YT YouTube2Text

YouTube2Text → Transcripts → Supplemental lecture: Migration and risk

Supplemental lecture: Migration and risk — Transcript

by CEPR & VideoVox Economics · 19,595 words · 2,864 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:05Okay, welcome back. Um, we're very
0:08excited uh to continue uh this uh
0:11course, STEG's virtual course on key
0:14concepts in macro development. Um, today
0:17we have uh Mushik Mubarak uh who will be
0:21talking about risk and uh migration. So,
0:25we're thankful for uh him and uh as I've
0:28mentioned before, everyone who's
0:30teaching this course is really doing
0:31this out of the kindness of their heart.
0:33Uh they're not getting uh paid big bucks
0:35for this or anything like that. Um that
0:38said, we are very thankful uh for uh the
0:42United Kingdom's Foreign Commonwealth
0:44and Development Office uh for funding um
0:48this course which is a part of STEG, the
0:51program on structural transformation and
0:53economic growth. uh we will be issuing
0:55calls actually uh later in the month uh
0:58for research. Um if you're new to the
1:01course uh all the course materials are
1:04on the website uh stag.cbr.org
1:08uh including uh mushik's um slides for
1:11today if you want to follow along. Uh
1:13the recording will also be there and and
1:15available on YouTube. Um the rules again
1:19are um if you're in the front row of
1:21students uh please mute your microphone
1:24but feel free to uh interrupt with any
1:27questions you might have as Mushvik uh
1:29uh teaches especially clarifying
1:32questions. It'll be helpful for others.
1:34Uh Mushik says he may actually ask you
1:36guys some questions. So be ready and pay
1:38attention. Um for those of you who are
1:40joining uh as um through the webinar uh
1:45the Q&A is open. Uh Doug and I are here.
1:48We will do our best. Oh, Mushik, you
1:50have a TA. Is your TA here today or not?
1:53Um yeah. So I I'll introduce
1:56I guess
1:58Okay. It's Is it okay? I'll let you
2:01introduce. Um we'll all be trying to
2:03answer questions and then um I guess
2:06it's Tuesday will be the TA session. Is
2:09that right?
2:11No, it's different this week because
2:12there's two this week. So the I'll have
2:14the TA session on Monday. on Monday.
2:17[clears throat]
2:17Okay. At 3 p.m. London time.
2:20Okay. Um, wonderful. And, uh, and then
2:24again tomorrow we have Ben Mole who will
2:26be giving a lecture. Um, so, uh, let me
2:30hand it over to Mushik now. Thank you.
2:32Okay.
2:33Hey, thank you, uh, Joe and Doug for
2:35organizing. This is a brilliant idea.
2:37Um, I it's not out of the goodness of my
2:39heart. I I appreciate being here and I
2:41appreciate being being asked to do this.
2:43Uh it's always a pleasure to interact
2:45with students all over the world and um
2:47you know um you know I'm going to share
2:50my screen first and I'm going to try and
2:53put the the chat window I'll just keep
2:57it open in front of me. Um actually, you
3:00know, one thing I'd love for you guys to
3:02do, those those of you who are in the
3:03front row, so you can probably chat um
3:06uh you want to just like um like write
3:09down what um maybe just just a couple of
3:11words on what country you're logging in
3:13from and uh where what program you're
3:15in? Um so, you know, it'll scroll to the
3:18chat. I I'll get to see um and yeah, let
3:22me introduce um a couple of people. So,
3:25first, uh Nate Barker, who just unmuted
3:27and spoke. He's a PhD student uh in uh
3:30an advanced PhD student in Yale's
3:32department of economics and he's taken
3:35the PhD development class with me along
3:37with you know people like Mark Rosenwick
3:39who also has made a lot of u uh
3:42contributions to this literature we'll
3:43talk about today um so so I encourage
3:46you to attend the TA session right uh so
3:48Nate's Nate's working on migration
3:51related um research topics in Ghana
3:55and oh and I'll also introduce Corey
3:57Vern know who actually works uh on my
4:00research team here at Yale and he can
4:02also maybe um like if there are so he's
4:05been working on related research with me
4:06so he can also like uh if there's Q&A
4:09answers that Cory you want to take
4:10please go ahead and type some answers if
4:12you like okay
4:14um okay and uh so let's uh get into the
4:18topic for today
4:21um so uh you know I I kind of took the
4:26uh the the words that were given to me
4:28migration and risk which were assigned
4:30by dog and Joe I took them seriously and
4:32I started thinking about all right so if
4:34I'm going to talk about migration and
4:35risk you know what are the important
4:38research agendas um uh and and research
4:42research topics in this in this area
4:44okay
4:45um so the first one is that migration
4:50itself may be a strategy to mitigate
4:52risk that people face that people may
4:54want to diversify sources of income by
4:56moving somewhere Right. And there's a
4:58large literature. I I'm just I'm just
5:00giving you one or two uh key citations
5:02from each of these bullet points, but
5:04there's like large literatures um behind
5:07all of this. So um so you know many
5:10people have thought about migration as
5:11risk mitigation. And this particular
5:13paper I highlighted here because it's a
5:15it's a fun interesting paper that shows
5:18that not only do people migrate in
5:20search of employment to diversify income
5:22sources, another thing they do to
5:24mitigate risk is that they marry farther
5:26away. So this is from rural India right
5:28and it shows that in more risky
5:30environments where there's more say
5:31weather fluctuations people are
5:33disproportionately more likely to marry
5:36uh into families that locate farther
5:38away which would have less correlated uh
5:41uh sort of income shocks.
5:44Okay. So that's one way in which
5:46migration and risk are related. Yet
5:48another way is that migration itself is
5:50risky because you're um uh you know
5:54you're having to go to a new place and
5:56there may be say risk of not finding
5:59employment. Okay. And this concept has
6:01actually played a key role in theories
6:04of development and structural
6:06transformation ever since we started
6:08theorizing about development and what
6:10what constitutes development structural
6:12transformation going all the way to the
6:131950s work by Nobel laurate Arthur Lewis
6:16and and and by Harrisono you know the
6:19idea of structural transformation is I
6:21mean I'm just going to paraphrase now
6:23you know you have um you know say people
6:26in a constant return to scale industry
6:28in the rural area something like
6:29agriculture Okay. And then there are new
6:31growth opportunities that open up in
6:33urban areas. So like there's
6:34industrialization and there's increasing
6:36returns of scale, right? And so in order
6:38to grow um countries have to uh induce
6:42labor to move from those rural areas
6:44that are relatively unproductive into
6:46these urban areas into this new um
6:49modern sector that's relatively more
6:51productive. Okay. And and then you know
6:54and then the wages are higher there.
6:56It's why why is there a wage gap between
6:58rural and urban areas? The one way that
6:59the models rationalize that is that oh
7:02this move would be risky and that's why
7:04you can sustain um lower wage rate in
7:08the rural area in inside the model just
7:11because that lower wage rate comes with
7:12some guarantee of getting the wage
7:14whereas the higher urban wage rate comes
7:15with some risk. Okay. Um and then
7:20another way in which risk and migration
7:22interact is that risk aversion often
7:26prevents us from investing in products
7:28and technologies and behaviors like
7:30rational response behaviors like
7:32migration. Okay. So that's uh so we'll
7:34talk a little bit about uh this today in
7:37detail. Uh I'll cover this paper um that
7:40I wrote with Garrett Brian and Shal
7:41Chosy who were at LSE and the University
7:43of Sydney. Right. And then recently
7:45David Lakos um uh also wrote a um
7:49overview piece where he talks about this
7:52idea um along with other ideas around
7:53why there's a migration and I but I saw
7:56that David is teaching as well as um and
7:59he has this on your reading list and
8:00therefore I didn't um I'm not going to
8:03spend too much time talking about that
8:04generally. Okay. Um and then um another
8:09way in which migration is linked to risk
8:12is that the act of migration may change
8:16the risk sharing or the insurance
8:17markets you have access to. Okay. So so
8:20uh I'm sure you've been exposed to this
8:22idea already that you know developing
8:25you know developing countries often
8:27rural and rare areas are very risky.
8:29Okay. And so people e either you have to
8:32then have formal insurance and if
8:33there's not u good um
8:37uh distribution of formal insurance
8:39markets in rural areas good penetration
8:41of those markets then you might have to
8:42rely on informal risk sharing so you
8:44know you and I are in a group together
8:47um you know in some years when I'm not
8:49doing well you help me out with the
8:50understanding other years me or my
8:52neighbors and friends will help you out
8:54okay that's what I mean by promotion
8:56okay and now if some people start
8:58migrating away that might actually
9:00change the resharing opportunities that
9:02are available to you, right? Uh so
9:05that's that's another way in which
9:06migration risk, right? And then the
9:08converse which is the presence of
9:11informal sharing or your reliance on
9:13informal sharing mayo prevent migration.
9:16Okay, how would that work? Right? So we
9:18don't have insurance. So like the rural
9:20area is relatively low wage. But the
9:23other thing about the rural areas or or
9:25or any any gi or any given area for that
9:28matter is that there's a lot of risk.
9:29There's a lot of fluctuation. So like
9:30one year you can get sick right or some
9:33family member gets sick suddenly you
9:35have a lot of health care costs right
9:37and you need some form of insurance to
9:39handle that. Now if in the absence of
9:41formal insurance like people might rely
9:43on their friends and neighbors right
9:45now. So that's uh so rural areas don't
9:47only offer wages. They al also offer
9:50those types of uh services that are
9:52available to you informal resharing with
9:54your friends and neighbors or your cast
9:55members or your clan members or your um
9:58tribe members whatever it is right and
10:00now if even though migration may be as
10:03I'm spoken to do it might generate uh
10:05income increases for your family it
10:07might generate growth for the economy
10:09um the the fact that if you migrate you
10:12lose all those services which is not
10:14easy for you to take with you to the
10:15urban area that might actually prevent
10:16the property migration So migration
10:18itself affects informal sharing but on
10:20the other hand informal sharing can
10:22prevent uh profitable migration. Um all
10:26right so okay so I I I was just u
10:30looking through the scroll and I see
10:31that you know there are people from all
10:33over the world which is great to see and
10:35and from many different types of
10:37institutions. I suspect that um relative
10:40to a normal classroom, we have a large
10:43much larger variety of um
10:46of backgrounds um uh in in this
10:49classroom. And so what I'm going to try
10:51and do is keep things somewhat general
10:54so that everybody gets something out of
10:56this, right? So I will point to like
10:58important ideas that I'm not uh that
11:00that you know those who want to like
11:03immediately start doing research in
11:04those areas you know you can get right
11:06into like what that um what those areas
11:09are but then there might be a lot of
11:10people here who are interested in macro
11:12and not necessarily on development and
11:14migration in which case just having um
11:18overview of this of this literature
11:19knowledge knowledge of this literature
11:21would be helpful. So, I'm going to try
11:22to keep things general, but I'm also
11:23happy to, you know, get emails from you
11:25after the fact and I'll I'll try my best
11:27to address those and of course use Nate
11:29uh for that purpose as well. Okay.
11:32Okay. So, these are the various uh
11:35important agendas. So, I'm going to talk
11:37about you know, we of course don't have
11:38time in 90 minutes to talk about five
11:40different literatures because each of
11:42these bullet points is actually a
11:44literature. Um it's not just single
11:46papers. So, so I I want to give you an
11:49overview of the environment we're
11:51talking about and why these particular
11:53questions are important. Okay. So, what
11:54you should have in mind is um and this
11:58the types of areas that developed
11:59economies spend a lot of time thinking
12:00about is rural agrarian areas of
12:02developing countries low and middle
12:04income countries. Okay. So, what
12:06characterizes those countries, right?
12:08It's agriculture remains uh primary
12:13uh uh uh uh primary form of income
12:18earning but more importantly a primary
12:20activity that many many people in those
12:22areas do right so the total value added
12:24generated by that sector is has been
12:26shrinking but it's still the case that
12:28in many countries across Africa and Asia
12:31um it it that sector still draws uh
12:34large numbers and and often the majority
12:36of population in that in those countries
12:38Okay. All right. So, what characterizes
12:40agriculture? Right. Uh, in a lot of
12:42places, it's weather dependent. It's
12:44like rainfall fed agriculture. Okay. And
12:45if it's weather dependent, that means
12:47that you're going to face large weather
12:48risk. Okay. So, there are some estimates
12:50u sort of more descriptive work that
12:52suggests that like 80 to 85% of
12:56interanual variability in livelihoods in
12:59rural India in the past 20 30 years has
13:02been due to uh the interanual
13:04variability has been due to fluctuations
13:06in weather. Right? So things like
13:08whether it rains enough or whether the
13:10first rainfall arrives on time and maybe
13:12more importantly whether the second
13:13rainfall arrives in time after you've
13:15planted the seed that would allow the
13:17seed to germinate, right? Okay, so it's
13:21a risky environment, right? And then the
13:22question is how do you mitigate that
13:24risk? Of course, the standard things we
13:26think about uh to mitigate risk is you
13:29might think about the credit market to
13:30do it. Okay, if you're doing poorly one
13:32particular year, you take a loan
13:35and when you're doing better, you pay
13:38back. Okay? Or another way to mitigate
13:41the risk uh other than credit is to say
13:43that oh I I can foresee that I will face
13:47some um some difficult years and
13:50therefore I should save up for those
13:51years. Okay. So but to use those markets
13:54you do need um you do need to worry
13:57about imperfections in those markets,
13:59right? And there are actually separate
14:00literatures that then talk about why are
14:03savings markets imperfect? Why don't
14:05people have good savings opportunities,
14:06right? Why are the returns on savings
14:09zero or negative in some places, right?
