identifying age penalty in women's wages: new method and evidence from germany
TRANSCRIPT
Identifying Age Penalty in Women’s Wages:
Identifying Age Penalty in Women’s Wages:New method and evidence from Germany
J. Tyrowicz L. van der Velde I. van Staveren
IAFFE @ ASSA 2017
Identifying Age Penalty in Women’s Wages:
Introduction
Why it matters?
Definitely: women have gradually better educational attainment
Arguably: sorting matters less (for many occupations)
⇒ raw aggregate gender wage gap should decline
which it does .... but really slowly ...
Aging process in Europe?
Is there an age pattern?Implications for efficient policies to address gender wage gap?
Identifying Age Penalty in Women’s Wages:
Introduction
Why it matters?
Definitely: women have gradually better educational attainment
Arguably: sorting matters less (for many occupations)
⇒ raw aggregate gender wage gap should decline
which it does ....
but really slowly ...
Aging process in Europe?
Is there an age pattern?Implications for efficient policies to address gender wage gap?
Identifying Age Penalty in Women’s Wages:
Introduction
Why it matters?
Definitely: women have gradually better educational attainment
Arguably: sorting matters less (for many occupations)
⇒ raw aggregate gender wage gap should decline
which it does .... but really slowly ...
Aging process in Europe?
Is there an age pattern?Implications for efficient policies to address gender wage gap?
Identifying Age Penalty in Women’s Wages:
Introduction
Why it matters?
Definitely: women have gradually better educational attainment
Arguably: sorting matters less (for many occupations)
⇒ raw aggregate gender wage gap should decline
which it does .... but really slowly ...
Aging process in Europe?
Is there an age pattern?Implications for efficient policies to address gender wage gap?
Identifying Age Penalty in Women’s Wages:
Introduction
Motivation
Adjusted gender wage gap for selected cohorts as they aged
.1.1
5.2
.25
.3.3
5A
djus
ted
gap
25 30 35 40 45 50 55 60Age
1940−1944 1950−1954 1960−1964
Controls: tenure, experience, small kids in the household, married, education level and year.
Identifying Age Penalty in Women’s Wages:
Introduction
Theory on age pattern in gender wage gap
Unequal distribution of activities within the household (Becker 1985)
Child bearing and child rearing and its expectation (Mincer andPolachek 1974, Goldin and Katz 2008, Goldin 2014)Gender bias in the measurement of human capital
Statistical discrimination from the employers (Dahlby 1983)
“Hysteresis effect” (Babcock et al. 2002, Blau and Ferber 2011)
“Double standard of aging” (Duncan and Loretto 2004, Neumarket al. 2015)
Identifying Age Penalty in Women’s Wages:
Introduction
Theory on age pattern in gender wage gap
Unequal distribution of activities within the household (Becker 1985)
Child bearing and child rearing and its expectation (Mincer andPolachek 1974, Goldin and Katz 2008, Goldin 2014)Gender bias in the measurement of human capital
Statistical discrimination from the employers (Dahlby 1983)
“Hysteresis effect” (Babcock et al. 2002, Blau and Ferber 2011)
“Double standard of aging” (Duncan and Loretto 2004, Neumarket al. 2015)
Identifying Age Penalty in Women’s Wages:
Introduction
Theory on age pattern in gender wage gap
Unequal distribution of activities within the household (Becker 1985)
Child bearing and child rearing and its expectation (Mincer andPolachek 1974, Goldin and Katz 2008, Goldin 2014)Gender bias in the measurement of human capital
Statistical discrimination from the employers (Dahlby 1983)
“Hysteresis effect” (Babcock et al. 2002, Blau and Ferber 2011)
“Double standard of aging” (Duncan and Loretto 2004, Neumarket al. 2015)
Identifying Age Penalty in Women’s Wages:
Introduction
Theory on age pattern in gender wage gap
Unequal distribution of activities within the household (Becker 1985)
Child bearing and child rearing and its expectation (Mincer andPolachek 1974, Goldin and Katz 2008, Goldin 2014)Gender bias in the measurement of human capital
Statistical discrimination from the employers (Dahlby 1983)
“Hysteresis effect” (Babcock et al. 2002, Blau and Ferber 2011)
“Double standard of aging” (Duncan and Loretto 2004, Neumarket al. 2015)
Identifying Age Penalty in Women’s Wages:
Introduction
Intended contribution
Explore the effects of the life-cycle in women’s earnings penalty
Extend the method proposed by DiNardo, Fortin and Lemieux(1996) to separate cohort, time and age effects.
Identifying Age Penalty in Women’s Wages:
Introduction
Intended contribution
Explore the effects of the life-cycle in women’s earnings penalty
Extend the method proposed by DiNardo, Fortin and Lemieux(1996) to separate cohort, time and age effects.
