Popular in Social Sciences

Books

City Time: On Being Sentenced to Rikers Island

Independent Scholar David A. Campbell

Culture and Society

The City as a Growth Machine: Toward a Political Economy of Place

New York University Harvey Moltoch

Economics

Gender and Economics

University of California Santa Barbara Shelly Lundberg

American Politics

Keeping the Republic

Boston College Marc Landy

Banks and Policy Organisations

Greening the financial sector

European Investment Bank Matteo Ferrazzi

British Politics

Entrepreneurial thinking: the politics and practice of policy impact

University of Sheffield Matthew Flinders

Culture and Society

Carceral Entrapments: Views from the Prison/Street Interface in India

Indian Institute of Technology Delhi Mahuya Bandyopadhya

The rise and fall of SES gradients

View more insights in Social Sciences - Economics - Health Economics

View more figures from University of California Los Angeles

View the page of Adriana Lleras-Muney

Read the Study

Image courtesy of interviewee. January 29, 2024

Generating Workspace

Export as PDF

This video video discusses the significant disparities in health outcomes between individuals from high socioeconomic status (SES) families and those from low SES families, particularly focusing on how these differences manifest from birth and continue to widen throughout childhood into adulthood. The speaker emphasizes that while newborns exhibit similar characteristics regardless of family income, disparities in health and development emerge as they grow.

One of the key approaches in the research is to utilize height as a measurable indicator of health across different ages, as it allows for more accurate comparisons over time than various other health metrics. The speaker references existing studies that indicate a steepening gradient in health disparities as children age, particularly highlighting that these gaps often lessen during adolescence due to various physiological factors, including the onset of puberty.

Furthermore, the text mentions research showing that while wealthier children generally start growing earlier and may stop sooner, poorer children can continue growing for longer, leading to a possible "catch-up" effect during adolescence. It discusses the complexities of child growth and health and suggests that adolescence is a critical period where interventions could potentially be effective in addressing health disparities.

The speaker concludes by reflecting on the relative lack of research focused on adolescence compared to early childhood development, suggesting a growing interest in understanding this phase's significance and potential for intervention. Overall, the text emphasizes the importance of socioeconomic factors in health outcomes and the potential for targeted interventions during key developmental periods.

Language:

Export as PDF

X-TIMESTAMP-MAP:LOCAL=00:00:00.000,MPEGTS=900000

00:00.605 --> 00:04.155
There is a very famous set of results out there.

00:04.535 --> 00:06.235
Um, let's just start with this.

00:07.015 --> 00:10.435
Uh, when you look at adults, there's a huge difference

00:11.015 --> 00:14.395
in all health outcomes between individuals

00:14.455 --> 00:16.355
who come from high SES families

00:16.355 --> 00:18.355
and those who come from low SES families.

00:18.535 --> 00:19.795
So rich people

00:20.015 --> 00:22.995
and poor people, they live very different lives

00:23.495 --> 00:26.395
and they have very different life expectancies

00:26.395 --> 00:28.675
with the rich, living much longer than the poor.

00:29.255 --> 00:32.495
And is it about the money that they have today

00:32.635 --> 00:34.655
or is it about the money that they had growing up?

00:34.725 --> 00:38.015
Well, there's a famous study that showed

00:38.285 --> 00:40.655
that when you look at babies at at birth,

00:41.005 --> 00:42.695
they don't look that different.

00:43.725 --> 00:45.015
They, they do look different,

00:45.035 --> 00:49.415
but at birth many things are similar between rich and poor.

00:50.195 --> 00:53.335
And then over time, as the children grow up,

00:53.625 --> 00:56.775
these differences between them start to emerge

00:56.875 --> 00:58.455
and they get bigger and bigger

00:58.555 --> 01:01.815
and bigger so that by the time you look at them as adults,

01:02.315 --> 01:05.695
there are gigantic, gigantic gaps.

01:05.995 --> 01:10.385
So, there's this kind

01:10.385 --> 01:12.625
of like the gradient is known as

01:12.625 --> 01:14.985
that the gradient is this expression to refer

01:14.985 --> 01:17.225
as the difference between the people at the top

01:17.245 --> 01:18.305
and the people at the bottom.

01:18.405 --> 01:20.345
And it's talked about as a gradient

01:20.345 --> 01:23.825
because there's, graded differences.

01:23.925 --> 01:27.545
So a little bit more income is associated

01:27.545 --> 01:28.545
with a little better health,

01:28.645 --> 01:30.785
and a little bit more is associated with even more.

01:31.245 --> 01:35.105
So there's kind of like a graded difference, by income.

01:35.485 --> 01:38.905
And you can, you can trace that back to childhood.

01:39.525 --> 01:41.945
And so somehow during the childhood process,

01:42.035 --> 01:43.225
these differences emerge.

01:43.445 --> 01:44.905
So this is where a study starts.

01:44.935 --> 01:47.865
There's this observation that there are large differences

01:47.925 --> 01:48.985
and they emerge at birth.

01:49.525 --> 01:52.785
And so the question is how do these differences,

01:53.165 --> 01:54.265
why do they emerge?

01:54.885 --> 01:58.345
And is there, and more importantly, is there some way to,

01:59.165 --> 02:00.985
offset that process?

02:01.355 --> 02:03.305
Right? And,

02:04.365 --> 02:07.105
and so the idea for this paper, this study is,

02:07.725 --> 02:08.785
is to look at height.

02:09.285 --> 02:13.985
So it, a very complicated issue here is

02:13.985 --> 02:15.265
that it's very difficult

02:15.285 --> 02:17.265
to compare people over the life course.

