RAJAGOPALAN: That’s true, actually.

CHIPLUNKAR: People are piecing together, stitching data, and making inferences based on what we find, but there is no real good panel data, especially for the youth.

RAJAGOPALAN: It’s also because of the level of informality, and most of our firms are tiny.

CHIPLUNKAR: Yes, exactly. Exactly. 

RAJAGOPALAN: It’s also the microfirm nature is one of the reasons retention is so bad, right? There isn’t that much place to move up. The best people usually want to move up. They don’t want to work in a four-people shop or a 10-people shop. If they get skilled, they want to move around. If they get married, they move. The number one reason for migration, especially for women, is marriage. 

There are so many things going on, and microfirms are micro because they don’t want to be formal, which means we can’t get the data. We have this problem. Your data problem is your regulatory problem is your job problem.

CHIPLUNKAR: Yes. It’s all intertwined.

RAJAGOPALAN: It’s all intertwined.

CHIPLUNKAR: But I think you need to start somewhere. [laughter]

Female Employment After Career Breaks

RAJAGOPALAN: No, no, no. Absolutely, you need to start somewhere. One of the other things that you’ve looked at when it comes to both skilling and women, and this is like an all-stars list of people working on female labor force participation. This is your paper with Ashwini Deshpande, Kanika Mahajan, Nandhini, Niharika Singh, and it’s forthcoming in the JDE. What I looked at is the design stage, because this is a preregistered RCT, and you’re looking at a small sliver of the labor market, which is highly educated women in formal employment in IT services. Just a disclaimer out there that this is not the entire labor market. This is not your general equilibrium paper kind of thing.

CHIPLUNKAR: Yes.

RAJAGOPALAN: You actually randomize them into three groups. There’s, of course, the control group that—this is women who are coming back from maternity leave, three years or more. There’s obviously a control group which basically just gets integrated into the workforce. Then there are two treatment groups: people who get some training and professional support, and then people who get professional support plus technical skill training. These are the three groups.

Now, I don’t know if you have results yet, but I just wanted to touch upon this. What are your priors on what you expect to find? If you have any results and you’re willing to share them, that would be great.

CHIPLUNKAR: I think the reason we got excited about this was, in some of my earlier work with vocational trainees, these were 18- to 21-year-olds. Where we worked was entirely in Uttar Pradesh. The number one thing when you ask the girls is, what about jobs? They’re clearly extremely excited about going and working in the labor market. But the number one thing they’ll tell us is that, as you said earlier, marriage. Like, “I’m going to get married, and then it’s going to be a whole different beast in terms of being able to bargain my way into the workforce later. Maybe I’m going to try this later once I have been married, and once I’ve settled in, then we can think about long-term work strategies.”

That got us excited. That was an anecdote. But then, once you look into the data, actually, women who want to come back after a career break, that population is nontrivial in India. It’s actually a large population of women. That solves many of these constraints of trying to move the needle on 21-year-olds. 

But then what if you’re a 25-, 26-year-old? You’ve been married, you probably have a kid or two, and they’re probably starting school soon. Now you’re at home, and you have a bunch of free time. That’s the set of population that we’re trying to study in this.

As you rightly said, this is not about informal work at all. That’s largely because of the partner that we’re working with, who has these training programs for these highly qualified, IT-skilled women. Nevertheless, I think it solves many of these problems. We can examine many of these. 

So, what are we doing? It’s still in the field. I don’t have results, unfortunately, to share, but stay tuned. I’m super excited about it.

RAJAGOPALAN: But what are your priors on something like this?

CHIPLUNKAR: We’re doing basically two things. One is to give them hard skills. 

RAJAGOPALAN: Yes.

CHIPLUNKAR: The one thing you can think about is skill depletion. I’m out of the labor for five years. This is IT, so it’s coding, especially now with AI, I guess, even more. One is just hard skills. Second is actually the softer side of things, right? Interview skills, CV-building skills, how to just answer a bunch of questions.

Now, having talked to a few participants, the one thing we are finding is that they’re searching more intensely because, as I said, this is in the field, so we don’t actually have outcomes on employment. This was just basically after they finished the training program. We just chatted with a few of them just to understand what they were thinking about and how they were going to talk about applying in the labor market. 

It looked like they were all going to search more “efficiently” in the sense that they know what they want, they know exactly where their skills land. Because now they’ve done the training program, they know exactly what skills they need to work on versus what they can actually remember, so on and so forth.

I think the soft skills is a huge part of this, because many of them, going in, before the training program, they were like, “I haven’t been in an interview for many years now. What questions will they ask? Who’s going to interview me?” Many of them are actually concerned that they might have to take a drop down on their career, and they were like, “How am I going to go and work with an 18-year-old analyst? I’ve been manager in my yada yada yada, and now I have to go and work as an analyst for a few years before I get promoted again.”

These are very real questions. To the extent that we can measure them and capture them either through qualitative and quantitative data, those are the things. My prior, basically, if you ask me, is, it’s not going to be as slam dunk of a result as we would hope, but I think on the margin, trying to understand who is this working for, I think will be the more interesting part for me.

RAJAGOPALAN: For me, actually, I love this study for a slightly different reason. India has maternity leave as good as Scandinavian countries, which seems a little bit ridiculous for a country at its GDP per capita. But that aside, that assumes that the biggest problem for women is not enough maternity leave, whereas the biggest barrier might be something like this.

CHIPLUNKAR: Absolutely.

RAJAGOPALAN: So, if we do have to focus our energies, whether as a firm giving an incentive or as the government designing policy, the policy solution may be somewhere else to bring women back into the workforce post maternity leave, and it may not be longer maternity leave. In fact, it may be something completely different and something much cheaper.

CHIPLUNKAR: I agree. There have been papers that have shown that once you think about how firms are going to respond, they reduce hiring of women in these— 

RAJAGOPALAN: Exactly, which they do anyway.

CHIPLUNKAR: In these fertility ages. Again, I think holistically looking at this is the right answer to trying to at least figure out what is the right way of designing policy support. I think reintegration is important. Again, in India, you multiply anything by infinity, and it becomes a really large number because just of the scale at which we operate. I think the scale also gives us a nice petri dish to be able to observe different kinds of constraints at really minute detail, which I don’t think any other country on earth has the luxury of doing. 

I think really drawing in lessons from almost the variation within the data is, I think, extremely informative to see how things might work under different constraints and what might work in one context and not. So, in this paper, for example, we have women who are across different cities. It’s not concentrated either in one city or not. It’s across different cities. It’s across a couple of different occupations as well.

I’m really excited to see what comes. We are in the end line phase right now. We’re actually six months out after the training program. We’re trying to gather data on their employment outcomes. We’ll see. Fingers crossed.

Female Labor Force Participation and Economic Growth

RAJAGOPALAN: No, I’m excited for this paper. More generally, everything we’ve been talking about has been like, “Let’s look at broad patterns in India. Let’s look at not just one supply-side constraint or demand-side constraint.” But the biggest thing about labor force participation, other than the multiplier effect, is what that leads to, which is obviously economic growth. That’s the big-ticket item that we’re all after. Women working has been either a symptom or a consequence or a cause. It’s complicated to disentangle these three of every single country that has had high growth for lots of years, right?

You have this great broad-based study that you’ve done with Tatjana Kleineberg. This is female labor force participation, but looking at it from the point of view of structural transformation across 100 countries. What is the lesson for us more broadly? India should be studied in its own right for various reasons, including female labor force participation is a really big problem. But what is it that we can learn from patterns across the world?

