Gender Inequality. The Role of Culture. Raquel Fernández NYU, NBER, CEPR, IZA, ESOP NYU

Transcription

1 Gender Inequality The Role of Culture Raquel Fernández NYU, NBER, CEPR, IZA, ESOP NYU Inequalities Dondena Centre, 10th Anniversary Workshop 29 April 2016 Fernández (NYU ) Gender Inequality 29 April / 43

2 Introduction You ve come a long way (baby) Gender inequality historically ubiquitous Men enjoyed greater freedoms & opportunities than women of similar social characteristics Men possessed more power within the family Men held preponderance of social positions with significant political, economic, cultural power Dramatic changes over time, especially last 150 years married women obtained right to property women obtained the vote employment discrimination mostly illegal Fernández (NYU ) Gender Inequality 29 April / 43

3 Introduction Women in parliament over time Source: OECD Employment Database (2015). Percentage of parliamentary seats in a single or lower chamber held by women Fernández (NYU ) Gender Inequality 29 April / 43

4 Introduction Female share of seats in national parliaments Source: OECD Employment Database (2015). Percentage of parliamentary seats in a single or lower chamber held by women Fernández (NYU ) Gender Inequality 29 April / 43

5 Introduction Employment over time Source: OECD Employment Database (2014). Full time equivalent employment rates, by sex The full-time equivalent employment rate is calculated as the employment/population ratio for yeas old, multiplied by the average usual hours worked per week per person in employment (both dependent and self-employment), and divided by 40. Fernández (NYU ) Gender Inequality 29 April / 43

6 Introduction Employment Gender Ratio Source: OECD Employment Database (2014). Full time equivalent employment rates, by sex The full-time equivalent employment rate is calculated as the employment/population ratio for yeas old, multiplied by the average usual hours worked per week per person in employment (both dependent and self-employment), and divided by 40. Fernández (NYU ) Gender Inequality 29 April / 43

7 Introduction Gender Wage Gap Source: OECD Indicators (2016). Gender wage gap: Total, of male median wage, 2012 The gender wage gap is unadjusted and is defined as the difference between median earnings of men and women relative to median earnings of men. Data refer to full-time employees and to self-employed. Fernández (NYU ) Gender Inequality 29 April / 43

8 Introduction Gender Education Gap Source: OECD Indicators (2015). Education at a Glance 2015 Fernández (NYU ) Gender Inequality 29 April / 43

9 Introduction Progress: why so fast? why so slow? Why has gender inequality changed so dramatically? urbanization declining fertility technological change In general, complicated interplay between changes in technology, in the economy, in institutions, and in culture Today s main topic: culture and gender Outline: show culture plays a role why does culture change? The role of learning some future directions for research Fernández (NYU ) Gender Inequality 29 April / 43

10 Introduction Culture Culture: A body of shared knowledge, understanding, and practice. By 1950s, Kroeber and Kluckhohn (1952) compiled over 150 definitions of culture... For our purposes: Differences in culture as systematic variation in beliefs and preferences across time, space, or social groups. Note: Culture need not be irrational nor static. Fernández (NYU ) Gender Inequality 29 April / 43

11 Introduction Economists and Culture Economists been reluctant to consider culture in their explanations. Why? Stigler-Becker dictum: De gustibus non est disputandum. Theory: Ad hoc analysis Empirical counterpart: culture as the residual as what is left unexplained (e.g., in a cross country regression) is not rigorous. Main challenge: Separating differences in beliefs and preferences from differences in environments (e.g. institutions, policies) or individual characteristics (e.g. education) Fernández (NYU ) Gender Inequality 29 April / 43

12 The Epidemiological Approach The Epidemiological Approach: Fernández 2007 Epidemiologists ask: Are cross-country health differences driven by genetic or environmental (including cultural) factors? Compare a health outcome (e.g. heart disease) for immigrants with that for natives. Is there convergence in health outcomes? - If convergence environment - If convergence is not observed, one cannot automatically conclude genetics. Fernández (NYU ) Gender Inequality 29 April / 43

