1. Parental Son Preference, Gender Role Attitudes, and Sharing of Housework in Korea Jisoo Hwang (HUFS) Chulhee Lee (SNU) Esther Lee (SNU)

2. Background Questions  How do cultural norms on gender roles affect within- family allocation of housework in Korea?  How do parental gender role attitudes (parental son preference) influence offspring gender norms?  Why is the gender disparity in within-family allocation of housework persistently high in Korea?

3. Relative Female Progress in Employment

4. Relative Female Progress in Wage

5. Percentages of Females among the Individuals Who Pass Major Government-Administered Exams for Prestigious Public- Service Positions Sources: Ministry of Public Administration and Security, Republic of Korea, Statistical Yearbook of Ministry of Public Administration and Security (MOPAS), 2001-2013; Ministry of Justice, Republic of Korea Press Release; Korean Women’s Development Institute; National Gender Statistics DB: Political and Social Participation; The website of Gender Statistics Information System (https://gsis.kwdi.re.kr).https://gsis.kwdi.re.kr

6. Source: Korean Time Use Survey 1999, 2004, 2009, 2014

7. Decline of Marriage

8. Quantity of Housework Done by Men and Fertility Feyrer, Sacerdote, and Stern, “Will the Stork Return to Europe and Japan?”

9. Economics Literature on Intra-Family Resource Allocation  Estimation of “Pareto weights” or “sharing rule” (Couprie 2007)  Identifying determinants of intra-family resource allocations  Relative earning power (Clark et al. 2004; Bonke 2009, 2015; Thibout 2015)  Remarriage market condition (Chiappori et al. 2002)  Division of household wealth after divorce (Chiappori et al. 2002)  Social or cultural factors (Zhang and Chan 1999; Clark et al. 2004; Hendy and Sofer 2010)

10. Effects of Cultural Variables  Possible pathways  Threat point or the sharing rule.  Preferences of family members  Constrains on the bargaining over within-family allocations  Measures employed in previous studies  Partner’s parental education and employment (Clark et al. 2004; Couprie 2007; Hwang 2015)  Money management practice (Clark et al. 2004)  Questions on gender role attitudes (Eun 2009)  Sharing of marriage costs (Hendy and Sofer 2010; Zhang and Chan 1999)  Problems  Results are mixed  Some measures could be endogenously determined by previous patterns of resource allocations (e.g. gender-role attitudes, money management)  Cultural norms? Other socioeconomic differences? (e.g. parental education and employment, cross-country comparisons)

11. Parental Influences on Gender Norms  Growing evidence regarding parental or ancestral influences on offspring cultural norms  Intergenerational transmissions of cultural beliefs about the appropriate roles of women in society (Fernandez, Fogli, and Olivetti 2004; Fernandez and Fogli 2006; Hwang 2015).  Measures:  Parental (especially mother’s) education and employment  Socioeconomic characteristics of the country of origin

12. Parental Son Preference and Sex Ratios at Birth  Long tradition of parental son preference in Korea  Sources of parental son preference  Cultural or religious values  Economic factors: likelihood of co-residence, lower labor-market status of females  Parental son preference is revealed in the rise in the sex ratios at birth after 1980.  Regional differences in the strength of son preference  The pattern of provincial variations in the sex ratio at birth persisted over time.  This might capture variations in parental son preference and different cultural backgrounds

13. Sex Ratio at Birth in South Korea, 1980-2013 Source: Vital Statistics of Korea: Births and Deaths, each year; Drawn from the website of Korean Statistical Information Service (http://www.kosis.kr).http://www.kosis.kr

14. Sex Ratios at Birth in Four Selected Regions Source: Vital Statistics of Korea: Births and Deaths, each year; Drawn from the website of Korean Statistical Information Service (http://www.kosis.kr).http://www.kosis.kr

15. Measure of Cultural Norms  The sex ratio at birth (SRB) from 1991 to 1994 in the province of birth is used as a measure of parental son preference.  1991-1994 SRB is applied to all birth cohorts regardless of the year birth year.  Provincial differences in parental son preference remained unchanged over time.  The regional differences are most clearly revealed in SRB in the early 1990s.

