Woman and mother: double employment penalty?

Master in Economics: Empirical Applications and Policies University of the Basque Country UPV/EHU Master Thesis Woman and mother: double employment penalty? Odra Quesada Campos Supervised by Sara de la Rica and Luc´ıa Gorj´on (ISEAK) July 30, 2020 Acknowledgments Hereby, I would like to thank to ISEAK Foundation for giving me this exceptional opportunity to collaborate with them in this paper. Their knowledge has enriched the analysis beyond all those numerical results obtained on this Master Thesis. They transmitted me their passion about this topic so, I hope I can do the same with the readers. Thank you, Sara and Luc´ıa, for all. ———————————– Odra Quesada Campos July 30, 2020 2 List of Figures 1 Evolution of gender gaps in employment levels in Spain by age groups . . . 8 2 Children and weekly hours worked by gender . . . . . . . . . . . . . . . . . 9 3 Couple and labor situation by gender . . . . . . . . . . . . . . . . . . . . . 19 4 Couple and weekly hours worked by gender . . . . . . . . . . . . . . . . . . 20 5 Children and labor situation by gender . . . . . . . . . . . . . . . . . . . . 21 6 Reasons of holding a part-time job among couples . . . . . . . . . . . . . . 22 7 Children, work schedule and type of contract by gender . . . . . . . . . . . 23 8 Occupational sex segregation, work schedule and gender . . . . . . . . . . . 24 9 Children and labor situation by gender - Second quarter of 2007 . . . . . . 32 10 Children and labor situation - Regional disparities . . . . . . . . . . . . . . 35 11 Estimated probability of being employed among parents - Differences betweenwomenandmen.............................. 37 A.1 Couple and type of contract by gender . . . . . . . . . . . . . . . . . . . . 43 A.2 Couple and labor situation by gender - High educated individuals . . . . . 43 A.3 Children and type of contract by gender . . . . . . . . . . . . . . . . . . . 44 A.4 Job segregation by sectors of activity, work schedule and gender . . . . . . 44 A.5 Couple and labor situation by gender - Second quarter of 2007 . . . . . . . 45 A.6 Children and labor situation - Regional disparities . . . . . . . . . . . . . . 46 A.7 Aragon...................................... 46 A.8 Asturias ..................................... 46 A.9 BalearicIslands................................. 46 A.10CanaryIslands ................................. 46 A.11Cantabria .................................... 46 A.12CastillayLaMancha.............................. 46 A.13CastileandLe´on ................................ 47 A.14Catalonia .................................... 47 A.15Valencia ..................................... 47 A.16Extremadura .................................. 47 A.17Galicia...................................... 47 A.18Madrid...................................... 47 A.19Murcia...................................... 48 A.20Navarra ..................................... 48 A.21LaRioja..................................... 48 3 List of Tables 1 Classification of the individuals within the target population. . . . . . . . . 14 2 Marginal Effects from estimating Equation (4) . . . . . . . . . . . . . . . . 26 3 Marginal Effects from estimating Equations (5), (7), (8), and (9) . . . . . . 28 4 Distribution of individuals among the second sub-sample - 2007 . . . . . . 31 5 Marginal Effects from estimating Equations (5), (7), (8), and (9) for the secondquarterof2007 ............................. 33 6 Differences in educational attainment among those couples with children employedinAragon. .............................. 38 A.2.1Results of probit models - First sub-sample . . . . . . . . . . . . . . . . . . 49 A.2.2Results of probit models - Second sub-sample - 2019 . . . . . . . . . . . . . 51 A.2.3Results of probit models - Second sub-sample - 2007 . . . . . . . . . . . . . 53 A.2.4Results of probit models by regions - Second sub-sample - 2007 . . . . . . . 55 A.2.5Results of probit models by regions - Second sub-sample - 2019 . . . . . . . 57 4 Contents 1 Introduction 7 1.1 Research hypothesis and motivation . . . . . . . . . . . . . . . . . . . . . . 7 1.2 Motherhood and labor market . . . . . . . . . . . . . . . . . . . . . . . . . 9 2 Literature review 10 3 Data and methodology 12 3.1 Descriptionofthedata............................. 12 3.1.1 Sub-sample 1: Singles and childless couples . . . . . . . . . . . . . . 13 3.1.2 Sub-sample 2: Couples with and without children . . . . . . . . . . 13 3.2 Methodology .................................. 15 4 Descriptive findings 18 5 Estimation results 25 6 Extensions 30 6.1 Temporalperspective.............................. 31 6.2 Regionalperspective .............................. 34 7 Summary and final remarks 38 A Appendices 43 A.1 Other descriptive graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 A.2 Othertableresults ............................... 49 5 Abstract This paper analyzes how females and males modify their labor situation after having a child to reconcile work and family life in Spain. The analysis aims to examine whether the heterogeneity in parents’ employment is explained by gender differences or by differences concerning sociodemographic characteristics, educational and occupational attainment, among others. Using 2019 data from the Spanish Labour Force Survey (SLFS), estimation focus on the differences by gender of four different outcomes: being employed, being parttime worker, having a temporary contract, and working less than 35 hours per week. We improve the national perspective by applying the same approach to the Spanish regions and changing the temporal perspective to a pre-crisis year like 2007. Changes in the employment patterns related to the arrival of children are only present on the females’ labor supply, also anticipated by changes when they get into a couple of relationships. Keywords: female employment, gender gaps, gender inequality, parenthood. 6 1 Introduction In the last decades, women have moved towards greater participation in the labor markets. Thus the feminization of the labor force is a stylized fact that characterizes all the OECD economies. However, there still exist differences among women and men related to lower female participation rate1. There are some countries where this gap is narrower, getting very close to a one-to-one female to male employment rate, but this type of gender inequality persists in most countries. Women participate more actively in the labor market due to changes in preferences, increasing opportunity costs, and changes in socio-demographic characteristics such as higher educational attainment. Thus, the gender employment gap has reduced in favor of women’s participation. However, more women at work do not necessarily imply a more egalitarian situation in terms of wages, hours worked, and labor stability. Another fact related to the process of female incorporation to the labor market is that, while their participation rates increase, fertility rates have declined in most developed countries [1]. In Spain, among other Southern European countries, low participation and low fertility rates coexist together in the economy, highlighting that fiscal policies need to be improved to allow women to work and have children. This negative relation between fertility and labor supply is a cause of concern among those European countries where both rates are lower than the rest, which is the case of Spain, Italy, and Greece. Since working generations finance with their work the pensions benefits of the previous ones, the sustainability of the pension system, which will cover theirs, depends on a new working generation: their children. We have focused the analysis on the differences in the labor market situation among women and men. However, we have also considered if there exist differences among them depending on their household situation (if they live with their parents, with flatmates or they live with a couple) and on their fertility situation (if they have or not children). 1.1 Research hypothesis and motivation This master thesis seeks to assess and quantify the factors that are driving the actual highly gendered employment patterns that have been proved to be necessary in order to determine empirically social and labor market policies. As we will focus on the labor market outcomes of females and males separately, we have several questions that we would like to answer along with this paper: how many individuals are employed, what type of jobs do they have, do these employment patterns differ if they are females and mothers 1Estimated by the International Labour Organization (ILOSTAT database) as the percentage of female population above 15 years old. 7 or males and fathers and, if it would be the case, the magnitude of such differences. The aim of this master thesis is twofold: First, we test whether there are differences in terms of the labor situation, type of contract, work schedules and, hours worked between females and males in Spain, comparing females and males that have no family commitments. Second, if those differences exist, to what extent they are due to maternity or paternity compering among couples with and without children. In this case, we will analyze if the adults’ employment patterns change with the arrival of children and, thus, we can evaluate if the adjustment is symmetric by gender. Spain is one of the developed countries where the process of female incorporation into the labor market has taken place most recently. This rising female employment rate has a positive effect reducing the gender employment gap2that is remarkably lower in the youngest cohort, even though it has not yet been eliminated anywhere, as we can observe in Figure 1. We can see a decreasing tendency years after the economic crisis and until 2013 among the different age-groups. This was mainly due to a decrease in males’ employment rate, given that construction, which was the economic sector hardest hit by the Great Recession, is considered as a male-dominated industry. In the case of Spain, the incorporation of women into the labor market has transformed the economy from a model based on the segregation of tasks by gender towards a dual-income household model [8]. Indeed, this mass incorporation of women into the workforce increased the female’s labor force participation rate up to 51.8% in 2019. Nevertheless, higher female employment rates have been unable to remove the actual gender gap in labor force participation. Figure 1: Evolution of gender gaps in employment levels in Spain by age groups 0% 5% 10% 15% 20% 2009 20192010 2011 2012 2013 2014 2015 2016 2017 2018 From 16 to 24 years old From 25 to 54 years old From 55 to 64 years old Source: Spanish Labor Force Survey (SLFS) - Own elaboration in Stata 14.0 2Difference in employment rates between men and women. 