Gender, Low-Paid Status, and Time Poverty in Urban China

ABSTRACT

Using synthetic data from the 2008 China Time Use Survey (CTUS) and the 2008 China Household Income Project (CHIP), this study estimates time-poverty rates and compares the profiles of time-poor men and women workers in urban China. In line with previous research, time poverty is defined as a lack of enough time for rest and leisure. Three time-poverty measures are adopted. By all three measures, women paid workers and low-paid workers account for a disproportionate share of the time poor. Regression analysis further shows that, other things being equal, workers who are women, low-paid, married, and who live with children or the elderly in counties with higher overtime rates and lower minimum wage standards are more likely to be time poor. Simulations indicate that enforcing working time regulations and raising minimum wage standards could be effective for reducing time poverty.

KEYWORDS:

• Time poverty
• gender
• low-paid workers
• unpaid care work
• China

JEL Codes:

• J16
• J22
• I32

INTRODUCTION

Time is an important determinant of human well-being and capacities. However, many workers do not have enough time to meet their basic needs for rest and leisure because of the excessively long hours of paid work and unpaid care work.¹ Clair Vickery (1977) termed this phenomenon “time poverty.” Time poverty has negative consequences for human well-being and capabilities. The prolonged deprivation of adequate rest and personal care can create emotional and physical strain and cause health deterioration (Park et al. 2001a, 2001b; Frijters, Johnston, and Meng 2009; Virtanen et al. 2011; Kim et al. 2013; Bannai and Tamakoshi 2014). The lack of freely disposable time impedes workers’ abilities to invest in human capital, build social networks, spend time with their families, and partake in cultural and political activities.

Time poverty has strong gender dimensions. Bearing the double burden of paid and unpaid care work, women tend to have less time for rest and leisure compared with men (a selective list of references includes Bardasi and Wodon [2010] for Guinea; Gammage [2010] for Guatemala; and Warren, Pascall, and Fox [2010]; and Burchardt [2008] for the United Kingdom), and the lack of freely disposable time hinders women's agency, well-being, and capabilities (Floro 1995; Robeyns 2003). Time poverty is also intertwined with income disparities. While workers with high earnings may choose to work long hours, low-paid workers face a more acute work–leisure trade-off. Because of their low earnings, low-paid workers must often work long hours to generate enough income to lift their families out of poverty. Low-paid workers are also more constrained in their ability to outsource housework and care services to the market. Thus, low-paid workers are more likely to be time poor. Recognizing the impact of time poverty on workers, the International Labour Organization regards the notion of allowing workers adequate time for rest and leisure as an important dimension of decent work (ILO 1999).

The issue of work burden and its relationship to time poverty is particularly important for China because Chinese workers, especially women workers, carry a heavy workload. Studies show that paid working hours in China are long relative to other parts of the world (Organisation for Economic Co-operation and Development [OECD] 2011). Most Chinese women workers are employed on a full-time basis, as their male counterparts are, but women also bear the primary responsibility for unpaid care work. As a result, the total work hours of Chinese women workers are much longer than that of their men coworkers (Dong and An 2015; Qi and Dong 2016). In addition, China's economic reforms have eroded the state's protection and support for workers with family responsibilities, thereby exacerbating gender- and income-based inequalities in working hours (Liu, Zhang, and Li 2009; Cook and Dong 2011). Nevertheless, the nonmonetary aspects of human deprivation and inequality are often neglected in mainstream policymaking.

This study examines the time poverty of men and women workers in urban China. The analysis is based on data from two large surveys: the 2008 China Time Use Survey (CTUS) and the 2008 China Household Income Project (CHIP). We first estimate time-poverty rates for various groups differentiated by gender and earnings status. We next investigate the correlates of time poverty and use the estimates to simulate the potential effects of strengthening labor market regulations on time poverty. Our study is the first to examine time poverty in China by using large-sample survey data. The analysis contributes to deepening our understanding of human deprivation as a multidimensional phenomenon and also provides valuable information for the design of gender-sensitive and inclusive public policies.

INSTITUTIONAL BACKGROUND

Feminist scholars have long argued that time-use decisions are influenced not only by economic considerations, but also by institutional factors (Folbre 2004; Burchardt 2008). Since China embarked on the transition from a planned economy to a market economy in the late 1970s, the overriding concern of the Chinese government has been to improve economic efficiency and stimulate growth. This development strategy places a greater weight on income and wealth than on the nonmonetary aspects of well-being and capabilities. In particular, public policies tend to take unpaid care work for granted and are lax in enforcing protective labor market regulations. These unbalanced policies overlook that the time and effort involved in the daily work of caring for oneself and others is a vital part of social reproduction and an essential foundation for the productive economy. The policy neglect of nonmonetary aspects of well-being and capabilities at the macro level influences the formulation and implementation of economic and social policies at the micro level, which directly affect the time budgets of women and men. Here, we review three related policy measures: minimum wage regulation, working time regulation, and provision of care services for children and the elderly.