16. Data  The 17 th wave of the Korean Labor and Income Panel Study (KLIPS)  Variables on time use and gender role attitudes  Variables on current individual and family characteristics  The entire waves of the KLIPS  Information on the place (province) of birth  The Annual Report on Live Births and Death Statistics  The sex ratio at birth for each province.  Sample Selection  952 dual-earner couples younger than 55

17. Model for Allocation of Housework in the Family

18. Multivariate Tobit Model

19. Descriptive Statistics: Household Characteristics Variables on Household Characteristics Ratio of schooling (Woman’s schooling/Man’s schooling)0.977 (0.148) Age difference (Man’s age – Woman’s age)2.503 (2.724) Number of children under 50.260 (0.531) Number of children age 5-180.966 (0.907) Number of adults family member2.382 (0.770) Ln(non-labor family income)3.212 (2.886) Live in metropolitan (other than Seoul)0.281 (0.450) Live in non-metropolitan area0.583 (0.493) Number of couples 945 Notes: Sample means and standard deviations are given in parenthesis.. “Live in metropolitan” indicates whether the couples live in one of the six metropolitan cities other than Seoul (i.e., Busan, Daegu, Daejeon, Incheon, Gwangju and Ulsan) or not.

20. Descriptive Statistics: Individual Characteristics ManWoman Variables on Individual Characteristics Sex ratio at birth of place of birth (POB)115.5 (4.555)(4.627) Years of schooling14.0513.53 (2.625)(2.347) Age44.3141.43 (6.704)(6.553) Wage worker0.8000.853 (0.400)(0.354) Hourly wage (unit: 10,000 KR won)1.7671.174 (1.271)(1.206) Usual work hours (hour/week)49.1741.90 (12.46)(12.92) Survey on working day0.8230.777 (0.382)(0.417) Notes: Sample means and standard deviations are given in parenthesis. The information on individual backgrounds (i.e., place of birth (POB), mother’s years of schooling) is collected from the 1998-2014 KLIPS.. “Usual work hours” reports the hours of work respondents usually spend in the labor market, not the hours on the day of the time use survey. We exclude respondents older than age 55 at the time of the 2014 survey from the sample.

21. Descriptive Statistics: Individual Time Use (minute/day) MenWomen Variables on Individual Time Use (minute/day) Housework time 40.286 (79.67) 211.24 (157.14) Housekeeping time17.46154.13 (46.68)(98.07) Childcare time20.5452.35 (57.24)(116.1) Family care time2.2864.762 (17.48)(40.50) Notes: Sample means and standard deviations are given in parenthesis. The information on time use is drawn from the 2014 KLIPS additional survey.

22. Housework Time by Gender and Man’s Place of Birth Not High5High 5 Women’s housework time 166.62 (99.31) 188.55 (135.05) Men’s housework time 24.76 (45.44) 28.18 (50.10) Source: 2014 KLIPS additional survey (time use survey)

23. Descriptive Statistics: Gender Index MenWomen 1. Agree: The ideal family is that the husband earns money and the wife looks after the home and family 0.381 (0.486) 0.372 (0.484) 2. Agree: Mother’s work has negative effects on preschool children 0.560 (0.497) 0.566 (0.496) 3. Disagree: Dual-earner couples should equally divide housework 0.275 (0.447) 0.127 (0.333) 4. Disagree: Husband’s and Wife’s incomes should be managed separately 0.741 (0.438) 0.675 (0.469) 5. Disagree: A house where a couple live together should be co-owned 0.560 (0.497) 0.359 (0.480) Gender index 2.516 (1.173) 2.099 (1.061) Observations945 Notes: Respondents were asked if they agree with six statements. We exclude one statement (“For gender equality between couples, women should work outside”) from our analysis because it is less straightforward if the statement is opposed to gender equality or not. Variables are coded as indicators with the value of 1 if the answer represents traditional gender role “Agree” or “Strongly agree” for Q1 and Q2, “Disagree” or “Strongly disagree” for Q3-Q5. “Gender index” is the summation of the five indicators.