CHIPLUNKAR: Yes.That’s precisely why we started working on this. On this paper, what we’re doing is basically looking at 100 countries over the last five decades or so. Really putting together rich data on the type of work that you’re doing, what occupations are you working in on the job ladder, what sectors you’re working on, and how does that correlate with economic development. The first part is just basically documenting a whole bunch of patterns, which I find extremely interesting.

The big pattern is, if you go back to development econ 101, the first pattern you see is that, oh, as countries develop, people move out of agriculture into manufacturing and then services. Now, of course, in the last 15 years, we’ve seen this jump from agriculture to services, but it’s really all of this. Turns out that if you actually look at this from a gender perspective, all of what I just said is a male-dominated story.

RAJAGOPALAN: Yes.

CHIPLUNKAR: For women, what the actual patterns\ we find is you get out of agriculture. This is, again, Claudia Goldin’s famous work.

RAJAGOPALAN: Exactly.

CHIPLUNKAR: So, none of this part is really new.

RAJAGOPALAN: It’s a U-shaped curve, and we have seen it before. [laughs]

CHIPLUNKAR: Exactly. None of this is actually new. What is, I think, exciting and new is, therefore, once we open up the box of, also, occupations. Again, if you think of it from an economic development point of view, it’s probably helpful for economic growth to get excess people off the farms and into the firms. 

But then the question is if all the women are going to become receptionists and all the men are going to become the managers. It has very different implications for economic growth than if men and women have equal chances of becoming a manager at a firm.

That part, I think, is what we dig into much more carefully in this paper, and it’s exactly what you find. In the low-income countries, there is a gap in terms of the probability that you are a manager versus if you are a worker, let’s say, but there’s no gender gap, specifically between manufacturing and services. What happens is, as countries develop, service sector gap basically closes in terms of managerial occupations. Rich countries are actually able to close this gender gap in managerial occupations, but that’s a service sector story.

RAJAGOPALAN: Yes.

CHIPLUNKAR: In manufacturing, the line is basically flat. Whatever you had in low-income countries in terms of male-to-female manager ratios is almost the same in rich countries as poor countries. Therefore, a sector plays a role. Again, this is not to say that we know what is going on. I think that we need more microeconomic studies to be able to unpack some of these micro foundations of what it is about services as opposed to manufacturing.

However, the other interesting thing is that if you look at clerical occupations, in rich countries, women are three times more likely to be working in these clerical occupations as opposed to men. Whereas in the poorer countries, again, there is a gender gap that favors men in terms of these occupations. 

Again, with development, how it percolates into gender is extremely uneven. A couple of times that we talked to a bunch of people about this, I think where people were pushing us to think about more is that some of this could just be economics. If you think about classic stories of comparative advantage, if you think about operating-machines-versus-computers stories, it could be skill acquisitions. In low-income countries, the first person to get out of school is a woman because the boy still needs to go to school, whereas the girl child can drop out. This could be skill differences. This could be comparative advantage differences, and so on and so forth, that could explain why these patterns are the way they are.

The next part is, again, going back to a more quantitative story. If we try to, again, similar to the Econometrica paper, build in all of this machinery and then say, “OK, giving it the best frontier models that are out there right now, can we explain the story completely by just economic gaps in terms of skills, comparative advantage, so on and so forth?” 

The answer is no, but then the quantification is interesting. Of course, this is all contingent on the model, and there are lots of asterisks attached, conditions apply, all of that there. But what we actually find is that gender barriers, which are these noneconomic barriers, to either work or earnings, explain about 25 percent to 30 percent the economic growth that we have experienced in these countries over the last 50 years. It’s not trivial. 

RAJAGOPALAN: Yes, it’s not trivial at all.

CHIPLUNKAR: Twenty-five percent to 30 percent of your growth is explained by the fact that the workplace or the labor market has become more gender-equal. Then the other 70 percent is because skill gaps have closed, because technology has developed. There’s a lot to say also about just good old-fashioned development, like get people in school, right? These kinds of things do work, but then there’s a nontrivial layer on top, which is 25 percent to 30 percent. That is just about changing some of these underlying constraints that women face, which are not really economic constraints.

RAJAGOPALAN: There are two sets of outliers in the global story. In the rich world, the Scandinavian countries are the outlier when it comes to gender participation. In the developing-country, and now very much in the middle-income or developed-country world, it’s the East Asian countries that have this incredible story where it was partially manufacturing-driven there for women. Is there something we can learn, maybe not from individual countries, but largely, these two clusters, which have some commonality, and the timing of it also matches in these countries? What is it that we can learn in India from these two ends of the spectrum?

CHIPLUNKAR: I think that’s a great question. Now, I’m completely going to make claims that are unfounded in any of the data or anything. But I think for example— 

RAJAGOPALAN: I’m thrilled. That’s what I do for a living. [laughter]

CHIPLUNKAR: Bangladesh and Vietnam being the two posters of really high growth in female labor force participation over the last few years.

RAJAGOPALAN: Sorry, as a random aside, Bangladesh is the number one reason I hate a lot of the supply-side-constrained literature. [laughter] It’s like anytime someone throws these norms and women in India are exotic and whatever and cultural, I’m like, “Bangladesh, Bangladesh, Bangladesh.”

CHIPLUNKAR: Oh, 100 percent. In fact, one of my advisers, I put it very nicely. It’s like, if your family is starving, people don’t really care whether it’s a man or a woman who’s working. There is a huge constraint in terms of these jobs that are being there. In fact, Farzana Afridi, I remember, has this amazing paper that I like on the fact that they connected a whole bunch of women to job portals. If I remember correctly from that paper, what they find is that women actually don’t take up these jobs, but their husbands are more likely to get better jobs. And so, that’s telling in terms of how technologies, for example, can percolate. Now, coming back to your original question, I think it’s just about good old—the larger scheme, to my understanding, seems to be from the garment sector again, which is a sector in which we have seen a lot of women labor force, even in India. Even in India. 

We’ve seen a lot of big firms enter these sectors which have been able to draw women into them. Maybe one lesson in all of this is probably if it is manufacturing, and if you want to view manufacturing from a gender-equity standpoint, then looking at sectors where there is possibility for women to come and contribute effectively.

Scandinavia is a completely different story. I think it’s probably not the right counterfactual to look at from the Indian perspective because, again, we have these laws. 

As you were saying earlier, maternity leave in India is amazing—on paper. Right? But then once you actually look at enforcement, once you actually look at how firms are able to get around it, for example, these issues become way more binding. Maybe instead of trying to emulate Scandinavia in terms of forming laws, maybe it’s also time to just, as you have argued in many of your pieces, just reduce regulatory burden on many of these things that really allow firms to grow.

Some of this growth in the short term might have consequences in the longer term that might be beneficial. That’s what I think in many of my conversations also. People are like, “Oh, but if you remove these things, look at who’s going to be affected by this.” I’m like, “Absolutely.” But there’s a distributional consequence. There’s a long-term/short-term tradeoff that you have to make. Whether you do it or not, I don’t think people even discuss it in greater detail.

RAJAGOPALAN: My bias for starting this, even though there are distributional consequences, and yes, this is the kind of awful thing economists say to each other and to econ students, but it’s true.

CHIPLUNKAR: Yes, yes.

RAJAGOPALAN: We say horrible things like, “The number of fire accidents in a factory, the optimal number is nonzero,” [laughter] or something like that. We are willing to live with it because we understand that there is a tradeoff between extreme safety regulation and extremely high and good care of maternity leave versus actually what happens on the ground and whether the firm can absorb the cost.

