The Determinants and the
Selection of Mexico-US Migrants
Daniel Chiquiar
Banco de México
May 30th, 2011
6th IZA/World Bank Conference
1
Index
1. Introduction
2. Self-selection of Mexico-US migrants
3. Trend in self-selection patterns
4. Possible hypothesis
5. Concluding remarks
2
Introduction
 The question of who migrates from Mexico to the US has been the
subject of several recent research papers.
 There has been mixed evidence on the self-selection of Mexican
immigrants in the US.
 The topic is relevant given that effective public policies related to
migration between Mexico and the US depend on the
characteristics of migrants.
• In particular, the skill level of Mexican migrants to the United States is a
relevant issue for labor markets and immigration policy.
Introduction
3
Borjas’ Self Selection Model
 Constant migration costs.
Wage
(net of migration costs)
Mexico
USA
Migrants
s*
Introduction
Schooling
 In poor countries (high returns
to skill, high wage dispersion),
the low-skilled have the
strongest incentive to migrate:
negative selection.
 Under the assumption that
wages are higher in the US than
in Mexico at low education
levels, and the opposite at high
education levels, we should
expect negative selection of
migrants.
4
Heterogeneity in migration costs
Wage
(net of migration costs)
Mexico
USA
Migrants
sL
Introduction
sU
 If
migration
costs
are
decreasing in schooling, we
could observe intermediate
selection.
 Migration costs preclude those
with low schooling from
migrating, and high returns to
schooling in Mexico dissuade
those with high schooling from
migrating.
Schooling
5
Index
1. Introduction
2. Self-selection of Mexico-US migrants
3. Trend in self-selection patterns
4. Possible hypothesis
5. Concluding remarks
6
Self-selection of Mexico-US migrants
Selectivity Pattern of Mexican Migrants to the US
Main findings
Selection on Education
Selection on Wages
Data
Chiquiar and Hanson (2005)
intermediate
intermediate
US and Mexican
Censuses
(1990 and 2000)
Kaestner and Malamud (2010)
intermediate
intermediate
MxFLS1/
Orrenius and Zavodny (2005)
intermediate
MMP2/
McKenzie and Rapoport (2010)
positive in regions with low
migration experience
negative in regions with high
migration experience
ENADID3/
Fernández-Huertas Moraga (2011)
negative
Ibarraran and Lubotsky (2007)
negative
Ambrosini and Peri (2011)
negative
1/ Mexican
Family Life Survey
2/ Mexican
Self-selection of Mexico-US migrants
Migration Project
3/ Encuesta
Nacional de la Dinámica Demográfica
negative
ENET
4/
Mexican Census
(2000)
negative
4/ Encuesta
MxFLS1/
Nacional de Empleo Trimestral
7
Chiquiar & Hanson (2005)
 Main finding is not consistent with a simple Borjas model:
intermediate selection of male immigrants in the US.
positive selection of female immigrants in the US.
 Mexican immigrants are more educated than non-migrants.
 Migrants would fall in the middle and upper portions of Mexico’s
wage distribution.
• Estimation of counterfactual wage distribution in Mexico.
 Data used: 1990 and 2000 Mexican and U.S. population censuses.
• Main concerns:
 Undercount of illegal immigrants.
 May overstate education due to additional schooling in the U.S.
 Findings in terms of observable characteristics only.
Self-selection of Mexico-US migrants
8
Kaestner & Malamud (2010)
 Find intermediate selection of male Mexican migrants with respect
to a skill index based on observed characteristics.
 Little evidence of selection on unobservables.
 No selection once proxy variables for costs of migration are taken
into account.
 Data used: Mexican Family Life Survey (MxFLS).
•
•
•
•
•
Panel data 2002 & 2005
Can observe earnings (and other characteristics) prior to migration.
Can identify migrants who moved their entire household.
Possible Ashenfelter dip.
Small sample size.
 Evidence that skills and costs of migrating are highly negatively
correlated.
Self-selection of Mexico-US migrants
9
Orrenius & Zavodny (2005)
 Find intermediate selection of illegal immigrants from Mexico with
respect to the education distribution.
• Stricter border enforcement is associated with a more positively selected
immigrant stream (higher average education).
• Higher U.S. wages and improved conditions in Mexico have the opposite
effect (lower average education of undocumented immigrants).
 Data used: Mexican Migration Project (MMP), 1982-1983, 1987-1997.
• Data concerns:
 Only migrants who have either returned to Mexico or who have at least one
household member remaining in Mexico.