Minimum wage regulation

China's minimum wage regulations were first introduced in the early 1990s and were enacted across the country in 2004. Each province's government determines the minimum wage standards, and, within each province, minimum wages vary across cities, depending on local market situations (Du and Wang 2008). The minimum wage standards in China are low. Zhaozhou Han and Wei Zhangjin (2011) estimate that the mean ratio of the minimum wage standard to the local average wage in thirty-five large and medium-sized cities in China is 26.8 percent, ranging from 17 percent in Beijing to 39.6 percent in Shijiazhuang. In comparison, the ratio of the minimum wage to the average wage in 2008 is, respectively, 45 percent in Australia and 50 percent in France (OECD n.d.). In the Employment Promotion Program for the Period of the 12th Five-Year Development Program (2011–5), the Chinese government proposed raising the minimum wages to 40–60 percent of the average wages (Zheng 2013). However, the growth of minimum wage standards has lagged behind the growth of wages (Zheng 2013; Xing and Xu 2016). Unlike other countries where minimum wage standards are stipulated based on hourly wages, China's minimum wage standards include two thresholds: a monthly threshold for full-time workers and an hourly threshold for part-time workers. Given that the vast majority of Chinese workers are employed on a full-time basis, the monthly threshold is applied to most workers. Although only a small fraction (3.5 percent) of full-time workers earn less than the legal monthly minimum wage (Ye, Gindling, and Li 2015), the double thresholds of minimum wage standards create a regulatory discrepancy that allows employers to circumvent the minimum wage regulation by requiring workers to work long hours.

Working time regulations

Adopted in 1994, China's Labor Law stipulates that workers shall work for no more than 8 hours a day and no more than 44 hours a week. The regulation also sets restrictions on and stipulates the compensation standards for overtime work. However, there are no means of enforcing the working time regulations in the nonpublic sector, where the vast majority of Chinese workers are employed (Liu, Zhang, and Li 2009).

The design of working time regulations has loopholes, as the compliance of the daily and weekly working hour regulations in some industries is not based on actual hours of work, but on the average daily and weekly work hours over a longer period.² In these industries, employers can avoid paying overtime by increasing work hours when the firm has large orders and reducing work hours when business is slow, as long as the average daily and average weekly working hours do not exceed the legally prescribed working hours standard.

Studies show that Chinese workers, on average, work longer hours than their counterparts in other developing countries (OECD 2011). The incidence of overtime work is prevalent, especially in the manufacturing and commercial service sectors (Verité 2004, 2012; Ye, Gindling, and Li 2015). Using an employer–employee matched dataset in Shanghai, Vinod Mishra and Russell Smyth (2013) found that employees worked longer hours in firms that employed a high proportion of women and migrant workers. They also found that workers who received lower pay worked longer hours. While low-paid workers had to work long hours to compensate for low wages, many workers with high earnings also worked overtime due to career pressures and work ethics that encourage men to chase after “success” (Irwin 2012).

Provision of care services

During the Mao era (1949–76), the state and employers played an important role in providing care services in cities (Cook and Dong 2011). In the reform era, care work has largely shifted to the market and the household. From 1998 to 2013, the share of public kindergartens for China as a whole decreased from 83 percent to 33 percent. In 2012, 73 percent of the kindergartens in cities were privately owned. The privatization of childcare services creates cost barriers to low-income families who want to enroll their children in childcare programs (Liu, Zhang, and Li 2009).

As in many countries worldwide, the provision of care for the elderly is primarily the responsibility of families in China. The Chinese government limits its role in eldercare provision to financing and operating nursing homes only for the elderly and disabled who have no family members to take care of them. In 2011, only 2.6 million out of the 120 million elderly people in China were living in nursing homes (Huang 2013). The aging Chinese population has further increased the care burden on families.

LITERATURE REVIEW: DEFINITION AND MEASUREMENTS OF TIME POVERTY

The concept of time poverty was first proposed by Vickery (1977). She argued that the design of poverty thresholds should consider not only the amount of income required to purchase the minimum goods and services from the market, but also the amount of time needed to process these goods and services in home production. If a household did not have enough time for home production after deducting the time spent on paid work to earn an income at the poverty threshold and the minimum time required for rest and leisure, then the household would face an income shortfall because it could not afford to outsource home production to the market under the existing poverty threshold. In this regard, a household is deemed time poor if it has a time deficit in home production.

In their study of Canadian households, Andrew S. Harvey and Arun K. Mukhopadhyay (2007) proposed a method for setting time-adjusted poverty lines under the assumption that housework and care services can be purchased from the market. This method first calculates the monetary value of the time deficit in home production and then deducts this value from household income before estimating poverty rates. Harvey and Mukhopadhyay (2007) found that, compared with other households, single-parent households and households with more children are more likely to have time deficits in home production and to be time poor. Adopting a similar approach, Ajit Zacharias, Thomas Masterson, and Rania Antonopoulos (2012) and Ajit Zacharias, Thomas Masterson, and Kijong Kim (2014) analyzed time and income poverty in Argentina, Chile, Mexico, and South Korea.

In developed countries, people increasingly complain about a “time crunch.” Robert E. Goodin et al. (2005) argued that many people work long hours in the labor market by choice, not strictly out of need, and public policies should focus on those who must work long hours in the labor market due to socioeconomic constraints. Hence, distinguishing between “free time” and “discretionary time” is necessary. Free time is the time left over after deducting the time actually spent on paid work, unpaid care work, and personal care from the total time in a given period, while discretionary time is the difference between the total amount of time in a given period and the minimum necessary time spent on paid work, unpaid care work, personal care, and leisure. Goodin et al. (2005) defined minimum necessary paid work time as the time needed to earn an income that corresponds to no less than the poverty line at the given wage rate. Minimum necessary unpaid care work time, personal care time, and leisure time are defined as the mean value of each item in the sample minus one standard deviation. In this study, time poor refers to individuals whose discretionary time has a negative value. Using 1992 Australian time-use data, Goodin et al. (2005) found that most people spend much more time than the minimum necessary on paid work and unpaid care work; only single mothers have negative discretionary time and are truly time poor.