The reason I have a bias against the regulatory barriers, other than just the fact that they’re stupid and they make people unfree, is growth compounds. And if you can get firms to start growing early, if you can get economic growth like little green shoots in a particular labor market, that really compounds over 10, 15, 20 years in a way that if you wait for things to happen and just say, “We’ll one day reach the level of development that will fit these bad regulations,” I don’t think you get the same result. I think we underestimate how much compounding matters in poorer areas.

The Trouble with Piecemeal Reforms

CHIPLUNKAR: I completely agree with that. Just to add on to that, I think the other part—talking to bureaucrats about—it’s not that people don’t know this or appreciate this. They might be underappreciating, but they know that this is the need of the hour. That’s where I think in economics, we need more of political economy as mainstream models that can account not only—for example, much of my work, in fact. 

You’ll be able to tell me, “Oh, these are the efficiency costs of an economy.” You go to a bureaucrat, and you tell this person, “Look, this is . . ..” They’re like, “Yes, we know this. Of course, women should be working. Of course, we should have larger firms. Of course we should . . ..”

It’s not like they don’t—now, you can tag a number on it and make them realize the severity of the issue. But I think where the rubber hits the road is basically, they’re like, “OK, I take this to my minister. This is not going to be politically feasible at all.” Right? That’s a problem that we haven’t really solved in India.

RAJAGOPALAN: The other part of it is when we have tried reforms, we haven’t tried wholesale reforms.

CHIPLUNKAR: Correct.

RAJAGOPALAN: The last time we had wholesale reforms was Rakesh Mohan and group removing License Permit Raj wholesale. You remove the entire command-and-control structure for a wide part of the economy. The trouble is, if you do this piecemeal, you don’t know which the binding constraint is. So, we can adjust maternity laws a little bit. It still doesn’t improve female labor force participation because you don’t know if the binding constraint is demand-side or supply-side or for women of reproductive age or what.

Unless we reform it in a more broad-based manner, I don’t think we are going to get the kind of political benefit that previous reforms saw. I think it’s also the way we advise our politicians. We make them do tiny, narrow things because we do RCTs on them or something. That’s just not how the rubber hits the road when it’s on the ground.

CHIPLUNKAR: In fact, the Econometrica paper does this very nicely because these frameworks allow you to say, of course, we don’t observe counterfactuals, but to the extent that you believe the estimated model, you can run counterfactuals. You can say, “In a simulated world, if I were to only remove one barrier at a time, would that really generate growth?” Basically, the Econometrica paper is very clear: no.

RAJAGOPALAN: Yes. No.

CHIPLUNKAR: Right? It’s not. The compounding is huge.

RAJAGOPALAN: Exactly.

CHIPLUNKAR: You really need multiple barriers being alleviated. Now, how much of that can policy push with versus not is a separate conversation. But to the extent that one needs to think about these big-bang reforms, again, in Indian labor force and just the labor market, I think is a no-brainer. 

Now, the question is, why aren’t politicians willing to bet their money on this and say, “Why don’t we push this through?” Now, the current government has put women as an agenda, a priority agenda. So, now there is some excitement that’s been generated around this.

RAJAGOPALAN: I’m a little alarmed about that. Your research has made me more alarmed, [laughter] and I’ll explain how. We have two big policy goals: One is formalization of firms, and the other is increasing female labor force participation. Your paper shows us that sometimes these goals may not be quite compatible, especially the Andhra Pradesh work, which is when you burden a particular firm with more goals, more policy objectives which are highly targeted, or more regulation. They actually flip back into informality. That’s what alarms me about this dual-goal problem where governments say we need more formality, and that’s going to be good for women. Now, let’s push the female labor force participation agenda to force more formality. 

Does that make sense?

CHIPLUNKAR: I think it does. And I think there’s a large part—again, I think that’s where some of these economic frameworks can be helpful because you can exactly model these tensions. Right? You can at least do a first-order—try and understand how these tensions might interact from a macroeconomic standpoint. I totally agree with you. Is formal sector really the goal, especially from a gender perspective? Not really, if flexibility, if informality is what is valued because that’s the society we live in.

RAJAGOPALAN: Yes.

CHIPLUNKAR: Obviously, everybody understands these are constraints.

RAJAGOPALAN: Or because people are too poor to pay for the goods that get produced in the formal sector.

CHIPLUNKAR: Yes, but some of them can be made available cheaply. For example, data is a good example. Clothes are another good example, where economies of scale might really reduce the costs. But to the larger point, absolutely. If you have a ton of laws and come down with a hammer on them and force people to do it, firms are not going to do it. They’re going to find ways to not do it. If anything, jugaad in India is great. [chuckles]

RAJAGOPALAN: Jugaad in India is great and bad because jugaad can only get you up to 10 workers.

CHIPLUNKAR: Exactly. Exactly. No, I meant more like formal-sector firms are going to figure out jugaad around this.

RAJAGOPALAN: Oh, that also, yes.

Frictions in the Job Market

RAJAGOPALAN: So, before we get to the jugaad and the political connections and what firms do in the formal part of the economy, I still want to stick with labor because there’s anotvvher set of papers that you have, which is not so much about women and demand and supply side necessarily, or even the regulatory barriers and the regulatory cholesterol that we face. It is frictions that exist in any job market.

CHIPLUNKAR: Correct.

RAJAGOPALAN: Right? The job market, we tend to treat it as this aggregate thing, but it’s actually hugely fractionalized when you start studying the microdata. There are a lot of matching problems, and there are a lot of allocation questions. Right? Not every software engineer is the same as another software engineer, though that is one particular job where there is a vast amount of similarity and overlap. But they’re still different.

Now, I want to discuss some of these papers, and I want to get to perhaps my most favorite paper of yours. [laughter] This is titled “Who Gets the Job?” This is with Erin Kelley and Greg Lane, and this is coming out in Restat shortly? The broad thing that you are trying to understand is information and how social networks are a conduit for information flows, in particular when it comes to job opportunities.

CHIPLUNKAR: Yes.

RAJAGOPALAN: Right? You have a lovely field experiment. This is an experiment in Bombay University where you look at a particular cohort across different subfields and colleges, and you randomly vary whether the job opportunity for that particular cohort is going to be rival or nonrival. 

CHIPLUNKAR: Correct.

RAJAGOPALAN: And by rival, it means they have to compete to get that job with the rest of the cohort. Nonrival is if you get the information, you’ll get the job, right? What happens to information flows within this group?

CHIPLUNKAR: Yes.

RAJAGOPALAN: What you find is not comforting. So, first, before we get to the results, maybe you can set up how you got to this experiment and what was interesting about it, because field experiments are fascinating. Then I have lots of questions about the result.

CHIPLUNKAR: [laughs] This came back to my college days, and on the job market as a PhD student as well, which is, in many cases, the way people know about jobs is, if you ask, look at literally any survey in literally any developing countries, friends and family. And so, there’s a lot of control over who gives that information. Abhijit, Esther, Matt Jackson, all of these guys have amazing RCTs in the Karnataka area that show gossip matters, nodes matter, and so on and so forth.

RAJAGOPALAN: Abhijit’s old work on rumors and so on, right?

CHIPLUNKAR: Exactly.

RAJAGOPALAN: That’s pre-RCT work, yes.

CHIPLUNKAR: Exactly. To couple that, I think there’s this entire literature on information flow in agriculture that basically shows that, “Oh, if my neighboring farmer does something with a new technology, I’m more likely to do it.” With jobs or with labor market information, the big wedge in this information flow is the fact that information is rival. Or can be rival. If I tell you about this great job opportunity and you decide to apply for it, then I’m competing with you for the same job. That gives me incentives to withhold this information from you. That’s basically—and we started thinking about where we wanted to go do this.

RAJAGOPALAN: No, and that also has macro effects. This wasn’t just—

CHIPLUNKAR: Yes.