 Recall bias (data collected retrospectively).
 Limited number of communities.
Self-selection of Mexico-US migrants
10
McKenzie & Rapoport (2010)
 Main Result: self-selection depends on migration history of the
community.
• Positive or neutral selection in communities with weak migration networks.
• Negative selection in communities with stronger networks.
 Data used: Encuesta Nacional de la Dinámica Demográfica (ENADID), 1997.
• Data concerns:
 Only migrants who have either returned to Mexico or who have at least one
household member remaining in Mexico.
 Hard to control for unobservable characteristics.
Self-selection of Mexico-US migrants
11
Fernández-Huertas Moraga (2011)
 Finds negative selection on wages and intermediate to negative
selection on education.
• For women, migrants are more educated than non-migrants.
• Selection is positive in rural Mexico.
 Data used: Encuesta Nacional de Empleo Trimestral (ENET), 2002-2004.
• Panel data.
 Earnings (and other characteristics) prior to migration.
 Possible Ashenfelter dip.
• Only migrants who have either returned to Mexico or who have at least one
household member remaining in Mexico.
• Low rates of labor market participation and lack of information on wages
due to the inclusion of younger men (aged 16 to 20).
Self-selection of Mexico-US migrants
12
Ibarraran & Lubotsky (2007)
 Find negative selection in terms of education.
• They estimate the education of Mexican migrants based on information
given by household members who remained in Mexico.
• Then conclude that migrants tend to be less educated than non-migrants.
 Data used: Mexican Census 2000.
• Had to make several assumptions to predict education attainment of
migrants.
Self-selection of Mexico-US migrants
13
Ambrosini & Peri (2011)
 Use the Mexican Family Life Survey (MxFLS), 2002-2005.
 Evidence of negative selection.
• On average, workers who later migrated earned 23% less than workers who
did not migrate.
• Note that this is a point estimate, not an analysis of the wage distribution.
Difference in (log) wages between
migrants and non-migrants before migration
Selection
Total Difference
Explained by Observable
Due to Unobservable
Total
Male
Female
-23%
-40%
10%
(3.9%)
(5.0%)
(4.0%)
-5%
-12%
-5%
(4.0%)
(5.2%)
(4.3%)
-18%
-27%
15%
(3.8%)
(4.7%)
(3.7%)
Source: Ambrosini & Peri (2011), Table 2.
Self-selection of Mexico-US migrants
14
Ambrosini & Peri (2011)
 Evidence on costs of migration, once controlling for the
migration earnings premium:
• Higher costs of migrating for highly educated.
• Living in a state close to the US border or having some assets
and lower debt make migration more likely.
• Having a relative in the US also makes migration more likely.
• Negative shocks to the household increase the probability of
migrating.
Self-selection of Mexico-US migrants
15
Ambrosini & Peri (2011)
Determinants of Migration Probabilities
Dependent Variable: percentage of migrants in the skill cell
Specifications
Migration Premium
(1)
(2)
(3)
(4)
(5)
0.018***
0.014**
0.014***
0.006**
0.009**
(0.001)
(0.001)
(0.001)
(0.002)
(0.002)
-0.006
-0.006
(0.003)
(0.004)
Living in Border States
0.07**
Living in Mexico City
-0.06**
Having a Relative in US
0.05**
(0.01)
(0.02)
(0.007)
Pre-migration income
0.002**
Pre-migration log of assets
(0.0001)
-0.002*
Pre-migration log of debt
(0.0001)
-0.02**
Pre-migration log of house value
(0.001)
0.038*
Share of individuals who experienced a family shock in the last year
(0.008)
-0.0006
Share of entrepreneurs
Age, education and family Status Dummies
Border and family network controls
Initial wealth controls
Number of cells
Observations
(0.0006)
No
No
No
Yes
No
No
Yes
Yes
No
Yes
Yes
Yes
Yes
Yes
Yes
72
4,246
72
4,246
72
4,246
72
4,246
72
4,246
***, **, * denote significance at the 1%, 5% and 10% level respectively.
Source: Ambrosini, JW. and Peri, G. (2011). The Determinants and the Selection of Mexico-US Migrants. Tables 6, 7, 8, and 9.
Self-selection of Mexico-US migrants
16
To keep in mind
 When comparing results across different studies some points
should be kept in mind:
• Flows vs stocks: some papers study selection of migration flows while
others focus on stocks of migrants.
