A life-course perspective on postponement and recuperation of fertility in Europe: long-term effects of

cheapecuadorianΔιαχείριση

28 Οκτ 2013 (πριν από 3 χρόνια και 9 μήνες)

60 εμφανίσεις






A life
-
course perspective on postponement and
recuperation of fertility in Europe: long
-
term effects of
economic recession on fertility in 17 EU
-
countries





Society for Longitudinal & Life
-
course studies

Developments & Challenges in Longitudinal Studies from Childhood

22
nd
/23
rd
/24
th

September 2010, Clare College, Cambridge, UK







Karel Neels






Sociology Department, Antwerp University






QASS
-
Programme
,
University College Brussels









Measuring ‘economic cycles’ and ‘fertility’...



Unemployment

and

consumer

sentiment

reflect

more

closely

impact

of

recession

than

more

general

indicators

as

GDP


Adsera

&

Menendez

(
2009
),

Van

Giersbergen

&

De

Beer

1997

in

Sobotka,

Skirbekk

&

Philipov

(
2010
)
;

Fokkema

et

al
.

(
2008
)


Unemployment

strongly

related

to

postponement

of

first

births


Adsera

(
2005
),

Santow

&

Bracher

et

al
.

(
2009
)

,

de

Cooman

et

al
.

(
1987
),

Rindfuss

(
1988
)

in

Sobotka

et

al
.

(
2010
)


Significant

impact

of

macro
-
level

unemployment

rate

persists

when

controlling

for

unemployment

spells

at

individual

level


Kravdal

(
2002
),

Hoem

(
2000
)

in

Sobotka

et

al
.

(
2010
)


Economic

recession

effects

vary

in

terms

of

gender,

age,

ethnic

background,

socio
-
economic

position

and

parity


Kravdal

(
2002
),

Mills

et

al
.

(
2005
),

Schmitt

(
2008
)

in

Sobotka

et

al
.

(
2010
)












Effect

of

recession

found

to

be

positive

among

lower

educated

women

and


negative

among

higher

educated


Francesconi

&

Goldt

(
2005
),

Andersson

(
2000
),

Liefbroer

(
2005
),

Schmitt

(
2008
),

Kravdal

(
2002
),

Meron

&

Widmer

(
2003
)

in

Sobotka

et

al
.

(
2010
)




Short
-
term

fertility

decline

often

interpreted

as

tempo

effect

with

little,

if


any,

impact

on

quantum,

but

research

actually

distinguishing

effects

on


tempo


and

quantum

currently

lacking


Sobotka,

Skirbekk

&

Philipov

(
2010
)




Macro
-
level

analysis

of

period
-
TFR

confounds

tempo

and

quantum


aspects

of

subsequent

transitions

that

may

be

differentially

affected


Ryder

(
1975
),

Bongaarts

&

Feeney

(
1998
)




Individual
-
level

hazard

models

contribute

to

unveiling

mechanisms


through

which

recession

affects

fertility

behavior,

but

difficult

to


generalize

into

aggregate

outcomes

for

fertility

trends

Measuring ‘economic cycles’ and ‘fertility’...





A life
-
course perspective on recession effects


Cohort

postponement
-
recuperation

model





Research Questions


R
1

Postponement


Does

economic

recession

affect

first

birth

hazards

of

all

age
-
groups

or

is

the

effect

more

pronounced

among

younger

age
-
groups?

R
2

Recuperation


Empirical

support

for

cohort

recuperation

effects

in

the

sense

that

the

negative

effect

of

economic

context

on

birth

hazards

at

younger

ages

is

offset

by

increasing

fertility

at

older

ages,

net

of

prevailing

economic

conditions

at

that

age

R
3

SES
-
Differentials


Do

socio
-
economic

groups

respond

differently

to

variation

in

economic

context?

Differences

more

pronounced

with

respect

to

postponement

of

recuperation?

