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

Impact of English instruction on labor market outcomes

View more insights in Social Sciences - Economics - Labor Economics

View more figures from University of Chicago

View the page of Oscar Galvez-Soriano

Read the Study

Is exposure to a foreign language conducive to better labor market outcomes? Oscar Galvez-Soriano discusses the effect of exposure to a foreign language in school on student achievement and labor market outcomes.

Image courtesy of interviewee. November 30, 2023

Generating Workspace

Export as PDF

This video video discusses a research project aimed at examining the impact of English instruction in Mexico's elementary schools on labor market outcomes. The study stems from two observations: a widespread belief in non-English speaking countries that English proficiency leads to better job opportunities and the implementation of English instruction in developing countries over the past 25 years.

The researchers note that while there is substantial literature on the returns of English language skills in English-speaking countries, there is a lack of evidence from non-English speaking nations, especially Mexico. They constructed a unique database linking elementary school students to labor market outcomes over ten years after exposure to English instruction.

The research highlights key issues, including the variability in the adoption of English instruction across different schools and cohorts, which could bias results. The findings indicate that the average effect of English exposure on wages is negligible, but for high-achieving individuals, the effect is positive. The researchers suggest that the benefits may stem from improved job prospects in roles requiring English.

Two principal mechanisms are explored: first, that English instruction does not significantly enhance other cognitive skills but does improve English proficiency, and second, that students exposed to English may opt for less physically demanding jobs. The researchers acknowledge that their work is ongoing and aim to refine their estimates further. Overall, they express enthusiasm about the research's potential insights into educational policies and labor market dynamics in Mexico.

Language:

Export as PDF

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

00:00.325 --> 00:02.385
We came up with this research idea

00:02.675 --> 00:04.745
after observing two interesting facts.

00:05.205 --> 00:07.465
So the first fact is that

00:08.595 --> 00:10.665
among non-English speaking countries

00:11.365 --> 00:14.745
is there is this common belief that the individuals

00:14.745 --> 00:19.665
who speak English are more likely to find very big jobs,

00:20.005 --> 00:22.865
or that to find very labor opportunities, right?

00:23.145 --> 00:25.825
Actually, we have this saying in Latin America

00:25.895 --> 00:27.185
that English opens doors.

00:28.125 --> 00:31.865
And the, I will say for the second fact,

00:33.165 --> 00:36.065
in the last 25 years, there have been efforts

00:36.755 --> 00:41.145
among developing countries to introduce English instruction

00:41.405 --> 00:43.785
as a subject in their public education systems.

00:44.605 --> 00:47.025
And they do this under the assumptions,

00:47.265 --> 00:50.385
a very similar assumption that the individuals

00:50.385 --> 00:52.865
who have exposure to English instruction will eventually

00:53.295 --> 00:54.665
acquire English abilities.

00:54.965 --> 00:58.265
And also when they incorporate to a labor market,

00:58.265 --> 01:00.985
they will find, very, paid jobs

01:01.085 --> 01:02.425
and very labor opportunities.

01:02.885 --> 01:04.585
The thing is that we don't have

01:05.055 --> 01:07.025
council evidence on this claim.

01:07.405 --> 01:10.825
So with this paper, we, we try to provide some

01:10.825 --> 01:14.465
of the first evidence from a developing non-English speaking

01:14.465 --> 01:15.665
country, which is Mexico.

01:16.235 --> 01:20.865
There is actually a large literature starting the returns

01:20.965 --> 01:22.865
to English language skills,

01:23.085 --> 01:26.465
but in the context of English speaking countries, right?

01:26.565 --> 01:28.985
And the classical references are she weak

01:28.985 --> 01:32.585
and Miller in for their contributions in the 1990s and

01:33.035 --> 01:34.825
after that, Blakely

01:34.845 --> 01:38.665
and Chin in 2004, with one of the first council,

01:38.985 --> 01:42.785
evidence on the effects of English skills on earnings,

01:43.245 --> 01:45.225
but among immigrants in the us.

01:45.605 --> 01:47.345
So this is important because most of the

01:48.145 --> 01:50.265
existing literature focus on this context

01:50.325 --> 01:52.505
of immigrants in English speaking countries.

01:52.675 --> 01:56.025
While there is just a few number of papers in the context

01:56.205 --> 01:58.105
of non-English speaking countries,

01:58.165 --> 02:01.745
and, even among these other papers, most

02:01.745 --> 02:04.105
of them focus in former British colonies.