02:17.325 --> 02:18.505
So how do you measure health?

02:19.165 --> 02:22.625
You know, well, when you're a baby, you look at,

02:24.235 --> 02:26.215
the color of their skin.

02:26.475 --> 02:30.295
You look at birth weight when you are like a 50-year-old,

02:30.435 --> 02:33.175
you look at blood pressure, you know,

02:33.585 --> 02:35.615
those are not the same measures, right?

02:36.155 --> 02:39.255
And so how do you, how do you trace this?

02:39.435 --> 02:43.175
You need some way of tracing the same outcomes over time in

02:43.175 --> 02:44.415
a way that's comparable.

02:45.195 --> 02:49.985
And one such outcome that you can look at is, is, is height.

02:50.765 --> 02:54.345
And so you can look at how shorter

02:54.845 --> 02:57.385
or tall somebody is when they're babies

02:57.645 --> 02:58.785
and when they're, and

02:58.785 --> 03:00.325
When they are adults.

03:00.705 --> 03:04.325
And so we thought that this would provide a very nice,

03:04.585 --> 03:09.245
way of, looking at what happens as pop as populations,

03:09.945 --> 03:11.285
as they, as they age.

03:12.295 --> 03:15.515
And so in this paper, we take advantage of the fact

03:15.515 --> 03:18.515
that this is a, this kind of data is available everywhere.

03:18.535 --> 03:20.675
So just to go back to the previous literature,

03:20.845 --> 03:23.995
these literature kind of tracing these, these, these gaps

03:24.095 --> 03:27.755
and how they emerge often looks at self-reported measures.

03:27.755 --> 03:30.995
So they ask the mom, is your kid in good health?

03:31.375 --> 03:33.395
And then, you know, your kid could be three months old

03:33.395 --> 03:35.235
or your kid could be 15 months old,

03:35.495 --> 03:37.275
and then they answer whatever the answer.

03:37.295 --> 03:39.835
And these measures, even though

03:39.855 --> 03:41.755
for some people might think they're suspect

03:41.755 --> 03:43.515
because they're self-reported, that they,

03:43.745 --> 03:45.075
they are quite predictive.

03:45.075 --> 03:48.195
Like if a mom were to say that a kid is not in good health,

03:49.145 --> 03:52.235
chances are that kid in fact will be,

03:52.415 --> 03:54.715
having more diseases would be more likely

03:54.715 --> 03:56.395
to be hospitalized for various reasons.

03:56.535 --> 04:00.195
So, so there is content to these answers, but,

04:00.655 --> 04:02.715
but it's hard to know how to scale them.

04:02.785 --> 04:04.995
It's hard to know if people answer it the same way

04:05.815 --> 04:07.675
across populations and over time.

04:07.975 --> 04:09.475
And so the advantage of height is

04:09.475 --> 04:12.995
that we actually have measurements of people,

04:13.535 --> 04:16.475
we take them regularly when you go to the doctor,

04:16.575 --> 04:19.315
and we have them for many people around the world.

04:20.255 --> 04:22.435
So that's what we wanted to do in this paper.

04:22.695 --> 04:24.995
Now, a curious,

04:25.175 --> 04:28.715
and difficult thing in this paper is that,

04:28.985 --> 04:32.155
there's a lot more data for children under the age of five

04:32.825 --> 04:35.675
than for children after, you know, so we,

04:35.735 --> 04:38.675
we know a lot less too about what happens

04:38.805 --> 04:40.355
after the age of five and,

04:41.645 --> 04:45.545
and a key issue for us, it was trying to understand

04:45.855 --> 04:48.905
what happens after the age of, after the age of five.

04:50.035 --> 04:51.695
And so I can come back talking about this,

04:51.715 --> 04:53.015
but this is where the paper starts.

04:53.035 --> 04:55.855
We have heights, it's measured, it's objective,

04:56.005 --> 04:57.975
it's comparable across all ages,

04:58.555 --> 05:01.375
and it's available for many, many people around the world.

05:01.715 --> 05:03.135
So we could study this question.

05:03.895 --> 05:06.655
I think for us, the starting point is the seminal paper by,

05:07.155 --> 05:11.215
case in Pax and, and Luki.

05:11.635 --> 05:14.535
And so they used two data sets from the United States.

05:15.155 --> 05:19.055
Uh, they're maybe more than two data sets, in fact.

05:19.395 --> 05:21.055
Um, and they

05:22.605 --> 05:25.945
looked at the average health as reported by the mother

05:25.965 --> 05:30.425
of children as a function of the, family income

05:30.805 --> 05:33.185
of the kid average over a number of years.

05:33.845 --> 05:38.145
And then they plotted the relationship between income

05:38.805 --> 05:41.665
and health, at different ages.

05:42.045 --> 05:45.145
And what they observed is that, as I mentioned

05:45.145 --> 05:48.025
before, if you look at very young children,

05:48.315 --> 05:50.945
there is a relationship where,

05:51.625 --> 05:54.585
children born in richer families are in better health than

05:54.745 --> 05:56.465
children born to poorer families.

05:56.845 --> 05:59.725
But that relationship gets steeper over time.

06:00.465 --> 06:02.325
And so these gaps between the rich

06:02.325 --> 06:05.285
and the poor, they just grow and grow and grow.

06:05.625 --> 06:09.085
And they con they observe that growth continue

06:09.975 --> 06:14.005
until age 65, at which point they start to,

06:14.735 --> 06:16.685
these gaps kind of start to diminish.