RAJAGOPALAN: —a cute experiment.

CHIPLUNKAR: No.

RAJAGOPALAN: Right? Because it changes the applicant pool fundamentally at the aggregate level in the labor market. 

CHIPLUNKAR: Absolutely.

RAJAGOPALAN: Sorry, keep going.

CHIPLUNKAR: No, absolutely. I think the reason we landed with college students was twofold. One was where are labor market information frictions probably the highest, where the consequences are going to be quite severe. One pool is college students, because they’re entering the labor market. It’s going to be their first job. We know from a lot of papers that this is an important phase of their careers. A, that, and B, was really to be able to work with a group of students where we could observe exactly how information was shared. Because these are college-going students. They’re in the same classroom. They know each other. Networks are very well-defined. For a lot of these reasons, we landed here.

RAJAGOPALAN: But more broadly, the information, I want to clarify, is not like a job posting. Because most people think of information as like a job posting or a classifieds ad or a LinkedIn post, whereas actually the information runs many layers deeper than that. One is about, is this within the feasibility set? The other is, is this respectable? Is this the kind of job that’s going to be good for me? Then the third layer of it is matching. Is this actually good for my skills? The fourth is rejection. Do I have a shot at it? There are layers upon layers and layers of information. 

CHIPLUNKAR: Correct.

RAJAGOPALAN: I love that you did this in Bombay University because that’s exactly the kind of stacked layers of information you can both study and parse out, right?

CHIPLUNKAR: Absolutely. Absolutely. And so, this job was not exactly a job from a firm standpoint. We were the firm, basically. The World Bank hired—they needed a bunch of interns. This was basically an internship opportunity for these students, which they valued a lot.

RAJAGOPALAN: Yes.

CHIPLUNKAR: The key thing that we wanted to vary was really try and see how much is this competition playing a role, number one. Number two, are there forces that can actually overcome competition? For example, you and I are great friends.  Even though I know we are going to be competing for the same job, the closeness of our friendship might mitigate some of these competitive concerns. 

RAJAGOPALAN: Yes.

CHIPLUNKAR: Right? And so— 

RAJAGOPALAN: Or you’re going to find out eventually anyway. If we are such close friends, I might as well tell you beforehand. [chuckles]

CHIPLUNKAR: Exactly. No, exactly.

RAJAGOPALAN: It could be both.

CHIPLUNKAR: No. Exactly. Exactly.

RAJAGOPALAN: Sorry for such a cynical view but reading your paper did not leave me optimistic about human race.

CHIPLUNKAR: No. I mean, yes . . .. Women, yes, I’m more optimistic about because you find that these—when we come to the results, we can talk more about the gender angle of this. But that’s where we started out. 

RAJAGOPALAN: Yes.

CHIPLUNKAR: The other wrinkle we wanted to add on this was the following: If this is going to affect the pool of applicants a firm is going to get, because the smartest guy in the room never knows about this job. It really, at the fundamental level, depends on the correlation between your ability and ability to gain information.

RAJAGOPALAN: Yes.

CHIPLUNKAR: Right? If these are positively correlated, which means that the smarter people also are more connected in their social networks, then this is not a problem because these are smarter people. They’re also . . .. But if you know a lot of Bollywood and just reality, [chuckles] the nerds are probably not the most connected in their social networks.

Rivalry, Information Flow, and the Nerd Effect

RAJAGOPALAN: But there’s a second element to it. What you find is something much better, which is the nerds are deliberately kept out of it.

CHIPLUNKAR: Exactly. No, but that was the surprising and perhaps exposed an unsurprising thing, which is what we randomize is basically who—so, after mapping out the social network and getting to know who your friends are in the classroom, so on and so forth. Really, what we randomize every week is who in your class gets information about this internship opportunity. Then we really very carefully track how that information flows across students within the classroom.

RAJAGOPALAN: Yes. It’s amazing.

CHIPLUNKAR: There is what you were saying. The one thing we find is that the nerds don’t get to know about the—are less likely, let’s say, to know about this job opportunity.

RAJAGOPALAN: Yes, so here, the nerds are people who are the higher-ability candidates. It’s not just a pejorative we’re using for the studious kids, right? [laughter] If someone gets information about a rival job where they have to compete, first, they’re less likely to share it overall.

CHIPLUNKAR: Yes.

RAJAGOPALAN: But within conditional upon sharing, they’re more likely to share it with people who are lower on the ability scale than them, than those higher on the ability scale than them, which basically means the nerds are left out. 

CHIPLUNKAR: Exactly. 

RAJAGOPALAN: That’s where we’re going with this.

CHIPLUNKAR: Yes, absolutely. I was a nerd, [chuckles] so in that sense, I feel like there’s much to be said— 

RAJAGOPALAN: We know.

CHIPLUNKAR: There’s much to be said about [laughter] some of these things. But jokes apart, I think that’s exactly right. I think because we know exactly the social network, we can actually do an analysis at the pair level, like the pair of friends, and really look at, on a one-to-one basis, who gets the information, and not at exactly this. 

Now, the surprising thing is that making a job nonrival—now, how do we do this is basically to say, look, if I tell Shruti, nonrival just means if you get this information, you’ve already got the job.

RAJAGOPALAN: Yes.

CHIPLUNKAR: Right?

RAJAGOPALAN: So, everyone who has the information has the job. They don’t have to compete.

CHIPLUNKAR: Exactly. That removes your incentive of just purely competing for this job, but we do encourage you to still share it, as we do in the rival case, encourage this sharing between your classmates. We find this “nerd gap,” let’s call it, disappears completely. It’s nothing to do with network. It’s purely competition. It’s just purely about the fact that I know this guy is smart. If I apply and this person applies, he or she’s going to get it.

RAJAGOPALAN: Which also brings me back to, if I had to scale up your experiment, [laughter] demand side really matters. 

CHIPLUNKAR: Yes.

RAJAGOPALAN: The number of jobs available overall in an economy really, really, really matters, right?

CHIPLUNKAR: One hundred percent. Absolutely. Yes.

CHIPLUNKAR: Nothing to add to that. Yes.

RAJAGOPALAN: It’s the same story over and over again. You’ve studied this using different methods. You’ve studied this in different sectors. 

It’s an incredible story. Now, one of the more interesting things that you found is this nerd effect. First, men are way more competitive than women. Men share less information. There is a bigger effect when it comes to the nerd effect. Men like to leave men of higher ability out. But these things disappear with women.

CHIPLUNKAR: Absolutely.

RAJAGOPALAN: What’s going on?

CHIPLUNKAR: I don’t know. You tell me. You’re the woman in our conversation. [laughs]

RAJAGOPALAN: I have grown up entirely in male networks in economics. [laughter] I don’t know any more if I have the same . . .. Women share information. We understand this. But I think one thing that might be going on, and this may have some supply-side factors related to it, is if two of my close friends also get the same job, my parents are more likely to allow me to go do it and things like that. Or it’ll be more fun, or we can travel together. 

You see this in a lot of field experiments which other people have done. I think that might be one part of it where sharing information and actually applying to the job together, interviewing together, or even doing the job together is “safer,” whether it’s in terms of social norms or everything else.

CHIPLUNKAR: Yes. No, in fact, Smit Gade has this really nice paper— 

RAJAGOPALAN: Exactly. That’s the one I was thinking about.

CHIPLUNKAR: —that looks at exactly this issue in terms of traveling together for a job. Fortunately for us, ours was online. There was no traveling or interviewing or anything like that that was the issue. But I agree. [chuckles] Having close male friends is a disaster if you’re looking for labor market information because they’re competitive. They don’t—

RAJAGOPALAN: Having close male friends is a disaster if they’re in the same field as yours and competing for the same pool of jobs.

