A distributional analysis of wage discrimination against migrant workers in China’s urban labour market

Abstract

Chinese internal migrants without a local hukou (household registration) are often discriminated against in the urban labour market. This study examines the impacts of such discrimination on wage differentials and the distribution among urban locals, urban migrants and rural migrants. It uses an extended analytical framework of segmented labour market to examine the multiple segmentations between urban residents and rural migrants and between locals and non-locals. The results show that, compared with urban locals, rural migrants only face discrimination above the medium-wage level, while urban migrants face discrimination below the medium-wage level, but to a much lesser degree. Owing to structural differences in employment, urban locals (rather than migrant workers) are discriminated against at other wage levels. The results suggest that the hukou system still plays an important role in segmenting China’s urban labour market. The degree of discrimination against urban migrants relative to urban locals is greater than that against rural migrants relative to urban migrants. This suggests that nowadays China’s urban labour market is mainly characterised by the segmentation between locals and non-locals, rather than the segmentation between urban residents and rural migrants, which was the case in the past.

Keywords

China migrant workers urban labour market wage discrimination

Introduction

A massive population migration into cities has been one of the most significant socio-economic transformations in post-reform China. In the past three decades, the so-called floating population, who have left their origin of hukou (household registration) for six months or more, has increased from 7 million in 1982 to 221 million in 2010 (National Bureau of Statistics of China (NBSC), 1982, 2011). The floating population includes urban migrants and rural migrants. Urban migrants possess urban hukou in cities other than their destinations, while rural migrants possess rural hukou from the countryside.

The floating population has contributed approximately 21% of GDP growth and 75% of urbanisation growth between 1978 and 1999 (Cai and Wang, 1999; Zhang and Song, 2003). However, the majority of rural migrants are treated differently from urban locals in the urban labour market (Guo and Iredale, 2004; Roberts, 1997). Rural migrants tend to earn a lower wage than urban locals, even in the same job (Lu and Song, 2006; Sylvie et al., 2008). In 2004 the average monthly wage of urban locals was 1335 yuan, or 2.48 times that of rural migrants (Yao et al., 2008). Although in recent years the overall wage level has gradually increased, the wage gap between urban locals and rural migrants has further widened. A survey of rural migrants in 2009 showed that the monthly wage of urban locals was 6394 yuan, which was 4.51 times higher than that of rural migrants (NBSC, 2009). Wage differentials are also significant among migrant workers. Another survey in 2009 showed that the income of urban migrants was 1.54 times that of rural migrants (National Population and Family Planning Commission of China (NPFPC), 2010).

China is not the only country experiencing large-scale migration resulting from industrialisation and urbanisation, and inequalities between urban locals and migrant workers. Developed economies such as the USA and the UK have also seen significant inequalities in occupational attainment, and differentials in wages between racial groups and between immigrants and natives (Massey et al., 1994; Zhou and Logan, 1989). The human capital theory suggests that these inequalities are due to the heterogeneity in productivity-related characteristics such as age, education, working experience and so on (Schultz, 1961). The empirical literature, however, suggests that wage differentials remain even after controlling for these productivity-related characteristics. A plausible explanation is that the remaining wage differentials are the products of discrimination (Borjas, 1995; Gustafsson and Li, 2000).

In developed countries it is well documented that native and immigrant workers work in different segements of the labour market. Immigrant workers work in the secondary sector, with limited opportunities for career advancement, and experience significant occupational discrimination (Constant and Massey, 2005; Kim and Tamborini, 2006). Several studies on income inequalities between native and immigrant workers and between racial groups in developed countries show that approximately 10–60% of wage differentials cannot be explained by the human capital gap, and that labour market discrimination is more intensive at low-income level (Arabsheibani and Wang, 2008; Guenter, 2000; Juan and César, 2008). These studies suggest that racial or immigrant status usually induces mistreatment and discrimination in the labour market.

This paper contributes to the literature in several ways. First, this study extends the empirical literature on labour market segmentation and discrimination against migrants in China, which has shown some distinctive features compared with those in developed economies. For example, although China’s urban labour market has gone through tremendous changes over the last 30 years of economic reforms, it remains relatively underdeveloped and informal. In addition, the hukou system has fostered a segmented urban labour market, which has resulted in disadvantaged socio-economic status for migrant workers (Goldstein and Goldstein, 1991; Knight and Song, 1999). Overall, this research area is underdeveloped despite some existing studies. For example, some studies argue that the institutional segmentation and the discrimination imposed by the hukou system are among the most important reasons for the wage differentials between urban locals and migrant workers (Dong and Bowles, 2002; Lu and Song, 2006; Zhang, 2006). Several decomposition studies suggest that approximately 25–50% of the earnings gap between rural migrants and urban locals is attributable to unexplained reasons, such as discrimination (Maurer-Fazio and Dinh, 2004; Meng and Zhang, 2001). Migrant workers are treated even more badly in the urban labour market, after taking into account urban locals’ non-wage bonus and non-financial benefits (see, for example, Lee, 2012).

Second, beyond existing research in China which mostly focuses on the mean wage gap between different groups, this study presents a distributional analysis of wage differentials. The distributional analysis of wage differentials has attracted significant attention over the last decade in the context of increasing wage inequalities in many countries. In China the overall Gini coefficient increased from 0.30 to 0.46 between 1978 and 2006 (Chen et al., 2010). In 2008 even larger inequalities are found among urban migrants and urban locals, who had Gini coefficients of 0.54 and 0.52, respectively; the Gini coefficient (0.43) among rural migrants was lower but still considerable (Guo and Cheng, 2010). However, little research has been conducted to estimate the degree of discrimination against migrant workers in wage distribution.

Third, this study empirically examines the theory that migrant workers may be more favoured by some employers who offer low-tier jobs, resulting in urban locals (rather than migrants) being discriminated against at low-wage levels in the urban labour market. It has been suggested that migrant workers are more qualified than urban locals in low-tier jobs (Meng and Zhang, 2001), but this suggestion has not been thoroughly examined in prior studies in terms of labour market outcomes such as wage distribution.

Finally, this study examines discrimination in a diversified, unique composition of migrant population in China. Most existing studies focus on the segmentation between urban residents and rural migrants. However, the proportion of urban migrants in the floating population has increased significantly in recent years (Zhang, 2007). Urban migrants have some advantages compared with rural migrants because of their urban hukou, but some disadvantages compared with urban locals owing to their non-local hukou. A recent official report on China’s migrant population development has begun to distinguish explicitly between urban migrants and rural migrants, and also discusses the general living conditions of the two groups (NPFPC, 2011). This has important implications for studying discrimination against migrant workers. However, few empirical studies have investigated the wage differentials and discrimination against urban migrants and rural migrants. This study extends the analytical framework of urban–rural dichotomy, which has been widely adopted in researching the mechanism of economic exclusion and discrimination against rural migrants in urban China (Knight and Song, 1999; Meng and Zhang, 2001). However, the framework either excludes urban migrants from the analysis, or does not distinguish urban migrants and rural migrants. Therefore it fails to adapt to the changing composition of migrants, which is more diversified than it was in the 1980s and early 1990s.