R
4

Cross
-
country

Differentials


Do

differences

in

institutional

context

give

rise

variations

in

postponement
-
recuperation

patterns





Data


Belgium




Census

(
2001
)

Maternity

histories

of

women

aged

>
14

years




Contextual

unemployment

data

1947
-
2009

(CSB

&

OECD)

France




GGS

Wave

1

(
2005
)
:

fertility

histories

of

men

and

women

aged

18
-
79




Contextual

unemployment

data

1967
-
2009

(INSEE

&

OECD)

Netherlands




GGS

Wave

1

(
2003
)

fertility

histories

of

men

and

women

aged

18
-
79




Contextual

unemployment

data

1952
-
2009

(CBS

&

OECD)

EU
-
17




European

Social

Survey

(
2006
)

year

of

first

birth




Macro
-
level

unemployment

rates

from

OECD,

1956
-
2010








Methods & Models



Discrete
-
time

event
-
history

models

of

first

birth

hazard


Postponement

model

(ages
15
-
49
)


-

cubic

effect

of

age

allowing

asymmetric

baseline

hazard

function


-

interaction

between

gender

and

baseline

function

(for

France

&

Netherlands)


-

interaction

between

unemployment

effect

(lagged

1

year)

and

5
-
year

age
-
groups


-

stratification

by

educational

attainment


-

stratification

by

country


Recuperation

Model

(ages

30
-
49
)


-

cubic

effect

of

age

allowing

asymmetric

baseline

hazard

function


-

interaction

between

gender

and

baseline

function

(France

&

Netherlands)


-

interaction

between

unemployment

(lagged

1

year)

and

5
-
year

age
-
groups


-

interaction

between

unemployment

(lagged

10

years)

and

5
-
year

age
-
groups


-

stratification

by

educational

attainment


-

stratification

by

country





Results on Postponement: Belgium


Unemployment

Rate

&

Synthetic
-
PPR
1


Belgium,

women

15
-
49
,

1960
-
2000



Results on Postponement: Belgium

Cross
-
correlation of Unemployment Rate with:

SPPR1

SPPR2

SPPR3

SPPR4

Lag (in years)

CC

SE

CC

SE

CC

SE

CC

SE

0

-
.430

.158

-
.386

.158

-
.367

.158

-
.296

.158

1

-
.612

.160

-
.199

.160

-
.160

.160

-
.085

.160

2

-
.500

.162

.006

.162

.109

.162

.107

.162

3

-
.319

.164

.246

.164

.329

.164

.333

.164

4

-
.068

.167

.369

.167

.386

.167

.357

.167

5

-
.113

.169

.315

.169

.335

.169

.330

.169

6

-
.078

.171

.172

.171

.146

.171

.252

.171

7

.031

.174

.097

.174

.107

.174

.057

.174

8

.154

.177

.180

.177

.220

.177

.070

.177

9

.021

.180

.300

.180

.228

.180

.123

.180

10

.098

.183

.160

.183

.220

.183

.185

.183

11

-
.007

.186

.099

.186

.223

.186

.245

.186

12

.002

.189

.122

.189

.173

.189

.195

.189

13

.095

.192

.102

.192

.234

.192

.215

.192

14

.164

.196

.079

.196

.166

.196

.205

.196

15

.127

.200

-
.043

.200

.073

.200

.068

.200

(1)
Cross
-
correlation

function

is

calculated

at

lags

from

0

to

15

years
.

Non
-
seasonal

differencing

relative

to

the

value

in

the

preceding

year

is

applied

to

both

series

to

eliminate

bias

in

the

cross
-
correlation

coefficients

resulting

from

upward

of

downward

trends
.


(2)
Sources
:

OECD

&

Statistics

Belgium,

2001

Census,

calculations

by

author
.