13 The Epidemiological Approach The Epidemiological Approach in Economics Does culture play a role in generating cross-country differences in an economic outcome (e.g. saving rates)? A different set of problems: The behavior of immigrants may be differentially affected by shocks (e.g., language, employment, etc.) Assimilation: beliefs influenced by mainstream culture Selection: immigrants are not the average individual Endowments: human capital differences, etc. Fernández (NYU ) Gender Inequality 29 April / 43

14 The Epidemiological Approach 11 Fernández (NYU ) Gender Inequality 29 April / 43

15 The Epidemiological Approach Why does women s LFP differ across countries? Structure of the economy - Wage differences - Production/sectoral composition - Policy differences (unemployment insurance, taxation...) - Institutional differences (divorce laws, childcare arrangements, welfare system, extended households...) What about differences in culture? Fernández (NYU ) Gender Inequality 29 April / 43

16 The Epidemiological Approach 13 Fernández (NYU ) Gender Inequality 29 April / 43

17 The Epidemiological Approach Fernández and Fogli (2006,2009), Fernández (2007,2008) Use the epidemiological approach to study married women s LFP and fertility. Examine an arguably less problematic sample: second-generation Americans. Shocks and unobserved economic factors less important. Problem of cultural assimilation is worse. Basic hypothesis: Women born in US face same economic and institutional environment. They have foreign-born parents potentially different cultures (beliefs about women s role, ideal family size, etc.) Exploit fact that culture is significantly more portable than institutions Does variation in culture help explain variation in LFP? Fernández (NYU ) Gender Inequality 29 April / 43

18 The Epidemiological Approach Why Test may Fail Cultural proxy = Female LFP in country of ancestry Even if culture matters, we may fail to find it with this test: Other economic and institutional factors affect female LFP in countries of ancestry they may offset cultural variation assimilation selection social incentives may be different culture is a social phenomena! Thus, if cultural proxy does not possess explanatory value, this does not imply that culture is irrelevant. Fernández (NYU ) Gender Inequality 29 April / 43

19 The Epidemiological Approach Challenge There also exist economic channels of intergenerational transmission: wealth, education, ability, etc. Have we adequately controlled for differences in economic variables? Fernández (NYU ) Gender Inequality 29 April / 43

20 The Epidemiological Approach Sample selection The 1970 US census (last census to ask individuals where their parents were born). We use father s birthplace as ancestry. Married women yrs old, born in US. Not living in institutions, not living on a farm, occuption not in agriculture. In which decade to measure female LFP in country of ancestry? ? - Because of data limitations Also argument for 1970 LFP. - Robust to later decades. Very high correlation over the decades ( ). Exclude USSR and 10 European countries which became centrally-planned economies around WWII - Culture may have changed significantly and parents were living in the US then. Fernández (NYU ) Gender Inequality 29 April / 43

21 1970 Hours Worked in U.S. The Epidemiological Approach Japan Cuba Philippines China Switzerland Denmark Portugal Mexico Germany Finland Lebanon Canada Norway Turkey Sweden Greece Netherlands Italy France Austria U.K. Spain Ireland Belgium Syria Log 1950 Female LFP (%) 24 Fernández (NYU ) Gender Inequality 29 April / 43

22 The Epidemiological Approach Dependent variable is Hours Worked in 1970 (i) (ii) (iii) Female 0.047** 0.041* 0.072** LFP 1950 (0.012) (0.016) (0.015) High School ** (0.520) (0.575) Some College ** (1.078) (1.034) College ** (0.492) (0.494) Husband * High School (0.730) Husband Some College (0.829) Husband ** College + (0.452) Husband ** Total Income (0.308) Obs Adj. R-sq Fernández (NYU ) Gender Inequality 29 April / 43 27

23 The Epidemiological Approach Potential Problem Can a strictly economic variable be responsible for the positive correlation btwn cultural proxy and economic outcome? Main source of concern: Unobserved human capital embodied either in the woman herself or in her ethnic network. Include GDP pc in 1950 (also other years) Fernández (NYU ) Gender Inequality 29 April / 43