8 Besides of the gender differences in employment, we also look for the existence of child penalties3, in terms of the type of contract, work schedules and hours worked, and if childbirth still ends up paid by women. 1.2 Motherhood and labor market Following childbirth, many families adopt gendered strategies to conciliate family and work responsibilities [21]. It has been documented that with the arrival of a child, mother and father increase their hours worked (paid and unpaid4). But the reallocation of time is not identical among both parents as Rapoport et al. (2011) [23] and Rizavi and Sofer (2010) [24] pointed out. Men typically increase the time they devote to paid work, while women decrease their paid working time or even exit the labor market [4], which implies an increase in the unpaid work. These differences in the supply of hours of work can be due to different preferences about work and home (in terms of division of household work and child care activities). Nevertheless, it would be hard to claim that the Spanish employment gap is due to women’s preferences when gaps in other European countries are so much lower. Those disparities in hours worked among couples are consistent with the Spanish evidence, as we can observe in Figure 2. Figure 2: Children and weekly hours worked by gender 91 2 4 2 94 2 3 1 96 2 2 1 97 1 2 1 77 7 11 5 66 10 17 7 69 8 16 8 71 6 14 9 0% 20% 40% 60% 80% 100% Couple with no children With children younger than 5 years old With children between 5 years old and 15 years old With children older than 15 years old Couple with no children With children younger than 5 years old With children between 5 years old and 15 years old With children older than 15 years old Males Females 35 hours or more 34-30 hours 29-20 hours Less than 20 hours Source: Spanish Labor Force Survey (SLFS) - Own elaboration in Stata 14.0 It is striking that being a mother is closely related to a decrease in the hours worked, 3Defined by Kleven et al. (2018) as the percentage by which women fall behind men due to children. 4Anxo et al. (2011) defined as time devoted to unpaid work, full range of care activities chores related with household work. 9 F() = 1 √2πZ −∞ e−t2 2dt Since binary regressions models are not linear, we cannot interpret directly the coefficients obtained from the estimations because they will not describe the relation between a variable and the outcome probability. We decide to show the results through discrete changes, which in our case means how much changes the predicted probability by changes from 0 to 1 for dummy variables, holding all the covariates constant, which can be also called marginal effects8, and that is define as: ∆Pr(yi= 1|xi) ∆xk,i =Pr(yi= 1|xi, xk,i = 1) −Pr(yi= 1|xi, xk,i = 0) (3) Below, we proceed to detail the estimations: •Being employed: We will estimate this probability separately for both subsamples. We included here as employed all those self-employed, entrepreneurs and, employees. We will control for gender, the interaction of gender and couple, and also, the iteration of gender and children9, and the vector of socio-demographic covariates includes nationality, age group, level of education and, the region of residence. P r(Employedi= 1) = F(β0+β1F emalei+β2Couplei+β3Femalei·Couplei +α0Socio-demographici)(4) P r(Employedi= 1) = F(β0+β1F emalei+β2Childreni+β3Femalei·Childreni +α0Socio-demographici)(5) We must say something about the variables included in the equations. In particular, it is very likely that the individual’s decision to participate in the labor market is influenced by the individual’s probability of being in the labor market and by her or his personal characteristics. The sets of variables included in the first equation, which estimates the probability of being employed, reflect individual characteristics such as age, nationality, educational attainment, and information concerning child status. Following Heckman [12], we will apply a two-step estimation procedure to control for the non-random probability of working. We use this type of double hurdle model 8We are referring to marginal effect to the discrete change instead of the common use of partial change. 9Couple means that the individuals are cohabiting with her/his partner and children means that the individual is mother or father. We will not take into account children’s years old in these regressions because we have restricted the age up to 30 years old. 16 because we assume that some of the covariates that affect the probability of working under the job attributes we want to analyze could have also affected the probability of being employed. Thus, we are running two models to estimate each labor output. The first model corresponds to the first step of Heckman and, it will be a choice or selection model, estimated using Equation (2). This will take into account how individuals select (positively or negatively) into employment, called sample selection. From this estimation, we will use the prediction of the probability of being salaried worker10, which will be incorporated into the second model as an independent variable. P r(Salaried workeri= 1) = F(β0+β1F emalei+β2Childreni+β3Femalei·Childreni +α0Socio-demographici)(6) The second model will estimate the probability of working part-time, holding a temporary contract, and working less than 35 hours per week, including the predicted probability of being employed. Following this procedure, we correct the estimation for sample selection, based on Heckman’s two-step consistent estimator. This sample correction takes care of the fact that working part-time or with a temporary contract, for example, differs across educational and occupational groups. With the significant differences in employment rates among educated individuals, the results of those outcomes could be distorted if we have not taken into account the apparent heterogeneity of the labor supply by gender and by personal status (unpartnered individuals versus partnered ones and, childless individuals versus those with children). •Working with a temporary contract: We estimate the likelihood of working with a temporary contract only for the second sub-sample, dividing them in two (Childless couples and couples with children). This model is controlled by all the characteristics mentioned above and also for some related to their job attributes like occupational category, sector of activity, and the probability of being a salaried worker. P r(Temporary contracti= 1) = F(β0+β1F emalei+β2Childreni+ β3Femalei·Childreni+β4c P r(Salaried workeri= 1) + α0Socio-demographici+ γ0Job attributesi) (7) •Working part-time: We estimate the probability of having a part-time job as we have just done in Equation (7) but, in this case, we also control for the type of 10We do not include the estimated probability of being employed due to the structure of the SLFS survey. They obtain the information related to work schedules, type of contract and, hours worked from salaried workers only. 17 contract. We estimate this labor outcome among couples to obtain the relation of having children and working part-time. P r(Working part-timei= 1) = F(β0+β1F emalei+β2Childreni+ β3Femalei·Childreni+β4c P r(Salaried workeri= 1) + β5Type of contracti+ α0Socio-demographici+γ0Job attributesi) (8) •Working less than 35 hours per week: We choose to control for the same characteristics that we use to estimate the probability of working part-time, focusing on those individuals within Childless couples and couples with children categories. P r(Working less than 35 hours/weeki= 1) = F(β0+β1F emalei+β2Childreni+ β3Femalei·Childreni+β4c P r(Salaried workeri= 1)+ β5Type of contracti+α0Socio-demographici+γ0Job attributesi) (9) Regarding the estimations of the probability of having a temporary contract, working part-time, and working less than 35 weekly hours, other explanatory variables are also included in the empirical estimation, such type of contract, occupational level, and sector of activity. Besides, some regional dummies have been included to capture location changes concerning our four dependent variables. Despite their low explanatory power, they have been included in the final estimation. 