02:04.905 --> 02:07.285
Uh, for example, we have the case of Asam Chin

02:07.285 --> 02:12.245
and Prakash, eh, they offer some evidence in the context

02:12.305 --> 02:15.885
of India and also Ericsson for the context of South Africa.

02:16.265 --> 02:19.445
But again, the thing is that these two economies are quite

02:19.445 --> 02:21.645
different from most of the developing work

02:21.645 --> 02:25.165
because of their status of former British colons.

02:25.215 --> 02:28.965
There are two important aspects that I would like to mention

02:29.115 --> 02:32.805
regarding the, the, the policy background

02:32.825 --> 02:35.965
and also the institutional, background in Mexico.

02:36.585 --> 02:40.325
The first one is that in, in Mexico, K

02:40.325 --> 02:44.525
to 12 schooling is divided in three levels, right?

02:44.895 --> 02:46.885
Elementary school,

02:47.525 --> 02:50.205
covering from first grade to sixth grade.

02:50.355 --> 02:54.325
Then middle school that goes from seventh to ninth,

02:54.785 --> 02:58.805
and finally a high school that goes from 10th to

02:58.885 --> 02:59.885
12th grade.

03:00.925 --> 03:05.375
Um, the second important aspect is that in, in Mexico,

03:05.595 --> 03:09.295
the enrollment rates in elementary school are high,

03:09.955 --> 03:11.935
at about 95%.

03:12.645 --> 03:17.055
Then for it drops to middle school to 85%.

03:17.525 --> 03:21.095
Then, it drops even more for high school and,

03:21.155 --> 03:24.775
and college going to 70 and then 30%.

03:25.235 --> 03:26.295
Now, this is relevant

03:26.295 --> 03:29.095
because in this paper, the policy that we,

03:29.605 --> 03:33.375
that we studied was implemented at the elementary school

03:33.375 --> 03:37.655
there, where the enrollment rates are particularly high.

03:38.275 --> 03:41.735
In this research, we construct a novel database

03:42.865 --> 03:44.655
connecting the universe

03:44.915 --> 03:47.055
of elementary school students in Mexico

03:47.595 --> 03:50.455
to the labor market outcomes more than 10 years

03:50.465 --> 03:52.495
after these kids have exposure

03:52.555 --> 03:55.895
to English instruction in school school,

03:55.895 --> 03:59.175
and we construct this dataset using three

03:59.175 --> 04:00.215
sources of information.

04:00.675 --> 04:02.215
The first one is test court data,

04:02.685 --> 04:04.255
then the Mexican school census,

04:04.715 --> 04:07.015
and we also use the social security

04:07.045 --> 04:08.575
data for the test court data.

04:08.715 --> 04:11.255
We, we observe the test scores

04:11.275 --> 04:14.335
for the individuals in two main subjects in

04:14.335 --> 04:15.655
Spanish and mathematics.

04:16.315 --> 04:19.095
And then for the Mexican school center,

04:19.125 --> 04:20.455
this is particularly relevant

04:20.455 --> 04:23.815
because this is the, the data that we use to construct the

04:24.535 --> 04:25.895
variable of exposure

04:25.955 --> 04:28.695
to English instruction measure in, in hours per week.

04:29.155 --> 04:33.335
And finally, with the social security data,

04:33.395 --> 04:36.455
we observe the universe of formal sector,

04:36.455 --> 04:37.855
workers in, in, in Mexico.

04:38.505 --> 04:42.375
There are like several aspects of this policy intervention,

04:42.465 --> 04:45.655
which makes it challenging to estimate the effect

04:45.655 --> 04:48.535
of exposure to English instruction on labor

04:48.535 --> 04:49.655
market outcomes, right?

04:49.655 --> 04:53.895
So I would like to mention about three, main, issues

04:53.965 --> 04:56.415
that we have with this, in this context.

04:56.555 --> 04:58.495
And the, the issue number one is

04:58.495 --> 05:01.335
that not all the elementary school,

05:01.335 --> 05:03.375
public elementary schools in Mexico decided

05:03.395 --> 05:06.615
to adopt the policy because this was a decision made

05:06.635 --> 05:09.815
by the principal of the each elementary school.

05:10.475 --> 05:13.695
And we have in this context the concern

05:13.695 --> 05:17.175
that maybe the beneficiary schools are systematically

05:17.175 --> 05:19.255
different from the non beneficiary ones

05:19.255 --> 05:23.055
because probably the former are schools with very teachers,

05:23.355 --> 05:27.415
with more information or with more resources, right?