Existing literature and analytical framework

In the existing literature the discrimination theory and the segmented labour market theory are widely adopted to examine labour market discrimination. The discrimination theory suggests that discrimination against a particular group originates from personal taste (Becker, 1957), or is the result of imperfect information (Aigner and Cain, 1977). For instance, some employers may prefer not to deal with a certain group. With imperfect information, some employers rely on easily observable characteristics, such as gender and race, to discriminate against a particular group on the basis of their previous experience or prevailing social norms.

The segmented labour market theory suggests that the source of labour market discrimination is a lack of competition due to social structures and institutional arrangements (Cain, 1976). It suggests that the labour market is segmented into primary and secondary labour markets (Doeringer and Piore, 1985). The primary segment is characterised by good pay, well-defined career ladders, favourable working conditions and job security, while the secondary segment is characterised by low pay, poor working conditions and a high mobility rate. Institutional constraints, such as regulations on migration, ethnicity, class relationship and residential status, restrict the movement between the two sectors (Smith, 2003). This results in a disadvantageous status of certain groups, who are largely confined to the secondary labour market (McDonald and Solow, 1981). The bulk of research has found sustained discrimination against minorities based on the segmented labour market theory (Ashenfelter, 1970; Gustafson et al., 1975).

A segmented labour market has also been observed in urban China (Appleton et al., 2004; Knight and Yueh, 2009). In the planned economy, China’s urban and rural labour forces were isolated owing to the hukou system that prohibited migration. While urban residents were entitled to employment, free housing and social security, rural residents were confined to the countryside and not allowed to migrate to urban areas (Sun and Fan, 2011). After the opening-up reforms, in order to develop private and informal sectors, a new labour administration was needed to support rapid economic growth in the cities (Fan, 2003). To attract cheap labour, rural residents was allowed to migrate temporarily and work in urban industries. From 1989 to 1993 the population of rural migrants increased from 30 million to 62 million (Li, 2008).

Although the restrictions on rural–urban migration have been relaxed considerably, the hukou system still functions as an ‘invisible wall’ between urban locals and migrant workers, resulting in an urban–rural dichotomy in the urban labour market (Maurer-Fazio and Dinh, 2004). For instance, without urban hukou, rural migrants are socially and economically separated from – and considered inferior to – local residents (Laurence, 2002). A large number of migrant workers are employed in informal or low-tier jobs, while urban locals enjoy secure jobs with higher wages, better working conditions and access to more social services (Meng and Zhang, 2001). Moreover, rural migrants are socially and/or residentially segregated from locals by living in low-income, inferior or deprived housing areas (Song et al., 2008).

Since the late 1990s local governments have driven a new segmentation between locals and non-locals. The fiscal decentralisation in the 1980s gave local governments the authority to administer local finance, and gave them the responsibility for maintaining a good level of momentum for socio-economic development and social stability. The reform of state-owned enterprises since 1997 had resulted in millions of urban workers being laid off, and caused serious unemployment and fiercer competition in the urban labour market. Consequently, both local government and urban locals regarded migrant workers as competitors to urban workers or troublemakers who brought in instability.

Local governments implemented discriminatory policies against migrants in response to the unemployment problem (Appleton et al., 2004). For example, some cities prohibited the employment of migrant workers in certain occupations, and even forced some enterprises to lay off migrant workers in favour of urban locals (Cai et al., 2001). These regulations were terminated recently after a series of laws and regulations issued by the central government that explicitly required local governments to enforce equal pay and equal job opportunities for migrant workers. However, local governments’ objectives of maintaining the segmentation between locals and non-locals has shifted from protecting employment opportunities for urban locals to shielding social security and welfare benefits for urban locals.

In view of the changing labour market, an integrative analytical framework, which examines the segmentation between urban residents and rural migrants and the segmentation between locals and non-locals, has been developed to study the socio-economic status of migrant workers (Guo and Zhang, 2012). This analytical framework suggests that, first, the urban–rural dichotomy in the urban labour market has largely disintegrated since the early 2000s, because the central government has shifted the policy focus from restricting migrants to integrating them in the cities. For instance, after the substantial reforms in hukou, employment and social security systems, most laws and regulations that restricted rural–urban migration and rural migrants’ occupational choices have been abolished. Nevertheless, the depth, scope and pace of hukou reform have not met many scholars’ expectations (Chan, 1996; Liu, 2005). It seems that the segmentation will linger and continue to play a role in shaping urban inequality and social stratification (Wang et al., 2002). Second, in the past 20 years, the segmentation between locals and non-locals has become increasingly dominant in the segmentation of China’s urban labour market as a result of the growing local protectionism in favour of locals.

Under this new analytical framework, urban migrants are also included in the analysis by using the so-called three-group analytical approach. This approach investigates the wage differentials among three groups, in other words, between urban locals and urban migrants, between urban locals and rural migrants, and between urban migrants and rural migrants. The wage differentials between the two groups are decomposed into two components. The explained portion measures the wage differentials due to the difference in mean productivity-related characteristics between the two groups, and the unexplained portion, which is often attributed to discrimination, measures the wage differentials due to the difference in returns to productivity-related characteristics between the two groups.

The three-group analytical approach better considers the reality of the two types of segmentation in China’s urban labour market and their interactions. The segmentation between urban residents and rural migrants still affects rural migrants, while the segmentation between locals and non-locals affects both rural migrants and urban migrants who are non-locals in their destinations.

To examine which type of segmentation plays a major role in the segmented urban labour market, the hypotheses of this study are:

H1: If urban migrants have a similar pattern of wage distribution to urban residents, while having a significantly better pattern of wage distribution than rural migrants, the segmentation between urban residents and rural migrants plays a major role in the urban labour market.

H2: If urban migrants have a similar pattern of wage distribution to rural migrants, while having a significantly worse pattern of wage distribution than urban locals, the segmentation between locals and non-local plays a major role in the urban labour market.

By examining these hypotheses, this study investigates to what degree discrimination against rural and urban migrants contributes to the wage distribution differentials.

Data and descriptive analysis

This study uses data from the research project entitled Migrant Labour in Large Chinese Cities. This project was funded by the Australian Research Council and conducted by researchers at Macquarie University (Sydney, Australia) and Nankai University (Tianjin, China) in Beijing, Shanghai, Tianjin and Guangzhou in 2008. The survey adopted a multi-stage stratified random sampling method. All districts in the four cities were taken as a sample frame. In each city, one urban and one suburban district were randomly selected; then two neighbourhood committees (juweihui) were randomly chosen from each district. Finally, 100 randomly selected households in each selected neighbourhood committee, including both local households and migrant households, were interviewed with the aid of questionnaires to collect individual and household data. The information collected in this survey includes individual members’ personal characteristics, wage and employment status, and so on. The survey finally received 1797 valid questionnaire responses. Among them, 1017 were rural migrants (57%), 378 were urban migrants (21%) and 397 were local residents (22%) (see Table 1).¹

Table 1.

Sample distribution by cities and groups.

