division of domestic labour and women s human capital esrc gender equality network project 4:...
TRANSCRIPT
Division of Domestic Labour and Women’s Human Capital
ESRC Gender Equality Network Project 4: Gender, Time Allocation and the “Wage Gap”
Jonathan Gershuny & Man Yee Kan
Centre for Time Use ResearchDepartment of Sociology
University of Oxford
Background – The Gender Wage Gap
• Introduction of the Equal Pay Act in 1970
• Gender wage gap has fallen but remained wide (EOC, 2001)
• In 1999, 18% pay gap in the UK between men and women working FT (c.f. 36% in 1973)
• 40% gap between female part-timers and male full-timers
Background – The Gender Wage Gap
Why gender pay gap?• Structure of labour market - e.g.
occupational segregation & part-time jobs • 62% of women working full-time and 90% of women working part-time are
employed in jobs mainly done by women (e.g. Paci et al. 1995).
• “Human capital” explanation – gender difference in educational qualifications, work skills and training (e.g. Mincer & Polachek, 1974)
Our project investigates the relationship between the division of domestic labour and human capital accumulation
Key hypothesis of the project
• A gendered division of domestic labour leads to a gender gap in wages
• Initial difference in human capital between partners gendered division of labour differentials in subsequent human capital formation
Project aims and focus of this presentation
For testing the main hypothesis of the project: • Creating a measure of human capital• Calibrating time use estimates for BHPSAims of this paper:• Changes in women’s and men’s time use
practices over the life course, and esp. since birth of first child
• Investigating the impacts of domestic div. of labour and on human capital accumulation
Defining human capital• “Human capital” refers to economically salient
embodied resources e.g. skills, educational attainment, and specific knowledge of the sort that might be considered by prospective employers as justifying offers of employment (Becker, 1993; Coleman, 1988)
• Sociological usage: “estimated value of economically salient work skills” – a key element of Weberian notion of “class situations”
• We use a Heckman regression model to estimate “shadow wage” for BHPS
Measuring human capital – Essex Score
Kan & Gershuny, ISER WP 2006 - 03
Data from British Household Panel Survey – 5,500 households, 10,300 individuals, in wave one– 1991 - present
Variables in the model:• Age, age squared • Dummies for education attainment• MOW scores (mean occupational wage, 2 digit categories,
standardised to 0-100)• Work, unemployment, family care status over past 48
months• Dummies for top MOW decile and deciles 7 to 9 • Product of MOW dummies and age, age squared• Sex in selection equation
Table 3. Essex Score by Gender and Employment Status in 1991
Men Women Employment status Mean SD N Mean SD N
Full-time employed 7.50 3.03 2872 6.24 2.62 1709 Part-time employed 5.91 3.97 220 5.15 2.02 1079 Unemployed 3.91 1.94 371 3.56 1.54 160 Non-employed 3.02 1.43 1358 2.83 1.19 2479 Total 5.89 3.35 4821 4.39 2.45 5427
Mean SD N
All (men+women) 5.09 3.01 10248 Note. The sample contains respondents aged over 15 in Wave 1 (1991) of BHPS. All values are unweighted.
Calibrating time use estimates I
• Kan & Gershuny,ISER WP 2006-19• British Household Panel Survey (BHPS) (1991 – 2005),
containing a rich set of demographic information, employment characteristics and history and so on.
• “Stylised questions” about normal weekly hours of paid and domestic work, frequency at various leisure activities collected since the 1994 wave.
• Home On-line Study (HoL) (1999 – 2001) - a smaller scale study, but contains both diary-based and questionnaire-based (stylised) time use data. The survey part asked the same/similar set of stylised questions about time use as the BHPS.
Calibrating time use estimates II
• Pooled sample of diary and survey data from the HoL adult respondents (N = 2,265)
• Regress diary-based time use estimates on stylised time use estimates
• Identify same stylised variables in BHPS• Parameter estimates from HoL used to
calibrate time use estimates for BHPS data
Calibrating time use estimates III
where k = 1 to 5, indicating the following five main activities respectively: (1) labour
market work, study and travel related to work/study; (2) routine housework, such as
cleaning, ironing and washing; (3) other household works, including caring for family
members, DIY, and shopping for household groceries; (4) sleep, personal care, and
rest; (5) consumption and leisure. kiM is the dependent variable indicating the number
of minutes per day spent at activity k, where 14405
1
k
kiM , calculated from the
respondent’s diary.
