childcare in urban China 12-2010 - Economics - The University of ...

6 downloads 45 Views 268KB Size Report
unearned income (Hallman et al. 2005). The mother's labor force participation increases with wages and decreases with the price of each type of childcare and ...
Women’s Labor Force Participation and Childcare Choices in Urban China during the Economic Transition

Fenglian Du and Xiao-yuan Dong

Department of Economics Working Paper Number: 2010-04

THE UNIVERSITY OF WINNIPEG Department of Economics 515 Portage Avenue Winnipeg, R3B 2E9 Canada

This working paper is available for download from: http://ideas.repec.org/s/win/winwop.html

Women’s Labor Force Participation and Childcare Choices in Urban China During the Economic Transition* Fenglian Du Inner Mongolia University Xiao-yuan Dong Corresponding Author Department of Economics, University of Winnipeg Winnipeg, MB Canada R3B 2E9 Phone: 204 786 9307 Fax: 204 774 4134 Email: [email protected] October 2010

*

We are grateful for the comments and suggestions of the editor and two anonymous referees. We also thank the Heinrich Böll Foundation Beijing Office for its financial support for the project entitled “Care for Children and Elders and Its Impact on Women in China” and the Ford Foundation for its support for the Economic Research and Training Program for young Chinese women economists. We also wish to thank Hongqin Chang for her assistance in organizing data.

1

Abstract: China’s transition from a centrally planned to a market economy has substantially eroded governmental support for childcare. This paper examines the labor force participation and childcare choices of urban Chinese women during the economic transition and explores the distributional implications of childcare reform. The analysis shows that following child care reform, access to informal caregivers became increasingly critical for women’s labor force participation. The rise of women’s dependence on informal caregivers apparently accounted for much of the decline in women’s labor force participation during the period from 1997 to 2006. In effect, child care reform heightened the tensions between income earning and child rearing for women who had no access to informal care providers and also could not afford to use formal care services.

JEL Classification code: H31; J13; J16; J22

2

1. Introduction The dramatic increase in the labor force participation of women with young children over recent decades has made child care an issue of considerable policy importance. It is widely appreciated that the availability of affordable out-of-home childcare options increases the likelihood that mothers will participate in the labor force and is therefore critical to gender equality in access to paid work and income security. Further, access to high-quality care and education for preschool children is imperative for early childhood development. Government support for and involvement in child care are deemed essential to providing equitable access to quality childcare and thereby modifying the effects of socioeconomic and gender-related inequities because private markets for child care tend to create cost barriers for low-income families (Gornick and Meyers 2003). It is, thus, not surprising that the governments of OECD countries, on average, spent 0.7 percent of GDP on childcare and pre-primary school programs in 1999 (Jaumotte 2004). In the developing world, the number of publicly subsidized or funded early childhood care and education (ECCE) programs has been steadily increasing with the growing awareness that childcare and early interventions are a powerful mechanism for alleviating poverty and overcoming future inequalities.1 Similar to socialist countries in Central and East Europe and the Soviet Union, China established a public childcare system under central planning. The public childcare system played an important role in Chinese women’s high labor force participation. However, the transition from a centrally planned to a market economy that was launched in 1978 has led to a dramatic decline in publicly funded childcare programs. This decline has adverse implications for women and children, particularly those in low-income

3

families. Although the deterioration of women’s position in the labor market following the economic transition in urban China is a well-known phenomenon,2 the role that childcare provision has played in that phenomenon is much less well understood. This paper aims to fill in this knowledge gap by examining the labor force participation and childcare choices of women with pre-school aged children in urban China using data from the China’s Health and Nutrition Survey (CHNS) for the period from 1991 to 2006. A large body of research has emerged that investigates the interrelationship between childcare and women’s labor force participation in developed market economies.3 However, the literature on childcare in developing and transition countries is relatively thin,4 and empirical studies of childcare and women’s labor force participation in China have just begun to emerge. Kilburn and Datar (2002) estimate the impact of the availability of childcare centers using data from the CHNS for the period 1991 to 1997. Using difference-in-differences and bivariate probit regression techniques, they find that the presence of childcare centers in a community has a positive effect on women’s labor force participation and the use of childcare. Du (2008) explores the relations between informal childcare arrangements and Chinese women’s employment. She presents evidence that co-residence with grandparents or having grandparents living nearby increases the labor force participation of women with young children. Maurer-Fazio et al. (2009) find that the labor force participation rate of urban women with preschool-age children fell dramatically between 1990 and 2000, and co-residence with grandparents became increasingly important for women to stay in the labor market. This paper makes two specific contributions to the literature on women’s work and childcare in China. First, the paper provides an overview of China’s childcare

4

reforms using information obtained from official documents, Chinese literature, and the authors’ field studies. Such background information, which has thus far been unavailable to existing studies, highlights the institutional context in which women craft their solutions to the conflicts between their paid work and child rearing responsibilities. Second, the paper examines the impact of the childcare reform on women’s labor force participation and childcare choices and explores its distributional implications. Our analysis obtains evidence that childcare reform has weakened the role of formal childcare programs in meeting working mothers’ need for childcare and has increased women’s dependence on informal care providers for labor force participation. In effect, childcare reform has heightened the tensions between income earning and child rearing for socioeconomically disadvantaged women who have no access to informal care providers and also cannot afford to use formal care services. The remainder of the paper is organized as follows. The next section provides an overview of economic reforms and childcare policy in China. Section 3 frames the research questions and explains the empirical methodology. Section 4 describes the data, defines the variables involved in the regressions, and presents summary statistics on the labor force participation of women with young children and patterns of childcare arrangements. The empirical results are presented in Section 5. The paper concludes with a summary of the findings. 2. Economic Transition and China’s Childcare Policy In Mao’s era (1949-1976), more women participated in the labor market in China than in any other country in the developing world (Croll 1983). Almost all working-age women joined the labor force and worked on a fulltime basis after completing school.

5

Chinese women, especially urban women, could not have had such a high level of labor force participation without a publicly funded childcare system. As early as 1952, the Chinese government issued “Regulations for Kindergartens (trial basis),” which called for governments at various levels and employers to develop nurseries and kindergartens. The Regulations stipulated that the child care program had two major goals: “to educate children so their health and cognitive capabilities will be fully developed prior to primary school and to reduce the burdens of childcare on mothers so they will have time to participate in political life, productive work, and cultural and educational activities” (He and Jiang 2008, 4). Through this program, resources were devoted to a publicly funded childcare system that provided care to children from the earliest months of their lives until they entered primary school. Compared to Central and Eastern European countries and to the Soviet Union, however, the management and financing of the Chinese childcare system was more decentralized.5 Childcare services in China were distributed through four mechanisms. First, the Ministry of Education ran a small number of showcase nurseries and kindergartens. Second, childcare was administered by the place of employment, which is termed in Chinese as danwei (and which involves three different types: qiye, productive enterprises; shiye danwei, public institutions such as schools and hospitals; and guojiajiguan, government offices). The employer was the principal provider of childcare to urban families because subsidized childcare was a main component of the pre-reform welfare package that was handled by the employer. Third, neighborhood committees (jiuweihui) were also involved in the provision of childcare to accommodate the needs of those parents whose employers, such as urban collectives, had no childcare programs.

6

Fourth, in rural areas, the provision of childcare programs was primarily the responsibility of communes and brigades. One major drawback of this decentralized provisioning system was its inadequate role for redistribution. Although the pre-reform childcare system permitted a broad-based distribution of services among urban families, it created a disparity between urban and rural sectors by distributing benefits predominantly among privileged urban populations. In the post-reform period, the Chinese government’s overriding concern has been to find the most efficient way of restructuring the economy, and equality, including gender equality, has not been considered a priority. As a result, economic reform has brought about two major changes in China’s childcare policy. First, post-reform policy discourse stresses the role of center-based childcare in promoting ECCE but downplayed its role supporting working women. The decoupling of the dual functions of childcare programs was a main assertion of the 1989 Regulations for Kindergartens according to Zhu and Wang (2005). In accordance with the new regulations, publicly-subsidized childcare programs would no longer cover children aged 0 to 2 years for whom the provision of education is considered unimportant. The second and the most far-reaching change has been the substantial cutback on government and employer supports for childcare programs. To increase labor market flexibility, urban welfare reform was launched to decouple state-owned enterprises (SOEs) from the provision of social welfare and services in the mid-1990s. Welfare reform transferred enterprises’ responsibility for income security, healthcare and pensions to various newly developed social insurance programs and transferred enterprise-run facilities such as schools, hospitals, and security forces to local governments. Nurseries