24 The Epidemiological Approach Potential Problem Can a strictly economic variable be responsible for the positive correlation btwn cultural proxy and economic outcome? Main source of concern: Unobserved human capital embodied either in the woman herself or in her ethnic network. Include GDP pc in 1950 (also other years) Parental education use different (smaller) data set (GSS) Fernández (NYU ) Gender Inequality 29 April / 43

25 The Epidemiological Approach Potential Problem Can a strictly economic variable be responsible for the positive correlation btwn cultural proxy and economic outcome? Main source of concern: Unobserved human capital embodied either in the woman herself or in her ethnic network. Include GDP pc in 1950 (also other years) Parental education use different (smaller) data set (GSS) Ethnic human capital 1940 Census Fernández (NYU ) Gender Inequality 29 April / 43

26 The Epidemiological Approach Potential Problem Can a strictly economic variable be responsible for the positive correlation btwn cultural proxy and economic outcome? Main source of concern: Unobserved human capital embodied either in the woman herself or in her ethnic network. Include GDP pc in 1950 (also other years) Parental education use different (smaller) data set (GSS) Ethnic human capital 1940 Census Cross-country measures of quality of education (Hanushek and Kimko) Fernández (NYU ) Gender Inequality 29 April / 43

27 The Epidemiological Approach Potential Problem Can a strictly economic variable be responsible for the positive correlation btwn cultural proxy and economic outcome? Main source of concern: Unobserved human capital embodied either in the woman herself or in her ethnic network. Include GDP pc in 1950 (also other years) Parental education use different (smaller) data set (GSS) Ethnic human capital 1940 Census Cross-country measures of quality of education (Hanushek and Kimko) Wage regressions if there is unobserved human capital, LFP variable should help explain wages. Fernández (NYU ) Gender Inequality 29 April / 43

28 The Epidemiological Approach Attitudes and Work I use two questions from the 1990 World Value Survey (Fernández 2007) 1. Being a housewife is just as fulfilling as working for pay. 2. Having a job is the best way for a woman to be an independent person. Fernández (NYU ) Gender Inequality 29 April / 43

29 The Epidemiological Approach Implement Epidemiological Approach Restrict sample to Europe (only 2-3 non-european cos. Approx indivs./country.) Run individual level Probit on whether individuals agreed (strongly or agree) with statement on country dummy. Control for demographics Use marginal effects associated with country dummies (or simply proportion who agreed) as proxies for culture. Does the country s marginal effect measured in 1990 have explanatory power in US 1970? This is the same epidemiological structure as before. Fernández (NYU ) Gender Inequality 29 April / 43

30 The Epidemiological Approach Being Housewife is as Fulfilling as Working for Pay Denmark Finland Portugal Turkey Austria Sweden Germany Norway Italy France Netherlands England Spain Ireland Belgium % Disagree or Strongly Disagree (WVS 1990) 38 Fernández (NYU ) Gender Inequality 29 April / 43

31 The Epidemiological Approach Job is Best Way for Women to be Independent Denmark Finland Portugal Germany Norway Netherlands England Austria Sweden Italy France Spain Ireland Belgium % Agree or Strongly Agree in WVS Fernández (NYU ) Gender Inequality 29 April / 43

32 Cultural Change Some Big Questions Why does culture change? Is there a role for policy? How do culture, economic incentives, and institutions interact and codetermine each other? Fernández (NYU ) Gender Inequality 29 April / 43

33 Cultural Change Fernández (NYU ) Gender Inequality 29 April / 43

34 Cultural Change Why has women s LFP changed? Technological change household technology: Greenwood, Seshadri, Yorukoglu (2005) work place technology (sectoral change): Goldin (1990), Rendall (2010) medical technology (contraception, maternal mortality, etc): Goldin & Katz (2002), Albanesi & Olivetti (2009), Knowles (2009) Wage changes returns to experience: Olivetti (2006). Knowles (2007) gender wage gap; Jones, Manuelli, & McGrattan (2003), Gayle & Golan (2010) Institutional change Changes in cost of child care: Attanasio, Low, & Sanchez-Marcos (2008) Divorce laws: Fernández and Wong (2012, 2014) What about changes in culture? Fernández (NYU ) Gender Inequality 29 April / 43