14:12Those are the types of questions that
14:14we'd have to ask to think about why is
14:16this market not solving that problem for
14:17us. Similarly, there's very large
14:19literature on credit market failures,
14:21right? That's where the some of the
14:23traditional ideas in development
14:24economics on moral hazard adverse
14:26selection come in. Right? is very
14:27difficult for us to identify good
14:29borrowers and that's why their credit
14:31markets often fail. So these you know we
14:34haven't been able to fully rely on these
14:36markets to solve this problem right. So
14:38another way to solve the problem be
14:40beyond credit and savings is insurance
14:42right insurance should be the first
14:44reaction you have to risk okay but again
14:48uh due to similar market failures
14:50there's also of course moral hazard
14:51adverse selection asymmetric information
14:53problems in insurance markets right and
14:55possibly other problems there's actually
14:56a big literature on this as well okay uh
14:59we don't see great penetration of formal
15:02insurance contracts in rural areas and
15:04the problems you know just to give you a
15:06flavor of how deep the problem like even
15:08when you bring in formal insurance right
15:11u so there are there have been um uh
15:14large scale projects done in India and
15:17Malawi right which show that you bring
15:18in insurance and even then like uh
15:22farmers are not willing to take it up
15:24right they seem reticent about taking it
15:26up okay and it gets even crazier like
15:29the project in Malawi by Shai Jane at
15:31the World Bank and Dean Young at the
15:32University of Michigan they say okay you
15:34know what I'm I'm going to have a credit
15:35contract and then I'll add like
15:37insurance on top of it. Okay, when you
15:39add insurance, the demand for credit
15:41actually falls, right? So, it's um it's
15:44really puzzling behavior. Okay, so
15:46again, so formal insurance markets have
15:48been very difficult for us to get going.
15:49So, there's another market failure there
15:51and in the absence of formal insurance,
15:52that's when we start relying more on
15:55informal risk sharing, right? And that's
15:57why there's such a large literature on
15:59informal risk sharing. And a seminal
16:01site for this um literature is uh Rob
16:06Townson's work. I think it was in
16:07econometric 1994
16:10where
16:12uh came up with ways to test u
16:14informative sharing. Oh, did somebody
16:15want to ask a question?
16:16Yeah. Uh hi. Uh so I was just wondering
16:19if there's if the risk is from like
16:22weather risk.
16:23It's a negative shock that's likely to
16:25be faced by almost all the people. So I
16:28don't understand how informal rat
16:29sharing works there.
16:30Sure. Excellent. Good question. Okay. So
16:32what Charlene is pointing out is that
16:34there are you know we should let's use
16:36the language from that literature. There
16:38are multiple types of risks right and I
16:40will characterize them as either
16:41aggregate risk or idiosyncratic risk.
16:44Okay. So idiosyncratic would be it
16:45affects one particular family but it
16:47might not affect your neighbor right but
16:49aggregate risk is something that affects
16:50you your neighbor and maybe everybody in
16:52the village. Okay. So idiosyncrat so
16:54Charlene's question is a good one which
16:56is idiosyncratic risk is easy to ensure
16:59through informal risk sharing right
17:01which is I my family member gets sick
17:04right that shouldn't be perfectly
17:06correlated with when my neighbor's
17:07family member gets sick right so then
17:10the neighbor and others help me out when
17:11my family member is sick with the health
17:13healthcare bills and in another year
17:15when they get sick I help them out okay
17:17uh so you you know like things like
17:19funeral societies right which you like
17:21institutions that you see in in like
17:23rural areas of Africa like that's a form
17:25of insurance against that. Okay. Now
17:27Charlene's question is that okay you
17:28might have aggregate risk like weather.
17:29Okay. Which is that when there's bad
17:31rainfall it's bad rainfall for
17:32everybody. Okay. Now one easy answer to
17:36that question but not a great answer is
17:38that oh you know there's occupational
17:39diversification within the village.
17:41Okay. So that's one way it could work.
17:42Right. But here's the much better answer
17:45which is that you know there isn't
17:46occupational diversification. So we
17:48can't just rely on that. You can't just
17:50say like because most people are engaged
17:51in farming and most people are growing
17:53similar things and therefore most people
17:54are care about the exact same rainfall
17:57patterns right the more important answer
18:00is that this literature has documented
18:02that these risk sharing networks are not
18:04village specific only okay so um so the
18:08the classic example of this is the cast
18:11system in India and Mark Rosenwig has as
18:15my colleague at Yale has done a lot of
18:16work on this and as well as Kaan Munchi
18:18who's done work and Kiban has written a
18:21um JP journal of economic perspectives
18:24overview piece on the role of the cast
18:26system in India where among other things
18:28he talks about its role in providing
18:30insurance okay and then when you so
18:32basically the think the way to think
18:33about a cast system is that you know
18:35you're born into this a subcast a jati
18:37okay and you're part of that network
18:39right and then you what you observe is
18:41that jatis actually span not only like
18:44not just villages but districts and
18:45sometimes even across states right so
18:47then you have like through a series of
18:49connections you're actually connected to
18:51somebody who might be in a different
18:52state right and then there's a series of
18:54transfers that may occur that indirectly
18:57ensures you when somebody else is doing
18:59well but you're not okay uh yeah but
19:01it's important you know in this
19:03literature it is definitely important
19:04for us to think about idiosyncratic
19:06versus aggregate risk right and again as
19:08I said you know like I'm just like
19:10touching the surface of many different
19:12literatures here a lot of this you know
19:14because my goal here for 90 minutes is
19:16like in a class like this where it's not
19:19like you know the same person lecturing
19:21you throughout. So everybody's bringing
19:22in a few different topics. My goal here
19:24is to try and expose you to some of
19:27these very important ideas animating
19:29development economics. Okay, especially
19:32around migration and risk and hopefully
19:35you know you'll find it somebody will
19:36find it interesting enough to like start
19:38digging uh much more deeply. Okay.
19:41All right. Um so we've talked about
19:44credit saving market failures. We've
19:45talked about lack of formal insurance
19:47and the uh value of as well as some
19:51limits to informal sharing. Okay. Now,
19:54another way to address
19:57uh address risk, right, is that you
19:59might want to diversify your income
20:01sources. So, you can't just rely on the
20:02local market, right? So, in this rural
20:04area, it's agrarian, but then there are
20:06certain seasons, right, where you don't
20:08need to hang out in the rural area. You
20:10might be able to go to the nearby city,
20:11go find work at a construction site or
20:14something, right? and you earn some
20:15money or maybe your family is engaged in
20:17agriculture and you send one migrant
20:19away somewhere else. Okay. And the value
20:21of that is that even if that migrant
20:23does not necessarily earn more money in
20:25expectation, the fact that they are
20:27drawing from a different labor market
20:30with different labor rainfall
20:31conditions, right? A different set of
20:34risk. Having that kind of
20:35diversification helps with managing that
20:38risk because essentially you're not
20:39putting all of your eggs in that same
20:42village basket. Okay. All right. So, so
20:45those are uh those are like some of the
20:48most important literatures in
20:49development economics on credit market,
20:51savings markets, insurance markets and
20:53migration, right? Those are four ways in
20:55which we think about risk mitigation.
20:58Now another way in which another thing
20:59that really matters here is be beyond
21:01risk right we also have to think about
21:04okay how do you get out of uh like uh
21:07how do you get out of poverty right how
21:08do you generate growth okay and again
21:11migration is an important path to that
21:14there's a separate literature that has
21:17you know and and some of the papers that
21:19you might want to read are by Michael
21:21Clemens who's at the center for global
21:23development David McKenzie who's at the
21:25world bank right that shows that you
21:29know m you know migration to a labor
21:32market that offers higher wages is one
21:35of the most important and quantitatively
21:37one of the like largest paths out of
21:40poverty. Right? The type of returns that
21:43migration can generate if you manage to
21:46find work in a much better labor market
21:48that offer higher wages. the returns
21:50that acrew to that family right they uh
21:53put to shame all the other returns that
21:56we have estimated in terms of making
21:57investments in health investments in
21:59education etc etc okay so migration also
22:03turns out to be a really important path
22:04to growth right but if that's true then
22:07we have to then of course figure out why
22:10are if if the returns of migration are
22:12that high why isn't everybody migrating
22:14out okay so that's going to be an
22:17important puzzle and I'll come back to
22:18this right talk to you about that
22:20literature and you know and David
22:21Lagakos's piece that I talked about
22:24is you know touches on a lot of those
22:26types of issues okay and another way to
22:29generate growth uh in which again risk
22:31plays a large role is in technology
22:33adoption right so you can actually think
22:35about migration as a technology right
22:37which is why don't I adopt migration as
22:39a strategy to generate growth okay
22:41similarly why don't I adopt new
22:43varieties of seeds as a strategy to
22:46generate growth so Doug uh for example
22:48has a recent paper that documents how
22:51important the green revolution was,
22:53right? Uh like essentially what's a
22:56green revolution? It's a set of new
22:58technologies, right? As they were
22:59adopted, right? You saw large increases
23:02in agricultural productivity. Okay? Now,
23:05if that's true, then we have to start
23:07thinking about, okay, why hasn't the
23:09green revolution happened in Africa yet?
23:11Right? Is it that the technology doesn't
23:13exist or is it that we're having trouble
23:15getting people to adopt technologies
23:16that do exist? Okay. Or we can even I
23:20mean even though the green revolution
23:21has happened, we should also ask the
23:23question why did the green revolution
23:24take 30 to 40 years, right? Cell phones
23:27didn't take that long. Okay. And and if
23:30if Doug's numbers are true or other many
23:32other numbers also they are true which
23:34is that um you know green revolution
23:37like tripled agricultural productivity
23:39in some areas. Okay. So if that's true
23:42then uh why did it take like 30 40 years
23:44for farmers to be convinced right? they
23:47it should have been spreading like
23:48wildfire, right? So that again again you
23:51know risk might have something to do
23:53with that right and your risk aversion
23:54might have something to do with that and
23:56we should be and that's you know uh
23:58that's a puzzle we should be thinking
24:00about as well and the reason to think
24:02about that puzzle is that you know even
24:03in areas where the green revolution did
24:05happen if we could speed up that process
24:07such that the diffusion of technologies
24:09instead of taking 30 years if it could
24:11take 10 years right people have done
24:13simulations and that speeding up that
24:15process would have enormous consequences
24:17for welfare
24:19Right? And therefore it's important for
24:20us to think about that as well. Right?
24:22Then risk might be an important uh part
24:24of that story as well. Okay. So now the
24:29research agendas that I'll spend time
24:32talking about in the next 60 65 minutes
24:34are I guess less than 60 minutes is
24:36these ones right why is there low
24:38technology adoption now is migration a
24:40profitable technology and if it is why
24:42don't people migrate okay and is risk
24:45aversion a deterrent to technology
24:46adoption and does migration change in
24:49formal risk sharing so I'll just touch
24:50on these um like something I'm
24:53emphasizing over and over again is that
24:54there are multiple papers in each bullet
24:56point. So, I'm just going to scratch the
24:58surface. Okay. Um, and one other thing
25:00I'll point out here is that when we talk
25:03about we talk about migration here down
25:04here as a way to generate growth, right?
25:07And we're talking about migration here
25:09as a way to mitigate risk. Okay? And so,
25:12migration, so always bear in mind that
25:14migration can take multiple different
25:16forms. The motivations for migration are
25:18different. Sometimes it's to search in
25:20search of higher wages. Sometimes it's
25:22that I'm having a bad year and I need to
25:25find ways to feed my family this year.
25:27Right? And as a result, the type of
25:30migration you do can be very different.
25:32Okay? So this type of migration is often
25:35internal within the same country. Right?
25:38It's fast migration where you go for a
25:40few weeks or a couple of months and you
25:41come right back. Right? You're still
25:43primarily a farmer in a rural area, but
25:46you're spending some of your time in the
25:48urban area. Okay? So you see a lot of
25:51what's called in in that literature
25:52seasonal circular migration right
25:57and short-term migration. So those are
25:59the times types of um uh key words that
26:02you'll see in the literature short-term
26:03seasonal circular whereas you also see
26:07permanent migration. So this is more
26:09like temporary migration and there's
26:10also permanent migration. By permanent
26:12we might mean you know people move to a
26:15city or people move to another country
26:16but it could also mean that there's a
26:18person who's moved for a few years.
26:21Okay. And sometimes what happens is that
26:24a person moves for a few years leaving
26:25their family behind with the
26:27expectations that they will return.
26:29Okay. So given all these different types
26:31of migration you know and sometimes a
26:33migration can be with documents and
26:34sometimes it's without documents. Right.
26:36So given that all these different types
26:38of migrations that occur, it could be
26:39permanent, could be seasonal, it could
26:40be short-term, could be few years or or
26:43or or um you're you're just not
26:45expecting to return. You might take your
26:47family with you or you might leave your
26:48family behind. Of course, the returns to
26:51those different types of migration is
26:53going to be different, right? The way
26:54that I choose to remit or share will
26:57depend on whether I'm leaving my wife
26:59and kids behind, right? Versus if I'm
27:02moving with my whole family, it's only
27:04my extended family that's behind, right?