• Recent vs not so recent migrants: if self-selection changes over time, results
may vary depending on when Mexican migrants arrived to the US.
 Costs of migration are still hard to identify with precision.
 Self-selection in return migration needs more attention.
Self-selection of Mexico-US migrants
17
Index
1. Introduction
2. Self-selection of Mexico-US migrants
3. Trend in self-selection patterns
4. Possible hypothesis
5. Concluding remarks
18
Trend in self-selection patterns
 The research previously described looks at self-selection at one
point in time.
 However, given possible changes of the self-selection patterns
over time, it is important to look at long term trends.
 Chiquiar and Hanson (2005) had already identified a shift towards
less positive selection between 1990 and 2000.
 Very recent data from the 2010 Mexican Census data allows us to
confirm this trend.
Trend in self-selection patterns
19
Trend in self-selection patterns
 We find that education levels have increased for both Mexican
residents and Mexican immigrants in the US, from 1990 to 2000,
and to 2010.
 However, Mexican immigrants have become less educated relative
to Mexican residents.
 This pattern is confirmed when we compute counterfactual wage
distributions.
Trend in self-selection patterns
20
Cumulative distribution of highest grade of schooling completed
Male
Female
1.0
1.0
Mexican Residents
0.8
Cumulative distribution
Cumulative distribution
Mexican Residents
0.6
0.4
0.2
1990
2000
0-4
0-8
0-9
0-11
0-12
0-15
Highest grade of schooling completed
0.2
0-16+
1990
0
1.0
Recent Mexican
Immigrants
Cumulative distribution
Cumulative distribution
0.4
2000
2010
0.0
0
0.8
0.6
2010
0.0
1.0
0.8
0.6
0.4
0.2
1990
2000
0.8
0-4
0-8
0-9
0-11
0-12
0-15
Highest grade of schooling completed
0-16+
Recent Mexican
Immigrants
0.6
0.4
0.2
2010
1990
2000
2010
0.0
0.0
0
0-4
0-8
0-9
0-11
0-12
0-15
Highest grade of schooling completed
0-16+
0
0-4
0-8
0-9
0-11
0-12
0-15
Highest grade of schooling completed
0-16+
Source: Mexican Census (1990, 2000, and 2010), US Census (1990 and 2000) and ACS (2009).
Trend in self-selection patterns
21
Self-selection on Education
Cumulative distribution of highest grade of schooling completed
Men
1.0
1990
1.0
2000
0.9
0.9
0.8
0.8
0.8
0.7
0.7
0.7
0.6
0.5
0.4
0.3
Mexican Residents
0.2
Cumulative distribution
0.9
Cumulative distribution
Cumulative distribution
1.0
0.6
0.5
0.4
0.3
Mexican Residents
0.6
0.5
0.4
0.3
Recent Mexican Immigrants
Recent Mexican Immigrants
0.1
0.1
0.1
0.0
0.0
0.0
0 0-4 0-8 0-9 0-11 0-12 0-15 0-16+
Highest grade of schooling completed
Mexican Residents
0.2
0.2
Recent Mexican Immigrants
2010
0 0-4 0-8 0-9 0-11 0-12 0-15 0-16+
Highest grade of schooling completed
0 0-4 0-8 0-9 0-11 0-12 0-15 0-16+
Highest grade of schooling completed
Source: Mexican Census (1990, 2000, and 2010), US Census (1990 and 2000) and ACS (2009).
Trend in self-selection patterns
22
Self-selection on Education
Cumulative distribution of highest grade of schooling completed
Women
1.0
1990
1.0
2000
0.9
0.9
0.8
0.8
0.8
0.7
0.7
0.7
0.6
0.5
0.4
0.3
Mexican Residents
Cumulative distribution
0.9
Cumulative distribution
Cumulative distribution
1.0
0.6
0.5
0.4
0.3
0.5
0.4
0.3
Recent Mexican Immigrants
0.1
0.1
0.1
0.0
0.0
0.0
0 0-4 0-8 0-9 0-11 0-12 0-15 0-16+
Highest grade of schooling completed
Mexican Residents
0.2
Recent Mexican Immigrants
Recent Mexican Immigrants
0.6
Mexican Residents
0.2
0.2
2010
0 0-4 0-8 0-9 0-11 0-12 0-15 0-16+
Highest grade of schooling completed
0 0-4 0-8 0-9 0-11 0-12 0-15 0-16+
Highest grade of schooling completed
Source: Mexican Census (1990, 2000, and 2010), US Census (1990 and 2000) and ACS (2009).