7

and kindergartens were not included in the list of services to be transferred to local governments, however, because childcare was considered a family responsibility. With the accelerating pace of reforms and increasing pressure for profits, the vast majority of Chinese enterprises ceased to offer subsidized childcare to employees. According to the Chinese enterprise social responsibility survey conducted in 2006, enterprises that still ran kindergartens accounted for less than 20 percent of SOEs and only 5.7 percent of all enterprises in the sample (see Table 1). Moreover, due to the decline of government funding for social services, many kindergartens provided by neighborhood communities were closed.6 The restructuring of the public sector also adversely affected childcare programs provided by public institutions and government offices, though the impact on those non-profit public organizations was not as dramatic as on those run by SOEs and neighborhood communities. Lacking adequate funding from the government, childcare centers that remained publicly funded relied heavily on service fees to cover their operational expenses. Many publicly funded childcare centers adopted a two-tier fee scheme that charged employee users subsidized prices and non-employee users market prices.7 In the rural sector, the privatization of township and village enterprises in the late 1990s weakened the local governments’ ability to finance public childcare programs. As a result, the number of kindergartens and kindergarten enrolments fell in rural areas from 10,700 and 1.6 million, respectively, in 1995 to 5,000 and 0.9 million, respectively, in 2003.8 In the face of this dramatic decrease in publicly-funded childcare programs, private childcare centers have grown rapidly to fill in the gap since the mid-1990s. In response to the changing patterns of childcare provision, the Chinese government issued the 2001 Guidelines for Kindergarten Education, which formally

8

endorsed a pluralistic approach to childcare, with “state-run kindergartens as the backbone and exemplar” and “social forces [an ideologically convenient term for market forces] as the primary providers.” Table 2 presents the trends of kindergartens for China as a whole and by region between 1997 and 2006. During this period, the number of publicly funded kindergartens (by the Ministry of Education and other non-private organizations) decreased by 65 percent. In contrast, private kindergartens grew rapidly, with a share that rose from 13.5 percent in 1997 to 57.8 percent in 2006, and the fastest growth of private kindergartens was observed in the poorest western region. For China as a whole, the number of kindergartens decreased by 28.5 percent between 1997 and 2006. The childcare reforms have raised concerns about availability, affordability, and quality of childcare programs in China.9 Publicly funded nurseries for children aged 0-2 years have largely disappeared. The pluralistic approach to childcare provisioning in conjunction with the legacy of employer-based welfare entitlements has created a twotiered system. In this system, publicly subsidized high-quality childcare services chiefly benefit already wealthy parents who are employees of the non-profit public organizations and large SOEs that still have childcare programs.10 In contrast, other parents must rely on fee-for service private or commercialized public kindergartens to meet their needs.11 Rising inequality in access to decent childcare has thus been documented by several authors (for instance, see Jin et al. 2005; Liu et al 2008). Childcare reforms have been accompanied by sweeping changes in other areas as well. The public-sector restructuring program, which was launched in the late 1990s, brought an end to the era of lifetime employment for state employees. In the post-

9

restructuring period, women have withdrawn from the labor force in large numbers (Dong et al. 2006; Maurer-Fazio, Hughes and Zhang 2007; Ding, Dong and Li 2009). Likewise, a growing number of urban workers, predominately women, have been pushed into the informal sector where jobs are typically temporary or part-time, insecure, and low-paying (Yuan and Cook 2010). The structural change of the Chinese economy from an agrarian to an industrial one creates new obstacles for women with young children to enter the labor market. The rapid growth of industrial production and high rates of urbanization separate the workplace from the home, which in turn increases the demand for non-parental care. However, the vast majority of women whose work is affected by the ongoing urbanization process have no access to publicly subsidized childcare programs under China’s two-tiered childcare system. In the remainder of the paper, we examine how urban Chinese women in different socioeconomic groups cope with the growing tension between paid work and childcare provision during the economic transition. 3. Research Questions and Empirical Methodology Our empirical analysis is designed to reveal the impact of childcare reform on women’s labor force participation and childcare choices by addressing three specific questions. The first question is whether the reform policy of decoupling the dual functions of childcare programs has indeed weakened their ability to meet working mothers’ need for childcare. The second question is whether childcare reform has heightened women’s dependence on access to informal care providers for their participation in the labor force given the government’s exit from childcare programs for younger children and the rising costs of childcare programs following child care reform.12

10

The third question is how childcare reform affects the options of women in different socioeconomic groups to deal with the competing demands of paid work and childcare when informal childcare providers are unavailable. Generally speaking, women who have no access to informal care providers have three possible options: a. withdraw from the labor force to take care of her child; b. work while monitoring the child or leaving the child unattended; or c. work while enrolling the child in a paid care program. Given the cost barriers to childcare programs, the probability of withdrawing from the labor market or combining paid work with child rearing should be relatively high for socioeconomically disadvantaged women. By reducing the availability of affordable childcare programs, child care reform is expected to exacerbate the tensions these women face. To address the first two questions, we estimate the determinants of women’s labor force participation and use of formal child care following Hallman et al. (2005). We assume that a Chinese mother makes the decisions to enter the labor force and use formal child care jointly to maximize her utility over market goods, home-produced goods, and leisure subject to a budget constraint and the mother’s time constraint. From the optimization program, the mother’s two decisions can be derived as reduced form functions of the prices of market-purchased and home-produced childcare, wages, and unearned income (Hallman et al. 2005). The mother’s labor force participation increases with wages and decreases with the price of each type of childcare and unearned income, whereas her demand for formal childcare increases with wages, unearned income and the price of home-produced childcare and decreases with the price of formal childcare.

11

Empirically, the mother’s two choices are specified as a bivariate probit model below: y1* = Xβ 1 + ε 1 ,

y1 = 1 if y1* > 0, 0 otherwise

y 2* = Xβ 2 + ε 2 ,

y 2 = 1 if y 2* > 0, 0 otherwise

(1)

where y1* and y 2* are unobserved latent variables for labor force participation y1 and choice of formal care y2; ε1 and ε2 are the error terms that are jointly normally distributed with means of 0, variances of 1, and correlation of ρ; β1 and β2 are unknown parameters; and X is a vector of exogenous variables. Following Hallman et al. (2005), we estimate both choices using a common set of exogenous determinants. The explanatory variables in X include the mother’s education and her age and its squared terms, which are proxy variables for her wages; her husband’s education, age and its squared term; unearned income; and access to farmland. The explanatory variables also include those measuring the household composition by sex and age as well as the living arrangements of elderly parents because these factors affect the availability of nonmarket maternal care substitutes and the domestic demand for the mother’s time.13 In addition, we control for community characteristics such as the availability and prices of formal childcare programs, 14 mean wages for babysitters, mean wages for female workers, labor force participation rates of women of prime age, the size of the formal sector, and variations across provinces and macroeconomic changes over time periods. To assess the impact of childcare reform and the role of informal caregivers, we examine proxy variables for grandmothers’ accessibility, i.e., the number of females aged 51 to 69 years in the household and binary indicators of whether a woman’s mother or mother in-law lives nearby as well as their interactive terms with the dummy variable for

12

the years of 2000, 2004 and 2006. The period from 2000 to 2006 represent the postchildcare reform period because the SOE welfare reform occurred in the mid-1990s, which led to the demise of broad-based, publicly subsidized child care programs. Given that informal care is a substitute for formal care, the three stand-alone proxy variables should have a negative effect on the use of formal care and a positive effect on a woman’s labor force participation. If childcare reform has shifted the focus of formal care programs from care provision toward early education, the interactive terms for formal care use should have a positive sign because informal care only plays a custodial role while formal care programs provide both care and early education. Additionally, a significant, negative estimate of the interactive informal care variable in the labor force participation equation is indicative of women’s growing dependence on informal care following the childcare reform. To tackle the third research question, we examine the mother’s choice of four mutually exclusive options: (0) does not participate in paid work; (1) works but does not use non-parental care; (2) works and uses informal (non-parental) care; and (3) works and uses formal care.15 Option (2) is available only to mothers who have access to informal care providers, and options (0), (1) and (3) represent three possible solutions for women who have no access to informal care. Each of these options has different implications to the well-being of the mother and her child. In accordance with Blau and Robins (1988), these discrete choices can be derived as corner solutions to the mother’s optimization problem similar to one described above. Thus, the utility that the mother derives from the choice of the jth option can be expressed in linear form as: U j = Xγ j + ν j

(2)

13

where Uj is the utility for choosing option j; X is the vector of variables that affect the mother’s choice as defined previously; γ is a vector of unknown parameters; and ν is a random disturbance. The mother chooses the option that yields the highest utility. The statistical model is derived as the probability that option j is chosen, which is Pr ob(U j > U k ) for all other k ≠ j.

(3)

Let Yi be a random variable that indicates the choice made among the four options. The probability that the mother chooses option j is written as Pr ob(Y = j ) = Pr ob(U j > U k ) = Pr ob[υ j − υ k > X (γ j − γ k )] for any j ≠ k (4) Assuming that the j disturbances are independent and identically distributed with a Weibull distribution and setting option (2) as the base category, equation (4) is specified as a multinomial logit model below:

 Pj ln  P2

  = Xγ j 

(5)

where γj, j = 0, 1, and 3 measure the marginal effects of explanatory variables in X on the log-odds ratio of option j to 2.16 The estimates of γj reveal how a mother’s socioeconomic characteristics such as her own education, husband’s education, and unearned income affect the mother’s response to the conflicts of paid work and child rearing when access to informal care is unavailable. The mother’s education is expected to have a positive effect on her participation in paid work and use of formal childcare. This positive effect results from a high level of education increasing the mother’s offered wage, which not only provides her with a better incentive to enter the labor force but also enhances her ability to pay for childcare. A mother with greater educational attainment is also more likely to work in the

14

formal sector where jobs are generally inflexible in hours and locations and where it is difficult to take care of children while working (Connelly et al. 1996). Moreover, a welleducated mother is expected to pay particular attention to her children’s early education and is therefore likely to choose formal childcare over self-provided childcare. Furthermore, education improves a mother’s opportunity to work in government organizations and large SOEs, which still provide subsidized childcare programs to their employees. The effects of husband’s education and unearned income are ambiguous. A mother who is married to a husband with less education or who has less unearned income is under more economic pressure to engage in paid work but also can less afford paid child care. Therefore, she is more likely to make a difficult choice to either exit the labor force or to participate in paid work without anyone else to take care of her child. An increase in unearned income should, however, unambiguously increase the use of formal childcare when all other things remain equal. In the post-childcare reform period, access to informal care providers should become more critical for reducing a mother’s probability of non-employment and a child’s risk of being left unattended.