27:06So that means it's not it's not like we
27:07can go to the literature and say okay
27:09tell me what the returns to migration
27:10are and we find a single number right
27:12the number shouldn't be the same one
27:14right so there's a there's a lot of
27:15detail even within these bullet points
27:18that has spawned uh multiple literatures
27:21sub literature'ses within migration okay
27:25[snorts] all right so let me start with
27:28this first bullet point about low
27:29technology adoption okay and I'm going
27:31to broaden this further because I want
27:33to sneak in like other even other
27:35literatures that some of you might find
27:36more interesting and might want to think
27:38about, okay, which is I'm using
27:40migration or agricultural technologies
27:43like new varieties of seeds or
27:45fertilizer as my examples for technology
27:47adoption, right? In what I've just told
27:50you, it turns out there are
27:52unfortunately too many examples of this
27:55of welfare improving technologies or
27:57technologies that we believe might be
27:59good for people that are not being taken
28:01up. Okay. So, by the way, I I I was
28:05careful to say apparently beneficial
28:07that we believe might be good for people
28:09because sometimes it turns out the
28:11technology actually sucks and we what we
28:14believe is wrong. Okay? And and we
28:16always have to u have to like be aware
28:20of that possibility, right? Which is
28:22that sometimes it's not a puzzle, okay?
28:23If you see a techn is not being taken
28:25up, it's because the techn is not good.
28:27We just like measured returns
28:29incorrectly. Okay? So okay but let me
28:32let me tell you about the types of
28:34puzzles that people have been working on
28:35and this is like not just in development
28:37economics it's in health economics
28:38public health it it spans in agriculture
28:40it spans a lot of different sectors okay
28:42so a lot of health products have this
28:44feature right like there's a small
28:46literature on insecticide treated bed
28:47nets so malaria is an important cause of
28:50low growth according to some research in
28:52subsouthern Africa okay we have a
28:54technology that actually prevents
28:55malaria transmission that's an
28:57insecticide treated bed nets okay but
28:59when researchers have tried to um
29:02disseminate bed nets and even if they
29:04offer it at like 80% subsidy so like 20%
29:06of market price you see that people
29:08aren't taking it up. Okay. Why is that a
29:09puzzle? Because if you do sleep under a
29:12net, the the early stage studies in
29:14terms of let's say the efficacy of the
29:16bed net, which might have been which
29:18which might have been a study that's
29:19done in a lab, nothing to do with
29:20economics, right? It's just like you
29:22look at how mosquitoes interact with the
29:24net, okay? You don't need economists to
29:27do that, but you find that oh, there's a
29:29lot less transmission, right? So once
29:32you once you learn that, then you're
29:33like, okay, so then if this bed net,
29:35let's say it cost $50, right? like in
29:38terms of the lower probability of
29:40contracting malaria, you would the
29:43bednet would more than pay for itself,
29:45right? Which is that okay, you're less
29:47likely to get malaria, therefore you're
29:48going to miss fewer days of work, right?
29:50So you're you're going to come out ahead
29:52economically. And the second thing is
29:53you're going to also avoid healthcare
29:54costs. Okay, so then you know you look
29:57at the first stage, let's say lablike
29:59studies and you say, okay, look that
30:01this is sensible technology. It should
30:02pay for itself. People should be buying
30:04it. Okay, and then the puzzle is out.
30:06then in that case why aren't people
30:07buying it you know not only at not at
30:10market price but at like 20% of the
30:12market price okay and that's when you
30:15know economics might come in right which
30:17is let's say we we need to now
30:18understand what is driving that
30:20household behavior why are they choosing
30:22to avoid a technology that should be
30:24really that we think should be really
30:26good for them right and that's where we
30:28need to start thinking about behavior
30:29right so similarly like you know you see
30:33low adoption of condoms or birth control
30:35technologies you the low adoption of
30:37vaccines that's something that we are
30:38having to deal with right now across the
30:40world like that's true in both
30:41developing and developed countries right
30:44okay so some health examples there are
30:47many good agricultural examples that
30:49we've already talked through there's
30:50also other health product examples like
30:52toilets okay so diaral disease is a big
30:55problem okay and one of a technology
30:58that prevents diaral disease
31:00transmission is to is using a toilet
31:01right but for that to work you actually
31:03have to use it and And um for decades on
31:08you know at at least until very recently
31:10and maybe even now like you have half of
31:12all rural Indians still defecating in
31:14the open. Okay. So now that's a big
31:17puzzle right that's a big public health
31:19threat. If you defecate in the open a
31:20fly can sit on the feces and go sit on
31:22somebody's food and that's how disease
31:23spreads right. And again it's a puzzle
31:26because oh look there's a technology
31:27that we know actually works. Right now,
31:30this is a case where it's going to be
31:31hard for us to argue that the technology
31:34sucks or it doesn't work, right? Because
31:37there's a lot of good like even without
31:39doing research, right? There's probably
31:41a lot of good evidence that all of us
31:43have that that's a technology that's
31:44that we think is useful, right? And in
31:47spite of that, half of people half half
31:49not using it. Okay? So, that's that's
31:51something we need to understand. Okay?
31:53Uh we've talked about farmers not being
31:55not taking a weather insurance and then
31:57migration is another example. Okay. So
31:59note that this example spans not only
32:01what we're talking about but a lot of
32:02other um development issues and the
32:06thing is that uh this could be a key
32:09constraint to development right so for
32:11example if you see that insurance is not
32:13being taken up like the story I told you
32:15from Malawi and it's true for India as
32:16well okay so now you're relying on okay
32:19why is insurance not taken up okay so
32:21when we write down a model we see oh the
32:23model is that oh there's already
32:25informal risk sharing and the nature of
32:28that risk sharing mediates a demand for
32:30formal insurance. Okay, so you're you're
32:32born into Ajati a cast, right? So you're
32:34already partially insured, right? And by
32:38the way, that's what the research shows
32:39like starting with towns in 1994 that
32:42people do have a lot of informal
32:43resharing. So they have a lot of
32:45insurance, but it's not complete
32:46insurance. Okay? And it exists, but but
32:48it doesn't fully insure you. All right?
32:50So you have that insurance. And then if
32:52I go to uh Michelle or Nate and I tell
32:56them uh oh, you know, here's a new
32:59product, right? Formal insurance. Do you
33:01want to take it and I'm going to sell
33:02you at sell sell it to you at actually
33:04fair rates, right? Michelle's response
33:06might be actually it's not actually it's
33:09not fair because I'm already partially
33:11insured. I was born into this cast and
33:13then Doug who keeps having problems
33:15every year, right? He keeps coming to me
33:17and asking for help, right? So, I'm
33:19going to continue having to help him.
33:21Okay. And if I buy your insurance, he's
33:23going to ask me for help even more,
33:25right? So, so that's why I might I might
33:28not be interested in buying that
33:29insurance. Okay. So, but if that's true,
33:32right? So, I'm so I keep relying on
33:34informal sharing. Then what ends up
33:36happening is that Michelle's like, "Oh,
33:37you know, the other thing that happens
33:39here is that if I invest and get rich,
33:42right, Doug shows up my up at my door
33:44all the time, right, and ask for help.
33:47And so that might keep me from investing
33:50from taking on entrepreneurial
33:51activities that you know uh that that
33:54might improve my um improve my economic
33:57prospects and generate growth in the
33:59area right so then that's why it's
34:01really important for us to figure this
34:02out right which is that if we you know
34:05we understand why that informal rearing
34:08network exists why that institution
34:11was designed the way it was right but
34:14for us to get development we also might
34:16need to figure out like how to move to a
34:19different equilibrium, right? Where we
34:22use more formalized instruments which
34:24may actually be good for growth, right?
34:26But that's also, you know, I'm not I'm
34:28not going to sit here and say with
34:30certainty, oh yeah, that's the path to
34:32growth, right? But that's something
34:34worth uh looking into, right? Um okay.
34:38And then you know, so in for this has
34:41animated a large literature in
34:42development economics. So there's a lot
34:43of like randomized control trials in
34:45developing countries that's been um uh
34:47that people have been using to um
34:52uh to to to study uh like behavioral
34:54models as to why there's so much
34:56hesitancy around um apparently welfare
34:59improving technologies. Okay. So let me
35:00ask you now like uh what what do you
35:02think it is?
35:04Why do you think there's like these
35:07technologies that seem really sensible
35:09like toilets or better varieties of
35:12seeds or bed nets or vaccines but people
35:14don't want to take it up?
35:17Any ideas?
35:22Um so we are you know trying to look
35:25from the economic perspective trying to
35:27model the utility functions from what we
35:29know but there might be certain aspects
35:31which these models are not able to
35:33capture. For example like if you look at
35:35the uh the toilets example right so
35:38there are certain religious beliefs in
35:39India which do not you know accept this
35:42kind of a thing and that is probably why
35:43people are not adopting it. So this
35:45behavioral aspect is what is really
35:47missing from these models and why
35:49adoption is not high. And when it comes
35:51to the seeds and variety of fertilizers
35:53again you know you need to have certain
35:56irrigation facilities available. So
35:58those kind of other inputs or
36:00complimentary things that are required
36:02that might play a role.
36:03Okay good. Okay. So let's let me respond
36:06in in backwards. Right. So, so when you
36:09when you say Nikita that there's some
36:10complimentary technology that's required
36:12that's not present that actually goes
36:13back to oh this technology is actually
36:15not that good. Okay, which is like the
36:18returns to the technology that we're
36:19measuring to be very positive that we're
36:21that's why there's a puzzle, right? Um
36:24it's because we are measuring the
36:25returns in a technology in a place where
36:27irrigation exists but like as as soon as
36:29you take it to a different context those
36:30returns don't exist anymore. So there's
36:32actually no puzzle or another way to say
36:34it is that the puzzle is not in this
36:37agricultural technology or the seed. the
36:40puzzle is how do we address the
36:42infrastructure problem or radiation
36:43problem right it might just take us
36:45somewhere else right um and then you
36:48know but having said that I would oh
36:51like let me actually let me go to the
36:52culture uh explanation right now okay
36:56first of all don't blame economist for
36:58low technology adoption so like if if
37:00your answer is correct okay then what
37:03should happen is that the people who are
37:05working on um you know toilet promotion
37:09and dissemination, right? Right. They
37:12should be uh carefully designing their
37:14messaging that addresses the religious
37:16issues, right? I mean, that's not that's
37:18not the fault of an economist. Maybe it
37:20just says that economists shouldn't be
37:22thinking about the problem like somebody
37:24else should should advise who uh how how
37:27to get around it. But but from an
37:29economics perspective though, right, if
37:32there is some cultural explanation,
37:34religious explanation, right? uh I think
37:37you know I'm I'm now I'm just making a
37:39general comment about approach to
37:40research okay and I think it's incumbent
37:42on us to then dig into where where did
37:45that cultural cultural belief come from
37:48right that's where we can make a maybe
37:50make a contribution right so just as an
37:52example I mean on on culture um say you
37:56know uh like how how does it how does it
37:58work out in um in the in in this example
38:00that we were working with with informal
38:02rearing and insurance okay so so what
38:05happens is you know you're you're you're
38:08born into a jati in India right you're
38:10part of this cash network okay and then
38:13remember like remember like the story of
38:15Michelle and Doug right so Michelle
38:17figures out like oh she's a good
38:19entrepreneur she has good ideas right
38:21and like and Doug keeps uh imposing on
38:25her and keeps asking her for help okay
38:27uh sorry I'm picking on Doug today but
38:30um but actually Doug I know Doug so he's
38:32kind of like that anyway so Um
38:36uh so keep keeps imposing on her. Okay.
38:39So then uh Michelle is going to be like,
38:41you know what? I've now realized that I
38:42am on average better than than Doug and
38:46it doesn't make sense for me to remain
38:47in this network. So what should she do?
38:48She's going to try and escape it, right?
38:50Maybe she'll migrate, right? He's like,
38:52I don't I don't I don't want to do it.
38:53Okay. But but like everybody like
38:55Michelle who's like above average in
38:57that network, they'll have that same
38:58incentive, right? Okay. And then if that
39:01happens then the informal resharing
39:03network doesn't work anymore because
39:05precisely the set of people right who
39:07are doing well they're going to try and
39:09escape and then the network would like
39:11unravel and only losers like Doug are
39:14not going to be able to like help each
39:15other out. Okay. So um so so then you
39:20know you need to then come up with some
39:22institution that prevents Michelle from
39:24departing. Okay. What's that
39:26institution? Now if you've watched any
39:28Bollywood movies you'll recognize this.
39:29that institution is oh hey Michelle if
39:32you try to depart and leave this network
39:34like nobody is going to marry your son
39:36or daughter right because you know
39:39because you know we're going to
39:40excommunicate you from from this and so
39:42that why I say Bollywood movies like
39:44every third movie is about this
39:46difficult love story where somebody
39:47falls in love with somebody from a
39:49different cast and it causes all sort of
39:50Romeo and Juliet type problems right so
39:53so then you um you have this um
39:57um you know you have this rule that oh
40:01um Michelle and and her family any every
40:05family member can only marry people who
40:07are part of the same cast right and so
40:09if you leave right we're not going to
40:11marry you and other cast won't accept
40:13you either because you're not part of
40:14their cast okay and then you have to say
40:16okay and then how do you enforce that
40:18right you have to enforce it for to keep
40:20Michelle from leaving right so how do
40:22you enforce it you say oh god said so
40:25right and maybe that's where the
40:27religious belief comes
40:29Right. So again I mean that's just an
40:31example nikita of like if we think that
40:34religion or some cultural belief is the
40:38key reason right and if you're quite
40:40confident in that so where can we
40:42contribute we can dig into it and figure
40:44out okay where did that come from? Where
40:46did that belief come from? Right. And
40:48perhaps there are some underlying
40:49economic incentives which can act as
40:52like levers we can push, right? If if
40:55this is really something that's holding
40:56back people's public health and
40:58development, right? So again if it's the
41:00case that in India so somehow you know
41:02you'd also have to explain that oh in
41:04India uh you know maybe it is religious
41:06belief because we're seeing this problem
41:08has been addressed quite well in
41:09Bangladesh right and there's some work
41:11in India that suggests that the problem
41:13is a little bit better addressed among
41:15Muslim communities and Hindu communities
41:16even though the Muslim communities are
41:18on average poorer right maybe there's
41:20something to it but we need to
41:21understand where where it's coming from
41:23all right so so those are a couple of
41:25explanations good any anything else on
41:27why people don't adopt why we see this
41:31anymore. I mean, Nikita already went
41:33into some um really sophisticated
41:35explanations. What are the most obvious
41:38things that come to your mind?