Belgium,

women

15
-
49
,

1960
-
2000






Cross
-
correlation

unemployment

&

1
st

birth

hazard

age

20

Belgium,

women

15
-
49
,

1960
-
2000

Results on Postponement: Belgium





No education & Primary
Education

Lower Secondary
Education
(1)

Higher Secondary
Education
(1)

Short Type

Tertiary

Education

(1)

Long Type

Tertiary Education
(1)

-

1
-

-

2
-

-

3
-

-

4
-

-

5
-

-

6
-

-

7
-

-

8
-

-

9
-

-

10
-

Age


Linear

2.044

C

2.258

C

2.578

C

2.910

C

3.353

C

3.892

C

5.223

C

5.323

C

10.066

C

10.222

C


Quadratic

0.951

C

0.943

C

0.937

C

0.927

C

0.929

C

0.915

C

0.914

C

0.908

C

0.890

C

0.884

C


Cubic

1.001

C

1.001

C

1.001

C

1.001

C

1.001

C

1.002

C

1.001

C

1.002

C

1.002

C

1.002

C

Age
-
groups (5 years) *
Current Unemployment Rate

(lagged by 1 year)


age 15
-
19

0.954

C

0.983

B

0.959

C

0.928

B

1.029

N


age 20
-
24

0.957

C

0.970

C

0.963

C

0.937

C

0.910

C


age 25
-
29

0.939

C

0.968

C

0.973

C

0.973

C

0.954

C


age 30
-
34

0.972

C

0.996

N

0.997

N

0.995

N

0.979

A


age 35
-
39

1.013

N

1.021

N

1.007
N

1.045

C

1.021

N


age 40
-
44

0.931

N

0.922

A

0.955

N

1.001

N

1.008

N


age 45
-
49

0.688

A

-

-

-

-
2LL

18374.7

18211.5

30814.7

30689.3

35496.7

35351.1

23838.6

23649.8

7051.6

6986.8

Legend:


N

p > 0.05;
A

p < 0.05;
B

p < 0.01 and
C

p < 0.001.

Note (1):

Analysis restricted to ages 15
-
44 years in models 3
-
8.

Source:

Statistics Belgium, 2001 Census, Calculations by author

Belgium,

women

15
-
49
,

1960
-
2000

Results on Postponement: Belgium





Educational Level:

ISCED 0
-
2

ISCED 3&4

ISCED5&6

ISCED ALL

Age centered


Linear

0,906

C

0,960

A

1,062

B

0,953

C


quadratic

0,976

C

0,974

C

0,970

C

0,976

C


Cubic

1,001

C

1,001

C

1,001

C

1,001

C

Female

1,367

C

1,161

A

1,353

C

1,225

C

Age centered*Female


Linear

0,936

C

0,912

C

0,922

C

0,936

C


Quadratic

1,003

A

1,005

B

1,002

N

1,004

C


Cubic

1,000

A

1,000

N

1,000

N

1,000

B

Age
-
groups (5 years)*
Current

Unemployment Rate (lagged by 1 year)


age 15
-
19

0,978

N

0,860

C

0,774

B

0,871

C


age 20
-
24

0,942

C

0,921

C

0,874

C

0,897

C


age 25
-
29

0,946

C

0,951

C

0,935

C

0,940

C


age 30
-
34

1,002

N

0,996

N

0,988

N

1,000

N


age 35
-
39

1,092

C

1,028

N

1,031

N

1,050

C


age 40
-
44

1,052

N

1,051

N

1,074

N

1,071

B


age 45
-
49

0,840

N

0,993

N

0,926

N

Constant

0,163

C

0,205

N

0,177

C

0,191

C

-
2LL

10365,5

15827,9

9446,4

36443,7

Legend:

N

p > 0.05;
A

p < 0.05;
B

p < 0.01 and
C

p < 0.001
-

Age centered at 28 years

Source: GGS France, 2005, Calculations by author.

France,

women

15
-
49
,

1968
-
2004

Results on Postponement: France





Educational Level:

