We explore the reasons behind the fall of female employment rates in transition economies and compare them to the evolution in advanced economies. Using a large set of micro level databases, we find that the mechanisms that lead to an increasing female presence in the labor market (higher education and postponing marriage) do not seem to play a role in transition economies.
1. How (Not) to Make Women Work?
How (Not) to Make Women Work?
Evidence from Transition Countries
Karolina Goraus Joanna Tyrowicz Lucas van der Velde
Faculty of Economic Sciences
University of Warsaw
25th IAFFE Annual Conference
Galway, 25 June 2016
2. How (Not) to Make Women Work?
Outline
1 Motivation
2 Data
3 Results
4 Conclusions
3. How (Not) to Make Women Work?
Motivation
Motivation
Literature emhasized substatial drop of women’s employment rates
in the process of transition (Brainerd 2000, Hunt 2002, Blau and
Kahn 2003)
women men
4. How (Not) to Make Women Work?
Motivation
Ratio of employment rates (women to men) increasing
much less in transition countries
Time effects estimates in regressions with country fixed effects
5. How (Not) to Make Women Work?
Motivation
Questions
What factors stand behind those changes in women’s
employment rates? What is the role of unemployment?
How the employment rates evolved for different cohorts?
What was the evolution of (adjusted) gender gaps in
employment rates? Did it differ between cohorts?
What was the role of the opportunity cost of working
(increasing tertiary schooling attendance vs. decreasing access
to child care facilities)?
6. How (Not) to Make Women Work?
Data
Varius sources of micro-level data
National censuses (acquired from Integrated Public Use
Microdata Series International)
International Social Survey Program
Living Standard Measurement Surveys of The World Bank
National Labor Force Surveys
European Union Labor Force Survey
European Community Household Panel
Life in Transition Survey
7. How (Not) to Make Women Work?
Data
Data on transition countries
Country LFS EU LFS Census LSMS ISSP LiTS
Albania 2002-2005 1989-2006
Armenia 2001 1989-2006
Azerbaijan 1995 1989-2006
Belarus 2008-2010 1999 1989-2006
Bosnia & Herz. 2001-2004 1989-2006
Bulgaria 1995-2012 2000-2012 1995-97, 2001-03 1993-1995 1989-2006
Croatia 1996-2012 1989-2006
Czech Republic 1998-2012 1993-1995 1989-2006
Estonia 1995-2012 1997-2012 1992-1995 1989-2006
FYR Macedonia 1989-2006
Georgia 1989-2006
Hungary 1997-2012 1990, 2001 1989-1995 1989-2006
Kazakhstan 1989-2006
Kyrgyzstan 1993, 1996-1998 1989-2006
Latvia 1998-2012 1995 1989-2006
Lithuania 1998-2012 1995 1989-2006
Moldova 1989-2006
Montenegro 1989-2006
Poland 1995-2012 1997-2012 1991-1995 1989-2006
Romania 1995-2012 1997-2012 1977, 1992, 2002 1989-2006
Russia 1991-1995 1989-2006
Serbia 2002-2004, 2007 1989-2006
Slovakia 1998-2012 1995 1989-2006
Slovenia 1996-2012 2002 1991-1995 1989-2006
Tajikistan 1999, 2003, 2009 1989-2006
Ukraine 1989-2006
Uzbekistan 1989-2006
8. How (Not) to Make Women Work?
Data
Data on benchmark countries
Country EU LFS ECHP ISSP
Austria 1995-2012 1995-2001 1989-1995
Belgium 1992-2012 1994-2001
Denmark 1992-2012 1994-2001
Finland 1995-2012 1996-2001
France 1993-2012 1994-2001
Germany 2002-2012 1994-2001 1989-1995
Greece 1992-2012 1994-2001
Ireland 1999-2012 1994-2001 1989-1995
Italy 1992-2012 1994-2001 1989-1995
Netherlands 1996-2012 1994-2001
Norway 1996-2012 1989-1995
Portugal 1992-2012 1994-2001
Spain 1992-2012 1994-2001 1993-1995
Sweden 1995-2012 1997-2001 1994-1995
Switzerland 1996-2012
UK 1992-2012 1994-2001
9. How (Not) to Make Women Work?
Data
Employment rate of women - time trend in OECD data
Time trends All countries Transition countries
Time 0.40*** 0.55*** 0.13*** -0.88***
(0.02) (0.08) (0.03) (0.13)
Time squared -0.01* 0.03***
(0.00) (0.00)
Constant 51.93*** 51.18*** 52.62*** 58.93***
(0.31) (0.50) (0.55) (0.91)
No of observations 629 629 211 211
R2 0.41 0.41 0.07 0.30
Number of countries 29 29 12 12
Employment rate of women - time trend in our data (replication)
Time trends All countries Transition countries
Time 0.38*** 0.45*** 0.11** 0.15
(0.03) (0.12) (0.05) (0.14)
Time squared -0.00 -0.00
(0.00) (0.00)
Constant 54.02*** 53.66*** 53.74*** 53.54***
(3.11) (3.16) (2.85) (2.92)
No of observations 901 901 422 422
R2 0.87 0.87 0.84 0.84
Number of countries 46 46 27 27
10. How (Not) to Make Women Work?
Results
Questions
What factors stand behind those changes in women’s
employment rates? What is the role of unemployment?
How the employment rates evolved for different cohorts?
