Internal Migration and EducationToward Consistent Data Collection Practices
for Comparative Research
AUDE BERNARD & MARTIN BELL
QUEENSLAND CENTRE FOR POPULATION RESEARCHUNIVERSITY OF QUEENSLAND, AUSTRALIA
Background
Significance of internal migration◦ 763 million people worldwide (4 times the # of international migrants)
◦ 10-14 times moves over the lifetime
◦ Young adults are the most mobile
◦ Linked to key life-course transitions, including education
◦ Integral to human development
◦ Main driver of population change
Key features◦ Most volatile component of population change◦ Difficult to measure – no international standards ◦ Theoretically challenging
Variations in levels of internal migration
0
10
20
30
40
50
60N
ew Z
eal
and
Sou
th K
ore
aU
SAA
ust
ralia Fi
jiC
anad
aP
anam
aC
hile
Swit
zerl
and
Sen
egal
Fran
ceC
ame
roo
nP
arag
uay
Isra
el
Jap
anM
on
golia
Bar
bad
os
Bo
livia
Kyr
gyzs
tan
Per
uU
rugu
ayG
uin
eaSo
uth
Afr
ica
Mo
rocc
oM
alta
Uga
nd
aC
amb
od
iaR
wan
da
Gre
ece
Bra
zil
Arg
enti
na
Mal
aysi
aC
ost
a R
ica
Tun
isia
Gh
ana
Gu
atem
ala
Do
min
ican
…P
ort
uga
lC
hin
aV
ietn
amIn
do
nes
iaEl
Sal
vad
or
Mau
riti
us
Cu
ba
Ho
nd
ura
sTh
aila
nd
St L
uci
aN
icar
agu
aEc
uad
or
Hai
tiP
hili
pp
ines
Spai
nIr
aqN
epal
Ven
ezu
ela
Mal
iN
ort
h K
ore
aM
exic
oEg
ypt
Ind
ia
% c
han
ged
ad
dre
ss in
th
e la
st f
ive
ye
ars
Bell et al. (2015), IMAGE project https://imageproject.com.au/
Spatial patterns of movementCHINA THAILAND
Charles-Edwards et al. (2016)
Education as a reason for moving
Level of education-related migration varies significantly between countries
Accounts for a small proportion of all moves
Few countries collect reasons for moving at their census
Sample size from surveys often too small to disaggregate by age
31
21
17
16
15
13
11
10
4
1
42
49
29
45
8
16
42
50
33
9
8
18
32
33
58
40
21
25
14
52
12
2
28
11
8
42
32
8
12
22
3
19
3
15
6
7
6
0% 20% 40% 60% 80% 100%
China
Thailand
Colombia
Indonesia
Iran
Egypt
Mexico
Cambodia
India
Iraq
education employment family Marriage other
Bernard et al. (forthcoming)
Reasons for long-distance migration (15-24 years)
Educational selectivity of migrants
Probability of moving increases sharply with education
Holds across all world regions
Educational gradient less pronounced in Latin America
1.7
2.8
3.7
0
1
2
3
4
5
Africa (n=20) Asia (n=17) Europe andNorth America
(n=13)
Latin America(n=18)
World (n=68)
Mig
rati
on
Ind
ex (
no
ed
uca
tio
n=1
)
Primary education Secondary education Tertiary education
Bernard et al. (forthcoming)
Educational selectivity of migrants
Significant variations between countries
The impact of tertiary education is what strongly differentiate countries
Reasons for this are poorly understood
MalaysiaThailand
IndonesiaChina
IranKyrgyzstan
VietnamArmenia
PhilippinesFiji
CambodiaJordan
MongoliaTurkey
IndiaIraq
Nepal
0 2 4 6 8 10
Migration index (no education=1)
Asia
Bernard et al. (forthcoming)
Impact of migration on children’s education Chronic mobility is detrimental to
educational performance
Need to better understand Type of moves (rural vs. urban)
Duration (temporary versus permanent)
Frequency of movement
Impact of regulatory framework (e.g. China’s Hukou system)
Impact on both enrolment and performance
Lu and Zhou (2013)
Rural-to-urban migration & language score in primary school in Beijing
Educational composition of migration flows and their impacts
Educational selectivity varies with flow types
The volume of particular flows affects the redistribution of human capital
Limitation of current data sources Few countries collect urban status of previous place of
residence
Definition of urban areas not consistent
Limited geographic detail
% of migrants with at least secondary education
0%
10%
20%
30%
40%
50%
60%
70%
urban to urban rural to urban
urban to rural rural to rural
Bernard et al. (forthcoming)
Economic Returns to Internal Migration
Positive effect of migration on earnings increase with education
Limited understanding how it varies over time and across countries ◦ Need comparable longitudinal datasets or retrospective residential histories
◦ Need high spatial and temporal resolution
◦ Urban status at both origin and destination
Internal migration data collection practices
Most countries collect some form of migration data ◦ 179/183 countries since 1995
Different collection instruments◦ Census (88 %)
◦ Population register or administrative source (28%)
◦ National surveys (61 %)◦ e.g. Demographic and Health Surveys (90 countries since 1980)
6 in 10 countries draw data from multiple sources
Bell et al. (2014) – IMAGE Repository
Measuring internal migration Different forms of data ◦ Transition
◦ Event
◦ Duration
Time interval (2000 UN Census round)◦ 1 year interval (20%)
◦ 5 year interval (37%)
◦ Undefined (23%)
◦ Lifetime (86%)
Spatial framework◦ All changes of address
◦ # of zones
Characteristics of migrants◦ Sex
◦ Age
◦ Education
Bell et al. (2014) – IMAGE Repository
Strength of each data source
Census Survey
- Enumeration of full the population
- High spatial resolution
- Long historical time series
- Large range of covariates, including education
- Individual-level migration data for 60 countries available in IPUMS
- Can collect detailed migration histories
- Can collect reasons for moving
- Large range of covariates, including education
- Information on place of origin
- Representative of whole population or tailored to particular groups (youths, refugees, ..)