	Urban locals	Urban migrants	Rural migrants	Subtotal (%)
Beijing	112	151	197	460 (25.67%)
Guangzhou	158	120	164	442 (24.67%)
Shanghai	68	45	281	394 (21.99%)
Tianjin	59	62	375	496 (27.68%)
Subtotal (%)	397 (22.15%)	378 (21.09%)	1017 (56.75%)	1792 (100%)

Considering the fact that a substantial proportion of migrant workers have to work overtime with little or no extra payment (Chan, 2002), we examined the hourly wage of urban locals and migrant workers. The average hourly wage is based on self-reported data in the survey, which excludes any non-cash income.² Table 2 presents the mean hourly wage and productivity-related characteristics for the three groups. The hourly wage of urban locals was 19.84 yuan, which was 1.27 times that of urban migrants and 2.19 times that of rural migrants.

Table 2.

Summary statistics.

	Urban locals (mean/percentage)	Urban migrants (mean/percentage)	Rural migrants (mean/percentage)
Hourly wage (yuan)	19.84	15.63	9.06
Male	49%	51%	59%
Age (years)	39.6	36.7	34.0
Education (years)	13.1	13.3	9.5
Job tenure (months)	96.5	39.7	52.8
Has formal job training	74%	56%	40%
Has labour contract	78%	57%	37%
Type of employer
State agency/state-owned enterprise (SOE)	50%	17%	10%
Collective-owned enterprise (COE)	8%	5%	4%
Private enterprise	22%	44%	37%
Self-employed	13%	30%	43%
Others	7%	4%	6%
White collar employee	51%	40%	15%
City
Beijing	28%	40%	19%
Tianjin	15%	16%	37%
Shanghai	17%	12%	28%
Guangzhou	40%	32%	16%

In this study, low, medium and high wages refer to an hourly wage below the 0.3 quantile, between the 0.3 and 0.7 quantile, and above the 0.7 quantile, respectively. Figure 1 presents the density and cumulative density of hourly wage distribution across the groups. The density diagram shows that the proportion of rural migrants with low and medium wages was higher than that of urban migrants and urban locals, while the proportions of urban locals and urban migrants with high wages were basically identical, and higher than that of rural migrants. The cumulative density diagram shows that urban locals earned a slightly higher hourly wage than urban migrants at most wage levels, with a much higher hourly wage than rural migrants. The wage differentials between urban locals and rural migrants increased gradually with the increase in wage levels. Results from a Wilcoxon rank-sum test (see Table 3) further show that the distributional wage differentials were significant between urban locals and rural migrants, and between urban migrants and rural migrants. This implies that migrant workers are not treated equally to their urban counterparts in earnings, and that rural migrants have the most disadvantaged wage status. The wage outcomes seem to support the notions of segmentation between urban residents and rural migrants, and segmentation between locals and non-locals.

Figure 1.

Density function and cumulative density function of logarithmic wage by three groups.

Table 3.

t-test on the differences in hourly wage and productivity-related characteristics.

	Rural migrants vs. urban locals		Urban migrants vs.urban locals		Rural migrants vs. urban migrants
	t	Pr(\|T\| > \|t\|)	t	Pr(\|T\| > \|t\|)	t	Pr(\|T\| > \|t\|)
Hourly wage^a	8.1560	0.0000	−0.1020	0.9190	8.6260	0.0000
Gender	−2.7702	0.0059	−0.5830	0.5603	2.0263	0.0434
Age	7.2402	0.0000	11.6082	0.0000	3.5156	0.0005
Education	−10.5556	0.0000	0.0492	0.9608	9.9749	0.0000
Job tenure	7.0792	0.0000	9.7509	0.0000	4.1172	0.0000
Training	9.4402	0.0000	5.0803	0.0000	−3.4511	0.0006
Contract	10.9863	0.0000	5.7618	0.0000	−3.7043	0.0002
Employer	−13.0063	0.0000	−9.3426	0.0000	2.0626	0.0399
White collar	9.1071	0.0000	1.7647	0.0784	−6.6333	0.0000
City	−9.7692	0.0000	1.8467	0.0652	−11.4824	0.0000

Note:

The Wilcoxon rank-sum test is used for the t-test on hourly wage.

As aforementioned, however, the wage differentials between each two groups may be in part determined by the differences in productivity-related characteristics. The statistics and t-tests for personal characteristics and employment status show that, at the 5% level, rural migrants’ personal characteristics and employment status were significantly different from that of urban locals and urban migrants (see Table 3). Nevertheless, between urban locals and urban migrants, the only significant differences were in age, job-related training, having a labour contract or not, type of industry and type of employer. Even if the abovementioned information were clear, it is still difficult to identify to what degree productivity-related characteristics and labour market discrimination contribute to the wage differentials along the distribution. To disentangle these two components, we employed quantile regression and decomposition approaches, which are discussed in the next section.

Econometric models

The Oaxaca-Blinder decomposition is one of the most widely adopted approaches for quantifying the contributions of group differences in productivity-related characteristics and discrimination to the mean wage differentials (Oaxaca, 1973). However, this approach is not effective for examining the distributional effects underlying wage differentials.

In recent years, to study the substantial increase in wage inequality in many countries, the original decomposition method has been improved and extended to examine distributional parameters beyond the mean (Fortin et al., 2010). More recently, the quantile-based regression and decomposition approaches were developed to untangle the sources of wage distribution differentials by Machado and Mata (2005). These approaches can reveal details of discrimination against migrant workers across the distribution of wages, particularly at the tails of the distribution. Since migrant workers are more preferred than urban locals in low-tier jobs (Meng and Zhang, 2001), the quantile approaches are appropriate for analysing whether migrants at low wage levels are being discriminated against.

Based on the Mincer earning function, the quantile regression model can be expressed as follows:³

Q_{\ln w_{i}} (θ | x_{i}) = x_{i} β_{θ} {+ ε}_{θ}

(1)

where lnw_i is the hourly wage (in logarithm) and x_i is a vector of personal characteristics and employment status for the ith respondent; β_θ is a vector of estimated coefficients of variables at the θ quantile; and ε is the error term.

There are two potential selectivity issues in earnings regression. One is the non-random occurrence of labour market participation, which means that not all people who can work are willing to find a job in the labour market. The selectivity problem would be prominent in a population in which the labour force participation rate is low, such as Australia and Germany in 1980s (Blau and Kahn, 2003; Miller and Rummery, 1991). Since the participation rates of urban locals and migrant workers in China are all higher than 80% (Park and Wang, 2010), this selection issue should not be problematic. One recent study on China using the standard Heckman selectivity correction also suggests that the labour participation process had no significant effect on earnings (Lee, 2012). The other issue relates to the fact that migrant workers are not random samples drawn from a rural population. This may bias the estimation of wage discrimination. Rural migrants usually have a higher earning capacity because they have better human capital and they are more motivated, resilient and ambitious than non-migrants in the countryside. Therefore, the wage discrimination might be even greater if rural migrants are randomly drawn in the rural areas. However, neither the present nor any other studies on migrants’ wages in China offer sufficient information to address this particular selection issue in econometric modelling because of the nature of observational data.

