Embed Size (px)
Citation preview
Albert Sabater and Elspeth Graham
University of St Andrews
Emigration and Fertility Decline in Spain
Since the Economic Recession:
A Population-Level Analysis
2nd Human Fertility Database Symposium
Population-level Fertility Research: State of the Art
WissenschaftsForum, Berlin
23-24 June 2016
In most studies that examine how economic uncertainty
relates to fertility the focus is on unemployment (Goldstein et
al. 2013; Schneider 2015; Sobotka et al. 2011)
Empirical evidence supports a procyclical relationship
between fertility rates and business cycle fluctuations,
partly reflecting postponement until conditions improve (Sobotka et al. 2011)
Migration-related factors have also been used to explain
short- and long-term fertility changes (Billari and Dalla-
Zuanna 2011; Sobotka 2008), albeit the focus is normally
on immigration not emigration
Background
Unemployment-emigration-fertility nexus
Unemployment
Fertility
Emigration
1
2
3
1 Rising unemployment usually associated with fertility decline
2 Unemployment may lead to emigration, although emigration is not necessarily
caused by unemployment
3 Rising emigration may affect fertility via compositional change, as a proportion of
the most fertile age groups emigrate
Unemployment, migration and fertility in Spain
-300
00
0-1
50
00
0
0
150
00
03
00
00
0
Net M
igra
tion
2008 2009 2010 2011 2012 2013Year
Total Men Women
51
01
52
02
53
0
Une
mplo
ym
en
t R
ate
20062001 2002 2003 2004 2005 2007 2008 2009 2010 2011 2012 2013Year
Total Men Women
.81
1.2
1.4
1.6
1.8
Tota
l F
ert
ility
Ra
te
2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013Year
Data & methods
Panel dataset (period 2009-2013 for 48 provinces)
TFR and ASFRs (outcome variables)
Women’s unemployment rate
Women’s total emigration (flow) by country of birth
Other controls: women’s total immigration (flow), female
population (stock) by country of birth, and mean age at birth-
first birth (diff)
(All variables are log-transformed, and all predictors are lagged by one year)
A two-step approach
Panel regressions (OLS with provincial FE) (Stata’s xtreg)
Spatial panel regressions (Spatial Durbin with FE)
(Stata’s xsmle)
Heterogeneity across provinces. Spain, 2009-2013
11
.21
.41
.6
Tota
l F
ert
ility
Ra
te
2009 2010 2011 2012 2013Year
Province Mean
10
20
30
40
50
60
Wom
en's
Un
em
plo
ym
ent
Ra
te
2009 2010 2011 2012 2013Year
Province Mean
-200
0-1
00
0
0
100
02
00
0
Net M
igra
tion
2009 2010 2011 2012 2013Year
Province Mean
Spatial dependence. TFR in Spanish provinces, 2013
0.96 - 1.17
1.17 - 1.28
1.28 - 1.34
1.34 - 1.52
-2-1
01
2
Sp
atia
lly lag
ged
TF
R
-2 -1 0 1 2TFR
WTFR Fitted values
(Moran's I=0.6594 and P-value=0.0010)
RESULTS
Panel data analysis with provincial fixed-effects,
Spain 2009-2013 - Beta coefficients for unemployment
Significance at the 95 confidence level (otherwise in light shade)
-0.80
-0.70
-0.60
-0.50
-0.40
-0.30
-0.20
-0.10
0.00
0.10
TFR 15-49 ASFR 20-24 ASFR 25-29 ASFR 30-34 ASFR 35-39
OLS with FE (basic model) OLS with FE (full model) Spatial Durbin with FE (full model)
Panel data analysis with provincial fixed-effects,
Spain 2009-2013 – TFR (outcome)
-0.15
-0.12
-0.09
-0.06
-0.03
0
0.03
0.06
0.09
0.12
0.15U
nem
plo
ymen
t
Mea
n a
ge a
t b
irth
Span
ish
E
Oth
er E
uro
pea
n E
Afr
ican
E
Lati
nam
eric
an E
Asi
an E
OLS with FE Spatial Durbin with FE
Significance at the 95 confidence level (otherwise in light shade)
Panel data analysis with provincial fixed-effects,
Spain 2009-2013 – TFR (outcome)
-0.15
-0.12
-0.09
-0.06
-0.03
0
0.03
0.06
0.09
0.12
0.15
Un
emp
loym
ent
Mea
n a
ge a
t b
irth
Span
ish
E
Oth
er E
uro
pea
n E
Afr
ican
E
Lati
nam
eric
an E
Asi
an E
Spatial Durbin with FE
Direct effect Indirect effect
Significance at the 95 confidence level (otherwise in light shade)
Higher levels of unemployment are associated with
lower fertility levels, although the effect is concentrated
and significant among the youngest age groups
The negative effect of unemployment on fertilty is
composed of direct and indirect effects on almost equal
terms
Higher levels of emigration are also associated with
lower fertility levels, although the effect is concentrated
and significant among two groups (Spanish and Asian)
The negative effect of emigration on fertilty is mainly
composed of direct effects
Findings (1)
Findings (2)
Our analyses indicate that unemployment and migration
data are not independent, but rather spatially dependent
(i.e. observations from one province exhibit values
similar to those from nearby provinces)
Spatial panel fixed-effect models corroborate that the
non-spatial panel fixed-effects models underestimate the
impact of unemployment and emigration
APPENDIX
Panel data analysis with provincial fixed-effects,
Spain 2009-2013 – ASFR 25-29 (outcome)
-0.15
-0.12
-0.09
-0.06
-0.03
0
0.03
0.06
0.09
0.12
0.15U
nem
plo
ymen
t
Mea
n a
ge a
t b
irth
Span
ish
E
Oth
er E
uro
pea
n E
Afr
ican
E
Lati
nam
eric
an E
Asi
an E
OLS with FE Spatial Durbin with FE
Significance at the 95 confidence level (otherwise in light shade)
Panel data analysis with provincial fixed-effects,
Spain 2009-2013 – ASFR 30-34 (outcome)
-0.15
-0.12
-0.09
-0.06
-0.03
0
0.03
0.06
0.09
0.12
0.15U
nem
plo
ymen
t
Mea
n a
ge a
t b
irth
Span
ish
E
Oth
er E
uro
pea
n E
Afr
ican
E
Lati
nam
eric
an E
Asi
an E
OLS with FE Spatial Durbin with FE
Significance at the 95 confidence level (otherwise in light shade)
Panel data analysis with provincial fixed-effects,
Spain 2009-2013 – ASFR 35-39 (outcome)
-0.15
-0.12
-0.09
-0.06
-0.03
0
0.03
0.06
0.09
0.12
0.15U
nem
plo
ymen
t
Mea
n a
ge a
t b
irth
Span
ish
E
Oth
er E
uro
pea
n E
Afr
ican
E
Lati
nam
eric
an E
Asi
an E
OLS with FE Spatial Durbin with FE
Significance at the 95 confidence level (otherwise in light shade)