06:17.225 --> 06:19.005
And they did this with different data sets

06:19.005 --> 06:22.645
that have different advantages, and disadvantages.

06:22.665 --> 06:24.925
But, but these are the patterns that they observe.

06:24.945 --> 06:26.405
And then this,

06:26.595 --> 06:29.725
this study was replicated subsequently in Canada

06:30.665 --> 06:33.845
by another set of excellent researches, Janet Curry and,

06:33.965 --> 06:34.965
and Mark Stabe.

06:35.225 --> 06:38.885
And they kind of found, that this was also true there,

06:39.015 --> 06:40.325
which is also interesting

06:40.325 --> 06:43.645
because, you know, kind of has different healthcare system

06:43.825 --> 06:45.525
and a, safety net.

06:46.025 --> 06:49.325
Um, and so there's been a lot of, a lot

06:49.325 --> 06:51.525
of work looking into, into this.

06:51.525 --> 06:55.685
So that's for us, this is the, the, the part of the part is,

06:56.065 --> 06:59.045
these observations that were made by,

06:59.145 --> 07:00.325
by these seminal studies.

07:00.855 --> 07:03.365
There are like three parts to this study.

07:04.225 --> 07:07.085
Um, the first takes advantage of the demographic

07:07.085 --> 07:10.645
and health surveys, which are a set of surveys conducted

07:10.645 --> 07:12.405
around the world in middle

07:12.465 --> 07:15.285
and low income countries where a set

07:15.285 --> 07:18.525
of surveyors interview representative samples

07:18.665 --> 07:19.765
of the population.

07:20.305 --> 07:21.645
And during that interview,

07:21.645 --> 07:24.085
they also take measurements of height and weight.

07:25.025 --> 07:27.045
Now, most of the height

07:27.045 --> 07:28.725
and weight that's available is

07:28.725 --> 07:30.565
for children under the age of five.

07:31.385 --> 07:35.445
In some of these surveys, they also measure the adults,

07:35.585 --> 07:36.965
the moms and the dads.

07:37.435 --> 07:41.525
With this data, we could look to see what happens

07:42.685 --> 07:44.425
to these differences between the rich

07:44.425 --> 07:47.305
and the poor with the age of the children.

07:47.565 --> 07:50.625
Now I should say, these surveys are not panel surveys.

07:50.625 --> 07:52.865
They're not following the same people over time.

07:53.325 --> 07:55.785
So all that we can see is, is the gap

07:55.785 --> 07:57.065
between the rich and the poor.

07:58.665 --> 08:02.525
Bigger as for older children than younger children.

08:02.905 --> 08:06.365
But we have data from millions of people around the world

08:07.155 --> 08:08.285
that we can look at.

08:08.705 --> 08:12.085
And so with this data, what we show is that in fact,

08:12.085 --> 08:14.005
we can replicate the patterns that,

08:14.155 --> 08:16.525
that the original studies found.

08:17.705 --> 08:19.045
We don't have income

08:19.045 --> 08:22.085
because you know, when you look at places like, like India

08:22.305 --> 08:26.325
or many countries in Africa, in these are countries

08:26.445 --> 08:29.565
that are still heavily agricultural for which a notion

08:29.565 --> 08:33.085
of income is, is difficult to income,

08:33.085 --> 08:34.525
is difficult to measure and obtain.

08:34.585 --> 08:37.085
So instead what we're doing is we're looking at maternal

08:37.155 --> 08:39.045
education, but our results are not

08:39.045 --> 08:40.205
particularly sensitive to this.

08:40.205 --> 08:43.125
We can use maternal education or paternal education.

08:43.125 --> 08:44.965
We can also use a measure of assets

08:45.595 --> 08:47.445
that is collected in the, in the data.

08:48.225 --> 08:49.805
So just think about,

08:50.755 --> 08:54.885
broadly speaking about the better off and less well

08:54.965 --> 08:55.965
Off families. And

08:55.965 --> 08:59.005
so in this data we can see very strongly

08:59.395 --> 09:03.365
that there are some small gaps

09:03.395 --> 09:04.485
between the better off

09:04.485 --> 09:07.805
and the worse off at age under at the age of one.

09:08.265 --> 09:10.805
And then by the, by the time we look at age five,

09:11.075 --> 09:13.645
there's a substantial gap in heights

09:13.795 --> 09:15.205
between, between children.

09:16.305 --> 09:19.085
Now, now the question is kind of

09:19.085 --> 09:20.085
what happens afterwards.

09:20.085 --> 09:21.685
So there were two issues with this.

09:22.225 --> 09:24.045
Uh, with this data,

09:24.375 --> 09:26.245
we're not following the same people over time

09:26.545 --> 09:28.405
and we cannot continue to follow children

09:28.695 --> 09:29.925
until they stop growing.

09:30.105 --> 09:33.165
So people stop growing sometime

09:34.075 --> 09:35.605
late in late adolescence

09:35.665 --> 09:37.845
or sometimes even into young adulthood.

09:38.025 --> 09:41.365
But we really wanna follow them until say, age 20 something.

09:42.955 --> 09:47.215
So in the second part of the study, we, made use of

09:47.985 --> 09:51.455
these five panel data sets where

09:52.145 --> 09:55.135
these are now surveys in which people go

09:55.635 --> 09:58.575
and they interview the same set of families over time,

09:59.215 --> 10:02.005
multiple times for many years,

10:02.705 --> 10:05.045
and they take in measurements of the children

10:05.265 --> 10:06.325
as the children grow.