05:27.875 --> 05:32.095
So we try to mitigate this issue by controlling

05:32.115 --> 05:34.815
for school fixed effects in a two-way,

05:34.815 --> 05:36.095
fixed effect specification,

05:36.785 --> 05:40.095
where the other dimension controls for cohort fixed effects.

05:40.675 --> 05:43.535
And the basic intuition is that by using this approach,

05:43.705 --> 05:47.175
we'll be able to compare students within the same school.

05:47.715 --> 05:51.055
But some of them have more exposure to English instruction

05:51.055 --> 05:53.655
because they belong to the younger cohorts,

05:53.705 --> 05:57.605
while others have less exposure or no exposure at all

05:57.605 --> 06:00.685
because these other kids, belong to the older cohort.

06:01.345 --> 06:03.365
So that's like the issue number one.

06:03.365 --> 06:07.725
The issue number two is that, the, this exposure

06:07.785 --> 06:11.005
to English instruction was different across cohorts.

06:11.265 --> 06:12.805
It was also different across schools.

06:13.345 --> 06:16.885
And then under these conditions, it is likely that some

06:16.885 --> 06:20.325
of the weights of, or the two-way fixed effects,

06:20.395 --> 06:24.405
estimator are negative, producing a bias estimate

06:24.425 --> 06:27.925
as suggested by Goman Bacon in his 2021 Journal

06:27.945 --> 06:28.965
of Econometrics paper.

06:29.705 --> 06:33.325
So in the current version of this paper, we try to,

06:33.725 --> 06:34.725
mitigate this issue.

06:35.105 --> 06:38.805
Uh, in, in in particular we follow some of the, recent,

06:39.105 --> 06:43.265
solutions proposed by De Chase Martin

06:43.285 --> 06:47.545
and FO in a very recent paper, also published in the Journal

06:47.545 --> 06:49.745
of Econometrics in, in this year 2023.

06:50.485 --> 06:54.905
Um, but we are also working on additional robust,

06:55.265 --> 06:58.265
estimates as those proposed by Kway and Santana

06:58.285 --> 07:01.385
and Southern Abraham, in both published in,

07:01.385 --> 07:03.305
in the Journal of of Econometrics.

07:03.645 --> 07:04.785
So that's the second issue.

07:04.875 --> 07:08.545
Issue number three is also also relevant for,

07:08.565 --> 07:13.385
for this research because we found with the existing,

07:13.845 --> 07:17.545
model that we are proposing that the exposure

07:17.645 --> 07:20.985
to English instruction reduces the likelihood

07:21.055 --> 07:24.225
that the individuals participate in the formal sector.

07:24.285 --> 07:25.425
And this is quite interesting,

07:25.605 --> 07:28.425
but it's also, troublesome

07:28.425 --> 07:33.385
because instead these individuals are more likely

07:33.445 --> 07:34.625
to be enrolled in school.

07:34.625 --> 07:35.785
And this is natural to think

07:35.785 --> 07:39.625
because the individuals that we study in this paper are,

07:40.705 --> 07:42.905
they are between 16 and 24.

07:43.815 --> 07:48.395
So we try to mitigate this selection into employment by

07:49.005 --> 07:51.995
focus on a low enrollment sample.

07:52.255 --> 07:53.275
And the main results

07:53.275 --> 07:55.915
that we obtain come from this low enrollment sample.

07:56.095 --> 07:57.155
Let me just say

07:57.155 --> 08:00.755
that we are currently working on alternative solutions to,

08:00.775 --> 08:01.915
to this issue.

08:02.335 --> 08:06.035
But, first, result that we obtain

08:06.565 --> 08:10.035
using preliminary solution that, that, I just mentioned,

08:10.375 --> 08:12.515
is that the effect of exposure

08:12.655 --> 08:15.355
to English instruction on wages is basically zero.

08:16.535 --> 08:20.785
Now, that's for the average work that turns out that

08:20.915 --> 08:22.185
among high achieving workers,

08:22.295 --> 08:23.945
this effect is actually positive.

08:24.405 --> 08:27.145
And the natural question is what are the mechanisms

08:27.145 --> 08:29.225
that explain these results, right?

08:29.325 --> 08:31.465
And we explored two main mechanisms.

08:31.525 --> 08:34.345
The first one is effects on cognitive abilities.

08:34.845 --> 08:37.905
The other one is effects on occupational choices.