Trend in self-selection patterns
23
Self-selection on wages
Actual and Counterfactual Wage
Densities, Men 1990
Counterfactual: Mex Imm in US
Actual: Mex Residents
Counterfactual: Mex Imm in US
0
0
.1
.2
.2
.3
.4
.4
.5
.6
Actual: Mex Residents
Actual and Counterfactual Wage
Densities, Men 2000
-3
-2
-1
0
lnw
1
2
3
-3
-2
Differences between Actual and
Counterfactual Wage Densities, Men
1
2
3
2
3
Double Differences
Men 2000-1990
.015
1990
0
lnw
-.04
-.01
-.02
-.005
0
0
.02
.005
.04
.01
2000
-1
-3
-2
1990 Skill prices
Trend in self-selection patterns
-1
0
lnw
1
2
3
-3
-2
1990 Skill prices
-1
0
lnw
1
24
Self-selection on wages
Male wage densities
1990 skill prices
Difference between Actual and Counterfactual
.04
1990
-.04
-.085
-.05
-.02
0
1990-2000
0
.05
.02
.085
2000
Double Differences
-2
-1
0
lnw
2
3
2
3
1990
-3
-2
-1
0
lnw
1
2
3
-3
-2
-1
0
lnw
1
2
3
.05
.02
.085
2010
1
.04
-3
-.04
-.085
-.05
-.02
0
0
1990-2010
-3
Trend in self-selection patterns
-2
-1
0
lnw
1
25
Self-selection on wages
Female wage densities
1990 skill prices
Difference between Actual and Counterfactual
.04
1990
.05
.02
.085
2000
Double Differences
-.04
-.085
-.05
-.02
0
0
1990-2000
-2
-1
0
lnw
2
3
2
3
1990
-3
-2
-1
0
lnw
1
2
3
.05
.02
.085
2010
1
.04
-3
-.04
-.085
-.05
-.02
0
0
1990-2010
-3
Trend in self-selection patterns
-2
-1
0
lnw
1
-3
-2
-1
0
lnw
1
2
3
26
Index
1. Introduction
2. Self-selection of Mexico-US migrants
3. Trend in self-selection patterns
4. Possible hypothesis
5. Concluding remarks
27
Migration costs
 What could be driving the shift towards less positive selection?
 Some of the obvious migration costs do not seem to be the
reason, since they are likely to be increasing over time.
 Tighter immigration policy from the US
 Higher cost of illegally crossing the border
Border Patrol Officers
Cost of Coyote
Thousands
Constant US Dollars, 2009
25
Source: Mexican Migration Project (MMP128) based on US Immigration
and Naturalization Service until 2005. From 2006, Department of
Homeland Security, CBP Fiscal Year Review Factsheets, 2006-2009.
Possible hypothesis
2008
2004
2000
1996
1992
500
1988
0
1984
1,000
2008
5
2005
1,500
2002
10
1999
2,000
1996
15
1993
2,500
1990
20
1980
3,000
Source: Mexican Migration Project (MMP128), Microdata.
28
Networks and probability of migration
 This implies that other important migration costs that affect
disproportionately the lower end of the skills distribution could be
decreasing over time.
• Networks are an important candidate.
Probability of migration
Logit estimation
Dep. var.: dummy variable for Mexican immigrant in the U.S.
Network
Men
Men
7.52***
4.33***
5.91***
(0.355)
(0.317)




















No
440,882
Yes
440,882
No
457,717
Yes
457,717
(0.300)
Controls
Education
Age
Married
Children
State fixed effects
Controls flexible by year
Observations
Women Women
3.20***
(0.370)
Data sources for Mexico:1990 and 2000 population censuses.
Data sources for the U.S.: 1990 and 2000 population censuses.
Network for 1990: % of households with at least one member with migration experience to the US within state of origin (source ENADID92).
Network for 2000: % of households with at least one member with migration experience to the US within state of origin (source ENADID97).
Possible hypothesis
29
Networks and probability of migration
Probability of migration
Logit estimation
Dep. var.: dummy variable for Mexican immigrant in the U.S.