4 Descriptive findings This section presents some of the main findings that describe the Spanish labor market situation from a gender perspective. As a European country, Spain participates in the current convergence of employment rates between males and females with the others OECD members. Nevertheless, gender inequality in terms of the level of employment is still present. As we can observe in Figure 3, when individuals have not yet emancipated, their level of employment is low and highly related to partial time jobs which can result due to demand market restrictions (for example, labor experience) or supply preferences, since maybe young individuals want to earn less because they have no expenses and they prefer having free time. Women are more associated with part-time jobs since the beginning of their labor careers. Once emancipation occurs, and both individuals move towards more economic independence, they participate more in the labor market. Until cohabitation, we cannot observe gender gaps in the employment rates among individuals, although there 18 are gender differences related to part-time jobs. On the contrary, women participate more than men, and also they show lower levels of inactivity with the absence of family commitments. Figure 3: Couple and labor situation by gender 54 12 23 11 66 8 16 10 82 5 8 5 84 6 7 2 44 23 23 9 63 18 13 7 72 15 9 4 69 15 10 6 0% 20% 40% 60% 80% 100% Singles living with their parents Singles living with others Singles (One-person family) Couples with no children Singles living with their parents Singles living with others Singles (One-person family) Couples with no children Males Females Employed - Full-time Employed - Part-time Unemployed Inactive Source: Spanish Labor Force Survey (Second quarter, 2019) - Own elaboration in Stata 14.0 Anxo et al. (2011) [4] stated that cohabitation for women of all ages means a heavy burden in terms of unpaid work. However, we cannot find such a negative relation between cohabitation and employment rates in our target population. In the case of women, they are 3 percentage points (p.p.) less allocated on full-time jobs compared to those females one-person family, but they have the same participation in part-time jobs. For males, the relationship between cohabitation and employment is not significantly higher than that of single males. It is important to remark that part-time work schedules are a matter of women, whether they cohabit with their partner or not. There are 12.8 points more females than males declaring that they work part-time because they do not find a full-time work11. They belong to that 59.12% who work part-time involuntary, forced by the labor market conditions. The second most declared reason, among those singles and childless couples, is follow teaching or training courses. Also, part-time jobs are more likely to be perceived as involuntary by women in the South than in northern Europe [7]. Being a woman is also associated with a lower level of hours worked12 compared to 11SLFS survey asks all those individuals who work within a part-time work scheme the reasons of having that type of working day. 12We have used as hours worked all those hours reported by employees that SLFS references as hours worked by contract. We will not be taking into account all those hours worked by self-employed, entrepreneurs, etc. 19 males with the same household situation. As we can observe in Figure 4, among those Singles (One-person family), only 78% of those female employees work more than 35 hours/week whereas in the case of males they reach a 92%. Thus, we have proved the existence of a gender gap in weekly hours worked close to 14 p.p. that keeps almost unchanged when they are childless couples. Figure 4: Couple and weekly hours worked by gender 79 4 10 7 87 3 6 4 92 2 4 2 91 2 4 2 61 5 19 15 73 6 14 7 78 7 11 4 77 7 11 5 0% 20% 40% 60% 80% 100% Singles living with their parents Singles living with others Singles (One-person family) Couples with no children Singles living with their parents Singles living with others Singles (One-person family) Couples with no children Males Females 35 hours or more 34-30 hours 29-20 hours Less than 20 hours Source: Spanish Labor Force Survey (SLFS) - Own elaboration in Stata 14.0 That penalty will be an obstacle in the females’ labor life because they will have to work against those differences when, perhaps, they decide to have a child, which it will be per se a penalty in their careers. This gap in hours worked will be closely related to the persistence of the gender pay gap. We have already analyzed which are the main differences by gender depending on their household situation, from those who are not already emancipated to those that are cohabiting with their partners. However, we can also analyze what happens with those who have decided to have children. Is there any difference in their labor situation comparing to those who have not? Furthermore, do there exist gender gap among those that are already parents? As we have mentioned, parenthood does have effects on employment levels. In the same manner, it is also proved by early literature that mothers adjust more their labor supply than their male counterparts. In this part of the section, we analyze the differences among those partnered cohabitant individuals depending on their fertility status (childless individuals or parents), again from a gender point of view. When both individuals cohabit together, women participate 6 points less than men, 20 Figure 5: Children and labor situation by gender 84 6 7 2 86 4 8 2 87 3 7 3 81 2 7 9 69 15 10 6 48 19 12 22 53 18 12 17 48 15 10 27 0% 20% 40% 60% 80% 100% Couple with no children With children younger than 5 years old With children between 5 years old and 15 years old With children older than 15 years old Couple with no children With children younger than 5 years old With children between 5 years old and 15 years old With children older than 15 years old Males Females Employed - Full-time Employed - Part-time Unemployed Inactive Source: Spanish Labor Force Survey (SLFS) - Own elaboration in Stata 14.0 as shown in the figure above. When they are also parents, the gender employment gap becomes more extensive, from 6 to 23 points. In the case of males, having children has no relation either with drops in the labor market participation or lower employment rates. However, becoming parents means that the time must be shared among one activity more, in this case, related to unpaid work. For women, we also have to take into account a crucial moment that pauses their labor careers if they were workers: the childbirth. The negative relation between fertility and females labor supply was theoretically established by Becker and Lewis (1973) [6] and empirically documented by Mincer (1985) [18], among other authors. Our results are also consistent with the literature, and we find this negative relation on the SLFS data of 2019, as we can see in Figure 4. Those women with children participate 17 points less than those childless so, this supports the existence of a motherhood penalty13 in terms of employment level comparing to those childless women. Part-time schedules are prevalent among those women with children as an instrument to conciliate family and work. From 67% of mothers who participate in the labor market, only 72% will do it through a full-time schedule. This correlation has been changing over time, and changes in social norms related to working mothers and against the stereotype of stay at home mums have improved it. Although the decision to have children is assumed to be equally distributed among parents, the decision of reconciliation is not. Family commitments are the main factor by 13Term coined by sociologists who argued that mothers are penalized, systematically, in pay, perceived competence and, benefits relative to childless women in their workplace. 21 which women are over-represented in part-time jobs and reduce working hours. Figure 6 indicates that 42% of women are involuntary part-time workers, and 36% have declared that they are not working full-time because of caring activities14. Figure 6: Reasons of holding a part-time job among couples 8.1% 2.3% 20% 62% 7.5% 36% 9.3% 42% 12% Male Female Family commitments Training and teaching activities Other reasons Did not find a full-time job Not wanting a full-time job Source: Spanish Labor Force Survey (SLFS) - Own elaboration in Stata 14.0 Also, with the arrival of children, many of them leave their jobs, becoming inactive population. Related to this, we have to remark that the inactivity level of those women with children is much higher compared with the rest of the categories. As children become teenagers, women increase their participation level, trying to recover all the proportion that they have just lost. This is a challenging process, and only 5 p.p. of women go back to employment compared to those who have younger kids. This cohort’s age explains the level of inactivity of women with children older than 15 years old. The median women of this cohort were born in the year 1967, where discriminating gender norms and culture assumed as incompatible the role of mother and worker. When the individuals are neither cohabiting nor having children, the differences in terms of employment level are negligible. We have shown in Figure 3 that Singles living with their parents, with others or a One-person family are not related to a lower level of employment whether they are females or males, but partial-time jobs and temporary contracts are a matter of women, with differences that are around 10 points with respect 14This includes caring sick, disabled, elderly