Tania Burchardt (2008) argued that both the lack of free time and the lack of discretionary time should be policy concerns. The former may reflect inflexible work schedules, high career pressures, and the financial burdens of housing, children's education, and pension contributions, whereas the latter relates to minimum wages and the availability of affordable childcare services. Stella Chatzitheochari and Sara Arber (2012) defined time poverty as a relative deprivation of free time and classified an individual as time poor if his or her free time is less than 60 percent of the median free time of those in the sample. Using British time-use data, Chatzitheochari and Arber found that women paid workers are more likely than men paid workers to be time poor.

Elena Bardasi and Quentin Wodon (2010) define time poverty based on the total amount of time spent on paid work and unpaid care work. Because everyone has the same amount of time, more time spent on paid and unpaid work means less time available for personal care and leisure. Bardasi and Wodon introduced two time-poverty measures to distinguish individuals who work long hours for pay by choice from those who work long hours for pay to meet their basic household needs due to low earnings or high domestic burdens. In the first measure, an individual is defined as time poor if his or her total work time exceeds the time-poverty threshold. The authors adopt two thresholds in the analysis, with one equal to 50 hours per week and one equal to 1.5 times the median paid work hours in the sample. In the second measure, an individual is defined as time poor if the individual works long hours for pay and lives in households that are income poor or that would become income poor if the individual were to reduce his or her paid work hours below the time-poverty threshold. Bardasi and Wodon (2010) found that, according to both definitions, less-educated individuals and women, especially women who live in rural areas and are single parents, are more likely to be time poor.

TIME POVERTY MEASURES FOR URBAN CHINESE WORKERS

In this study, we define time poverty by following the approach of Bardasi and Wodon (2010). We introduce three time-poverty measures: TP1 is based on actual total work time; TP2 is based on the minimum necessary total work time; and TP3 is the intersection between TP1 and TP2. The three measures are defined below: (1) (2) (3) where APT and AUPT stand for actual paid work time and actual unpaid care work time, respectively; NPT is the minimum necessary paid work time that enables an individual to earn an income that prevents the household from falling below the poverty line, given this individual's hourly wage and his or her share of total household labor income.³

The time-poverty threshold of 68.4 hours per week is specified under the assumption that, given the total of 168 hours in a week, if an individual works (whether for pay or not) more than 68.4 hours per week, he or she will not have enough time to meet the minimum required for personal care (8 hours of sleep and 4 hours of eating, drinking, and personal hygiene per day) and leisure (15.6 hours per week, that is, 60 percent of the median hours of leisure in the sample).

We calculate the minimum necessary paid work time (NPT) in Equation 2(2) as follows: (4) where PL is the income poverty line; SIZE is the equivalent-scale of household size, namely, the square root of the number of individuals in the household; NINC is the household's nonlabor income; SHARE is an individual's share of total earned income by the household; W is the individual's hourly wage; and CT is the time spent on commuting to work. Because China has no official poverty line for urban households, we adopt a relative poverty line, which is 60 percent of the median equivalent-scale household income per capita in a province in the sample.⁴ The choice of the provincial median income instead of the full sample median income takes account of regional economic disparities. We also estimate time-poverty rates by using alternative time-poverty thresholds and income-poverty thresholds and find that the patterns of the alternative estimates are substantively similar to those presented in Tables 2 and 3.⁵

The time-poor individuals defined by TP1 include those who work long paid or unpaid hours, though not due to pressures to meet their households’ basic needs. TP2 and TP3 limit the time poor to those who need to work long paid hours to escape income poverty. Following Bardasi and Wodon (2010), TP3 defines the time poor as workers who are not only time poor, but also income poor or at risk of becoming income poor if these workers were to reduce their paid working time to below the minimum necessary paid working time for above-poverty living standards. One shortcoming of TP3 is that it overlooks the “hidden income-constrained time-poor” individuals, that is, those who are income poor but not time poor and who would become time poor if they were to increase their paid working time above the minimum working time for above-poverty living standards. These hidden time-poor individuals are unable to work as many hours as needed perhaps because of poor health or a lack of options regarding their work hours in the labor market. TP2 includes both actual and hidden income-constrained time poor, thereby presenting a more complete picture of the income–time tradeoffs that low-income workers face. In the remainder of the study, we use the terms “time poverty” for TP1, “income and time poverty” for TP2, and “income-constrained time poverty” for TP3.

DATA

We draw the data used in this study from the 2008 CTUS and the 2008 CHIP. Our analysis focused on women and men who are between ages 15 and 64 and who are employed in the urban sector. The 2008 CTUS is the first large-scale time-use survey in China. The survey covers 37,142 individuals between ages 15 and 74 years from 16,661 households in ten Chinese provinces. The survey collected time-use data using a time-diary approach. The survey asked the respondents to record what they did, where they were, and with whom they were spending time in each 10-minute interval for the previous 24 hours on a weekend day and a weekday. The recorded activities were subsequently post-coded into nine one-digit, sixty-one two-digit, and 113 three-digit categories. The nine one-digit categories include personal care and maintenance (0); paid employment (1); household production in primary industry (2); household-based production in manufacturing and construction industries (3); household-based services to generate income (4); housework for households’ own consumption (5); care for household members (children and the elderly, sick, or disabled), help to other households, and community volunteer services (6); study and training (7); and recreation, leisure, and social contact (8). The survey included traveling time in each one-digit category and coded it as a three-digit activity. In this study, we classify all activities into four categories: personal care (corresponding to one-digit category 0), paid work (corresponding to one-digit categories 1, 2, 3, and 4), unpaid care work (corresponding to one-digit categories 5 and 6), and leisure time (corresponding to one-digit categories 7 and 8).