4 Data, Variables, Women’s Work, and Childcare Arrangements 4.1 Data Our empirical analysis uses data from the six waves (1991, 1993, 1997, 2000, 2004 and 2006) of the CHNS.17 Each iteration of the CHNS covers about 4000 households and 16,000 individuals from nine provinces, namely, Heilongjiang, Liaoning, Shandong, Henan, Jiangsu, Hubei, Hunan, Guizhou, and Guangxi. The survey provides rich socioeconomic information on the individuals, households and communities in the sample. In this paper, we use data on married women who are younger than 46 years old

15

and have a child under age 7 and husband present. We confine our analysis to urban settings (cities, suburban areas, and county centers) because the patterns of childcare and employment differ between urban areas and rural villages in developing countries. After excluding the observations with missing information, our sample consists of a total of 1,353 woman-year observations for analysis.18

4.2 Variables For the empirical analysis, we classify non-parental child care into informal and formal care. Informal care refers to care provided by resident family members (primarily grandparents), non-resident relatives, neighbors, relatives, and hired nannies, whereas formal childcare refers to center-based programs such as nurseries, kindergartens and preschools. The dependent variables of the bivariate probit model in (1), labor force participation y1 and choice of formal care y2, are defined as y1 is equal to one if a mother is currently working and zero otherwise; and y2 is equal to one if she uses formal care and zero otherwise. We next divide working mothers into three categories: those who do not use non-parental care, those who use informal care (not formal care), and those who use formal care. Combining the three categories of working mothers with non-working mothers yields the dependent variables for the multinomial logit model in (5). The explanatory variables in X are defined as follows. Parents’ education is measured by years of schooling. Unearned income is the sum of the husband’s earnings and the family’s non-labor income measured as yuan per month. Access to farmland is measured by a dummy indicator for those who have farmland. Household composition is measured by the numbers of males and females at different age intervals. The living arrangements of elderly parents are measured by four binary indicators of whether the

16

woman lives as a neighbor to her mother, father, mother in-law or father in-law who does not need care. With respect to community characteristics, the availability of formal child care is measured by two dummy variables for the presence of childcare centers for children aged 0 to 2 and aged 3 to 6 in a community. The prices of childcare for the two age groups are the median prices of childcare centers in a city measured in yuan per month. The childcare prices are derived at the city level instead of the community level because some communities have no centers. The price for non-center based care is measured by the mean wages of babysitters and nannies in a community in yuan per day. Female wages are the mean wages of female workers in a community measured in yuan per day. Female labor force participation is derived as the mean labor force participation of women aged between 19 and 50 in a community in the sample excluding observation i. The proxy for the size of the formal sector is the percentage of the labor force employed in enterprises with 20 or more employees. We also control for the degree of urbanization using two dummy variables for communities in suburban areas and county centers, and we control for regional differences and policy changes over time using provincial and year dummy variables. All monetary variables are discounted by the urban consumer price index in 2006 provided by CHNS, and all are in log form. The summary statistics of the variables for mothers’ characteristics by working status and childcare arrangements are presented in the appendix.

4.3 Women’s Work and Childcare Arrangements The data indicate that at the beginning of the period under investigation, that is, in 1991, the majority of women in cities and county centers engaged in wage-salary work in the public sector, and most women in suburban areas worked on family farms producing

17

vegetables, eggs and diary/meat products for the cities. During the sample period from 1991 to 2006, cities and county centers saw dramatic institutional changes, with a growing proportion of the labor force moving into the private and informal sectors. In contrast, suburban areas were at the forefront of urbanization, with workers moving out of agriculture in large numbers due to declining access to farmland. These institutional and structural changes in the labor market had important consequences for the work of women with young children. Table 3 presents the labor force participation rates of parents with preschool-age children. We note that more than 90 percent of mothers in the sample participated in market work in 1991. Over the period from 1991 to 2006, labor force participation rates declined for both fathers and mothers, but the change was particularly dramatic for mothers: from 99.3 percent to 91.6 percent for fathers and from 90.7 percent to 69.7 percent for mothers. As a result, the gender difference in labor force participation increased from 8.6 to 19.9 percentage points. Among mothers, a larger decline in labor force participation was observed for those with children aged 0-2 (by 32.6 percentage points) than those with children aged 3-6 (by 14.1 percentage points). Table 4 presents the child care arrangements for all mothers by their work status. According to the survey, all mothers provided child care, although working mothers, on average, spent fewer hours on childcare than did non-working mothers (33 versus 43 hours per week). An appreciable proportion of fathers participated in child care, and the probability of paternal childcare is higher for working mothers than for non-working mothers (49.9 percent versus 31.6 percent). There are four major types of non-parental child care. Among working mothers, these include public formal child care (13.8

18

percent), private formal child care (19.3 percent), resident household members that are primarily grandparents (32.3 percent), and non-resident relatives, neighbors and hired baby-sitters (28.6 percent). Among nonworking mothers, 5.6 percent use public formal child care, 11.3 percent use private child care, 19.3 percent use the services of a resident household member and 23.6 percent use the services of a non-resident relative or neighbor. Although non-working mothers have less need for child care, they may use maternal care substitutes for reasons such as educating children in pre-schools, allowing grandparents to spend time with grandchildren, and giving themselves the time to fulfill other domestic duties or to enjoy leisure time. Table 5 presents the summary statistics of employment types, working hours, and earnings and child care hours from all sources for working women. Overall, half of the working mothers engage in regular wage-salary work; about 8 percent are owners of small business or contractors; 36.7 percent are self-employed (including farmers); and the rest work as temporary or unpaid family workers. Noticeably, the primary employment of working mothers varies across childcare arrangements. More than 70 percent of the mothers who use formal care have regular wage-salary positions, whereas regular wage earners account for only 50 percent of those who use informal care and less than 16 percent of those who do not use non-parental care. Conversely, self-employment is the primary occupation for more than 60 percent of those who do not use non-parental care but only 37 percent of those who use informal care and 16 percent of those who use formal care. Thus, not surprisingly, earnings are lowest among working mothers who do not use non-parental care even though the number of weekly work hours is similar across the three groups. As expected, the children of the working mothers who do not use non-

19

parental care receive substantially less care than the children of the other two groups; the total number of care hours from all sources for the former is, on average, less than half that of the other two groups (31.6 hours versus 77.5 and 77.4 hours per week). These statistics indicate that working mothers who did not use non-parental care either took children to work or left children unattended when working.

5. Empirical Results 5.1 Estimates of the Bivariate Probit Model Table 6 presents the estimates of the bivariate probit regressions that describe a mother’s joint decisions to work and to use formal child care for the full sample and the sample excluding suburban areas. The latter sample highlights the choices of mothers in highly urbanized areas where few women have the option to combine work with childcare. The Wald tests for ρ presented at the bottom of the table reject the null hypothesis that the mother’s labor force participation and choice of formal care are independent for both regressions. Thus, the bivariate probit model is a more appropriate specification than the single probit model.

A. Determination of Formal Childcare Utilization The equations for formal childcare utilization are designed to provide insights into the question of whether childcare reform has weakened the role of formal childcare programs as a means to provide care services. We discern the changing role of formal childcare programs with three proxy variables for accessibility of grandmother (number of females in the household aged 51 to 69 years and whether the woman lives close to her mother or mother in-law) and their interactive terms with the dummy variable for the

20

post-reform period (2000-2006). For the full sample, the three stand-alone informal care variables all have a negative sign and are significant at the 5 percent level or higher, whereas their interactive terms all have a positive sign and the variable of having a mother living nearby is significant at the 5 percent level. For the sample that excludes suburban areas, the estimates have a similar sign pattern and all of the informal care variables and their interactive terms are significant at the 5 percent level with the exception of having a mother in-law living nearby and its interactive term. Grandmothers appear to have been an important substitute for formal childcare programs prior to childcare reform, and the substitutability of grandmothers for formal childcare programs has declined in the post-reform period. The finding that more children are enrolled in childcare programs regardless of whether non-paid care providers (grandmothers) are available implies that childcare programs have become more important as a means to provide early education than to meet working mothers’ need for care services. The effects of the policy that decoupled the functions of childcare are also revealed by the finding that the utilization of formal care is constrained by the availability of childcare programs for children aged 0 to 2 years in a community (significant at 0.063 percent for the full sample and at the 5 percent level for the sample that excludes suburban areas) but not by the availability of childcare programs for children aged 3 to 6 years. Turning to other estimates, we find that mothers’ socioeconomic characteristics are important determinants of the use of formal care; both maternal education and unearned income are positively and significantly associated with choosing a formal care program. The use of formal care is negatively correlated with access to farmland, but the estimate is only significant for the full sample. The probability of using formal care also

21

increases with the age of the youngest child but decreases with the number of school-age boys (7 to 14 years old). Families with more school-age children have a tighter budget constraint and thus can less afford to use formal care for pre-school age children. With the exception of the presence of childcare programs for children aged 0 to 2 years, none of the community characteristic variables are significant.