children and/or adults 22 to males. Among unpartnered individuals, 5.7% of employed males are working part-time while the same situation for females increases up to 17.2%. Figure 7: Children, work schedule and type of contract by gender 74 3 19 4 75 2 20 3 81 1 16 2 84 1 14 1 57 10 25 9 56 19 15 10 58 19 15 9 63 19 12 6 0% 20% 40% 60% 80% 100% Couple with no children With children younger than 5 years old With children between 5 years old and 15 years old With children older than 15 years old Couple with no children With children younger than 5 years old With children between 5 years old and 15 years old With children older than 15 years old Males Females Indefinite - Full-time Indefinite - Part-time Temporary - Full-time Temporary - Part-time Source: Spanish Labor Force Survey (SLFS) - Own elaboration in Stata 14.0 Having children is negatively related to the level of temporary and part-time contracts that males held, so we cannot observe any adjustment in their labor supply when they become fathers. Therefore, women are more exposed than men to job insecurity since the beginning of their labor careers, as we can observe in Figure 7, with presence in parttime jobs through both, indefinite and temporary contracts. The type of contract is one of the significant factors that determine job quality [8]. Some authors have suggested that part-time and temporary work serves as a transition out mechanism from unemployment to a regular job in the future. Nevertheless, in Spain, temporary employment seems not to be so ’temporary’, representing an obstacle to reach permanent jobs, leaving workers with low prospects of improvement in their career paths [22]. Temporary contracts mean uncertainty to workers, and this labor incertitude is more related to young people and with women. Looking deeper into these differences that seem to be attached to women, we find that it could be related to occupational gender-based segregation. All the occupations are presented in Figure 8, desegregated in 9 high-level aggregations15. We can identify those gendered occupations, which are the ones that most people relate to females or males. 15It have been made by aggregation of the corresponding 1-digit occupations, based on the classification of occupations, CNO11. We have also used to support the Standard Occupational Classification and Coding Structure done by the U.S. Bureau of Labor Statistics. See more details: https://www.bls.gov/ soc/soc_2010_class_and_coding_structure.pdf 23 Figure 8: Occupational sex segregation, work schedule and gender 1 7 4 88 1 9 2 88 2 10 4 85 3 28 1 68 7 27 4 61 23 27 5 45 9 46 3 42 12 52 3 33 22 42 5 31 0% 20% 40% 60% 80% 100% Agriculture and manufacturing skilled workers Protection and security services workers Machine operators Directors and management Support profesionals Elementary occupations Scientific, intelectual technicians and profesionals Accounting, administrative, and office employees Customer interaction and other service-oriented occupations Female Part-time Female Full-time Male Part-time Male Full-time Source: Spanish Labor Force Survey (SLFS) - Own elaboration in Stata 14.0 Analyzing all individuals who belong to the target population from both sub-samples, we can clearly observe an asymmetry in the distribution by occupations that are concerned with sex segregation. We denote by ”gender-dominated” occupations, following Anker (1998) [2], as those occupations were, at least, 80% of workers are either women or men. Another way to determine them is to include all those occupations where females’ presence is lower than their participation in the labor force. 47% of the workers in our sample and target population are females. So, it can be the case that, in a hypothetical situation with no occupational segregation, we might expect about 47% of workers in every occupation to be women16. It can be determined as male-dominated occupation agriculture and manufacturing skilled workers, protection and security services, and machine operators. In the same way, we can say that customer interaction and other service-oriented occupations and accounting, administrative and office employees are female-dominated occupations. We can also distinguish gender inequality in the distribution of workers by sector of activity, also called sectorial gender segregation (see Figure A.4). This shows us that we must take into account the actual differences by gender in the Spanish labor market, to be able to obtain consistent results. This fact is crucial for being related to the females’ career and promotion path, also called pipeline problems17. Wolfinger, Manson, and Goulden 16See more details: https://inequality.stanford.edu/sites/default/files/Pathways_SOTU_ 2018_occupational-segregation.pdf 17The pipeline theory suggests that increasing the number of women in male-dominated fields should lead to more equality in the labor market. 24 (2007) [16] found that marriage and children adversely affect the likelihood that women obtain tenure-track positions in Academia. Nevertheless, this also happens in other sectors where promotion is needed to obtain better job conditions through increments in salary, participating in training programs, increasing health insurance coverage, etc. This has nothing to do with educational heterogeneity since it has been tested that women have higher levels of education than their male counterparts. Although public sector jobs help women’s employability, it increases at the same time the level of occupational sex segregation [17]. In Spain, jobs in the public sector are well remunerated, and they offer the possibility to reduce working hours. One drawback of this could be the under-representation of women in other sectors. On the other side, although the expansion of the service sector has benefited women by providing increased job opportunities [13], it is a sector with a high incidence of job insecurity in Spain (part-time work schedules and high level of temporary contracts). An important remark is that work schedules are highly related to the type of sector in which the individual works. We cannot declare any causal relationship but, which is a fact is that part-time schedules do not appear on male-dominated industries, whether they are highly related to those female-dominated. Furthermore, as Anker et al. (2003) [3] states, occupational sex segregation harms the efficiency and flexibility of the labor markets so, its consequences go further the gender inequality in the labor markets. 5 Estimation results Before presenting the results of the estimations that we have proposed in Section 3, we will clarify what we can find in those estimations and how we will analyze those results. First, to estimate the relationship between cohabitation and the probability of being employed, we work with the first sub-sample of singles and childless couples. This allows us to address the importance of the couple in the labor situation and its differences by gender. Secondly, to estimate the association between having children and the probability of being employed, having a temporary contract, working part-time and working less than 35 hours per week, we use the second sub-sample of couples, with and without children. In this case, we can measure the gender differences within couples, if there exist any, in terms of employment. With these estimations, we can capture how having children can be negatively or positively related to the parents’ labor situation. For an easier interpretation of the coefficients from the probit model, marginal effects are reported. First, we analyze if there is any difference among those individuals without family commitments (in terms of childbearing and nurture) to test our first hypothesis related to cohabiting and being employed by gender. Is cohabitation associated with 25 Figure 9: Children and labor situation by gender - Second quarter of 2007 91 2 4 2 92 2 3 3 92 1 3 4 84 1 3 12 72 11 7 10 40 18 7 35 45 17 7 31 38 12 5 44 0% 20% 40% 60% 80% 100% Couple with no children With children younger than 5 years old With children between 5 years old and 15 years old With children older than 15 years old Couple with no children With children younger than 5 years old With children between 5 years old and 15 years old With children older than 15 years old Males Females Employed - Full-time Employed - Part-time Unemployed Inactive Source: Spanish Labor Force Survey (SLFS) - Own elaboration in Stata 14.0 We can differentiate between these levels of inactivity two relations. On the one hand, having children was related to higher inactivity due to norms, culture, or stereotypes, considering that that inactivity did not fit with the economic boom of 2007, where very high employment rates characterized males and childless females’ labor supply. On the other hand, age is an important factor in the level of inactivity. The average age of those women with children older than 15 years old is 43 years old. As they were born during the 1960s, they could adopt the role of stay at home mums. They presented 17 points more of inactivity than those mothers in 2019. We estimate those gender differences on the access to employment using Equation 5 with the data from the second quarter of 2007. In the same way, we apply the two-step procedure that we have shown in Section 3.2 to quantify those penalties in terms of holding a temporary contract, working part-time, and working less than 35 weekly hours. We control for the same number of covariates21 to obtain the differences in employment levels and labor conditions among couples, from a gender perspective. The results of those estimations are displayed in Table 5. Column [1] reveals that having children in 2007 is negatively associated with the probability of being employed but only in women’s cases. This means that in a family with children, the woman was less likely to be employed in 2007, compared to their spouses or partners, than they were in 2019. 