10:07.335 --> 10:12.155
And so in this data, we then observe

10:13.405 --> 10:17.715
two things for the children under the age of five,

10:18.055 --> 10:21.075
we see the same pattern as they grow up,

10:21.375 --> 10:23.115
the gaps in height emerge.

10:23.855 --> 10:26.275
But the most interesting thing that we find is that

10:26.985 --> 10:31.075
when we look at children in adolescence, some

10:31.075 --> 10:33.595
of these gaps actually shrink a bit.

10:34.535 --> 10:37.435
And so that was very surprising to to us.

10:38.315 --> 10:40.215
And it's very interesting also

10:40.215 --> 10:42.815
because there is this question of like,

10:42.835 --> 10:45.215
can we do something about these, these gaps?

10:46.035 --> 10:49.135
And so lemme just back out a little bit

10:49.155 --> 10:50.215
and talk a little,

10:50.355 --> 10:53.015
and talk about, the physiology of height.

10:54.285 --> 10:57.385
If you have children, you may remember

10:58.175 --> 11:00.825
that in fact they grow a lot when they're babies

11:01.845 --> 11:03.265
and then they continue to grow.

11:03.925 --> 11:07.225
But then there's an important growth spurt

11:07.225 --> 11:10.705
that happens in adolescence when people go through puberty.

11:11.845 --> 11:16.025
And so, there are in fact these growth curves

11:16.025 --> 11:19.025
that were developed,

11:19.935 --> 11:21.065
that describe this.

11:21.065 --> 11:25.265
And so the rate at which you grow is very fast when you're

11:25.265 --> 11:26.625
an infant and then it falls

11:26.725 --> 11:29.065
and then you grow at kind of a constant rate.

11:29.485 --> 11:32.225
And then you have that growth spur in adolescence when you

11:32.225 --> 11:34.705
grow fast again, and then you stop growing.

11:35.805 --> 11:38.465
And so it was really important to look at

11:38.465 --> 11:39.505
that adolescent period

11:39.505 --> 11:42.985
because that's when another maybe 20 to 30%

11:42.985 --> 11:45.305
of your final height gets determined.

11:46.895 --> 11:51.035
And what we found is that this gap between the better off

11:51.035 --> 11:54.085
and the worse off it rises continuously.

11:54.505 --> 11:56.725
And then at some point it kind of shrinks

11:57.105 --> 11:59.165
and we do still see differences,

12:00.165 --> 12:03.185
but at the end when people have stopped growing,

12:03.885 --> 12:06.225
but they're smaller than they, than their ones were.

12:06.725 --> 12:10.675
And so in, in this data, this is, kind

12:10.675 --> 12:12.355
of like a second finding here,

12:12.355 --> 12:16.515
because now we are tracking the same families

12:16.575 --> 12:18.235
and we are tracking them all the way

12:18.285 --> 12:19.355
until they stop growing.

12:19.695 --> 12:21.675
And so we can see the full evolution

12:21.895 --> 12:23.715
of these, of these patterns.

12:24.055 --> 12:26.315
And then we, we discover that,

12:26.625 --> 12:30.795
that these gaps are somehow attenuated in

12:30.795 --> 12:32.195
that second growth spurt.

12:32.955 --> 12:35.375
And so then we try to investigate, well,

12:35.515 --> 12:38.255
why are they attenuated and isn't this interesting?

12:38.635 --> 12:40.615
And can we leverage that for policy?

12:41.595 --> 12:44.375
And, and so we kind of look at two sets

12:44.375 --> 12:45.735
of explanations there.

12:46.925 --> 12:49.985
Uh, one has to do with how adolescents are treated,

12:50.605 --> 12:52.665
and the second has to do with, again,

12:52.665 --> 12:54.025
the physiology of height.

12:54.165 --> 12:58.815
And so what we end up finding, which is interesting

13:00.275 --> 13:03.485
also, is that a lot of these differences have to do

13:03.485 --> 13:06.685
with the onset of purity.

13:06.945 --> 13:09.805
So what is happening is that,

13:10.585 --> 13:14.685
and so it's very difficult to measure the onset of puberty.

13:14.685 --> 13:16.725
If you have children, you observe it yourself,

13:17.075 --> 13:20.765
like the boys just get taller and they start getting hair.

13:20.825 --> 13:22.045
And there are all these markers.

13:22.305 --> 13:24.885
But in surveys, you know, we don't really have that.

13:25.065 --> 13:27.245
So, but we, what we do have

13:27.715 --> 13:30.885
that is well measured is the age at which girls

13:31.175 --> 13:32.405
start to menstruate.

13:33.675 --> 13:38.015
And that doesn't, that's not the start of, puberty,

13:38.235 --> 13:40.255
but it is a good marker in

13:40.255 --> 13:43.415
that it tells you if things are happening earlier or later.

13:43.715 --> 13:46.175
And that has been changing very substantially

13:46.395 --> 13:48.935
around the world and in the past two centuries.

13:49.275 --> 13:53.545
So there are very large, differences across countries

13:53.545 --> 13:57.705
and within countries, in terms of when that happens.

13:58.855 --> 14:03.315
And so it turns out then a lot of developing countries,

14:04.255 --> 14:09.195
children from advantage backgrounds have,

14:09.495 --> 14:11.995
their onset of puberty earlier.

14:12.895 --> 14:14.195
And so they stopped growing

14:15.095 --> 14:17.955
and the children who are from poor disadvantaged

14:17.955 --> 14:19.315
backgrounds continue to grow.