24. Effects of Gender Norms on Housework Time Woman’s housework timeMan’s housework time (1)(2)(3)(4)(5)(6) Man’s POB sex ratio3.422*** 3.347***1.721 1.779 (1.185) (1.179)(1.644) (1.642) Woman’s POB sex ratio-0.260 -0.164-0.845 -0.859 (1.165) (1.156)(1.631) Man’s gender index 7.942**8.287** -0.102-4.259 (3.773)(3.681) (2.047)(5.154) Woman’s gender index 2.5181.702 1.1533.927 (4.160)(4.045) (2.257)(5.340) No. of children under 5107.0***100.2***106.4***74.75***33.76***75.42*** (12.08)(9.905)(12.06)(11.74)(5.373)(11.77) No. of children age 5-1830.31***26.29***29.84***20.89***5.557**21.31*** (5.265)(5.191)(5.239)(6.731)(2.816)(6.774) No. of adults family member4.6541.8203.353-20.93**-1.292-20.46* (6.393)(6.765)(6.306)(10.53)(3.670)(10.59) Ln(non-labor family income)-1.102-0.323-1.0520.1390.5460.126 (1.616)(1.476)(1.618)(2.036)(0.801)(2.031) Man’s job: self-employed3.8293.7413.941-14.34-1.877-14.69 (10.23)(10.63)(10.24)(15.96)(5.766)(15.91) Woman’s job: self-employed16.9414.8215.4928.50*9.35829.13* (10.80)(11.91)(10.78)(16.93)(6.460)(16.97) Metropolitan city (except Seoul)-30.85*-21.89-32.65*-13.33-4.859-13.49 (16.78)(14.00)(16.80)(19.50)(7.594)(19.59) Not metropolitan city-51.94***-46.52***-52.75***13.723.19113.39 (15.51)(12.74)(15.54)(17.09)(6.911)(17.16)

25. Woman’s housework timeMan’s housework time (1)(2)(3)(4)(5)(6) Man’s years of schooling-3.177-2.699-1.9042.7254.8361.967 (6.949)(7.252)(7.017)(10.37)(3.934)(10.31) Woman’s years of schooling5.7375.3134.62811.640.70412.39 (7.273)(7.544)(7.331)(10.90)(4.093)(10.87) Ratio of years of schooling-83.04-78.97-68.16-28.1621.06-38.50 (79.95)(97.10)(80.64)(138.6)(52.68)(137.5) Age (woman ; man)0.4920.09150.388-2.829**-0.882*-2.862** (0.954)(0.877)(0.959)(1.123)(0.476)(1.121) Age difference-2.855*-2.523-2.830*4.190*1.656*4.184* (1.637)(1.592)(1.623)(2.281)(0.905)(2.283) Man’s wage1.8261.0631.427-4.682-2.157-4.554 (4.918)(5.643)(4.925)(6.993)(3.061)(6.966) Woman’s wage-6.461-7.559-6.740*-10.23-1.723-10.12 (4.018)(6.538)(3.978)(10.90)(3.547)(10.89) Man’s wage*woman’s wage-2.350-0.982-1.6062.086-0.1931.848 (3.281)(3.917)(3.301)(4.901)(2.125)(4.912) Man: weekend48.71***46.58***48.47***-130.2***-61.30***-130.6*** (16.86)(13.31)(16.83)(18.43)(7.222)(18.50) Woman: weekend-194.3***-191.7***-194.0***-20.53-21.44***-20.49 (17.16)(12.25)(17.11)(15.76)(6.646)(15.78) Constant-30.01328.1***-65.91-72.9034.35-64.62 (164.8)(108.1)(164.0)(217.6)(58.63)(216.2) Observation945 Effects of Gender Norms on Housework Time (cont’d)

26. Summary of Baseline Results  Women married to more traditional men (whose parents had stronger son preference) tend to spend more time on housework.  Difference between Incheon and Kyungbuk (about 20) would produce more than one hours of additional housework by wife.  More traditional gender views of the husband increase the housework work time of the wife.  A man’s gender norms do not affect his own housework time.  The two measures of men’s gender norms independently affect women’s household work time.  How to explain the result?  “who he is” (habit, skill, model family) vs. “what he thinks” (education, social pressure, intention)