B. Determination of Mothers’ Labor Force participation The labor force participation equations address the question of whether child care reform has increased women’s reliance on access to informal care for labor force participation. To answer this question, we again examine the proxy variables for grandmother accessibility and their interactive terms with the post-reform dummy variable. For the full sample, all three stand-alone informal care variables are insignificant, but the interactive terms of having a mother or a mother in-law living nearby are significant at the 5 percent level. For the sample that excludes suburban areas, the variable of having a mother living nearby and its interactive term are both significant at the 10 percent level or higher while the variable of having a mother in-law living nearby is insignificant but its interactive term is significant at the 5 percent level. Quantitatively, for the full sample, having a mother living nearby increases the probability that a mother participates in the labor force by 5.5 percent before childcare reform and 24.9 percent (up by 18.9 percentage points) after the reform, and the respective effects of having a mother in-law living nearby are 2.6 percent and 12.8 percent (up by 10.2 percentage points in the post-reform period). For the sample that excludes suburban areas, the effect of having a mother living nearby is 74 percent prior to the reform and rises to 93.2 percent (up by 19.2 percentage points) in the post-reform

22

period, and the respective effects of having a mother in-law living nearby are 3.9 percent and 19.3 percent (up by 15.4 percentage points). These estimates support the conjecture that childcare reform has increased working women’s dependence on informal care providers to meet their need for child care.19 Given that 90 percent of the women in the sample did not live with or close to their mothers, the rising dependence on maternal grandmothers is associated with a 17 percentage point decline in women’s labor force participation and 90 percent of the decline between 1997 and 2006. With respect to the proxy variables for paid child services, the median prices of child care programs are insignificant for both regressions, perhaps because they are too imprecise to capture the real costs paid by individual families as we pointed out before. The presence of childcare programs is also insignificant, which indicates that the availability of paid care services does not represent a constraint for women’s labor force participation. Instead, the issue appears to be a lack of affordable childcare services for working mothers. With respect to other covariates, we find that a mother’s own education and her husband’s education both have positive effect on her decision to work and are highly significant for both regressions. Women’s own age is also important but her husband’s age is not. Unearned income is insignificant for both regressions. Access to farmland has a positive effect on woman’s labor force participation (significant at the 10 percent level) for the full sample, but this access is insignificant for the sample that excludes suburban areas. Among household demographics, we find that the age of the youngest child is insignificant for the full sample but has a significant positive effect on the probability that a mother works in the cities and county centers. We also find that the probability that a

23

mother works increases with the number of females aged from 40 to 50 (for the full sample only) but decreases with the number of males aged between 15 and 25 and between 26 and 39 for both regressions. Regarding the living arrangement of male elderly parents, we find that the probability that a mother works decreases if her father in-law lives nearby (for the full sample) or if her father lives nearby (for the city and county center sub-sample). In addition, the probability that a mother works is higher in communities with a higher female labor force participation rate, in communities where a larger proportion of the labor force is employed by enterprises with 20 or more employees, and in suburban areas.

5. 2 Estimates of the Multinomial Logit Model Table 7 presents the multinomial logit estimates of the discrete choice model by (5) with option 2 (mother works and uses informal care) as the omitted category. As with the bivariate probit estimates, the results of the multinomial logit model are similar in substance between the full sample and the sample that excludes suburban areas. Due to the page limitation, we only report the results for the full sample. The regression is highly significant according to the log likelihood ratio test presented at the bottom of the table. We first examine the estimates of access to informal care providers. As expected, access to female maternal care substitutes (females aged 51 to 69 years in the household and mothers or mothers in-law living nearby) decreases the odds ratio of all three options (0 that a mother does not work, 1 that a mother works but does not use non-parental care, and 3 that a mother works and uses formal care) relative to option 2 that a mother works and uses informal care. Compared to options 1 and 3, option 0 that a mother does not work is particularly sensitive to the availability of informal care providers.20 A woman is

24

also more likely to exit the labor market if there are more males aged 15 to 25 and 26 to 39 in the household or if her father or father in-law lives nearby. To assess the effect of childcare reform, we take a look at the interactive variables of the number of females aged 51 to 69 years and mother in-law living nearby to the multinomial logit regression.21 The interactive variable for mother in-law living nearby has a negative sign and is significant at the 5 percent level or higher for options 0 and 1, and the interactive variable for females aged 51 to 69 years also has a negative sign and is significant for option 1 at the 10 percent level. These estimates thus support for the conjecture that the lack of access to an elderly female maternal care substitute increases the risk of non-employment for a mother as well as the risk of being left alone for the child of a working mother in the post-reform period. Regarding childcare availability in a community, we find that the presence of childcare programs for children aged 0-2 has a positive effect on the probability of option 1 (work, do not use non-parental care) and option 3 (work, use formal care) relative to option 2 (work, use informal care), and both estimates are significant at the 1 percent level. The odds ratio of option 1 to option 2 appears to be higher in communities that have childcare programs for young children. This result may be because, as we will show later, children who are left alone when mothers work tend to be older children whose mothers are typically from low-income families and, therefore, cannot afford to send the children to these programs. Nevertheless, an improvement of the availability of childcare programs for young children is attributable to a substantial shift from informal care use by working mothers to formal care, which raises the log of the odds ratio of option 3 to option 2 by 0.800. As with the bivariate probit model, the presence of childcare programs

25

for children aged 3 to 6 years in a community has no significant effect on a mother’s decision to enter the labor force and her choice of childcare arrangements. Holding constant access to informal caregivers and daycare programs, how do women’s socioeconomic characteristics influence their options to solve the competing demands of paid work and childcare? We note that a mother’s education has a negative effect on the probability of options 0 that the mother does not work and 1 that the mother works but does not use non-parental care. Relative to option 2 that the mother works and uses informal care, an additional year of schooling decreases the log of the odds ratio by 0.194 for option 0 and by 0.081 for option 1. Maternal education has no significant impact on the odds ratio of using formal care to informal care for working mothers (option 3 versus option 2). Moreover, an increase in the husband’s education decreases the probability that a woman does not work; an additional year of husband’s schooling decreases the log of the odds ratio of option 0 to option 2 by 0.173. With respect to unearned income, we note that an increase in unearned income increases the odds ratio of option 0 to option 2 (significant at the 10 percent level) as well as option 3 to option 2 (significant at the 5 percent level). Quantitatively, a 10 percent increase in unearned income raises the log of the odds of options 0 and 3 relative to option 2 by 0.018 and 0.028, respectively. These estimates indicate that, among the women who have no access to informal caregivers, women from low-income families are more likely to combine work with childcare; in contrast, women from affluent families can better afford to provide care to children by either withdrawing from the labor force or using formal child care. Thus, the lack of access to affordable non-parental care not only limits the ability of

26

women in low-income families to engage in high-paying work but also deprives their children of adequate care and early education. Turning to other estimates, we note that the age of the youngest child is positively correlated with the odds ratio of options 0, 1 and 3 relative to option 2, which indicates that younger children are more likely to be placed under the care of informal maternal substitutes (primarily grandmothers) than to be cared for by their non-working mothers, left alone, or sent to formal care programs. Given that the estimates for options 1 and 3 are larger than that for option 0 (0.268 and 0.592 versus 0.156), working mothers’ children who do not receive non-parental care or who are sent to formal care programs are typically older than children with non-working mothers. The husbands of women who work and do not use non-parental care appear to be older relative to the husbands of women who choose other options. Finally, the probability that a mother does not participate in paid work is lower in communities with a lower labor force participation rate for women of prime age and in suburban areas.

