Accounting for Spatial Heterogeneity in Educational Outcomes and International Migration in Mexico

15
Accounting for Spatial Heterogeneity in Educational Outcomes and International Migration in Mexico Edith Gutierrez Landy Sanchez Silvia Giorguli El Colegio de Mexico

description

Accounting for Spatial Heterogeneity in Educational Outcomes and International Migration in MexicoEdith Yolanda Gutierrez-Vazquez,Landy Lizbeth Sanchez-Peña, Silvia Elena Giorguli-Saucedo - El Colegio de Mexico

Transcript of Accounting for Spatial Heterogeneity in Educational Outcomes and International Migration in Mexico

Page 1: Accounting for Spatial Heterogeneity in Educational Outcomes and International Migration in Mexico

Accounting for Spatial Heterogeneity in Educational Outcomes and International Migration in Mexico

Edith Gutierrez Landy Sanchez Silvia Giorguli

El Colegio de Mexico

Page 2: Accounting for Spatial Heterogeneity in Educational Outcomes and International Migration in Mexico

EDUCATIONAL OUTCOMES AND INTERNATIONAL MIGRATION IN MEXICO: A BRIEF REVIEW

To have in mind:• Educational attainment and international migration

• Positive effects: remittances improve chances of school to work transition

• Negative effects: the “culture of migration”: migration as a better social mobility mechanism than education

Mexico-US Migration is characterized by:• Strong regional component due to historical trajectories• since the 90’s the stream spread diversified across the

countryEducational achievement is also strongly diverse across México

• How these two spatial patterns relates? • How can be captured regional differences?

Page 3: Accounting for Spatial Heterogeneity in Educational Outcomes and International Migration in Mexico

TESTING SPATIAL HETEROGENEITY HYPOTHESIS: CLASSICAL VERSUS SPATIAL APPROACHES

• Classical definition for migratory regions (Durand &

Massey, 2003):• Based on the historical intensity of Mexico-US flows and on

migration prevalence ratio Hypothesis:

International migration disincentives educational achievement, regional variations depending on historic experience: stronger effects in traditional and border regions and smaller in regions where outflow started recentlyRegions defined by migration prevalence at a

given point in time

Page 4: Accounting for Spatial Heterogeneity in Educational Outcomes and International Migration in Mexico

TESTING SPATIAL HETEROGENEITY HYPOTHESIS: CLASSICAL VERSUS SPATIAL APPROACHES

• Spatial approach• Based on spatial heterogeneity in the relationship between education,

international migration and labor market.

Hypothesis:International Migration will have negative effect on educational

outcomes but the variations will be due to historical migratory trajectories and to employment and educational infrastructure: strong effects of historical migration experience regions will decrease in regions with a good labor market performance and vice versa

Regions are defined based on Geographically Weighted Regression results, not solely by migration prevalence

Page 5: Accounting for Spatial Heterogeneity in Educational Outcomes and International Migration in Mexico

TESTING SPATIAL HETEROGENEITY HYPOTHESIS: METHODOLOGICAL ISSUES

Both Classical and Spatial Hypothesis imply spatial processes of:

• Dependence:• Autocorrelation within regions between local educational

outcomes and education, migration and employment trade offs

• Heterogeneity:• Significant differences in the effects of migration or labor on

educational attainment across regions

Need a dependence and structural heterogeneity spatial model to decide which is the best approach to define regions

Page 6: Accounting for Spatial Heterogeneity in Educational Outcomes and International Migration in Mexico

GENERAL METHODOLOGICAL STRATEGY

• Both hypothesis require spatial analysis techniques and suggest a spatial dependence process:1. Corroborating spatial effects:

1. OLS regression 2. Moran’s I3. Local Indicators of Spatial Association

2. Proving differences across regions:1. Spatial Regimes model with a spatial dependence term and

a heteroskedasticity correction2. Migration parameter significance 3. Chow-Wald Test4. Coefficients Stability Test

Page 7: Accounting for Spatial Heterogeneity in Educational Outcomes and International Migration in Mexico