The survey also gathered data on each respondent's age, sex, ethnicity, marital status, educational attainment, occupation, and income from the previous month. However, the information on income and household structure is inaccurate or incomplete. The survey reports monthly incomes in a categorical measure, and it lacks information on nonlabor income. As for household structure, the survey did not question household members younger than 15 years of age or older than 75 years of age. We can only infer the presence of household members ages 0–6, 7–15, or older than 75 from the respondent's report of “who you were with,” but we do not know the size and composition of the household. To calculate TP2 and TP3, we supplement the data from the 2008 CTUS with data from the 2008 CHIP.

The 2008 CHIP is the fourth wave of the CHIP survey, which aims to investigate poverty and income distribution in China.⁶ The survey provides rich information on employment, earnings, household incomes and assets, and household demographic structures. The 2008 CHIP covered 46,469 individuals from 13,001 households in nine Chinese provinces.

We statistically match the data for analysis from the 2008 CTUS and the 2008 CHIP using a procedure developed by Hyunsub Kum and Thomas Neal Masterson (2010).⁷ As major disparities exist on the level of economic development across provinces, ensuring that we match the observations to the correct province is important. Therefore, we restrict our data matching to five provinces that were covered by both surveys: Zhejiang, Anhui, Henan, Guangdong, and Sichuan. We describe the data-matching procedure in the Appendix.

The sample for analysis consists of 5,243 individuals who are between ages 15 and 64 and are employed or self-employed in the urban sector, with women accounting for 45.4 percent of the observations. We present the descriptive statistics of the variables used in regressions in Appendix Table A2.⁸

RESULTS

Time allocation and time poverty

Table 1 presents the patterns of time allocation for men and women workers in the labor market in urban China. Similar to their men workers, most women workers work for pay full time; the average weekly paid working hours of women workers is only 2.7 hours fewer than those of men workers. However, compared with men workers, women workers spend, on average, 11 more hours per week on unpaid care work. Thus, women workers work 8.7 more hours per week than men workers. As the statistics at the bottom of Table 1 indicate, working overtime is common in the workplace: approximately 24 percent of the workers in the labor market work more than 44 hours per week for pay and 19 percent work more than 48 hours for pay, and the gender gap in the overtime rate is small.

Table 1 Patterns of time allocation of men and women workers in urban China (hours per week)

	Men	Women
Actual time spent on main activities (hours/week)
Personal care	81.71(11.18)	81.15 (10.03)
Paid work (APT)	45.60 (15.99)	42.94 (14.66)
Net paid work time	37.57 (15.64)	35.69 (14.40)
Commuting time (CT)	8.03 (6.31)	7.25 (5.65)
Unpaid care work (AUPT)	9.94 (9.68)	21.30 (12.04)
Leisure time	30.74 (13.71)	22.61 (11.94)
% of workers overworked
Net paid work time (APT) >44	26.27 (44.02)	21.66 (41.20)
Net paid work time (APT) >48	20.98 (40.72)	15.97 (36.64)
Observations	2,851	2,392

Note: Numbers in parentheses are standard deviations.

Table 2 presents time-poverty rates for men and women workers in the labor market according to three alternative measures. For the overall sample, 27.4 percent are time poor (TP1); 22.3 percent are income and time poor (TP2); and 11.5 percent are income-constrained time poor (TP3). Comparatively, the time-poverty rate based on a much lower time-poverty line of 50 paid work hours per week is 18.7 percent in Guinea (Bardasi and Wodon 2010), and the time-poverty rate based on a relative poverty line of 60 percent of the median leisure time is 20 percent in the UK (Chatzitheochari and Arber 2012). The time-poverty rate appears higher in China than in Guinea and the UK.

Table 2 Time-poverty rates of men and women workers in urban China (%)

	Men	Women	Gap (women–men)
Total
TP1=1	18.91	37.58	18.67
TP2=1	17.05	28.64	11.59
TP3=1 (TP1=1 and TP2=1)	7.02	16.89	9.87
TP1=1, TP2=0	11.89	20.69	8.8
TP1=0, TP2=1	10.03	11.75	1.72
TP1=0, TP2=0	71.06	50.67	−20.39
Observations	2,851	2,392

By all three measures, time-poverty rates are much higher for women workers than men workers, with a gender gap of 18.7 percentage points by TP1, 11.1 percentage points by TP2, and 9.9 percentage points by TP3. Of the workers, 15.9 percent are time poor but not income poor (TP1=1 and TP2=0) and 10.8 percent are not time poor but income poor (TP1=0 and TP2=1). Interestingly, the gender gap in the former is noticeably larger than the latter (8.8 versus 1.7 percentage points), indicating that relative to men, women are more disadvantaged by time deprivation than by income deprivation.

We compare the time-use patterns of the time poor and the non-time poor, as defined by TP1, and find that for both women and men, the time poor spend more time on both paid and unpaid work than the non-time poor (see Table A3). The paid work hours of the time poor are extremely long, 65.6 hours per week for men and 52 hours for women; their total work hours are also extremely long, 79 hours per week for men and 78 hours for women. The gap in total work time between the time poor and the non-time poor is large: 29 hours per week for men workers and 22 hours for women workers.