CHIPLUNKAR: That’s a well-caveated statement. [laughter] Let’s put it that way. Yes, but that’s exactly right if you’re looking for jobs. We’ve talked to a few . . .. There’s no clear answer. It’s some of what you say. It’s some of the fact that, “Oh, look, I’m actually close friends with”—they mean, “She’s a close friend of mine. Why would I not tell her this information?”

RAJAGOPALAN: Yes.

CHIPLUNKAR: Which is pure altruism as opposed to, like, “Oh, but this is only a one-week opportunity. It’s fine.” At the end of the day, there’s some reciprocity to it. “She has helped me out in the past, so I’m going to do the same. . ..” There are a multitude of reasons that, at least from these qualitative conversations, we learn about. Now why men don’t value it, I have no idea.

RAJAGOPALAN: My hunch is just the Indian job market in terms of really good jobs.

CHIPLUNKAR: But these are internships.

RAJAGOPALAN: I know, but there is a mindset problem, right?

CHIPLUNKAR: Yes.

RAJAGOPALAN: The mindset comes from what is going on more broadly.

CHIPLUNKAR: Yes.

RAJAGOPALAN: We have too few seats, whether it’s at IIT, whether it’s at Bombay University, or whether it is Infosys hiring or anywhere else. Everything is a race.

CHIPLUNKAR: Agreed.

RAJAGOPALAN: You can’t blame them in thinking at a very local level that this is a zero-sum game.

CHIPLUNKAR: Yes.No, absolutely. No, absolutely.

RAJAGOPALAN: There’s more pressure on men. No question.

CHIPLUNKAR: Yes, absolutely. As I said, there’s a combination of all of these things. But the surprising thing is, again, when you make jobs nonrival, none of this matters.

RAJAGOPALAN: None of this matters. Exactly.

CHIPLUNKAR: None of this matters.

RAJAGOPALAN: Right? So, that’s the thing. If there are enough jobs, men would also behave in a sane way.

CHIPLUNKAR: Exactly.

RAJAGOPALAN: Is this why all your coauthors are women, because you’ve done this study and you know exactly what you’re doing? [laughter]

CHIPLUNKAR: No, it’s a happy coincidence.

RAJAGOPALAN: There are lots. [laughter] You’ve worked with a lot of different people from a lot of different institutions and countries.

CHIPLUNKAR: Yes.

RAJAGOPALAN: Whether it’s working on female labor force participation or not, you have an extraordinary number, above average for sure, of female coauthors.

CHIPLUNKAR: Thank you, thank you. I’m just grateful they want to work with me. [laughs]

RAJAGOPALAN: I’m wondering if this study has something to do with that. You’re like, “The men are just going to cut me down [laughter] because of the nerd effect.”

CHIPLUNKAR: That’s true.

RAJAGOPALAN: You’re like, “The women coauthors are at least—” [laughter]

CHIPLUNKAR: I’ve really internalized my own research to a degree that I don’t know about it. [laughter]

RAJAGOPALAN: “At least my female coauthors will let me know the opportunities.”

CHIPLUNKAR: That’s true. [laughter] No, I’m just very happy. I’m grateful that they want to work with me. [laughter] But, someone else did point to me a few weeks back that, “Oh, you’ve got a lot of amazing female coauthors.”

RAJAGOPALAN: Did they land on the same reason?

CHIPLUNKAR: No, this is the first time I’m hearing about this reason. It’s probably not them. It’s me who has internalized my own research. [chuckles]

RAJAGOPALAN: Micro foundations.

CHIPLUNKAR: Yes, exactly. [laughter] Exactly. No, but, sorry, but just to continue that conversation, I think that the other thing that we find with this, which I find fascinating, is that this is not queued just from a college information stick.

RAJAGOPALAN: Exactly.

CHIPLUNKAR: Because we actually give them the jobs. We actually monitor their entire performance on the job. It really does affect outcomes. The pool shrinks in terms of the quality, the performance of the job. The last wrinkle that we add to this, which I find interesting, is we say, “OK, if I’m a firm, what am I going to do to attract talent? Well, I’m going to probably offer higher wages.”

RAJAGOPALAN: Higher wages. So, you double the wages.

CHIPLUNKAR: So, we actually offer double the wages.

RAJAGOPALAN: Yes.

CHIPLUNKAR: And so, we randomize who gets these. But you know what? If you make jobs more attractive, competition goes up. 

RAJAGOPALAN: Exactly.

CHIPLUNKAR: Right?

RAJAGOPALAN: So, they’re even less likely to share them. Yes.

CHIPLUNKAR: It exacerbates some of these effects. It overcomes some of these constraints in terms of just information flow because people are now more interested more generally in these jobs. But it exacerbates these competitive effects. The traditional ways in which one would say, “Oh, high-quality jobs, just offer higher wages and people will come.” Yes, people will come, but there is selection in terms of—

RAJAGOPALAN: But who is coming?

CHIPLUNKAR: —who is coming. I think this is nice to be able to study that very cleanly from an experimental standpoint.

RAJAGOPALAN: Yes, and also matters who your node is within an information network. Right? That’s the biggest thing. Your nodes better be really high quality and more likely women.

CHIPLUNKAR: No, exactly. Absolutely. There was a lot to learn from it from how information, from a policy standpoint should—for example, online job portals or these new technologies, are they really overcoming some of these information barriers because now people are finding it? I think that’s an interesting question to study.

Challenges to Placement and Retention

RAJAGOPALAN: Yes. OK. I want to stick with the information thing. This online portal thing is super interesting because this gets me to your next paper, which is actually a much older paper. This is your paper with Abhijit Banerjee also studying labor market frictions. This is very concrete, old-school matching exercise. 

CHIPLUNKAR: Yes.

RAJAGOPALAN: Here you also have a middleman. You have people looking for a job. You have firms hiring. Then you have a middleman, which is a placement coordinator or a hiring manager or something like that.

Now, there is a question not just of information flow between the appropriate opportunity from the point of view of the applicant. It’s also for the firm, not just in aggregate terms, but for this particular applicant. Now the placement coordinator and what precise information they have about the particular candidate and if they can actually match them. Because matching is really the problem we need to solve, not just hiring random people, right?

CHIPLUNKAR: Yes.

RAJAGOPALAN: What you guys do is when you ask hiring managers or placement coordinators or HR to figure out what are the preferences of individual candidates., they actually do a terrible job of figuring that out [laughter] even though that’s their primary role.

CHIPLUNKAR: Absolutely.

RAJAGOPALAN: So, first of all, what is the information friction going on? Because on paper, a lot of these jobs—and you’re looking at a standard degree kind of thing.  You know that a skill from this person in this college is going to be about at this level. This is really about: Do they wish to work this job? Do they wish to work these hours? Do they wish to travel so much? What is it about all this information which is not as explicit as maybe the skill and the degree signal, and how do we overcome it?

CHIPLUNKAR: Yes.I think the placement officers was an interesting exercise because they play a huge role, as you were saying, in terms of putting— 

RAJAGOPALAN: I didn’t know. I thought they were doing nothing.

CHIPLUNKAR: No. [laughter] One would argue effectively, [laughter] but to the extent that they’re incentivized—

RAJAGOPALAN: They could do a lot.

CHIPLUNKAR: They could do a lot. 

RAJAGOPALAN: Yes.

CHIPLUNKAR: They’re incentivized to do a lot. I think the key part was trying to figure out preferences are multidimensional. The thing that they get right, unsurprisingly, is that higher-paid jobs are preferred more. Duh. That’s not very hard to figure out. Where do they get it wrong is basically all the nonmonetary dimensions.