kiM
),,,,
,,,,,*,
,,*,,,(
iiii
iiiiiii
iiiiiii
leisureeleisuredleisurecleisureb
lesiureawashcleancookshopparentemphrs
househrsemphrsparentageparentagesqagef
Unstandardized Coefficients of OLS Models of Time Use on Categorized Activities – Employed Women only
Variable
Labour market
work, study, travelling
Routine housework
Other unpaid household
works
Sleep, personal care, and
rest Consumption
and leisure Sum of row Intercept 253.86 -30.83 72.60 737.93 406.45 1440 hwork -1.57 1.84 0.92 -0.80 -0.39 0 paidtr 3.83 -0.52 -0.65 -1.23 -1.43 0 married -5.19 -3.10 21.84 9.10 -22.65 0 age 1.29 3.38 0.72 -4.15 -1.24 0 agesq -0.04 -0.02 0.00 0.04 0.02 0 shop 31.51 -2.88 14.58 -6.03 -37.18 0 cook 13.44 16.34 -32.55 -4.95 7.72 0 clean -44.48 14.42 13.56 -6.02 22.52 0 wash -9.67 8.12 3.83 -0.67 -1.61 0 parent -201.84 -21.94 387.81 -34.54 -129.48 0 paidtpar 0.87 -0.13 -1.67 0.65 0.29 0 agepar 3.60 0.90 -6.59 0.23 1.86 0 leisura1 11.81 -1.02 -4.13 2.75 -9.40 0 leisura2 27.52 -10.25 -22.25 9.09 -4.12 0 leisurb1 5.82 -11.08 -0.49 15.75 -10.00 0 leisurb2 -2.33 -1.69 0.33 1.02 2.67 0 leisurc1 27.86 -22.07 33.10 -31.85 -7.04 0 leisurc2 25.16 -10.48 -7.37 -17.64 10.33 0 leisurd1 -47.39 19.22 -22.69 10.10 40.76 0 leisurd2 -40.68 28.81 -18.08 9.68 20.27 0 leisure1 4.79 -1.49 -8.77 -8.50 13.98 0 leisure2 2.32 -4.40 6.29 -15.16 10.94 0 mileact -36.94 9.32 -37.28 79.65 -14.75 0 R2 0.656 0.428 0.329 0.196 0.383
Note: The sample is pooled from all three waves of the Home On-line Study, 1999 – 2001.
Aims of this paper• Examine changes in time use practices of
married men and women over the life course, and their implications for the gender wage gap
H1: Specialization in the domestic division of labour by gender increases over the lifecourse
H2: The gender specialization in the dom. div of labour has negative impacts on women’s human capital accumulation
Figure 1. Time use practices over the lifecourse, women aged 19-40
Paid work
Routine housework
Care and other domestic work
Sleep and rest
Consumption and leisure
0
200
400
600
800
1000
1200
1400
Stay single(n=3518)
Acquire partner(n=489)
Stay partnered, nochild (n=2675)
Stay partnered,acquire child
(n=354)
Stay partnered,keep child(n=6866)
Stay partnered,child leaves/hasgrown up (n=98)
Min
ute
s p
er
da
y
Consumption and leisure
Sleep and rest
Care and other domestic work
Routine housework
Paid work
Figure 2. Time use practices over the lifecourse, men aged 19-40
Paid work
Routine houseworkCare and other domestic work
Sleep and rest
Consumption and leisure
0
200
400
600
800
1000
1200
1400
Stay single(n=5350)
Acquire partner(n=528)
Stay partnered, nochild (n=2457)
Stay partnered,acquire child
(n=358)
Stay partnered,keep child(n=5483)
Stay partnered,child leaves/hasgrown up (n=61)
Min
ute
s p
er
da
y
Consumption and leisure
Sleep and rest
Care and other domestic work
Routine housework
Paid work
Figure 3. Time spent on paid work and unpaid domestic work before and after the birth of first child, women aged 19-40 (n=747)
Paid work
Routine housework
Care and other domestic work
0
100
200
300
400
500
600
Year before 1 2 3 4 5
Year after childbirth
Min
ute
s p
er d
ay
Care and other domestic work
Routine housework
Paid work
Figure 4. Time spent on paid work and unpaid domestic work before and after the birth of first child, men aged 19-40 (n=671)
Paid work
Routine housework
Care and other domestic work
0
100
200
300
400
500
600
Year before 1 2 3 4 5
Year after childbirth
Min
ute
s p
er
day
Care and other domestic work
Routine housework
Paid work
Figure 5. Proportion of unpaid domestic work to all work after the birth of first child
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Year before 1 2 3 4 5
Year after childbirth
Non employed women before childbirth
Spouses of non-employed women before childbirth
Women who stayed in employment after childbirth