We employed the quantile decomposition method to estimate the degree of discrimination:

Δ_{θ} \ln w_{i j} = Q_{θ} (\ln w_{i}) - Q_{θ} (\ln w_{j}) = [Q_{θ} (\ln w_{i}) - Q_{θ} (\ln w_{i - j})] + [Q_{θ} (\ln w_{i - j}) - Q_{θ} (\ln w_{j})]

(2)

where $Δ_{θ} \ln w_{i j}$ represents the raw wage differentials between group i and group j at the θ quantile; $Q_{θ} (\ln w_{i - j})$ is the counterfactual wage distribution; that is, the conditional wage distribution of group j if they had identical marginal returns on various skills as has group i.

The first term on the right-hand side of the equation is the explained part of the wage gap that is attributable to the group differences in personal characteristics and employment status, and the second term is the unexplained part of the wage gap that is attributable to group differences in returns on productivity-related characteristics. This unexplained portion of the wage gap is often attributed to discrimination in labour economics.

To be consistent with other decomposition methods, a practical concern associated with quantile decomposition employed in this study is the index number problem or ‘path dependence’ in modelling (see examples in Bourguignon and Ferreira, 2005; Démurger et al., 2009). The decomposition results may differ, depending on which group’s wages are chosen as the non-discriminatory norm. To ensure the robustness of results, this paper addresses this issue by reporting a simple average of different sets of decomposition results.

Results

Although China’s urban labour market has undergone considerable reform in the last 30 years, the legacy of the hukou system, including the segmentation between urban residents and rural migrants and the segmentation between locals and non-locals, continues to play an important role in determining earnings and eligibility for access to employee benefits. Both urban migrants and rural migrants might suffer wage discrimination to different degrees, because they do not have local or urban hukou. This section presents the results from quantile regression of wage determinants and wage decomposition to estimate the degree of discrimination against migrant workers in wage distribution.

Multivariate analysis of wage determinants

Before proceeding to the wage decomposition, it is necessary to compare the wage determinants across different groups. Table 4 presents the estimated effects of independent variables on the hourly wage of urban locals, urban migrants and rural migrants. The constant terms for urban locals are greater than that for urban migrants and rural migrants at most wage levels, indicating that there exists a large unexplained wage premium for urban locals.

Table 4.

Quantile regression on hourly wage by three groups.

	Quantiles
	0.1	0.3	0.5	0.7	0.9
Urban locals
Male (ref: female)	0.2248**	0.2660***	0.2615***	0.2131**	0.4251***
Age	−0.0152	−0.0140	−0.0050	0.0499	−0.0724
Age²	0.0000	0.0001	0.0000	−0.0006	0.0010*
Education (ref: primary school and below)
College and above	0.4580	0.4369	1.0588***	1.2111***	1.1820***
Senior high school	0.2250	0.0756	0.4838**	0.5363	0.5111
Junior high school	0.1638	−0.0982	0.3636	0.6660*	0.7226*
Has training (ref: no)	0.0881	−0.0090	−0.1005	−0.2036	−0.4406**
Has contract (ref: no)	0.2025	0.0626	0.1194	0.1229	0.2108
Job tenure	0.0007	0.0010*	0.0013**	0.0008	0.0006
Employer (ref: self-employed)
State agency/SOE	−0.3455*	−0.1639	−0.5180**	−0.5149**	−0.9968***
COE	−0.5998**	−0.3161	−0.5301**	−0.6445**	−0.9869**
Private enterprise	−0.2591	−0.1144	−0.3615	−0.3654	−0.5605
Others	−0.5522*	−0.3625	−0.4876	−0.4360	−0.9562
White collar (ref: no)	0.0404	0.0891	0.1255	0.0810	0.1573
City (ref: Tianjin)
Beijing	0.3281***	0.3127***	0.3622**	0.6497***	0.4314**
Shanghai	0.4041***	0.5293***	0.4740***	0.6603***	0.2449
Guangzhou	0.2144	0.5926***	0.6457***	0.9393***	0.6730***
Constant	1.3500	1.5828**	1.3620**	0.3662	3.6484***
R ²	0.1496	0.2239	0.2407	0.2476	0.3110
Observations	326
Urban migrants
Male (ref: female)	0.0838	0.0524	0.0601	0.1398**	0.0541
Age	0.0991**	0.1269***	0.1378***	0.0783**	0.0878
Age²	−0.0013**	−0.0016***	−0.0018***	−0.0009*	−0.0010
Education (ref: primary school and below)
College and above	0.8919*	0.4616	0.3933	0.7695**	0.9723**
Senior high school	0.6251	0.1145	−0.0130	0.3657	0.8465**
Junior high school	0.4891	−0.1223	0.0317	0.1636	0.7732**
Has training (ref: no)	0.2367**	0.2170***	0.2597**	0.2457**	0.2951
Has contract (ref: no)	0.2533*	0.1931**	0.1984**	0.0154	−0.2508
Job tenure	0.0000	0.0000	0.0000	0.0000	0.0000
Employer (ref: self-employed)
State agency/SOE	−0.0131	−0.1031	0.0499	−0.1739	−0.6039**
COE	0.1385	−0.0111	−0.1024	−0.3426***	−0.8499**
Private enterprise	−0.0733	−0.0277	−0.0956	−0.3073**	−0.3484
Others	0.1607	0.0337	−0.1631	−0.1466	−0.1326
White collar (ref: no)	0.4151***	0.2265**	0.1858**	0.2357**	0.8127***
City (ref: Tianjin)
Beijing	0.2579	0.4721***	0.5818***	0.4966***	0.7460***
Shanghai	0.6213**	0.7975***	0.7821***	0.6219***	0.6731***
Guangzhou	0.2095	0.3290***	0.5875***	0.5633***	0.9159**
Constant	−1.5245	−1.2379	−1.1679	−0.0096	−0.0476
R ²	0.2663	0.2630	0.2228	0.2330	0.2219
Observations	324
Rural migrants
Male (ref: female)	0.2966***	0.1600***	0.1090***	0.0558	0.0566
Age	0.0523**	0.0268	0.0070	0.0010	0.0548***
Age²	−0.0008**	−0.0004*	−0.0002	0.0000	−0.0007***
Education (ref: primary school and below)
College and above	0.5745***	0.5878***	0.5197***	0.7304***	0.9445***
Senior high school	0.0796	0.2545***	0.1894***	0.2253**	0.2970**
Junior high school	0.0163	0.1232*	0.0689	0.1564**	0.1882
Has training (ref: no)	0.0564	0.1587***	0.1847***	0.1981***	0.1596
Has contract (ref: no)	0.1902**	0.1140**	0.1739***	0.1790***	0.1228
Job tenure	0.0010	0.0013**	0.0010**	0.0011**	0.0005
Employer (ref: self-employed)
State agency/SOE	−0.1204	−0.1234*	−0.2568***	−0.2834***	−0.3619*
COE	−0.1782	−0.2363**	−0.5019***	−0.6062***	−0.4006**
Private enterprise	−0.0729	−0.1398**	−0.2677***	−0.4011***	−0.4594***
Others	0.0046	−0.1694*	−0.2959***	−0.4685***	−0.6283***
White collar (ref: no)	0.3604***	0.3435***	0.3872***	0.2885***	0.2847**
City (ref: Tianjin)
Beijing	0.2548**	0.2943***	0.2928***	0.4192***	0.3180***
Shanghai	0.3883***	0.4481***	0.4005***	0.4468***	0.3634***
Guangzhou	0.0665	0.1034	0.1301	0.1961**	0.3950**
Constant	−0.0087	0.7102**	1.4254***	1.6864***	1.1764***
R ²	0.1633	0.1939	0.2139	0.1990	0.1833
Observations	879

Notes:

P < 0.1, **P < 0.05, ***P < 0.01.