How much do you prefer characteristic x of a particular job, and especially whether that’s with respect to distance, whether that’s with respect to the work hours, how active the job is. Do you really have to go on a delivery scooter halfway across Lucknow versus a cushy office job where you just type on a computer all day in an air-conditioned office? These kinds of amenities, I think people—there are two things. One is that preferences are varied, even amongst individuals.

RAJAGOPALAN: And we assume as economists that they are given to us—

CHIPLUNKAR: Correct.

RAJAGOPALAN: —from the people. We’re not distorting their preferences.

CHIPLUNKAR: Correct. Exactly. Students themselves have varied preferences across these jobs. Whatwe learned from that experiment was basically, yes. One part of it was there’s information friction, especially on the nonmonetary dimension, and so that begs the obvious question. Suppose we know students’ preferences because we measured them very carefully. What would happen if we just give preferences to these placement officers? Fortunately, we do find they do pay attention to what we were telling them.

RAJAGOPALAN: They won’t go looking for the information, but once you give them the information, they’ll act on it.

CHIPLUNKAR: Exactly. They did act on it. People were placed in jobs that—if you take a locus of how far from your most preferred job—because that means it’s multidimensional, so we have an index and all that, but they’re closer to their preferred index, but they don’t actually stick to their jobs. There were two things that were surprising. One part was the fact that, why don’t placement officers weight this to begin with?

I think when we talked to a lot of these placement officers, one part was they had no clue. They accepted the fact that, “Sure, you might have these.” One part of it, coming back to your earlier point, is that you might have these preferences. We are never going to be able to find the jobs that fit these preferences because the way many of these placements actually work is bulk placements. There’s one employer who wants to hire 40 people, and so I don’t give two hoots about your preferences. You’re going to get what you get.

RAJAGOPALAN: But there are margins on which you could still make it work is what I got from the paper.

CHIPLUNKAR: A few, a few. But the primary thing on what placement officers told us was the reason they underinvest in this kind of exploration is because in many cases their jobs are just bulk hiring. Or in Lucknow for this kind of—this was in UP, and so we had Lucknow, Delhi, Kanpur. These were the main placement areas. They basically were constrained with the kinds of options that they were getting to place their students in, which is why they underinvested in this. The second part is that we actually followed these guys over time, the students themselves, once they were— 

RAJAGOPALAN: Exactly.

CHIPLUNKAR: Technically if you were placed in these jobs, you should love it and you should be happy about this. Literally six or nine months later, I would say very, very, very few people are actually sticking to these jobs.

RAJAGOPALAN: Which goes back to your earlier point of they don’t want to invest in skilling because this churn is real.  And the churn comes from mismatch.

CHIPLUNKAR: Correct. But this was after reducing mismatch.You still don’t have retention, right? That was fascinating. There were two or three reasons that came up. I’ll go from the most logical story to the—the one thing people told us was they were trying for public exams.

RAJAGOPALAN: Yes.

CHIPLUNKAR: Government exams.

RAJAGOPALAN: Kunal Mangal and all his research—

CHIPLUNKAR: Niharika and Kunal had this entire research

RAJAGOPALAN: —has shown us how horribly that distorts the labor market.

CHIPLUNKAR: Exactly. Again, these were 21-year-olds. This was UP again. Again, context might differ and this might have worked very differently in other contexts. But at least in this context, the one thing they wanted to do was to appear for government exams and they were not interested in some random delivery job, which begs the question, why did they want to take the vocational training course to begin with? I have some theories on that that I’m happy to talk about later. But that was reason number one.

Reason number two, especially, and this is where gender again comes in very interestingly: Women are actually more likely to stick to these jobs than men. The reason was very different. Men, it was the first time that they were living out of their household. The Raja Babu syndrome [chuckles] was shattered. Was shattered. So, they would rather come back to their village and just hang out in their homes than actually take these employment opportunities in these cities that were nearby.

RAJAGOPALAN: The cities are cruel.

CHIPLUNKAR: They are.

RAJAGOPALAN: In all fairness.

CHIPLUNKAR: We asked a lot about who they were living with. They were usually sharing apartments with four or five people. They had to cook their own food.

RAJAGOPALAN: That’s pretty awful.

CHIPLUNKAR: Work hours were grueling. It was very difficult for them to adjust to this life in a very short time.

RAJAGOPALAN: So, the aspiration they have of what kind of job they want and the kind of job they get in this particular job market, there’s a huge gap. So, suddenly, you feel a little bit better about applying to the government job lottery.

CHIPLUNKAR: Exactly. For many of the men, this was what was going on in terms of trying to, government jobs, a combination of just the labor—the job itself actually wasn’t—we asked them this distinction between the quality of life versus the work itself. People expected the work to be what it is, reassuringly, because they were in a vocational training program. They’d seen some of this, at least in theory. That wasn't where the update was. 

Most of the update in terms of their priors came from just living in a city and just trying to understand that “in a 20,000-rupee job, I’m probably not going to get the quality of life that I envision myself doing.”

RAJAGOPALAN: Which is heartbreaking.

CHIPLUNKAR: Which is heartbreaking. And then for women, on the other hand, the story—we were surprised and cautiously optimistic that women are actually sticking to these jobs way more. But for them, it was the first time they were allowed to step out of their household, and they were relishing it.

RAJAGOPALAN: Yes.

CHIPLUNKAR: Right? It was—

RAJAGOPALAN: And it probably means that they’ve already overcome the previous barrier. They’ve figured out their PG situation or they’re living with a family member or an extended family member or something.

CHIPLUNKAR: Exactly.Of course, there were constraints in terms of like, for example, women were more likely to go to Lucknow than Delhi, for example. There were these obvious constraints. But to the extent that they were able to do that, it was the first time that they actually relished their own freedom. In fact, we find that women were more likely to send a larger fraction of their salary back home as opposed to the men. One of it was, again, an income story, that they were earning 25,000 rupees in these jobs and they used to send their regular income back home. It was almost like, very unfortunately, the price that they put on the fact that they could live in these cities and work in these cities. The big drop, of course, that we still see after that is just marriage.

RAJAGOPALAN: Marriage, yes. 

CHIPLUNKAR: The minute that they have these talks—and these are qualitative conversations we’ve had with a few—we didn’t follow them large because no one stuck to their jobs. We didn’t have a much longer follow-up on this, but to a few people who were able to call and talk to them, marriage was still the biggest barrier. The minute the family started talking about marriage, they quit their jobs. They were back home.

RAJAGOPALAN: Yes. I’m thinking about this, given what you just said. When I was reading the paper, I thought it was exactly the way you had set it up, which is, it is the placement officer who doesn’t have the information about all this. Now I’m wondering if one big problem is the individual applicants are either not happy to reveal their preference or they don’t even know their preference because they are not that familiar with this kind of a job market, especially in a big city, given what’s happening with the structural transformation. 

How much of that is playing a role in this? Women are just clearer about what their preferences are because their life has been mapped out for them. Families just tell them, “You’re allowed to study, you’re allowed to work for a few years, then you’re going to get married by this age, and this is it.”

CHIPLUNKAR: To your first part of how much do we believe these preferences? Within the experiment, we had a sub-experiment where we actually incentivized individuals in the following way: We told them, “Look, you can take this here.” The way we actually got these preferences were real-world jobs that were offered to previous cohorts. We mapped them out more carefully to look at what dimensions were varying in these jobs, and we presented students with a list of about eight to 10 jobs.

We were like, “Why don’t you just go and rank them in terms of order of preferences?” We didn’t want to make explicit certain dimensions that we wanted to prime them on, etc. These were real-world job offers, so these weren’t hypothetical scenarios of ideal offers and so on and so forth. 