21Socio-demographic characteristics, job attributes, probability of being salaried worker, and type of contract depending on the estimated model. 32 Table 5: Marginal Effects from estimating Equations (5), (7), (8), and (9) for the second quarter of 2007 Model 6 Model 7 Model 8 Model 9 Employed Temporary Part-time Less than 35 hours/week Female (ref. Male) -0.177*** 0.0498*** 0.0440*** 0.0499*** (0.000645) (0.000811) (0.000483) (0.000571) Children (ref. Without children) 0.00458*** -0.0627*** -0.00885*** 0.00691*** (0.000582) (0.000577) (0.000399) (0.000439) Female and children (ref. Male without children) -0.169*** -0.0371*** 0.0463*** 0.0252*** (0.000747) (0.00109) (0.000770) (0.000808) Immigrant (ref.native) -0.0711*** 0.207*** -0.0130*** -0.0101*** (0.000388) (0.000591) (0.000197) (0.000290) Age (ref. 46-60 years old) Less than 29 years old 0.0505*** 0.250*** 0.0290*** 0.0368*** (0.000379) (0.000839) (0.000432) (0.000522) 30-45 years old 0.0784*** 0.126*** 0.0261*** 0.0297*** (0.000262) (0.000466) (0.000250) (0.000304) Education (ref. Secondary education) Lower than secondary -0.134*** 0.0112*** -0.00953*** -0.0168*** (0.000345) (0.000714) (0.000350) (0.000434) Terciary 0.124*** 0.0341*** 0.0168*** 0.0591*** (0.000235) (0.000738) (0.000426) (0.000574) Temporary job (ref. Indefinite) 0.0612*** 0.0717*** (0.000235) (0.000286) Occupations (ref. Pink collar) White collar -0.109*** -0.0452*** -0.0674*** (0.000765) (0.000157) (0.000230) Blue collar 0.102*** -0.0520*** -0.0712*** (0.000518) (0.000209) (0.000276) Grey collar -0.151*** -0.0297*** -0.0513*** (0.000634) (0.000398) (0.000446) Gold collar -0.0619*** -0.0258*** -0.0212*** (0.000451) (0.000169) (0.000245) Sector of activity (ref. Primary) Services -0.144*** 0.00988*** 0.0112*** (0.000837) (0.000499) (0.000694) Industry -0.158*** -0.0238*** -0.0429*** (0.000589) (0.000376) (0.000510) Construction 0.0797*** -0.0353*** -0.0548*** (0.00102) (0.000317) (0.000441) Public Sector (Education, Health, etc) -0.0414*** -0.0273*** -0.00699*** (0.000871) (0.000380) (0.000673) Probability of being salaried worker -0.457*** -0.152*** 0.0715*** (0.00392) (0.00198) (0.00100) Observations 14,561,199 8,821,228 8,821,228 8,416,548 Note: Regions of residence dummies are also included to control the estimation but they are not reported for the sake of simplicity. Weighted observations. Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 In particular, they had 16.9 p.p. less probability of working than her partner when both share the same socio-demographic characteristics and family commitments. This alarming 33 situation reflects that after 12 years, women only have cut off 7 points of child penalty on females’ participation. In 2019, women were 11.4 points less likely to be employed than their couples when both individuals have children, controlling for individual, sociodemographic, and job characteristics. In terms of having part-time jobs and hours worked, child penalties were already on the Spanish labor market, which worsened over time. While in 2007 women with children were 4.6 p.p. more likely to have a part-time job than their couples, in 2019 this increases 4 points more, which can be obtained by comparing the coefficient of Children in column [3] from Table 3 and Table 5 (in both cases, the coefficient is statistically different from zero). Another striking result was found looking into the evolution of the hours worked. Women had lower performance in terms of hours devoted to paid work than their partners when both have family commitments. In 2007, females were 2.5 p.p. more likely to work less than 35 hours per week, which increased 5 points in 2019. Therefore, the main conclusion reached concerning those differences between males’ and females’ labor situation over time is that, after 12 years, gender gaps persist in the Spanish labor market. Besides, we observe that the gender gap is lower in terms of access to employment and more related to norms and culture, whereas differences by gender in the participation on part-time jobs and hours worked are lower and more associated with the economic cycle. 6.2 Regional perspective We turn now to the gender employment gap discussed in the previous sections, but this time we will compare over the Spanish regions. This can be an indicator of the relative performance of women in the different regional labor markets. We address this issue from a static overview across them, and then we will analyze how the situation has changed in the last 12 years, comparing the regional results obtained in 2007 with those in 2019. There are substantial differences across regions, as it is shown in Figure 10. It has been analyzed the spatial distribution of the unemployment levels, regional disparities, and persistence. To investigate this issue further from a gender perspective, we will present the labor market situation of two different regions22: Andalusia and Basque Country. Even though both economies are very different in terms of unemployment rates, the gender gap in the employment levels is still present in each of them. This reveals that women with children are associated with lowering employment rates than those childless, whether they allocate 22We have analyzed in the same way the other 15 Autonomous Communities, and we have excluded the 2 Autonomous Cities of Ceuta and Melilla, for not having enough observations. See more details: Appendix A.1 -Figures A.7 to Figure A.21. 34 in a high-performance regional economy, which is the case of the Basque Country, or in a low one like Andalusia. Moreover, males’ labor situation is related to an improvement in their occupational levels when they become fathers. The case of Andalusia is somewhat surprising, as we can observe in Figure 10a. Although the Spanish economy was booming, in Andalusia, women’s labor situation was characterized by very high levels of inactivity and unemployment. This can support the importance of social norms, traditional roles, and stereotypes related to the inactivity level among women. Figure 10: Children and labor situation - Regional disparities 80 8 8 4 81 5 11 3 82 3 11 4 76 3 9 12 64 16 12 8 40 20 17 24 42 19 20 20 35 11 16 38 0% 20% 40% 60% 80% 100% Couple with no children With children younger than 5 years old With children between 5 years old and 15 years old With children older than 15 years old Couple with no children With children younger than 5 years old With children between 5 years old and 15 years old With children older than 15 years old Males Females Employed - Full-time Employed - Part-time Unemployed Inactive (a) Andalusia 78 13 7 1 89 5 6 1 93 2 3 1 85 2 4 9 66 21 5 8 50 26 4 20 57 24 5 14 56 17 6 21 0% 20% 40% 60% 80% 100% Couple with no children With children younger than 5 years old With children between 5 years old and 15 years old With children older than 15 years old Couple with no children With children younger than 5 years old With children between 5 years old and 15 years old With children older than 15 years old Males Females Employed - Full-time Employed - Part-time Unemployed Inactive (b) Basque Country Source: Spanish Labor Force Survey (SLFS) - Own elaboration in Stata 14.0 As Ortbals (2010) [20] states, in the late 1970s, the daily reality of women in Andalusia was rife with illiteracy and unemployment and was centered on their marriages. The 35 average female’s age represented in Figure 10a is 43 years old, which means that their labor force participation can be closely related to their mother’s roles [11]. Undoubtedly, there exists gender employment gap across all the regions, and parttime is highly related to women, as we can observe in Figures A.7 until Figure A.21. However, we can see clearly some different patterns in, for example, the case of Navarre23. The motherhood penalty, which is the difference in the employment levels of childless women compared to mothers, is more pronounced due to the starting point of the childless women: all those women in the Navarre’s sub-sample are employed, which reveals that having children is positively associated to inactivity and unemployment. This is also important in terms of working part-time. Only 13% of childless women were working part-time whereas, after having children, this number doubles. To understand the forces that lie behind those regional disparities, we quantify child penalties in the access to employment by gender among couples for each of the Spanish Autonomous Communities, applying the same methodology, and controlling for the same covariates24 (See Section 3.2). For simplicity, we present the main results of the estimations in a graph to better compare the results obtained for the two years of interest. We select the coefficients that represent the difference of the relationship between having children and the access to the employment for mothers with the same relation for their spouses or partners in terms of probabilities among regions, and we plot them in Figure 11. When this coefficient is negative, and both individuals of the couple are also parents, the probability of being employed of the woman is lower compared to those of her husband or partner, when they share similar socio-demographic characteristics as nationality, age, and educational attainment. In general, it is striking that gender differences have declined due to the probability of being employed among parents over time and regions. If we focus, for example, in the case of Basque Country, we obtain that women are less likely to be employed compared with their spouses or partner when both are parents, and these gender differences were higher in 2007. Only Cantabria and Navarre presents equal opportunities for accessing the labor market among couples, whether the individual is mother or father. Those coefficients were no statistically different from zero, which means that having children is not related to differences in employment levels by gender. Focusing on the case of Andalusia, differences among couples depending on the gender increased in 12 years. Mothers are 5 points less likely to be employed than they were in 2007, compared to fathers among couples with children. Worth mentioning is the 23See Figure A.20 24In each regression, we control by socio-demographic characteristics and job attributes, except for not introducing regional dummies. 