Men
Men
Women
Women
0
7.63***
4.26***
7.69***
4.88***
(0.331)
(0.388)
(0.346)
(0.404)
1-4
6.73***
3.68***
4.87***
2.37***
(0.337)
(0.389)
(0.354)
(0.403)
5-8
7.16***
3.85***
4.89***
2.22***
(0.311)
(0.369)
(0.328)
(0.384)
9
7.96***
4.84***
5.48***
2.93***
(0.345)
(0.393)
(0.366)
(0.414)
10-11
7.90***
4.68***
6.36***
3.65***
(0.375)
(0.422)
(0.411)
(0.459)
12
8.19***
4.99***
6.16***
3.58***
(0.331)
(0.379)
(0.346)
(0.396)
13-15
7.77***
4.56***
6.28***
3.69***
(0.364)
(0.412)
(0.387)
(0.435)
16+
7.57***
4.41***
5.82***
3.33***
(0.378)
(0.425)
(0.409)
(0.451)
Controls
Education
Age
Married
Children
State fixed effects




















Controls flexible by year
No
Yes
No
Yes
440,882
440,882
457,717
457,717
Network interacted with
highest grade of schooling completed
Observations
Data sources for Mexico:1990 and 2000 population censuses.
Data sources for the U.S.: 1990 and 2000 population censuses.
Network for 1990: % of households with at least one member with migration experience to the US within state of origin (source ENADID92).
Network for 2000: % of households with at least one member with migration experience to the US within state of origin (source ENADID97).
Possible hypothesis
30
Networks and probability of migration
Percentage of households with at least
one member that migrated to the US
to work or search for a job in any year
14
Percentage of households that had
at least one migrant to the US
during the last 5 years
6
13.00 %
13
5
4.93 %
4
% 12
2.83 %
% 3
11.56 %
2
11
1
0
10
1992
1997
Source: Demographic Dynamics National Survey
Mexico (ENADID) 1992 and 1997
Possible hypothesis
2000
2010
Source: Mexican Census 2000 and 2010
31
Networks and self-selection on wages
Male, 1990-2000
Original estimation
Actual: Mex Residents
Counterfactual: Mex Imm in US
0.6
Actual: Mex Residents
Counterfactual: Mex Imm in US
0.6
1990
0.5
0.5
0.4
0.4
0.3
0.3
0.2
0.2
0.1
0.1
Self-selection 1990
2000
Self-selection 2000
0.07
0.03
2000
1990
-0.01
-0.05
0
0
-3
-2
-1
0
lnw
1
2
3
-0.09
-3
-2
-1
0
lnw
1
2
3
-3
-2
-1
0
lnw
1
2
3
Estimation including networks
0.6
Self-selection 1990
Actual: Mex Residents
Counterfactual: Mex Imm in US
Actual: Mex Residents
Counterfactual: Mex Imm in US
0.6
1990
0.5
0.07
2000
0.5
0.4
1990
0.03
0.4
0.3
0.3
0.2
0.2
0.1
0.1
0
0
-3
-2
-1
Possible hypothesis
0
lnw
1
2
3
Self-selection 2000
2000
-0.01
-0.05
-0.09
-3
-2
-1
0
lnw
1
2
3
-3
-2
-1
0
lnw
1
2
3
32
Networks and self-selection on wages
Male, 1990-2010
Original estimation
Actual: Mex Residents
Counterfactual: Mex Imm in US
0.6
Actual: Mex Residents
Counterfactual: Mex Imm in US
0.7
1990
0.5
0.6
0.4
0.5
Self-selection 1990
2010
0.07
0.03
1990
2010
0.4
0.3
Self-selection 2010
-0.01
0.3
0.2
0.2
0.1
-0.05
0.1
0
-0.09
0
-3
-2
-1
0
lnw
1
2
3
-3
-2
-1
0
lnw
1
2
3
-3
-2
-1
0
lnw
1
2
3
Estimation including networks
Actual: Mex Residents
Counterfactual: Mex Imm in US
0.6
Actual: Mex Residents
Counterfactual: Mex Imm in US
0.8
1990
2010
0.7
0.5
Self-selection 1990
0.07
0.6
0.4
Self-selection 2010
0.03
0.5
0.4
0.3
1990
2010
-0.01
0.3
0.2
0.2
0.1
-0.05
0.1
0
0
-3
-2
-1
Possible hypothesis
0
lnw
1
2
3
-0.09
-3
-2
-1
0
lnw
1
2
3
-3
-2
-1
0
lnw
1
2
3
33
Networks and self-selection on wages
Differences in original
estimation
Self-selection 1990
Male
Double differences
Differences including
networks
Self-selection 1990
Self-selection 2000
Self-selection 2000
Original
Including Networks
0.02
0.07
0.07
1990 vs 2000