41:40Um I I think
41:43uh Shutaru, why don't you go ahead
41:44first?
41:45I think the benefit doesn't come in the
41:48short term and it's a more long-term
41:50thing. So if some people have a distinct
41:54like discount factor, they won't take up
41:57in the short term and that continues
41:59indefinitely for their life.
42:01Right. So good. So basically what you're
42:04saying is that people don't have money
42:06up front when the benefits are coming in
42:08the future. Right? So when I was saying
42:09that the bednet is um uh that the bed
42:12net is um is is sensible and it's
42:15puzzling people are not adopting it.
42:17It's just that those health avoided
42:19healthcare costs comes 5 years later,
42:21right? But you need to pay that $50
42:22today. And what Chutaro is essentially
42:25saying is that people don't have that
42:27$50 in their pocket,
42:29right? They don't have the liquidity,
42:32right? But that is not for an economist
42:34that's not going to be a complete
42:35answer, right? Because if this is true,
42:37Shataru, what should you do? Can
42:39somebody else tell us? Like if Shataru's
42:41explanation is correct, like what's the
42:43market solution to that?
42:48Charlene, go ahead.
42:49Access to credit for
42:51Yeah, exactly. So, what you do is, you
42:53know, I know that if I pay $50 today,
42:55I'll get $100 in return. It's a good
42:57investment. That's what I mean by the
42:58bedet being a sensible investment. Then
43:01I should be able to go to a bank and say
43:03that give me the 50 bucks. I'll get a
43:05bed net and I'll pay you back 60 and
43:06I'll come out ahead. Right? So that must
43:08mean again there's a credit market
43:09failure. Similarly, there's probably
43:11also like no, you know, it's not like
43:13I'm getting surprised by the fact that
43:15there's a bad net in the market today,
43:17right? Like if I think that's a good
43:19investment, I should have like every
43:20year saved up $5 until I got the 50 to
43:25to invest in it, right? So there must be
43:26some other saving failure as well,
43:28right? But it's all related to lack of
43:30money being the problem, right? Uh
43:32anything else?
43:34Uh but if it's a combination of lack of
43:37liquidity and credit market constraints.
43:39Yes. Yes. Exactly. So what we're saying
43:41is you know if if you take answer we
43:43have to say oh credit market failures
43:45are having to do something to do with
43:46it. That's that's exactly what I'm
43:48saying. Yeah.
43:50Okay.
43:52So another common reason that comes up
43:54other than lack of money when you ask
43:55these questions is that people aren't
43:58fully informed, right? They don't see
43:59the connection between the toilet and uh
44:03diarrheal disease, right? They don't see
44:04the connection between an improved cook
44:06stove and uh and the and the coughing
44:09symptoms or respiratory infections that
44:11they have, right? And as soon as we or
44:14or people don't know about the benefits
44:15of the new agricultural technology,
44:17right? And we need to address those.
44:19Okay. And then with benefits of the new
44:21agricultural technology, right? We you
44:23know already uh Nikita talked about
44:25taste and tradition. There might also be
44:26things outside of rational costbenefit
44:28decision- making like some decisions are
44:30habitual right and then with agriculture
44:33technology you know it might also be
44:34that oh I don't know if it's going to
44:36work right I believe you if you're
44:38telling me that you know you tested it
44:40in the eerie lab in Manila and it worked
44:42well but like what if given my farming
44:45conditions like my soil conditions my
44:47rainfall conditions my farming
44:48techniques it doesn't work for me I
44:51cannot afford a crop failure whereas I
44:53have this traditional technology that
44:54I'm using my mother and my grandmother
44:57used and it's working well
45:00right so risk might have something to do
45:02with it okay then you know we've talked
45:04about some substitute informal solutions
45:06and there might be already shataru
45:08hinted at this present bias preferences
45:10right and then nikita also hinted at
45:13like some kinds of complimentarities or
45:15interlinkages between different types of
45:17technologies or conditions okay
45:20and oh and one other thing to keep in
45:22mind okay there's also a small
45:23literature here you know my My goal
45:25again is to just like
45:27expose you to it and then you'll have to
45:29do do the digging yourself. Sometimes,
45:31you know, when we think about the
45:32household as a decision maker, it could
45:34be that there's one person in the
45:36household who's the problem, right? So,
45:38think about improved cook stoves which
45:39prevent respiratory infections. It's
45:41women who benefit from the new
45:42technology because they're the ones who
45:44are cooking, right? But if they're not
45:46the financial decision maker, if their
45:47husband is, right? You might see this
45:49puzzling decision of people not adopting
45:52technology. But it's not so puzzling
45:54when you think about the husband making
45:55the decision for him. It's a negative
45:57net return technology but not for the
45:59wife, right? So there might be an
46:01externality that's within the household.
46:03Okay. And you know why you know why are
46:06we going through this and why is there
46:08literatures on this? Because
46:09disentangling different reasons for
46:11adoption has important implications for
46:12policy, right? So for example, if really
46:15it's just an information failure, then
46:17going around subsidizing the technology
46:19may not be sensible to do, right?
46:21because you can reach a lot more people
46:22with the same budget if you just um uh
46:25if you just run information campaigns,
46:26okay, which is much cheaper. But in the
46:28on the other hand, if it's something to
46:29do with the liquidity or credit
46:31constraints, then just running
46:32information campaigns might be a waste
46:33of money. Okay, so and one way that
46:37we've tried to address these problems is
46:39like marketing
46:41um you know these randomized control
46:43trials. We market the product or
46:45behavior of the technology to a few
46:46thousand households and then we randomly
46:49vary how you know what source of
46:51aversion you're trying to target like
46:52whether it's an information campaign or
46:54some subsidies or access to credit or
46:56providing some money back guarantees
46:58like insurance right to to disentangle
47:02these different reasons and understand
47:03what the underlying issue is. And then
47:06in this particular research approach
47:07what ends up happening is like
47:09randomization is often helpful to un you
47:12know to get good data on
47:14um like what exactly the source of
47:16version is but then you also sometimes
47:18need large samples right because you
47:20want to make statistically precise
47:21inferences on like what know which of
47:24these technologies are working right and
47:26then uh what's kind of in the um getting
47:30to the state-of-the-art in development
47:32economics is that like designing these
47:33experiments not only to answer this
47:35first stage question I just laid out but
47:37also then thinking about okay fine if if
47:40like providing credit does work right
47:42let's now think about how that would
47:44look as policy
47:47right like if we if we were to like
47:49introduce a big credit market for
47:52toilets right uh or we start developing
47:56the supply chain so that there are
47:58toilet sellers toilet part sellers in
47:59villages right what kind of equilibrium
48:02level changes might that occur and here
48:04this This is where a lot of the given
48:06the the topic of of this overall the arc
48:09of the course this is where a lot of you
48:11know where the kind of the macro
48:13literature and the
48:15uh development economics literature
48:17right is intersecting. Okay. So the
48:20types of papers that you're seeing is
48:21that development economists who have
48:23these skills in fieldwork and running
48:26these experiments and generating high
48:28quality uh data and high quality
48:32variation in that data. Right? So much
48:35more sort of applied micro techniques
48:36right. So they might you know you might
48:38end up running that experiment right and
48:40generate that data but if but that data
48:43by itself is not going to be large
48:45enough or ambitious enough for us to
48:47answer the question oh what are going to
48:49be the equilibrium effects of this what
48:50are going to be the effects on people
48:51who are not the direct beneficiaries but
48:53other people who are interacting right
48:55so just as an example if I were if I
48:58wanted to think about migration as the
49:00technology okay that why are people not
49:03adopting it and the way I do it is that
49:05I market migration and I try to get
49:07people to migrate. Okay. So, of course,
49:09there's going to be a direct set of
49:10effects on the people I gave the bus
49:12tickets to to encourage them to migrate.
49:15Okay? And we can study that and and and
49:17randomized control trial might be
49:19sufficient to study that, right? But
49:21what ends up happening is that if you
49:23really think about that as policy,
49:24right? If you really are interested in
49:26addressing that problem at scale, what
49:28might end up happening is that you give
49:30lots of people these bus stickers, many
49:32people migrate. Now that's going to have
49:34an effect on land owners who rely on
49:37those laborers for various like
49:39agricultural tasks right and we need to
49:42understand what happens to that rural
49:44economy. It's also going to have an
49:46effect on city residents right who now
49:50see an influx of migrants and they they
49:52might be competing with the migrants. So
49:54we need to understand what those effects
49:55are and those are the types of things
49:57where increasingly what you're observing
49:58is you take that applied micro like
50:01generated by people who do that style of
50:04work right you combine it with
50:06macroeconomic modeling right where of
50:09course there's more assumptions that
50:10need to be made but you model these
50:12processes these interactions between the
50:15people who receive the intervention as
50:18well as others who interact in those
50:19same labor markets or in land markets or
50:21whatever it is right and then you try
50:23and get a more comprehensive picture of
50:25what the uh what the effect of the
50:28program might look like at scale, right?
50:30And a lot of the um you know really good
50:34papers that we were seeing coming out in
50:36the last five, seven years have this
50:39feature where you're bringing in the
50:41tools from different fields, right?
50:44You're combining them to answer more
50:47sort of ambitious questions.
50:49Okay.
50:51Um so I was just um uh you know
50:55selecting papers for the annual like uh
50:58CR conference and the NBR conference in
51:01development right and you see some of
51:03the strongest papers right have have
51:05this kind of a feature I mean of course
51:06it takes a lot longer to write because
51:08you're trying to you can't just run the
51:10randomized control trial you also have
51:12to do additional work in order to uh you
51:15know be more ambitious and answer like
51:17questions that about effects in the
51:19aggregate uh but um but but it's a style
51:23of style of work that I think has um is
51:25is receiving a lot of value and credit
51:28in the profession.
51:31Okay. So let's um let's now uh I want to
51:36give you the migration example to to
51:38just give you an example of very quickly
51:39go through in a in a particular paper
51:41what this ends up looking like. Okay. So
51:44uh so you know let me just um maybe uh
51:47may maybe just skip over this just to
51:49say you know what do we mean by risk
51:51aversion as a source of uh low
51:54technology adoption right it's not that
51:56I'm wor you know that the product like I
51:59think the product is bad and therefore
52:00I'm not adopting it right it's that
52:04I in expectation I think the product is
52:07good this is something that will work
52:08for me but I still choose not to
52:10experiment with it right so under what
52:13conditions would I have beliefs that
52:15this is my this is probably going to
52:17make my life better off and even then I
52:20cannot take it up right that's the
52:22that's the field that I I want to get
52:24into and then to talk to you a little
52:25bit about okay
52:28so um so the ex specific example I'll
52:32use is migration to stick with the uh
52:35the topic of the lecture today migration
52:37and risk okay and uh the the let me let
52:44me explain
52:46the the setting we're in. Okay, so think
52:48about poor rural areas of developing
52:50countries and actually let me do it in
52:52the form of some graphs. Okay, so this
52:54is just data nationally represented data
52:55from Bangladesh. Uh the blue is the rest
52:58of the country and red is this very poor
53:01rural area in north north Bengal. Okay.
53:03So generally speaking this this ronpur
53:06area is poorer than the rest of the
53:08country. But importantly there's a
53:10particular season in which it gets
53:12especially poor. Okay where income fall
53:16and what is that season? So essentially
53:18here is where you harvest the crop and
53:20here is where you plant it. So you plant
53:21the main rice planting is here and you
53:23harvest here. Okay. So during this
53:25period you're just waiting for the crop
53:27to grow. Okay. And so most of the people
53:29who are engaged in agriculture in South
53:31Asia, including in Bangladesh, they're
53:32actually landless, right? So they're
53:34supplying their labor on other people's
53:36farms. So while everybody's waiting for
53:37the crop to grow, the demand for their
53:39labor falls. So wages fall during this
53:41period. Like there's a few tasks here
53:43like weeding etc. But it's not uh like
53:45it's not where people earn a lot of
53:46money. During harvest, there's a lot of
53:48tasks. During planting, there's a lot of
53:49tasks. Yeah. So you see overall
53:51consumption dropping, but during that
53:54period, food expenditure goes up. And
53:55the reason is something somewhat
53:57independent which is that the price of
54:00rice spikes up during that period. Okay.
54:03And this probably has to do with the
54:04fact that you know
54:07it's been the longest time since the
54:08last harvest, right? Just before the new
54:11harvest. And there's probably some other
54:13market failures here in in like storage
54:16and so forth. Okay. All right. So that
54:18means if you combine these what you see
54:20is that the price is going up. people
54:22are having to spend more money on food,
54:25right? And they're but but overall we do
54:28see consumption going down. I'll show
54:29you that in a second. And over in
54:30overall like non-food consumption goes
54:32down by a lot. And so given the price
54:34increase, you see quantity consumed
54:36going going down, right? And in our
54:38survey data in North Bengal, this shows
54:40up where exactly during that period,
54:42September, October, November, right? You
54:44have much larger increases in people who
54:47are experiencing some food insecurity.