5.3 Simulation Results Based on the estimates presented in Tables 6 and 7, we obtained a series of predictions to indicate how mothers’ labor force participation and choice of childcare arrangements vary with daycare availability, parents’ education, and unearned income (see Table 8). We first look at the predictions for the availability of childcare programs for children aged 0-2 and unearned income using the estimates of the bivariate probit model. With respect to availability of childcare programs, we compare the baseline probability that a mother uses formal care at the mean values (43.6 percent of the observations in the communities that have daycares for children aged 0-2) with the

27

probability of assuming that all women live in communities that have such programs. The simulation results show that the improvement in the availability of childcare programs increases the probability that a mother uses formal care by 5.1 percentage points for the full sample and 7.7 percentage points for cities and county centers. Regarding unearned income, we compare the probabilities at the first and fourth quintiles. The probability of using formal care for a mother in the top quintile is higher than that for a mother in the bottom quintile by 7.1 percentage points for the full sample and 10.0 percentage points for the cities and county centers. It is evident that formal care programs are less accessible to less affluent families and that the disparity in access to formal care is greater in the cities and county centers than in the suburban areas. We next examine the predictions for mothers’ four discrete choices with respect to their own and husbands’ education and unearned income using the multinomial logit estimates. The baseline option probabilities are calculated using the sample means of the explanatory variables. The baseline predictions show that the option of combining work with child care is not limited to the suburban areas and that about 19 percent of the women in the sample that excludes suburban areas work even though no one else provides care to their children. We calculate the predicted probabilities for a family with parents who both have only a primary school education (6 years of schooling) and a family whose parents are both university graduates (16 years). For the full sample, the probability for the family with low educational attainment is 26.2 percent for option 0, 28.1 percent for option 1, 24.1 percent for option 2, and 21.6 percent for option 3; in contrast, the respective option probabilities for the family with high educational attainment are 1.2 percent, 18.6 percent, 42.7 percent and 37.5 percent for each respective

28

option. The gaps in options 0, 1, 2 and 3 between the two educational groups are 25.0, 9.6, and -18.6, -15.9 percentage points, respectively. Compared to the full sample, the gap between the two educational groups for the sample that excludes suburban areas is relatively large in terms of the probability that the mother does not work (34.9 percentage points) and the probability that the mother works and uses formal care (-29.1 percentage points). The gap is relatively small in terms of the probability that the mother works but does not use non-parental care (6.8 percentage points). For unearned income, we again compare the probabilities for the families at the first and the fourth quintiles of unearned income. The probability that the mother does not work is slightly higher at the fourth quintile than at the first quintile by 1.7 percentage points for the full sample and 0.5 percentage points for the sample that excludes suburban areas. The differences between the low- and high-income groups are more striking in the probabilities for options 1 and 3. Specifically, the probability that the mother works but does not use non-parental care for the fourth quintile is lower than that for the first quintile by 7 percentage points for the full sample and 10.6 percentage points for the sample that excludes suburban areas. Conversely, the probability that the mother works and uses formal care for the fourth quintile is higher than for the first quintile by 8.2 percentage points for the full sample and by 11.1 percentage points for the sample that excludes suburban areas. Comparing the probabilities for options 1 and 3 between the full and the non-suburban samples, we again note that inequality in childcare provision is greater in the cities and county centers than in suburban areas.

6. Conclusion

29

China’s transition from a centrally planned economy to a market economy has substantially eroded the government’s support for child care programs and shifted responsibility of child care primarily to the family. In this paper, we provide an overview of childcare reform in China and examine its impact on women’s labor force participation and child care choices and its distributional implications. Our empirical results show that the decline in government support for daycare programs, especially programs for young children, has weakened the role of formal childcare programs as a means to provide care services and consequently increased women’s dependence on informal caregivers to remain in the labor market. The rise of women’s dependence on their own mothers for childcare appears to account for much of the decline in women’s labor force participation during the period from 1997 to 2006. The lack of access to affordable non-parental care has forced women, especially those in socioeconomically disadvantaged families, to make a difficult choice to either leave the labor market or to work even though there is no one else to provide care to their children. These difficult choices have important implications for social inequalities, including gender inequality and for the wellbeing of women and their children. Chinese childcare reform offers important lessons for policy makers. To promote growth with equity and social harmony, the government should not limit its involvement in childcare to a regulatory and priority-setting role and shift care responsibility predominantly to the market and the family. Instead, policy measures must be introduced to remove barriers to quality childcare for low-income families. These policy interventions may include means-tested child care subsidies to parents and the development of publicly-funded childcare programs in low-income communities, less

30

developed regions, and rural areas. An improvement in the availability of daycare programs for young children should help reduce working women’s dependence on informal caregivers. To assess the effectiveness and costs of alternative policy options for Chinese policy markers, more rigorous analysis on the impact of the availability of child care programs and childcare costs should be a priority of future research.

31

Reference Appleton, Simon, John Knight, Lina Song, and Qingjie Xia. 2002. “Labor Retrenchment in China: Determinants and Consequences.” China Economic Review 13, no. 2-3: 252-275. Blau, David M., and Philip K. Robins. 1988. “Child-Care Costs and Family Labor Supply.” Review of Economics and Statistics 70, no. 3: 374-81. Blau, David M., and Hagy, Alison P. 1998. “The Demand for Quality in Child Care.” Journal of Political Economy 106: 104-39. Carneiro, Petro, and James Heckman. 2003. “Human Capital Policy.” NBER working paper no. 9495. Cleveland, Gordon, Morley Gunderson, and Douglas Hyatt. 1996. “Child Care Costs and the Employment Decision of Women: Canadian Evidence.” Canadian Journal of Economics 29:132-51. Connelly, Rachel.1992. “The effect of child care costs on married women’s labor force participation.” Review of Economics and Statistics 74: 83-90. Connelly, Rachel, Deborah S. DeGraff, and Deborah Levison.1996. “Women’s employment and child care in Brazil.” Economic Development and Culture Change 44: 619-656. Corter, Carl, Zeenat Janmohammed, Jing Zhang, and Jane Bertrand. 2006. “Strong foundations: early childhood care and education.” Paper commissioned for the EFA Global Monitoring Report 2007. Croll, Elisabeth. 1983. Chinese Women Since Mao. London: Zed Books.

32

Ding, Sai, Dong, Xiao-Yuan, and Li, Shi. 2009. “Married Women’s Employment and Family Income Inequality during China’s Economic Transition.” Feminist Economics 15, no. 3: 163-190. Dong, Xiaoyuan, Jianchun Yang, Fenglian Du, and Sai Ding. 2006. “Women’s Employment and Public Sector Restructuring: the Case of Urban China.” In Unemployment in China: Economy, Human Resources and Labor Markets, ed. Grace O.M.Lee and Malcolm Warner. London: Routledge. Doiron, Denise, and Guyonne Kalb. 2002. “Demand for Childcare Services and Labour Supply in Australian Families.” Australian Economic Review 35, no. 2: 204-213. Du, Fenglian, and Xiao-Yuan Dong. 2009. “Why Do Women Have Longer Durations of Unemployment than Men in Post-Restructuring Urban China?” Cambridge Journal of Economics 33:233-252. Du, Fenglian. 2008.

“Family Structure, Child Care and Women’s Labor Supply:

Evidence from urban China.” World Economic Papers 2: 1-12. Giles, John, Albert Park, and Fang Cai. 2006. “Reemployment of Dislocated Workers in Urban China: The Roles of Information and Incentives.” Journal of Comparative Economics 34, no. 3: 582–607. Gornick, Janet C., and Marcia K. Meyers, 2003. Families That Work. New YorK: Russell Sage Foundation. Gustafsson, Siv, and Frank Stafford. 1992. “Child Care Subsidies and Labor Supply in Sweden.” Journal of Human Resources 27(special issue on child care): 204-230.

33

Hallman, Kelly, Agnes R. Quisumbingm, Maries Ruel, and Benedicte de la Bariere. 2005. “Mothers’ Work and Child Care: Findings from the Urban Slums of Guatemala City.” Economic Development and Cultural Change 53, no. 4: 855-85. He, Jianhua, and Yongping Jiang. 2008. “An Analysis of China’s Childcare Policy and Current Situation from the Perspective of Supporting Women and Balancing Family and Work.” Studies in Preschool Education (xueqianjiaoyuyanjiu) 8: 3-7. Heckman, James J. 1974. “Effects of Child-Care Programs on Women's Work Effects.” Journal of Political Economy 82, no. 2: 136-163. Jaumotte, Florence. 2004. “Labor Force Participation of Women: Empirical Evidence on the Role of Policy and Other Determinants in OECD Countries.” OECD Economic Studies 37: 51-108. Jin, Qinghua, Yan Liu, Yan Zhang, and Qiong Li. 2005. “A Survey of Current Pre-school Education for Children from Urban Low-income Families in Beijing.” International Journal of Early Years Education 13, no. 2: 157-169. Kimmel, Jean. 1995. “The Effectiveness of Child Care Subsidies in Encouraging the Welfare-to-Work Transition of Low-income Single Mothers.” American Economic Review 85, no. 2: 271-275. Kilburn, M. Rebecca,and Ashlesha Datar. 2002. “The Availability of Child Care Centers in China and Its Impact on Child Care and Maternal Work Decisions.” Labor and Population Program, RAND Working Paper Series 02–12. Liu, Bohong, Yongying Zhang, and Yani Li. 2008. “Balancing the Conflicts between Family and Work: Problems and Policy in China.” Internal document of All China Women’s Federation.

34

Lokshin, Michael. 2004. “Household Childcare Choices and Women's Work Behavior in Russia.” Journal of Human Resources 39:1094-1115. Lokshin, Michael, and Monica Fong. 2006. “Women’s Labour Force Participation and Child Care in Romania” Journal of Development Studies 42, no. 1: 90–109. Lokshin, Michael, Elena Glinskaya, and Marito Garcia. 2004. “The Effect of Early Childhood Development Programs on Women’s Labor Force Participation and Older Children’s Schooling in Kenya.” Journal of African Economics 13, no. 2: 240-276. Maurer-Fazio, M, Rachel Connelly, Lan Chen and Lin Tang. 2009. “Childcare, Eldercare and Labor Force Participation of Urban Women 1982-2000.” Paper presented at the ASSA conference, San Francisco, January 3. Maurer-Fazio, Margaret, James Hughes, and Dandan Zhang. 2007. “Gender, Ethnicity, and Labor Force Participation in Post-reform Urban China.” Feminist Economics 13, no. 3 & 4: 159 – 187. Motiejunaite, Akvile, and Zhanna Kravchenko. 2008. “Family Policy, Employment and Gender Role Attitudes: a Comparative Analysis of Russia and Sweden.” Journal of European Social Policy 18: 38-49. Powell, Lisa M. 2002. “Joint Labor Supply and Childcare Choice Decisions of Married Mothers.” Journal of Human Resources 37, no. 1: 106-128. Quisumbing, Agnes R., Kelly Hallman, and Marie T. Ruel. 2007. “Maquiladoras and Market Mamas: Women’s Work and Childcare in Guatemala City and Accra.” Journal of Development Studies 43, no. 3: 420-455.