Standard errors are omitted because of space constraints; they are readily available from the authors.

The significance of the gender variable reflects discrimination against females, which is more severe among urban locals. Being male increases the hourly wage for urban locals by 24–53% at all wage levels, while for rural migrants only by 11–35% below medium-wage level. This is consistent with recent findings that the degree of gender discrimination was lower in the private sector, where rural migrants are disproportionately concentrated, because private employers tend to reward productivity-related factors (Liu et al., 2000).

At most wage levels, the hourly wage of urban migrants and rural migrants follows an inverted U-shaped pattern as age increases, reaching a turning point at around 40 and 30 years for the two groups, respectively. The inverted U-shaped trend is more prominent among urban migrants than rural migrants. One potential reason is that most jobs for rural migrants are labour-intensive, low-skilled and physically demanding, and thus favour younger workers.

Education has a significant effect among urban locals and urban migrants at a high wage level, and rural migrants at all wage levels. This is consistent with the fact that the return to education was significantly higher for urban locals than for rural migrants because of the lower quality of education in rural areas (Maurer-Fazio and Dinh, 2004). Having college and above education nearly doubles urban locals’ hourly wages, and increases the hourly wage for urban migrants by 1.5 times and for rural migrants by 70–100%. In addition, the increase in return to education with wage levels suggests that the capabilities of high-wage earners are better utilised at work.⁴

Receiving occupational training has a positive effect on the hourly wage of migrant workers at most wage levels. Having job training increases the hourly wage by 25% for urban migrants and by 20% for rural migrants. One potential reason is that training can more significantly improve the productivity of urban migrants who already have better human capital.

Having a labour contract has a significant effect at most wage levels among rural migrants, and at and below medium wage levels among urban migrants. Signing a labour contract increases the hourly wage by 22–26% for urban migrants, but only by 11–19% for rural migrants. This is consistent with the fact that some rural migrants signed formal labour contracts possibly at the cost of higher wages (Cheng et al., 2013) or without more significant wage increases (Freeman and Li, 2013) after the implementation of the Labour Contract Law in 2008. Employers may have reduced employees’ wages in order to recover the additional non-wage benefits required by labour contracts. In addition, some employers increased fees for dormitory and meals and penalties for breaching rules or making mistakes at work (Wang et al., 2009). Consequently, the positive effect of a labour contract on earnings at other wage levels is offset, and migrant workers’ actual hourly wages may have even declined.

Current job tenure, which reflects work experience, significantly affects the hourly wage of urban locals and rural migrants at the medium-wage level. One additional month on a current job contributes to a 0.1% higher hourly wage. The type of employer has a significant effect at most wage levels among urban locals and rural migrants and at a high-wage level among urban migrants. Self-employed workers earn the highest hourly wage among all types. Relative to self-employment, working in state-owned enterprises decreases the hourly wage by 30–63% for urban locals, by 45% for urban migrants and by 11–30% for rural migrants. Working in collective-owned enterprises decreases the hourly wage for urban locals by 40–63%, for urban migrants by 30–57% and for rural migrants by 20–45%. Working in private enterprises decreases the hourly wage for urban migrants by 26% and for rural migrants by 13–37%. One explanation is that individuals with better endowments, especially those among migrant workers, are more likely to be self-employed in the informal sectors and obtain higher wages (Meng, 2001). In addition, the definition of wages in this study does not take into account the fringe benefits that urban locals and formal sector workers are more likely to enjoy (Knight and Yueh, 2004). Being a white collar worker increases the hourly wage for urban migrants by 20–125% and for rural migrants by 33–47%. The effect of locations on the hourly wage among urban migrants is generally greater than urban locals and rural migrants. This implies that besides economic factors, migrant workers’ hourly wages might also be affected by relevant employment policies imposed by local governments and other locality-specific factors.

The results of F-tests suggest that there are significant differences in the determinants of hourly wages between three groups at most wage levels.⁵ The structural difference in returns on productivity characteristics indicates that migrant workers are treated differently in the urban labour market. As mentioned above, the different treatment may be the outcome of labour market segmentation and prejudiced attitudes of urban residents and local governments. Based on the coefficients of the three groups presented in Table 4, we employed the quantile decomposition approach to evaluate the degree of wage discrimination.

Decomposition of distributional wage differentials

Table 5 presents the decomposition results of distributional wage differentials between urban locals and migrant workers. Surprisingly, urban locals rather than migrant workers face discrimination at some wage levels. The discovery of so-called ‘reverse discrimination’ in this study is different from some previous studies on mean wage differentials in urban China.

Table 5.

Quantile decomposition of distributional wage differentials between urban locals and migrant workers.

	Quantiles
	0.1	0.2	0.3	0.4	0.5	0.6	0.7	0.8	0.9
Rural migrants vs. urban locals
Raw wage differentials	0.19	0.22	0.24	0.27	0.36	0.31	0.41	0.47	0.48
Composition effects^a	0.20	0.28	0.32	0.35	0.34	0.35	0.37	0.31	0.38
Wage structure effects^b	0.18	−0.05	−0.02	−0.10	0.13	0.10	0.20	0.42	0.60
Interaction	−0.20	0.00	−0.05	0.03	−0.12	−0.14	−0.16	−0.25	−0.51
Discrimination (%)	97.07	−24.80	−8.20	−38.30	37.62	31.90	48.05	88.38	125.78
Urban migrants vs. urban locals
Raw wage differentials	−0.19	−0.14	−0.12	−0.09	−0.07	−0.04	−0.05	−0.05	−0.13
Composition effects^a	0.17	0.13	0.11	0.13	0.16	0.13	0.14	0.13	0.15
Wage structure effects^b	−0.04	−0.03	−0.02	−0.02	−0.02	−0.01	0.04	0.08	0.03
Interaction	−0.32	−0.24	−0.21	−0.21	−0.21	−0.15	−0.23	−0.26	−0.31
Discrimination (%)	18.74	22.15	17.16	16.26	32.52	28.51	−90.61	−154.87	−20.90
Rural migrants vs. urban migrants
Raw wage differentials	0.38	0.34	0.36	0.36	0.39	0.44	0.46	0.42	0.61
Composition effects^a	0.41	0.36	0.39	0.40	0.38	0.38	0.44	0.46	0.49
Wage structure effects^b	0.02	0.06	0.05	0.04	0.11	0.18	0.10	0.03	0.16
Interaction	−0.05	−0.07	−0.08	−0.07	-0.09	−0.13	−0.09	−0.07	−0.04
Discrimination (%)	6.05	16.31	13.85	9.78	27.61	41.28	21.65	7.04	26.57

Notes:

Composition effects indicate the explained part of the wage gap that is attributable to the group differences in mean personal and employment characteristics listed in Table 2.

Wage structure effects indicate the unexplained part of the wage gap that is attributable to the group differences in returns on productivity-related characteristics.