54:49They're saying that we often miss meals
54:52or reduce our portion sizes in order to
54:54cope. Okay, so there's like there's
54:56poverty, but there's also seasonal
54:58dimension to poverty. That's the setting
54:59we're in. Okay, and this is not just a
55:01recent phenomenon. This is a paper from
55:03Nature published in 1884. Thankfully,
55:06the British when they were in South
55:08Asia, they collected a lot of data,
55:09right? So, we actually saw seasonal
55:12patterns and mortality even back then,
55:14right? So there's a series of papers
55:15from over 100 years ago that suggest it
55:18and this also shows up in lots of
55:21countries around the world where there's
55:22predictable seasonal patterns in food
55:25insecurity in stunting and wasting etc
55:28etc. Okay.
55:30Um all right so there's a failure of
55:32smooth consumption smoothing. Okay. So
55:34why are these what are markets that are
55:36failing? As we already said, credit
55:37savings, right? That's or labor markets
55:40are probably not specially well
55:42integrated, right? That's why the
55:43traditional solutions of credit,
55:45savings, insurance, migration may not be
55:48working very well, right? And and so
55:51what are the solutions to this? Like,
55:52you know, you could think of subsidies
55:54for migration as a solution, which is
55:55the one I'll talk to you about, right?
55:57You could also think about seasonally
55:59time consumption loans, making it easier
56:01to remit funds for people who are
56:03migrants, right? or like some technical
56:06fixes for storage, right? Or making it
56:08easier for people to save. There could
56:09be a many of solutions. The one we'll
56:12talk talk a lot about today is subsidies
56:14for migration. Okay?
56:16So, so let me actually um uh you know
56:20get into what the uh what the project
56:23does, right? So, first thing we do is to
56:27is to provide is to run that randomized
56:29control trial is to provide subsidies
56:31for migration. Okay. And with that we
56:33document that there seems to be pretty
56:35large positive returns to subsidizing
56:37seasonal migration. Okay. So we get
56:40people you know a lot of people take up
56:42these uh these grants or loans and they
56:45send a migrant right and then we observe
56:47that when they do send a migrant their
56:49families do a lot better and they choose
56:50to migrate again. So that's the first
56:52part of the paper. The second part of
56:54the paper then says okay if this is true
56:56that people are migrating when you offer
56:59them the subsidy then why the heck
57:01weren't they migrating to begin with why
57:04did we have to run this program right so
57:05then we have to explain using a model
57:08and in this particular model risk
57:09aversion will play a large role which is
57:11why I chose to cover it in this
57:12particular lecture uh in this model you
57:15know with risk aversion and some
57:18individual specific learning that has to
57:19occur at the destination you can explain
57:21these patterns in the data
57:23Okay, that answers the question. Oh
57:25yeah, people weren't migrating for an
57:27identifiable reason and once we run this
57:29intervention, right, it makes sense that
57:31they started doing it. Okay, so then we
57:35have to go back and then test the model,
57:37right? So given that the model was
57:38written with these facts already in
57:41mind, you can't call them tests of the
57:42model, right? So then we once you write
57:45the model, right, the model has other
57:47implications for what you should see in
57:49the data like the heterogeneity you
57:50should see in the data. So we go back
57:52and test the model. Okay. And then it
57:54turns out that even after you test the
57:56model, this is where the macro comes in.
57:58Okay, it's really difficult to make full
58:01quantitative sense of everything we're
58:03seeing in the data. Right? Essentially,
58:05the issue is that the migration has such
58:07positive returns is very difficult for
58:09us to say, oh yeah, just because people
58:11are risk averse, they must not be doing
58:12it. Yeah. Another way to say it is that
58:14if people were that risk averse, maybe
58:16these are the same people we we should
58:18see not getting up from bed.
58:20Okay. Um and and so we need to go back
58:23and um and um kind of reincquire about
58:28this and and here you know we basically
58:32leave the first paper as um you know
58:35yeah it's it's it's somewhat of a puzzle
58:37still. Okay. And then and we now have a
58:41second paper where we've made a bit of
58:43progress on explaining these results a
58:46little bit more in a quantitative sense.
58:48Okay. [snorts] All right. So here's the
58:50experiments. So we did this with uh in
58:53100 villages, 19 households per village,
58:55right? In some villages, we offered a
58:56grant for people to migrate. So
58:58basically the contract was just this.
59:00Okay, I'll give you 600 taka, right? If
59:03you send a migrant. So only condition is
59:04that you got to try it out. Okay, I'm
59:06not imposing who within the household
59:08should go except that, you know, we want
59:10to prevent child labor, right? Uh I'm
59:13not um saying how long you should go for
59:16where you should go. I'm not trying to
59:17connect you to a specific employer. I
59:19just want you to get get on a bus during
59:21this period. Try it out, right? And if
59:23it doesn't work out, you can come back.
59:24I'm just covering the roundtrip bus
59:26fair. And by the way, if you like it's
59:27600 T at home and if you show up at the
59:29destination, we have people there and
59:31you can get a little bit more money.
59:32Okay, so another 200. Okay, so we do
59:35that in one set of villages, another set
59:37of the exact same contract except that
59:40we do it in the form of a loan rather
59:41than a grant. So you got to pay back in
59:43January during the harvest. Okay, but
59:45zero interest loan, right? And then now
59:47these two of course automatically comes
59:49with some information because you can't
59:50walk into a house and say oh here's
59:52money for you to migrate but I can't
59:54tell you anything about it right so we
59:55have to tell people okay that oh these
59:57are the cities that people often migrate
59:59to we've done surveys in those cities
1:00:01and we see these are the average wages
1:00:02these are the sectors people work in etc
1:00:04right so then the thought was oh maybe
1:00:07that information may have some
1:00:08independent effect on
1:00:11uh on uh on on people's migration
1:00:13behavior so let's test that out so we'll
1:00:15just have uh arm where we just provide
1:00:18that information without the money to
1:00:19see like whether we can generate the
1:00:21benefits through information alone and
1:00:23then there's a control group where we're
1:00:24collecting data and not running those
1:00:25interventions. Okay. All right. And then
1:00:28why this amount of money is because it
1:00:31pays for round trip bus fair plus a few
1:00:33days of food. So we're just insuring the
1:00:34trip. Okay. All right. And then we
1:00:36tracked households in the short and long
1:00:37run. So what did we find? So first thing
1:00:40is that you know 35.9% of people migrate
1:00:43in control group. If you provide just
1:00:45the information 35.9% of people migrate
1:00:48it has a precise zero effect. Okay. So
1:00:50if you think about it like if a third of
1:00:53households are sending a migrant anyway
1:00:55right then Mushri coming in the village
1:00:57and say you can migrate that's not going
1:00:59to do anything because people say that
1:01:01yeah I already knew one of my neighbors
1:01:03migrated. Okay so however if you if you
1:01:07uh give um the loan or the grant your
1:01:10migration rate jumps to like 57 to 60%.
1:01:12Okay, so you're able to induce about 25%
1:01:16additional people to go to the loaner
1:01:18program. Okay, that's result number one.
1:01:20And result number two is that now that
1:01:21we have random variation migration, we
1:01:23can use it to study the effect of
1:01:24migration. And the bottom line is that
1:01:26if you're induced to migrate by the
1:01:27program, then your family consumes 600
1:01:29calories per person per day more. Right?
1:01:32So often during this lean period,
1:01:34they're often reduced to two meals a
1:01:36day. And so essentially, it's moving to
1:01:37three meals a day from two. Okay? That's
1:01:40the bottom line. Like basically you have
1:01:42some positive effects right from from
1:01:44induced migration. And the third thing
1:01:46is that we track people a year later. We
1:01:49don't you know there's a subset of
1:01:50villages we left like we don't mess with
1:01:53them again but you observe even in those
1:01:55villages a lot of the people we induced
1:01:57to migrate choose to remigrate like
1:01:59migration rate remains significantly
1:02:01higher in those villages right so not
1:02:03all 25 extra people went back but like
1:02:0612 out of the 25 did about half of them
1:02:07did choose to go back and this slowly
1:02:10dissipates over time and one clue as to
1:02:13what's happening here is that when we
1:02:14ask people okay why did you go right a
1:02:17lot of the people say they went back and
1:02:19worked for the same employer that they
1:02:20met and suddenly so the kind of the
1:02:21asset that they grow is a connection to
1:02:23an employer but it says okay come back
1:02:24next year like you pro proven yourself
1:02:27to be reliable worker come back next
1:02:28year you can have the same job okay like
1:02:30you go to a construction site there's
1:02:32like some work happening and like the
1:02:35foreman says okay the project will
1:02:37continue next year and uh and just come
1:02:40back okay but these connection seems to
1:02:42decay over time as as projects disappear
1:02:44or something and you have to reestablish
1:02:46them
1:02:47um I have Quick question.
1:02:49Did you when you were collecting the
1:02:51data look at who was migrating within
1:02:53the family?
1:02:54Yes. And it's always Yeah, we we we have
1:02:57a lot of information on that and it's
1:02:58always males who migrate and in the
1:03:02majority of cases like the household
1:03:04head, the adult male, right? Uh I think
1:03:06that's about twothirds of the cases and
1:03:08the rest is like the eldest son.
1:03:11Okay. Yeah. I was wondering because
1:03:13obviously there's a lot of like gender
1:03:14norms in South Asia that prevent like
1:03:16Yeah. So and and again like early on we
1:03:19were talking about how there's many
1:03:20different kinds of migration. So in
1:03:22Bangladesh there's also now a different
1:03:23type of migration of young women going
1:03:25to work in garment factories right but
1:03:27often you see like the women who live
1:03:31within commuting distance of factories
1:03:34have much larger access to these jobs
1:03:36right but in some cases women are
1:03:37actually moving to live in dorms and
1:03:39work.
1:03:40Yeah.
1:03:40But that's like a much more skilled
1:03:42migration than this one. Right. like you
1:03:44really need some literacy, numeracy
1:03:46skills, etc. Whereas the type of
1:03:47migration we're talking about here, the
1:03:48seal migration. Um it's your
1:03:50agricultural workers who are working at
1:03:52construction sites doing things that
1:03:54really use upper body strength, right?
1:03:56Yeah.
1:03:56And and their gender I think gender
1:03:58norms also matter because these men are
1:04:00going into either living at the
1:04:01construction site or like living in um
1:04:05informal housing nearby.
1:04:07And I did have a follow-up question just
1:04:08by the nature of the work that they're
1:04:10doing since they're doing something
1:04:11that's non-aggricultural.
1:04:12Did you see any of them staying more
1:04:15permanently or was it all pure
1:04:17seasonality?
1:04:17Yeah, we'll Yeah, we'll actually come
1:04:19back. No, so we'll we'll come back to
1:04:20this later, but like we you know, we've
1:04:23tracked these households for a long
1:04:24time. So like about 9 years later, we go
1:04:26back and figure out, oh, in the
1:04:27treatment village versus control
1:04:29village, is it any harder for us to find
1:04:30these households now? Right? It turns
1:04:32out that no. Um so there's no permanent
1:04:36um like switch to permanent migration
1:04:38even though like that's that's what my
1:04:40expectation was as well. Yeah,
1:04:42but here's and if anything, we're seeing
1:04:45actually a negative coefficient there,
1:04:46which is that the households that were
1:04:50exposed to seasonal migration and
1:04:51started doing this more, right, are less
1:04:53likely to move away, right? And the way
1:04:56that I'm expose theorizing here to be
1:04:58clear, but the way that I'm I think
1:05:00about it is that like, you know, you've
1:05:02now
1:05:04given them a technology like this asset
1:05:07that they grew, a connection, etc. that
1:05:09makes the rural area more livable. Yeah,
1:05:12because there's a way for me to solve a
1:05:13rural specific problem, right? And so
1:05:15that might be why you're you're not
1:05:17being pushed out as much.
1:05:19Is there I mean is there also from these
1:05:22areas some form of more permanent like
1:05:24rural to urban migration? I'm just
1:05:26curious that's also
1:05:28happening concurrently with this
1:05:30seasonality.
1:05:31AB: Absolutely. So that's the overall
1:05:33pattern not only in Bangladesh but
1:05:35across all of Asia and Africa is
1:05:37urbanization, right? And with that like
1:05:40poverty is also urbanizing. So a lot of
1:05:42people are being pushed out, right? And
1:05:45you see more and more of the poor now
1:05:47living in urban areas, right?
1:05:48Yeah.
1:05:49And the third thing I'll point out is
1:05:51that in little you know in Bangladesh
1:05:54probably in like the Meong Delta in
1:05:55Vietnam, right? There's also a
1:05:57phenomenon of climate migration which is
1:06:00that um the uh this the context has
1:06:05really changed like what used to be
1:06:06profitable is no longer profitable right
1:06:09and so in Bangladesh like there's
1:06:10increased salinity in both southern
1:06:13Bangladesh and in the Mikong Delta and
1:06:15so in Bangladesh the response has been
1:06:17some of it has has been probably
1:06:18migration in drips right but the other
1:06:20has been like changing what you do
1:06:22there's a bit more agriculture shrimp
1:06:23farming etc.
1:06:25Mhm. Oh, thank you.
1:06:27Um, okay. Um,
1:06:29yeah.
1:06:30Uh, aside from the control,
1:06:34um, why did you do the uh the other
1:06:38segments when it's possible that the uh
1:06:42motivation could be a combination of
1:06:44factors?
1:06:46Yeah. So, one reason you know u so
1:06:49there's two ways to answer this. So, why
1:06:51not just have this and this only?
1:06:53Exactly.
1:06:54Okay.
1:06:54Yeah. So the reason
1:06:57a a migrant may be motivated by a
1:07:00combination of factors basically.