The way we incentivized them was to say, “Look, here are 10 jobs,” and it was factually true, “which had been offered to previous cohorts. Your preference is going to matter because we’re going to ask the placement officer to search in these dimensions depending on what you have.” We increased the stakes of randomly just ranking these job offers. We find no difference in terms of whether the incentivized and the non-incentivized individuals in terms of the rankings that we get on these 10 jobs actually don’t differ.

RAJAGOPALAN: Do they worry that they won’t get the job at all if there is too much of a strong preference for one kind of thing versus another?

CHIPLUNKAR: Potentially, that could be going on. But again, these were jobs that were offered to previous cohorts, so they were jobs that had come by in previous batches. It could be that if everybody just says “job one,” then we’re all going to compete for job one, and so maybe—

RAJAGOPALAN: No, not so much that. It’s like, “Oh, these people are going to think I’m picky and I have these preferences.” This is me speaking as a woman. Women thinking if we express too much about what our strong preferences are, no one’s going to hire us.

CHIPLUNKAR: Potentially, but again, this wasn’t a preference elicitation in terms of we are going to tell employers this is what you did.

RAJAGOPALAN: Exactly, this is just “tell us.”

CHIPLUNKAR: This is more about tell us because we want to match you to jobs that you actually want to work in. It was set up in that way, but we find no real difference in any of this. It’s not really the fact that people weren’t willing to—and again, if you talk to these students, they actually want jobs that they want. They’re really excited about this coming from a vocational training program. They want to go and experience the city and they want to really live in these labor markets. Much of it, I think, was the update afterwards.

In fact, I remember the two key things Abhijit was excited about in terms of trying to unpack further was should therefore we have internship programs in the college where people actually go to these labor markets that they’re going to work in and actually have firms hire on a temporary basis.

RAJAGOPALAN: Which is how we all did it.

CHIPLUNKAR: Exactly.

RAJAGOPALAN: We all figured out what our preferences are and what our skill-matched set is based on internships.

CHIPLUNKAR: Exactly. That was one. The second, is then trying to look at it more from how should the placement process itself work for these vocational training programs to be able to—because clearly this is not a case where retention meant you were going to get better jobs in Lucknow. This was your first—by the way, to be clear, there were some people who did very well.

RAJAGOPALAN: Yes, of course.

CHIPLUNKAR: They really went to these jobs. They loved it. They found better opportunities for themselves. There are success stories. I don’t want to under-emphasize the success story part of this as well, but on average, that wasn't the story—

RAJAGOPALAN: That wasn’t the story.

CHIPLUNKAR: —that was coming out of it. I’m just focusing on the average, as opposed to there were a few students who did really well on this as well in terms of just breaking through the Lucknow market and finding themselves better opportunities.

The second part is what should we do about the public sector in terms of how it is either distorting—in this case, clearly distorting labor market opportunities. Niharika and Kunal are digging into this more deeply, but these were the two things on how— and then you couple this with the social network experiment in terms of if there are these few well-sought-after jobs, then it really matters, it gives you power over who holds that information and how that is conveyed to other people.

On the Potential of Job Portals and Leapfrogging

Interestingly, let me add one quick thing, which is in some follow-up work that is paused right now, but we hope to pick it up at some point, was I was really excited to see whether job portals specifically can break this barrier. If this is all about the fact that, look, the placement officer just doesn’t know who in Lucknow, and I just have to go and contact Domino’s, who wants to hire 50 delivery people—job portals can solve this problem because you know exactly how the Lucknow labor market can look like through jobs that are posted on these job portals.

We did a small pilot in Jharkhand, and basically, what we find is the following: We find that actually placement officers do put in more effort. To be clear, what did we try doing? We basically train students on how to search on these job portals.

The experiment was, we just show up, we give them a two-day training on a job portal, we help them create their profiles, we’ll tell them how to filter, what to look for in jobs, and so on and so forth. Then three months later, we follow them up and we ask them, “OK, how did you get your job?”

Placements through online job portals is very, very low. It’s not that people are—and they apply to jobs. They search for jobs. They apply to jobs. We ask them, “How did you find your job?” They were like, “Oh, our placement officer, our placement manager basically put . . .”

RAJAGOPALAN: Networks.

CHIPLUNKAR: Networks matter. It could be the fact. Now I think the reason is positive because we are trying to unpack what could be driving this. There are two channels. One channel is, “I’ve just increased your outside option as a student, so you’re no longer at my mercy as the placement officer. Now I have to work hard to earn my incentive because now you can just go and find a job.”

RAJAGOPALAN: Oh, nice. I like that story a lot.

CHIPLUNKAR: The second option is basically, “Purely giving you information doesn’t matter, and my real monopoly is not on the information. It’s about the connection to the manager at Domino’s who’s going to hire you. Therefore, you might do everything you want on the job portal, but you’re not going to get that job. But me knowing a manager at Domino’s . . ..will” But these are the opposite end of the spectrum in terms of what it means.

RAJAGOPALAN: Seeing what your other research has shown, it’s probably the latter.

CHIPLUNKAR: No. Exactly. Now we are trying to dig into that a little bit more carefully and trying to understand whether it’s a monopoly on information or connections as opposed to an outside option that’s going up for you, which is why I’m working hard to earn this incentive.

RAJAGOPALAN: You’ve done other work on information. This is literally your paper again with Penny Goldberg on the digital age and places leapfrogging in terms of technology by getting 2G, 3G networks, and using these phones to have more information. 

The information story is not quite as compelling for the structural transformation and also matching the skills with the jobs. Am I reading too much into your paper with Penny Goldberg on leapfrogging and how that result is really mixed? You find a few more opportunities through information, but they’re not great opportunities, and they’re certainly not the kinds that you’re talking about, which is, “Through a placement officer, I can get matched to something higher.”

CHIPLUNKAR: Yes, no, absolutely. I think that’s a better summary of the paper. [laughter]

RAJAGOPALAN: Why don’t you tell me about the paper? [laughter]

CHIPLUNKAR: No, no, it’s absolutely right. I think— 

RAJAGOPALAN: Sorry, I’ve read all your work all together much more recently than you have, so this tends to happen.

CHIPLUNKAR: As I’ve told you before we started this podcast, I think the way you’re threading through all of this work, I would have never been able to do that. I’m really grateful that you are able to see and connect all of this, which you’re obviously fantastic at doing that. You’re better at mirroring my own research to me, which I’m now discovering is great in terms of learning a lot of things about the labor market.

RAJAGOPALAN: You read Marginal Revolution and you know Tyler and Alex. One of the things I learned from them very early on, maybe this was 15, 20 years ago, is don’t look at papers, but look at literatures. I took this very seriously very early on. I actually like reading papers in clusters. I find that reading them in clusters actually tells you a lot more about what’s going on.

CHIPLUNKAR: Fascinating.

RAJAGOPALAN: And what we do as economists is pattern recognition and putting the pieces together. Also, most people are doing partial equilibrium stuff when it comes to these kinds of big questions. It’s just easier for me now, I think having done this for many, many years, to just read papers as literature. Yours has, to me, a very, very clear throughline.

CHIPLUNKAR: It’s fantastic.

RAJAGOPALAN: But going back to your paper with Penny, this is really so relevant for India because you’re looking at this leapfrogging technology and putting a phone in the hands of people. One, does it matter for structural transformation, and what are the differences between men and women, which is back to the big theme?

CHIPLUNKAR: No, exactly. Does it matter for structural transformation? No, it’s what we find. But there is a nuance.

RAJAGOPALAN: It matters for jobs.