35

Ribar, David C. 1995. “A Structural Model of Child Care and the Labor Supply of Married Women.” Journal of Labor Economics 13, no. 3: 558-597. -------------------- 1992. “Child Care and Labor Supply of Married Women: Reduced Form Evidence.” Journal of Human Resources 27: 134-165. Viitanen, Tarja K. 2005. “Cost of Childcare and Female Employment in the UK.” LABOUR 19: 149-170. Wetzels, Cecile. 2005. “Supply and Price of Childcare and Female Labor Force Participation in the Netherlands.” LABOUR 19:171–209. Wong, Rebeca, and Ruth E. Levine. 1992. “The Effect of Household Structure on Women’s Economic Activity and Fertility: Evidence from Recent Mothers in Urban Mexico.” Economic Development and Cultural Change 41, no. 1: 89-102. Wrohlich, Katharina. 2006. “Labor Supply and Child Care Choices in a Rationed Child Care Market.” IZA working paper, no. 2053. World Bank. 2006. World Development Report 2006: Equity and Development. Now York: Oxford University Press. Yuan, Ni, and Sarah Cook. 2010. “Gender Patterns of Informal Employment in Urban China.” in Gender Equality and China’s Economic and Social Transformation: Informal Employment and Care Provision, eds. Xiao-yuan Dong and Sarah Cook. Beijing: Economic Science Press. Zhu, Jiaxiong, and X. Christine Wang. 2005. “Contemporary Early Childhood Education and Research in China.” In International Perspective on Research in Early Childhood Education, ed. Olivia Saracho and Bernard Spodek. Greenwich: Information Age Publishing.

36

TABLE 1 CHILDCARE PROVISION IN CHINESE MANUFACTURING ENTERPRISES, 2006 Ownership

No. firms

% have daycare

All firms

1,246

SOEs & collectives

262

19.6

Private & share-holding companies Foreign firms & joint ventures

781

2.7

260

3.1

5.7

Source: The Chinese Enterprises’ Social Responsibility Survey.

37

TABLE 2 NUMBER OF KINDERGARTENS IN CHINA, 1997-2006 1997

2000

2005

2006

182,485

175,836

124,402

130,495

30,694

35,710

25,688

26,877

127,148

95,809

29,879

28,192

24,643

44,317

68,835

75,426

13.5

25.2

55.3

57.8

Total

----

98,511

67,448

69,528

Private

----

22,553

32,202

34,313

% of total

----

22.9

47.7

49.4

Total

----

42,429

30,110

32,437

Private

----

12,490

19,364

22,000

% of total

----

29.4

64.3

67.8

Total

----

34,896

26,844

28,530

Private

----

9,274

17,269

19,113

% of total

----

26.6

64.3

67.0

Total Ministry of Education Other non-private organizations* Private % of private childcares By region East region

Central region

West region

Note: * kindergartens run by work units, neighborhood communities, and township and village governments. Source: Ministry of Education, 1997-2006 Education Statistics.

38

TABLE 3 LABOR FORCE PARTICIPATION RATES FOR PARENTS WITH CHILDREN AGED 0-6 YEARS, 1991-2006 1991

1993

1997

2000

2004

2006

All Fathers

0.993

0.954

0.991

0.959

0.906

0.916

All Mothers

0.907

0.883

0.886

0.815

0.713

0.697

Gender gap

0.086

0.071

0.105

0.144

0.193

0.199

Mothers by age of the youngest child Age 0-2

0.892

0.845

0.859

0.741

0.652

0.566

Age 3-6

0.913

0.897

0.908

0.859

0.758

0.772

Observations

428

265

201

157

157

145

Source: Tables 3-8 are based on the CHNS.

39

TABLE 4 CHILD CARE ARRANGEMENTS FOR MOTHERS WITH CHILDREN AGED 06, BY WORK STATUS Use non-parental childcare

Use parental childcare

Formal childcare

Informal childcare

Public

Other resident family members

Private

NonMother resident relatives / neighbors

Working mothers (observations = 1141) % using this type 13.8 19.3 32.3 28.6 100.0 of care Mean hours of care per child per 43.8 45.2 41.9 41.9 31.5 week Median hours of care per child per 48.0 48.0 29.0 48.0 21.0 week Non-working mothers (observations = 212) % using this type 5.6 11.3 19.3 23.6 100.0 of care Mean hours of care per child per 38.4 39.9 42.5 35.2 43.1 week Median hours of care per child per 36.0 40.0 28.0 40.0 35.0 week Notes: Some mothers use more than one type of care and therefore the sum of

Father

49.9

22.2

14.0

31.6

20.2

14.0

percentages is greater than 100 percent. The mean and median care hours are for the respective care provider.

40

TABLE 5 TYPE OF EMPLOYMENT, WORK HOURS, EARNINGS, AND CHIOLDCARE HOURS FOR WORKING MOTHERS All working Working mothers mothers Not use non- Use informal parental care child care 100.0 29.64 37.13

Distribution

Use formal Child care 33.13

Type of employment (%) Wage-salary work

50.96

30.37

50.12

70.29

Owner of small business/contractor Self employed

8.01

5.83

9.59

7.71

36.67

60.43

36.69

15.71

Temporary 4.37 work/unpaid work Hours worked per week last year

3.37

3.60

6.29

Mean

42.26

42.35

41.56

45.61

Mean (yuan/year)

5,514

5,087

5,888

5,472

Median (yuan/year)

3,440

2,896

3,699

3,571

Total earnings last year

Amount of time provided by all childcare types (hours per week per child) Maternal care

22.1

23.0

24.6

18.4

Paternal care

8.5

8.6

8.8

8.1

Resident nonparental care Care outside home*

14.0

0.0

30.3

8.3

19.0

0.0

13.8

42.6

Total

63.6

31.6

77.5

77.4

Note: *includes both time spent in formal childcare programs and being cared by nonresident relatives, neighbors, and hired baby-sitters.

41

TABLE 6 DETERMINANTS OF MOTHERS’LABOR FORCE PARTICIPATION AND USE OF FORMAL CHILD CARE Bivariate Probit Estimates Full Sample Labor force participation

∂F ( y1 = 1) ∂x Education Age Age2 Husband’s education Husband’s age Husband’s age2 Log unearned income Have farmland Age of youngest child Number of males age 7-14 Number of females age 7-14 Number of females age 1525 Number of males age 15-25 Number of males age 26-39 Number of females age 2639 Number of female age 4050 Number of males age 40-50 Number of females age 5169

0.013 (0.003)*** 0.091 (0.033)*** -0.001 (0.000)*** 0.014 (0.003)*** -0.005 (0.016) 0.000 (0.000) -0.012 (0.008) 0.049 (0.030)* 0.006 (0.005) -0.015 (0.029) 0.000 (0.022) 0.042 (0.027) -0.058 (0.024)** -0.070 (0.029)** 0.014 (0.039) 0.109 (0.046)** -0.053 (0.042) 0.041 (0.032)

Exclude suburban areas Use formal child care

∂F ( y 2 = 1) ∂x 0.010 (0.006)* -0.038 (0.067) 0.001 (0.001) 0.002 (0.006) -0.012 (0.036) 0.000 (0.000) 0.052 (0.019)*** -0.173 (0.052)*** 0.093 (0.010)*** -0.134 (0.051)*** 0.005 (0.040) 0.019 (0.040) -0.024 (0.052) 0.030 (0.053) 0.016 (0.052) -0.019 (0.061) -0.067 (0.067) -0.104 (0.049)**

Labor force participation

∂F ( y1 = 1) ∂x 0.017 (0.004)*** 0.051 (0.035) -0.001 (0.001) 0.015 (0.004)*** -0.010 (0.020) 0.000 (0.000) -0.015 (0.011) -0.004 (0.052) 0.015 (0.006)** 0.004 (0.040) -0.012 (0.031) 0.006 (0.032) -0.068 (0.033)** -0.081 (0.042)* -0.025 (0.052) 0.081 (0.058) -0.092 (0.058) 0.052 (0.039)

Use formal child care

∂F ( y 2 = 1) ∂x 0.023 (0.008)*** -0.058 (0.097) 0.001 (0.002) 0.003 (0.008) 0.012 (0.048) 0.000 (0.001) 0.087 (0.030)*** -0.138 (0.093) 0.114 (0.016)*** -0.142 (0.075)* 0.076 (0.059) -0.030 (0.052) 0.029 (0.081) 0.099 (0.079) -0.033 (0.068) -0.012 (0.105) 0.006 (0.107) -0.187 (0.067)***