36 positive relation between motherhood and the access to employment that we can observe in Aragon and La Rioja for the last year, 2019. In both, women had higher probabilities of having a job than their partners, comparing among a sample of couples. Figure 11: Estimated probability of being employed among parents - Differences between women and men. -24 -11 -22 -11 -19 -6.8 -17 -7.7 -16 -7.7 -15 -20 -15 -2.6 -14-14 -12-11 -12 9.7 -12 -5.5 -12 10 -10 -14 -9.9 -7.9 -3.8 2.7 -3.4 0 0 0 -25p.p. -20p.p. -13.5p.p.2 -10p.p. -6.5p.p.1 0p.p. 5p.p. 10p.p. Valencia Catalonia Murcia Basque Country Galicia Andalusia Canary Islands Castile and León Madrid Aragon Extremadura La Rioja Castilla-La Mancha Balearic Islands Asturias Navarre Cantabria 1 National Average - 2019 2 National Average - 2007 2007 2019 Source: Spanish Labor Force Survey (SLFS) - Own elaboration in Stata 14.0 We explore which can be the underlying reasons for this to occur in one of the three regions where there is no child penalty related to the employment level. One of the reasons could be the existence of gender differences in the educational attainment25. We present in Table 6 all those individuals that are living with their partners and children in Aragon for the two years of interest. It is very striking that the distribution of mothers that are employed are 9 p.p. more located in tertiary education than their spouses or partners, as we can see in column [3] from Table 6. These differences become higher when we analyze the same target individual in Aragon for 2019: woman and mother living with her husband or partner and with her child or children. They are 6 points more educated than those mothers in 12 years before and 15 points more than their couples in 2019. To sum up, one of the explanations we find resides on those educational level disparities. Those mothers are more educated (higher presence on tertiary education) than their husbands or partners, and education increases 25Other possible explanations could be the lack of observations and the presence of multicollinearity. However, we are not able to differentiate between these three. 37 Table 6: Differences in educational attainment among those couples with children employed in Aragon. 2007 2019 Males Females Gender differences Males Females Gender differences Low educated 15.4% 9.7% 5.7 p.p. 3.3% 1.9% 1.3 p.p. Medium educated 63.0% 59.6% 3.4 p.p. 73.8% 60.2% 13.6 p.p. High educated 21.6% 30.7% -9.1 p.p. 23.0% 37.9% -14.9 p.p. their opportunity costs, so if they look actively for a job, they find more chances than their couples. 7 Summary and final remarks This paper is motivated by the existence of differences in the employment rates between men and women. We test the existence of gender gaps in different labor outcomes among singles and couples. Although men and women share the same characteristics, we find that they obtain different outcomes, which is a signal of gender discrimination on the Spanish labor market. Those differences are higher if the woman is also a mother so, while women try to close the employment gap, they have to deal with the cost of motherhood in their labor careers. If we analyze a couple with children, the mother has lower probabilities of being employed than the father, when both have similar sociodemographic characteristics like nationality, age, education, and region of residence. As we have mentioned throughout this paper, this is due to differences in the time devoted to unpaid work, where we can include those activities related to a family commitment, child care, and household tasks. However, the entry to employment is not the only labor market outcome that we have analyzed and where women are under-represented. As we have mentioned in this study, one key factor contributing to a broader or a narrower gender pay gap is the existence of differences in the females’ and males’ hours worked. We have seen that women are more likely to participate in employment through less stable labor conditions, which means that they are more present on part-time and temporary contracts and have a higher likelihood of being on that unstable labor situation. With this paper, we contribute to understanding the differences between females’ labor situation in Spain compared to their partners’ ones, subject to their own characteristics. Also, descriptive analysis has shown significant disparities depending on an individual’s life cycle stage, gender, and her or his family commitments. Our results support the existence of two penalties in the labor market, in terms of employment levels and labor 38 conditions: being a woman and being a mother. We quantify them, taking into account the number of covariates to control for and changing the temporal and territorial perspective. Despite considerable gender convergence, substantial gender inequality remains in Spain, over time and regions. Males have more significant chances for entering the labor market than their partners when both have similar characteristics. Spain must boost sustainability to the pension system. Increasing the female’s participation in the labor market and transforming how this increase is negatively related to fertility rates can be the solution for its improvement. The means for this is related to stimulate female employment through mother friendly policies. It is widely known that extending the length of education and childcare services provided by the governmental institutions to children between 0 and 3 years old contributes to closing gender gaps and increasing fertility rates. Regarding this, we have found on the results of the year 2007 that whether the economy is expanding, gender gaps are still on the table. At the same time, part-time jobs were more related to women, although they shared characteristics with their spouses or partners. We cannot declare any causal relationship between being a mother and working part-time, but differences in their preferences cannot only explain the proportion of those women. Demand-side restrictions and involuntary part-time work can be the underlying reasons, as we have mentioned in this paper. Improve gender equality through non-gendered norms and roles, eradication of stereotypes, education, and culture can be the right solution for reaching it and maintain it in the long run because of the existence of the inter-generational transmission of child penalties. So, increasing males’ time devoted to household tasks and childcare can boost females’ participation in the labor market and make it persist over time. Even though our findings contribute to a better understanding of Spain’s employment situation from a gender perspective, it is not without limitations. On the one hand, although we have waves of representative individuals of the Spanish population, we do not have panel data that allows us to analyze the dynamics within the labor market, following individuals over their life cycle, giving us more precise information about their labor transitions. On the other hand, the lack of wages in the SLFS data do not allow us to estimate the inequality in earnings and the factors driving it in Spain. Also, it would be interesting to have enough information about same-sex couples to determine if gender roles have a more significant impact on the relation between children and employment. Last but not least, we propose further research on the inclusion of the elderly care activities on time devoted to unpaid work. As gender differences can come from a different distribution of time devoted to unpaid work among couples, it would be interesting to include those caring activities like elderly care to the ones just analyzed here as child care. 39 References [1] Aaberge, R., Colombino, U., Del Boca, D., Ermisch, J., Francesconi, M., Pasqua, S., and Strøm, S. Women’s participation in the labor market and fertility: The effects of social policies. Women at Work. An Economic Perspective (2005), 121–153. [2] Anker, R. Gender and jobs: Sex segregation of occupations in the world. International Labour Organization, 1998. [3] Anker, R., Melkas, H., Korten, A., et al. Gender-based Occupational Segregation in the 1990s. International Labour Office, 2003. [4] Anxo, D., Flood, L., Solaz, A., Mencarini, L., Pailh´ e, A., and Tanturri, M. Gender differences in time use over the life course in france, italy, sweden, and the us. Feminist Economics 17 (07 2011), 159–195. [5] Becker, G. S. Fertility and the economy. Journal of Population Economics 5, 3 (1992), 185–201. [6] Becker, G. S., and Lewis, H. G. On the interaction between the quantity and quality of children. Journal of political Economy 81, 2, Part 2 (1973), S279–S288. [7] Boeri, T., Del Boca, D., and Pissarides, C. Women at work: an economic perspective: a report for the fondazione rodolfo debenedetti, 2005. [8] Conde-Ruiz, J. I., and Marra de Artinano, I. Gender gaps in the spanish labor market. Estudios sobre la Econom´ıa Espa˜nola 32 (2016), 1–103. [9] Davia, M. A., and Legazpe, N. The role of education in fertility and female employment in spain: A simultaneous approach. Journal of Family Issues 35, 14 (2014), 1898–1925. [10] De la Rica, S., and Gorj´ on, L. The impact of family-friendly policies in spain and their use throughout the business cycle. IZA journal of European labor studies 5, 1 (2016), 9. [11] Farr´ e, L., and Vella, F. The intergenerational transmission of gender role attitudes and its implications for female labour force participation. Economica 80, 318 (2013), 219–247. [12] Heckman, J. J. Sample selection bias as a specification error. Econometrica: Journal of the econometric society (1979), 153–161. 