27. Women’s Time Spent on Housekeeping, Child Care and Family Care All Women who have at least one child (under age 18) (1)(2)(3)(4)(5)(6) House- keepingChild careFamily care House- keepingChild careFamily care Man’s POB sex ratio1.836**0.9040.3741.895**2.063*0.561 (0.744)(0.815)(0.347)(0.886)(1.082)(0.481) Woman’s POB sex ratio-0.640-0.09630.349-1.260-0.4820.508 (0.735)(0.805)(0.343)(0.871)(1.062)(0.474) 0.0097 (0.0542) 0.0088 (0.0572) -0.516*** (0.0846) -0.562*** (0.081) -0.0451 (0.738) -0.0623 (0.0823) Observation945 683

28. Robustness Checks (Women’s housework time) Sample size (1)(2)(3) Dependent variables: Women’s housework time A. Dummy variable Man’s POB: High595131.91***30.62***28.47*** (10.87)(10.79)(10.66) Woman’s POB: High5 -8.510-8.293-6.693 (11.00)(10.91)(10.81) B. Have at least 1 child Man’s POB sex ratio6834.690***4.502***4.259*** (1.542)(1.539)(1.488) Woman’s POB sex ratio -1.111-1.012-0.705 (1.518)(1.505)(1.475) C. OLS estimation Man’s POB sex ratio9453.132** 2.952** (1.209) (1.262) Woman’s POB sex ratio -0.348 -0.146 (1.107) (1.190) D. SUR estimation Man’s POB sex ratio9453.110***3.046***2.898*** (1.103)(1.100)(1.078) Woman’s POB sex ratio -0.329-0.234-0.0653 (1.091)(1.088)(1.067) Hours of work No Yes Gender index NoYes

29. Robustness Checks (Men’s housework time) Sample size (1)(2)(3) Dependent variables: Men’s housework time A. Dummy variable Man’s POB: High595115.0615.3721.40 (15.89) (16.15) Woman’s POB: High5 -11.56-11.64-16.29 (16.42) (16.53) B. Have at least 1 child Man’s POB sex ratio6832.4632.4732.838 (1.884)(1.883)(1.969) Woman’s POB sex ratio -1.655-1.657-1.943 (1.836)(1.835)(1.900) C. OLS estimation Man’s POB sex ratio9450.5940.5870.626 (0.510)(0.508)(0.569) Woman’s POB sex ratio -0.0674-0.0565-0.0352 (0.609)(0.612)(0.596) D. SUR estimation Man’s POB sex ratio9450.6190.6200.663 (0.598) (0.593) Woman’s POB sex ratio -0.0556-0.0562-0.0393 (0.592) (0.588) Hours of work No Yes Gender index NoYes

30. Summary of Additional Results  For the entire sample, men’s gender norms largely affect wife’s time spent on housekeeping.  For the sample with at least one child aged 18 or under, the wife’s time spent on housekeeping and child care are significantly affected by the husband’s gender norm variable.  The results are robust to changes in the measure of gender norms, sample selection, and model.

31. Additional Issues (1) Indirect Effect through Changes in Market Time?  Indirect effect through changes in hours of work?  Husband’s gender norms can influence the amount of market time of wife.  Housework time can be affected by hours of market work.  Working hours regressions  Man’s POB sex ratio has a weak effect on the wife’s working hours.  The husband’s gender index has a strong negative effect on the wife’s working hours.  Housework time regressions including working hours  If the amount of market work is controlled, the effect of man’s POB sex ratio remains unchanged.  In contrast, the coefficient for the gender index becomes insignificant.

32. A1. Effects of Gender Norms on Working Hours Woman's usual working hours (average per week) Woman's actual minutes of work on the survey day (1)(2)(3)(4)(5)(6) Man's POB sex ratio-0.0868-0.0778-3.051*-2.899* (0.109)(0.110)(1.607)(1.586) Woman's POB sex ratio0.1060.09420.8740.691 (0.107) (1.564)(1.536) Man's gender index-1.288***-1.244***-15.40***-15.08*** (0.373)(0.382)(5.703)(5.712) Woman's gender index0.09050.0362-1.879-2.202 (0.411)(0.418)(5.658)(5.671) ControlsYes Observations945953945 953945 Notes: Usual working hours (Column 1-3) indicate the average working hours per week answered in the section of the individual survey. Actual minutes of work (Column 4-6) measure time spent on working on a day before the survey, which is reported in the section of time use survey. Mean of usual working hours is 41.97 hours per week and mean of actual minutes of work is 381.27 minutes. Robust standard errors are reported in parenthesis. *** p<0.01, ** p<0.05, *p<0.1