08:38.685 --> 08:41.065
Now, regarding cognitive abilities, what we,

08:41.445 --> 08:43.505
we provide evidence that exposure

08:43.525 --> 08:46.465
to English instruction does not affect,

08:46.835 --> 08:49.865
other cognitive skills such as language or mathematics.

08:50.365 --> 08:53.025
And this supports the idea that maybe the effect is going

08:53.025 --> 08:55.365
through the acquisition of English skills.

08:55.465 --> 08:59.125
And in the companion paper, I show evidence on this.

08:59.425 --> 09:01.565
Um, but using a different data set,

09:01.925 --> 09:03.125
a different policy change,

09:03.145 --> 09:04.685
but also in the context of Mexico.

09:05.265 --> 09:06.965
And another piece of information

09:07.105 --> 09:09.725
and piece of evidence that supports this idea

09:09.725 --> 09:12.885
of the acquisition of English skills is that for this paper,

09:13.025 --> 09:16.365
we show that the exposure to English instruction

09:16.965 --> 09:20.085
increases the probability that individuals find jobs

09:20.115 --> 09:21.685
that require English skills.

09:22.545 --> 09:26.725
Now, regarding the second mechanism, what we find is

09:26.725 --> 09:28.925
that the individuals who have exposure

09:28.945 --> 09:33.085
to English instruction substitute physically demanding jobs

09:33.425 --> 09:36.885
for other type of industries that require less,

09:37.205 --> 09:40.965
physical, physical activity in the context of

09:42.205 --> 09:43.605
a non-English speaking country.

09:44.885 --> 09:48.485
Exposure to English language instruction at school may lead

09:48.485 --> 09:50.125
to find better jobs,

09:50.745 --> 09:52.925
but also very working conditions

09:53.345 --> 09:56.125
as the individuals acquire English abilities.

09:56.545 --> 09:59.725
And they find jobs in economic industries

09:59.725 --> 10:02.525
that require precisely the same type of skills.

10:03.465 --> 10:07.685
But again, I would like to to say that, this is a,

10:08.005 --> 10:09.285
a work in progress.

10:09.295 --> 10:13.405
We're still trying to improve some of the estimates

10:13.405 --> 10:14.725
that we are obtaining.

10:14.725 --> 10:16.405
There are many challenges that that,

10:16.405 --> 10:19.165
that we are still trying to, to deal with.

10:19.705 --> 10:21.205
And, but,

10:21.305 --> 10:24.445
but this is a very exciting, exciting researcher

10:24.445 --> 10:26.605
to are currently, doing

10:26.785 --> 10:29.565
and I'm glad that I had the opportunity to share it with,

10:29.595 --> 10:30.925
with you in, in this platform.

00:00 - We came up with this research idea after observing two interesting facts. So the first fact is that among non-English speaking countries is there is this common belief that the individuals who speak English are more likely to find very big jobs,

00:20 - or that to find very labor opportunities, right? Actually, we have this saying in Latin America that English opens doors. And the, I will say for the second fact, in the last 25 years, there have been efforts among developing countries to introduce English instruction as a subject in their public education systems. And they do this under the assumptions, a very similar assumption that the individuals who have exposure to English instruction will eventually

00:53 - acquire English abilities. And also when they incorporate to a labor market, they will find, very, paid jobs and very labor opportunities. The thing is that we don't have council evidence on this claim. So with this paper, we, we try to provide some of the first evidence from a developing non-English speaking country, which is Mexico. There is actually a large literature starting the returns

01:20 - to English language skills, but in the context of English speaking countries, right? And the classical references are she weak and Miller in for their contributions in the 1990s and after that, Blakely and Chin in 2004, with one of the first council, evidence on the effects of English skills on earnings, but among immigrants in the us. So this is important because most of the existing literature focus on this context

01:50 - of immigrants in English speaking countries. While there is just a few number of papers in the context of non-English speaking countries, and, even among these other papers, most of them focus in former British colonies. for example, we have the case of Asam Chin and Prakash, eh, they offer some evidence in the context of India and also Ericsson for the context of South Africa. But again, the thing is that these two economies are quite different from most of the developing work

02:21 - because of their status of former British colons. There are two important aspects that I would like to mention regarding the, the, the policy background and also the institutional, background in Mexico. The first one is that in, in Mexico, K to 12 schooling is divided in three levels, right? Elementary school, covering from first grade to sixth grade. Then middle school that goes from seventh to ninth, and finally a high school that goes from 10th to

02:58 - 12th grade. the second important aspect is that in, in Mexico, the enrollment rates in elementary school are high, at about 95%. Then for it drops to middle school to 85%. Then, it drops even more for high school and, and college going to 70 and then 30%. Now, this is relevant because in this paper, the policy that we, that we studied was implemented at the elementary school