40 [13] ILO.World employment report 2001: life at work in the information economy. International Labour Organization - Geneva, 2001. [14] Kleven, H., Landais, C., Posch, J., Steinhauer, A., and Zweim¨ uller, J. Child penalties across countries: Evidence and explanations. In AEA Papers and Proceedings (2019), vol. 109, pp. 122–26. [15] Kleven, H., Landais, C., and Søgaard, J. E. Children and gender inequality: Evidence from denmark. American Economic Journal: Applied Economics 11, 4 (2019), 181–209. [16] Marry, C., and Pochic, S. The “glass ceiling” in france: is the public sector more egalitarian than the private sector? Cadernos de Pesquisa 47, 163 (2017), 148–167. [17] Melkas, H., and Anker, R. Occupational segregation by sex in nordic countries: An empirical investigation. Int’l Lab. Rev. 136 (1997), 341. [18] Mincer, J. Intercountry comparisons of labor force trends and of related developments: an overview. Journal of labor economics 3, 1, Part 2 (1985), S1–S32. [19] OECD.OECD Employment Outlook: June 1999. Organisation for Economic Cooperation and Development, 1999. [20] Ortbals, C. D. The potential of local women’s associations in andalusia: Pursuing culture, enriching lives and constructing equality. South European Society and Politics 15, 2 (2010), 203–223. [21] Pascal, A. Report on the gender initiative: Gender equality in education, employment and entrepreneurship, 2011. [22] Petrongolo, B. Gender segregation in employment contracts. Journal of the European Economic Association 2, 2-3 (2004), 331–345. [23] Rapoport, B., Sofer, C., and Solaz, A. Household production in a collective model: some new results. Journal of Population Economics 24, 1 (2011), 23–45. [24] Rizavi, S. S., and Sofer, C. Household division of labor: Is there any escape from traditional gender roles? [25] Uunk, W., Kalmijn, M., and Muffels, R. The impact of young children on women’s labour supply: a reassessment of institutional effects in europe. Acta sociologica 48, 1 (2005), 41–62. 41 87 4 4 5 85 5 5 4 88 2 7 4 80 2 8 10 64 13 15 8 43 15 11 30 47 20 11 23 44 16 11 29 0% 20% 40% 60% 80% 100% Couple with no children With children younger than 5 years old With children between 5 years old and 15 years old With children older than 15 years old Couple with no children With children younger than 5 years old With children between 5 years old and 15 years old With children older than 15 years old Males Females Employed - Full-time Employed - Part-time Unemployed Inactive Figure A.19: Murcia 88 11 1 89 6 5 1 86 3 6 5 86 2 1 11 87 13 51 25 6 19 46 29 12 13 55 21 1 23 0% 20% 40% 60% 80% 100% Couple with no children With children younger than 5 years old With children between 5 years old and 15 years old With children older than 15 years old Couple with no children With children younger than 5 years old With children between 5 years old and 15 years old With children older than 15 years old Males Females Employed - Full-time Employed - Part-time Unemployed Inactive Figure A.20: Navarra 93 5 1 89 2 8 1 88 5 4 3 87 0 4 9 80 9 7 4 43 30 8 19 50 30 6 14 53 22 7 18 0% 20% 40% 60% 80% 100% Couple with no children With children younger than 5 years old With children between 5 years old and 15 years old With children older than 15 years old Couple with no children With children younger than 5 years old With children between 5 years old and 15 years old With children older than 15 years old Males Females Employed - Full-time Employed - Part-time Unemployed Inactive Figure A.21: La Rioja 48 A.2 Other table results Table A.2.1: Results of probit models - First sub-sample Model 1 Employed Female (ref. Male) -0.00683*** (0.000577) Couple (ref. Single) 0.0572*** (0.000507) Female and couple (ref. Male and single) -0.0891*** (0.000938) Immigrant (ref.native) -0.0583*** (0.000527) Age (ref. 46-60 years old) Less than 29 years old -0.0862*** (0.000988) 30-45 years old 0.0143*** (0.000392) Education (ref. Secondary education) Lower than secondary -0.113*** (0.00111) Terciary 0.0791*** (0.000379) Regions (ref. Andalucia) Aragon 0.0599*** (0.000799) Asturias 0.0375*** (0.00114) Balearic Islands 0.0499*** (0.000858) Canary Islands 0.0115*** (0.000818) Cantabria 0.0545*** (0.00132) Castilla y La Mancha 0.0419*** (0.000803) Castilla y Le´on 0.0441*** (0.000752) Catalonia 0.0446*** (0.000523) Valencia 0.0272*** (0.000636) Extremadura -0.0252*** (0.00147) Galicia 0.0313*** (0.000788) Madrid 0.0608*** (0.000505) Murcia 0.0324*** (0.000948) Navarra 0.0909*** (0.000853) Basque Country 0.0470*** Continued on the next page. . . 49 Model 1 Employed (0.000755) La Rioja 0.0700*** (0.00137) Ceuta and Melilla 0.0316*** (0.00300) Observations 3,346,713 Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 50 Table A.2.2: Results of probit models - Second sub-sample - 2019 Model 2 Model 3 Model 4 Model 5 Employed Temporary Part-time Less than 35 hours/week Female (ref. Male) -0.105*** 0.0680*** 0.0604*** 0.0639*** (0.000666) (0.000748) (0.000575) (0.000647) Children (ref. Without children) 0.000341 -0.0243*** -0.0512*** -0.0469*** (0.000548) (0.000554) (0.000550) (0.000585) Female and children (ref. Male without children) -0.114*** -0.0820*** 0.0860*** 0.0713*** (0.000737) (0.000841) (0.000775) (0.000828) Immigrant (ref.native) -0.131*** 0.0545*** -0.00409*** -0.0209*** (0.000382) (0.000777) (0.000511) (0.000534) Age (ref. 46-60 years old) Less than 29 years old -0.0224*** 0.242*** 0.0389*** 0.0575*** (0.000527) (0.000877) (0.000549) (0.000653) 30-45 years old 0.0665*** 0.112*** 0.0412*** 0.0617*** (0.000239) (0.000421) (0.000327) (0.000364) Education (ref. Secondary education) Lower than secondary -0.169*** -0.00442*** -0.0407*** -0.0656*** (0.000535) (0.000951) (0.000495) (0.000506) Terciary 0.133*** 0.0131*** 0.00982*** 0.0420*** (0.000214) (0.000727) (0.000574) (0.000673) Temporary job (ref. Indefinite) 0.0831*** 0.0877*** (0.000300) (0.000335) Occupations (ref. Pink collar) White collar -0.0604*** -0.0648*** -0.0907*** (0.000733) (0.000326) (0.000347) Blue collar 0.125*** 0.0172*** 0.0152*** (0.000425) (0.000266) (0.000311) Grey collar -0.115*** -0.0524*** -0.0559*** (0.000524) (0.000484) (0.000586) Gold collar -0.0209*** -0.0284*** -0.0379*** (0.000423) (0.000262) (0.000305) Sector of activity (ref. Primary) Services -0.170*** 0.112*** 0.129*** (0.000703) (0.000680) (0.000791) Industry -0.171*** -0.00665*** -0.00977*** (0.000410) (0.000673) (0.000793) Construction -0.0274*** 0.0516*** 0.0413*** (0.000749) (0.00105) (0.00112) Public Sector (Education, Health, etc) -0.0282*** 0.0774*** 0.112*** (0.000736) (0.000872) (0.00103) Probability of being salaried worker -0.375*** -0.236*** 0.0316*** (0.00413) (0.00309) (0.00102) Regions (ref. Andalusia) Aragon 0.104*** -0.0492*** 0.0239*** -0.00764*** (0.000433) -0.000841 -0.000894 -0.000876 Asturias 0.0311*** -0.0374*** 3.03E-05 -0.0330*** (0.000756) -0.000942 -0.000776 -0.000708 Balearic Islands 0.0931*** -0.0436*** -0.0358*** -0.0598*** (0.000466) -0.000856 -0.000544 -0.000398 Canary Islands 0.0192*** -0.0477*** -0.0472*** 0.0509*** (0.000503) -0.000605 -0.000328 -0.00136 Cantabria 0.0694*** -0.0321*** 0.0280*** 0.0142*** (0.000772) -0.00117 -0.00113 -0.000666 Castilla y La Mancha 0.0418*** -0.0644*** 0.0106*** 0.0486*** Continued on the next page. . . 