Table 3 presents the time-poverty rates of low-paid and non-low-paid workers to elucidate the connection between time poverty and low earnings. Using the ILO definition, we define low-paid workers as those who earn less than 60 percent of the median wage of full-time workers in a province. In our sample, most low-paid workers are unskilled workers in manufacturing, construction, and service industries and migrant workers. Low-paid workers account for 17.2 percent of men workers and 24.9 percent of women workers in the sample. The high incidence of low-paid workers is indicative of low minimum wage standards. Indeed, the minimum hourly wages of the five provinces under investigation, on average, only amount to 17 percent of the mean and 27 percent of the medium hourly wage of our sample. As we would expect, low-paid workers have much higher time-poverty rates than non-low-paid workers, with a between-group gap of 32.7 percentage points for TP1, 39 percentage points for TP2, and 26.2 percentage points for TP3. Overall, 53.3 percent of the low-paid workers are time poor, 53.2 percent are income and time poor, and 32.3 percent are income-constrained time poor.

Table 3 Time-poverty rates of low-paid and non-low-paid workers, by gender

	Men	Women
TP1 (%)
Low-paid workers	46.53	58.89
Non-low-paid workers	13.17	30.51
TP2 (%)
Low-paid workers	49.80	56.04
Non-low-paid workers	10.25	19.54
TP3 (%)
Low-paid workers	26.33	37.25
Non-low-paid workers	3.01	10.13
Observations	2,851	2,392

Note: Number of low-paid workers is 1,086; number of non-low-paid workers is 4,157.

Correlates of time poverty

We explore the characteristics of time-poor men and women workers by estimating a linear probability model for three binary time-poverty indicators: TP1, TP2, and TP3. The explanatory variables of the regression model include individual characteristics, such as the individual's pay status, marital status, age, and education; household characteristics, such as the presence of preschool and school-age children, the presence of the elderly, and single- or double-earner household status; and county-level variables, such as GDP per capita in log form, overtime rates (the proportion of workers, including both urban and rural workers in the county, who spend more than 44 hours per week on paid work), minimum wages in log form, and the share of GDP from tertiary industry.⁹ For each time-poverty measure, we first separately estimate the model for men and women workers and then do so for men and women workers combined.¹⁰

Table 4 presents the ordinary least squares (OLS) estimates of the linear probability model of time-poverty determination. As indicated by summary statistics, other things being equal, women workers are more likely than men workers to be time poor, with a female effect of 16, 9, and 8 percentage points for TP1, TP2, and TP3, respectively. Moreover, the probability of being time poor is much higher for low-paid workers than for non-low-paid workers: for all workers as a whole, 29 percentage points higher for TP1, 39 percentage points higher for TP2, and 26 percentage points higher for TP3. Being a low-paid worker is a statistically significant correlate of time poverty for both sexes.

Table 4 OLS estimates of the linear probability model of correlates of time poverty in urban China

	TP1			TP2			TP3
	Men	Women	All	Men	Women	All	Men	Women	All
Variables	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
Female			0.16***			0.09***			0.08***
			(0.01)			(0.01)			(0.01)
Low-paid	0.32***	0.27***	0.29***	0.40***	0.39***	0.39***	0.23***	0.28***	0.26***
	(0.02)	(0.02)	(0.02)	(0.02)	(0.02)	(0.02)	(0.02)	(0.02)	(0.02)
Married	0.08**	0.11***	0.09***	0.01	−0.01	0.00	0.05**	0.02	0.04*
	(0.03)	(0.04)	(0.02)	(0.03)	(0.04)	(0.02)	(0.02)	(0.03)	(0.02)
Presence of children	0.08***	0.13***	0.10***	0.12***	0.21***	0.16***	0.07***	0.14***	0.10***
under age 6	(0.03)	(0.04)	(0.02)	(0.02)	(0.04)	(0.02)	(0.02)	(0.03)	(0.02)
Presence of children	0.03*	0.14***	0.08***	0.10***	0.17***	0.13***	0.02**	0.13***	0.07***
ages 7–15	(0.02)	(0.02)	(0.01)	(0.02)	(0.02)	(0.01)	(0.01)	(0.02)	(0.01)
Presence of elderly	−0.01	0.02	0.00	0.12***	0.13***	0.13***	0.04***	0.07***	0.05***
above age 74	(0.02)	(0.03)	(0.02)	(0.02)	(0.02)	(0.02)	(0.01)	(0.02)	(0.01)
Double earners	−0.00	−0.01	−0.00	−0.03**	−0.00	−0.02	−0.01	0.01	−0.00
	(0.02)	(0.03)	(0.01)	(0.01)	(0.02)	(0.01)	(0.01)	(0.02)	(0.01)
Ages 25–34	−0.02	0.09**	0.06*	0.03	0.07*	0.06*	0.03	0.07**	0.06**
	(0.05)	(0.04)	(0.03)	(0.04)	(0.04)	(0.03)	(0.03)	(0.03)	(0.02)
Ages 35–44	0.02	0.14***	0.09***	0.02	0.08**	0.05	0.03	0.08**	0.06**
	(0.05)	(0.04)	(0.03)	(0.05)	(0.04)	(0.03)	(0.03)	(0.03)	(0.03)
Ages 45–54	−0.02	0.12***	0.06*	0.02	0.08*	0.06*	0.00	0.07**	0.04
	(0.05)	(0.04)	(0.03)	(0.05)	(0.04)	(0.03)	(0.03)	(0.03)	(0.03)
Ages 55–64	−0.03	0.14	0.05	0.02	0.07	0.06	0.02	0.10	0.07**
	(0.05)	(0.09)	(0.04)	(0.05)	(0.07)	(0.04)	(0.03)	(0.06)	(0.03)
Lower secondary	−0.05	−0.13**	−0.09**	0.04	0.03	0.04	−0.04	−0.03	−0.04
school	(0.05)	(0.07)	(0.04)	(0.04)	(0.06)	(0.04)	(0.04)	(0.05)	(0.04)
Upper secondary	−0.06	−0.15**	−0.10**	0.10**	0.08	0.09**	−0.03	−0.00	−0.02
school	(0.05)	(0.07)	(0.04)	(0.04)	(0.06)	(0.04)	(0.04)	(0.05)	(0.04)
Junior college and	−0.10*	−0.21***	−0.15***	0.03	0.02	0.03	−0.06	−0.05	−0.05
above	(0.05)	(0.07)	(0.04)	(0.04)	(0.06)	(0.04)	(0.04)	(0.05)	(0.04)
County GDP pc (log)	−0.00	0.03	0.01	−0.03**	−0.02	−0.03*	−0.02	−0.00	−0.01
	(0.02)	(0.02)	(0.01)	(0.02)	(0.02)	(0.01)	(0.01)	(0.02)	(0.01)
Over-working rate	0.54**	0.91***	0.71***	−0.07	−0.06	−0.06	0.04	0.35*	0.17
	(0.21)	(0.28)	(0.18)	(0.20)	(0.24)	(0.18)	(0.14)	(0.21)	(0.13)
Minimum wage (log)	−0.04	−0.13***	−0.08***	0.02	0.02	0.02	0.00	−0.02	−0.01
	(0.04)	(0.05)	(0.03)	(0.03)	(0.04)	(0.03)	(0.02)	(0.04)	(0.02)
GDP share of service	0.10	0.18	0.14	0.07	0.15	0.11	0.07	0.10	0.08
industry	(0.12)	(0.16)	(0.10)	(0.12)	(0.15)	(0.11)	(0.08)	(0.13)	(0.07)
Constant	0.32*	0.55**	0.35**	0.20	−0.03	0.05	0.12	0.01	0.03
	(0.19)	(0.24)	(0.16)	(0.16)	(0.21)	(0.15)	(0.12)	(0.18)	(0.11)
R^2	0.12	0.12	0.15	0.21	0.20	0.21	0.14	0.16	0.16
Observations	2,851	2,392	5,243	2,851	2,392	5,243	2,851	2,392	5,243

