Document

advertisement
Reversing the reversal?
The cross-country correlation between
female labour market participation and
fertility revisited
Anna Matysiak and Tomáš Sobotka
“Family dynamics, fertility choices, and family policy”
Oslo 9-10.10.2014
Background
• Many publications since around 2000 (esp. 2000-2005): the initially
negative cross-country correlation between female labour force
participation (FLFP) and period total fertility rates (TFR) reversed
from negative to positive in (Western) Europe and
OECD countries around mid-1980s
Source: Engelhardt and Prskawetz (2004)
Source: Rindfuss et al. (2003)
Background
The reversal received many interpretations:
• Cross-country differences in policies, work-family
incompatibility (Rindfuss et al. 2004)
• Labour market rigidities in Southern Europe (Ahn and Mira
2002, Adsera 2005)
• Time series correlations remain negative or turn
insignificant (Engelhardt et al. 2004, Kögel 2004)
• Split correlation (Lesthaeghe and Permanyer 2014)
Revisiting the correlation
The positive TFR – FLFP correlation has become one of the key
stylised facts about fertility in the developed world
• How solid is this link?
Our contribution:
• Looking at the recent data through 2012 (most previous research
extends until around 2000)
• Refining the measurement of the LFP: excluding early reproductive
(time in education) and post-reproductive ages
• Addressing the split correlation
• Looking at smaller regions (NUTS2)
• Adding a cohort dimension; addressing possible tempo effect in
period TFR
• Extending the analysis to CEE countries (wherever possible)
Data
Focus:
• Period: 1975-2012
• Cohorts: 1950-1975
• 37 countries: EU-15, Norway and Switzerland, 5 outside Europe
(Australia, Canada, Japan, New Zealand, US), 14 Central & Eastern
Europe (some analyses only)
FLFP by age (focus: 25-39 and 25-44):
• from LFS and population censuses, mostly from OECD Employment
Database, ILO Laborsta Database + national statistical offices
• cohort measures (age 25-39) reconstructed from period FLFP
Period and cohort TFRs:
• computations based on Eurostat data, national statistical offices,
Human Fertility Database; small part of cohort TFR projected for
cohorts 1970-75
Changing the age groups of LFP
measures
The cross-country correlation
between FLFP and TFR, 1975-2012
Data for “Western” OECD countries, different age specifications of FLFP
0.80
0.60
TFR x FLFP correlation
0.40
0.20
0.00
-0.20
-0.40
25-39 all
-0.60
15-64 all
25-44 all
-0.80
1975
1980
1985
1990
1995
Year
2000
2005
2010
Why has the correlation disappeared?
Why has the correlation disappeared?
Labour force participation rates
(%), women aged 25-39
Period total fertility rate
100.0
2.80
90.0
2.60
80.0
2.40
70.0
60.0
2.20
50.0
2.00
40.0
30.0
1.80
20.0
1.60
10.0
0.0
LOW
MEDIUM-LOW
MEDIUM-HIGH
HIGH
1970
1972
1974
1976
1978
1980
1982
1984
1986
1988
1990
1992
1994
1996
1998
2000
2002
2004
2006
2008
2010
2012
1970
1972
1974
1976
1978
1980
1982
1984
1986
1988
1990
1992
1994
1996
1998
2000
2002
2004
2006
2008
2010
2012
1.40
LOW
MEDIUM-LOW
MEDIUM-HIGH
HIGH
The high LFP countries (LFPR in 1980 >70%): Denmark, Finland, Sweden
The medium-high LFP countries (LFPR in 1980 60-70%): Portugal, Belgium, Canada, France, United States,
Norway
The medium-low LFP countries (LFPR in 1980 50-60%): Japan, Australia, Austria, United Kingdom, West Germany
The low LFP countries (LFPR in 1980 <50%): Ireland, Spain, Greece, Netherlands, New Zealand, Italy, Luxembourg
Why has the correlation disappeared?
Simple simulation: what would happen if we “froze” the TFR and /or FLFP at their 1995
levels?
cross-country correlation between
period TFR and FLFP (ages 25-39)
0.6
Until mid-1990s:
Trends in correlation
driven by changes in
the TFR
0.4
0.2
0
From around 2000:
Trends in correlation
driven by changes
(convergence) in
FLFP levels
-0.2
-0.4
-0.6
-0.8
observed
year
LFP frozen at 1995
TFR frozen at 1995
Changing the country coverage
and the analytical units
Split - correlation
2010
1998
What if we include CEE
countries?
No correlation would be observed in the 1990s and 2000s if the CEE
countries were taken into account.
What if we go down to subnational
level?
The correlation at the regional (NUTS-2) level in EU-15, Norway and Switzerland has
been negative (in both country groups) but has been weakening in recent years.
The developments in CEE were different.
Note: Non-European countries are excluded from this analysis
The cross-region correlation is based on within country variation in TFR and FLFPR
Using cohort measures
Why study cohort correlation?
• Life-cycle indicators of fertility & labour involvement
• Unaffected by short-term ups and downs that can influence period
data
• Solving the “tempo effect” problem in the period TFR: cohort fertility
gives the real measure of fertility “quantum” unaffected by the
changes in timing of childbearing
The cohort “story”: much weaker
correlation, but similar trends
The cross-country correlation between cohort TFR and cohort FLFP as compared with the
period correlation, 23 OECD countries
1.00
0.80
Period correlation (lagged by 30 years: year = cohort + 30)
0.40
0.20
0.00
-0.20
Cohort correlation
-0.40
-0.60
-0.80
-1.00
1950
1952
1954
1956
1958
1960
1962
1964
1966
1968
1970
1972
1974
1976
1978
1980
1982
TFR x FLFP correlation
0.60
Birth cohort
The cohort
correlation peaks
in the 1970
cohort at a low
level of 0.28
Discussion
Conclusions
•
•
•
The period cross-country correlation for „Western” OECD countries
disappears after 2010 when FLFP analysed for the key reproductive &
post-education ages only (25-39)
This trend driven by a strong cross-country convergence towards a
similar high FLFP levels around 80%
The finding of the strong positive period cross-country correlation for
„Western” OECD countries in the 1990s was partly driven by:
– Measures used (period indicators, broad age groups in LFP measures)
– Analytical units used (negative although weakening correlation at the
regional level)
– Countries covered (no correlation after adding CEE countries, split
correlation)
•
Weakening cross-regional within-country correlation is appealing and
requires further investigations
Discussion
Future: The correlation may come and go as the relationship
becomes volatile due to similar levels of LFP
 small changes in LFP or TFR may trigger shifts in correlations
Adsera (2004) had the most accurate foresight
It is likely, however, that the positive correlation between fertility and
female labor participation may fade away (…)
As women in countries with the lowest participation rates gradually enter
the labor force, female participation rates will slowly converge across
developed countries. However, if their fertility does not increase (due to
lack of changes in labor market institutions),the relation between fertility
and participation in the cross-section of OECD countries should become
flat in the near future.
Adserà, A. 2004. “Changing fertility rates in developed countries. The impact of
labour market institutions.” Journal of Population Economics 17(1): 1-27.
anna.matysiak@oeaw.ac.at
tomas.sobotka@oeaw.ac.at
Research presented here was funded by the European
Research Council under the European Union’s Seventh
Framework Programme (FP7/2007-2013) / ERC Grant
agreement n° 284238 (EURREP).
EURREP website: www.eurrep.org
Download