03:33 - there, where the enrollment rates are particularly high. In this research, we construct a novel database connecting the universe of elementary school students in Mexico to the labor market outcomes more than 10 years after these kids have exposure to English instruction in school school, and we construct this dataset using three sources of information. The first one is test court data,

04:02 - then the Mexican school census, and we also use the social security data for the test court data. We, we observe the test scores for the individuals in two main subjects in Spanish and mathematics. And then for the Mexican school center, this is particularly relevant because this is the, the data that we use to construct the variable of exposure

04:25 - to English instruction measure in, in hours per week. And finally, with the social security data, we observe the universe of formal sector, workers in, in, in Mexico. There are like several aspects of this policy intervention, which makes it challenging to estimate the effect of exposure to English instruction on labor market outcomes, right? So I would like to mention about three, main, issues that we have with this, in this context.

04:56 - And the, the issue number one is that not all the elementary school, public elementary schools in Mexico decided to adopt the policy because this was a decision made by the principal of the each elementary school. And we have in this context the concern that maybe the beneficiary schools are systematically different from the non beneficiary ones because probably the former are schools with very teachers, with more information or with more resources, right?

05:27 - So we try to mitigate this issue by controlling for school fixed effects in a two-way, fixed effect specification, where the other dimension controls for cohort fixed effects. And the basic intuition is that by using this approach, we'll be able to compare students within the same school. But some of them have more exposure to English instruction because they belong to the younger cohorts, while others have less exposure or no exposure at all because these other kids, belong to the older cohort.

06:01 - So that's like the issue number one. The issue number two is that, the, this exposure to English instruction was different across cohorts. It was also different across schools. And then under these conditions, it is likely that some of the weights of, or the two-way fixed effects, estimator are negative, producing a bias estimate as suggested by Goman Bacon in his 2021 Journal of Econometrics paper. So in the current version of this paper, we try to,

06:33 - mitigate this issue. in, in in particular we follow some of the, recent, solutions proposed by De Chase Martin and FO in a very recent paper, also published in the Journal of Econometrics in, in this year 2023. but we are also working on additional robust, estimates as those proposed by Kway and Santana and Southern Abraham, in both published in, in the Journal of of Econometrics. So that's the second issue.

07:04 - Issue number three is also also relevant for, for this research because we found with the existing, model that we are proposing that the exposure to English instruction reduces the likelihood that the individuals participate in the formal sector. And this is quite interesting, but it's also, troublesome because instead these individuals are more likely to be enrolled in school. And this is natural to think

07:35 - because the individuals that we study in this paper are, they are between 16 and 24. So we try to mitigate this selection into employment by focus on a low enrollment sample. And the main results that we obtain come from this low enrollment sample. Let me just say that we are currently working on alternative solutions to, to this issue. But, first, result that we obtain

08:06 - using preliminary solution that, that, I just mentioned, is that the effect of exposure to English instruction on wages is basically zero. Now, that's for the average work that turns out that among high achieving workers, this effect is actually positive. And the natural question is what are the mechanisms that explain these results, right? And we explored two main mechanisms. The first one is effects on cognitive abilities.

08:34 - The other one is effects on occupational choices. Now, regarding cognitive abilities, what we, we provide evidence that exposure to English instruction does not affect, other cognitive skills such as language or mathematics. And this supports the idea that maybe the effect is going through the acquisition of English skills. And in the companion paper, I show evidence on this. but using a different data set, a different policy change,

09:03 - but also in the context of Mexico. And another piece of information and piece of evidence that supports this idea of the acquisition of English skills is that for this paper, we show that the exposure to English instruction increases the probability that individuals find jobs that require English skills. Now, regarding the second mechanism, what we find is that the individuals who have exposure to English instruction substitute physically demanding jobs

09:33 - for other type of industries that require less, physical, physical activity in the context of a non-English speaking country. Exposure to English language instruction at school may lead to find better jobs, but also very working conditions as the individuals acquire English abilities. And they find jobs in economic industries that require precisely the same type of skills. But again, I would like to to say that, this is a,

10:08 - a work in progress. We're still trying to improve some of the estimates that we are obtaining. There are many challenges that that, that we are still trying to, to deal with. And, but, but this is a very exciting, exciting researcher to are currently, doing and I'm glad that I had the opportunity to share it with, with you in, in this platform.

Faculti AI can make mistakes