Notes: Numbers in parentheses are heteroscadesticity-robust standard errors (clustered at household level). Columns 4, 8, and 12 are the coefficient estimates of interaction terms between female dummy and explanatory variables in regressions for all workers as a whole. ***, **, * denote statistical significance at the 1, 5, and 10 percent levels, respectively.

With respect to family structure, being married is statistically significantly correlated with time poverty for both sexes and with income-constrained time poverty for men workers. The presence of preschool-age children and school-age children increases the likelihood of time poverty for all three measures, and the children effect is significantly larger for women than for men in all cases except for the presence of preschool-age children for TP1. The presence of seniors who are older than 74 has a statistically significant positive effect on income and time poverty and income-constrained time poverty, but no effect on time poverty for both sexes, and the effect on income-constrained time poverty is slightly larger for men than for women (at the 10 percent level of significance). This finding means that workers with a greater responsibility to provide for elderly parents are more vulnerable to both income and time poverty.

Turning to human capital characteristics, we note that age is a statistically significant determinant of time poverty for women workers; compared with women ages 15–24 (the reference group), women ages 25–54 are more likely to be time poor, and women ages 35–54 are more likely to be income and time poor and income-constrained time poor. Interestingly, age is not a significant correlate of time poverty for men by all three measures. Thus, the double burden of paid work and unpaid care responsibility appears to be a problem particularly for women in the age group that generally contains mothers and younger grandmothers. Education is negatively correlated with time poverty, and the correlation is statistically significant for women. Unlike women, men's probability of being time poor is uncorrelated with education.

Regarding county-level variables, a county's GDP per capita has a statistically significantly negative association with the income and time poverty of only men workers. Workers in counties where the incidence of overtime work is more prevalent are more likely to be time poor. While it is statistically insignificant, the gender difference in overtime work effects is numerically large, 0.54 for men and 0.91 for women. Bearing the primary responsibility for unpaid care work, excessively long paid work hours make women workers more vulnerable to time poverty than men workers. Moreover, minimum wage standards have a significant negative correlation with time poverty for women. This result suggests that increasing minimum wage standards would have a larger protective effect for women workers than for men workers because women workers account for a larger share of the workers at the lower end of the wage distribution. Lastly, the share of tertiary industry is not significantly associated with any type of time poverty for either sex. In light of the fact that time poverty is concentrated among workers of low socioeconomic status, the expansion of market services does not appear to be an effective solution for time poverty.

Policy simulations

The finding that low-paid status, overtime work rates, and minimum wage standards are statistically significant correlates of time poverty suggests that labor market regulations may have a role to play in alleviating time poverty. Using these estimates, we simulate the potential effects of strengthening labor market regulations, and we present the results in Table 5.¹¹ For the minimum wage adjustment, we explore the effects of raising the minimum wage to 60 percent of the provincial median wage, which is in line with the policy target of the 12th Five-Year Development Program. This adjustment would reduce low-paid rates to zero and raise the mean minimum wage at the provincial level by 61 percent. Since low-paid status is statistically significant for all three indicators, this policy adjustment would affect all three types of time poverty. Simulations presented in Table 5 show that raising the minimum wage to 60 percent of the median wage would decrease time-poverty rates of women and men, respectively, by 13.8 and 7.9 percentage points for TP1, by 8.2 and 6.0 percentage points for TP2, and by 7.8 and 3.9 percentage points for TP3. It is not surprising that the effects of this policy adjustment are markedly larger for women than for men, given the fact that women account for a disproportionate share of low-paid workers.