- Ability to modify questions
- Detailed temporal coverage (longitudinal data collected annually or retrospective life histories)
Limitations of each data source
Census Survey
- Covariates measured at the end observation of period
- Do not pick up return or multiple moves
- Coarse temporal coverage (most censuses are decennial)
- Cross-sectional data
- Limited information on place of origin
- Some countries do not disseminate data
- Spatial detail often coarse
- Variability in format limits comparability of migration
- Ongoing national surveys rare in developing countries
- Longitudinal data collected over many years
- Small sample size limits disaggregation by age or region
- Sampling error
- Recall error
- Panel attrition
Utility of each data sourceCensus Survey
Migrant selectivity
- Large range of covariates, including education but measured at the end of observation period
- Large range of covariates, including education
- Analysis of causal relationships
Spatial analysis & redistribution of human capital
- High spatial resolution - Enumeration of the full population
permits spatial disaggregation - Limited information on place origin
- Limited spatial resolution- Sample size restricts spatial
disaggregation - Information on place of origin
Trend analysis- Historical trends but infrequent observations and data is cross-sectional
- Recent trends with annual observations
Cross-nationalcomparisons
- Individual-level migration data for 60 countries available in IPUMS
- Variability in format limits comparability of migration
Recommendations for cross-national comparisons Migration data should be collected in a comparable format
◦ Internal migration best measured as an event or over a fixed interval (ideally 1 or 5 years)◦ Changes of address should be collected ◦ Place of residence, past and present, should be coded to the smallest geographical units feasible◦ Usual residence should be defined using a threshold criterion of 6 months.
Longitudinal surveys ◦ long-term investment & problem of attrition
Retrospective residential histories◦ Cost-effective and immediate results (collected in one wave as part of on-going survey)◦ Collect full residential histories (All domestic and international moves since birth, address, date, reasons for moving)◦ Collect histories in other life domains (education, employment, family)◦ Increasingly common
◦ Europe (SHARE, ELSA ), China (CHARLS ) and US (HRS )◦ Comparable questionnaires ◦ Life-history grids improve recall
Annotated reference listBell, M, Blake, M, Boyle, P, Duke-Williams, O, Rees, P, Stillwell, J & Hugo, G 2002, 'Cross-national comparison of internal migration: issues and measures', Journal of the Royal Statistical Society: Series A (Statistics in Society), vol. 165, no. 3, pp. 435-64. This paper reviews key issues in comparing internal migration between countries and proposes a set of summary measures for the cross-national comparison of migration
Bell, M, Charles-Edwards, E, Kupiszewska, D, Kupiszewski, M, Stillwell, J & Zhu, Y 2015, 'Internal Migration Data Around the World: Assessing Contemporary Practice', Population, Space and Place, vol. 21, no. 1, pp. 1-17. This paper reviews contemporary migration data collection practices around the world and provides a set of recommendations to enhance the utility and comparability of internal migration data.
Bell, M, Charles-Edwards, E, Kupiszewska, D, Kupiszewski, M, Stillwell, J & Zhu, Y 2015, 'Internal migration and development: comparing migration intensities around the world', Population and Development Review, vol. 41, no. 1, pp. 33-58. This paper proposes a robust method to compare levels of internal migration between countries and assembles a league table of 96 countries.
Bernard A., Bell, M, & Cooper J. (Forthcoming) Internal migration and education: A cross-national comparison Background paper prepared for the 2018 Global Education Monitoring Report, UNESCO, Paris This paper examines the links between education and internal migration in 58 countries from around the world.
Bernard, A 2017, 'Cohort Measures of Internal Migration: Understanding Long-Term Trends', Demography, vol. 54, no. 6, pp. 2201-21. This paper proposes a set of cohort measures of internal migration to can be applied to retrospective survey data to analyses migration trends and patterns between successive cohorts and countries. It demonstrates the utility of a cohort approach by applying the proposed measures to retrospective residential histories collected in England.
Bernard, A 2017, 'Levels and patterns of internal migration in Europe: A cohort perspective', Population Studies, vol. 71, no. 3, pp. 293-311. This paper uses retrospective survey data from 14 European countries to explore cross-national variations in the level and patterns of internal migration. It reveals the demographic mechanism underpinning enduring regional variations in migration level.
Lu, Yao, and Hao Zhou. "Academic achievement and loneliness of migrant children in China: School segregation and segmented assimilation." Comparative education review 57.1 (2012): 85-116. This paper compares the academic achievement of children from migrant and non-migrant families by type of school in Beijing.
Vidal, Sergi, and Janeen Baxter. For the sake of the children? A longitudinal analysis of residential relocations and academic performance of Australian children. Life-Course Centre Working Paper Series No. 2016-14, Brisbane, Australia This paper assesses the effect of residential relocation on the academic performance of children in Australian and found a very limited impact.