42

Number of females age 5169 x after 1997 Number of males age 51-69

0.105 (0.072) -0.029 (0.052) -0.009 (0.085) 0.034 (0.086) -0.249 (0.089)*** 0.379 (0.155)** 0.250 (0.085)*** -0.150 (0.061)** 0.142 (0.096) 0.012 (0.059) 0.086 (0.050)* 0.017 (0.048) 0.001 (0.048) -0.007 (0.036) 0.005 (0.037) -0.022 (0.037) 0.101 (0.104) 0.000 (0.001) -0.091 (0.058) -0.025 (0.047) yes

0.026 (0.047) -0.007 (0.043) -0.040 (0.050) 0.113 (0.093) 0.740 (0.171)*** 0.192 (0.111)* -0.859 (0.153)*** -0.039 (0.055) 0.154 (0.060)** -0.018 (0.057) 0.027 (0.025) 0.000 (0.024) 0.028 (0.023) -0.008 (0.028) 0.014 (0.019) -0.003 (0.023) 0.308 (0.087)*** 0.000 (0.000)

0.184 (0.094)** -0.061 (0.080) -0.017 (0.115) -0.137 (0.142) -0.385 (0.174)** 0.597 (0.231)*** 0.304 (0.169)* -0.140 (0.090) 0.087 (0.131) 0.025 (0.085) 0.149 (0.059)** -0.055 (0.057) -0.009 (0.065) -0.013 (0.054) 0.056 (0.050) -0.039 (0.060) -0.017 (0.166) -0.001 (0.001)

Regional dummy

0.014 (0.039) -0.008 (0.026) -0.040 (0.035) 0.002 (0.040) 0.052 (0.096) 0.189 (0.095)** -0.140 (0.087) 0.026 (0.047) 0.102 (0.049)** -0.114 (0.049)** 0.032 (0.021) 0.001 (0.020) 0.007 (0.019) 0.019 (0.018) 0.018 (0.017) -0.033 (0.022) 0.241 (0.071)*** 0.001 (0.000)** 0.057 (0.031)* 0.013 (0.026) yes

----0.005 (0.024) yes

----0.018 (0.054) yes

Year dummy

yes

yes

yes

yes

Rho

0.169

0.323

(0.079)

(0.088)

Wald test of Rho = 0

4.394

11.608

p-value

0.036

(0.001)

Observations

1142

656

Number of females age 70 and older Number of males age 70 or older Mother lives nearby Mother lives nearby x after 1997 Father lives nearby Mother in-law lives nearby Mother in-law lives nearby x after 1997 Father in-law lives nearby Community has daycare for age 0-2 Community has daycare for age 3-6 Log median price of daycare for age 0-2 Log median price of daycare for age 0-2 Log daily wage for babysitters/nannies Log mean female wage per day LFP of married women aged 19 – 50 % work in enterprises with 20 or more employees Suburban areas County centers

43

Notes: Standard errors presented in parentheses are robust to heteroscedasticity and intracommunity clustering. ***, **, and * denote significance levels of 1%, 5% and 10%, respectively. Each regression also includes the intercept. The reference group includes those who live in urban areas, in communities with no daycare for children in the respective age group and do not live as a neighbor to their mother, father, mother in-law or father in-law. The omitted class for household composition is children aged 6 years or younger.

44

TABLE 7 MULTINOMIAL LOGIT ESTIMATES OF MOTHERS’ WORK AND CHILDCARE CHOICES (Full Sample) Mother does not work

Mother works but does not use nonparental care

Mother works and uses formal care

Relative to the base category (mother works and uses informal care)

Education Age Age2 Husband’s education Husband’s age Husband’s age2 Log unearned income Have farmland Age of youngest child Number of males age 7-14 Number of females age 7-14 Number of females age 15-25 Number of males age 15-25 Number of males age 26-39 Number of females age 26-39 Number of female age 40-50 Number of males age 40-50 Number of females age 51-69

∂ ln( P0 / P2 ) ∂x

∂ ln( P1 / P2 ) ∂x

∂ ln( P3 / P2 ) ∂x

-0.193 (0.047)*** -0.892 (0.459)** 0.013 (0.007)* -0.175 (0.048)*** 0.180 (0.264) -0.001 (0.004) 0.182 (0.108)* -0.820 (0.329)** 0.159 (0.074)** -0.003 (0.397) -0.088 (0.354) -0.642 (0.386)* 0.854 (0.396)** 0.964 (0.449)** -0.281 (0.466) -1.512 (0.649)** 0.780 (0.587) -1.262 (0.452)***

-0.084 (0.038)** 0.164 (0.426) -0.003 (0.007) -0.018 (0.039) 0.647 (0.295)** -0.009 (0.004)** -0.116 (0.090) 0.126 (0.275) 0.267 (0.061)*** -0.018 (0.317) -0.168 (0.262) -0.018 (0.250) 0.172 (0.338) 0.044 (0.426) 0.139 (0.388) -0.149 (0.444) 0.510 (0.513) -1.281 (0.307)***

0.003 (0.041) 0.148 (0.437) -0.003 (0.007) -0.011 (0.040) 0.030 (0.253) 0.000 (0.004) 0.294 (0.113)*** -1.067 0.294)*** 0.590 (0.064)*** -0.912 (0.382)** -0.081 (0.283) 0.115 (0.253) 0.004 (0.357) 0.245 (0.387) 0.114 (0.369) -0.208 (0.443) -0.162 (0.498) -1.185 (0.297)***

45

Number of females age 51-69 x after 1997 Number of males age 51-69 Number of females age 70 and older Number of males age 70 or older Mother lives nearby Father lives nearby Mother in-law lives nearby Mother in-law lives nearby x after 1997 Father in-law lives nearby Community has daycare for age 0-2 Community has daycare for age 3-6 Log median price of daycare for age 0-2 Log median price of daycare for age 3-6 Log mean wage for babysitters per day Log mean female wage per day LFP of married women aged 19 – 50 % work in enterprises with 20 or more employees Suburban areas County centers Regional dummies Year dummies

-0.314 (0.551) 0.039 (0.330) 0.171 (0.502) -0.301 (0.614) -2.358 (1.369)* 2.820 (1.380)** -0.600 (0.517) -1.840 (0.716)*** 1.796 (0.514)*** 0.110 (0.309) -0.072 (0.297) -0.317 (0.295) -0.014 (0.226) -0.156 (0.262) 0.564 (0.289)* -2.320 (0.757)***

-0.916 (0.487)* -0.251 (0.273) -0.442 (0.432) -1.410 (0.565)** -2.551 (0.875)*** 1.677 (0.951)* -0.158 (0.374) -1.257 (0.609)** 0.751 (0.403)* 0.821 (0.259)*** -0.084 (0.038)** 0.164 (0.426) -0.003 (0.007) -0.018 (0.039) 0.647 (0.295)** -0.009 (0.004)**

0.052 (0.447) -0.254 (0.269) -0.206 (0.433) -0.206 (0.456) -1.612 (0.769)** 1.988 (0.865)** -0.959 (0.429)** -0.035 (0.624) 0.533 (0.451) 0.800 (0.245)*** 0.018 (0.250) -0.265 (0.261) 0.192 (0.199) 0.136 (0.218) 0.077 (0.250) 1.391 (0.768)*

-0.011 (0.004)** -0.668 (0.363)* -0.114 (0.336) yes

-0.116 (0.090) 0.126 (0.275) 0.267 (0.061)*** yes

-0.005 (0.003)* -0.448 (0.300) -0.039 (0.263) yes

yes

yes

yes

2

0.260

2

802.35

Pseudo R

LR test (χ ) p-value Observations

0.0 1142

Notes: Standard errors presented in parentheses are robust to heteroscedasticity and intracommunity clustering. ***, **, and * denote significance levels of 1%, 5% and 10%, 46

respectively. Each regression also includes the intercept. The reference group includes those who live in urban areas, in communities with no daycare for children in the respective age group, and those who do not have mother, father, mother in-law or father in-law living nearby. The omitted class for household composition is children aged 6 years or younger.

47

TABLE 8 SIMULATION OF THE EFFECTS OF AVAILABILITY OF DAYCARES, EDUCATION, AND UNEARNED INCOME Bivariate Probit Model Full sample

Exclude suburban areas

Labor force Use of formal Labor force Use of formal participation child care participation child care Effects of availability of daycares for children age 0-2 in a community Baseline* 100% available

0.916 0.254 0.935 0.305 Effects of unearned income

At the first quintile** At the fourth quintile

-------

0.223 0.293 Multinomial Logit Model

0.916 0.932

0.322 0.401

-------

0.277 0.378

Non-working mother

Working mother Not use non- Uses informal Uses formal parental care care care Effects of education

Full sample Baseline*** 0.109 0.274 Both parents have primary 0.270 school education (6 years) 0.258 Both parents are college graduates (16 years) 0.011 0.189 Sample excluding suburban areas Baseline 0.107 0.190 Both parents have primary school education (6 years) 0.359 0.185 Both parents are college graduates (16 years) 0.010 0.117 Effects of unearned income

0.338

0.279

0.254

0.216

0.450

0.348

0.308

0.395

0.211

0.245

0.337

0.536

Full sample At the first quintile 0.100 0.306 0.351 0.243 At the fourth quintile 0.117 0.239 0.320 0.324 Sample excluding suburban areas At the first quintile 0.106 0.240 0.313 0.341 At the fourth quintile 0.111 0.146 0.291 0.452 Notes: *43.6% of women in the full sample and 50.1% for cities and county centers live in the communities that have daycares for children aged 0 to 2 years. ** The first and 48

fourth quintiles are, respectively, 8.17 and 9.52 for the full sample and 8.28 and 9.43 for the sample of excluding suburban areas. *** The average years of schooling are 8.81 for mothers and 9.60 for fathers for the full sample and the respective figures are 9.50 and 10.18 for the sample that excludes suburban areas.