Identifying Age Penalty in Women’s Wages:
Method
DiNardo, Fortin and Lemieux decomposition (1996)
Given a joint distribution of wages and characteristics of the form
f (wi ) =
∫fi (w |x) f (x |g = i)dx (1)
(where i represents the gender: men or women)
then a counterfactual wage structure using a reweighting parameter Ψ(x)may be represented as
f (w cf ) =
∫ff (w |x) Ψj(x)fj(x |g = f )dx . (2)
Conveniently, Ψ(x) can be recovered using probit models.
Identifying Age Penalty in Women’s Wages:
Method
DiNardo, Fortin and Lemieux decomposition (1996)
Given a joint distribution of wages and characteristics of the form
f (wi ) =
∫fi (w |x) f (x |g = i)dx (1)
(where i represents the gender: men or women)
then a counterfactual wage structure using a reweighting parameter Ψ(x)may be represented as
f (w cf ) =
∫ff (w |x) Ψj(x)fj(x |g = f )dx . (2)
Conveniently, Ψ(x) can be recovered using probit models.
Identifying Age Penalty in Women’s Wages:
Method
Methodology
By setting alternative Ψ(x), we define counterfactual distributions, e.g.
traditional: male ˆdistribution with female characteristics
our approach:male ˆdistribution if female characteristics were constant as we age+female ˆdistribution if female characteristics were constant over time
if sample of men and women is constant ⇒ also unobservablecharacteristics
⇒ how gender wage gaps change, as men and women age
Identifying Age Penalty in Women’s Wages:
Method
Methodology
By setting alternative Ψ(x), we define counterfactual distributions, e.g.
traditional: male ˆdistribution with female characteristics
our approach:male ˆdistribution if female characteristics were constant as we age
+female ˆdistribution if female characteristics were constant over time
if sample of men and women is constant ⇒ also unobservablecharacteristics
⇒ how gender wage gaps change, as men and women age
Identifying Age Penalty in Women’s Wages:
Method
Methodology
By setting alternative Ψ(x), we define counterfactual distributions, e.g.
traditional: male ˆdistribution with female characteristics
our approach:male ˆdistribution if female characteristics were constant as we age+female ˆdistribution if female characteristics were constant over time
if sample of men and women is constant ⇒ also unobservablecharacteristics
⇒ how gender wage gaps change, as men and women age
Identifying Age Penalty in Women’s Wages:
Method
Methodology
By setting alternative Ψ(x), we define counterfactual distributions, e.g.
traditional: male ˆdistribution with female characteristics
our approach:male ˆdistribution if female characteristics were constant as we age+female ˆdistribution if female characteristics were constant over time
if sample of men and women is constant ⇒ also unobservablecharacteristics
⇒ how gender wage gaps change, as men and women age
Identifying Age Penalty in Women’s Wages:
Method
Methodology
By setting alternative Ψ(x), we define counterfactual distributions, e.g.
traditional: male ˆdistribution with female characteristics
our approach:male ˆdistribution if female characteristics were constant as we age+female ˆdistribution if female characteristics were constant over time
if sample of men and women is constant ⇒ also unobservablecharacteristics
⇒ how gender wage gaps change, as men and women age
Identifying Age Penalty in Women’s Wages:
Method
Method
The raw gender wage gap in any age (∆j) is the sum of explained andunexplained component:
∆j = f (w |m, j)− f ′(w |f , j)︸ ︷︷ ︸Explained component
+ f ′(w |f , j)− f (w |f , j)︸ ︷︷ ︸Unexplained component
Hence, ∆j −∆i =∫fm,j(w |x) ((f (x |m, i)− f (x |m, j)
−(f (x |f , j))− f (x |f , i)))dx︸ ︷︷ ︸Change in explained component
+
∫(fm,i (w |x)− fm,j(w |x)
−(ff ,i (w |x)− ff ,j(w |x))) (f (x |f , i)︸ ︷︷ ︸Change in unexplained component
+ Change in residuals
Identifying Age Penalty in Women’s Wages:
Method
Method
The raw gender wage gap in any age (∆j) is the sum of explained andunexplained component:
∆j = f (w |m, j)− f ′(w |f , j)︸ ︷︷ ︸Explained component
+ f ′(w |f , j)− f (w |f , j)︸ ︷︷ ︸Unexplained component
Hence, ∆j −∆i =∫fm,j(w |x) ((f (x |m, i)− f (x |m, j)
−(f (x |f , j))− f (x |f , i)))dx︸ ︷︷ ︸Change in explained component
+
∫(fm,i (w |x)− fm,j(w |x)
−(ff ,i (w |x)− ff ,j(w |x))) (f (x |f , i)︸ ︷︷ ︸Change in unexplained component
+ Change in residuals
Identifying Age Penalty in Women’s Wages:
Data
Data
(West) German nationals aged 25-59 – SOEP
Period: 1984-2008.