51 Model 2 Model 3 Model 4 Model 5 Employed Temporary Part-time Less than 35 hours/week (0.000464) -0.000575 -0.000577 -0.000818 Castile and Le´on 0.0729*** -0.0439*** 0.0402*** 0.0325*** (0.000418) -0.000674 -0.000733 -0.000663 Catalonia 0.100*** -0.0746*** 0.0137*** 0.0245*** (0.000272) -0.00056 -0.000567 -0.000573 Valencia 0.0628*** -0.0244*** 0.0214*** 0.0288*** (0.000326) -0.000559 -0.00051 -0.000904 Extremadura 0.0104*** -0.0448*** 0.0302*** 0.0167*** (0.000682) -0.000796 -0.000813 -0.000704 Galicia 0.0750*** -0.0281*** 0.0131*** 0.00208*** (0.000400) -0.000686 -0.000615 -0.000574 Madrid 0.0844*** -0.0896*** -0.00673*** 0.00028 (0.000295) -0.000514 -0.000489 -0.000721 Murcia 0.0500*** -0.0366*** 0.000331 0.104*** (0.000508) -0.000722 -0.000616 -0.00153 Navarra 0.102*** -0.0169*** 0.0870*** 0.0806*** (0.000593) -0.00119 -0.0014 -0.000994 Basque Country 0.0948*** -0.00443*** 0.0613*** 0.130*** (0.000391) -0.000841 -0.000892 -0.00213 La Rioja 0.104*** -0.0289*** 0.0804*** -0.0416*** (0.000802) -0.00152 -0.00181 -0.00154 Ceuta and Melilla -0.0323*** -0.0121*** -0.0346*** -0.412*** (0.00181) -0.00225 -0.00131 -0.00346 Observations 13,482,918 8,825,174 8,825,174 8,652,114 Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 52 Table A.2.3: Results of probit models - Second sub-sample - 2007 Model 6 Model 7 Model 8 Model 9 Employed Temporary Part-time Less than 35 hours/week Female (ref. Male) -0.177*** 0.0498*** 0.0440*** 0.0499*** (0.000645) (0.000811) (0.000483) (0.000571) Children (ref. Without children) 0.00458*** -0.0627*** -0.00885*** 0.00691*** (0.000582) (0.000577) (0.000399) (0.000439) Female and children (ref. Male without children) -0.169*** -0.0371*** 0.0463*** 0.0252*** (0.000747) (0.00109) (0.000770) (0.000808) Immigrant (ref.native) -0.0711*** 0.207*** -0.0130*** -0.0101*** (0.000388) (0.000591) (0.000197) (0.000290) Age (ref. 46-60 years old) Less than 29 years old 0.0505*** 0.250*** 0.0290*** 0.0368*** (0.000379) (0.000839) (0.000432) (0.000522) 30-45 years old 0.0784*** 0.126*** 0.0261*** 0.0297*** (0.000262) (0.000466) (0.000250) (0.000304) Education (ref. Secondary education) Lower than secondary -0.134*** 0.0112*** -0.00953*** -0.0168*** (0.000345) (0.000714) (0.000350) (0.000434) Terciary 0.124*** 0.0341*** 0.0168*** 0.0591*** (0.000235) (0.000738) (0.000426) (0.000574) Temporary job (ref. Indefinite) 0.0612*** 0.0717*** (0.000235) (0.000286) Occupations (ref. Pink collar) White collar -0.109*** -0.0452*** -0.0674*** (0.000765) (0.000157) (0.000230) Blue collar 0.102*** -0.0520*** -0.0712*** (0.000518) (0.000209) (0.000276) Grey collar -0.151*** -0.0297*** -0.0513*** (0.000634) (0.000398) (0.000446) Gold collar -0.0619*** -0.0258*** -0.0212*** (0.000451) (0.000169) (0.000245) Sector of activity (ref. Primary) Services -0.144*** 0.00988*** 0.0112*** (0.000837) (0.000499) (0.000694) Industry -0.158*** -0.0238*** -0.0429*** (0.000589) (0.000376) (0.000510) Construction 0.0797*** -0.0353*** -0.0548*** (0.00102) (0.000317) (0.000441) Public Sector (Education, Health, etc) -0.0414*** -0.0273*** -0.00699*** (0.000871) (0.000380) (0.000673) Probability of being salaried worker -0.457*** -0.152*** 0.0715*** (0.00392) (0.00198) (0.00100) Regions (ref. Andalucia) Aragon 0.109*** -0.0780*** 0.0504*** 0.00384*** (0.000451) (0.000786) (0.000865) (0.000707) Asturias -0.00134* -0.0911*** -0.00345*** 0.0125*** (0.000752) (0.000760) (0.000504) (0.000765) Balearic Islands 0.106*** -0.0844*** 0.00536*** -0.0222*** (0.000475) (0.000777) (0.000592) (0.000433) Canary Islands 0.0512*** -0.0434*** -0.00891*** -0.00709*** (0.000488) (0.000684) (0.000358) (0.000849) Cantabria 0.0401*** -0.0812*** -0.00385*** 0.0260*** (0.000878) (0.00103) (0.000660) (0.000624) Castilla y La Mancha 0.0372*** -0.0669*** 0.0179*** 0.0249*** Continued on the next page. . . 53 Model 6 Model 7 Model 8 Model 9 Employed Temporary Part-time Less than 35 hours/week (0.000496) (0.000618) (0.000494) (0.000581) Castilla y Le´on 0.0476*** -0.0941*** 0.0327*** 0.0316*** (0.000456) (0.000542) (0.000528) (0.000525) Catalonia 0.102*** -0.121*** 0.0338*** 0.0389*** (0.000283) (0.000495) (0.000486) (0.000503) Valencia 0.0605*** -0.0715*** 0.0374*** 0.0276*** (0.000342) (0.000487) (0.000441) (0.000818) Extremadura -0.00367*** -0.0155*** 0.0242*** 0.0347*** (0.000731) (0.000946) (0.000668) (0.000610) Galicia 0.0706*** -0.0567*** 0.00992*** 0.0105*** (0.000408) (0.000611) (0.000431) (0.000450) Madrid 0.0798*** -0.107*** 0.0232*** 0.0280*** (0.000316) (0.000479) (0.000424) (0.000730) Murcia 0.0532*** -0.0357*** 0.0140*** 0.0928*** (0.000541) (0.000784) (0.000551) (0.00139) Navarra 0.0944*** -0.0589*** 0.0922*** 0.0845*** (0.000685) (0.00110) (0.00132) (0.000810) Basque Country 0.0702*** -0.0626*** 0.0635*** 0.0719*** (0.000446) (0.000665) (0.000696) (0.00171) La Rioja 0.0730*** -0.111*** 0.0644*** 0.00923*** (0.000986) (0.00113) (0.00149) (0.00194) Ceuta and Melilla -0.0910*** -0.0794*** 0.0315*** -0.264*** (0.00227) (0.00213) (0.00190) (0.00236) Observations 14,561,199 8,821,228 8,821,228 8,416,548 Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 54 Table A.2.4: Results of probit models by regions - Second sub-sample - 2007 Andalusia Aragon Asturias Balearic Islands Canary Islands Cantabria Castilla-La Mancha Castile and Le´on Catalonia Female (ref. Male) -0.285*** -0.123*** -0.264*** -0.148*** -0.196*** -0.352*** -0.277*** -0.304*** -0.0661*** (0.00170) (0.00371) (0.00522) (0.00306) (0.00344) (0.00595) (0.00288) (0.00366) (0.00128) Children (ref. Without children) -0.00266* -0.0427*** -0.158*** 0.0352*** 0.000621 -0.00531 -0.00514* -0.0595*** 0.0720*** (0.00157) (0.00273) (0.00358) (0.00305) (0.00300) (0.00592) (0.00272) (0.00306) (0.00131) Female and children (ref. Male without children) -0.151*** -0.119*** -0.0375*** -0.0995*** -0.149*** 0.00297 -0.102*** -0.136*** -0.221*** (0.00197) (0.00438) (0.00597) (0.00369) (0.00390) (0.00702) (0.00339) (0.00430) (0.00158) Immigrant (ref.Native) -0.00112 0.00194 0.0656*** -0.0939*** 0.0370*** -0.0845*** -0.0855*** -0.0994*** -0.119*** (0.00112) (0.00162) (0.00363) (0.00172) (0.00153) (0.00507) (0.00224) (0.00203) (0.000810) Age (ref. 46-60 years old) Less than 29 years old 0.0700*** 0.0108*** 0.0423*** 0.0530*** 0.0911*** 0.109*** 0.0754*** 0.0296*** 0.0550*** (0.00101) (0.00214) (0.00325) (0.00191) (0.00168) (0.00302) (0.00175) (0.00208) (0.000732) 30-45 years old 0.102*** 0.0846*** 0.122*** 0.0118*** 0.100*** 0.0748*** 0.0425*** 0.0575*** 0.0849*** (0.000697) (0.00127) (0.00191) (0.00157) (0.00137) (0.00239) (0.00116) (0.00135) (0.000591) Education (ref. Secondary education) Lower than secondary -0.155*** -0.113*** -0.179*** -0.151*** -0.142*** -0.110*** -0.143*** -0.125*** -0.0919*** (0.000788) (0.00190) (0.00260) (0.00224) (0.00160) (0.00352) (0.00154) (0.00154) (0.000756) Terciary 0.188*** 0.0869*** 0.149*** 0.0175*** 0.0815*** 0.167*** 0.140*** 0.196*** 0.107*** (0.000638) (0.00106) (0.00191) (0.00179) (0.00143) (0.00198) (0.00100) (0.00102) (0.000482) Observations 2,710,909 404,590 308,815 351,820 591,928 173,731 750,222 652,994 2,411,603 55 Valencia Extremadura Galicia Madrid Murcia Navarre Basque Country La Rioja Female (ref. Male) -0.125*** -0.282*** -0.126*** -0.166*** -0.219*** -0.267*** -0.135*** -0.249*** (0.00187) (0.00599) (0.00299) (0.00162) (0.00421) (0.00541) (0.00269) (0.00827) Children (ref. Without children) 0.0150*** -0.113*** -0.00521** -0.0349*** 0.0175*** -0.0259*** 0.00714*** 0.0134 (0.00171) (0.00418) (0.00254) (0.00132) (0.00395) (0.00478) (0.00243) (0.00845) Female and children (ref. Male without children) -0.241*** -0.118*** -0.160*** -0.121*** -0.192*** -0.0340*** -0.168*** -0.117*** (0.00218) (0.00674) (0.00334) (0.00191) (0.00479) (0.00606) (0.00320) (0.00973) Immigrant (ref.Native) -0.108*** -0.259*** -0.163*** -0.0380*** -0.00192 -0.0195*** -0.115*** -0.0833*** (0.00102) (0.00517) (0.00282) (0.000838) (0.00179) (0.00278) (0.00258) (0.00409) Age (ref. 46-60 years old) Less than 29 years old 0.0670*** -0.0421*** 0.0415*** 0.0254*** 0.0245*** 0.0115*** 0.0310*** 0.0842*** (0.00108) (0.00364) (0.00177) (0.00104) (0.00216) (0.00323) (0.00200) (0.00364) 30-45 years old 0.0780*** 0.0884*** 0.0665*** 0.0620*** 0.101*** 0.0616*** 0.0832*** 0.0862*** (0.000779) (0.00185) (0.00106) (0.000642) (0.00155) (0.00193) (0.00112) (0.00314) Education (ref. Secondary education) Lower than secondary -0.143*** -0.164*** -0.113*** -0.125*** -0.157*** -0.0873*** -0.124*** -0.165*** (0.00112) (0.00247) (0.00146) (0.00100) (0.00191) (0.00291) (0.00177) (0.00408) Terciary 0.110*** 0.253*** 0.121*** 0.0865*** 0.159*** 0.0861*** 0.0717*** 0.0867*** (0.000737) (0.00143) (0.000994) (0.000561) (0.00127) (0.00157) (0.00104) (0.00282) Observations 1,630,878 340,735 812,166 1,945,723 461,666 188,052 677,448 103,723 56 Table A.2.5: Results of probit models by regions - Second sub-sample - 2019 Andalusia Aragon Asturias Balearic Islands Canary Islands Cantabria Castilla-La Mancha Castile and Le´on Catalonia Female (ref. Male) -0.285*** -0.123*** -0.264*** -0.148*** -0.196*** -0.352*** -0.352*** -0.304*** -0.0661*** (0.00170) (0.00371) (0.00522) (0.00306) (0.00344) (0.00595) (0.00595) (0.00366) (0.00128) Children (ref. Without children) -0.00266* -0.0427*** -0.158*** 0.0352*** 0.000621 -0.00531 -0.00531 -0.0595*** 0.0720*** (0.00157) (0.00273) (0.00358) (0.00305) (0.00300) (0.00592) (0.00592) (0.00306) (0.00131) Female and children (ref. Male without children) -0.151*** -0.119*** -0.0375*** -0.0995*** -0.149*** 0.00297 0.00297 -0.136*** -0.221*** (0.00197) (0.00438) (0.00597) (0.00369) (0.00390) (0.00702) (0.00702) (0.00430) (0.00158) Immigrant (ref.Native) -0.00112 0.00194 0.0656*** -0.0939*** 0.0370*** -0.0845*** -0.0845*** -0.0994*** -0.119*** (0.00112) (0.00162) (0.00363) (0.00172) (0.00153) (0.00507) (0.00507) (0.00203) (0.000810) Age (ref. 46-60 years old) Less than 29 years old 0.0700*** 0.0108*** 0.0423*** 0.0530*** 0.0911*** 0.109*** 0.109*** 0.0296*** 0.0550*** (0.00101) (0.00214) (0.00325) (0.00191) (0.00168) (0.00302) (0.00302) (0.00208) (0.000732) 30-45 years old 0.102*** 0.0846*** 0.122*** 0.0118*** 0.100*** 0.0748*** 0.0748*** 0.0575*** 0.0849*** (0.000697) (0.00127) (0.00191) (0.00157) (0.00137) (0.00239) (0.00239) (0.00135) (0.000591) Education (ref. Secondary education) Lower than secondary -0.155*** -0.113*** -0.179*** -0.151*** -0.142*** -0.110*** -0.110*** -0.125*** -0.0919*** (0.000788) (0.00190) (0.00260) (0.00224) (0.00160) (0.00352) (0.00352) (0.00154) (0.000756) Terciary 0.188*** 0.0869*** 0.149*** 0.0175*** 0.0815*** 0.167*** 0.167*** 0.196*** 0.107*** (0.000638) (0.00106) (0.00191) (0.00179) (0.00143) (0.00198) (0.00198) (0.00102) (0.000482) Observations 2,710,909 404,590 308,815 351,820 591,928 173,731 173,731 652,994 2,411,603 57