33. Gender Norms, Working Hours, and Housework Time Woman's housework timeMan's housework time (1)(2)(3)(4) Man's POB sex ratio3.176***2.631**2.1222.153 (1.145)(1.072)(1.662)(1.626) Woman's POB sex ratio0.0110-0.00348-1.061-1.095 (1.130)(1.071)(1.647)(1.593) Man's gender index5.7024.097-2.378-3.630 (3.633)(3.543)(5.068)(5.049) Woman's gender index1.9071.5843.6572.455 (3.966)(3.764)(5.216)(5.116) Woman's usual working hours-2.055***1.898*** (0.321)(0.449) Man's usual working hours0.180-2.174*** (0.338)(0.560) Woman's actual minutes of work-0.384***0.133*** (0.0385)(0.0454) Man's actual minutes of work0.0711**-0.281*** (0.0337)(0.0462) ControlsYes Observation945

34. Additional Issues (2) Migration and Effects of Current Place of Residence  Effect of place of birth? Effect of place of residence?  The result could be driven by the effects of the cultural environment of the place where a person grew up.  Regressions including the sex ratios at birth (1991-1994) of the places of current residence and of residence at age 14.  No significant effect of the sex ratio of current place of residence  The effect of man’s POB sex ratio remains unchanged if the sex ratio of the current residence is controlled.  The sex ratio of man’s place of residence at age 14 has a significant positive effect on the wife housework, but the magnitude is smaller than that of POB.  Regressions including mover dummy interacted with the sex ratio in the place of residence at age 14.  Controlling for the place of birth, the place of residence at age 14 doesn’t appear to have additional effects on gender norms.

35. Effects of Place of Birth, of Adolescence and Current Living Place Woman's housework timeMan's housework time (1)(2)(3)(4)(5)(6) Man's POB sex ratio3.860***1.972 (1.254)(1.805) Woman's POB sex ratio0.159-0.605 (1.235)(1.799) Current region's sex ratio-1.1521.682-0.6630.120 (1.399)(1.171)(2.019)(1.334) Man's age14 sex ratio2.813**1.292 (1.200)(1.742) Woman's age14 sex ratio0.314-1.135 (1.162)(1.720) ControlsYes Observation945953944945953944 Notes: Current region’s sex ratio indicates the sex ratio at birth from 1991 to 1994 in the province in which couples were living in 2014. Man’s(woman’s) age 14 sex ratio represents the sex ratio at birth from 1991 to 1994 in the province where the husband(wife) lived at age 14. Control variables are the ones included in table 3. Robust standard errors are reported in parenthesis. Significant level: *** p<0.01, ** p<0.05, *p<0.1

36. A2. Effect of Mobility on Housework Time Woman's housework timeMan's housework time (1)(2)(3)(4) Man's POB sex ratio3.306***3.344***1.6931.624 (1.209)(1.193)(1.672)(1.663) Woman's POB sex ratio-0.329-0.161-0.606-0.827 (1.158)(1.173)(1.643)(1.649) Man's indicator: move-263.8-103.2 (337.3)(463.7) Woman's indicator: move46.63439.2 (256.8)(379.3) Man move*age14 sex ratio2.3450.05340.864-0.0388 (2.970)(0.109)(4.073)(0.131) Woman move*age14 sex ratio-0.489-0.0815-3.870-0.0494 (2.223)(0.0972)(3.314)(0.136) ControlsYes Observation945953944945 Notes: Move indicator is equal to 1 if respondents lived in the different region (province) from the place of birth at age 14. The sex ratio at birth from 1991 to 1994 in the province which respondents moved to is used in the interaction terms. Control variables are the ones included in table 3. Robust standard errors are reported in parenthesis. Significant level: *** p<0.01, ** p<0.05, *p<0.1

37. Additional Issues (3) Marriage Selection  Results driven by selections in marriage according to men’s cultural norms?  In marriage market, more traditional cultural norms of males would have to be compensated by other positive attributes (e.g. higher wage).  Married males from places with higher SRB could be positively selected in terms of productivity.  Test if marriage selection is stronger among men from places with higher SRB.  Estimate wage equation including interaction between marriage and SRB of place of birth.  The result does not support the marriage selection hypothesis.