SOEP has great retention rates
Over 7 000 individuals are observed for a decade or longer.25% of the original sample observed on every year.Almost 70 000+ complete observations (exclusion gender symmetric)
Dependent variable: real hourly wages
Rich set of covariates: education, tenure, experience full and parttime, household characteristics, occupations, industries, type ofemployment...
Identifying Age Penalty in Women’s Wages:
Data
Data
(West) German nationals aged 25-59 – SOEP
Period: 1984-2008.
SOEP has great retention rates
Over 7 000 individuals are observed for a decade or longer.25% of the original sample observed on every year.Almost 70 000+ complete observations (exclusion gender symmetric)
Dependent variable: real hourly wages
Rich set of covariates: education, tenure, experience full and parttime, household characteristics, occupations, industries, type ofemployment...
Identifying Age Penalty in Women’s Wages:
Data
Data
(West) German nationals aged 25-59 – SOEP
Period: 1984-2008.
SOEP has great retention rates
Over 7 000 individuals are observed for a decade or longer.25% of the original sample observed on every year.Almost 70 000+ complete observations (exclusion gender symmetric)
Dependent variable: real hourly wages
Rich set of covariates: education, tenure, experience full and parttime, household characteristics, occupations, industries, type ofemployment...
Identifying Age Penalty in Women’s Wages:
Data
A quick look at the sample
0
.2
.4
.6
.8
Pro
port
ion
Married Small kids Higher education Employment
1984 1990 1996 2002 2008 Men
Aged: 25−34
Identifying Age Penalty in Women’s Wages:
Data
A quick look at the sample
0
.2
.4
.6
.8
Pro
port
ion
Married Small kids Higher education Employment
1984 1990 1996 2002 2008 Men
Aged:35−44
Identifying Age Penalty in Women’s Wages:
Data
A quick look at the sample
0
.2
.4
.6
.8
Pro
port
ion
Married Small kids Higher education Employment
1984 1990 1996 2002 2008 Men
Aged:45−59
Identifying Age Penalty in Women’s Wages:
Results
Adjusted gender wage gap across age and cohorts
Bar: a period in the sample, colors preserve bar colors. Line: women’s participationrate at the right axis.
Identifying Age Penalty in Women’s Wages:
Results
Double decomposition: changes in the adjusted gap
Initial year Avg. changeInitial Age 1984 1989 1994 1999 2004 with age
25-29 0.04 0.07 0.09 0.01 0.05 0.0530-34 0.10 0.03 0.03 0.03 -0.02 0.0335-39 -0.04 0.15 0.00 -0.04 -0.02 0.0140-44 0.17 -0.02 0.00 0.01 -0.01 0.0345-49 -0.11 0.01 0.06 0.08 0.05 0.0250-54 -0.03 0.03 -0.14 -0.05 -0.01 -0.04
What to do about non-working years?
Include working for a wage in Ψ(x)
Identifying Age Penalty in Women’s Wages:
Results
Double decomposition: changes in the adjusted gap
Initial year Avg. changeInitial Age 1984 1989 1994 1999 2004 with age
25-29 0.04 0.07 0.09 0.01 0.05 0.0530-34 0.10 0.03 0.03 0.03 -0.02 0.0335-39 -0.04 0.15 0.00 -0.04 -0.02 0.0140-44 0.17 -0.02 0.00 0.01 -0.01 0.0345-49 -0.11 0.01 0.06 0.08 0.05 0.0250-54 -0.03 0.03 -0.14 -0.05 -0.01 -0.04
What to do about non-working years?
Include working for a wage in Ψ(x)
Identifying Age Penalty in Women’s Wages:
Results
Double decomposition: changes in the adjusted gap
Initial year Avg. changeInitial Age 1984 1989 1994 1999 2004 with age
25-29 0.04 0.07 0.09 0.01 0.05 0.0530-34 0.10 0.03 0.03 0.03 -0.02 0.0335-39 -0.04 0.15 0.00 -0.04 -0.02 0.0140-44 0.17 -0.02 0.00 0.01 -0.01 0.0345-49 -0.11 0.01 0.06 0.08 0.05 0.0250-54 -0.03 0.03 -0.14 -0.05 -0.01 -0.04
What to do about non-working years?