1:07:01Sure. So the one reason is that this is
1:07:04actually you should think of this as
1:07:06cash plus information. Okay. So if
1:07:09because the information comes
1:07:10automatically with it. Okay. So if this
1:07:12is cache plus info and I only showed you
1:07:14cach plus info is different from
1:07:16control. You might come back and tell me
1:07:18oh maybe it was just the info. The cash
1:07:19didn't have anything to do with it.
1:07:21Okay. There are some for forms of
1:07:23intersections in
1:07:24Yeah, exactly. So that's why having this
1:07:26is useful for us to isolate the cache
1:07:28dead matter.
1:07:29Thank you. Absolutely. Yeah. Yes.
1:07:32And was there any specific reason for
1:07:34not remigrating like did they specify
1:07:37anything?
1:07:38Uh so I think you know it it would it
1:07:41probably has a lot to do with why they
1:07:43weren't migrating to begin with right.
1:07:45So let me actually get into the model to
1:07:46explain why people might not be
1:07:48migrating. Okay. uh and then that you
1:07:51know I think some of those reasons
1:07:53remain relevant even after the
1:07:55intervention. Okay. So so essentially
1:07:58okay uh I'm going to give you like the
1:08:00in the paper which uh is part of your
1:08:02reading list is a more complicated
1:08:04model. I'm going to give you like the
1:08:05really simplified version of it. Okay.
1:08:07Uh in in two slides okay so imagine that
1:08:10we we have a situation where there's
1:08:11some guaranteed income at home of Y,
1:08:14right? But there's some uncertain
1:08:16returns from migration. So either you
1:08:18get y + g it's like some good outcome or
1:08:21you get y + b a bad outcome and b could
1:08:23be negative. Okay. And there's a
1:08:26probability of the good outcome there's
1:08:28a probability of the bad outcome and the
1:08:31good good outcome is better than the bad
1:08:32outcome. Okay. So so then you have a
1:08:35situation where the good outcome makes
1:08:38you better off relative to staying at
1:08:40home but the bad outcome makes you worse
1:08:41off relative to stay at home. Okay. So
1:08:43now the value functions for m value
1:08:46function for migration would be the
1:08:48probability that you get the good
1:08:49outcome plus the utility associated with
1:08:51it. Right? So the expected returns plus
1:08:53the probability you get the bad outcome
1:08:54times the um uh times the times the
1:08:58return. So now the thing is that when
1:09:00you get the bad return you don't go back
1:09:02again. You learn that it's bad right?
1:09:04Your skills don't fare well at the
1:09:05destination. So in the first year you
1:09:07end up with y plus b but thereafter you
1:09:09just get the home return. Right? That's
1:09:11why you have this complication here.
1:09:13Okay. And then you discount all of it.
1:09:14They won't mess up. Whereas in if you
1:09:17stay at home, you just always stay at
1:09:18home.
1:09:20So in a situation like this, right, you
1:09:22can have a single crossing condition,
1:09:25right? So if B is sufficiently bad.
1:09:27Okay. So the idea is that this is what
1:09:30you know. So here no the concavity of
1:09:32the utility function will play a role.
1:09:34Okay. So what are we seeing? Why are
1:09:36people not migrating here? So what we
1:09:38should observe is really poor people
1:09:39don't migrate. Okay. And what is the
1:09:42intuition for why the model tells you
1:09:44really poor people don't migrate? It's
1:09:46because like when you're extremely poor,
1:09:48okay,
1:09:50if you get the bad outcome, you're in a
1:09:52very steep part of the utility function
1:09:55and the bad outcome could be really bad.
1:09:56So in the actual paper, we have a
1:09:58subsistence constraint, right? So think
1:10:00about so what's happening in that in
1:10:02that model is that you know you're very
1:10:04close to your subsistence constraint,
1:10:05but you're surviving, right? Now imagine
1:10:09that there's a 95% chance you get the
1:10:10good outcome. You think that the city is
1:10:12better, right? I believe these guys the
1:10:14city offer much higher wages. Okay, but
1:10:16if if there's only a tiny tiny chance
1:10:18that there's a bad outcome, but that
1:10:20tiny chance is associated with the out,
1:10:22you know, I'm barely above subsistence
1:10:24and now I hit my subsistence constraint
1:10:26and somebody might die, right? Then I
1:10:28might the utility loss associated with
1:10:30that may be infinite and that dominates
1:10:32my decision-m, right? So just that
1:10:35simple fact will tell you why people may
1:10:38stay away from
1:10:41um from profitable technologies because
1:10:44they're so poor they can't handle that
1:10:46risk. Okay. So this is a this is this is
1:10:48kind of a poverty trap right that you're
1:10:50so poor you're so close to subsistence
1:10:52that you cannot take up technologies
1:10:54that in expectation will make you better
1:10:56off and keeping you poor in the process.
1:10:59Okay so risk aversion can actually lead
1:11:01to poverty traps like this. Okay. So
1:11:03that's the sense in which you know I
1:11:05mean I I but why I chose this paper uh
1:11:07chose this particular paper to talk
1:11:08about with respect to migration and risk
1:11:10like you can migration can be profitable
1:11:13but people just can't take it up
1:11:16so so then like okay so now by by the
1:11:19way I showed you the empirical results
1:11:21and then the model was motivated by the
1:11:23results. So why does this model um
1:11:25rationalize what we saw in the data in
1:11:27the experiment is because what you'd see
1:11:29is that okay there are a bunch of poor
1:11:31people who like if you were to get them
1:11:34to go they would do better like we see
1:11:36in the experiment but they're still not
1:11:38going right and um but then if you give
1:11:44them a if you just give them information
1:11:46it doesn't help because the risk remains
1:11:49right but if you give them cash and you
1:11:51say that look I'm going to put you away
1:11:52from subsistence constraint you can try
1:11:54it out come back you're going to be no
1:11:57worse off right just experiment then
1:11:59they'll go right so that's that's why
1:12:01that's why the experiment works okay so
1:12:04note that if this is the right story
1:12:06then the experiment is saying you know
1:12:08it's not a cash grant that was needed
1:12:10what you needed is insurance the cash
1:12:12grant happens to provide that insurance
1:12:15right so uh as long as you're giving
1:12:18people money without forcing them to pay
1:12:21back
1:12:22right? Uh under bad conditions that's
1:12:24fine. It's just that we happen to
1:12:26provide the money. We're saying you you
1:12:27don't have to pay back at all. Like even
1:12:28the credit like you know it was run by
1:12:30an NGO that's normally poverty focused
1:12:33right. Uh and so there was an
1:12:35understanding that like you know if
1:12:37you're in a lot of trouble there's
1:12:38implicit insurance there. So um whereas
1:12:43you know that might have been too
1:12:45expensive a way to run the program. A
1:12:47better way to run the program might have
1:12:48been uh okay you know what try this out.
1:12:51If it works out, right, you can pay me
1:12:54back, but if it doesn't work out, you
1:12:55don't have to, right? Uh, of course,
1:12:58that that can have moral hazard, so we'd
1:13:00have to think about that carefully. But
1:13:03we did actually run that program, and
1:13:04you see it working exactly the same way.
1:13:07Okay. So sorry and
1:13:10I think honestly that's my concern with
1:13:12the with this kind of social experiment
1:13:15because u it's [laughter]
1:13:20why the experiment when we can actually
1:13:22use a survey method for example just to
1:13:25maybe because the actions are more
1:13:28behavioral than
1:13:30um so so I mean giving people boss
1:13:34ticket to move from one place to the
1:13:35other could just trigger it goes there I
1:13:38I mean the there are other underlining
1:13:40factors which are personal and quite
1:13:43behavioral that could make determine
1:13:45whether the person is going to be
1:13:46successful when they move for one reason
1:13:48or the other.
1:13:50Okay, I understand. Uh so you're
1:13:52actually saying two different things
1:13:53here because there might be other
1:13:54factors, right? But the fact that there
1:13:56might be other factors is precisely why
1:13:57you need to run an experiment.
1:13:59Okay, because if I ask questions,
1:14:00they'll say, "Oh yeah, all these five
1:14:02things matter, right?" whereas the
1:14:03experiment teaches us whether or not
1:14:05there is one particular policy that will
1:14:07be useful. Okay. So, uh this is going to
1:14:09be a longer conversation. I suspect I
1:14:11won't make you happy in the next uh one
1:14:13minute that I'll spend answering your
1:14:14question but we can talk about it later
1:14:16but and I also I also don't want to take
1:14:18up the people's time but let me say the
1:14:20following that I want you to think think
1:14:21through a little bit okay about your
1:14:23concerns about experiment uh in this
1:14:25particular case. So people are
1:14:27seasonally poor their kids are not
1:14:29getting enough to eat. you know, as I
1:14:31said, they're eating two meals a day.
1:14:33People are down to in in the in the data
1:14:35about 14-500 calories per day. Okay. All
1:14:38right. So, so one, you know, you could
1:14:41say, oh, yeah, if you experiment with
1:14:43experiment with people, there might be
1:14:44some ethical complications here. Okay.
1:14:47But the other thing I do want you to
1:14:49think about, I do want everybody to
1:14:50think about is that if I know, right,
1:14:52and this is a problem we've known for
1:14:55known about since the 1970s at least,
1:14:58okay? If I know that kids are not
1:15:01getting enough to eat right and in spite
1:15:04of the fact that the government is
1:15:05running some other programs like free
1:15:07food distribution etc. And in spite of
1:15:08that right I have this data we've had
1:15:12this data here. Sorry.
1:15:15Yeah. That many families are saying
1:15:18they're missing meals. Okay. So there is
1:15:21also an ethical implication of not doing
1:15:24something right. not saying okay let's
1:15:27try out some program which may help
1:15:30right so so just like saying that oh you
1:15:32know social experiment is uh is
1:15:35concerning because it might have some
1:15:37unintended negative effect on some
1:15:39family right not doing something also
1:15:42has guaranteed negative effects on many
1:15:45many families right and there's some e
1:15:47you know there's an ethical obligation
1:15:49there too that we have to we have to be
1:15:50thinking about so I'll just say that and
1:15:52this is something that uh you know uh
1:15:55we'll probably have to think about and
1:15:57and take time to discuss with our
1:15:58friends and family etc.
1:16:05Okay. So, so okay, what's uh I mean
1:16:09what's happening here? Uh I'm just going
1:16:10to explain the model in without any
1:16:12equation or anything. It's like imagine
1:16:14that I asked you to buy a lottery
1:16:17ticket. Okay, where I say look I'm going
1:16:18to sell you this lottery ticket. There's
1:16:20about a 80% chance it'll work out,
1:16:22right? That's actually the success rate
1:16:23that I see in my data. Okay. And the 80%
1:16:26chance that it works out, you know, you
1:16:27end up getting a huge return. You get
1:16:29like a 300 400% return. And there's a
1:16:3220% chance that it won't work out and
1:16:33you lose your you lose your money. And
1:16:35the lottery ticket costs $10, right?
1:16:37Would you buy that ticket?
1:16:40It's a pretty good deal, right? We
1:16:42should be we should be buying that
1:16:43ticket. But the puzzle here is that
1:16:45people aren't buying that ticket. Okay?
1:16:46And the explanation is imagine if I now
1:16:49tell you that yeah it's the same you
1:16:50know it's $10 okay and there's a 80%
1:16:54chance that you'll do better your family
1:16:55will do better right but in the 20%
1:16:57chance you lose your $10 right that
1:17:00might not get you know those $10 may be
1:17:02what uh is the difference between being
1:17:05able to buy food or not
1:17:08right if I tell you that if you're in
1:17:10that situation right or another way for
1:17:12me to do it is to say okay look I'm
1:17:14going to give you a million dollar
1:17:15lottery
1:17:16right? The 80% chance you'll become $5
1:17:18million, but there's a um 20% chance
1:17:22you'll lose a million dollars, right? We
1:17:24might not be able to buy that ticket.
1:17:26And that's what that's what's happening
1:17:28here. Okay? So,
1:17:31uh so then let's see if in the data we
1:17:34observe these these types of um like the
1:17:37implications of this model. So one of
1:17:38the implications of this model is that
1:17:40we should see that the people who are
1:17:42not migrating are the ones who are very
1:17:44close to subsistence. Right? So in the
1:17:47x-axis that's a measure I have that's a
1:17:49measure in the x-axis is what
1:17:53fraction of your total expenditure in
1:17:54the household is just on food. Okay?
1:17:58Because we collect a detailed
1:17:59expenditure. So people out here are are
1:18:01the ones who are like closest citizens.
1:18:03They they have no money other than
1:18:04buying food. All right? So what you see
1:18:08is that and this is the set of people
1:18:10this is a distribution the blue is the
1:18:11distribution of people who migrate and
1:18:12the red is distribution of people who
1:18:14don't right okay right what you observe
1:18:15is that migrant families are away from
1:18:18subsistence right right so if you're
1:18:20very close to subsistence you are more
1:18:22likely to not migrate in the treatment
1:18:25villages we get rid of that so the
1:18:28entire effect is being driven by people
1:18:30who are out here
1:18:32okay exact I mean that's what the model
1:18:34predicts right And um and then you know
1:18:38another way to see this is that if you
1:18:40look at you know the migration rate
1:18:42differences right like you see that
1:18:45under treatment most of the migrants are
1:18:48actually coming from over here where
1:18:49you're spending most of your money on
1:18:51food okay and then there's a you know
1:18:54what uh in the interest of time I'm not
1:18:55going to go through other implications
1:18:57and other tests but there's a large part
1:18:59of the paper that third section of the
1:19:00paper is about thinking through all
1:19:02these different implications of the
1:19:03model and do we see it in the data okay
1:19:05now let me just skip over it. Uh
1:19:09and
1:19:11um and then as I said, you know, it's
1:19:14going to be very difficult for us to
1:19:17uh rationalize everything in the data
1:19:19because I mean people would have to be
1:19:21essentially what the macro style model
1:19:23tells you is that people would have to
1:19:25be extremely riskaverse for us to make
1:19:27sense of all of this why they're not
1:19:28migrating without the treatment but they
1:19:30choose to migrate with the treatment.