The degree of reverse discrimination (see the rows titled ‘Discrimination’) against urban locals compared with rural migrants ranges from 8.2% to 38.3% at or below medium-wage levels (at the 0.2–0.4 quantiles), while the degree of reverse discrimination against urban locals compared with urban migrants ranges from 20.9% to 154.87% at high-wage levels (at and above the 0.7 quantile). One explanation is that the employment structures for urban locals and rural migrants are different because urban locals shun most low-tier jobs. To reduce cost and increase efficiency, some employers prefer to employ rural migrants who are willing to do these jobs (see, for example, Fan, 2002; Wong et al., 2007). Consequently, employers who offer jobs below the medium-wage level discriminate against some urban workers. In contrast, urban migrants with better human capital are more qualified in high-wage and high-skilled jobs, and are thus more favoured by employers.

At other wage levels, the degree of discrimination against rural migrants compared with urban locals ranges from 31.9% to 125.78% at the 0.5 quantile and above. The latter percentage (125.78%), which is higher than 100%, indicates that the wage gap produced by discrimination could be greater than the raw wage gap. The degree of discrimination against urban migrants compared with urban locals ranges from 16.26% to 32.52% at the 0.6 quantile and below.

The higher degree of discrimination against rural migrants reflects the fact that they face discrimination arising from the segmentations between locals and non-locals and between urban residents and rural migrants, while urban migrants only face discrimination arising from the segmentation between locals and non-locals. At most wage levels, the degree of distributional discrimination against rural migrants estimated in this study is higher than that reported in previous research on mean wage differentials varying from 20% to 50% (Lee, 2012; Maurer-Fazio and Dinh, 2004). This indicates that the degree of discrimination is higher after taking into account the diversification in the composition of migrant workers and the wage inequality within each group. The results from this study may more accurately identify the mistreatment of migrant workers.

The decomposition within migrant workers (in other words, rural migrants compared with urban migrants) shows that, compared with urban migrants, the degree of discrimination against rural migrants is positive at all wage levels in the range of 6.05% to 41.28%. Referring to the hypotheses, the results show that the degree of discrimination against urban migrants compared with urban locals is greater than the degree of discrimination against rural migrants compared with urban migrants. This suggests that the segmentation between locals and non-locals is more dominant nowadays than the segmentation between urban residents and rural migrants, which was more dominant at the earlier stage of reform.

Robustness checks of decomposition

A limitation of the above results is that the decomposition does not consider in-kind income, welfare entitlements and employee benefits that many urban locals enjoy (Fan, 2008). Therefore, the degree of discrimination against migrant workers may be underestimated if these non-wage benefits are considered. To address this concern we conducted a robustness check by employing the method in Yue et al. (2010) which considers the proportion of non-wage values in total real income when decomposing the wage differentials between monopolistic and competitive industries.

Following this method, we assumed that the underestimated proportions of urban locals’ income were identical across the distribution, while the underestimated values increased as wages increased. This assumption is reasonable since non-wage benefits such as bonuses and employee benefits are also determined by workers’ human capital and employment status. The specific procedure is first to multiply the wage of urban locals by an adjustment factor larger than one, while holding that of rural or urban migrant workers constant, and then to decompose the wage differentials between locals and migrants.

According to our examinations of other surveys, such as the 2008 Chinese General Social Survey, the 2005 China Urban Labour Survey and the 2008 Rural Migrant Workers in Urban Pearl River Delta Survey, the non-cash income, including in-kind income, bonuses and welfare entitlements, contributed 8–25% to urban locals’ real income, while the contribution to migrants’ real income was relatively small. Therefore, this study uses three adjustment factors (that is, 1.1, 1.2 and 1.5) to decompose the differentials based on the presupposition that urban locals’ real income is undervalued by 9%, 17% and 33%, respectively.⁶

In general, the results are consistent with those based on observed wages (see Table 6). The only difference is that migrant workers are discriminated against at almost all levels compared with urban locals, and the degree of discrimination becomes greater with the increment of undervalued degree in distribution.

Table 6.

Discrimination percentage from adjusted quantile decomposition of distributional wage distribution between urban locals and migrant workers.

	Quantiles
	0.1	0.2	0.3	0.4	0.5	0.6	0.7	0.8	0.9
Rural migrants vs. urban locals
Observed wage	97.07	−24.8	−8.2	−38.3	37.62	31.9	48.05	88.38	125.78
1.1-fold	101.03	−13.38	−2.32	−20.04	35.57	38.63	60.01	90.19	129.49
1.2-fold	120.45	−2.67	2.02	−9.78	46.25	57.9	76.72	99.61	146.24
1.5-fold	150.69	4.89	10.27	15.41	70.03	78.45	90.05	136.78	171.89
Urban migrants vs. urban locals
Observed wage	18.74	22.15	17.16	16.26	32.52	28.51	−90.61	−154.87	−20.9
1.1-fold	35.29	39.06	25.77	20.54	40.37	35.91	−83.41	−131.14	−17.05
1.2-fold	42.91	53.48	33.98	33.18	54.88	43.27	−67.99	−119.91	−11.67
1.5-fold	62.02	70.33	60.17	55.39	67.22	60.06	−58.24	−100.55	−0.51
Rural migrants vs. urban migrants
Observed wage	6.05	16.31	13.85	9.78	27.61	41.28	21.65	7.04	26.57
1.1-fold	8.98	20.78	21.07	16.54	33.06	49.82	30.22	13.02	33.48
1.2-fold	10.54	27.09	26.59	25.39	45.15	55.35	47.19	22.11	50.27
1.5-fold	19.76	38.39	34.79	42.17	58.07	70.13	59.31	35.74	66.53

In summary, the decomposition results from both the original specification and robustness checks suggest that the hukou system remains active in the segmentation of China’s urban labour market. The segmented urban labour market has resulted in substantial discrimination against migrant workers at most wage levels compared with urban workers. It has also resulted in reduced labour productivity because urban locals may receive a wage higher than their marginal productivity of labour (MPL), while migrant workers may gain a wage lower than their MPL.

In addition, urban migrants are also treated differently from urban locals. A comparison between the wage discrimination against urban migrants compared with urban locals and the discrimination against rural migrants compared with urban migrants suggests that nowadays the segmented labour market is mainly driven by the segmentation between locals and non-locals, rather than by the segmentation between urban residents and rural migrants.

Conclusion

This study presents a comprehensive analysis of discrimination against migrant workers in wage distribution by adopting an extended framework which considers the dual segmentations between urban residents and rural migrants and between locals and non-locals. Urban migrants are included in this framework to examine the interaction between these two segmentations and their relative importance in the segmented labour market.

The results confirm that there are unfair labour market conditions in urban China due to the two segmentations that were induced by the hukou system. Compared with urban locals, rural migrants are only discriminated against above the medium-wage level, while urban migrants are discriminated against below the medium-wage level. The greater degree of wage discrimination against rural migrants reflects the fact that they face discrimination arising from both segmentations in the urban labour market, while urban migrants only face discrimination arising from the segmentation between locals and non-locals.