ISCED 0
-
2

ISCED 3&4

ISCED5&6*

ISCED ALL

Age centered


Linear

0,998

N

1,129

C

1,197

C

1,082

C


Quadratic

0,974

C

0,974

C

0,970

C

0,975

C


Cubic

1,001

C

1,001

C

1,001

A

1,001

C

Female

1,575

C

1,324

C

0,829

N

1,382

C

Age centered*Female


Linear

0,930

C

0,960

B

1,021

N

0,938

C


Quadratic

1,003

N

1,002

N

1,008

N

1,004

B


Cubic

1,000

N

1,000

A

0,999

N

1,000

C

Age
-
groups (5 years)*
Current

Unemployment Rate (lagged by 1 year)


age 15
-
19

0,890

B

0,845

B

-

0,815

C


age 20
-
24

0,898

C

0,850

C

0,767

C

0,842

C


age 25
-
29

0,905

C

0,871

C

0,799

C

0,864

C


age 30
-
34

0,959

A

0,905

C

0,893

C

0,921

C


age 35
-
39

1,012

N

0,903

C

0,902

A

0,940

C


age 40
-
44

0,969

N

0,997

N

0,894

N

0,980

N


age 45
-
49

0,956

N

1,021

N

0,776

N

0,960

N

Constant

0,180

C

0,178

C

0,156

C

0,179

N

-
2LL

11889,1

18535,0

3077,9

34338,5

Legend
:

N

p

>

0
.
05
;

A

p

<

0
.
05
;

B

p

<

0
.
01

and

C

p

<

0
.
001
.

*

Model

restricted

to

ages

20
-
49

due

to

lack

of

variation

in

the

age

category

15
-
19
.

Source: GGS Netherlands, 2002
-
2004, Calculations by author.

Netherlands,

women

15
-
49
,

1962
-
2002

Results on Postponement: Netherlands





Results on recuperation: Belgium

Belgium,

women

30
-
49
,

1960
-
2000


Cross
-
correlation

unemployment

&

1
st

birth

hazard

age

30





Results on recuperation: Belgium



No education

& Primary Education

Lower Secondary
Education
(1)

Higher Secondary
Education
(1)

Short Type

Tertiary Educaton
(1)

Long Type

Tertiary Education
(1)

-

1
-

-

2
-

-

3
-

-

4
-

-

5
-

-

6
-

-

7
-

-

8
-

-

9
-

-

10
-

Age


Linear

4.646
N

5.855
N

1.683
N

1.093
N

0.025
N

0.027
N

1.035
N

1.093
N

0.422
N

0.631
N


Quadratic

0.929
N

0.920
N

0.965
N

0.986
N

1.913
N

1.188
N

1.006
N

1.004
N

1.053
N

1.031
N


Cubic

1.001
N

1.001
N

1.001
N

1.000
N

0.997
N

0.997
N

1.000
N

1.000
N

0.999
N

0.999
N

Age
-
groups (5 years) *
Current Unemployment Rate

(unemployment rate lagged by 1 year)


age 30
-
34

0.964
B

0.971
A

0.966
B

1.007
N

1.009
N

1.020
A

0/995
N

1.032
N

1.001
N


age 35
-
39

.0968
N

0.975
N

1.018
N

0.988
N

0.977
N

1.040
B

1.011
N

1.039
N

0.970
N


age 40
-
44

0.934
N

0.830
A

0.920
N

0.991
N

0.945
N

1.053
N

1.009
N

1.052
N

1.109
N

Age
-
groups (5 years) *
Past Unemployment Rate

(unemployment rate lagged by 10 years)


age 30
-
34

0.979
N

1.036
B

0.997
N

1.041
C

1.053
B


age 35
-
39

0.994
N

0.992
N

1.013
N

1.047
A

1.101
A



age 40
-
44

1.171
A

1.072
N

1.061
N

1.065
N

0.951
N

-
2LL

2964.4

2957.5

3693.2

3684.0

4235.0

4234.0

4568.9

4548.4

1994.9

1976.1

Legend:

N

p > 0.05;
A

p < 0.05;
B

p < 0.01 and
C

p < 0.001.

Note (1):

Analysis restricted to ages 15
-
44 in models 3
-
8.