CHIPLUNKAR: Yes. To be clear, what is this paper? This is basically putting together data on 3G coverage. These are smartphones. Improvement of that over a bunch of countries and over time. Think about this more as like, “I live in a particular region, and this region now has access to 3G. What do I do with that new technology that comes in?” The good thing is that we have data also on 2G, which is the not smartphones. 

RAJAGOPALAN: Barely phones.

CHIPLUNKAR: Dumb phones, as someone told me in a talk. Because they’re not smartphones, they’re dumb phones. That’s not so much about cell phone access as much as it is about internet on the smartphone.

RAJAGOPALAN: Yes, so it’s actually information and not networks, which is the important distinction.

CHIPLUNKAR: Or communication. It could be WhatsApp.

RAJAGOPALAN: Exactly.

CHIPLUNKAR: It could be all these online job brokers.

RAJAGOPALAN: Whereas with the older version of the phones, we were just talking to people we knew, effectively.

CHIPLUNKAR: Exactly.Absolutely. What do we find? We basically find two things. One is that employment goes up. People are more likely to be employed. People are more likely to work in wage jobs. Of course, because this is cross-country, we’re very limited in terms of the depth that we can go in. But the patterns seem to be very, very broadly consistent across countries. Now there a is literature also country-specific, and people are finding and echoing some of these findings again and again, which is basically the fact that people are more likely to be employed. They’re more likely to transition from unpaid employment into paid-wage jobs.

RAJAGOPALAN: Here, just to caveat, unpaid employment is basically jobs around the house, you’d be doing work on the farm, which is technically unpaid, but it is employment.

CHIPLUNKAR: Absolutely. Exactly. So, it’s household work.

RAJAGOPALAN: Exactly.That matters a lot for poorer countries and poorer families, which are at one end, not yet into the structural transformation pipeline, which is why it matters so much.

CHIPLUNKAR: Yes. In fact, and it matters for women because that’s where we find the biggest gender divergence. What we find in terms of the gender implications of this is that why does it not affect structural transformation? Because men are actually moving out of these unpaid household-type of jobs into either wage-paid jobs or many of them actually also into owner-operated enterprises, so self-employment.

Women, on the other hand, female labor force participation is going up. More women entering the workforce, but basically, they’re working on the job ladder that the men have just vacated. Basically, they’re the ones filling up on these farm opportunities. Some of them do go into wage jobs in the service sector.

There’s, again, a spectrum of doing that, but the large picture seems to be—everyone in that sense is climbing up the job ladder. It’s just not gender-neutral in terms of how that progress is being made. On net, if you look at average, this is great because on average, everyone’s better off.

RAJAGOPALAN: Exactly.The result you find is that a 10 percentage point increase in 3G coverage raises female labor force participation by about 4.9 percentage points.

CHIPLUNKAR: Exactly.

RAJAGOPALAN: Which is not as much as the men, but it’s nothing to sneeze at.

CHIPLUNKAR: No, absolutely. Absolutely. But again, it matters where these women are going and working in the workforce. Now, we actually have a follow-up work that we’re just trying to wrap up soon in Mexico. We’ve taken one example where we have great data. We have a good way of identifying these causal effects. Then we can actually dig more into formal sector versus informal sector, earnings versus not, and so on and so forth.

One thing that I’m fairly sure will survive all the subsequent analysis that’s going to happen on the paper before we can release it publicly is that men are actually moving into better-paying jobs. Because there is an entry of women on the extensive margin, also an entry into unpaid jobs, on average, earnings of women are not actually going up. We’re trying to unpack some of those results. But it’s fascinating, again, in terms of how technology can empower certain things versus the inequalities that might still persist despite this technology.

Again, with Mexico, it’s really nice because there is time-use data on people just being asked, “How do you use your smartphones?” Again, we find some gender disparity in terms of information, communication. Fortunately for us, everybody uses it for entertainment, which is great. [chuckles] Also, for example, information communication channels matter more for women than they matter for men.

Defining the Goalposts

RAJAGOPALAN: No, and this, again, I don’t mean to keep harping on this, but again, this study shows so much that if you don’t figure out or crack your labor-demand puzzle relaxing the supply-side constraint for women is only going to reduce their wages.

CHIPLUNKAR: Yes, absolutely.

RAJAGOPALAN: Right? This is, again, something I see over and over again in what you’re doing. You’re looking at labor supply, labor demand, all kinds of different factors. But if you don’t solve labor demand—I’m not saying you’re doing women a disservice; I think having a job, even if it’s slightly less paid, is better than having no job as far as female labor force participation is concerned, especially in the long run for economic growth. But you are going to suppress their wages. [chuckles]

CHIPLUNKAR: No, absolutely. I think the Econometrica paper makes that point very cleanly. If you only reduce in a counterfactual simulation where you only—suppose in the best-case scenario, we eliminate all labor supply barriers. You go from over 24 percentage point, female labor force participation all the way up to 70 percent. But if there are no jobs, because all these demand-side constraints exist—

RAJAGOPALAN: Then they’re going to be treated horribly and paid very little.

CHIPLUNKAR: Yes. That’s a natural extension of this to make. Again, there are different goalposts. I think in much of the conversations that I’ve read/interacted with, I think people have their own goalpost on what welfare means. It’s always such a hard thing. Economics takes one stand on it, but people more generally, what do you mean? Is it about, as you were saying—there’s a really nice paper actually with my coauthor, Erin Kelley, and others have in Bangladesh on the psychological value of work. Which basically shows that just—and these are refugees, and so it’s a different context and a different place of the world—but basically showing that just the fact that people have a job really matters to them from a mental perspective and so on and so forth.

RAJAGOPALAN: We’re seeing it in the developed world with all the AI conversation. 

CHIPLUNKAR: Exactly, exactly.

RAJAGOPALAN: People are like, “No, we need to work.” Just abundance and productivity not going to hack it.

CHIPLUNKAR: Absolutely. Again, I think the goalpost is what is—there’s no reason to pick one over the other, but I think sticking to one and defining that, and then looking at and evaluating a policy based on that, I think is much more constructive, at least in terms of my thinking. Because then you can understand exactly the tradeoffs with impacting whatever measure of welfare or job or earnings. It speaks to what you were saying. Sure, you’ll get more women to work, and maybe that’s what you’re gearing for. But if jobs don’t go up, then wages are going to go down. There’s only so much that you can say about that.

RAJAGOPALAN: And you’re complicating things for them.

CHIPLUNKAR: And you have to be fine with that, right? Coming back to why we need more frameworks and why we need to take them more seriously, usually, it’s like throwing the baby out of the bathwater because people are like, “Oh, this is a model. What do we learn from this, and why should we care about this? This is not the world.” True, but also I think it gives you these insights that are very hard to get unless you have really rich, long-term data from a RCT perspective that you can collect.

RAJAGOPALAN: Google Maps is a model, [laughter] and I learn a lot from it without having to walk every street. There is something to be said [laughter] for well-put-together models from actual, real empirical work. This was such a pleasure. Actually, I feel bad, Gaurav, or maybe good. We’ve covered only about half your papers, [laughter] so you have to come back and talk to us about all the political economy work, which I’m even more excited about. This is still George Mason University, and we love to talk about rent-seeking and political corruption. 

CHIPLUNKAR: Yes.

RAJAGOPALAN: Hopefully, you will come back. But this was such a pleasure. Really, kudos to you for just looking at the labor force participation problem in different countries, different sectors, different income levels, different methods. You’re really looking at it from many, many different lenses, and I got a lot out of it. Thank you for doing this.

CHIPLUNKAR: Pleasure has been all mine, Shruti. Thank you for having me.