14:20.095 --> 14:23.975
And so there's a certain amount of catch up that happens

14:24.235 --> 14:27.375
as a result of this, as in, in some sense it's mechanical.

14:27.915 --> 14:29.895
Now what's interesting about this is that

14:31.195 --> 14:33.535
in rich countries, this is not true.

14:34.035 --> 14:37.175
So what's not true is that,

14:38.155 --> 14:42.975
in rich countries, poor girls have their adolescents, their,

14:43.115 --> 14:44.855
menstruation earlier.

14:45.715 --> 14:48.455
And so there is been kind of a reversal

14:48.475 --> 14:49.535
of that relationship.

14:49.715 --> 14:50.715
So

14:51.115 --> 14:53.125
This kind of brings up a set of puzzles

14:53.315 --> 14:55.045
that are out there in the literature.

14:55.265 --> 14:57.565
So if you look back in the 19th century,

14:58.625 --> 15:01.645
the our best estimates suggest that the onset

15:01.645 --> 15:05.685
of menstruation for girls was around 15 to 16 years of age.

15:06.145 --> 15:09.645
And the onset of menstruation today in rich countries is

15:09.645 --> 15:13.045
something closer to 12 to 13, maybe even less.

15:13.505 --> 15:15.885
So, the estimates are that the onset

15:15.885 --> 15:18.925
of menstruation has fallen very substantially

15:18.985 --> 15:21.165
and it continues to go down.

15:21.945 --> 15:24.805
And so it's, and it's not well understood why

15:24.805 --> 15:28.605
that's occurring, but it's continuing to occur in the US

15:28.625 --> 15:31.285
and it's happening also in Europe and in many countries.

15:31.705 --> 15:34.525
And so there's been this transition where, people

15:35.185 --> 15:38.485
are becoming adults earlier, we don't know why.

15:39.225 --> 15:42.405
And in that transition, there's been a reversal,

15:42.505 --> 15:46.525
in the relationship between socioeconomic status

15:47.225 --> 15:48.565
and, and puberty.

15:48.825 --> 15:52.575
So it used to be that the rich got

15:53.365 --> 15:54.695
into adulthood earlier,

15:55.635 --> 15:59.535
and now the opposite is true, at least in rich countries.

15:59.535 --> 16:02.455
So in the poor countries that we look at, what occurs is

16:02.455 --> 16:04.255
that because the poor

16:05.035 --> 16:07.935
or disadvantage continued to grow for longer,

16:08.245 --> 16:13.015
that allows them some amount of catch up, supposedly a,

16:13.075 --> 16:17.535
I've been hypothesized to be, a mechanism

16:18.005 --> 16:22.215
that poorly fed malnourished populations tend to grow

16:22.315 --> 16:26.385
for longer than well fed populations.

16:26.405 --> 16:28.185
And that's maybe the explanation.

16:28.725 --> 16:31.545
Um, so ultimately we find that a lot

16:31.545 --> 16:33.865
of these reduction in these gaps happens

16:34.095 --> 16:37.945
because of when the onset of adolescence occurs.

16:38.605 --> 16:41.025
We had a very large set of hypotheses

16:41.025 --> 16:43.945
because in many countries, as soon as people

16:45.255 --> 16:49.035
or in traditional societies anyway, as soon as people went

16:49.035 --> 16:50.595
through the adolescent growth spur

16:50.615 --> 16:54.555
or started going to, that they were eligible for,

16:55.345 --> 16:56.515
they were entering adulthood

16:56.735 --> 16:58.235
and they would be treated very differently.

16:58.335 --> 17:01.435
And so we had hypothesized that maybe,

17:01.435 --> 17:04.755
because of this differential treatment, that

17:04.755 --> 17:07.275
that would somehow lead to smaller gaps.

17:07.375 --> 17:08.875
And so that maybe there was something

17:08.875 --> 17:11.355
that the poor was doing that was beneficial for the children

17:11.695 --> 17:13.955
or something that the rich were doing that was detrimental

17:13.975 --> 17:17.235
to their children, that somehow was leading to these,

17:17.415 --> 17:20.075
to these, declines in this gap.

17:20.855 --> 17:24.925
Um, and unfortunately we did not find any evidence

17:25.705 --> 17:28.965
except for one thing that I should mention briefly, which is

17:28.965 --> 17:33.565
that we did, did find that in some countries,

17:34.065 --> 17:37.005
so in some countries, girls are married very early as soon

17:37.005 --> 17:38.045
as they have their period,

17:38.385 --> 17:41.445
and that we did find is detrimental for their height.

17:42.145 --> 17:45.405
Um, but that practice is not more common

17:45.455 --> 17:46.805
among the rich than the poor.

17:46.945 --> 17:48.645
So it doesn't ultimately explain

17:49.235 --> 17:50.645
Kind of our, our results.

17:51.325 --> 17:52.685
I would say two things.

17:52.905 --> 17:56.615
One is that in the last maybe 20 years

17:56.675 --> 17:58.735
or something, there's been a tremendous amount

17:58.735 --> 18:02.695
of focus on the first years of life, like the first three,

18:02.695 --> 18:07.135
maybe the first five years because of this, you know,

18:08.815 --> 18:11.665
because of all these findings that have come to show

18:11.665 --> 18:12.705
that even by the age

18:12.705 --> 18:15.065
of five there are already these very large gaps.

18:15.165 --> 18:19.065
And, and, and we, we, our work is consistent with this.