Source:

Statistics

Belgium
, 2001 Census, Bewerking door auteur

Belgium,

women

30
-
49
,

1960
-
2000





Belgium,

women

30
-
49
,

1960
-
2000


Comparison

synthetic

and

cohort

parity

progression

Results on recuperation: Belgium





Educational Level:

ISCED 0
-
2

ISCED 3&4

ISCED5&6

ISCED ALL

Age centered


Linear

0,921

N

1,106

N

0,892

A

0,947

N


Quadratic

1,006

N

1,012

N

1,001

N

1,006

N


Cubic

1,000

N

0,996

B

1,000

N

0,999

N

Female

1,109

N

0,759

N

0,972

N

0,959

N

Age centered*Female


Linear

0,895

N

0,932

N

0,938

N

0,939

N


quadratic

0,990

N

1,001

N

0,983

N

0,991

N


Cubic

0,999

N

1,001

N

1,000

N

1,000

N

Age
-
groups (5 years)*
Current

Unemployment Rate (lagged by 1 year)


age 30
-
34

1,106

N

0,990

N

0,987

N

1,000

N


age 35
-
39

1,191

A

0,870

B

0,980

N

0,991

N


age 40
-
44

1,104

N

0,872

N

1,178

N

1,017

N


age 45
-
49

0,684

N

0,832

N

1,054

N

0,851

N

Age
-
groups (5 years)*
Past

Unemployment Rate (lagged by 10 years)


age 30
-
34

0,960

N

1,063

B

1,071

B

1,056

B


age 35
-
39

0,890

A

1,117

A

1,083

N

1,041

N


age 40
-
44

0,885

N

1,071

N

0,870

N

0,967

N


age 45
-
49

1,341

N

1,289

N

0,951

N

1,160

N

Constant

0,027

C

0,078

C

0,061

C

0,055

C

-
2LL

1832,1

3119,3

2985,4

8015,5

Legend:
N

p > 0.05;
A

p < 0.05;
B

p < 0.01 and
C

p < 0.001


Age centered at 35
years

-

Source: GGS France, 2005, Calculations by author.

France,

men

and

women

aged

30
-
49
,

1977
-
2004

Results on recuperation: France





France,

men

and

women

aged

30
-
49
,

1977
-
2004


Unemployment

&

lagged

cohort

parity

progression

Results on recuperation: France





Educational Level

ISCED 0
-
2

ISCED 3&4

ISCED5&6

ISCED ALL

Age


linear

0,924

N

0,859

C

0,840

B

0,868

C


quadratic

1,009

N

0,983

B

0,983

N

0,990

B


Cubic

0,998

N

1,001

N

1,001

N

1,000

N

Female

1,060

N

0,885

N

0,977

N

0,943

N

Age centered*Female


Linear

0,896

N

0,897

B

0,927

N

0,904

C


Quadratic

0,996

N

0,996

N

0,988

N

0,995

N


Cubic

1,000

N

1,000

N

1,001

N

1,000

N

Age
-
groups (5 years)*
Current

Unemployment Rate (lagged by 1 year)


age 30
-
34

0,940

A

0,931

C

0,912

A

0,934

C


age 35
-
39

0,983

N

0,799

C

0,804

B

0,858

C


age

40
-
44

0,881

N

0,958

N

0,745

N

0,907

N


age 45
-
49

1,098

N

1,204

N

0,454

N

1,068

N

Age
-
groups (5 years)*
Past

Unemployment Rate (lagged by 10 years)


age 30
-
34

1,068

B

1,037

A

1,010

N

1,039

B


age 35
-
39

1,020

N

1,191

C

1,162

B

1,143

C


age 40
-
44

1,018

N

1,101

N

1,329

A

1,118

A


age 45
-
49

1,075

N

0,864

N

1,679

N

0,984

N

Constant

0,056

C

0,101

C

0,122

C

0,087

N

-
2LL

2770,7

6860,3

1779,4

11485,9

Significance Levels:

N

p > 0.05;
A

p < 0.05;
B

p < 0.01 and
C

p < 0.001.

Source: GGS Netherlands, 2002
-
2004, Calculations by author.

Netherlands,

men

and

women

aged

30
-
49
,

1962
-
2002

Results on recuperation: Netherlands





Netherlands,

men

and

women

aged

30
-
49
,

1962
-
2002


Unemployment

&

lagged

cohort

parity

progression

Results on recuperation: Netherlands





Netherlands,

men

and

women

aged

30
-
49
,

1962
-
2002

Postponement & Recuperation in EU
-
17

Model 1

Model 2

Model 3

Postponement

Recuperation: gross effect

Recuperation:net effect

Covariates

Exp(b)

Sig.