0.015
2000
0.01
1990
0.03
0.03
1990
0.005
2000
-0.01
-0.01
-0.05
-0.05
0
-0.005
-0.01
-0.015
-0.09
-0.09
-3
-2
-1
0
lnw
1
2
-3
3
Differences in original
estimation
Self-selection 1990
-2
-1
0
lnw
1
2
-3
3
Differences including
networks
Self-selection 1990
Self-selection 2010
-2
-1
0
lnw
1
2
3
Double differences
Self-selection 2010
Original
Including Networks
0.04
0.07
0.07
0.03
0.03
1990 vs 2010
0.03
0.02
1990
2010
-0.01
-0.01
-0.05
-0.05
0.01
1990
2010
0
-0.01
-0.02
-0.03
-0.04
-0.09
-0.09
-3
-2
Possible hypothesis
-1
0
lnw
1
2
3
-3
-2
-1
0
lnw
1
2
3
-3
-2
-1
0
lnw
1
2
3
34
Networks and self-selection on wages
Differences in original
estimation
Self-selection 1990
Differences including
networks
Self-selection 1990
Self-selection 2000
0.12
0.12
0.07
0.07
0.02
0.02
1990
2000
-0.03
-0.03
-0.08
-0.08
-0.13
-0.13
-3
-2
-1
0
lnw
1
2
Self-selection 2000
Original
0
1990
2000
-0.02
-0.04
-2
-1
0
lnw
1
2
3
-3
Differences including
networks
Self-selection 2010
Self-selection 1990
0.08
0.08
0.03
0.03
Including Networks
1990 vs 2000
-3
3
Double differences
0.02
Differences in original
estimation
Self-selection 1990
Female
-2
-1
0
lnw
1
2
3
Double differences
Self-selection 2010
Original
Including Networks
1990 vs 2010
0.04
0.02
0
2010
-0.02
2010
-0.02
-0.02
-0.07
-0.07
1990
-0.04
1990
-0.12
-0.12
-3
-2
-1
Possible hypothesis
0
lnw
1
2
3
-0.06
-3
-2
-1
0
lnw
1
2
3
-3
-2
-1
0
lnw
1
2
3
35
Possible demand effect
 Additionally, a demand effect could be playing a role.
• If the US demands unskilled Mexican workers, and Mexicans have become
more educated over time, the flow of unskilled Mexicans that previously
looked like positive selection is now starting to look like negative selection.
schooling
Demand for immigrants
Possible hypothesis
36
Possible demand effect
 The proportion of Mexican residents with 5-8 and 9 years of schooling has
decreased. Meanwhile, the percentage of recent Mexican immigrants with
such schooling level has not changed significantly.
Highest grade of schooling completed by males (%)
Mexican Residents
30%
1990
2000
Recent Mexican Immigrants
35%
2010
2000
1990
U.S. Natives
45%
2010
2010
40%
30%
25%
2000
1990
35%
25%
15%
Percentage
Percentage
Percentage
20%
20%
30%
25%
20%
15%
10%
15%
10%
10%
5%
5%
5%
0%
0%
0
1--4
5--8
9
10--11 12 13--15 16+
Highest grade of schooling completed
Possible hypothesis
0%
0
1--4
5--8
9
10--11 12 13--15 16+
Highest grade of schooling completed
0
1--4
5--8
9
10--11 12 13--15 16+
Highest grade of schooling completed
37
Possible demand effect
Highest grade of schooling completed by males (%), 1990 and 2010
35%
30%
Mexican Residents 1990
Mexican Residents 2010
Mexican immigrants 1990
Mexican Immigrants 2010
Percentage
25%
20%
15%
10%
5%
0%
0
1-4
5-8
9
10-11
12
Highest grade of schooling completed
13-15
16+
38
Index
1. Introduction
2. Self-selection of Mexico-US migrants
3. Trend in self-selection patterns
4. Possible hypothesis
5. Concluding remarks
39
Concluding remarks
 Contrary to a simple Borjas model, there is evidence of
intermediate selection of Mexican immigrants to the US. However,
some studies have also found that migrants may be negatively
selected.
 New data on the 2010 Mexican census has shown that there may
be a shift towards less positive selection over time.
 It may be worth studying the reasons behind this shift, especially
whether networks are playing a role, or whether there is a
demand effect.
Concluding remarks
40