18:19.325 --> 18:22.145
Um, but I think one thing that's interesting,

18:22.145 --> 18:25.145
and I think our work shows this in a, in a lot of

18:25.765 --> 18:27.905
new work is starting to focus on this.

18:27.935 --> 18:31.785
It's just the importance of the adolescent period for, for,

18:31.885 --> 18:32.945
for outcomes and

18:33.045 --> 18:35.825
how there are many things that change for people.

18:35.925 --> 18:38.025
So I think for me, one of the things

18:38.025 --> 18:42.065
that I think is important is to understand that yes, a lot

18:42.065 --> 18:44.705
of important things are determined by age five,

18:44.885 --> 18:48.225
but not everything is determined by age five, thankfully.

18:48.845 --> 18:51.145
And it seems like adolescence is a very

18:51.545 --> 18:52.625
critical period as well.

18:53.045 --> 18:55.585
And there seems to be at least the possibility

18:55.585 --> 18:58.545
that this is a window of opportunity during which

18:59.135 --> 19:01.505
some other interventions can occur

19:01.805 --> 19:05.065
and they can meet maybe quite successful.

19:05.765 --> 19:08.265
So I don't know, we don't have evidence on

19:08.265 --> 19:09.625
what those interventions could be,

19:09.645 --> 19:12.785
but at least we do observe that something can happen

19:13.085 --> 19:16.105
and that these in these gaps could,

19:16.155 --> 19:19.145
could at least shrink in, in principle during this period.

19:19.765 --> 19:22.345
The second thing that I would say about it is

19:22.345 --> 19:27.245
that there is this need to understand these issues

19:27.245 --> 19:28.245
around puberty.

19:28.245 --> 19:32.005
And I think there's been a lot written about this in terms

19:32.005 --> 19:33.565
of like menstruation for example,

19:33.595 --> 19:35.725
like understanding why these, this is happening

19:35.725 --> 19:37.165
and this, is this a good development

19:37.165 --> 19:38.445
or is this a bad development?

19:38.445 --> 19:40.365
Like what are the underlying causes

19:40.945 --> 19:43.765
of these changes in the onset of, of menstruation

19:43.785 --> 19:47.325
and there's, and and say similarly for, for puberty.

19:47.705 --> 19:51.405
And so I think that kind of raises a, a, a puzzle there.

19:51.545 --> 19:53.965
Um, but I think there's more, there is more work

19:53.965 --> 19:57.005
that's being done by a lot of people trying to look at

19:57.525 --> 19:59.485
interventions, for example, in adolescents,

19:59.485 --> 20:01.845
but it's a nascent literature.

20:01.865 --> 20:06.285
So in some sense we open up more questions than we answer.

20:06.665 --> 20:09.045
If you go back and you look at the literature in

20:09.045 --> 20:13.005
adolescence, it is, I think substantially smaller

20:13.595 --> 20:15.885
than the literature that's devoted to like

20:16.705 --> 20:19.205
the childhood period, you know, up

20:19.265 --> 20:20.685
to age five and all of this.

20:20.705 --> 20:24.605
So there's, I think, much better understanding, of

20:24.605 --> 20:26.405
what happens in those first five years

20:26.465 --> 20:27.925
and much less understanding.

20:28.265 --> 20:30.685
And I think for a long time, and for example, Heckman

20:30.685 --> 20:34.615
and his group have, I think put out a lot of ideas

20:34.615 --> 20:37.175
that have helped emphasize this early childhood

20:37.175 --> 20:38.575
period as being very important.

20:38.635 --> 20:43.285
But I think we're now having more studies that are showing

20:43.315 --> 20:46.405
that things during adolescents can be done,

20:46.405 --> 20:47.605
that will matter.

20:48.105 --> 20:49.525
But it is very new.

20:49.905 --> 20:53.245
So we just don't know enough about what kind

20:53.245 --> 20:57.285
of interventions can be successful during this time period.

00:00 - There is a very famous set of results out there. let's just start with this. when you look at adults, there's a huge difference in all health outcomes between individuals who come from high SES families and those who come from low SES families.

00:18 - So rich people and poor people, they live very different lives and they have very different life expectancies with the rich, living much longer than the poor. And is it about the money that they have today or is it about the money that they had growing up? Well, there's a famous study that showed that when you look at babies at at birth, they don't look that different. They, they do look different,

00:45 - but at birth many things are similar between rich and poor. And then over time, as the children grow up, these differences between them start to emerge and they get bigger and bigger and bigger so that by the time you look at them as adults, there are gigantic, gigantic gaps. So, there's this kind of like the gradient is known as that the gradient is this expression to refer as the difference between the people at the top

01:17 - and the people at the bottom. And it's talked about as a gradient because there's, graded differences. So a little bit more income is associated with a little better health, and a little bit more is associated with even more. So there's kind of like a graded difference, by income. And you can, you can trace that back to childhood. And so somehow during the childhood process, these differences emerge.

01:43 - So this is where a study starts. There's this observation that there are large differences and they emerge at birth. And so the question is how do these differences, why do they emerge? And is there, and more importantly, is there some way to, offset that process? Right? And, and so the idea for this paper, this study is, is to look at height.

02:09 - So it, a very complicated issue here is that it's very difficult to compare people over the life course. So how do you measure health? You know, well, when you're a baby, you look at, the color of their skin. You look at birth weight when you are like a 50-year-old, you look at blood pressure, you know, those are not the same measures, right? And so how do you, how do you trace this?

