Hasan Tekgüç (MAÜ), Değer Eryar (İEÜ) & Dilek Cindoğlu (MAÜ)
Turkish Labor Market Network Meeting
BETAM Bahçeşehir University,
December 2, 2014
2.12.2014
Türkiye İşgücü Piyasası Araştırma Ağı Konferansı, Bahçeşehir Ü.
• Gender wage gap is widely studied, including Turkey.
• However, we believe that the empirical findings may not be reliable, b/c
• The gender discrepancy of labor force participation.
• Exception to discrepancy is women with higher education.
• Disaggregating only by gender leads to misleading findings.
• We further disaggregate by education level when estimating wage gap and taking into account selection bias problem.
2.12.2014
Türkiye İşgücü Piyasası Araştırma Ağı Konferansı, Bahçeşehir Ü.
• Cindoğlu & various co-authors (sociology literature): Tertiary eduction enables access to respectable/clean jobs such that women working in these jobs are conferred higher status.
• Working class women: they work outside of the house because «unfortunately they have to ».
• White collar women: work gives them status and can benefit from increased status within household
• Royalty (1998) and Theoddossiou & Zangelidis (2009) study job mobility by disaggregating both for gender and education level.
• Unobserved characteristics (omitted variable problem)
• Mobility behavior differs by education
2.12.2014
Türkiye İşgücü Piyasası Araştırma Ağı Konferansı, Bahçeşehir Ü.
Education and Female Labor Force Participation
2.12.2014
Türkiye İşgücü Piyasası Araştırma Ağı Konferansı, Bahçeşehir Ü.
Why 20-54? Household Labor Force Survey(HLFS) 2004 & 2011
2.12.2014
Türkiye İşgücü Piyasası Araştırma Ağı Konferansı, Bahçeşehir Ü.
Labor Force participation, HLFS 2011 (20 – 54)
2.12.2014
Türkiye İşgücü Piyasası Araştırma Ağı Konferansı, Bahçeşehir Ü.
Wage according to education level, HLFS 2004 & 2011 (20 – 54 years old)
2.12.2014
Türkiye İşgücü Piyasası Araştırma Ağı Konferansı, Bahçeşehir Ü.
Empirical Literature on Turkey
• Dayıoğlu and Kasnakoğlu (1997)
• female-to-male monthly wage ratio 96 % (1987 survey)
• Only human capital characteristics
• Dayıoğlu and Tunalı (2004)
• female to male wage ratio of 98 % (1988) survey & 85 % ( 1994)
(corrected for selection bias).
• Human capital + some workplace characteristics such as firm size
& industry
• Tansel (2004)
• Female to male wage ratio is 73% in private sector, 78% in SEEs and almost nonexistent in the public sector according to 1994
Household Expenditure Survey (corrected for selection bias)
• Discrimination is mostly observed in private sector
• The key factors are returns to education and experience
2.12.2014
Türkiye İşgücü Piyasası Araştırma Ağı Konferansı, Bahçeşehir Ü.
• İlkkaracan and Selim (2007)
• No correction for selection bias
• Female male wage ratio is around 70 % (1994: employment and Wage
Structure Survey)
• Mostly manufacturing, electricity, gas, and mining
• Detailed workplace characteristics drop the unexplained part of the gap
•
•
All the above results are based on Blinder-Oaxaca composition (mean wage differential)
Aktaş and Uysal (2012)
• Almost no gap at the bottom of the distribution and higher wages for women at the top (2006: employment and Wage Structure Survey)
• Different returns to education is crucial in explaining the gap rather than the workplace characteristics
2.12.2014
Türkiye İşgücü Piyasası Araştırma Ağı Konferansı, Bahçeşehir Ü.
• Step 1 - selection model:
• LFP = marital status + urban + NUTS2 + age + # young child + others ’ income + education (if necessary)
• Step 2 - Wage estimation:
• Ln(wage_hour) = marital status + urban + metro + NUTS1 + age + major + tenure + tenure2 + public + firm-size + occupation + admin + inv. Mills ratio [population weights]
• Step 3 - Wage decomposition:
• 𝑅 =
𝑀
𝑌
𝑊
𝑋
𝑀
𝑋
𝑊 𝛽
𝑀
𝑋 ′
𝑊 𝛽
𝑀 𝛽
𝑊
Endowment
Discrimination
2.12.2014
Türkiye İşgücü Piyasası Araştırma Ağı Konferansı, Bahçeşehir Ü.
Dependent Variable: Monthly Wage
Türkiye İşgücü Piyasası Araştırma Ağı Konferansı, Bahçeşehir Ü.
2.12.2014
2.12.2014
Türkiye İşgücü Piyasası Araştırma Ağı Konferansı, Bahçeşehir Ü.
Selection or not? Disaggregation or not?? HLFS, 2011
2.12.2014
Türkiye İşgücü Piyasası Araştırma Ağı Konferansı, Bahçeşehir Ü.
Does the observed discrimination really exist?
2.12.2014
Türkiye İşgücü Piyasası Araştırma Ağı Konferansı, Bahçeşehir Ü.
2004 & 2011 comparison
2.12.2014
Türkiye İşgücü Piyasası Araştırma Ağı Konferansı, Bahçeşehir Ü.
Private versus public
2.12.2014
Türkiye İşgücü Piyasası Araştırma Ağı Konferansı, Bahçeşehir Ü.
Summary: There is discrimination and it is rising
• Both correcting for selection and disaggregating for education levels lead to more reliable results.
• Both for tertiary (15 %) and less educated (21 %) women endowment explains only a minor part ( 1 %) of observed differences.
• Tertiary education:
•
•
•
Excluding women working part-time and in informal sector reduces wage gap.
Controlling for more detailed major/subject does not reduce wage gap!
Controlling for occupation and admin. tasks reduces wage gap.
• 2011 alt-gruplar:
• Total wage gap is the same both for the public and private sector , however, the role of discrimination is much more important in the public sector
• Increasing wage gap between 2004 and 2011 has been mostly due to discrimination
• Glass Ceiling exists.
2.12.2014
Türkiye İşgücü Piyasası Araştırma Ağı Konferansı, Bahçeşehir Ü.
Glass Ceiling:
2.12.2014
Türkiye İşgücü Piyasası Araştırma Ağı Konferansı, Bahçeşehir Ü.
Other MENA countries:
2.12.2014
Türkiye İşgücü Piyasası Araştırma Ağı Konferansı, Bahçeşehir Ü.
2.12.2014
Türkiye İşgücü Piyasası Araştırma Ağı Konferansı, Bahçeşehir Ü.
2.12.2014
Türkiye İşgücü Piyasası Araştırma Ağı Konferansı, Bahçeşehir Ü.
2.12.2014
Türkiye İşgücü Piyasası Araştırma Ağı Konferansı, Bahçeşehir Ü.
2.12.2014
Türkiye İşgücü Piyasası Araştırma Ağı Konferansı, Bahçeşehir Ü.
2.12.2014
Türkiye İşgücü Piyasası Araştırma Ağı Konferansı, Bahçeşehir Ü.
2.12.2014
Türkiye İşgücü Piyasası Araştırma Ağı Konferansı, Bahçeşehir Ü.