Table 5 Simulation: Effects of labor market regulations on time poverty (%)

	Men	Women	Gap (women–men)
Actual time-poverty rates
TP1=1	18.9	37.6	18.7
TP2=1	17.1	28.6	11.6
TP3=1	7.0	16.9	9.9
Raising the minimum wage to 60 percent of provincial median monthly wage
Predicted time-poverty rates
TP1	11.0	23.8	12.8
Net change	−7.9	−13.8	−5.9
TP2	11.0	20.5	9.4
Net change	−6.0	−8.2	−2.2
TP3	3.1	9.1	5.9
Net change	−3.9	−7.8	−3.9
Effectively enforcing work-time regulations
Predicted time-poverty rates
TP1	12.7	24.3	11.7
Net change	−6.2	−13.2	−7.0

Note: Predicted time-poverty rates are derived by using the OLS estimates of low-paid status, overtime work rates, and minimum wage standards that are statistically significant.

We assess the effects of enforcing work-time regulations by setting the overtime-work rate per county under study to zero. This adjustment would only affect time poverty (TP1), since this variable is statistically significant only for this indicator. The simulation results presented at the bottom of Table 5 show that the effective enforcement of work time regulations would lower TP1 by 6.2 percentage points for men workers and by 13.2 for women workers. The larger effect for women is due to the fact that the coefficient estimate of overtime work rates is much larger for women than for men. The simulation results suggest that strengthening labor market regulations is an effective means of reducing time poverty, and it has a stronger protective effect for women than for men.

CONCLUSIONS

This study estimates time-poverty rates and analyzes profiles of the time poor among Chinese workers using synthetic data from two nationally representative surveys. Extending the work of Bardasi and Wodon (2010), the study introduces three time-poverty measures. By all three measures, time poverty is prevalent among Chinese workers. Women workers and low-paid workers are more likely to be time poor than men workers and non-low-paid workers, respectively. Regression analysis further shows that, other things being equal, workers who are women, low paid, and married and who live with children and/or the elderly in counties with higher overtime rates and lower minimum wage standards are more likely to be time poor. Moreover, women's care responsibility makes them more vulnerable to time poverty and more sensitive to the violation of work-hours regulations, compared with men workers.

To explore policy solutions, we simulate the potential effects of raising minimum wage standards and enforcing work-time regulations on time poverty. The simulation results confirm that raising minimum wages and enforcing work-time regulations can statistically significantly lower time poverty rates and narrow the gender time-poverty gap.

The findings of this study call for a comprehensive approach to address time poverty, which requires not only policy measures that reduce women's unpaid care work and encourage more equitable distribution of unpaid care work within the household, but also protective labor market regulations. The state should play a greater role in the provision of care services to make such services affordable for low-income families. The formulation of labor market regulations should strike a proper balance between firms’ concerns about competitiveness and workers’ needs to have adequate time to care for themselves and their dependents. These policy measures are crucial, not only to improve workers’ well-being and minimize socioeconomic inequalities, but also to sustain economic growth in the long run. The analysis also calls for more balanced development strategies, which pay equal attention to material and non-material aspects of human well-being.

ACKNOWLEDGMENTS

This work was carried out with the aid of a grant from the International Development Research Center of Canada (Project no. 107579) and a grant from the National Natural Science Foundation of China (Grant no. 71573146). The views expressed in this paper are those of the authors and do not in any way implicate the National Bureau of Statistics of China – the organization that conducted the 2008 CTUS. We also benefited greatly from the comments and suggestions of Fiona MacPhail on an earlier version of the paper.

Additional information

Notes on contributors

Liangshu Qi

Liangshu Qi is Associate Professor at the School of Economics and Management at Tsinghua University, China. Her research interests cover a wide range of issues, including regional development, rural finance, health economics, intrahousehold allocation, and gender equality. Her current research focuses on time-use analysis and public policies.

Xiao-yuan Dong

Xiao-yuan Dong is Professor in the Department of Economics at the University of Winnipeg, Canada, and Adjunct Professor at the National School of Development, Peking University, China. She has served on the board of directors of the International Association of Feminist Economics and is an Associate Editor of Feminist Economics. Her research focuses on China's economic transition and development, with an emphasis on labor and gender issues.

Notes

1 In this study, unpaid care work refers to housework and caregiving activities at home.

2 See “Measures for the Examination and Application of Non-Fixed Working Hours and Comprehensive Calculation of Working Hours in Enterprises,” an administrative regulation issued by the Ministry of Labor in 1994 as a supplement of the Labor Law (Available in Chinese on the website of the Ministry of Labor and Social Security, that is, the former Ministry of Labor at: http://www.mohrss.gov.cn/SYrlzyhshbzb/zcfg/flfg/gz/201705/t20170522_271153.html).

3 Like Bardasi and Wodon (2010), we do not specify the minimum necessary unpaid care work time due to data limitations. We assume that the amount of time spent on unpaid care work reflects the household's needs. Given that housework and care for family members consume time and energy, but have no explicit monetary reward, people would not spend time on these activities if they were not needed.

4 The threshold of 60 percent of the median income is adopted in parallel with the threshold of 60 percent of the median free time for time-poverty measurement. This income threshold is adopted in the UK and the European Union.

5 These alternative estimates of time poverty are available upon request.

6 The CHIP was conducted by a team of international and domestic economists in collaboration with China's National Bureau of Statistics.

7 Zacharias, Masterson, and Antonopoulos (2012) and Zacharias, Masterson, and Kim (2014) also supplemented time-use surveys with information from household surveys using statistical matching methods.