35 Cultural Change 10 Fernández (NYU ) Gender Inequality 29 April / 43

36 Cultural Change Hypothesis: cultural change as a process of learning (Fernández 2014) Cultural change as the evolution of beliefs in a rational inter-generational learning process. S-shape figure of female LFP may be clue that process similar to tech. diffusion may be taking place (albeit much slower) Fernández (NYU ) Gender Inequality 29 April / 43

37 Cultural Change Technology Diffusion Fernández (NYU ) Gender Inequality 29 April / 43

38 Cultural Change S-Shapes: England & France LFP France Great Britain Data for Great Britain is from Table 1 in Costa (2000). Data for France is compiled from Tables 1 and 2 in Reboud (1985). Year Fernández (NYU ) Gender Inequality 29 April / 43

39 Cultural Change Cultural Change as Learning True cost (to psyche, marriage, children) of a woman working is uncertain Observe other women working and learn (update beliefs) about cost Percentage of women who work and distribution of beliefs evolve endogenosly I calibrate learning model to US statistics and see how well it performs in generating LFP I show that the contribution of endogenously changing beliefs regarding the cost of women working is quantitatively important Implications of model: learning can be sped up by technological and institutional change role for policies that incentivize LFP when learning (and cultural change) is slow hold for a policy-maker that has same beliefs as population (i.e., results do not rely on government being better informed) Fernández (NYU ) Gender Inequality 29 April / 43

40 Cultural Change Learning from Experience: Rosie the Riveter Fernández, Fogli, Olivetti (2004) Isolate the idiosyncratic component in the variation across US states in men s mobilization rates during WWII. In states with higher mobilization rates, more women worked. We find that this effect carries over to next generation More women work in the next generation as well This is consistent with learning: Preferences/attitudes transmitted from mothers to daughters/sons. Learning from the experience of previous generation. Employers learning about women as workers. Fernández (NYU ) Gender Inequality 29 April / 43

41 Cultural Change Learning from Experience: Female Leaders Quotas and Women Leaders Since 1998, in W. Bengal a random 1/3 of village council-leader positions reserved for women. Survey evidence (Beaman et al 2008) suggests that exposure to female leaders weakened gender stereotypes associated with various activities Men were less apt to associate women with domestic roles and more with leadership roles. More importantly: 10 years after the quotas were put into place, women were more likely to run and win elections for unreserved seats if the village seat had been reserved in the prior 2 elections. Fernández (NYU ) Gender Inequality 29 April / 43

42 Cultural Change Exposure to Other Social Norms: Learning by Watching Brazilian soaps portrayed families that were much smaller than in reality. Variation in the timing of entry of the TV station used to show that exposure to soap operas is correlated with lower fertility. La Ferrara, Chong and Duryea (2012) When cable TV was introduced in rural parts of India, women in exposed villages were less likely to express son preference, fertility fell, and autonomy increased. Jensen and Oster (2009) Fernández (NYU ) Gender Inequality 29 April / 43

43 Future Work Future Directions for Research Interaction between culture, economics, and institutions: how best to intervene Fernández (NYU ) Gender Inequality 29 April / 43

44 Future Work Future Directions for Research Interaction between culture, economics, and institutions: how best to intervene What are the implications for policy, for cultural/economic evolution of a significant gender gap regarding women s role? Fernández (NYU ) Gender Inequality 29 April / 43

45 Future Work Attitudes towards Education Source: Gender Equality Universally Embraced, But Inequalities Acknowledged. Pew Research Centers Global Attitudes Project, July 1, Fernández (NYU ) Gender Inequality 29 April / 43

46 Future Work Future Directions for Research Interaction between culture, economics, and institutions: how best to intervene What are the implications for policy, for cultural/economic evolution of a significant gender gap regarding women s role? Will women s greater education relative to men bring about deep change? Fernández (NYU ) Gender Inequality 29 April / 43

47 Future Work Marriage and Relative Earnings The sample includes married couples where both the husband and wife earn positive income and are between 18 and 65 years of age. SIPP From Bertrand, Kamenica, and Pan (2015). Fernández (NYU ) Gender Inequality 29 April / 43