1:19:32Okay. So, so, so therefore, right, we
1:19:36need to start thinking about, okay, what
1:19:38else might there be in the model, right,
1:19:41that helps us understand why people are
1:19:43behaving this way. Okay, so you you you
1:19:46know, you can add multiple things to the
1:19:48model and one of the things that seem to
1:19:50make a really big difference and helps
1:19:52us understand this data better is that
1:19:54there are some non-monetary welfare
1:19:57costs of migration that that our data on
1:20:00calories and income gains are not
1:20:01picking up. Right? The idea is that you
1:20:04know you are migrating right and you get
1:20:07more income but you're but this guy has
1:20:09to go and live in an urban slum and it's
1:20:11really unpleasant to be in that slum
1:20:13okay so the returns to migration the
1:20:15aggregate returns is actually not just
1:20:16the monetary returns it's a little bit
1:20:18smaller because of that right or it
1:20:21could lead to family separation right so
1:20:22we basically you know include that in
1:20:25the model and able to explain the data a
1:20:27lot better right and within that then we
1:20:29start collecting data on okay what is
1:20:31that non-monetary this utility. It's not
1:20:33the family separation that doesn't seem
1:20:35to matter to people too much because
1:20:36it's just a few weeks away, right? But
1:20:39what seems to matter is the bad living
1:20:40conditions.
1:20:43Okay? So therefore, if you really do
1:20:44want to solve this problem at large
1:20:46scale, you also have to think about the
1:20:48urban housing problem. Okay. All right.
1:20:51And then uh with let me just spend
1:20:56um a two more minutes and then I'm going
1:20:59to stop uh two more minutes talking
1:21:02about um what uh like when you start
1:21:07thinking about scaling this program and
1:21:08thinking about this as a policy, right?
1:21:11What are the other things that you
1:21:12should worry about? Okay, so first is
1:21:15like for us to understand why were you
1:21:16know we need to understand the question
1:21:17why were people not migrating to begin
1:21:19with? And we think that we have some
1:21:20ideas there. Okay.
1:21:23Now, here's the here are the things that
1:21:25we don't yet have good ideas on. So, one
1:21:26is like if you scale this up, let's say
1:21:29you move from 2,000 households to like
1:21:3030,000. Okay. So, now you're going to be
1:21:33inducing lots of people in the same
1:21:35village to migrate, right? Okay. And
1:21:39when you do that, that might have some
1:21:41effects on that village that we haven't
1:21:43yet thought through. Okay. So for
1:21:45example when people leave okay let's say
1:21:47it's the skilled workers who feel most
1:21:50confident leaving okay and the village
1:21:52is left with just the unskilled people
1:21:54and there the village production
1:21:56function could have some complimentarity
1:21:57between skilled and unskilled workers
1:21:59right in that case we might
1:22:01inadvertently being making the people
1:22:03who cannot migrate we're making them
1:22:04much worse off
1:22:06right and we have to think through that
1:22:08carefully okay second when this is
1:22:12coming back to risk sharing okay so the
1:22:14people who migrate they are in these
1:22:16networks sharing risk with others right
1:22:19okay now if I induce them to migrate and
1:22:21that phrase those risk sharing
1:22:23connections I may be inadvertently
1:22:25making these other people worse off
1:22:28on the other hand right if the if the
1:22:31few weeks of being away the temporary
1:22:32migration does not fray the connection
1:22:35okay then I might actually make the
1:22:38network better off because
1:22:41now some members of the network are
1:22:43drawing their income from a less
1:22:47correlated income source. They're
1:22:48drawing it from the city. And it helps
1:22:50me address the question that Charlene
1:22:52asked right at the beginning, right?
1:22:54Which is how do I address aggregate
1:22:56risk? By diversifying the locations of
1:22:59where people are, I'm actually
1:23:00diversifying the entire network and it
1:23:02might help. Okay? So, it could go either
1:23:05way, right? And then now when you move
1:23:07from from 30,000 to like 300,000 people
1:23:10migrating then you have to start
1:23:12worrying about okay what's going to
1:23:13happen in the city right maybe these
1:23:16guys you know that it will strain the
1:23:17absorptive capacity of the city or these
1:23:19guys will flood the urban slums and it's
1:23:22going to cause sanitation problems etc
1:23:24right or they're going to take away jobs
1:23:26from people who are in the city right
1:23:28there might be more crime like a lot of
1:23:30city residents worry about and complain
1:23:31about migrants right in fact in
1:23:34development policy circles there's a
1:23:36thought of anti-migration bias where
1:23:39policy makers often think about
1:23:40migration as a problem they need to
1:23:42repel.
1:23:43Right? So therefore if you want to
1:23:46really think about the aggregate
1:23:48implications of this and the policy
1:23:49implications of this the research that
1:23:52we did is not enough. You have to
1:23:53actually also think about all these
1:23:56different types of things that happen
1:23:57right including like you know what we
1:24:00added to the model which is oh maybe
1:24:02it's just family separation and then
1:24:04that's going to cause a lot of problems
1:24:05if the father is away from the children
1:24:07right or husbands and wives are away
1:24:09from each other if the father you know
1:24:12takes up a new family in the city that
1:24:14could cause other types of problems that
1:24:15we are not thinking about right so
1:24:17somehow you know we've been doing this
1:24:19research in Nepal Indonesia Bangladesh
1:24:21now in Indonesia That's like always a
1:24:23big concern as soon as we talk about
1:24:25migration like oh what if this guy
1:24:26marries somebody else
1:24:28uh for some reason that's I keep hearing
1:24:31that in Indonesia. Yeah. So what we
1:24:33ended up doing okay so so now Abulad's
1:24:37question might be oh you know what this
1:24:38is this is concerning because all these
1:24:40negative things could happen. Okay. So
1:24:42one way to react to that is like yeah
1:24:45negative things could happen therefore
1:24:46we shouldn't do anything let them be.
1:24:48Okay. Another way to address this is to
1:24:51say, yeah, negative things could happen
1:24:53and we need to do this carefully and um
1:24:56and responsibly and ask all these
1:24:59questions and get rigorous data on it.
1:25:01And that's what we that's what I've been
1:25:03spending my life on for the last like 12
1:25:04years that and raising two kids. So um
1:25:09so there is um so basically we started
1:25:12with this okay the economic direct
1:25:13economic effects on the beneficiaries
1:25:16right then we started writing other
1:25:17papers on like a variety of non-economic
1:25:20effects okay like this utility like what
1:25:23happens with female empowerment so is it
1:25:25the case that when men leave right women
1:25:27are more empowered or is it the case
1:25:29that when men leave and they are now
1:25:31earning a lot more money right that puts
1:25:33women at more risk right because one
1:25:36reason for domestic violence is the
1:25:37asymmetries in economic power within the
1:25:40family. Right. Right. So, we're studying
1:25:42that too. Right. What are hap what's
1:25:45happening to health effects in those
1:25:46families? Right. Is it the guy is
1:25:48bringing back diseases?
1:25:50Right. Uh because I mean in Bangladesh
1:25:52this is not so as much of an issue but
1:25:54like if you were to replicate this in
1:25:55some places in Africa where STDs are
1:25:58much more common than in Bangladesh,
1:25:59right? You might think of that as a
1:26:00first order concern. Okay. Um, and then
1:26:04not only this is on direct
1:26:05beneficiaries, but you also have to
1:26:06think about spillover effects on others
1:26:08in rural areas. And you also have to
1:26:09think about urban spillovers, including
1:26:12the question that I think Sarah asked
1:26:13earlier, which is, is this leading to a
1:26:16fundamental change in where people live?
1:26:18Okay, so essentially this is what we've
1:26:20been doing. And then the idea is that
1:26:23once you uh attack this problem from all
1:26:26these different angles, right? then you
1:26:29might feel comfortable that you've now
1:26:32done a thorough and responsible job of
1:26:35telling advising policy makers on oh
1:26:37here are the positive here are the
1:26:38negative effects right and if you want
1:26:41to think of migration and occupational
1:26:44or ris divers or spatial diversification
1:26:47as a solution to poverty and seasonal
1:26:50poverty and risk in rural areas right
1:26:52these are the concerns you might have so
1:26:54one concern I would definitely have is
1:26:56that if you're going to pursue
1:26:58strategies like this, we do need to
1:26:59invest in urban housing. Okay. And
1:27:02again, Abu Dwad, the the my reaction is
1:27:06not, yeah, urban housing is a problem
1:27:09and people are upset about it and
1:27:11therefore we shouldn't do this. My my my
1:27:13reaction is there's a lot of seasonal
1:27:16hunger and this is bad for kids in rural
1:27:18areas. We need to address it and if we
1:27:20want to address it well, then we also
1:27:22need to invest in urban housing.
1:27:25Mushik.
1:27:26Yeah,
1:27:26we should probably wrap up to give a
1:27:28little bit of time for questions.
1:27:29Oh, yeah. Yeah, sure. Sorry.
1:27:30Especially from the people that are in
1:27:32the uh webinar.
1:27:33Okay.
1:27:34Yeah. And I'm I'm happy to stay stay
1:27:36back a bit and and keep answering
1:27:38questions.
1:27:43Okay.
1:27:47How do how do we pick
1:27:49um
1:27:51u are you Oh, I didn't mean to say you
1:27:53had to stop talking at that exact
1:27:55minute, but I mean if
1:27:58I I was actually done.
1:27:59Oh, you were done. Okay. Um if we have a
1:28:01couple questions from the front row and
1:28:02then we'll see if there's from the Q&A.
1:28:08I have a couple about the model in your
1:28:11paper.
1:28:12Um I'm just wondering what your thoughts
1:28:14are on the space in the model for some
1:28:17sort of uncertainty aversion on top of
1:28:19the risk aversion or like there's a
1:28:21distribution over the mu g and mub but
1:28:25also kind of what if the perceived
1:28:29probabilities aren't the truth like if
1:28:31migrants don't completely share their
1:28:33experience with great
1:28:35job searching or wages. Yeah. Uh both
1:28:38are things we thought through, right? So
1:28:40one is on incorrect beliefs. So you'll
1:28:42see in like the fourth section of the
1:28:44paper and the paper is just ridic. It's
1:28:46like a it's like a book now 80 80 pages
1:28:49and um so I'm um yeah so this is not
1:28:52like a reaction that you should have
1:28:54read the paper because like I I don't
1:28:58need to wish that on you but to towards
1:29:00the end you'll see that we we thought
1:29:02about like oh are there like
1:29:03systematically incorrect beliefs etc.
1:29:05and we actually collected data on that
1:29:07and that turns out to be not true,
1:29:09right? People generally had the right
1:29:11belief. So that's that's not something
1:29:13that would really help us explain what's
1:29:15going on here. And then it's also the
1:29:17case that other than a subsistence
1:29:19constraint either through uncertainty
1:29:21aversion or through like loss aversion,
1:29:24right? You could explain the uh the data
1:29:27better. And in fact, Michael Kramer,
1:29:29Gautam Ralph in their um they have a
1:29:32nice review paper on behavioral
1:29:33economics and development and they take
1:29:35our paper and model and they have an
1:29:37explanation for it that's based on loss
1:29:39aversion, right? And the thing is that
1:29:41yeah that that can also explain the
1:29:43data. The problem is that that's not how
1:29:45we designed the experiment to get at
1:29:46that particular behavioral factor. So
1:29:48that was like a more of the art of
1:29:50choosing what to put in a model where
1:29:52you know if if it's something that
1:29:54wasn't really testable, right? We we
1:29:56didn't prioritize that.
1:29:58Thanks.
1:30:02I have a question
1:30:05of the live. Please go ahead.
1:30:06Yeah. Uh I just wanted to go back to the
1:30:08results. So around transportation costs.
1:30:11I was wondering if you could say
1:30:13something about it especially on the
1:30:14selection of those um um 1900 um uh
1:30:20households whether there was some
1:30:21heterogeneity in transportation costs in
1:30:24which case you know the 600 takas could
1:30:26maybe give some sort of heterogeneity in
1:30:29intensive margin on the of the treatment
1:30:31or was it the case that they were
1:30:33roughly kind of the same facing the same
1:30:35transportation cost?
1:30:36So within a village they were roughly
1:30:38facing the same transportation cost.
1:30:40Okay. But but I but there's um uh you
1:30:44know more useful answers to your
1:30:45question. So one in terms of just
1:30:46selection of the of the people who
1:30:48participated. Uh so this is a problem
1:30:51remember on people who are landless
1:30:53right like those whose like labor demand
1:30:55falls during the lean period. So we
1:30:59targeted people who are landless. So
1:31:01think about a poor region of the country
1:31:02and the bottom 60% of population there
1:31:05right entirely landless folks. Okay. And
1:31:08uh and on transportation costs, what
1:31:10else in the variation, what else was I
1:31:12going to say? Um yeah. Okay. So what I
1:31:16was going to say is that in terms of
1:31:17like with the multiple stages of scale
1:31:20up, what we've learned is that you know
1:31:23if we are if risk aversion is really the
1:31:26problem here, right? Then it makes no
1:31:28sense for us to run this program in
1:31:31areas that are like like within an hour
1:31:33of the city, right? Okay. Where the
1:31:35transportation costs are low. And the
1:31:36reason is that I'm saying that oh people
1:31:38are too scared to experiment here right
1:31:40when are they too scared to experiment
1:31:42if there's a real loss associated with
1:31:43it. If you're in an hour from the city
1:31:45one day when you're not so busy you
1:31:46might go in and try to find work at the
1:31:48construction site right for those types
1:31:50of people we don't need to come up with
1:31:52an external intervention right whereas
1:31:55you know where the types of locations
1:31:57we're working we precisely always look
1:31:59for places where it's too far for you to
1:32:01just like solve the problem for
1:32:03yourself.