02:39 - You need some way of tracing the same outcomes over time in a way that's comparable. And one such outcome that you can look at is, is, is height. And so you can look at how shorter or tall somebody is when they're babies and when they're, and When they are adults. And so we thought that this would provide a very nice, way of, looking at what happens as pop as populations, as they, as they age.

03:12 - And so in this paper, we take advantage of the fact that this is a, this kind of data is available everywhere. So just to go back to the previous literature, these literature kind of tracing these, these, these gaps and how they emerge often looks at self-reported measures. So they ask the mom, is your kid in good health? And then, you know, your kid could be three months old or your kid could be 15 months old, and then they answer whatever the answer. And these measures, even though

03:39 - for some people might think they're suspect because they're self-reported, that they, they are quite predictive. Like if a mom were to say that a kid is not in good health, chances are that kid in fact will be, having more diseases would be more likely to be hospitalized for various reasons. So, so there is content to these answers, but, but it's hard to know how to scale them. It's hard to know if people answer it the same way

04:05 - across populations and over time. And so the advantage of height is that we actually have measurements of people, we take them regularly when you go to the doctor, and we have them for many people around the world. So that's what we wanted to do in this paper. Now, a curious, and difficult thing in this paper is that, there's a lot more data for children under the age of five than for children after, you know, so we,

04:35 - we know a lot less too about what happens after the age of five and, and a key issue for us, it was trying to understand what happens after the age of, after the age of five. And so I can come back talking about this, but this is where the paper starts. We have heights, it's measured, it's objective, it's comparable across all ages, and it's available for many, many people around the world. So we could study this question.

05:03 - I think for us, the starting point is the seminal paper by, case in Pax and, and Luki. And so they used two data sets from the United States. they're maybe more than two data sets, in fact. and they looked at the average health as reported by the mother of children as a function of the, family income of the kid average over a number of years. And then they plotted the relationship between income and health, at different ages.

05:42 - And what they observed is that, as I mentioned before, if you look at very young children, there is a relationship where, children born in richer families are in better health than children born to poorer families. But that relationship gets steeper over time. And so these gaps between the rich and the poor, they just grow and grow and grow. And they con they observe that growth continue until age 65, at which point they start to,

06:14 - these gaps kind of start to diminish. And they did this with different data sets that have different advantages, and disadvantages. But, but these are the patterns that they observe. And then this, this study was replicated subsequently in Canada by another set of excellent researches, Janet Curry and, and Mark Stabe. And they kind of found, that this was also true there, which is also interesting

06:40 - because, you know, kind of has different healthcare system and a, safety net. and so there's been a lot of, a lot of work looking into, into this. So that's for us, this is the, the, the part of the part is, these observations that were made by, by these seminal studies. There are like three parts to this study. the first takes advantage of the demographic and health surveys, which are a set of surveys conducted

07:10 - around the world in middle and low income countries where a set of surveyors interview representative samples of the population. And during that interview, they also take measurements of height and weight. Now, most of the height and weight that's available is for children under the age of five. In some of these surveys, they also measure the adults,

07:35 - the moms and the dads. With this data, we could look to see what happens to these differences between the rich and the poor with the age of the children. Now I should say, these surveys are not panel surveys. They're not following the same people over time. So all that we can see is, is the gap between the rich and the poor. Bigger as for older children than younger children. But we have data from millions of people around the world

08:07 - that we can look at. And so with this data, what we show is that in fact, we can replicate the patterns that, that the original studies found. We don't have income because you know, when you look at places like, like India or many countries in Africa, in these are countries that are still heavily agricultural for which a notion of income is, is difficult to income, is difficult to measure and obtain.

08:34 - So instead what we're doing is we're looking at maternal education, but our results are not particularly sensitive to this. We can use maternal education or paternal education. We can also use a measure of assets that is collected in the, in the data. So just think about, broadly speaking about the better off and less well Off families. And so in this data we can see very strongly

08:59 - that there are some small gaps between the better off and the worse off at age under at the age of one. And then by the, by the time we look at age five, there's a substantial gap in heights between, between children. Now, now the question is kind of what happens afterwards. So there were two issues with this. with this data,

09:24 - we're not following the same people over time and we cannot continue to follow children until they stop growing. So people stop growing sometime late in late adolescence or sometimes even into young adulthood. But we really wanna follow them until say, age 20 something. So in the second part of the study, we, made use of these five panel data sets where these are now surveys in which people go

09:55 - and they interview the same set of families over time, multiple times for many years, and they take in measurements of the children as the children grow. And so in this data, we then observe two things for the children under the age of five, we see the same pattern as they grow up, the gaps in height emerge. But the most interesting thing that we find is that when we look at children in adolescence, some

10:31 - of these gaps actually shrink a bit. And so that was very surprising to to us. And it's very interesting also because there is this question of like, can we do something about these, these gaps? And so lemme just back out a little bit and talk a little, and talk about, the physiology of height. If you have children, you may remember that in fact they grow a lot when they're babies

11:01 - and then they continue to grow. But then there's an important growth spurt that happens in adolescence when people go through puberty. And so, there are in fact these growth curves that were developed, that describe this. And so the rate at which you grow is very fast when you're an infant and then it falls and then you grow at kind of a constant rate. And then you have that growth spur in adolescence when you

11:32 - grow fast again, and then you stop growing. And so it was really important to look at that adolescent period because that's when another maybe 20 to 30% of your final height gets determined. And what we found is that this gap between the better off and the worse off it rises continuously. And then at some point it kind of shrinks and we do still see differences, but at the end when people have stopped growing,