Spouses of women who stayed in employment after childbirth
Women who quitted employment after childbirth
Spouses of women who quitted employment after childbirth
(n=132)
(n=132)
(n=460)
(n=460)
(n=110)
(n=110)
Figure 6. Proportion of domestic work done by women after the birth of first child
0.56
0.58
0.6
0.62
0.64
0.66
0.68
0.7
0.72
0.74
0.76
0.78
Year before 1 2 3 4 5
Year after childbirth
Couples where the female partner was non-employed beforechildbirth
Couples where the female partner stayed in employment afterchildbirth
Couples where the female partner quitted employment afterchildbirth
(n=132)
(n=460)
(n=110)
Recap: Hypotheses
H1: Specialization in the domestic division of labour by gender increases over the lifecourse
H2: The gender specialization in the dom. div of labour has negative impacts on women’s human capital accumulation
Figure 7. Potential hourly wage before and after the birth of first child
0
1
2
3
4
5
6
7
8
9
10
Year before 1 2 3 4 5
Year after childbirth
Po
ten
tial
ho
url
y w
age
(in
GB
P)
Non-employed women before childbirth
Spouses of non-employed women before childbirth
Women who stayed in employment after childbirth
Spouses of women who stayed in employment after childbirth
Women who quitted employment after childbirth
Spouses of women who quitted employment after childbirth
(n=132)
(n=132)
(n=460)
(n=460)
(n=110)
(n=110)
Table 1. OLS Models of Potential Wage on Domestic Work Participation after Childbirth
Women Men
B Robust
SE B
Robust
SE B
Robust
SE B
Robust
SE
Share of domestic work -3.641*** 0.552 -1.805** 0.612
Weekly domestic work time -0.005*** 0.001 -0.003** 0.001
Partner’s weekly domestic work
time 0.002** 0.001 0.001* 0.001
Number of children -0.219* 0.093 -0.027 0.086 0.284** 0.097 0.270** 0.100
Years since the birth of first child 0.055* 0.025 0.017 0.024 0.165*** 0.026 0.166*** 0.027
Last year’s potential wage 0.803*** 0.025 0.768*** 0.025 0.872*** 0.021 0.871*** 0.021
Partner’s potential wage 0.068*** 0.019 0.064** 0.018 0.109*** 0.025 0.113*** 0.025
Constant 3.222*** 0.446 2.383*** 0.301 0.778** 0.286 0.161 0.326
R2 0.766 0.776 0.768 0.768
Note: Data from the British Household Panel Survey, 1994 - 2005. N =2,997 couples. In the OLS models, standard errors take account of multiple observations of individuals.
Table 2. OLS and Fixed Effect Models of Potential Wage on Domestic Work Participation, Married and
Cohabiting men and women
Women Men
OLS Model OLS Model FE Model OLS Model OLS Model FE Model
B B B B B B
Share of domestic work -5.030*** -0.185 -5.780*** -0.224
Weekly domestic work time -0.011*** -0.012***
Partner’s weekly domestic work
time 0.002***
0.003***
Number of adults -0.252*** -0.228*** 0.089*** 0.053 0.069 0.268***
Number of children -0.246*** 0.228*** -0.142*** 0.010 0.139* 0.302***
Partner’s potential wage 0.349*** 0.307*** 0.129*** 0.567*** 0.542*** 0.206***
Constant 7.234*** 6.476*** 5.254*** 6.253*** 5.217*** 6.157***
R2/ Between groups R2 0.243 0.298 0.215 0.212 0.218 0.145 Note: Data from the British Household Panel Survey, 1994 - 2005. N = 22,858 for women, and 21,425 for men. The OLS models include dummies for year; standard errors take account of multiple observations of individuals.
Conclusion• Women’s time use practices change to a
greater extent than men’s after lifecycle events• Women become more specialized in domestic
work, and men’s in paid work over time• Men and women have more or less the same
total work (paid+unpaid domestic)But - Women’s potential wage suffers because:• Significant, negative relationship between
proportion of domestic work on human capital
Further work• Further analyses on the causal
relationship between domestic division of labour and potential wage:
• Instrumental variable approach
• Models with lagged dependent variables
• Graphical chain models