1:32:07Thank you.
1:32:09Uh I have a quick question about like I
1:32:12was just thinking about the sustain like
1:32:13a sustainable solution and the credit
1:32:16market seems like a plausible way. So I
1:32:18was just wondering in the credit whether
1:32:20it just like how was the repayment rate
1:32:23or was default more
1:32:26so yeah two answers. So one is that um
1:32:29we see the success rate of migration is
1:32:31about 80%. By success rate I mean like
1:32:3480% of people earned enough like found
1:32:35the job quickly enough and earned enough
1:32:37such that it was a sensible thing for
1:32:38them to do. Okay. Um and you know
1:32:41depending on like what measure you use
1:32:43like you can directly ask them oh did
1:32:44you earn more than you thought you would
1:32:46or whatever right you get success rates
1:32:48of somewhere between 70 and 80%. And the
1:32:50credit repayment rate was exactly 80%.
1:32:54Okay. Um so um yeah he's been very
1:32:59patient. has been like in front of me
1:33:01and not not bothering me at all and
1:33:03until the end of the lecture. Uh so uh
1:33:07so yeah so that so I think okay so
1:33:09that's one answer and the second answer
1:33:11trill is that um
1:33:14you know when you're thinking about the
1:33:16scalability of this yes I mean if you do
1:33:18it as a credit program it does um of
1:33:22course you know that money can be
1:33:23recycled into the program the repayment
1:33:25because 80% of people repay but on the
1:33:27other hand when I actually started
1:33:29working on the implementation scale up
1:33:31because I also wear different hats not
1:33:33as a researcher but actually working on
1:33:35this program, right? What we learn is
1:33:37that like the way that micro credit
1:33:39organizations are set up, right? It
1:33:41doesn't really make much sense for them
1:33:43to spend a lot of time collecting a
1:33:45onetime repayment of $10. Like it costs
1:33:47us something like $12 to to to get back
1:33:52that $10, right? So So then you might
1:33:55just end up running it as a grand
1:33:57program.
1:34:00Um
1:34:04Can I ask a question about
1:34:06sorry go ahead.
1:34:07Okay. Uh so uh two questions from from
1:34:11the urban urban like the the the city
1:34:14perspective. First it's do you see the
1:34:19total number of say residents including
1:34:22the migrants in the cities being
1:34:25fluctuated uh in with the seasonal
1:34:29patterns or across different years. And
1:34:31second
1:34:33um you you mentioned about the the urban
1:34:36living condition as a as a important
1:34:39factor uh affecting whether uh migrants
1:34:42will uh get their utility from. I was
1:34:45wondering if you seen responses from
1:34:48migrants either from a qualitative or
1:34:52qualitative that other aspects of public
1:34:55services publicities
1:34:57um affecting their the migrants
1:34:59decision. Yeah. And and these two can be
1:35:02you know interdependent as the if if the
1:35:06total amount of people uh demanding for
1:35:08certain kind of services is fluctuating
1:35:10a lot it's going to be hard for local
1:35:12government to you know set up a capacity
1:35:16um to meet those demands that's
1:35:18fluctuating across different time.
1:35:20Yeah. So uh so first of all yes uh in
1:35:24general I mean not from just from this
1:35:26program which is relatively small scale
1:35:28but in general given the urbanization
1:35:30the nature of urbanization in the
1:35:31country there's definitely seasonal
1:35:33fluctuation in the population in cities
1:35:35right and this has been a well-known
1:35:37fact for like at least 20 years as I was
1:35:39growing up in Dhaka that there were
1:35:41certain parts of the year that's saw saw
1:35:44huge influxes because remember a third
1:35:46of households were already sending
1:35:47migrants right that's in the control
1:35:49village right and that's been like we've
1:35:50been tracking that for 10 years that's
1:35:52always been true. Third of households
1:35:53send migrant and they send it during a
1:35:55period they send the migrant during a
1:35:57period when there's not much to do in
1:35:58the city. to this particular season,
1:36:00right? So you observe that like at
1:36:02constru, you know, if you talk to
1:36:04construction site foremen or if you talk
1:36:05to or like rickshaw garage owners,
1:36:08right? There are certain months of the
1:36:10year when business is better than
1:36:11others, right?
1:36:13[snorts]
1:36:13And and so u like the urban housing
1:36:16right uh it does need to pay attention
1:36:19right to to that like regardless of my
1:36:22any intervention we do right which is
1:36:24that this is uh there are periods when
1:36:27when the city's absorptive capacities
1:36:29will need more strain than others right
1:36:31and then we when we did we asked about
1:36:34um like in this other paper the the
1:36:36paper with lagos we we did like go back
1:36:39and ask migrants about like what are the
1:36:41key deterrents Right? And that's where
1:36:44housing came up. And the key aspect of
1:36:46housing that they came up is access to
1:36:47sanitation. Right? So they are willing
1:36:50to give up a lot more in wages as long
1:36:53as they can get access to a house where
1:36:56they might have like indoor sanitation
1:36:58as opposed to like trying to figure out
1:37:00where to go.
1:37:02I have a quick question about um I guess
1:37:05Sarah when Radika was trying to ask a
1:37:07question before. Let me take her and
1:37:08then I'll I'll take your
1:37:11question. My question was about uh
1:37:13postmigration of climate refugees. So
1:37:16like where situations where it's not
1:37:18driven by selection on who migrates um
1:37:20and then where they migrate to but also
1:37:22like where the entire family migrates um
1:37:25and also like grow you know we're
1:37:27talking about kids going malnutrition
1:37:30like growing up as an urban poor in
1:37:32Islam versus rural poor. Um is there a
1:37:36good book on that or who would you
1:37:38recommend?
1:37:38Yeah. Uh no, I think this uh I think the
1:37:42area of climate migration is something
1:37:45that yes, you should work on. It's uh
1:37:48there's not much work in that area. You
1:37:51it's it's a it's an important topic.
1:37:53It's going to become increasingly
1:37:54important and uh and there's not that
1:37:57much going on yet.
1:38:00And like and I guess some of the like
1:38:02where should you do that work? I suspect
1:38:04the places where it' be like this is my
1:38:07guess uh where it would be most
1:38:09effective to do that work right now are
1:38:11like low-lying delta small island
1:38:13nations right where which which are
1:38:15already experiencing some of the effects
1:38:19right of changes in sea sea level rise
1:38:21because I think sea level rise has been
1:38:22one of the primary effects that we've
1:38:24experienced so far and the secondary
1:38:26effect that we're beginning to
1:38:27experience is probably weather pattern
1:38:29changes
1:38:30so I said just to better like just a
1:38:33follow on that question. I think my
1:38:34question was more about like uh not just
1:38:37like the primary household um head
1:38:39migrating but also like families
1:38:42migrating.
1:38:43Yeah.
1:38:43Um and I think that's I mean that's
1:38:47partly driven by primate triages but
1:38:50like there's also places where like the
1:38:51entire family moves.
1:38:53Sure. So I mean if uh so there's
1:38:56definitely going to be an interaction
1:38:57with uh you know climate migration and
1:39:00whether families move, right? But but if
1:39:02I were to just take your question as oh
1:39:04let's look at um sort of differential
1:39:07effects of whole family move versus
1:39:09single person moving right then already
1:39:12you we have lots of variation in that
1:39:14one could play with like there are some
1:39:16countries that explicitly don't allow
1:39:18you to move with your family like South
1:39:19Asians like the biggest migration
1:39:21corridor in the world are South Asians
1:39:23working in richer Asian countries okay
1:39:27and at and and most of those job
1:39:29categories you're explicitly not allowed
1:39:31to bring your family, right? If you go
1:39:33to the UAE or Qatar or Malaysia or
1:39:35Singapore, right, you have to live in
1:39:36migrant dorms, right? You're not allowed
1:39:38to bring in and and then even for
1:39:40high-skll migrants into those countries,
1:39:42right? There's no path to citizenship or
1:39:44residency. And so even if you bring your
1:39:46family because you're, you know, as a
1:39:48doctor, engineer, you go work in these
1:39:50countries, right? Is your your kids are
1:39:52not allowed to remain. Okay? So there's
1:39:54a lot of variation in that already going
1:39:55on, right? And the second on the climate
1:39:57migration side itself, right? So one you
1:40:00know like when you say whole family
1:40:02moves I think you have in mind that oh
1:40:04the the place has become unlivable
1:40:07right and therefore everybody has to
1:40:09move right now because your land is now
1:40:10flooded right but that's not the
1:40:12situation we are in right now thankfully
1:40:14that's not the modal situation the
1:40:16situation we are in right now is that
1:40:18because of changes in weather patterns
1:40:20or your you know too much flooding or
1:40:23your crop getting destroyed because of
1:40:24salinity right it's not that the place
1:40:27has become completely unlivable it has
1:40:29become uh a place that where your
1:40:31economic livelihood strategies don't
1:40:33work anymore,
1:40:35right? So, so that doesn't necessitate
1:40:37like full full family migration. Yep.
1:40:40Okay. Sarah,
1:40:42um
1:40:44just we'll take Sarah's question and
1:40:45we'll have to end it because it's
1:40:46getting late in the day in London and I
1:40:49don't want to keep Mandy and Lauren and
1:40:50and Kirsty work too long. Thanks.
1:40:54All right. Um thank you. Um so my
1:40:56question was actually on like wage
1:40:58premiums. So I'm wondering if if I mean
1:41:01I guess I don't know how much you've
1:41:02done in like data collection in the
1:41:04urban environment but are I mean oftent
1:41:07times right because um migrants are
1:41:10temporary they're paid either a a lower
1:41:12wage or they're you know like there's
1:41:14price wage discrimination between
1:41:17permanent urban residents and then like
1:41:19the migrants and I was wondering if you
1:41:21were observing that and if I mean
1:41:23clearly it's still profitable above the
1:41:25rural wages but I was wondering if there
1:41:28is like some wage discrimination between
1:41:30the permanent.
1:41:31Yeah. So, I don't know if it's wage
1:41:32discrimination. I wouldn't jump to that
1:41:34word, but it is definitely true that
1:41:36there are
1:41:38gaps in average wages that a seasonal
1:41:41migrant earns versus a city resident
1:41:43earns versus somebody who is uh like not
1:41:48necessarily a city resident, but a
1:41:49long-term res like migrant of the city,
1:41:51right?
1:41:52And that's like aggregate or is that um
1:41:55like for the same job? Are they Do you
1:41:58have that data?
1:41:58Uh yeah, I I I think like if you were I
1:42:02this is a guess. I I don't have a lot of
1:42:04data right on it, but I think
1:42:06it's like if you were to decompose that
1:42:07wage gap, right?
1:42:09Um you'll be able to explain some of it
1:42:11by looking at the type of work people
1:42:14are doing. So just as an example, like
1:42:16within a construction site,
1:42:18right, migrant seasonal migrants are
1:42:20coming in. They don't quite know how to
1:42:23use the concrete mixer,
1:42:25right? And so they have lower wage jobs
1:42:27whereas the people who are there they
1:42:29have some skills that's specific to that
1:42:31particular site. Okay. That explains the
1:42:33wage gap. But even if you condition on
1:42:35that I suspect you'll see a bit of a
1:42:38wage difference.
1:42:39Yeah.
1:42:39And one reason you'll see the wage
1:42:41difference though is that the seasonal
1:42:43migrants may get their wages in the form
1:42:46of like oh I can sit I can stay in this
1:42:48construction site
1:42:50that takes care of my housing. So I'm
1:42:51getting like a compensation in a
1:42:53different form. So it's also very
1:42:54difficult for us to
1:42:56do apples to apples comparisons.
1:42:59Great. Thank you.
1:43:02Thank you all. Um especially Mushik. I
1:43:05also I should have given thanks to uh
1:43:07Lauren and Kirsty Lauren Wearing Kirsty
1:43:10McMill and Mandy Chen for helping
1:43:12organize from CR. Why don't we all
1:43:15unmute and give Mushik a round of
1:43:17applause.
1:43:22I I like that. We should do that in all
1:43:24Zoom meetings. So like at least one
1:43:26piece of the real world that that we can
1:43:29replicate. Thank you very much. Oh, it
1:43:31was it was my pleasure. Really great
1:43:33questions. Thank you for your engagement
1:43:34and as I said, you know, please do feel
1:43:36free to email and you know and Nate is
1:43:40going to host the follow-up session on
1:43:42Monday. Uh so attend that. Uh
1:43:44we have Ben Mole talking about models of
1:43:47um heterogeneous agents which arise with
1:43:49when you have risk uh tomorrow. So well
1:43:52hopefully we'll see you all there
1:43:53tomorrow. Byebye.
1:43:54And sorry just a one thing to to add. I
1:43:56I actually gave the wrong date. It's on
1:43:57Tuesday as the email sent out by Lauren.
1:43:59So I'll I'll see all of you on Tuesday
1:44:01that come. Sorry about that.
1:44:04Thanks everyone. Thanks Mushik.

About this transcript

This page contains the full transcript of Supplemental lecture: Migration and risk by CEPR & VideoVox Economics, generated from the public captions YouTube serves with the video. The transcript has 19,595 words across 2,864 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to choose a caption track available on the video.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.