12:03 - but they're smaller than they, than their ones were. And so in, in this data, this is, kind of like a second finding here, because now we are tracking the same families and we are tracking them all the way until they stop growing. And so we can see the full evolution of these, of these patterns. And then we, we discover that, that these gaps are somehow attenuated in

12:30 - that second growth spurt. And so then we try to investigate, well, why are they attenuated and isn't this interesting? And can we leverage that for policy? And, and so we kind of look at two sets of explanations there. one has to do with how adolescents are treated, and the second has to do with, again, the physiology of height. And so what we end up finding, which is interesting

13:00 - also, is that a lot of these differences have to do with the onset of purity. So what is happening is that, and so it's very difficult to measure the onset of puberty. If you have children, you observe it yourself, like the boys just get taller and they start getting hair. And there are all these markers. But in surveys, you know, we don't really have that. So, but we, what we do have that is well measured is the age at which girls

13:31 - start to menstruate. And that doesn't, that's not the start of, puberty, but it is a good marker in that it tells you if things are happening earlier or later. And that has been changing very substantially around the world and in the past two centuries. So there are very large, differences across countries and within countries, in terms of when that happens. And so it turns out then a lot of developing countries, children from advantage backgrounds have,

14:09 - their onset of puberty earlier. And so they stopped growing and the children who are from poor disadvantaged backgrounds continue to grow. And so there's a certain amount of catch up that happens as a result of this, as in, in some sense it's mechanical. Now what's interesting about this is that in rich countries, this is not true. So what's not true is that, in rich countries, poor girls have their adolescents, their,

14:43 - menstruation earlier. And so there is been kind of a reversal of that relationship. So This kind of brings up a set of puzzles that are out there in the literature. So if you look back in the 19th century, the our best estimates suggest that the onset of menstruation for girls was around 15 to 16 years of age. And the onset of menstruation today in rich countries is

15:09 - something closer to 12 to 13, maybe even less. So, the estimates are that the onset of menstruation has fallen very substantially and it continues to go down. And so it's, and it's not well understood why that's occurring, but it's continuing to occur in the US and it's happening also in Europe and in many countries. And so there's been this transition where, people are becoming adults earlier, we don't know why. And in that transition, there's been a reversal,

15:42 - in the relationship between socioeconomic status and, and puberty. So it used to be that the rich got into adulthood earlier, and now the opposite is true, at least in rich countries. So in the poor countries that we look at, what occurs is that because the poor or disadvantage continued to grow for longer, that allows them some amount of catch up, supposedly a, I've been hypothesized to be, a mechanism

16:18 - that poorly fed malnourished populations tend to grow for longer than well fed populations. And that's maybe the explanation. so ultimately we find that a lot of these reduction in these gaps happens because of when the onset of adolescence occurs. We had a very large set of hypotheses because in many countries, as soon as people or in traditional societies anyway, as soon as people went through the adolescent growth spur

16:50 - or started going to, that they were eligible for, they were entering adulthood and they would be treated very differently. And so we had hypothesized that maybe, because of this differential treatment, that that would somehow lead to smaller gaps. And so that maybe there was something that the poor was doing that was beneficial for the children or something that the rich were doing that was detrimental to their children, that somehow was leading to these,

17:17 - to these, declines in this gap. and unfortunately we did not find any evidence except for one thing that I should mention briefly, which is that we did, did find that in some countries, so in some countries, girls are married very early as soon as they have their period, and that we did find is detrimental for their height. but that practice is not more common among the rich than the poor. So it doesn't ultimately explain

17:49 - Kind of our, our results. I would say two things. One is that in the last maybe 20 years or something, there's been a tremendous amount of focus on the first years of life, like the first three, maybe the first five years because of this, you know, because of all these findings that have come to show that even by the age of five there are already these very large gaps. And, and, and we, we, our work is consistent with this.

18:19 - but I think one thing that's interesting, and I think our work shows this in a, in a lot of new work is starting to focus on this. It's just the importance of the adolescent period for, for, for outcomes and how there are many things that change for people. So I think for me, one of the things that I think is important is to understand that yes, a lot of important things are determined by age five, but not everything is determined by age five, thankfully.

18:48 - And it seems like adolescence is a very critical period as well. And there seems to be at least the possibility that this is a window of opportunity during which some other interventions can occur and they can meet maybe quite successful. So I don't know, we don't have evidence on what those interventions could be, but at least we do observe that something can happen and that these in these gaps could,

19:16 - could at least shrink in, in principle during this period. The second thing that I would say about it is that there is this need to understand these issues around puberty. And I think there's been a lot written about this in terms of like menstruation for example, like understanding why these, this is happening and this, is this a good development or is this a bad development? Like what are the underlying causes

19:40 - of these changes in the onset of, of menstruation and there's, and and say similarly for, for puberty. And so I think that kind of raises a, a, a puzzle there. but I think there's more, there is more work that's being done by a lot of people trying to look at interventions, for example, in adolescents, but it's a nascent literature. So in some sense we open up more questions than we answer. If you go back and you look at the literature in adolescence, it is, I think substantially smaller

20:13 - than the literature that's devoted to like the childhood period, you know, up to age five and all of this. So there's, I think, much better understanding, of what happens in those first five years and much less understanding. And I think for a long time, and for example, Heckman and his group have, I think put out a lot of ideas that have helped emphasize this early childhood period as being very important.

20:38 - But I think we're now having more studies that are showing that things during adolescents can be done, that will matter. But it is very new. So we just don't know enough about what kind of interventions can be successful during this time period.

Faculti AI can make mistakes