49

APPENDIX CHARACTERISTICS OF MOTHERS WITH DIFFERENT CHILD CARE ARRANGEMENTS All working mothers

Working mothers Not use Use noninformal parental child care child care 7.59 9.20 30.61 28.67

Years of schooling 8.90 Age 29.82 Husband’s years of schooling 9.72 8.95 Husband’s age 31.72 32.60 Log unearned income (yuan) 8.76 8.51 Have farm land 0.327 0.500 Household demographics and living arrangements Age of youngest child 3.31 3.48 Number of males age 06 0.61 0.63 Number of females age 0-6 0.50 0.58 Live with mother 0.037 0.077 Live with father 0.032 0.062 Live with mother in-law 0.360 0.186 Live with father in law 0.287 0.139 Mother lives nearby 0.038 0.021 Father lives nearby 0.031 0.021 Mother in-law lives nearby 0.143 0.228 Father in-law lives nearby 0.105 0.180 Community characteristics Has daycare for age 0-2 0.436 0.373 Has daycare for age 3-6 0.595 0.515 Log median price of daycare for age 0-2 (yuan/month) 4.38 4.37 Log median price of daycare for age 3-6 (yuan / month) 4.27 4.28 Log mean wage for babysitter/ nannies 1.87 1.87

Use formal child care

Nonworking mothers

9.75 30.41

8.24 29.46

9.87 30.34

10.24 32.47

8.87 32.04

8.79 0.355

8.96 0.140

8.92 0.250

2.48

4.09

2.95

0.63

0.58

0.56

0.47 0.009 0.009 0.562 0.471 0.042 0.026

0.45 0.032 0.029 0.288 0.214 0.048 0.045

0.52 0.047 0.024 0.311 0.236 0.033 0.038

0.111

0.103

0.175

0.068

0.079

0.165

0.367 0.579

0.569 0.685

0.425 0.627

4.46

4.29

4.56

4.30

4.21

4.36

1.92

1.83

2.12

50

(yuan/day) Log mean wage of female workers (yuan/day) LFP rate of married women aged 19 - 50 % work in enterprises with 20 or more employees Live in urban areas Live in suburban areas Live in county centers Observations

2.52

2.48

2.57

2.50

2.71

0.858

0.864

0.849

0.862

0.728

40.3 0.228 0.431 0.341 1141

32.6 0.145 0.553 0.302 338

40.6 0.202 0.456 0.341 425

46.9 0.331 0.294 0.376 378

28.6 0.193 0.349 0.458 212

51

ENDNOTES

1

Carnerio and Heckman (2003) argue that public interventions in early childhood

development are more effective as a preventive measure compared with policies that seek to remedy deficits incurred in the early years for children from socioeconomically disadvantaged households. According to the World Development Report 2006, early childhood development programs have a high “rate of return” ($2 to $5 for every dollar invested). 2

Studies have documented that in the post-transitional period, women were often laid off

at much higher rates than men and experienced greater difficulty finding reemployment (Appleton et al. 2002; Giles, Park and Cai 2006; Du and Dong 2009). Additionally, a large number of women have withdrawn from the labor market (Dong et al. 2006; Maurer-Fazio, Hughes and Zhang 2007). 3

A selective list of such studies includes Heckman (1974), Blau and Robin (1988),

Connelly (1992), Robar (1992, 1995), Kimmel (1995), and Blau and Hagy (1998) in the United States; Gustafsson and Stafford (1992) in Sweden; Cleveland, Gunderson and Hyatt (1996) and Power (2002) in Canada; Doiron and Guyonne (2002) in Australia; Vitttanen (2005) in the UK; Van Den Brink and Groot (1997) and Wetzels (2005) in the Netherlands; and Wrohlich (2006) in Germany. 4

Among a small number of rigorous empirical studies are Wong and Levine (1992) in

Mexico; Connelly, DeGraff and Levison (1996) in Brazil; Lokshin, Glinskaya and Garcia (2004) in Kenya; Lokshin (2004) in Russia; Hallman, et al. (2005) and Quisumbing, Hallman and Ruel (2007) in Guatemala; and Lokshin and Fong (2006) in Romania.

52

5

For instance, in the Soviet Union, childcare programs were organized and funded by the

municipal authorities and their coverage was universal (Motiejunaite and Kravchenko 2008). 6

Based on a survey conducted in the western district of Beijing, Zhang and Wu (2006)

find that the number of kindergartens run by neighborhood communities fell from 50 in 1987 to 25 in 1998 and to 17 in 2002. 7

A survey of 14 provinces revealed that about 36 percent of public kindergartens charged

non-employee users “sponsorship fees” ranging from 1,000 yuan to 10,000 yuan per year. The sponsorship fees of some elite kindergartens in Beijing were as high as 30,000 yuan per year (Liu, Zhang and Li 2008). 8

See the Ministry of Education, Educational Statistics Yearbook of China, various issues.

9

See Corter et al. (2006), He and Jiang (2008), and Liu et al. (2008).

10

From interviews with managers in some northern Chinese cities in 2007, we found that

the organizations that continued to run childcare programs were either non-profit public organizations and government offices or large SOEs in industries that disallow private entry such as resources and communication. For instance, one aluminum company provided 2 million yuan of subsidies to the childcare, whereas a large steel company provided 16 million yuan. These subsidies did not include the free space and facilities that the companies also provided to childcare programs. However, most SOEs in light industries were privatized and ceased to provide childcare services to employees after privatization. The end to employer-provided or subsidized childcare changed the composition of the workforce in some companies. One wool textile company had previously employed a large number of women with young children as shop-floor

53

workers. Since privatization, however, almost all of the shop-floor female workers in this company have been young, single women. 11

For example, a survey shows that average fees ranged from 600 to 800 yuan per month

for publicly financed kindergartens and 1,300 to 1,500 yuan per month for private kindergartens in Beijing (He and Jiang 2008). 12

We focus on the role of informal care instead of the availability and affordability of

paid childcare programs for two reasons. First, a change in the availability of paid childcare programs does not automatically change a woman’s decision to work because the majority of working women use informal sources of child care at little or no direct cost. Second, the CHNS data provides no reliable information to assess the real costs of paid childcare to individual women because they largely depend upon factors such as the sector a mother or her spouse is employed in rather than the average prices of child care centers in a community reported in the dataset. 13

For purpose of tractability, we assume that household composition and parents’ living

arrangements are exogenous. To address the concern that recent fertility may be endogenous to a mother’s labor force participation, the number of children aged 6 years or younger is chosen as the omitted class for household composition measures and the age of the youngest child is introduced only as a covariate control variable. 14

Due to the lack of suitable instrumental variables, the availability of formal childcare is

treated as exogenous. We address the concern that the presence of formal child care programs may be driven by a high demand of working mothers by controlling for local female wages, the labor force participation rate of prime-age women, and the size of the formal sector in the local market.

54

15

As we show in the next section, some non-working mothers use non-parental care. We

combine all non-working mothers into one group because an additional division would create problems for the multinomial logit estimation. 16

Recognizing that the multinomial logit model imposes an unrealistic assumption that

the odds ratio Pj/Pk is independent of the other alternatives, we also estimate equation (5) as a multinomial probit model. The results are similar to those presented in Table 7. 17

The CHNS is jointly sponsored by the Carolina Population Center at the University of

North Carolina at Chapel Hill, the Institute of Nutrition and Food Hygiene of China and the Chinese Academy of Preventive Medicine. Detailed information about the CHNS is available at the website www.cpc.unc.edu/china/home.html. 18

Although the data include a panel of households and individuals, they are analyzed as

repeated cross-sections to avoid biases related to attrition and cohort as the panel ages over time. Moreover, given that the time lag between two consecutive CHNS surveys is approximately three or four years, a panel of women with children aged 0 and 6 contains too few observations to render the analysis meaningful. 19

Arguably, there may be other forces driving the positive association of women’s labor

force participation with proximity to grandmothers, such as the development of housing markets in the late 1990s. The rising residential mobility that may partly explain this result could not be a dominant force relative to the declining affordability of paid child care programs because housing rental markets remain underdeveloped in China and purchasing a new house to be close to grandmothers is much more expensive than enrolling children in childcare programs.

55

20

The proxy variables for female maternal care substitutes are insignificant in the labor

force participation equation but significant for option 1 in the multinomial logit equation prior to the child care reform because the former does not account for those mothers who work in spite of no access to informal care givers. 21

The interactive term of having a mother living nearby is dropped from the multinomial

logit model because the variable does not have sufficient variation over all options for the post-1997 period.

56