Contrary to some previous studies on the mean wage differentials in China’s urban labour market, urban locals (rather than migrant workers) face discrimination at other wage levels mainly because of the difference in employment structure. Rural migrants are more desired by employers to undertake low-wage jobs characterised by labour-intensive, low-skilled and hazardous conditions which are typically shunned by urban locals, while urban migrants are more favoured to undertake high-wage jobs characterised by human capital-intensive and high-skilled conditions. In addition, at most wage levels, the degree of discrimination against rural migrants estimated in this study is greater than that reported in previous research. This indicates that taking the diverse composition of migrant workers and wage inequality into account might allow for more accurate identification of mistreatment for migrant workers.

When we compared discrimination against urban migrants and rural migrants, we found that the degree of discrimination against urban migrants compared with urban locals was greater than that against rural migrants compared with urban migrants, which suggests that the segmentation between locals and non-locals has played a leading role in the current urban labour market. This reflects a profound transformation from a hukou-dominated urban–rural dichotomy in China’s urban labour market for approximately the first 20 years following the economic reforms, to a segmentation between locals and migrants in recent years. Similar or even stronger results could also be obtained if non-cash income is taken into consideration, as the robust checks demonstrate.

The present study indicates that while there have been fundamental changes in the hukou system since the 1990s, which resulted in an increasing population mobility and a significant socio-economic development, the reforms still lagged far behind economic development. The hukou system continues to affect social and labour stratifications and imposes severe discrimination against migrants. Discrimination causes large wage differentials between urban locals and migrants and reduces the labour productivity in urban China. This study points to the need for further reform of the hukou system and effective anti-discrimination labour market policies to promote equal pay and equal access to employment. Furthermore, urban migrants should be considered when urban labour market policies and regulations are formulated.

This study empirically extends the application of discrimination theory and segmented labour market theory to urban migrants and to a transitional economy, and contributes a clearer characterisation of migrant workers’ experience in the labour market. It should be noted that, as with most studies on labour market discrimination, this study also has some limitations in measuring the degree of wage discrimination. The unexplainable proportion of wage differentials, which have been generally interpreted as discrimination in most research on wage and employee benefit differentials, may include such factors as workers’ ability to access information and to adapt to urban environments, or other non-productivity-related factors. They may bias the estimation of degree of wage discrimination. However, it is generally accepted that a large proportion of unexplained wage differentials should be attributed to the different treatment for migrant workers (Meng and Zhang, 2001).

Footnotes

Acknowledgements

The authors thank the anonymous reviewers for their useful comments.

Funding

This project was funded by the Australian Research Council (DP0773060).

1.

Different from some other surveys focusing solely on migrant workers, this study aims to compare the three groups in a unified sampling framework within the same neighbourhoods. However, it should be noted that some migrant workers also live in factory dormitories, construction sites and so on. Therefore, the data collected in this study does not represent the general residential pattern of migrant workers. Recently there have been some studies focusing on urban neighbourhoods similar to those in our study (see, for example, Cheng and Wang, 2013).

2.

Local residents might have more grey income and non-income fringe benefits than migrants. However, probing into and measuring these ‘grey incomes’ and non-income components comprises a common and long-lasting problem facing most, if not all, household surveys in China. Nonetheless, in the robustness check of decomposition later, we consider such potential bias and the results are consistent.

3.

Readers who are interested in the quantile regression can refer to a methodology paper by .

4.

We ran the same specification on monthly wages and found that rural migrants have a higher return to education than urban locals, but this is largely due to their longer working hours.

5.

F-test results are available from the authors.

6.

For example, the adjustment factor 1.1 is estimated based on the 2008 Chinese General Social Survey data, which was jointly administrated by Renmin University and Hong Kong University of Science and Technology. In this survey urban locals’ average hourly income (9.15 yuan; including wages, in-kind income, bonuses and welfare entitlements) is 1.1 times the hourly wage (8.39 yuan).

References

Aigner

Cain

(1977) Statistical theories of discrimination in labor markets. Industrial and Labor Relations Review 30: 175–187.

Appleton

Knight

Song

. (2004) Contrasting paradigms: Segmentation and competitiveness in the formation of the Chinese labour market. Journal of Chinese Economic and Business Studies 2: 185–205.

Arabsheibani

Wang

(2008) Asian–white male wage differentials in the United States. Applied Economics Letters 17: 37–43.

Ashenfelter

(1970) Changes in labor market discrimination over time. The Journal of Human Resources 5: 403–430.

Becker

(1957) The Economics of Discrimination. Chicago, IL: University of Chicago Press.

Blau

Kahn

(2003) Understanding international differences in the gender pay gap. Journal of Labour Economics 21: 106–144.

Borjas

(1995) Assimilation and changes in cohort quality revisited: What happened to immigrant earnings in the 1980s? Journal of Labor Economics 13: 201–245.

Bourguignon

Ferreira

(2005) Decomposing changes in the distribution of household incomes: Methodological aspects. In: Bourguignon

Ferreira

Lustig

(eds) The Microeconomics of Income Distribution Dynamics. Oxford: Oxford University Press.

Cai

Wang

(1999) The sustainability of economic growth and the labour contribution in China. Economic Research Journal 34: 62–68.

10.

Cai

Wang

(2001) Household registration system and labour market protection. Economic Research Journal 12: 41–49.

11.

Cain

(1976) The challenge of segmented labor market theories to orthodox theory: A survey. Journal of Economic Literature 14: 1215–1257.

12.

Chan

(2002) The culture of survival: Lives of migrant workers through the prism of private letters. In: Link

Madsen

Pickowicz

(eds) Popular China: Unofficial Culture in a Globalizing Society. Boulder, CO: Rowman and Littlefield, pp. 163–188.

13.

Chan

(1996) Post-Mao China: A two-class urban society in the making. International Journal of Urban and Regional Research 20: 134–150.

14.

Chen

Dai

. (2010) The trend of the Gini coefficient of China. Brooks World Poverty Institute Working Paper Series, No. 109.

15.

Cheng

Wang

(2013) Do neighbourhoods have effects on wages? A study of migrant workers in urban China. Habitat International 38: 222–231.

16.

Cheng

Wang

Chen

(2013) Labor contract, trade union membership and workplace relations: A study of migrant workers in Guangdong province, China. In: Hao

Chen

(eds) Social Issues in China: Gender, Ethnicity, Labor, and the Environment. New York: Springer, pp. 183–206.

17.

Constant

Massey

(2005) Labor market segmentation and the earnings of German guestworkers. Population Research and Policy Review 24: 489–512.

18.

Démurger

Gurgand

. (2009) Migrants as second-class workers in urban China? A decomposition analysis. Journal of Comparative Economics 37: 610–628.

19.

Doeringer

Piore

(1985) Internal Labor Markets and Manpower Analysis. Armonk, NY: M.E. Sharpe.

20.

Dong

X-y

Bowles

(2002) Segmentation and discrimination in China’s emerging industrial labor market. China Economic Review 13: 170–196.

21.

Fan

(2002) The elite, the natives, and the outsiders: Migration and labor market segmentation in urban China. Annals of the Association of American Geographers 92(1): 103–124.

22.

Fan

(2003) Rural–urban migration and gender division of labor in transitional China. International Journal of Urban and Regional Research 27: 24–47.

