Women’s Labor Force Participation in Urban China, 1990–2010

Abstract:

Using a series of survey and census/mini-census data, this paper explores the trends of women’s labor force participation (WLFP) and estimates the key factors affecting women’s labor supply in urban China between 1990 and 2010. Our results show that WLFP in urban China had changed dramatically from 1990 to 2010. The rates of WLFP had dropped drastically during the 1990s and early 2000s. In 2003, WLFP showed a steady trend of increase. Since 2003, WLFP had become more and more responsive to incomes from other family members. Specifically, women from poorer families have been more likely to join the labor force over time. We argue that Chinese women’s labor supply had been driven by different social or economic forces in different historical stages. From 1990 to 2003, the sharp decline of WLFP may have been caused by institutional transformation (market transition) starting from early 1980s and the radical layoff policy of publicly-owned enterprises (POEs) in the late 1990s. The steady increase of WLFP since 2003 may have been driven by the surge of living costs (e.g., housing, education, and healthcare) in urban China.

Notes

Since the early 1990s, the Chinese government had carried out the reform of publicly- owned enterprises (POEs), including the state-owned enterprises (SOEs) and collective-owned enterprises (COE). One of the most influential policies is the layoff policy, which is about the retrenchment of millions of workers who had been considered as surplus. As a result, a huge number of urban workers had been laid off or forced to retire earlier than the legal retirement age.

According to Naughton (2007), between 1993 and 2003, about 30 million SOE workers were laid off, 38 percent of the entire labor force, and almost 50 million urban workers of all kinds including workers of collective owned enterprises.

See Wu and Zhang (2010) and Yeung (2013) for reviews of the background and consequences of the reform of education (especially higher education).

See Ma, Lu, and Quan (2008) and Zhang and Kanbur (2005) for detailed information about the China’s healthcare reforms.

The layoff policy ended in 2003 for most provinces in China. According to a previous study, the nationwide rate of labor force participation (LFP) reached the lowest level in 2003, and in 2004, showed a trend of increase (Cai 2008).

Other factors may also contribute to the decreasing effect of education on employment over time. For example, China has witnessed a dramatic expansion in higher education during the late 1990s and early 2000s. As a possible consequence of the expansion, the supply of highly educated labor may surplus the demand, which could lead to the predicament of employment for college graduates (this happened in recent years).

See Wu and He (2015) for detailed information about 1990 and 2000 Chinese census data.

The Chinese General Social Survey (CGSS) data was collected jointly by the Survey Research Center of Hong Kong University of Science and Technology and the National Survey Research Center at Renmin University of China. See Bian and Li (2012) for the research agenda, sample design, and other information from this project.

We followed the previous study (Wu 2010b) to make the age restriction, considering this range as the prime working ages for women in urban China.

In this section, we presented figures to demonstrate the trends of women’s labor force participation (WLFP). Please refer to Appendix 1 for specific numbers in each year.

The continuous decrease of LFP for women with a college degree may be partly due to the rapid expansion of higher education and the predicament of employment of recent college graduates.

We excluded the 1990 and 2000 census data from our analysis in this part because the Chinese Population Censuses in 1990 and 2000 had not collected personal or family income information.

According to Allison (1999), comparing logit coefficients across groups (datasets) may be problematic because the residual variances of the equations for different groups (datasets) may be different. Researchers propose various ways for dealing with these issues (Williams 2009; see also Hong 2014 for a review), but unfortunately the proposed ways have problems of their own (Williams 2009), and make the comparison process very complicated (Hong forthcoming). We use the linear probability model (LPM) as an alternative modeling strategy to check if the pattern still holds. Using the ordinary least squares (OLS) procedure, LPM assumes that the probability of LFP is linear. We find that the LPM models produce the same pattern as that of the logistic models (results are not shown but are available from the authors on request). That is, the magnitude of human capital (years of schooling) decreases over time between 2003 and 2010. In the current study, we prefer logistic regression models to LPM because LPM violates the linear regression assumptions.

e^0.207 = .23, meaning that, other things being equal, having one additional year of schooling increases the odds of entering the labor market by 23 percent.

The pattern still holds if we use LPM to do the estimation (coefficients of LPM models are not shown but are available from the authors on request).

Additional information

Notes on contributors

Yuxiao Wu

Yuxiao Wu is a professor in the Department of Sociology, School of Social and Behavioral Sciences at Nanjing University, Nanjing, China. His research interests include social stratification and mobility, labor markets, sociology of education, and quantitative research methods.

Dongyang Zhou

Dongyang Zhou is a Ph.D. candidate in the Department of Sociology, School of Social and Behavioral Sciences at Nanjing University, Nanjing, China. His research interests include social stratification and inequality, with a focus on gender inequality in higher education in China.