Exp(b)

Sig.

Exp(b)

Sig.

Baseline hazard function

Linear

2.283

***

2.274

***

2.279

***

Quadratic

.962

***

.962

***

.962

***

Cubic

1.001

***

1.001

***

1.001

***

Sex of Respondent

Female

8.812

***

9.443

***

9.463

***

Sex*baseline interaction

linear*female

.778

***

.754

***

.753

***

quadratic*female

1.011

**

1.013

**

1.014

**

cubic*female

1.000

-

1.000

-

1.000

*

Country

(D
etailed country
-
differentials omitted)

Current unemployment rate

age 15
-
19

.976

**

age 20
-
24

.977

***

age 25
-
29

.976

***

age 30
-
34

.991

-

1.003

-

age 35
-
39

.993

-

.991

-

age 40
-
44

.963

-

1.007

-

age 45
-
49

.863

-

.852

-

Past unemployment rate

age 30
-
34

1.020

***

1.017

**

age 35
-
39

1.033

***

1.039

**

age 40
-
44

1.015

-

1.007

-

age 45
-
49

.957

-

1.058

-





Discussion & Conclusion


SES

Differentials

in

Fertility

Postponement




unemployment

exerts

a

negative

effect

on

first

birth

hazards,

particularly

among


younger

women




between
-
country

variation

of

SES
-
differentials

in

postponement
:


-

low

SES

in

Belgium

affected

between

ages

15
-
34
,

whereas

high

SES


predominantly

affected

between

ages

20
-
29


-

more

uniform

effect

of

economic

conditions

in

France,

adversely

affecting

all


SES
-
groups

between

ages

15
-
29


-

effect

most

pronounced

in

the

Netherlands,

negatively

affecting

low

SES


between

ages

15
-
34

and

high

SES

between

ages

15
-
39


Negative

effect

of

unemployment

likely

to

reflect

difficult

entry

into

labour

market
:



-

more

pronounced

among

younger

age

groups


-

less

pronounced

on

2
nd

and

higher
-
order

births








Discussion & Conclusion


SES

Differentials

in

Fertility

Recuperation




empirical

evidence

for

cohort

recuperation
-
effects

at

older

ages,

net

of

prevailing


unemployment

rates

at

that

point

in

the

life
-
course




between
-
country

variation

in

SES

differentials

stronger

for

recuperation


-

in

France,

lagged

cohort

parity

progression

suggests

limited

long
-
term

effect

of


economic

context

on

fertility,

regardless

SES


-

in

Belgium

pronounced

SES
-
differentials

in

recuperation

with

only

higher


educated

recuperating

postponed

births
.

SES

differentials

in

fertility

may

be


linked

to

strong

differentials

in

uptake

of

childcare

arrangements,



.


-

In

Netherlands

prevailing

economic

conditions

after

age

30

overrule


recuperation

effects,

with

cohort

completed

fertility

of

first

births

being

strongly


shaped

by

variations

in

economic

context

for

all

SES


Period

measures

as

PTFR
1

but

also

SPPR
1

are

biased

measures

of

quantum

as

they

accumulate

postponement

effects

between

ages

15
-
39

while

ignoring

recuperation







Limitations & Future Work


Limitations




Need

to

control

aggregate
-
level

unemployment

rate

for

unemployment

at


individual

and

household

level




Specify

different

time
-
lags

for

recuperation

by

SES




Additional

checks

on

robustness

of

findings

required

by

limiting

time

frame




Explore

interactions

between

unemployment

effect

and

gender


Future

work




extend

indicators

of

macro
-
economic

context

to

GDP,

inflation,






extend

analysis

of

macro
-
economic

context

to

second

and

higher
-
order

births




elaborate

between
-
country

variation

and

SES
-
differentials

to

(changes)

in

social


policy

context

and

SES
-
differentials

in

uptake

of

arrangements




expand

number

of

countries









Thank you for your attention!