VARIABLES

Dimension Measure Educational outcomes

Expected years of schooling after elementary education

International Migration Intensity migration index

Labor Market Female participation rateProportion of employed population with income belor two minimum wagesProportion of employed population in the manufacturing sector

Cotrolling VariablesQuality and Availability of Educational services

Internal Migration

Urbanization level

Page 8: Accounting for Spatial Heterogeneity in Educational Outcomes and International Migration in Mexico

Educational Outcomes

International Migration

Moran’s I 0.6239

Moran’s I 0.3713

Results: Significant spatial dependence, heteroskedasticity issues and significant clusters across the country

Page 9: Accounting for Spatial Heterogeneity in Educational Outcomes and International Migration in Mexico

International Migration

Female Labor

Low-income Workers

Industrialization level

Page 10: Accounting for Spatial Heterogeneity in Educational Outcomes and International Migration in Mexico

Traditional Regions

Spatially Defined Regions

Page 11: Accounting for Spatial Heterogeneity in Educational Outcomes and International Migration in Mexico

OLS RESULTS: DEPENDENCE AND HETEROSKEDASTICITY TESTS

Traditional Regions

Spatially Defined Regions

Heteroskedasticity test, both regional definitions:

• The Koenker-Bassett has a 1% significance level

Test SignificanceRobust LM Error 0.001Robust LM Lag 0.010

Test SignificanceRobust LM Error 0.100Robust LM Lag 0.001

Page 12: Accounting for Spatial Heterogeneity in Educational Outcomes and International Migration in Mexico

RESULTS SPATIAL MODEL WITH STRUCTURAL CHANGE AND GROUPWISE HETEROSKEDASTICITY

Significance Level of Individual Coefficients Stability Test for the Traditional and the Spatially defined Regions

VariableTradition

alSpatially defined

CONSTANT 0.00 0.00International migration 0.27 0.05Female participation rate 0.00 0.00Low-income workers 0.00 0.00Industrialization level 0.00 0.00

Educational profile of teachers 0.34 0.82

Without high schools vs technical schools 0.18 0.16Without high schools vs technical-general track schools 0.26 0.11

Inter-municipal migration 0.94 0.37

Rural vs Rural-Urban 0.23 0.54

Rural vs Urban 0.83 0.54

Rural vs Metropolitan area 0.21 0.08

Page 13: Accounting for Spatial Heterogeneity in Educational Outcomes and International Migration in Mexico

SPATIALLY DEFINED REGIONS RESULTS SPATIAL ERROR MODEL WITH STRUCTURAL CHANGE AND GROUPWISE HETEROSKEDASTICITY

Parameters and Significance

Variable Border Traditional Central Southeast

International migration-3.39*** -2.44*** -1.68*** -2.39*

Female participation rate 2.14* 4.57*** 1.64*** -0.22

Low-income workers-2.47*** -1.48** -5.68*** -6.26***

Industrialization level-2.76*** -1.93*** -0.60* 0.88

Page 14: Accounting for Spatial Heterogeneity in Educational Outcomes and International Migration in Mexico

SPATIALLY DEFINED REGIONS RESULTS SPATIAL LAG MODEL WITH STRUCTURAL CHANGE AND GROUPWISE HETEROSKEDASTICITY

Parameters and Significance

VariableNorth and

Center Northwest WestGulf and

South

International migration -1.853*** -3.7145*** -3.3764*** -2.3594***

Female participation rate 5.3197*** 0.0373 3.5359*** 0.8843***

Low-income workers -1.307*** -2.35366** -1.3685** -5.1795***

Industrialization level -2.419*** -1.81616* -1.6712** -0.1766

Page 15: Accounting for Spatial Heterogeneity in Educational Outcomes and International Migration in Mexico

CONCLUSIONS

• Negative effects of international migration on educational outcomes

• Results support hypothesis raised from an interaction between education, migration and labor market

• Regions based on spatial-varying links between dimensions studied are more appropriate to capture heterogeneity and diffusion processes than those defined previously by migration historicity

• Need to use proper geostatistical methods to test and develop hypotheses that imply spatial effects

• Regions are essential to consider how the relationships between sociodemographic variables shape geographical disparities