38. Wage Equation Estimation

39. Mechanisms (1) Household Production Technology  Why longer total housework time for the family headed by more traditional husbands?  Household production technologies requiring more housework time of the wife?  i.e. outsourcing (domestic helper, eating out, ordering foods, prepared meals, etc.) and time-saving home appliances  The information on “eating out” reported in the KLIPS is problematic. Other variables are not available.  Regression examining how the man’s POB sex ratio affected the probability that the household had an electric washer in 1985  About 23% of the households with married couples aged 55 and younger had an electric washer in 1985.  An increase in the husband’s POB sex ratio is associated with 0.34~0.55 % points decrease in the probability of having an washer.

40. Sex Ratio at Birth in the County of Birth and Probability of Having Washer in 1985 Source: 2% micro sample of the 1985 census. Notes: Education refers to the years of schooling. Occupation variables include dummy variables for not working, manual, service, clerical (omitted category), and professional. Variables on city size include rural areas and small cities (omitted category), five metro cities, and Seoul. Robust standard errors are reported in parenthesis. Significant level: *** p<0.01, ** p<0.05, *p<0.1 Married Males 55 or YoungerMarried Females 55 or Younger (1)(2)(3)(4)(5)(6) POB sex ratio-0.0034***-0.0041***-0.0055***-0.0037***-0.0047***-0.0057*** (0.0002) Age, schooling, occupationNoYes NoYes Household size, city sizeNo YesNo Yes Observation132,313 143,645 Dependent variable mean0.233 0.221

41. Mechanisms (2) Formation of Housework Skills  Males raised in a traditional family did not have opportunities for developing skills required for household works.  Wives of these men spend more time on house work to compensate their husband’s lower productivity in home production.  Under a particular condition, the results on washer regression reject this hypothesis.  Other things being equal (no differences in preferences, etc.), less productive men would be more likely to purchase a washer.  Looking for a better way of testing the hypothesis  It is difficult to find measures of productivity in household work.

42. Possible Changes across Cohorts  Do the effects of the variables on gender norms become weaker as parental son preference diminishes over time?  Regressions separately conducted for each age group  The effects of the variables on gender norms become weaker for younger couples.  The effect of the man’s POB sex ratio is significantly positive for couples aged 36 to 55 (born in the 1960s and 1970s); it becomes insignificant for those aged 35 or younger (born after 1980).  The effect of gender index loses statistical significance for those born after 1970.

43. Housework Time by Husband Age Group Woman's housework timeMan's housework time Man's age: 46-55 Man's age: 36-45 Man's age <=35 Man's age: 46-55 Man's age: 36-45 Man's age <=35 (1)(2)(3)(4)(5)(6) Man's POB sex ratio4.105***4.024**2.1954.324-0.7354.861 (1.243)(1.804)(7.255)(2.635)(2.347)(4.316) Woman's POB sex ratio0.584-3.133*5.5360.9080.192-6.256 (1.184)(1.676)(6.651)(2.579)(2.157)(3.946) Man's gender index7.576*8.2591.240-7.374-7.379-2.508 (4.231)(5.742)(15.88)(9.171)(7.121)(10.55) Woman's gender index-1.2482.7933.0369.5532.10014.68 (4.222)(6.809)(16.95)(9.581)(7.366)(12.70) ControlsYes Observation423405124423405124 Dependent variable mean172.41233.33264.6820.0751.4170.89

44. Some Findings New to the Literature  The husband’s gender norms affects the wife’s housework time. Variables on gender norms do not affect own housework time.  Variables on parental son preference and own gender role attitudes of men independently affect the wife’s housework time.  Total housework time is longer for the families with more traditional men. Different household production technology might be an explanation.  The effect of the husband’s gender norms on housework time allocation becomes weaker over time.

45. Implications  Because of the importance of parental influences, changes in gender norms could be slower than socioeconomic changes.  Persistence or slow change in traditional gender norms among males could be an explanation for the high gender inequality in within-family time allocation in Korea.  This circumstance might explain why it is so difficult to change marriage and fertility behaviors quickly with policies for changing economic incentives.  The future prospect is brighter given the secular decline in son preference in Korea.