11. How (Not) to Make Women Work?
Results
Employment rate of women and overall unemployment rate
Employment rate of women (standardized) ILO OECD EUROSTAT
Unemployment rate (standardized) -0.5760*** -0.5684*** -0.4774***
(0.0550) (0.0304) (0.0428)
Transition country dummy 0.3316** 0.0689 -0.1694**
(0.1513) (0.0746) (0.0715)
Transition x unemployment rate 0.3798** 0.2293*** 0.2139***
(0.1883) (0.0569) (0.0686)
Constant -0.1591*** -0.0365 0.0584
(0.0466) (0.0257) (0.0432)
No of observations 515 1,338 632
R2 0.266 0.310 0.250
12. How (Not) to Make Women Work?
Results
Women’s employment rates by age
Advanced economies Transition countries
13. How (Not) to Make Women Work?
Results
Women’s employment rates by age
Transition countries - NMS Transition countries - other
14. How (Not) to Make Women Work?
Results
Decomposition of changes in female employment rate
15. How (Not) to Make Women Work?
Results
Questions
What was the evolution of (adjusted) gender gaps in
employment rates? Did it differ between cohorts?
What was the role of the opportunity cost of working
(increasing tertiary schooling attendance vs. decreasing access
to child care facilities)?
16. How (Not) to Make Women Work?
Results
How to measure discrimination?
17. How (Not) to Make Women Work?
Results
Research method
Oaxaca-Blinder decomposition
¯yM
− ¯yF
= ˆβM
(¯xM
− ¯xF
) + (βM
− βF
)¯xF
Decomposition of Npo
δ = δM + δX + δA + δF
δM - can be explained by differences between matched and
unmatched males
δX - can be explained by differences in the distribution of
characteristics of males and females over the common support
δA - unexplained part of the gap
δF - can be explained by differences between matched and
unmatched females
18. How (Not) to Make Women Work?
Results
Empirical analysis
Two stages
1 Obtaining comparable measures of gender discrimination in
employment rates (∆A) - Npo (2008) decompositions.
one per country-year
separately for cohorts working under transition and those that
entered after transition (two per country-year)
2 Using gender gap estimates as explained variables, whereas
country-year characteristics as explanatory variables. Identify
the correlates (better yet: determinants) of the stark
differentials in measured ∆A.
19. How (Not) to Make Women Work?
Results
Adjusted gender employment gap - time patterns
Calendar years Years from transition
(1) (2) (3) (4)
Transition country -0.6922*** -0.2105***
(0.0806) (0.0599)
Time -0.0366*** -0.0244*** 0.0152*** -0.0273***
(0.0110) (0.0052) (0.0030) (0.0035)
x transition country 0.0586*** 0.0461*** 0.0009 0.0418***
(0.0122) (0.0061) (0.0049) (0.0042)
Time2 0.0006* 0.0003 -0.0002*** 0.0001***
(0.0004) (0.0002) (0.0000) (0.0000)
x transition country -0.0013*** -0.0009*** -0.0002 -0.0005***
(0.0004) (0.0002) (0.0002) (0.0001)
Constant 1.0916*** 0.5734*** 0.6680*** 0.9435***
(0.1121) (0.0406) (0.0989) (0.0544)
Country F.E. No Yes No Yes
Observations 1,184 1,184 1,184 1,184
R-squared 0.287 0.754 0.268 0.754
20. How (Not) to Make Women Work?
Results
Time trend shapes
21. How (Not) to Make Women Work?
Results
Adjusted gender employment gap - institutional factors
(1) (2) (3) (4)
ln GDP per capita -0.39***
(0.0428)
x transition 0.44***
(0.032)
Persons with tertiary -1.28***
in % of population (0.13)
x transition 1.05***
(0.19)
Women with tertiary -1.92***
in % tertiary (0.22)
x transition 1.71***
(0.24)
Constant 0.31*** 0.73*** 0.64*** 0.97***
(0.0489) (0.05) (0.05) (0.09)
Observations 1184 1087 1184 1184
R-squared 0.71 0.78 0.73 0.73
22. How (Not) to Make Women Work?
Results
Adjusted gender employment gap - institutional factors
(5) (6) (7) (8)
% of households with 0.32**
small children (0.13)
x transition -0.33
(0.22)
Access to early
childhood facilities
x transition -0.02***
(0.004)
% of children in
kindergardens
x transition -0.003**
(0.001)
Employment rate -2.08***
of women (0.11)
x transition 0.88***
(0.16)
Constant 0.41*** 0.72*** 0.65*** 1.21***
(0.06) (0.10) (0.10) (0.05)
Observations 870 310 327 1184
R-squared 0.69 0.64 0.64 0.80
23. How (Not) to Make Women Work?
Results
Adjusted GEG - cohort effects in transition countries
(1) (2) (3) (4) (5)
Cohorts working before transition 0.064*** 0.16*** 0.32*** 0.19*** 0.29***
(0.01) (0.03) (0.07) (0.03) (0.05)
Persons with tertiary in % of population 0.09
(0.16)
x cohorts working before transition -0.56***
(0.19)
Women with tertiary education in % tertiary -0.29***
(0.08)
x cohorts working before transition -0.52***
(0.12)
% of households with small children 0.66***
(0.15)
x cohorts working before transition -0.28*
(0.16)
Employment rate of women -0.84***
(0.10)
x cohorts working before transition -0.50***
(0.11)
Constant 0.24** 0.45*** 0.60*** 0.22** 0.81***
(0.11) (0.07) (0.08) (0.09) (0.08)
Observations 1352 1352 1352 1233 1352
R-squared 0.35 0.36 0.38 0.36 0.45
24. How (Not) to Make Women Work?
Conclusions
Conclusions
Employment rates evolution in transition countries - low
explanatory power of unemployment rates, importance of
”entries” and ”exits”
Adjusted gaps initially smaller in transition countries, but then
stable
Relation between gender gaps in employment and institutional
factors less clear for transition countries
Younger cohorts face lower adjusted gaps, institutional factors
play bigger role for older cohorts
25. How (Not) to Make Women Work?
Conclusions
Thank you for your attention