8 One shortcoming of the 2008 CTUS and 2008 CHIP is that migrant workers are underrepresented because both surveys are based on households, while a large proportion of migrant workers live in factory dormitories. Indeed, there are only thirty-seven migrant workers in our sample. Therefore, our analysis focuses on nonmigrant urban workers.

9 County-level data are collected from China's Statistical Yearbook of various issues, statistical communiqués, and government websites.

10 To test the gender differences, we also estimate a model for all workers with a stand-alone gender dummy and its interactive terms with each explanatory variable. Gender difference is not statistically significant in most cases, except for the presence of preschool-age and school-age children and the elderly.

11 Strictly speaking, we cannot establish the counterfactual time poverty rates of the proposed policy adjustment by using the estimates of correlations. Nevertheless, the simulation results provide valuable insights into the scope of policy reforms.

APPENDIX

Data matching procedure

Prior to data matching, the CTUS data file and the CHIP data file had 8,877 and 6,896 individuals, respectively, between ages 15 and 74 years in the urban sector in the five provinces under study. Data matching seeks to assign each individual in the CTUS data file statistically accurate information on labor earnings, household income, and household size. We separately match for individuals and for households. To match for individuals, we first divide individuals in the CTUS and the CHIP data files by sex, working status (if they have a stable job), and province to form twenty pairs of matching cells. We then perform propensity score matching on each pair of corresponding matching cells. The variables used to calculate propensity scores include age, years of schooling, and income categories.

To match for households, we first divide households in the two data files by province to form five pairs of matching cells. We then perform propensity score matching on each pair of corresponding matching cells. The variables used to calculate propensity scores include the number of household members who bring in labor income, the income categories of the primary and the household's secondary income earners, the number of adults ages 20 to 65 in the household, the presence of children ages 0 to 6 in the household, the presence of children ages 7 to 15 in the household, and the presence of elderly household members ages 75 and older.

In terms of the matching algorithm, observations in the CTUS are treated as dependent variables (CTUS=1, CHIP=0). We use the nearest neighbor rule with replacement, which matches an observation from the CHIP file as a partner for an observation in the CTUS file that is the closest measured by propensity score, and observations from the CHIP to be used as a match more than once, thus making the order of matching irrelevant. The value of the variable of interest for the individual from the CHIP is assigned to its matching partner from the CTUS. We present summary statistics of the matching results in Table A1. These statistics indicate that the data from the two surveys are matched very well, with the probability of exact matching higher than 90 percent for all relevant variables except for age.

Table A1 Data matching results

Comparison of the matching variables
	CTUS	CHIP	% of exact matching
Individual income category	4.16 (0.88)	4.16 (0.88)	99.96
Age	39.48 (8.76)	39.45 (8.73)	65.21
Schooling year	12.65 (2.59)	12.65 (2.53)	97.73
Number of household members who have income	2.04 (0.55)	2.00 (0.48)	93.11
Income category of the primary income earner in the household	4.42 (0.81)	4.40 (0.82)	98.82
Income category of the secondary income earner in the household	3.52 (1.12)	3.52 (1.17)	98.97
Number of adults ages 20–65 in the household	2.09 (0.51)	2.13 (0.52)	90.48
Presence of children ages 0–6	0.10 (0.30)	0.09 (0.29)	98.28
Presence of children ages 7–15	0.36 (0.48)	0.34 (0.47)	90.92
Presence of elderly ages 75 and above	0.17 (0.38)	0.15 (0.36)	97.79
Descriptive statistics of the imputed variables
	Mean	S.D.	Min	Max
Individual labor earnings	2192.18	1730.75	80	20000
Household income	66965.25	58108.03	5080	480000
Household size	3.04	0.90	1	6

Table A2 Summary statistics of the variables involved in regressions by gender

	Men		Women
	Mean	S.D.	Mean	S.D.
Low-paid	0.17	0.38	0.26	0.43
Married	0.92	0.27	0.88	0.32
Presence of children ages 0–6	0.10	0.31	0.09	0.29
Presence of children ages 7–15	0.35	0.48	0.37	0.48
Presence of elderly above age 74	0.18	0.38	0.17	0.38
Double earners	0.62	0.49	0.73	0.44
Ages 25–34	0.22	0.41	0.30	0.46
Ages 35–44	0.40	0.49	0.42	0.49
Ages 45–54	0.29	0.45	0.21	0.41
Ages 55–64	0.07	0.25	0.02	0.12
Lower secondary school	0.17	0.38	0.16	0.37
Upper secondary school	0.35	0.48	0.39	0.49
Junior college and above	0.46	0.50	0.42	0.49
County GDP pc (log)	10.36	0.66	10.34	0.67
Overtime rate	0.14	0.04	0.14	0.04
Minimum wage (log)	6.36	0.34	6.35	0.34
GDP share of service industry	0.38	0.09	0.38	0.09
Observations	2,851	2,392

Table A3 Time allocation of time poor and non-time poor among men and women workers (hours per week)

	Men		Women
	Time poor	Non-time poor	Time poor	Non-time poor
Paid work	65.6 (39.0%)	40.9 (24.4%)	52.0 (30.9%)	37.5 (22.3%)
Unpaid work	13.4 (8.0%)	9.1 (5.4%)	26.0 (15.5%)	18.5 (11.0%)
Personal care	72.5 (43.2%)	83.9 (49.9%)	75.6 (45.0%)	84.5 (50.3%)
Leisure time	16.5 (9.8%)	34.1 (20.3%)	14.5 (8.7%)	27.5 (16.4%)
Observations	539	2,312	899	1,493

Note: Time poor refers to those who are in time poverty by TP1.