Include working for a wage in Ψ(x)
Identifying Age Penalty in Women’s Wages:
Results
Double decomposition: changes in the adjusted gap
Initial year Avg. changeInitial Age 1984 1989 1994 1999 2004 with age
25-29 0.04 0.07 0.10 0.04 0.07 0.0630-34 0.04 0.02 0.07 0.04 0.01 0.0435-39 -0.02 0.15 0.00 -0.03 0.00 0.0240-44 0.17 0.02 -0.02 0.09 0.04 0.0645-49 -0.13 0.03 0.18 0.11 0.07 0.0550-54 -0.04 0.05 -0.16 -0.06 -0.03 -0.05
Identifying Age Penalty in Women’s Wages:
Results
Double decomposition: changes in the adjusted gap
Initial year Avg. change No EInitial Age 1984 1989 1994 1999 2004 with age controls
25-29 0.04 0.07 0.10 0.04 0.07 0.06 0.0530-34 0.04 0.02 0.07 0.04 0.01 0.04 0.0335-39 -0.02 0.15 0.00 -0.03 0.00 0.02 0.0140-44 0.17 0.02 -0.02 0.09 0.04 0.06 0.0345-49 -0.13 0.03 0.18 0.11 0.07 0.05 0.0250-54 -0.04 0.05 -0.16 -0.06 -0.03 -0.05 -0.04
Identifying Age Penalty in Women’s Wages:
Conclusions
Take home message
Adjusted gender wage gap ...
grows with age
non-monotonically
also in post-reproductive age
Interpretation
Consistent with human capital ... to some extent
Question: is there a case for human capital story in thepost-reproductive age?
Identifying Age Penalty in Women’s Wages:
Conclusions
Take home message
Adjusted gender wage gap ...
grows with age
non-monotonically
also in post-reproductive age
Interpretation
Consistent with human capital ... to some extent
Question: is there a case for human capital story in thepost-reproductive age?
Identifying Age Penalty in Women’s Wages:
Conclusions
Summary
1 A new method for identifying age effects in adjusted GWG
2 New evidence for Germany, a country with relatively high inequality,stable over time
Policy implication 1: if Germany is typical, aggregate GWG willincrease as societies age (composition effects)Policy implication 2: overlapping penalties?
Where to now?
International context: UK, US, Canada, Russia, Korea
Hours flexibility story (Goldin 2014)
Identifying Age Penalty in Women’s Wages:
Conclusions
Summary
1 A new method for identifying age effects in adjusted GWG
2 New evidence for Germany, a country with relatively high inequality,stable over time
Policy implication 1: if Germany is typical, aggregate GWG willincrease as societies age (composition effects)
Policy implication 2: overlapping penalties?
Where to now?
International context: UK, US, Canada, Russia, Korea
Hours flexibility story (Goldin 2014)
Identifying Age Penalty in Women’s Wages:
Conclusions
Summary
1 A new method for identifying age effects in adjusted GWG
2 New evidence for Germany, a country with relatively high inequality,stable over time
Policy implication 1: if Germany is typical, aggregate GWG willincrease as societies age (composition effects)Policy implication 2: overlapping penalties?
Where to now?
International context: UK, US, Canada, Russia, Korea
Hours flexibility story (Goldin 2014)
Identifying Age Penalty in Women’s Wages:
Conclusions
Summary
1 A new method for identifying age effects in adjusted GWG
2 New evidence for Germany, a country with relatively high inequality,stable over time
Policy implication 1: if Germany is typical, aggregate GWG willincrease as societies age (composition effects)Policy implication 2: overlapping penalties?
Where to now?
International context: UK, US, Canada, Russia, Korea
Hours flexibility story (Goldin 2014)
Identifying Age Penalty in Women’s Wages:
Conclusions
Questions or suggestions?
Thank you for your attention
Identifying Age Penalty in Women’s Wages:
Conclusions
Babcock, L., Gelfand, M., Small, D. and Stayn, H.: 2002, Propensity to initiatenegotiations: A new look at gender variation in negotiation behavior, IACM 15thAnnual Conference.
Becker, G. S.: 1985, Human capital, effort, and the sexual division of labor, Journal ofLabor Economics 3(1), pp. S33–S58.
Blau, F. D. and Ferber, M. A.: 2011, Career plans and expectations of young womenand men: The earnings gap and labor force participation, Journal of HumanResources 26(4), 581–607.
Dahlby, B.: 1983, Adverse selection and statistical discrimination: An analysis ofcanadian automobile insurance, Journal of Public Economics 20(1), 121–130.
Duncan, C. and Loretto, W.: 2004, Never the right age? gender and age-baseddiscrimination in employment, Gender, Work & Organization 11(1), 95–115.
Goldin, C.: 2014, A grand gender convergence: Its last chapter, The AmericanEconomic Review 104(4), 1091–1119.
Goldin, C. and Katz, L. F.: 2008, Transitions: Career and family life cycles of theeducational elite, The American Economic Review 98(2), 363–369.
Mincer, J. and Polachek, S.: 1974, Family investments in human capital: Earnings ofwomen, Journal of Political Economy 82(2), pp. S76–S108.
Neumark, D., Burn, I. and Button, P.: 2015, Is it harder for older workers to find jobs?new and improved evidence from a field experiment, National Bureau of EconomicResearch, Working Paper No. 21669 .