23.

Fan

(2008) China on the Move: Migration, the State, and the Household. London: Routledge.

24.

Fortin

Lemieux

Firpo

(2010) Decomposition methods in economics. National Bureau of Economic Research Working Paper Series, No. 16045.

25.

Freeman

(2013) How Does China’s New Labor Contract Law Affect Floating Workers? National Bureau of Economic Research.

26.

Goldstein

(1991) Permanent and Temporary Migration Differentials in China. Honolulu, Hawaii: East-West Center.

27.

Guenter

(2000) Native-Immigrant Wage Differentials in Germany – Assimilation, Discrimination, or Human Capital? Universitaet Augsburg, Institute for Economics.

28.

Guo

Cheng

(2010) Labour market disparity, poverty, and inequality in urban China. China Perspectives 84: 16–31.

29.

Guo

Iredale

(2004) The impact of hukou status on migrants’ employment: Findings from the 1997 Beijing Migrant Census. International Migration Review 38: 709–731.

30.

Guo

Zhang

(2012) The urban labor market status of China’s floating population: A three-group approach. Population Research 36: 3–14.

31.

Gustafson

Gwartney

Haworth

(1975) Earnings, productivity, and changes in employment discrimination during the 1960s. The American Economic Review 65: 158–168.

32.

Gustafsson

(2000) Economic transformation and the gender earnings gap in urban China. Population Economics 13: 305–329.

33.

Juan

C-D

César

R-G

(2008) Analysis of wage differences between native and immigrant workers in Spain. Spanish Economic Review 10: 109–134.

34.

Kim

Tamborini

(2006) The continuing significance of race in the occupational attainment of whites and blacks: A segmented labor market analysis. Sociological Inquiry 76: 23–51.

35.

Knight

Song

(1999) The Rural–Urban Divide Economic Disparities and Interactions in China. Oxford; New York: Oxford University Press.

36.

Knight

Yueh

(2004) Job mobility of residents and migrants in urban China. Journal of Comparative Economics 32: 637–660.

37.

Knight

Yueh

(2009) Segmentation or competition in China’s urban labour market? Cambridge Journal of Economics 33: 79–94.

38.

Koenker

Hallock

(2001) Quantile regression. Journal of Economic Perspectives 16(4): 143–156.

39.

Laurence

JCM

(2002) Urban transformation in China, 1949–2000: A review and research agenda. Environment and Planning A 34: 1545–1569.

40.

Lee

(2012) Decomposing wage differentials between migrant workers and urban workers in urban China’s labor markets. China Economic Review 23: 461–470.

41.

(2008) Rural Migrant Workers in China: Scenario, Challenges and Public Policy. Geneva: Policy Integration and Statistics Department International Labour Office.

42.

Liu

P-W

Meng

Zhang

(2000) Sectoral gender wage differentials and discrimination in the transitional Chinese economy. Journal of Population Economics 13: 331–352.

43.

Liu

(2005) Institution and inequality: The hukou system in China. Journal of Comparative Economics 33: 133–157.

44.

Song

(2006) Rural–urban migration and wage determination: The case of Tianjin, China. China Economic Review 17: 337–345.

45.

McDonald

Solow

(1981) Wage bargaining and employment. The American Economic Review 71: 896–908.

46.

Machado

JAF

Mata

(2005) Counterfactual decomposition of changes in wage distributions using quantile regression. Journal of Applied Econometrics 20: 445–465.

47.

Massey

Arango

Hugo

. (1994) An evaluation of international migration theory – The North-American case. Population and Development Review 20: 699–751.

48.

Maurer-Fazio

Dinh

(2004) Differential rewards to, and contributions of, education in urban China’s segmented labor markets. Pacific Economic Review 9: 173–189.

49.

Meng

(2001) The informal sector and rural–urban migration – A Chinese case study. Asian Economic Journal 15: 71–89.

50.

Meng

Zhang

(2001) The two-tier labor market in urban China. Journal of Comparative Economics 29: 485–504.

51.

Miller

Rummery

(1991) Male–female wage differentials in Australia: A reassessment. Australian Economic Papers 30: 50–69.

52.

National Bureau of Statistics of China (1982) 1982 Population Census of China. Beijing: China Statistics Press.

53.

National Bureau of Statistics of China (2009) A Note on the Income of Urban Residents in the First Half of the Year. Beijing: National Bureau of Statistics of China

54.

National Bureau of Statistics of China (2011) Press Release on Major Figures of the 2010 National Population Census. National Bureau of Statistics of China.

55.

National Population and Family Planning Commission of China (2010) Current living situation of migrant population in China: A pilot suvery of migrant population in five major cities. Population Research 34: 6–18.

56.

National Population and Family Planning Commission of China (2011) Report on China’s Migrant Population Development. Beijing: China Population Publishing House.

57.

Oaxaca

(1973) Male–female wage differentials in urban labor markets. International Economic Review 14: 693–709.

58.

Park

Wang

(2010) Migration and urban poverty and inequality in China. China Economic Journal 3: 49–67.

59.

Roberts

(1997) China’s ‘tidal wave’ of migrant labor: What can we learn from Mexican undocumented migration to the United States? International Migration Review 31: 249–293.

60.

Schultz

(1961) Investment in human capital. American Economic Review 51: 1–17.

61.

Smith

(2003) Labour Economics. London: Taylor and Francis.

62.

Song

Zenou

Ding

(2008) Let’s not throw the baby out with the bath water: The role of urban villages in housing rural migrants in China. Urban Studies 45: 313–330.

63.

Sun

Fan

(2011) China’s permanent and temporary migrants: Differentials and changes, 1990–2000. Professional Geographer 63: 92–112.

64.

Sylvie

Marc

. (2008) Migrants as second-class workers in urban China? A decomposition analysis. Groupe d’Analyse et de Théorie Economique (GATE), Centre national de la recherche scientifique (CNRS), Université Lyon 2, Ecole Normale Supérieure.

65.

Wang

Zuo

Ruan

(2002) Rural migrants in Shanghai: Living under the shadow of socialism. International Migration Review 36: 520–545.

66.

Wang

Appelbaum

Degiuli

. (2009) China’s new labour contract law: Is China moving towards increased power for workers? Third World Quarterly 30: 485–501.

67.

Wong

DFK

Song

(2007) Rural migrant workers in urban China: Living a marginalised life. International Journal of Social Welfare 16(1): 32–40.

68.

Yao

Xue

(2008) The progress of the research on the floating population in China. Urban Problems 6: 69–76.

69.

Yue

Sicula

(2010) Are the high salaries of employees in monopoly industries justified? Social Science in China 3: 77–93.

70.

Zhang

Song

(2003) Rural–urban migration and urbanization in China: Evidence from time-series and cross-section analyses. China Economic Review 14: 386–400.

71.

Zhang

(2006) The local residents-migrants divide of urban social security and the floating population’s absence of social security. China Opening Herald 6: 51–56.

72.

Zhang

(2007) Urban–rural segregation to regional segregation. Population Research 31: 16–23.

73.

Zhou

Logan

(1989) Returns on human capital in ethnic enclaves: New York City’s Chinatown. American Sociological Review 54: 809–820.